journal of issn 1998-2062 - painsa · 2019. 3. 18. · journal of the south african chapter of the...

Journal of

The South African Chapter of the IASP

ISSN 1998-2062

Volume 14 Number 1

2019

Postoperative pain, analgesia, and recovery—bedfellows that cannot be ignored

What makes surgical nerve injury painful? A 4-year to 9-year follow-up of patients with intercostobrachial nerve resection in women treated for breast cancer

Psychosocial factors associated with persistent pain in people with HIV: a systematic review with meta-analysis

Postoperative opioids, endocrine changes, and immunosuppression

Journal of The South African Chapter of the IASPVolume 14 Number 1

Editorial

All correspondence to the editor should be addressed to: [email protected]

Patient recovery following surgery is influenced by many factors. The issue at hand has always been exactly which factors could we anticipate, and which factors could we modify so as to get the best post-surgical outcome.

The original work in this regard was done by Hendrik Kehlet who has visited our shores and has been a speaker at both SASA and Painsa. His original publication on the factors affecting post-operative morbidity and mortality caused renewed interest in the subject and this was followed by many guidelines and new special interest groups.

The PROSPECT group gathered data and made recommendations regarding eleven surgical procedures. Advice is given on topics from pre-operative evaluation and management through to the recommended anaesthetic

technique, pain interventions, and post-operative management so as to optimize the post-surgical response and outcome for each of the procedures. This collaboration is ongoing, and more procedures will be added to the existing list.

The next development was that of Enhanced Recovery After surgery (ERAS). Once again, the various contributors evaluated surgical procedures and their outcomes but this time included methodology and suggestions to not only improve the surgical outcome but also advice on how to shorten hospitalization.

I have included the latest review by Kehlet as published by IASP in their Biennial Review of Pain. As the author states, “the article is not intended to be systematic review on the complex problem of the pathogenesis and interaction between pain and postoperative recovery, but rather to represent a selective update on recent developments with an emphasis on the multidisciplinary challenges that lie ahead not only to improve postoperative pain management, but also the necessary strategies for the integration of pain management into enhanced recovery programs (enhanced recovery after surgery.”

We need to realise that everything we wish to achieve for our patients following surgery is based on simple principle which, if addressed, will improve patient outcome. These is a multimodal situation and requires that the following factors be considered.

1. Pre-operative optimization of the patient and his/her pathological state2. Attenuation of the surgical stress response3. Optimal pain relief4. Mobilization and exercise5. Commencement of oral nutrition

These are not strange foreign concepts but rather factors which we should be able to manage in our own practices.

I trust that this review serves to update us as to the state-of-the-art principles that we must incorporate into our peri-operative patient management.

Dr. Milton RaffBSc MB ChB FFA(SA)

1

EDITORDr M Raff

BSc (WITS), MBChB (Pret), FFA (SA)

EDITORIAL BOARDProf H Meyer

MBChB(Pret) MPraxMed(Pret) MFGP(SA)

Prof C L Odendaal MBChB, MMed(Anest),

GFN(SA)

Prof D MitchellBSc Hons, MSc, PhD

(all University of the Witwatersrand)

Dr S BaumannBA. Mb.Ch.B.(U.C.T.),

P.G.C.E.(University College of Wales), M.R.C.Psych.(London),

F.C.Psych (S.A.)

Mrs P BergerBSc Physio (Wits), Acup (SA)

Prof E FrohlichMD(Tel-Aviv), DA(SA),

FCA(SA), Master (Med) Pain Management (Syd)

This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically the rights for translation, reprinting reuse of illustrations, broad-casting, reproduction of CD-Rom, microfilm, online publication, or in any other way, and storage in data banks.

The use of registered names trademarks etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt for the relevant laws and regulations and therefore free for general use.

Product liability: the publishers cannot guarantee the accuracy of any information about the publication of medications contained in this publication. In every individual case, the user must check such information by consulting the relevant literature.

PUBLISHER / MEDSPEC PUBLISHING / ADVERTISING & RATES

Reni Rouncivell, Tel: (012) 657 2327 Fax: 086 561 5122, Cell: 082 441 6904, e-mail: [email protected], Private Bag X1036, Lyttelton, South Africa 0140

SUBSCRIPTIONS & ACCOUNTS

Chantel du Toit, Tel: 082 385 5524, e-mail: [email protected]/[email protected]

BIENNIAL REVIEW OF PAINPostoperative pain, analgesia, and recovery—bedfellows that cannot be ignoredHenrik Kehlet

4

RESEARCH PAPERWhat makes surgical nerve injury painful? A 4-year to 9-year follow-up of patients with intercostobrachial nerve resection in women treated for breast cancerLaura Mustonen, Tommi Aho, Hanna Harno, Reetta Sipilä, Tuomo Meretoja, Eija Kalsoa

10

SYSTEMATIC REVIEW AND META-ANALYSISPsychosocial factors associated with persistent pain in people with HIV: a systematic review with meta-analysisWhitney Scott, Chinar Arkuter, Kitty Kioskli, Harriet Kemp, Lance M. McCrackena, Andrew S.C. Rice, Amanda C. de C. Williams

21

PAIN CLINICAL UPDATESPostoperative opioids, endocrine changes, and immunosuppressionSimon Haroutounian

37

2

Biennial Review of Pain

Postoperative pain, analgesia, andrecovery—bedfellows that cannot be ignoredHenrik Kehlet

1. Introduction

Surgical injury is followed by pain, risk of organ dysfunctions,morbidity, prolonged hospital stay (length of stay [LOS]),readmissions, and delayed convalescence. However, despitean explosion in our understanding of the pathophysiology ofacute pain, the development of a few new analgesics,77 novelmethods of drug delivery, and introduction of pain-reducingminimal invasive surgical techniques, acute postoperative painmanagement remains a huge challenge for optimization.26

This is emphasized by the fact that postoperative painmanagement allowing for early mobilization is a prerequisitefor improving recovery and decreasing the risk of complica-tions.34 The complex problems involved in postoperative pain,analgesic interventions, and outcome have been emphasizedin several surveys over the past decades, but apparently withonly small improvements, despite the existence ofseveral international guidelines for perioperative painmanagement.9,26,58

The present article is a narrative review based on a personalengagement over the past 30 years in acute postoperative painresearch, outcome pathophysiology, and strategies forimprovement.28–30,34 The article is not intended to bea systematic review on the complex problem of the pathogen-esis and interaction between pain and postoperative recovery,but rather to represent a selective update on recent develop-ments with an emphasis on the multidisciplinary challengesthat lie ahead not only to improve postoperative painmanagement, but also the necessary strategies for theintegration of pain management into enhanced recoveryprograms (enhanced recovery after surgery [ERAS]) (Fig. 1and Table 1).30

2. Surgical stress

The surgical injury leads to a complex cascade of responsesmediated by afferent nervous stimuli28 together with complexinflammatory–immunological responses,1 both of which maybe important for pain and subsequent risk of organ dysfunc-tion. In this context, the role of different anesthetic–analgesic

techniques to reduce afferent stimuli and thereby pain hasbeen well established for decades.28,34,38 However, morerecent studies have emphasized the role of the inflammatory–immunological responses for not only recovery of postopera-tive pain, but also all over functional recovery.17 The futurechallenges, therefore, should focus on a more detailedunderstanding of which responses to modify and how muchto reduce pain and enhance recovery (Fig. 2). In this context,the use of minimally invasive surgery is well established toreduce pain, but more importantly also to reduce theinflammatory–immunological responses, thereby serving asan optimal basis for further improvements.72 Nevertheless,despite the efficacy of regional and other analgesic techniquestogether with minimally invasive surgery, there is a need fordevelopment of further pharmacological techniques to modifyundesirable stress responses. In this context, the use of highdoses of glucocorticoids (.15 mg dexamethasone or equiv-alents) before the skin incision has repeatedly been demon-strated to have profound analgesic effects and furthermore toreduce postoperative fatigue and other undesirable patho-physiological responses to surgery13,47,50,64,70 and conse-quently with the potential to enhance recovery. Because safetydata are promising,27,70 there is an urgent need for procedure-specific, dose-finding studies and the potential for repeatdosing in certain high-pain surgeries/patients.

3. Principles for perioperative pain management

Based on the concept of “fast-track surgery” or ERAS, neitherpain nor recovery can be sufficiently managed by a single-modality treatment because of the multifactorial problembetween pain management and recovery.29,30,34,38,75 Conse-quently, and because of the undesirable efficacy–safety ratio ofopioids, the concept of multimodal opioid-sparing analgesia wasintroduced more than 2 decades ago.33 The concept has beenpromising,76 but not widely used.46 Furthermore, several ques-tions remain to be answered regarding the number of nonopioidanalgesic techniques to be recommended, the safety aspects,and especially, the need for a procedure-specific analysisbecause different procedures have different pain responses,pain consequences, and painmechanisms.25 In addition, amajorchallenge persists to analyze the enormous amount of post-operative analgesic studies not only on a procedure-specificbasis, but also with regard to efficacy vs side effects. Thisespecially applies to randomized controlled trials when a givenanalgesic is studied when combined with simple evidence-basedbasic multimodal analgesia (peripheral local anesthetics, para-cetamol, and nonsteroidals) vs pure placebo-controlled stud-ies.26 Finally, a more critical assessment of the scientific method(bias) in most analgesic studies is required.15,16 Consequently,

Sponsorships or competing interests that may be relevant to content are disclosed

at the end of this article.

Section of Surgical Pathophysiology, Rigshospitalet, Copenhagen University,

Copenhagen, Denmark

Address: Section of Surgical Pathophysiology, Rigshospitalet, 7621, Blegdamsvej

9, DK-2100 Copenhagen, Denmark. Tel.: 145 3545 4774; fax: 145 3545 6543.

E-mail address: [email protected] (H. Kehlet).

PAIN 159 (2018) S11–S16

© 2018 International Association for the Study of Pain

http://dx.doi.org/10.1097/j.pain.0000000000001243

September 2018·Volume 159·Number 9·Supplement 1 www.painjournalonline.com S11

Copyright � 2018 by the International Association for the Study of Pain. Unauthorized reproduction of this article is prohibited.4

there is a major need for reevaluation of current guidelines forperioperative pain management,26 subsequently to be integratedinto ERAS programs.75

4. Pain assessment

For many years, the classic postoperative pain assessment hasused a variety of validated scales (visual analogue scale, numericrating scale, etc.) which obviously is important when assessingefficacy of analgesics. However, when discussing pain manage-ment and postoperative recovery, it has been emphasized formany years that pain assessment must include assessment ina procedure-specific relevant activity,63 which unfortunately oftenis not performed. In addition, there is a need for future pain andrecovery studies with a focus on the postoperative paintrajectories and resolution on a procedure-specific basis.3,21 Asa supplement to specific pain assessment including pain duringwell-defined function, opioid requirements have been a commonoutcome in many analgesic studies. However, future studiesneed to focus on the clinical consequences of the achievedopioid-sparing and not milligram opioid spared per se.18 In thiscontext, recent studies have suggested that baseline early

postoperative opioid consumption in the control group needsmore consideration when discussing interpretation of analgesicefficacy in postoperative pain studies compared with procedure-specific considerations.4,14

In addition to the assessment of pain during well-definedfunction on a procedure-specific basis, we need to prioritizefuture analgesic studies on a patient-specific basis becausepostoperative pain responses (and recovery?) may be dependenton preoperative nociceptive function, preoperative psychosocialrisk factors (catastrophizing and anxiety) and preoperative opioiduse, all well known to be high-pain responders and requiringadditional analgesic interventions.18 Finally, the improved painassessment algorithms must be combined with not only sub-jective, but also objective assessment of postoperative recovery(see below).

5. Pain/analgesia and postoperativeorgan dysfunctions

5.1. Cardiopulmonary dysfunction

The pathogenesis of the classic pulmonary complications(atelectasis and pneumonia) seen with traditional care is mostly

Figure 1. Concept of “fast-track” or “enhanced recovery.”

Table 1

Pathogenic factors to be considered for optimizing postoperative recovery based on the question “Why is the patient in hospital

today?”

Organ dysfunction (“surgical stress”)

Hypothermia-induced morbidity

Pain

PONV/ileus

Fluid excess/hypovolemia

Cognitive dysfunction/sleep disturbances

Immobilization/semistarvation

Blood management

Fatigue (early/late)

Traditions (tubes, drains, restrictions, etc.)

PONV, postoperative nausea and vomiting.

S12 H. Kehlet·159 (2018) S11–S16 PAIN®


related to pain, fluid overload, and immobilization, all of which aremanageable in a fully implemented ERAS program and wherepulmonary complications have been significantly reduced even inabdominal and thoracic procedures.5,37,38 These results againemphasize themultifactorial pathogenesis of postoperative organdysfunction and recovery and where sufficient analgesia repre-sents only 1 factor30 (Fig. 1 and Table 1).

The pathogenesis of cardiac complications (arrhythmias andinfarction) is multifactorial,19 but where pain, activation of thesympathetic nervous system, hemodynamics, the inflammatoryresponse, and decreased pulmonary function (hypoxemia) mayplay a major role for the oxygen supply–demand ratio. Conse-quently, cardiac complications may be reduced by sufficientanalgesic techniques and afferent blockade28,29,39,60 as well asthe other factors included in ERAS programs. The specific role ofhypoxemia in relation to sleep disturbances and mobilizationrequires further study (see below) but where pain, analgesictechniques, fluid management, and hemodynamic optimizationare relevant pathogenic factors. It is beyond the scope of thisarticle to review the multifold pathogenic factors and interven-tional studies to reduce postoperative myocardial infarction.59

5.2. Postoperative nausea and vomiting and paralytic ileus

Pain per se will contribute to postoperative nausea and vomiting,but the pathogenesis includes additional factors such as theneuroinflammatory responses and use of opioids—all modifiablein a fully implemented ERAS program. Postoperative paralyticileus represents a specific surgical stress response with undesir-able consequences such as discomfort, pulmonary complica-tions, and insufficient oral nutrition. The mechanisms are wellestablished as well as the effect of interventional techniques31,71

(Fig. 3), and where a combination of these techniques includingoptimized opioid-sparing analgesia with thoracic epidural anal-gesia, systemic opioid-sparing techniques, use of peripheralopioid antagonists, and minimally invasive surgery have provenefficient thereby allowing for early oral feeding, which otherwisemay enhance recovery.

5.3. Thromboembolic complications

Postoperative immobilization may be followed by an increasedrisk of thromboembolic and pulmonary complications.

Consequently, improved pain management combined withstrategies for an enhanced recovery program30 will allow forearlymobilization and a subsequent lower risk of thromboemboliccomplications24 as well as pulmonary complications.5,38

5.4. Cognition and sleep

It is well established that major surgery and especially in elderlypatients may be followed by a risk of delirium and late cognitivedysfunction.2,53,62 Although the mechanisms still remain to beestablished44 (Fig. 4), they aremultifactorial including pain, opioiduse, sleep disturbances, and neuroinflammation. Consequently,ERAS programs have been shown to reduce the risk of clinicallyrelevant delirium in hip and knee replacement and colonicsurgery.45,54 Also, the problems of late, more subtle cognitivedysfunction may be reduced by the same approach,43 but morestudies are required on a combination of optimized opioid-sparing analgesia, reduction of inflammatory–immunologicalresponses,68 combined with early mobilization and discharge.In this context, the role of postoperative sleep disturbances needsfurther studies because the profound changes in sleep architec-ture so far have not been demonstrated to be avoidable, although

Figure 2. Clinical techniques to modify surgical stress responses. PONV, postoperative nausea and vomiting.

Figure 3.Current techniques to reduce the postoperative ileus. ↑ prolongationand ↓ enhanced recovery of the ileus.



a shorter duration of sleep disturbances was observed in a well-established fast-track hip and knee ERAS program.42 Altogether,the specific role of sleep disturbances for cognitive problemsrequires further study because reappearance of rapid eyemovement sleep may be associated with episodic hypoxemiaand thereby potentially contributing to cardiac complications andcognitive dysfunction.29 Finally, postoperative sleep disturbancesshould receive more attention because sleep deprivation maysensitize the nociceptive system10,61 and because subjective-reported sleep problems are associated with more severepostdischarge pain and prolonged recovery in orthopedic jointreplacement patients.12

5.5. Persistent postsurgical pain

An overwhelming amount of data has shown that most surgicalprocedures may be followed by a risk of persistent painsyndromes.36,65 The pathogenesis is multifold including thesurgical approach (nerve-sparing surgery), preoperative psycho-social risk factors (high-pain responders),8,36,73 and potentiallypain management.65 Unfortunately, the concept of “preemptive”or “preventive” analgesia to reduce central sensitization andpersistent pain has overall been disappointing.8,36,65 Futureanalgesic strategies to reduce persistent postsurgical pain willrequire a focus on modification of the neuroinflammatoryresponses.22,23,78

5.6. Pain management in enhanced recovery aftersurgery programs

Because optimized postoperative pain management representsa prerequisite to enhance recovery,34 it was widely assumed thatan improved outcome would automatically be followed byprovision of more efficient analgesia. However, during the past3 decades, the results from plenty of perioperative analgesicstudies, including the effects of “acute pain” services,74 havebeen rather disappointing by not being able to demonstrateimproved analgesia per se to led to major improvement ofoutcomes (morbidity, LOS, convalescence, and persistentpain).8,36,41,51,55,56,75 These somewhat disappointing findingsare again explained by themultifactorial problem of postoperativerecovery and where pain/analgesia represents only 1 factor (Fig.1 and Table 1). Consequently, future studies are required whereoptimized analgesia is integrated on a procedure-specific basis

into a fully implemented ERAS program. Such studies haveuniformly shown a major reduction in LOS and “medical”complications without increasing readmissions,37,48 emphasiz-ing the need for the integrated recovery approach.67 Althoughoptimized pain management combined with an ERAS programreduces the need for postoperative hospitalization (LOS), severalclinically relevant problems require further attention with regard topostdischarge regain of muscle function, and return to normalactivities and work. Therefore, future studies should continue toask the question “Why is the patient in hospital today?”29

(Table 1) combined with other pathogenic factors to enhancerecovery.37 In this context, an ERAS program may lead tosubstantial reduction in perioperative opioid use69 and therebypotentially reduce the significant problem of short- and long-termuse of opioids after surgery6,66 by specifying pain expectations forpatients and standardizing postoperative opioid prescribing.

Furthermore, an important question to be addressed is how toassess postdischarge recovery either by subjective reports(patient-reported outcome measures, which includes “pain”questions)20,40 or by the many new instruments that canobjectively assess recovery (sleep, activity, cardiovascularfunction, hypoxemia, etc.)52 Importantly, preliminary data frommajor joint arthroplasty have shown a discrepancy betweenimprovements in patient-reported outcomes compared with lackof sufficient recovery when assessed by objective measure-ments.11,49,57 The explanations hereto may be several, calling forfuture postoperative recovery studies on the relative role ofpostdischarge pain vs other factors.

6. Summary

The complex clinical problem of postoperative pain managementper se and the potential to improve recovery and reducemorbiditycontinues to represent a major challenge, and further empha-sized by the change in patient demographics including moreelderly and high-risk surgical patients. Nevertheless, the progressin the scientific approach to perform and interpret analgesicstudies made within the past decades,18 combined with moreefficient and safe multimodal opioid-sparing analgesic techni-ques and with an integrated multidisciplinary approach to use theoptimized analgesia into an ERAS program is promising.However, despite the progress demonstrated with this approach,significant challenges exist to improve in-hospital and post-discharge painmanagement in relation to functional recovery and

Figure 4. Pathogenesis of postoperative delirium and cognitive dysfunction.

S14 H. Kehlet·159 (2018) S11–S16 PAIN®


to refine the many different components of ERAS programs.32,37

Consequently, theremay be light at the end of the tunnel, but it willrequire an intensified multidisciplinary collaboration7,35 betweenbasic pain researchers, anesthesiologists, surgeons, surgicalnurses, and physiotherapists, altogether within an enhancedrecovery program framework.37

Conflict of interest statement

The author has no conflict of interest to declare.

Article history:Received 28 February 2018Received in revised form 6 April 2018Accepted 9 April 2018

References

[1] Alazawi W, Pirmadjid N, Lahiri R, Bhattacharya S. Inflammatory andimmune responses to surgery and their clinical impact. Ann Surg 2016;264:73–80.

[2] Aldecoa C, Bettelli G, Bilotta F, Sanders RD, Audisio R, Borozdina A,Cherubini A, Jones C, Kehlet H, MacLullich A, Radtke F, Riese F, SlooterAJ, Veyckemans F, Kramer S, Neuner B, Weiss B, Spies CD. EuropeanSociety of Anaesthesiology evidence-based and consensus-basedguideline on postoperative delirium. Eur J Anaesthesiol 2017;34:192–214.

[3] Andersen LO, Gaarn-Larsen L, Kristensen BB, Husted H, Otte KS, KehletH. Subacute pain and function after fast-track hip and knee arthroplasty.Anaesthesia 2009;64:508–13.

[4] Andreae MH, Pace NL. A novel approach to synthesize the evidence onanalgesic adjuvants for postoperative pain. Anesth Analg 2018;126:377–81.

[5] Basse L, Thorbol JE, Lossl K, Kehlet H. Colonic surgery with acceleratedrehabilitation or conventional care. Dis Colon Rectum 2004;47:271–7.

[6] Brummett CM, Waljee JF, Goesling J, Moser S, Lin P, Englesbe MJ,Bohnert ASB, Kheterpal S, Nallamothu BK. New persistent opioid useafter minor andmajor surgical procedures in US adults. JAMA Surg 2017;152:e170504.

[7] Cannesson M, Ani F, Mythen MM, Kain Z. Anaesthesiology andperioperative medicine around the world: different names, same goals.Br J Anaesth 2015;114:8–9.

[8] Chapman CR, Vierck CJ. The transition of acute postoperative pain tochronic pain: an integrative overview of research on mechanisms. J Pain2017;18:359.e351–359.e338.

[9] Chou R, Gordon DB, de Leon-Casasola OA, Rosenberg JM, Bickler S,Brennan T, Carter T, Cassidy CL, Chittenden EH, Degenhardt E, GriffithS, Manworren R, McCarberg B, Montgomery R, Murphy J, Perkal MF,Suresh S, Sluka K, Strassels S, Thirlby R, Viscusi E, Walco GA, WarnerL, Weisman SJ, Wu CL. Management of postoperative pain: a ClinicalPractice Guideline from the American Pain Society, the AmericanSociety of Regional Anesthesia and Pain Medicine, and the AmericanSociety of Anesthesiologists’ Committee on Regional Anesthesia,Executive Committee, and Administrative Council. J Pain 2016;17:131–57.

[10] Chouchou F, Khoury S, Chauny JM, Denis R, Lavigne GJ. Postoperativesleep disruptions: a potential catalyst of acute pain? SleepMedRev 2014;18:273–82.

[11] Cooper NA, Rakel BA, Zimmerman B, Tonelli SM, Herr KA, Clark CR,Noiseux NO, Callaghan JJ, Sluka KA. Predictors of multidimensionalfunctional outcomes after total knee arthroplasty. J Orthop Res 2017;35:2790–8.

[12] Cremeans-Smith JK, Millington K, Sledjeski E, Greene K, Delahanty DL.Sleep disruptions mediate the relationship between early postoperativepain and later functioning following total knee replacement surgery.J Behav Med 2006;29:215–22.

[13] de la Motte L, Kehlet H, Vogt K, Nielsen CH, Groenvall JB, Nielsen HB,Andersen A, Schroeder TV, Lonn L. Preoperative methylprednisoloneenhances recovery after endovascular aortic repair: a randomized,double-blind, placebo-controlled clinical trial. Ann Surg 2014;260:540–9.

[14] Doleman B, Sutton AJ, Sherwin M, Lund JN, Williams JP. Baselinemorphine consumption may explain between-study heterogeneity inmeta-analyses of adjuvant analgesics and improve precision andaccuracy of effect estimates. Anesth Analg 2018;126:648–60.

[15] Fabritius ML, Geisler A, Petersen PL, Nikolajsen L, Hansen MS, KontinenV, Hamunen K, Dahl JB, Wetterslev J, Mathiesen O. Gabapentin for post-operative pain management—a systematic review with meta-analysesand trial sequential analyses. Acta Anaesthesiol Scand 2016;60:1188–208.

[16] Fabritius ML, Mathiesen O, Wetterslev J, Dahl JB. Post-operativeanalgesia: focus has been on benefit—are we forgetting the harm?Acta Anaesthesiol Scand 2016;60:839–41.

[17] Gaudilliere B, Fragiadakis GK, Bruggner RV, NicolauM, Finck R, TingleM,Silva J, Ganio EA, Yeh CG, Maloney WJ, Huddleston JI, Goodman SB,Davis MM, Bendall SC, Fantl WJ, Angst MS, Nolan GP. Clinical recoveryfrom surgery correlates with single-cell immune signatures. Sci TranslMed 2014;6:255ra131.

[18] Gilron I, Carr DB, Desjardins PJ, Kehlet H. Current methods andchallenges for acute pain clinical trials. Pain Rep 2018. doi: 10.1097/PR9.0000000000000647. [epub ahead of print].

[19] Helwani MA, Amin A, Lavigne P, Rao S, Oesterreich S, Samaha E, BrownJC, Nagele P. Etiology of acute coronary syndrome after noncardiacsurgery. Anesthesiology 2018. doi: 10.1097/ALN.0000000000002107.[epub ahead of print].

[20] Hossain FS, Konan S, Patel S, Rodriguez-Merchan EC, Haddad FS. Theassessment of outcome after total knee arthroplasty: are we there yet?Bone Joint J 2015;97-B:3–9.

[21] Houle TT, Miller S, Lang JE, Booth JL, Curry RS, Harris L, AschenbrennerCA, Eisenach JC. Day-to-day experience in resolution of pain aftersurgery. PAIN 2017;158:2147–54.

[22] Ji RR, Nackley A, Huh Y, Terrando N, Maixner W. Neuroinflammation andcentral sensitization in chronic and widespread pain. Anesthesiology2018. doi: 10.1097/ALN.0000000000002130. [epub ahead of print].

[23] Ji RR, Xu ZZ, Gao YJ. Emerging targets in neuroinflammation-drivenchronic pain. Nat Rev Drug Discov 2014;13:533–48.

[24] Jorgensen CC, Jacobsen MK, Soeballe K, Hansen TB, Husted H,Kjaersgaard-Andersen P, Hansen LT, Laursen MB, Kehlet H.Thromboprophylaxis only during hospitalisation in fast-track hip andknee arthroplasty, a prospective cohort study. BMJ Open 2013;3:e003965.

[25] Joshi GP, Kehlet H. Procedure-specific pain management: the road toimprove postsurgical painmanagement?Anesthesiology 2013;118:780–2.

[26] Joshi GP, Kehlet H. Guidelines for perioperative pain management: needfor re-evaluation. Br J Anaesth 2017;119:720–2.

[27] Jørgensen CC, Pitter FT, Kehlet H. Safety aspects of preoperative high-dose glucocorticoid in primary total knee replacement. Br J Anaesth2017;119:267–75.

[28] Kehlet H. Modification of responses to surgery and anesthesia by neuralblockade: clinical implications. In: Cousins MJ, Bridenbaugh PO, editors.Neural blockade in clinical anesthesia and management of pain.Philadelphia: Lippincott, 1987. p. 145–88.

[29] Kehlet H. 1993 John J. Bonica Lecture. Postoperative pain relief. A lookfrom the other side. Reg Anesth 1994;19:369–77.

[30] Kehlet H. Multimodal approach to control postoperative pathophysiologyand rehabilitation. Br J Anaesth 1997;78:606–17.

[31] Kehlet H. Postoperative ileus—an update on preventive techniques. NatClin Pract Gastroenterol Hepatol 2008;5:552–8.

[32] Kehlet H. ERAS implementation—time to move forward. Ann Surg 2018;267:998–9.

[33] Kehlet H, Dahl JB. The value of “multimodal” or “balanced analgesia” inpostoperative pain treatment. Anesth Analg 1993;77:1048–56.

[34] Kehlet H, Dahl JB. Anaesthesia, surgery, and challenges in postoperativerecovery. Lancet 2003;362:1921–8.

[35] Kehlet H, Delaney CP, Hill AG. Perioperative medicine–the second roundwill need a change of tactics. Br J Anaesth 2015;115:13–14.

[36] Kehlet H, Jensen TS, Woolf CJ. Persistent postsurgical pain: risk factorsand prevention. Lancet 2006;367:1618–25.

[37] Kehlet H, Jorgensen CC. Advancing surgical outcomes research andquality improvement within an enhanced recovery program framework.Ann Surg 2016;264:237–8.

[38] Kehlet H,Wilmore DW. Evidence-based surgical care and the evolution offast-track surgery. Ann Surg 2008;248:189–98.

[39] Khan SK, Malviya A, Muller SD, Carluke I, Partington PF, Emmerson KP,Reed MR. Reduced short-term complications and mortality followingEnhanced Recovery primary hip and knee arthroplasty: results from6,000 consecutive procedures. Acta Orthop 2014;85:26–31.

[40] Kleif J, Waage J, Christensen KB, Gogenur I. Systematic review of theQoR-15 score, a patient- reported outcome measure measuring qualityof recovery after surgery and anaesthesia. Br J Anaesth 2018;120:28–36.

[41] Kopp SL, Borglum J, Buvanendran A, Horlocker TT, Ilfeld BM, MemtsoudisSG, Neal JM, Rawal N, Wegener JT. Anesthesia and analgesia practicepathway options for total knee arthroplasty: an evidence-based review by the



American andEuropeanSocieties of Regional Anesthesia andPainMedicine.Reg Anesth Pain Med 2017;42:683–97.

[42] Krenk L, Jennum P, Kehlet H. Postoperative sleep disturbances afterzolpidem treatment in fast-track hip and knee replacement. J Clin SleepMed 2014;10:321–6.

[43] Krenk L, Kehlet H, Hansen TB, Solgaard S, Soballe K, Rasmussen LS.Cognitive dysfunction after fast-track hip and knee replacement. AnesthAnalg 2014;118:1034–40.

[44] Krenk L, Rasmussen LS, Kehlet H. New insights into the pathophysiologyof postoperative cognitive dysfunction. Acta Anaesthesiol Scand 2010;54:951–6.

[45] Kurbegovic S, Andersen J, Krenk L, Kehlet H. Delirium in fast-trackcolonic surgery. Langenbecks Arch Surg 2015;400:513–16.

[46] Ladha KS, Patorno E, Huybrechts KF, Liu J, Rathmell JP, Bateman BT.Variations in the use of perioperative multimodal analgesic therapy.Anesthesiology 2016;124:837–45.

[47] Lindberg-Larsen V, Ostrowski SR, Lindberg-Larsen M, Rovsing ML,Johansson PI, Kehlet H. The effect of pre-operative methylprednisoloneon early endothelial damage after total knee arthroplasty: a randomised,double-blind, placebo-controlled trial. Anaesthesia 2017;72:1217–24.

[48] Liu VX, Rosas E, Hwang J, Cain E, Foss-Durant A, Clopp M, Huang M,Lee DC, Mustille A, Kipnis P, Parodi S. Enhanced recovery after surgeryprogram implementation in 2 surgical populations in an integrated healthcare delivery system. JAMA Surg 2017;152:e171032.

[49] Luna IE, Kehlet H, Peterson B, Wede HR, Hoevsgaard SJ, Aasvang EK.Early patient-reported outcomes versus objective function after total hipand knee arthroplasty: a prospective cohort study. Bone Joint J 2017;99-B:1167–75.

[50] Lunn TH, Kristensen BB, Andersen LO, HustedH, Otte KS, Gaarn-LarsenL, Kehlet H. Effect of high-dose preoperative methylprednisolone on painand recovery after total knee arthroplasty: a randomized, placebo-controlled trial. Br J Anaesth 2011;106:230–8.

[51] McIsaac DI, Cole ET, McCartney CJ. Impact of including regionalanaesthesia in enhanced recovery protocols: a scoping review. Br JAnaesth 2015;115(suppl 2):ii46–ii56.

[52] Michard F, Gan TJ, Kehlet H. Digital innovations and emergingtechnologies for enhanced recovery programmes. Br J Anaesth 2017;119:31–9.

[53] Needham MJ, Webb CE, Bryden DC. Postoperative cognitivedysfunction and dementia: what we need to know and do. Br JAnaesth 2017;119(suppl 1):i115–25.

[54] Petersen PB, Jorgensen CC, Kehlet H. Delirium after fast-track hip andknee arthroplasty—a cohort study of 6331 elderly patients. ActaAnaesthesiol Scand 2017;61:767–72.

[55] Popping DM, Elia N, Van Aken HK, Marret E, Schug SA, Kranke P, WenkM, Tramer MR. Impact of epidural analgesia on mortality and morbidityafter surgery: systematic review and meta-analysis of randomizedcontrolled trials. Ann Surg 2014;259:1056–67.

[56] Sanderson BJ, Doane MA. Transversus abdominis plane catheters foranalgesia following abdominal surgery in adults. Reg Anesth Pain Med2018;43:5–13.

[57] SchotanusMGM, Schoenmakers DAL, Sollie R, Kort NP. Patient-specificinstruments for total knee arthroplasty can accurately predict thecomponent size as used preoperative. Knee Surg Sports TraumatolArthrosc 2017;25:3844–8.

[58] Schug S, Palmer GM, Scott DA, Halliwell R, Trinca J. Acute painmanagement: scientific evidence. 4th ed. Melbourne: Australian and NewZealand College of Anaesthetists and Faculty of Pain Medicine, 2015.

[59] Sessler DI, Meyhoff CS, Zimmerman NM, Mao G, Leslie K, Vasquez SM,Balaji P, Alvarez-Garcia J, Cavalcanti AB, Parlow JL, Rahate PV,Seeberger MD, Gossetti B, Walker SA, Premchand RK, Dahl RM,

Duceppe E, Rodseth R, Botto F, Devereaux PJ. Period-dependentassociations between hypotension during and for four days afternoncardiac surgery and a composite of myocardial infarction anddeath: a substudy of the POISE-2 trial. Anesthesiology 2018;128:317–27.

[60] Sharrock NE. Challenging conventions to make a difference in patientcare: the 2017Gaston Labat Award Lecture. Reg Anesth PainMed 2018;43:50–6.

[61] Sivertsen B, Lallukka T, Petrie KJ, Steingrimsdottir OA, Stubhaug A,Nielsen CS. Sleep and pain sensitivity in adults. PAIN 2015;156:1433–9.

[62] Slooter AJC. Delirium, what’s in a name? Br J Anaesth 2017;119:283–5.[63] Srikandarajah S, Gilron I. Systematic review of movement-evoked pain

versus pain at rest in postsurgical clinical trials and meta-analyses:a fundamental distinction requiring standardized measurement. PAIN2011;152:1734–9.

[64] Steinthorsdottir KJ, Kehlet H, Aasvang EK. Surgical stress response andthe potential role of preoperative glucocorticoids on post-anesthesia careunit recovery. Minerva Anestesiol 2017;83:1324–31.

[65] Steyaert A, Lavand’homme P. Prevention and treatment of chronicpostsurgical pain: a narrative review. Drugs 2018;78:339–54.

[66] Sun EC, Darnall BD, Baker LC,Mackey S. Incidence of and risk factors forchronic opioid use among opioid-naive patients in the postoperativeperiod. JAMA Intern Med 2016;176:1286–93.

[67] Tan M, Law LS, Gan TJ. Optimizing pain management to facilitateenhanced recovery after surgery pathways. Can J Anaesth 2015;62:203–18.

[68] Terrando N, Eriksson LI, Ryu JK, Yang T, Monaco C, Feldmann M,Jonsson Fagerlund M, Charo IF, Akassoglou K, Maze M. Resolvingpostoperative neuroinflammation and cognitive decline. Ann Neurol2011;70:986–95.

[69] Thiele RH, Rea KM, Turrentine FE, Friel CM, Hassinger TE, McMurry TL,Goudreau BJ, Umapathi BA, Kron IL, Sawyer RG, Hedrick TL.Standardization of care: impact of an enhanced recovery protocol onlength of stay, complications, and direct costs after colorectal surgery.J Am Coll Surg 2015;220:430–43.

[70] Toner AJ, Ganeshanathan V, Chan MT, Ho KM, Corcoran TB. Safety ofperioperative glucocorticoids in elective noncardiac surgery: a systematicreview and meta-analysis. Anesthesiology 2017;126:234–48.

[71] van Bree SH, Nemethova A, Cailotto C, Gomez-Pinilla PJ, Matteoli G,BoeckxstaensGE.New therapeutic strategies for postoperative ileus. NatRev Gastroenterol Hepatol 2012;9:675–83.

[72] Watt DG, McSorley ST, Horgan PG, McMillan DC. Enhanced recoveryafter surgery: which components, if any, impact on the systemicinflammatory response following colorectal Surgery?: a systematicreview. Medicine (Baltimore) 2015;94:e1286.

[73] Weinrib AZ, Azam MA, Birnie KA, Burns LC, Clarke H, Katz J. Thepsychology of chronic post-surgical pain: new frontiers in risk factoridentification, prevention and management. Br J pain 2017;11:169–77.

[74] Werner MU, Søholm L, Rotbøll-Nielsen P, Kehlet H. Does an acute painservice improve postoperative outcome? Anesth Analg 2002;95:1361–72.

[75] White PF, Kehlet H. Improving postoperative pain management: what arethe unresolved issues? Anesthesiology 2010;112:220–5.

[76] Wick EC, Grant MC, Wu CL. Postoperative multimodal analgesia painmanagement with nonopioid analgesics and techniques: a review. JAMASurg 2017;152:691–7.

[77] Yekkirala AS, Roberson DP, Bean BP, Woolf CJ. Breaking barriers tonovel analgesic drug development. Nat Rev Drug Discov 2017;16:545–64.

[78] Zhang H, Li F, Li WW, Stary C, Clark JD, Xu S, Xiong X. Theinflammasome as a target for pain therapy. Br J Anaesth 2016;117:693–707.

S16 H. Kehlet·159 (2018) S11–S16 PAIN®


Research Paper

What makes surgical nerve injury painful? A 4-yearto 9-year follow-up of patients withintercostobrachial nerve resection in womentreated for breast cancerLaura Mustonena,b, Tommi Ahoa, Hanna Harnoa,b, Reetta Sipilaa, Tuomo Meretojac, Eija Kalsoa,*

AbstractNerve injury during breast cancer surgery can cause neuropathic pain (NP). It is not known why some, but not all, patients developchronic postsurgical neuropathic pain (CPSNP) after the same nerve injury. In this study, we examined 251 breast cancer survivorswith surgeon-verified intercostobrachial nerve resection to identify factors that associate with CPSNP. The patients were recruitedfrom a previous study of 1000women treated for breast cancer in 2006 to 2010. This enabled us to analyze preoperative factors thatassociate with future CPSNP. The patients were re-examined in 2014 to 2016 to diagnose CPSNP using the revised NP diagnosticcriteria. Preoperative assessments were pain in the area to be operated on, any chronic pain condition, depressive symptoms,anxiety, sleep, and experimental cold pain sensitivity using the cold pressor test (CPT). Follow-up assessments were CPT,psychological factors, sleep, any chronic pain, and basic laboratory tests. One hundred thirty-seven (55%) patients withintercostobrachial nerve resection fulfilled CPSNP diagnostic criteria after 4 to 9 years. Of them, 30 patients (22%) had moderate tosevere pain in self-reports and 86 (63%) presentedmoderate to severe evoked pain at examination. Preoperative pain in the surgicalarea, other chronic pains, and breast-conserving surgery were associated with future CPSNP. Other chronic pains, increasedpsychological burden, and insomnia, both before surgery and at the follow-up, were associated with CPSNP. Preoperative CPT didnot associate with future CPSNP. Patients with established CPSNP showed increased pain sensitivity in CPT and higher levels ofinflammatory markers, suggesting that central sensitization and inflammation may associate with the maintenance of CPSNP.

Keywords: Neuropathic pain, Intercostobrachial nerve, Breast cancer, Cold pressor tolerance, Central sensitization, Anxiety,Depression

1. Introduction

The reported pooled prevalence estimates of chronic post-surgical neuropathic pain (CPSNP) after breast cancer surgeryare 14% to 31% in all patients and 33% to 58% in those reporting

persistent postsurgical pain (PPSP).16,17 The intercostobrachialnerve (ICBN) is frequently resected in breast cancer surgery,leading to sensory abnormalities.44 However, the evidence isconflicting about how the ICBN is handled and subsequentpain.2,3,14,32,44

Neuropathic pain (NP) is defined as pain caused by a lesion ora disease of the somatosensory nervous system.12 In the revisedcriteria for the diagnosis of NP, surgeon-verified nerve injury is oneof the confirmatory tests for definite NP, in addition to sensoryabnormalities and pain in the corresponding area.12

Multiple mechanisms encompassing both the peripheral andcentral nervous systems have been identified in the pathophys-iology of NP,9 but these do not explain why some patients, but notall, develop NP despite the same etiology.9,20,25 This also appliesto ICBN resection and CPSNP.20

Previous prospective studies analyzing the association ofpreoperative patient-related factors and CPSNP have shown thatpain, opioid use, poorer neuropsychological function, female sex,and anxiety associate with future CPSNP.4,11,25,26,32

In other types of NP, predictive factors are more difficult to assessbecause the onset of nerve injury is not as well-defined as in CPSNP.However, other NP conditions provide cross-sectional information ofneuropathic patients with or without pain. In diabetes, the role ofinflammation is of significant interest in both polyneuropathy andNP.34 Interestingly, different inflammatory profiles have been reportedin painful vs nonpainful diabetic polyneuropathy.10,45 To ourknowledge, there are no studies on the role of inflammation in painful



L. Mustonen and T. Aho contributed equally to this work.

a Division of Pain Medicine, Department of Anesthesiology, Intensive Care and Pain

Medicine, University of Helsinki, Helsinki University Hospital, Helsinki, Finland,b Clinical Neurosciences, Neurology, University of Helsinki and Department of

Neurology, Helsinki University Hospital, Finland, c Breast Surgery Unit, Compre-

hensive Cancer Center, University of Helsinki, Helsinki University Hospital, Helsinki,

Finland

*Corresponding author. Address: Pain Clinic, Helsinki University Hospital, P.O. Box

140, Helsinki, 00029 HUS, Finland. Tel.: 1 358 9 4717885. E-mail address:

[email protected] (E. Kalso).

Supplemental digital content is available for this article. Direct URL citations appear

in the printed text and are provided in the HTML and PDF versions of this article on

the journal’s Web site (www.painjournalonline.com).

PAIN 160 (2019) 246–256

Copyright© 2018 The Author(s). Published byWolters Kluwer Health, Inc. on behalf

of the International Association for the Study of Pain. This is an open-access article

distributed under the terms of the Creative Commons Attribution-Non Commercial-

No Derivatives License 4.0 (CCBY-NC-ND), where it is permissible to download and

share the work provided it is properly cited. The work cannot be changed in any way

or used commercially without permission from the journal.


246 L. Mustonen et al.·160 (2019) 246–256 PAIN®

10

vs nonpainful traumatic nerve injury. However, higher cytokine levelsin the cerebrospinal fluid have been measured in patients withtraumatic NP compared with healthy controls.7

Lipid profiles differ in HIV patients with or without neuropathy.33

However, their role has not been clarified in painful vs nonpainfulneuropathy. Elevated plasma glucose levels have been reportedto associate with daily chronic pain,31 but not with painfulcompared with nonpainful diabetic polyneuropathy.34

No study has assessed the role of general pain sensitivity,measured with the cold pressor test (CPT) before surgery, to predictdevelopment ofCPSNP.However, patientswith establishedCPSNPhave been reported to be more pain sensitive in the CPT comparedwith those who do not have NP after similar nerve transection.42

To study factors that may differentiate patients with or withoutCPSNP after ICBN resection, we used the data from a previousstudy of 1000 patients treated for breast cancer21 and invitedthose having a surgeon-defined ICBN resection for a newexamination 4 to 9 years after surgery to confirm whether thepatients had the diagnosis of definite CPSNP or not. To studygroup differences, we repeated the preoperative assessments ofpain sensitivity and tolerance using CPT, presence of any chronicpain, distress, and self-reported sleep disturbances. In addition,we measured inflammatory biomarkers, glucose, and lipid levelsin plasma at the follow-up.

2. Patients and methods

2.1. Original cohort: preoperative and treatment-related variables

The patients in the current study were recruited from a previousoriginal cohort of 1000 women operated on for unilateral breastcancer during 2006 to 2010 at the Helsinki University Hospital.The patient selection and study procedures of this original cohorthave previously been described in detail.21 In brief, womenscheduled for surgery of nonmetastasized unilateral breastcancer without neoadjuvant treatment or immediate breastreconstruction were invited. The preoperative and follow-up visitassessments are listed in Table 1.

Experimental heat pain sensitivity was analyzed with a 16 3 16-mm thermode (TSA-II NeuroSensory Analyzer; Medoc Ltd, RamatYishai, Israel). In the heat pain test, the patients reported their painintensity with a Numerical Rating Scale (NRS 0-10) after a 5-secondstimulation with 43˚C and 48˚C. Cold pain sensitivity and tolerancewere assessed using the CPT. Patients immersed their contralateral(to the side to be operated on) hand into circulating cold water (12-4˚C) bath (JULABO USA Inc, Allentown, PA) up to the wrist for themaximum time tolerated but no longer than 90 seconds (referred toas cold pain tolerance). During the CPT, patients reported painintensity every 15 seconds and at the end of CPT on anNRS 0 to 10(referred to as cold pain sensitivity).

The surgical procedure was either mastectomy or breast-conserving surgery (BCS) with sentinel lymph node biopsy(SLNB) or axillary lymph node dissection (ALND). Clinicallynode-negative patients with radiologically unifocal tumors notexceeding 30mm in size underwent SLNB. Generally, all patientswith tumor-positive sentinel nodes underwent completion ALND.Patients with large (.30 mm) or multifocal tumors in breastimaging, as well as clinically node-positive patients, underwentdirect ALND of Berg levels I and II. Level III was also dissected ifclinically suspicious nodes were present. Surgery was performedor directly supervised by experienced breast surgeons. Theoperating surgeon documented whether ICBN was preserved,totally or partially resected, or not visualized during surgery.

Anesthesia was standardized with remifentanil, propofol, androcuronium. Postoperatively, the patients were given acetamino-phen 1 g every 8 hours, and they were titrated pain-free withintravenous oxycodone, first by the research nurse at thepostanesthesia care unit (PACU) and then with patient-controlledanalgesia on the ward.21 Data on pain intensity before oxycodonetitration and consumption of oxycodone during the 2-hour period atthe PACU were collected. Data concerning oncological treatments,reoperation, and breast reconstructions were collected.

2.2. Current cohort: patients with injury of theintercostobrachial nerve

To address the question of why some patients but not all developCPSNP after similar nerve injury, we, in this study, includedpatients with a surgeon-verified, total, or partial ICBN resection. Inthe original cohort of 1000 patients, 440 patients underwenteither total or partial ICBN resection. Forty-one patients (9.3%)had died, and 38 patients (8.6%) had reached 75 years of age atthe time of recruitment and were therefore not invited. Eightpatients (1.8%) were excluded for other reasons (eg, no breastcancer at final histology). Thus, 353 patients (353/440, 80.2%)were eligible for the follow-up visit. The research nurse contactedthese patients through telephone to ask about their willingness toparticipate in the study. Of the 353 patients, 37 (10.5%) could notbe reached and 65 patients (18.4%) declined. Thus, 251 patients(251/440, 57%) participated. The participants (N 5 251) andeligible nonparticipants (N 5 102) did not differ in terms ofpreoperative and treatment-related variables (see supplementaryTable 1, supplemental digital content, available at http://links.lww.com/PAIN/A661, which demonstrates comparison of par-ticipants and eligible nonparticipants). Figure 1 illustrates thecomplete patient flow.

The study was approved by the Coordinating Ethics Board ofthe Helsinki and Uusimaa Hospital District and registered inClinicalTrials.gov (NCT02487524). All patients gave informedwritten consent.

2.3. Follow-up visit—clinical examination and gradingcriteria for definite neuropathic pain due to injury ofintercostobrachial nerve

The 251 patients with surgeon-verified injury to the ICBNunderwent a thorough sensory examination of the upper bodyat the follow-up. Sensory examination consisted of testing tactilesensation by a cotton tuft, static allodynia by finger compression,dynamic allodynia by a painter’s brush, pinprick sensation bya sharp wooden cocktail stick, and cold and warm sensation bya metal roller. The affected side was compared with thecontralateral side and the surrounding skin. The examinationconsisted of the following sensory modalities: hypoesthesia(diminished sensitivity), hyperesthesia (heightened sensitivity),dysesthesia (unpleasant sensation), and allodynia (pain evokedby normally painless stimuli). If evoked pain was observed duringexamination, the patients were asked to rate the pain intensity(NRS 0-10). The examining neurologist was blinded to the ICBNstatus of the patients during the sensory evaluation. To identifypain in the surgical area, we used the pain intensity rating 1 orhigher (NRS 0-10) in at least one of the following 2measures: BriefPain Inventory (BPI) for the worst pain during past week or evokedpain in the clinical sensory examination. The patient located thepain to a body map drawing and the examining neurologistlocated the sensory findings to a similar body map drawing of theupper body.

January 2019·Volume 160·Number 1 www.painjournalonline.com 247

11

We used the revised stepwise grading criteria for NP12 toidentify patients with CPSNP. The steps include A) a history ofrelevant neurological lesion and neuroanatomically plausible paindistribution, B) that the pain associates with sensory signs in thesame neuroanatomical distribution, and C) that a diagnostic testconfirms the lesion in the somatosensory nervous system.According to the recent revision of the grading system,a surgeon’s report of nerve resection is equivalent to a diagnostictest in the case of postsurgical neuropathic pain.12

All patients in this study had a history of breast cancer surgeryand ICBN resection. The region of interest was the innervationarea of the ICBN: axilla, medial upper arm, flank of chest, andlateral breast.2 Patients were classified as “unlikely CPSNP” if nopain was present or the localization of pain was not neuro-anatomically plausible for the surgical area (criterion A not met).Patientswith pain in the surgical areawithout sensory abnormalityin the corresponding region were classified as possible CPSNP(criterion B not met). The patients with either self-reported orevoked pain in the surgical area with at least one sensoryabnormality in the corresponding region were classified asdefinite CPSNP if these findings occurred in the area of ICBNinnervation2 (criteria A, B, and C fulfilled). If the patients had painand sensory abnormalities in the surgical area, but outside theICBN innervation (eg, medial breast), they were classified asprobable CPSNP (criterion C not met).

We excluded the possible CPSNP (N 5 32) and probableCPSNP (N5 15) groups from the analyses to avoid possible biascaused by uncertain NP diagnosis (Fig. 1). In the final analysis wehad only patients with definite CPSNP (CPSNP group) or unlikelyCPSNP (non-CPSNP group). Two patients from each group wereexcluded because of ongoing cancer treatments. We analyzeddifferences in these 2 groups to understand what makes a similarnerve injury painful or not. Thus, we had a nested case–control

design including patients having had a previous ICBN resectionwith or without current CPSNP.

2.4. Demographic factors, questionnaires, and coldpressor test

Demographic factors preoperatively and at the follow-up visit arepresented in Table 1. Preoperatively, the question of insomnia wasincluded after the study had started, and therefore, data are missingfrom 33 patients. For pain intensity, we considered NRS $4/10 asmoderate to severe pain. At the follow-up visit, BPI for other painswas included after the start of the study, and therefore, data aremissing from 26/251 (10%) patients. The patients reported otherchronic pains with an open question and a pain drawing. Based onthese, other pains at the follow-up were categorized as follows:headaches, pain in the joints, back, neck, or other area.

For the psychological and sleep questionnaire outcomes, weused cutoff values for clinically relevant outcomes to report theproportion of patients with clinically significant symptoms. Thesewere used as follows: $10 for at least mild and $19 for at leastmoderate depressive symptoms in the Beck’s Depression InventoryII (BDI II)5,29; 8 to 10 for borderline and $11 for clinically significantanxiety/depression in HADS6,29;$40 for clinically significant anxietyin STAI39; $8 for mild to severe insomnia in ISI30; and $30 forclinically significant catastrophizing in PCS.40 Cronbach’s alphas forthese are reported in the supplementary Table 2 (available at http://links.lww.com/PAIN/A661).

The fasting blood samples for glucose level, lipids, vitamin D, andinflammatory markers (high-sensitivity CRP [hs-CRP]; orosomucoid[ORM]) were drawn at the follow-up and analyzed according to thestandard laboratory protocol (HUSLAB, Helsinki, Finland). The CPTwas performed similarly preoperatively and at the follow-up, by thesame research nurse, protocol, and equipment.

Table 1

Assessments at preoperative and follow-up visits.

Preoperative Follow-up 4–9 y after operation

Demographics Age, BMI Age, BMI, marital status, education, smoking,

alcohol consumption, and use of medication (self-

report)

Pain and sensory changes in the operative area

Neuropathic pain diagnosis N/A According to revised diagnostic grading criteria

(Finnerup et al., 2016)

Clinical sensory examination of the upper

body

N/A Static mechanical allodynia, light touch, dynamic

touch, pinprick, and thermal sensation

Pain intensity in the operative area 0-10 NRS, worst pain past week 0-10 NRS, worst pain past week

Evoked pain in the operative area N/A 0-10 NRS, at clinical examination

Other pain

Other chronic pain conditions Yes/No Pain in the following locations (yes/no): back, joints,

neck, head, and other

Intensity of other pain 0-10 NRS, worst pain past week 0-10 NRS, worst pain past week

Mood and sleep

Depressive symptoms BDI II BDI II, HADS-D

Anxiety STAI state and trait HADS-A

Insomnia Not at all/at least once a wk/every night ISI

Pain catastrophizing N/A PCS

Experimental pain

Heat pain Heat pain at 43˚C and 48˚C N/A

Cold pain CPT CPT

Blood samples N/A GHb-A1c, lipids, 25-hydroxyvitamin-D, hs-CRP,

and ORM

BDI II, Beck’s Depression Inventory II; BMI, body mass index; CPT, cold pressor test; GHb-A1c, glycated hemoglobin A1c; HADS-A, Hospital Anxiety and Depression Scale–Anxiety; HADS-D, Hospital Anxiety and Depression

Scale–Depression; hs-CRP, high-sensitivity C-reactive protein; ISI, insomnia severity index; N/A, not assessed; NRS, numerical rating scale; ORM, orosomucoid; PCS, pain catastrophizing scale; STAI, State-Trait Anxiety Inventory.


12

2.5. Statistical analysis

We used Student t test for normally distributed continuousvariables in pairwise comparisons and repeated measures.Mann–Whitney U test and x2-tests were used for non-normallydistributed and categorical variables, respectively. Spearman’srho (rs) was used for correlations and Cronbach’s alpha forreliability assessments.

TopredictCPSNPwithpreoperative andbreast cancer treatment–related clinical variables, the variables reaching P , 0.05 in the

bivariate analysis were entered as predictors in a logistic regression

analysis using the forward stepwisemethod. In addition,we tested the

model with the backward stepwise method to control for multi-

collinearity. We combined the type of breast surgery and radiotherapy

as a single categorical variable for the regression analysis because

nearly all patientswithBCS receive radiotherapy.Continuousvariables

were not categorized for this analysis. To detect the effect of possible

multicollinearity, we tested entering the susceptible variables (BDI II,

STAI state, and trait) one by one and in different combinations.

Figure 1. Patient selection and clinical grading for chronic postsurgical neuropathic pain. Surgical area refers to the breast, axilla, upper side of the chest, andmedial arm in the operated side. The area of ICBN resection refers to the lateral side of the breast, axilla, upper side of the chest, and medial arm in the operatedside. CPSNP, chronic postsurgical neuropathic pain; ICBN, intercostobrachial nerve.


13

To assess the role of preoperative thermal pain sensitivity inpredicting future CPSNP, we conducted a logistic regressionanalysis. Preoperatively measured variables for age, body massindex (BMI), chronic pain (yes/no), depression (yes/no, cutoff$19in BDI II), and anxiety (yes/no, cutoff $40 in STAI state) wereentered as covariates to control for possible confounding.

To analyze the cold pain sensitivity and tolerance in establishedCPSNP 4 to 9 years after breast cancer surgery, we conducteda Cox-regression analysis. We used (1) time to withdrawal and (2)time to NRS 10 during CPT as the time to event. Data were right-censored if the participant endured the CPT the maximum of 90seconds or if NRS values did not reach 10 during CPT. Variablesmeasured at the research visit 4 to 9 years after surgery, includingage, BMI, other pain of at least moderate intensity (yes/no),depression (yes/no, cutoff $19 in BDI II), and anxiety (yes/no,cutoff $11 in HADS-A), were added as covariates to control forpossible confounding. We performed an interaction analysis forinflammatory markers (hs-CRP and ORM) and CPSNP on CPTparameters using cross-product terms in Cox models. Inflam-matory markers were inserted into the model as continuousvariables.

Two-tailed P # 0.05 was considered statistically significant.Statistical analyses were performed using SPSS 22.0 version forWindows (SPSS Inc, Chicago, IL).

3. Results

3.1. Clinical grading of chronic postsurgical neuropathic painand patients with definite chronic postsurgicalneuropathic pain

Figure 1 shows the distribution of patients to unlikely, possible,probable, and definite CPSNPgroups. In definite CPSNPpatients(135), evoked pain at examination presented in 114 (84%)patients and wasmostly static allodynia (99%). Pain intensity wasmoderate to severe in 30 of 135 (22%) BPI reports.

3.2. Factors associating with established chronicpostsurgical neuropathic pain: patients with and withoutchronic postsurgical neuropathic pain 4 to 9 yearsafter surgery

The intergroup comparisons of demographic factors, other pains,psychological factors and sleep questionnaires, and laboratoryparameters are shown in Table 2. The groups were homogenousin terms of age, time from surgery, and other demographicfactors. Body mass index was significantly higher in CPSNPpatients compared with non-CPSNP patients (Table 2).

The CPSNP patients had significantly more other painconditions than non-CPSNP patients (Table 2), particularly joint,back, and neck pains. The CPSNP patients hadmore other painsand reported higher intensities for other chronic pains: moderateto severe pain was reported by 52/118 (44%, 17 missing values)of CPSNP patients compared with 12/60 (20%, 5missing values)by non-CPSNP patients (P 5 0.003).

Only a few patients in either group reported current use of NPmedications (ie, tricyclics, serotonin–norepinephrine reuptakeinhibitors, and gabapentinoids), whereas the use of other painmedications such as nonsteroidal anti-inflammatory drugs(NSAIDs) was more prevalent among CPSNP patients comparedwith non-CPSNP patients (Table 2).

The CPSNP group reported significantly more symptoms in allquestionnaires related to psychological factors and sleep(Table 2). A total of 60/131 (45%, 4 missing values) of theCPSNP patients reported at least mild depressive symptoms in

BDI II compared with 14/65 (22%) of non-CPSNP patients (P 50.001). A total of 33/134 (25%, one missing value) of CPSNPpatients compared with 7/65 (11%) of non-CPSNP patientsshowed borderline or clinically significant levels of anxiety (P 50.025). A total of 67/132 (51%, 3missing values) CPSNP patientsand 20/65 (31%) non-CPSNP patients suffered from at least mildinsomnia (P 5 0.008).

The CPSNP patients had significantly higher levels of hs-CRPand ORM (Table 2). There were no differences between thegroups in other biochemical parameters (Table 2). The levels ofhs-CRP correlated positively with BMI in both groups: rs 0.374and P, 0.001 in CPSNP group; rs 0.344 and P5 0.005 in non-CPSNP group. Orosomucoid and BMI showed a weaker positivecorrelation: rs 0.175 and P5 0.045 in CPSNP group; rs 0.239 andP 5 0.055 in non-CPSNP group.

3.3. Preoperative and treatment-related factors associatingwith future chronic postsurgical neuropathic pain

Table 3 depicts the intergroup comparison of factors related tocancer and its treatment. In both groups, most patients (CPSNP:124/135, 92%; non-CPSNP: 57/65, 88%) had undergone axillaryclearance. Partial resections of ICBN were more frequent thantotal resections. The CPSNP patients had undergone BCS andreceived radiotherapy more frequently than the non-CPSNPpatients. There was no difference between the groups regardingadministration of chemotherapy or hormonal therapy (Table 3).

Preoperatively, future CPSNP patients showed higher BMI,reported more pain in the surgical area and elsewhere, had moredepressive symptoms and anxiety, and reported insomnia morefrequently than future non-CPSNP patients (Table 4). In addition,CPSNP patients presented with higher immediate postoperativepain intensity ratings and higher oxycodone consumption at thePACU. In multivariate analysis, preoperative pain in the surgicalarea, the presence of chronic pain conditions, and BCS as thetype of breast surgery were associated with increased risk ofCPSNP (Table 4). The effect of the type of breast surgeryremained significant even after controlling for radiotherapy. Theresults remained unaltered despite the method of stepwiselogistic regression used. Similarly, entering variables susceptiblefor multicollinearity (BDI II, STAI state and trait) one by one or indifferent combinations did not affect the outcome. No significantdifferences in cancer type, number of metastatic lymph nodes,nerve resection type, reoperations, or late reconstructions weredetected (Table 3).

3.4. General pain sensitivity and chronic postsurgicalneuropathic pain: preoperative and postoperative coldpressor test and preoperative heat pain assessment

Compared with non-CPSNP patients, CPSNP patients pre-sented with a significantly lower cold pain tolerance and highercold pain sensitivity postoperatively, but not preoperatively(Table 5). Both patient groups showed a significant increase ofwithdrawal times (CPSNP: mean difference 14.2 seconds, P ,0.001; non-CPSNP: mean difference 15.6 seconds, P, 0.001),although no significant mean differences were detected betweengroups (P 5 0.716). Preoperative heat pain (48˚C) intensitypredicted CPSNP, although the association remained non-significant after multivariate adjustment (Table 6).

A Cox-regression analysis of the postoperative CPT (Table 7)showed that patients with CPSNP aborted the CPT significantlyearlier than non-CPSNP patients (Fig. 2). The associationremained unaltered after multivariate adjustment. Patients in the


14

Table 2

Patient demographics and clinical features 4–9 years after breast cancer surgery.

CPSNP (n 5 135) Non-CPSNP (N 5 65) P

Demographics

Age, mean (SD), y 60.4 (8.01) 61.4 (8.89) 0.45*

Time from index surgery, mean (SD), mo 77.9 (13.31) 77.7 (13.35) 0.93*

BMI, mean (SD), kg/m2 26.2 (3.90) 24.1 (3.95) <0.001*Marital status, no. (%) 0.96†

Married or cohabiting 76 (56.3) 38 (58.5)

Single 22 (16.3) 10 (15.4)

Divorced or widowed 37 (27.4) 17 (26.2)

Educational level, number (%) 0.78†

Low 18 (13.3) 9 (13.8)

Moderate 33 (24.4) 13 (20.0)

High 84 (62.2) 43 (66.2)

Smoking, number (%) 0.82†

Never smoked 66 (48.9) 30 (46.2)

Smoker 24 (17.8) 14 (21.5)

Ex-smoker 45 (33.3) 21 (32.3)

Alcohol consumption, number (%)§ 0.56†

Abstinent 20 (14.8) 13 (20.0)

,6 doses per week 84 (62.2) 36 (55.4)

$6 doses per week 30 (22.2) 16 (24.6)

Other pain conditions

Worst other pain past week, median (IQR), 0-

10 NRS‖

3 (1-5) 0 (0-2) <0.001‡

Joint pain, number (%) 82 (60.7) 21 (32.3) <0.001†Back pain, number (%) 42 (31.1) 11 (16.9) 0.033†Neck pain, number (%) 47 (34.8) 5 (7.7) <0.001†Headache, number (%) 15 (11.1) 2 (3.1) 0.056†

Other pain, number (%) 67 (49.6) 15 (23.1) <0.001†Overlapping pain conditions, number (%) <0.001†No other pain 15 (11.1) 33 (50.8)

1 or 2 other pain conditions 80 (59.3) 27 (41.5)

3 or more other pain conditions 40 (29.6) 5 (7.7)

Current use of pain medication

Tricyclic antidepressant, gabapentinoids, or

SNRI, number (%)

6 (4.4) 3 (4.6) 0.96†

NSAID, acetaminophen, or mild opioid,

number (%)

25 (18.5) 3 (4.6) 0.008†

Mood and sleep

BDI II, median (IQR) 9 (5-14) 5 (2-9) <0.001‡HADS-A, median (IQR) 5 (3-7) 3 (1-5) <0.001‡HADS-D, median (IQR) 3 (1-6) 1 (0-3) 0.001‡PCS, median (IQR) 6 (1-13) 1 (0-8) <0.001‡ISI, median (IQR) 8 (4-12) 4 (2-9) 0.001‡

Laboratory parameters{GHb-A1C, mean (SD), % (4.0-6.0) 5.6 (0.4) 5.5 (0.3) 0.36*

Cholesterol (total), mean (SD), mmol/l (,5.0) 5.5 (1.1) 5.6 (0.9) 0.47*

LDL, mean (SD), mmol/L (,3.0) 3.3 (0.9) 3.4 (0.8) 0.69*

HDL, mean (SD), mmol/L (.1.20) 2.01 (0.55) 2.03 (0.53) 0.75*

Triglyserides, mean (SD), mmol/L (,1.70) 1.13 (0.46) 1.20 (0.57) 0.37*

hs-CRP, median (IQR), mg/L (0.05-3.00) 0.99 (0.37-2.42) 0.53 (0.27-1.14) 0.005‡ORM, mean (SD), mg/L (500-1200) 852 (237) 781 (178) 0.019*25-Hydroxyvitamin-D, mean (SD), nmol/L

(.50)

76 (25) 82 (22) 0.094*

* Student t test.

† Chi-square test.

‡ Mann–Whitney U test.

§ One answer is missing from the NP group. One dose corresponds to 12 g of pure alcohol.

‖ Seventeen values are missing from the NP group and 5 from the non-NP group.

{ The normative values are shown in parenthesis.

P values, 0.05 are shown in bold. BDI II, Beck’s Depression Inventory II; BMI, body mass index; CPSNP, chronic postsurgical neuropathic pain; GHb-A1C, glycated hemoglobin A1C; HADS-A, Hospital Anxiety and Depression

Scale–Anxiety; HADS-D, Hospital Anxiety and Depression Scale–Depression; HDL, high-density lipoprotein; hs-CRP, high-sensitivity C-reactive protein; IQR, interquartile range; ISI, insomnia severity index; LDL, low-density

lipoprotein; NSAID, nonsteroidal anti-inflammatory drug; ORM, orosomucoid (alpha-1-acid glycoprotein); PCS, pain catastrophizing scale; SNRI, serotonin–norepinephrine reuptake inhibitors.


15

CPSNP group were also more likely to score NRS 10 during CPTcompared with non-CPSNP patients (Fig. 2).

There was a significant association for cold pain tolerance andhs-CRP (HR5 1.06, CI: 1.02-1.10,P5 0.005) but not for ORM (P5 0.201). There was no significant interaction between CPSNPand hs-CRP (P 5 0.831) or ORM (P 5 0.201) on cold pressortolerance.

4. Discussion

4.1. Main findings

Of the 440/1000 patients with surgeon-verified ICBN resection,251/440 (57%) were examined 4 to 9 years after index breastcancer surgery. Fifty-five percent of these patients (137/251)fulfilled the diagnostic criteria for definite CPSNP. Twenty-twopercent (30/137) of them reportedmoderate to severe pain in self-report (BPI), and 63% (86/137) had moderate to severe evokedpain at clinical examination. Compared with the non-CPSNPpatients, CPSNP patients had more depressive symptoms,anxiety, pain catastrophizing, impaired sleep, and other pains.They also had higher levels of inflammatorymarkers and increasedsensitivity in the CPT, suggesting a possible role of centralsensitization. Preoperatively, CPSNP patients showed more

psychological distress, insomnia, and more pain both in thesurgical area and in other locations than the non-CPSNP patients.

4.2. Prevalence of chronic postsurgical neuropathic painafter breast cancer surgery

TheprevalenceofNPafter breast cancer surgery ranges from33%to 58% in those patients who have PPSP,17 depending on whenand how the diagnosis of NP is made. The diagnosis is usuallybased on validated questionnaires in cross-sectional studies. Sofar, few prospective assessments with clinical examination havebeen conducted with follow-ups to 1 year.17,32 Our study is thefirst, to our knowledge, to use the revised NP grading criteria,combining a thorough clinical sensory examination with surgeons’report of nerve injury to reach the diagnostic level of “definite”CPSNP. Extensive sensory examination of multiple modalities,especially reports of evoked pain, allowed identification of CPSNPpatients who might have been unnoticed in previous studies. Inaddition, our study had a 4-year to 9-year follow-up to providefurther evidence of CPSNP being a long-lasting consequence ofnerve injury. This is in line with a previous study that showed painand sensory disturbances to be a marked problem after breastcancer surgery during a follow-up of 5 to 7 years. In that study,PPSP was reported by over a third of the patients.27

Table 3

Characteristics of breast cancer and its treatment.

CPSNP (n 5 135) Non-CPSNP (N 5 65) P

Histology, number (%) 0.27*

Ductal 94 (69.6) 40 (61.5)

Lobular 27 (20.0) 13 (20.0)

Other 14 (10.4) 12 (18.5)

Gradus, number (%) 0.40*

I 27 (20.0) 14 (21.5)

II 61 (45.2) 23 (35.4)

III 47 (34.8) 28 (43.1)

Size of the tumor, median (IQR), mm 18 (13-25) 18 (14-25) 0.96†

No. of lymph nodes evacuated, mean (SD) 20.8 (10.35) 20.8 (9.42) 0.98‡

No. of metastatic lymph nodes, median (IQR) 1 (1-4) 1 (1-2) 0.066†

Breast surgery type, number (%) <0.001*Mastectomy 69 (51.1) 50 (76.9)

BCS 66 (48.9) 15 (23.1)

Axillary surgery type, number (%) 0.35*

SLNB 11 (8.1) 8 (12.3)

ALND 124 (91.9) 57 (87.7)

Type of ICBN resection, number (%) 0.093*

Partial 88 (65.2) 50 (76.9)

Total 47 (34.8) 15 (23.1)

Reoperation, number (%) 17 (12.6) 10 (15.4) 0.59*

Late reconstruction, number (%) 45 (33.3) 25 (38.5) 0.48*

Chemotherapy, number (%) 120 (88.9) 56 (86.2) 0.58*

Radiotherapy, number (%) 102 (75.6) 38 (58.5) 0.013*

Endocrine therapy, number (%) 116 (85.9) 54 (83.1) 0.60*

Tamoxifen, number (%) 92 (68.1) 42 (64.6) 0.62*

Aromatase inhibitor, number (%) 90 (66.7) 47 (72.3) 0.42*

* Chi-square test.

† Mann–Whitney U test.

‡ Student t test.

P values , 0.05 are shown in bold. ALND, axillary lymph node dissection; BCS, breast-conserving surgery; CPSNP, chronic postsurgical neuropathic pain; ICBN, intercostobrachial nerve; IQR, interquartile range; SLNB,

sentinel lymph node biopsy.


16

4.3. Type of surgery and chronic postsurgicalneuropathic pain

Axillary lymph node dissection has emerged as an important riskfactor for PPSP in multiple studies.3,15,27,28 In this cohort, ALNDwas significantly more frequent than SLNB in both groupsbecause ICBN resection is typically performed in ALND but not inSLNB. However, this study demonstrates that CPSNP may alsooccur in SLNB patients with ICBN resection. The small number ofSLNB patients in our cohort does not allow for multivariateanalyses of type of axillary surgery. The type of ICBN resection,partial or total, did not associate with a higher risk of having NPagreeing with previous studies.14

Breast-conserving surgery was associated with a higherprevalence of CPSNP than mastectomy in multivariate analysis,even after controlling for radiotherapy. Previous studies have alsoreported high incidence of PPSP after BCS, especially in theipsilateral arm,3,41 which may indicate a role for ICBN lesions inthe subsequent pain. The association between the type of breastsurgery and CPSNP ismultifactorial. The access and visualizationof the axilla is usually better in mastectomy than in BCS.Mastectomy and BCS patients differ in terms of breast cancer

characteristics–mastectomy patients have, on average, largertumors and more metastatic lymph nodes in the axilla. However,these factors did not differ between CPSNP and non-CPSNPgroups (Table 3).

4.4. Preoperative factors associating with future chronicpostsurgical neuropathic pain

Preoperatively, future CPSNP patients presented more pain,anxiety, and depressive symptoms than non-CPSNP patients.Preoperative pain in the surgical area and other chronic painshave previously been shown to predispose to PPSP.15,28,36,37

Similarly, we found other chronic pain and preoperative pain in thesurgical area to predispose to CPSNP. Chronic pain patientsoften have significant symptom overlap, which makes assess-ment of depressive symptoms challenging.22 Therefore, theintergroup differences in BDI II scores in our cohort may partlyreflect the chronic pain load in the patients who developedCPSNP. Moreover, depressive symptoms and anxiety did notpresent as statistically significant in the multivariate analysis forthe prediction of CPSNP, and the impact of preoperative pain andother chronic pains seems to override their effect in our model.

Table 4

Logistic regression model of the associations of preoperative and treatment-related factors with CPSNP after ICBN resection.

Bivariate analysis Stepwise logistic regression analysis

CPSNP (n 5 135) Non-CPSNP (n 5 65) P B OR (95% CI) P

BMI at the time of surgery, mean (SD), kg/m2 25.4 (3.90) 23.8 (3.64) 0.006*

Type of breast operation and radiotherapy,

number (%)

0.002†

BCS with radiotherapy§ 64 (47.4) 15 (23.1) 1 [reference]

Mastectomy without radiotherapy 31 (23.0) 27 (41.5) 21.79 0.17 (0.06-0.44) ,0.001

Mastectomy with radiotherapy 37 (28.1) 23 (35.4) 21.41 0.24 (0.09-0.65) 0.005

Worst pain in the surgical area past week

preoperatively, median (IQR), 0-10 NRS

2 (1-3) 0 (0-1) ,0.001‡ 0.83 2.29 (1.55-3.39) ,0.001

Worst other pain past week preoperatively,

median (IQR), 0-10 NRS

2 (1-4) 0 (0-3) ,0.001‡

Presence of any chronic pain condition

preoperatively, number (%)

40 (29.6) 4 (6.2) ,0.001† 1.64 5.16 (1.53-17.34) 0.008

Pain in the surgical area upon arrival to PACU,

median (IQR), 0-10 NRS║3 (0-5) 0 (0-4) 0.015†

Oxycodone consumption at PACU, mg/kg 0.18 (0.10) 0.14 (0.10) 0.010*

BDI II preoperatively, median (IQR){ 9 (4-13) 5 (2-9) 0.002‡

STAI state preoperatively, mean (SD)# 41.1 (10.41) 37.8 (11.74) 0.041*

STAI trait preoperatively, mean (SD)** 37.9 (9.39) 34.7 (10.70) 0.034*

Insomnia preoperatively, number (%)†† 0.027‡

Not at all 50 (37.0) 37 (56.9)

At least once a week 40 (29.6) 16 (24.6)

Every night 24 (17.8) 5 (7.7)

Nagelkerke pseudo-R2 was 0.364. Hosmer and Lemeshow Test for the model suggests a good fit to data as P5 0.423 (.0.05). Preoperative and treatment-related variables with P, 0.05 in the bivariate analysis were

included in the logistic regression analysis. Thirty-six patients (28 from the NP group and 8 from the non-NP group) had missing values in some of the variables. Only cases with complete sets of data were included in the

analysis. The missing values were not imputed. Preoperative insomnia was the source of most missing values in the regression analysis. The outcome of the analysis remained unaltered whether or not preoperative insomnia

was included.

* Student t test.

† Chi-squared test.

‡ Mann–Whitney U test.

§ Two patients from the NP group had undergone BCS without radiotherapy and were excluded from the analysis.

║ Value missing from one non-NP patient.

{ Value missing from one NP patient.

# Values missing from one NP and one non-NP patient.

** Values missing from 2 NP patients.

†† Values missing from 21 (15.6%) NP and 7 (10.8%) non-NP patients.

BCS, breast-conserving surgery; BDI II, Beck’s Depression Inventory II; BMI, body mass index; CI, confidence interval; CPSNP, chronic postsurgical neuropathic pain; IQR, interquartile range; OR, odds ratio; PACU,

postanesthesia care unit; STAI, State-Trait Anxiety Inventory.


17

4.5. Chronic pain load

Chronic postsurgical neuropathic pain patients seem to accu-mulate pain conditions. They had significantly more multisitepains than non-CPSNP patients. This suggests that inherentpatient-related risk factors play a role in both the developmentand maintenance of CPSNP. Interestingly, increased anxiety,pain catastrophizing, depressive symptoms, and impaired qualityof sleep in the CPSNP patients may reflect a similar biopsy-chosocial profile previously reported in patients with chronic andoverlapping pain conditions.24

4.6. Lipids, glucose, and inflammatory markers

Lipid profiles and glucose levels did not differ between CPSNPand non-CPSNP patients. A similar finding was reported inpatients having painful or nonpainful diabetic polyneuropathies.34

The CPSNP patients showed higher levels in inflammatorymarkers (hs-CRP and ORM) compared with non-CPSNPpatients. In addition, higher levels of hs-CRP, but not ORM,associated with increased cold pain tolerance in CPT, but with nointeraction with CPSNP. Subclinical inflammation with high hs-CRP levels has previously been associated with lower paintolerance in CPT.1,35

Orosomucoid is an acute-phase protein with various immu-nomodulatory functions, and it has been associated with neuro-inflammation.13,18 To our knowledge, it has not been studied inNP patients before. Further studies are needed to assess its rolein the pathogenesis of NP.

Our results suggest a role for low-grade inflammation in themaintenance of CPSNP, in line with previous evidence for otherNP conditions.8,10,38 A recent study also showed that NP patientshave increased levels of proinflammatory cytokines in thecerebrospinal fluid compared with healthy individuals.7 Interest-ingly, neuroinflammation has also been associated with de-pression, anxiety, and impaired sleep,43 which all associated withCPSNP in our cohort.

4.7. Cold pain sensitivity and tolerance

Sensitization and chronic inflammation have been suggested aspossible mechanisms in NP. The CPT was assessed pre-operatively and re-assessed 4 to 9 years later. To our knowledge,there are no other studies to show consistent CPT resultspreoperatively and several years after surgery. Previously, a goodtest–retest reliability of the CPT was shown within a 2-weekinterval.23 In this study, years after the initial CPT, patientstolerated the cold water significantly longer (Table 5). We wereunable to identify an explanation for this. The follow-up periodwas several years, and the patients had received oncologicaltreatments, which may have affected the cold pressor tolerance.Nearly 90% of the patients in both CPSNP and non-CPSNPgroups had received chemotherapy.

Sensitivity to preoperative CPT was not associated withCPSNPat 4 to 9 years postoperatively. However, postoperatively,the CPSNP patients were significantly more sensitive and lesstolerant in the CPT than the non-CPSNP patients. A previouscross-sectional study on CPT and PPSP suggested that otherchronic pains might override the effect of PPSP on CPT.19 Ourresults, however, showed that CPSNP patients were moresensitive in the CPT, even after multivariate adjustment ofconfounding factors including other pains. In line with this, a smallprevious study showed that patients having NP after ulnar ormedian nerve transection had decreased pain tolerance andincreased pain sensitivity in CPT, compared with non-NPpatients.42 These results together with ours may suggest a rolefor central sensitization in NP.

4.8. Strengths and limitations of the study

Strengths of this study are the relatively large and homogenouspatient cohort, with a long follow-up, with rich clinical data and

Table 6

Logistic regression analysis of preoperative experimental pain measures to predict CPSNP after ICBN resection.

Unadjusted model Fully adjusted model*

B OR (95% CI) P B OR (95% CI) P

Cold pain (n 5 187)

Withdrawal time, s 20.01 0.99 (0.98-1.00) 0.15 20.01 0.99 (0.98-1.00) 0.27

Pain intensity at withdrawal, 0-10 NRS 0.04 1.04 (0.89-1.21) 0.63 0.01 1.00 (0.85-1.19) 0.97

Heat pain (n 5 199)

Pain intensity at 43˚C, 0-10 NRS 0.21 1.24 (0.92-1.67) 0.16 0.15 1.16 (0.85-1.60) 0.35

Pain intensity at 48˚C, 0-10 NRS 0.15 1.16 (1.01-1.33) 0.03 0.10 1.10 (0.96-1.28) 0.18

Hosmer and Lemeshow Test: cold pain model P 5 0.714; heat pain model P 5 0.272 (.0.05).

* Models are adjusted for the following preoperative variables: age, BMI, chronic pain (no/yes), depression (yes/no, cutoff$19 in BDI II), and anxiety (yes/no, cutoff$40 in STAI state). P values, 0.05 are shown in bold.

B, unstandardized regression weight, BMI, body mass index; BDI II, Beck’s Depression Inventory II; CI, confidence interval; CPSNP, chronic postsurgical neuropathic pain; ICBN, intercostobrachial nerve; NRS, numeric rating

scale; OR, odds ratio; STAI, State-Trait Anxiety Inventory.

Table 5

General pain sensitivity: preoperative and postoperative cold

pressor test and preoperative heat pain test.

CPSNP Non-CPSNP P

Preoperative measurements

Heat pain intensity at 43˚C,


0 (0-1) 0 (0-0) 0.14*

Heat pain intensity at 48˚C,


3 (2-5) 3 (1-4) 0.020*

CPT withdrawal time†, median

(IQR), s

38 (21-89) 51 (27-90) 0.12*

CPT pain intensity at

withdrawal†, median (IQR),

0-10 NRS

9 (8-9) 9 (8-9) 0.89*

Postoperative measurements after

4-9 y

CPT withdrawal time, median

(IQR), s

65 (33-90) 90 (44-90) 0.019*

CPT pain intensity at withdrawal,


9 (8-10) 8 (6-9) 0.003*

* Mann–Whitney U test.

† Ten values are missing from the CPSNP and 3 from the non-CPSNP group.

P values, 0.05 are shown in bold. CPSNP, chronic postsurgical neuropathic pain; CPT, cold pressor test;

IQR, interquartile range.


18

new insights into factors predisposing to and associating withCPSNP. All patients were clinically examined and classifiedaccording to latest NP grading criteria. The study provides newsupport for the role of central sensitization in CPSNP as shownwith decreased CPT tolerance in the CPSNP patients.

A limitation of this study is that all patients had been treated forcancer and the results can therefore not be directly translated toother traumatic nerve injuries. Second, some of the variableswere only cross-sectional, and no conclusions regarding causalrelationship can be drawn, eg, regarding the possible role ofpreoperative inflammation. Third, the data of chronic and otherpains were collected more thoroughly at the research visit 4 to 9years from surgery than preoperatively. Although this does notallow for direct comparison between the time points, we candemonstrate a clear difference between CPSNP and non-CPSNP groups at both time points. Fourth, due to the long andvarying follow-up period, we cannot exclude the possibility thatsome of the non-CPSNP could have fulfilled the NP diagnosticcriteria at some point during the follow-up. However, the timespan from the index operation did not differ significantly betweenthe CPSNP and non-CPSNP groups.

5. Conclusions

The CPSNP and non-CPSNP patient groups differed preopera-tively in terms of other chronic pain, sleep, and psychologicalfactors. The CPSNP patients showed enhanced pain sensitivityand decreased pain tolerance in CPT only after they haddeveloped NP, suggesting central sensitization. In addition tothe mounting load of chronic pain, systemic inflammation mayalso have contributed to this. Our results suggest that several,possibly interlinked, patient-related risk factors may play a signif-icant role in the development and maintenance of chronic NPafter ICBN resection. These factors should be considered whenattempting to improve prevention and management of CPSNP.


T. Meretoja has received a grant from Cancer Foundation Finlandsr. The remaining authors have no conflicts of interest to declare.

Acknowledgements

The authors thank our research nurse Eija Ruoppa for excellentwork, and Jari Lipsanen, MPsy, and adjunct professor VesaNiskanen for their expert advice in statistical analyses. Theauthors also thank Les Hearn, MSc, for reviewing the Englishlanguage of themanuscript. The authors are grateful to all patientswhoparticipated in the study. This studywas funded by EuropeanUnion FP7 (# Health_F2-2013-602891), NeuroPain.Author contributions: E. Kalso, T. Meretoja, H. Harno, and R. Sipilacontributed to the study design and data collection. H. Harnoconducted the clinical examination, and R. Sipila conducted thepsychological assessment of the patients. L.Mustonen and T. Ahocontributed to the data analysis and the making of the figures andtables. L. Mustonen, T. Aho, H. Harno, T. Meretoja, and E. Kalsocontributed to the writing of the manuscript. All authors wereinvolved in data interpretation, review, and approval of the finalversion of the manuscript.

Figure 2. Cox-regression analysis. Survival curves for cold pain tolerance (A)and sensitivity (B). CPSNP, chronic postsurgical neuropathic pain; NRS,Numeric Rating Scale.

Table 7

Cox-regression analysis of cold pressor test in CPSNP and

non-CPSNP patients.

HR 95% CI P

Cold pressor tolerance

Unadjusted model

CPSNP (yes) 1.70 1.10-2.61 0.016Fully adjusted* model

CPSNP (yes) 1.59 1.01-2.51 0.047Age (y) 0.99 0.97-1.01 0.268

BMI (kg/m2) 0.99 0.94-1.04 0.653

Other pain (yes) 1.18 0.79-1.78 0.419

Anxiety (yes) 0.94 0.86-1.03 0.190

Depression (yes) 1.04 0.99-1.07 0.064

Maximum pain intensity NRS 5 10

Unadjusted model

CPSNP (yes) 2.09 1.13-3.89 0.019Fully adjusted* model

CPSNP (yes) 2.12 1.11-4.07 0.023Age (y) 0.97 0.94-0.99 0.020BMI (kg/m2) 0.95 0.89-1.02 0.188

Other pain (yes) 0.89 0.50-1.58 0.689

Anxiety (yes) 0.93 0.82-1.05 0.234

Depression (yes) 1.05 0.99-1.11 0.090

* Adjusted for age, BMI, other pain of at least moderate intensity (yes/no), depression (yes/no, cutoff$19 in

BDI II), and anxiety (yes/no, cutoff $11 in HADS-A). P values , 0.05 are shown in bold.

BDI II, Beck’s Depression Inventory II; BMI, body mass index; CI, confidence interval; CPSNP, chronic

postsurgical neuropathic pain; HADS-A, Hospital Anxiety-Depression Scale-Anxiety; HR, hazard ratio.


19

Appendix A. Supplemental digital content

Supplemental digital content associated with this article can befound online at http://links.lww.com/PAIN/A661.

Article history:Received 28 January 2018Accepted 4 September 2018Available online 17 September 2018

References

[1] Afari N, Mostoufi S, Noonan C, Poeschla B, Succop A, Chopko L,Strachan E. C-reactive protein and pain sensitivity: findings from femaletwins. Ann Behav Med 2011;42:277–83.

[2] Andersen KG, Aasvang EK, Kroman N, Kehlet H. Intercostobrachial nervehandling and pain after axillary lymph node dissection for breast cancer.Acta Anaesthesiol Scand 2014;58:1240–8.

[3] Andersen KG, Duriaud HM, Jensen HE, Kroman N, Kehlet H. Predictivefactors for the development of persistent pain after breast cancer surgery.PAIN 2015;156:2413–22.

[4] Attal N, Masselin-Dubois A, Martinez V, Jayr C, Albi A, Fermanian J,Bouhassira D, Baudic S. Does cognitive functioning predict chronic pain?Results from a prospective surgical cohort. Brain 2014;137:904–17.

[5] Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J. An inventory formeasuring depression. Arch Gen Psychiatry 1961;4:561–71.

[6] Bjelland I, Dahl AA, Haug TT, Neckelmann D. The validity of the HospitalAnxiety and Depression Scale. An updated literature review.J Psychosom Res 2002;52:69–77.

[7] Backryd E, Lind AL, ThulinM, Larsson A, Gerdle B, Gordh T. High levels ofcerebrospinal fluid chemokines point to the presence ofneuroinflammation in peripheral neuropathic pain—a cross-sectionalstudy of two cohorts of patients compared to healthy controls. PAIN2017;158:2487–95.

[8] Calvo M, Dawes JM, Bennett DL. The role of the immune system in thegeneration of neuropathic pain. Lancet Neurol 2012;11:629–42.

[9] Colloca L, Ludman T, Bouhassira D, Baron R, Dickenson AH, Yarnitsky D,Freeman R, Truini A, Attal N, Finnerup NB, Eccleston C, Kalso E, BennettDL, Dworkin RH, Raja SN. Neuropathic pain. Nat Rev Dis Primers 2017;3:17002.

[10] Doupis J, Lyons TE, Wu S, Gnardellis C, Dinh T, Veves A. Microvascularreactivity and inflammatory cytokines in painful and painless peripheraldiabetic neuropathy. J Clin Endocrinol Metab 2009;94:2157–63.

[11] Duale C, Ouchchane L, Schoeffler P; EDONIS Investigating Group,Dubray C. Neuropathic aspects of persistent postsurgical pain: a Frenchmulticenter survey with 6-month prospective follow-up. J Pain 2014;15:24.e1–e20.

[12] Finnerup NB, Haroutounian S, Kamerman P, Baron R, Bennett DL,Bouhassira D, Cruccu G, Freeman R, Hansson P, Nurmikko T, Raja SN,Rice AS, Serra J, Smith BH, Treede RD, Jensen TS. Neuropathic pain: anupdated grading system for research and clinical practice. PAIN 2016;157:1599–606.

[13] Fournier T, Medjoubi NN, Porquet D. Alpha-1-acid glycoprotein. BiochimBiophys Acta 2000;1482:157–71.

[14] FreemanSRM,Washington SJ, Pritchard T, Barr L, BaildamAD, BundredNJ. Long term results of a randomized prospective study of preservationof the intercostobrachial nerve. EJSO 2003;29:213–5.

[15] Gartner R, Jensen MB, Nielsen J, Ewertz M, Kroman N, Kehlet H.Prevalence of and factors associated with persistent pain following breastcancer surgery. JAMA 2009;302:1985–92.

[16] Haroutiunian S, Nikolajsen L, Finnerup NB, Jensen TS. The neuropathiccomponent in persistent postsurgical pain: a systematic literature review.PAIN 2013;154:95–102.

[17] Ilhan E, Chee E, Hush J,Moloney N. The prevalence of neuropathic pain ishigh after treatment for breast cancer: a systematic review. PAIN 2017;158:2082–91.

[18] Jo M, Kim JH, Song GJ, Seo M, Hwang EM, Suk K. Astrocyticorosomucoid-2 modulates microglial activation and neuroinflammation.J Neurosci 2017;37:2878–94.

[19] Johansen A, Schirmer H, Stubhaug A, Nielsen CS. Persistent post-surgical pain and experimental pain sensitivity in the Tromsø study:comorbid pain matters. PAIN 2014;155:341–8.

[20] Jaaskelainen SK, Teerijoki-Oksa T, Virtanen A, Tenovuo O, Forssell H.Sensory regeneration following intraoperatively verified trigeminal nerveinjury. Neurology 2004;62:1951–7.

[21] Kaunisto MA, Jokela R, Tallgren M, Kambur O, Tikkanen E, Tasmuth T,Sipila R, Palotie A, Estlander AM, Leidenius M, Ripatti S, Kalso EA. Pain in1,000 women treated for breast cancer: a prospective study of painsensitivity and postoperative pain. Anesthesiology 2013;119:1410–21.

[22] Knaster P, Estlander AM, Karlsson H, Kaprio J, Kalso E. Diagnosingdepression in chronic pain patients: DSM-IVmajor depressive disorder vsbeck depression inventory (BDI). PLoS One 2016;11:e0151982.

[23] Koening J, Jarczok M, Ellis R, Bach C, Thayer J, Hillecke T. Two-week test-retest stability of thecoldpressor taskprocedure at twodifferent temperaturesas a measure of pain threshold and tolerance. Pain Pract 2014;14:E126–35.

[24] Maixner W, Fillingim RB, Williams DA, Smith SB, Slade GD. Overlappingchronic pain conditions: implications for diagnosis and classification.J Pain 2016;17(9 suppl):T93–T107.

[25] Martinez V, Uceyler N, Ben Ammar S, Alvarez JC, Gaudot F, Sommer C,Bouhassira D. Clinical, histological, and biochemical predictors ofpostsurgical neuropathic pain. PAIN 2015;156:2390–8.

[26] Masselin-Dubois A, Attal N, Fletcher D, Jayr C, Albi A, Fermanian J,Bouhassira D, Baudic S. Are psychological predictors of chronic postsurgicalpain dependent on the surgical model? A comparison of total kneearthroplasty and breast surgery for cancer. J Pain 2013;14:854–64.

[27] Mejdahl MK, Andersen KG, Gartner R, Kroman N, Kehlet H. Persistentpain and sensory disturbances after treatment for breast cancer: 6 yearnationwide follow-up study. BMJ 2013;346:f1865.

[28] Meretoja TJ, Leidenius MH, Tasmuth T, Sipila R, Kalso E. Pain at 12months after surgery for breast cancer. JAMA 2014;311:90–2.

[29] Meretoja TJ, Andersen KG, Bruce J, Haasio L, Sipila R, Scott NW, Ripatti S,Kehlet H, Kalso E. Clinical prediction model and tool for assessing risk ofpersistent pain after breast cancer surgery. J Clin Oncol 2017;35:1660–7.

[30] Morin CM, Belleville G, Belanger L, Ivers H. The insomnia severity index:psychometric indicators to detect insomnia cases and evaluate treatmentresponse. Sleep 2011;34:601–8.

[31] Mantyselka P, Miettola J, Niskanen L, Kumpusalo E. Chronic pain,impaired glucose tolerance and diabetes: a community-based study.PAIN 2008;137:34–40.

[32] Pereira S, Fontes F, Sonin T, Dias T, FragosoM, Castro-Lopes J, Lunet N.Neuropathic pain after breast cancer treatment: characterization and riskfactors. J Pain Symptom Manage 2017;54:877–88.

[33] Phillips TJ, BrownM, Ramirez JD, Perkins J,Woldeamanuel YW,Williams AC,OrengoC,Bennett DL,Bodi I, CoxS,MeierC,KrumovaEK,RiceAS.Sensory,psychological, and metabolic dysfunction in HIV-associated peripheralneuropathy: a cross-sectional deep profiling study. PAIN 2014;155:1846–60.

[34] Raputova J, Srotova I, Vlckova E, Sommer C, Uceyler N, Birklein F, RittnerHL, Rebhorn C, Adamova B, Kovalova I, Kralickova Nekvapilova E, ForerL, Belobradkova J, Olsovsky J, Weber P, Dusek L, Jarkovsky J, BednarikJ. Sensory phenotype and risk factors for painful diabetic neuropathy:a cross-sectional observational study. PAIN 2017;158:2340–53.

[35] Schistad EI, Stubhaug A, Furberg AS, Engdahl BL, Nielsen CS. C-reactive protein and cold-pressor tolerance in the general population: theTromsø Study. PAIN 2017;158:1280–8.

[36] Schou Bredal I, Smeby NA, Ottesen S,Warncke T, Schlichting E. Chronicpain in breast cancer survivors: comparison of psychosocial, surgical,and medical characteristics between survivors with and without pain.J Pain Symptom Manage 2014;48:852–62.

[37] Sipila R, Estlander AM, Tasmuth T, Kataja M, Kalso E. Development ofa screening instrument for risk factors of persistent pain after breastcancer surgery. Br J Cancer 2012;107:1459–66.

[38] Sommer C, Leinders M, Uceyler N. Inflammation in the pathophysiologyof neuropathic pain. PAIN 2018;159:595–602.

[39] Spielberger CD, Gorsuch RL, Lushene PR, Vagg PR, Jacobs AG. Manualfor the state-trait anxiety inventory (form Y). Palo Alto: ConsultingPsychologists Press, Inc, 1983.

[40] Sullivan MJL, Bishop S, Pivik J. The pain catastrophizing scale:development and validation. Psychol Assess 1995;7:524–32.

[41] Tasmuth T, von Smitten K, Kalso E. Pain and other symptoms during thefirst year after radical and conservative surgery for breast cancer. Br JCancer 1996;74:2024–31.

[42] Taylor KS, Anastakis DJ, Davis KD. Chronic pain and sensorimotordeficits following peripheral nerve injury. PAIN 2010;151:582–91.

[43] Walker AK, Kavelaars A, Heijnen CJ, Dantzer R. Neuroinflammation andcomorbidity of pain and depression. Pharmacol Rev 2013;66:80–101.

[44] Warrier S, Hwang S, Koh CE, Shepherd H, Mak C, Carmalt H, SolomonM. Preservation or division of the intercostobrachial nerve in axillarydissection for breast cancer: meta-analysis of randomised controlledtrials. Breast 2014;23:310–16.

[45] Ziegler D, Rathmann W, Dickhaus T, Meisinger C, Mielck A. Neuropathicpain in diabetes, prediabetes and normal glucose tolerance; theMONICA/KORA augsburg surveys S2 and S3. Pain Med 2009;10:393–400.


20

Systematic Review and Meta-Analysis

Psychosocial factors associated with persistentpain in people with HIV: a systematic reviewwith meta-analysisWhitney Scotta,*, Chinar Arkutera, Kitty Kiosklia, Harriet Kempb, Lance M. McCrackena,c, Andrew S.C. Riceb,Amanda C. de C. Williamsd

AbstractChronic pain remains a prevalent and disabling problem for people living with HIV in the current antiretroviral treatment era.Psychosocial treatments may have promise for managing the impact of this pain. However, research is needed to identifypsychosocial processes to target through such treatments. The current systematic review and meta-analysis examined theevidence for psychosocial factors associated with pain, disability, and quality of life in people living with HIV and persistent pain.Observational and experimental studies reporting on the association between one or more psychosocial factors and one or morepain-related variables in an adult sample of people living with HIV and pain were eligible. Two reviewers independently conductedeligibility screening, data extraction, and quality assessment. Forty-six studies were included in the review and 37 of these provideddata for meta-analyses (12,493 participants). “Some” or “moderate” evidence supported an association between pain outcomes inpeople with HIV and the following psychosocial factors: depression, psychological distress, posttraumatic stress, drug abuse, sleepdisturbance, reduced antiretroviral adherence, health care use, missed HIV clinic visits, unemployment, and protectivepsychological factors. Surprisingly, few studies examined protective psychological factors or social processes, such as stigma.Therewere few high-quality studies. These findings can inform future research and psychosocial treatment development in this area.Greater theoretical and empirical focus is needed to examine the role of protective factors and social processes on pain outcomes inthis context. The review protocol was registered with PROSPERO (CRD42016036329).

Keywords: HIV, Pain, Systematic review, Psychosocial factors

1. Introduction

HIV remains a significant global health concern with 36.7 millionpeople living with HIV worldwide.130 The availability of combinedantiretroviral therapy (cART) has drastically improved life expec-tancy.9,93,120 In well-resourced countries, and increasingly in lesswell-resourced regions, the shift in HIV from a terminal illness to

a chronic condition has led to a focus on disease and symptommanagement.59

Chronic pain is a common symptom in people with HIV. Datafrom one systematic review indicate that 54% to 83% of peoplewith HIVmay experience clinically meaningful persistent pain, andthese estimates seem to be stable from the pre- to current-cARTera.80 Neuropathic pain is a frequent complication of HIV and/orantiretroviral therapy.15 Approximately 42% to 66%of peoplewithHIV have peripheral sensory neuropathy (HIV-SN), and around54% to 78% of these experience neuropathic pain.84,88,129

Importantly, pain in people with HIV is associated with increaseddisability and reduced quality of life.27

There are few pharmacological options for managing chronicHIV-related pain. A systematic review of 14 randomizedcontrolled trials (RCTs) of pharmacotherapy for painful HIV-SNfound efficacy only for topical capsaicin, smoked cannabis, andsubcutaneous nerve growth factor.85 However, nerve growthfactor is not clinically available, capsaicin is not feasible in lower-resourced settings, and a subsequent review of cannabisshowed no effect on neuropathic pain and concerns aboutlong-term side effects.32 Additional negative RCTs of pregabalin,capsaicin, and amitriptyline have been published.17,24,103

In the wider literature, psychological approaches are commonin chronic pain management.34 Psychological treatments, in-cluding cognitive–behavioural therapy (CBT), are associated withimproved functioning and mood for chronic pain that is primarilymusculoskeletal.124 However, research on psychological treat-ments for pain in HIV is less well developed. Only 2 RCTs haveexamined CBT for people with HIV and chronic pain, but



a Health Psychology Section, Institute of Psychiatry, Psychology, and Neurosci-

ence, King’s College London, London, United Kingdom, b Pain Research Group,

Department of Surgery and Cancer, Faculty of Medicine, Imperial College London,

London, United Kingdom, c INPUT Pain Management Unit, Guy’s and St Thomas’

NHS Foundation Trust, London, United Kingdom, d Research Department of

Clinical, Educational, and Health Psychology, University College London, London,

United Kingdom

*Corresponding author. Address: Health Psychology Section, Institute of Psychiatry,

Psychology, and Neuroscience, King’s College London, 5th Floor Bermondsey

Wing, Guy’s Hospital Campus, London SE1 9RT, United Kingdom. Tel.: 144

2071885421. E-mail address: [email protected] (W. Scott).

Supplemental digital content is available for this article. Direct URL citations appear

in the printed text and are provided in the HTML and PDF versions of this article on

the journal’s Web site (www.painjournalonline.com).

PAIN 159 (2018) 2461–2476


of the International Association for the Study of Pain. This is an open access article

distributed under the Creative Commons Attribution License 4.0 (CCBY), which

permits unrestricted use, distribution, and reproduction in anymedium, provided the

original work is properly cited.


December 2018·Volume 159·Number 12 www.painjournalonline.com 2461

21

interpretation of these trials is hampered by small samples118 andhigh dropout rates.29 An observational study of CBT for HIV-related pain showed similarly poor treatment completion.21,113

There is a clear need for improving psychological treatments forpeople with HIV-related pain.

Improving psychological approaches for chronic pain in HIVwillrequire consideration of the psychosocial complexities associ-ated with HIV. For example, stigma, mental health problems, andsubstance abuse may influence pain and treatment engagementin people with HIV.36,69,71,111,123,132 However, research has notsystematically examined psychosocial factors associated withpain in this context. The systematic review by Parker et al. (2014),which estimated the prevalence of pain in HIV, described 5studies reporting psychosocial factors. However, that review didnot specifically include assessment of psychosocial factors in theeligibility criteria. Furthermore, 33 potentially eligible studies wereexcluded due to low-quality ratings,80 which limits our un-derstanding of the range of psychosocial factors examined in thiscontext. Therefore, we conducted a systematic review andmeta-analyses to examine the associations between psychosocialfactors and persistent pain in HIV. Because the aims of the reviewwere exploratory, we did not formulate specific hypotheses aboutthe associations between these variables.

2. Methods

The review protocol was registered with PROSPERO (http://www.crd.york.ac.uk/PROSPERO/display_record.php?ID5CRD42016036329).

2.1. Inclusion and exclusion criteria

2.1.1. Inclusion

(1) People with HIV aged 18 years and older.(2) The original protocol specified the study must have a “(sub)

sample with average pain duration of $3 months.” Afterpiloting this criterion, a large number of studies did not defineor report pain duration. We contacted authors to enquireabout pain chronicity; however, these data were generally notavailable. Given the potentially high prevalence of chronic painin HIV,80 we decided to include studies with (sub)samples ofambiguous pain duration, provided that “pain” vs “no pain”subgroup analyses were reported in studies for which chronicpain was not an eligibility criterion or which did not report painduration.

(3) Data on presence of pain, pain intensity, functioning, and/orquality of life.

(4) Data on one or more psychosocial variable, representing anypotentially modifiable cognitive, affective, behavioural, orinterpersonal process. Adherence to antiretroviral therapyand health care use variables represent modifiable behaviourpatterns. Therefore, we considered these as psychosocialvariables eligible for this review.

(5) Observational (cross-sectional, case-control, or prospective)or experimental studies (RCTs) reporting between- or within-groups associations between pain and psychosocial variablesin a (sub)sample with pain.

(6) Any language, from any region, from 1981 onwards (the datethat HIV was identified in the literature).

(7) Studies (published and unpublished) with an available full-text.Where only abstracts or trial registration summaries wereavailable, the authors provided unpublished data for thereview. Unpublished studies are commonly included in

systematic reviews, given recognition of overestimation ofeffects in published research.62 In addition, studies conductedin lower-resourced countries where HIV is particularlyprevalent may not always proceed through to publication.Therefore, the inclusion of unpublished studies and disserta-tions may help overcome this disparity and allows us toconsider potential contextual differences.

2.1.2. Exclusion

(1) Studies only measuring associations between unchangeabledemographic factors (eg, age, ethnicity) and pain. Whilepiloting the eligibility criteria, we found a number of studiesreporting history of injecting drug use as participants’ HIV riskfactor. Given lack of further information about substanceabuse history or current abuse, we excluded studies for whichinjecting drug use history was the only psychosocial factor.Likewise, we excluded studies reporting only average units ofalcohol consumed, rather than alcohol abuse.

(2) Qualitative studies.

2.2. Search strategy

We searched the following databases during March 2016:Medline, EMBASE, CINAHL, PsycINFO, Cochrane, and Web ofScience. We also searched ISRCTN, clinicaltrials.gov, and EUClinical Trials Register. Reference lists of eligible studies weresearched and key authors were contacted. We reran the searchin August 2017. The search included terms for the targetpopulation (HIV or AIDS), outcome (chronic pain), and exposuremeasurement (psychosocial factors). Relevant search termswere identified from previous reviews on pain in HIV,80,124

psychosocial factors in HIV,99 and psychosocial factors in chronicpain40 (Appendix A, available at http://links.lww.com/PAIN/A643).

2.3. Data extraction

Two reviewers (W.S. and C.A.) independently screened titles/abstracts and full-texts for eligibility. The following data wereextracted from eligible studies: year; design; country; samplesize; demographics (ie, age, sex, and race/ethnicity); clinicalfactors (ie, HIV duration, use of ART, CD41 count and viral load,and pain duration and type); assessment of pain and psychoso-cial variables; and, statistical analyses. In cases where bothcross-sectional and prospective data reported the same (or anoverlapping) cohort and variables, the prospective analyses wereextracted. Data were extracted from all studies by W.S., andindependently by C.A. and K.K. who each extracted data fromapproximately half of the studies. Disagreements regardingeligibility and data extraction were discussed to reach consensusand, where discrepancies remained, W.S. discussed these withthe wider team. The reviewers were not blinded to the authorshipof the studies reviewed.

2.4. Quality assessment

We assessed the methodological quality of studies using anadapted version of quality assessment tools used in previoussystematic reviews of observational studies relevant to pain andHIV.3,40,80 The quality assessment tool contained itemsassessing: study purpose, recruitment, response rates,sample description, assessment measurements, data analy-sis, and confounding/matching (Appendix B, available at

2462 W. Scott et al.·159 (2018) 2461–2476 PAIN®

22

http://links.lww.com/PAIN/A643). Additional items assessedfeatures specific to prospective designs. Thus, quality scoresdiffered for cross-sectional and prospective studies. In somecases, the overall study design did not correspond to the nature ofthe data extracted for the purpose of this review. In addition, insome cohorts, a cross-sectional design was used to examine onepsychosocial variable, whereas a prospective design was used toexamine another variable using the same sample. In all cases, thequality assessmentwas applied to the design used for the nature ofthe data extracted for a given psychosocial variable. Qualityassessment items were rated as “positive” (1), “negative” (0), or“unclear” (?), and total scoreswere computed and classified as low(,50%), medium (50%-80%), and high (.80%).3,40 W.S. com-pleted quality assessment ratings for all studies, whereas C.A. andK.K. each independently completed the quality assessment forapproximately half of the studies. The strength of evidence wasassessed according to the levels outlined by Ariens et al.3 ina systematic review of observational studies of psychosocial risk

factors for neck pain: (1) Strong: consistent results in multiple high-quality prospective and/or case-control studies; (2) Moderate:consistent results in multiple prospective and/or case-controlstudies; (3) Some Evidence: findings in one prospective or case-control study, or consistent findings in multiple cross-sectionalstudies with at least one high-quality study; and (4) Inconclusive:Inconsistent findings in multiple studies or consistent findings inmultiple low-quality cross-sectional studies.3

2.5. Data synthesis

Meta-analyses were conducted using Stata 15.0 where therewere at least 2 studies43 with the same design and effectestimate of the association between the same pain (eg,intensity) and psychosocial variables (eg, depression). Wetook a broad approach to the meta-analyses,37,45 andgrouped psychosocial variables on the basis that theyreflected conceptually similar underlying constructs with

Figure 1. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA75) flow diagram.


23

Table 1

Summary of the evidence for psychosocial variables.

Study quality Psychosocial assessment Study design/analyses Main summary

Depression

High: (n 5 2)79,84 BDI* (n 5 11); BSI (n 5 2) Prospective (HR) (n 5 2; Fig. S1,

http://links.lww.com/PAIN/A643)

Depression consistently positively associated

with pain presence, intensity, and interference,

and negatively with quality of life in prospective

and cross-sectional studies.

Medium: (n 5 19)2,27,29†,31,48,50,60,61,67,88,89,91,

92,96,97,102,104,114,116,119

CDC (n 5 1); CES-D (n 5 6) Prospective (OR) (n 5 2; Fig. S2,


Evidence level: moderate

Low: (n 5 8)8,28,39,55,57,95,101,110 CIDI* (n 5 1); DAPOS (n 5 1) Cross-sectional (OR) (n 5 9, includes 1

cross-sectional case-control; Fig. 2)

DASS (n 5 1); GAIN (n 5 1) Cross-sectional (SMD) (n 5 13; Fig. 3/text)

HADS (n 5 1); HSC (n 5 1) Cross-sectional (correlation) (n5 8; Fig. S8,

http://links.lww.com/PAIN/A643/text)

PHQ (n 5 2); PRISM (n 5 1)

Nonvalidated (n 5 1)

Psychological distress

High: (n 5 1)84 BSI (n 5 2); CSQ (n 5 1) Prospective (OR) (n 5 1; in text) Distress consistently positively associated

with pain presence, intensity, and

interference in cross-sectional analyses, but

this was not consistent in the 2 prospective

studies.

Medium: (n 5 11)2,29†,31,44,47,50,61,66,86,88,91,104

GAIN (n 5 1); HADS (n 5 1) Prospective (correlation) (n 5 1; in text) Evidence level: some

HSC*(n 5 1); INTRP (n 5 1) Cross-sectional (OR) (n 5 5; Fig. 4)

PASS (n 5 1); PCS* (n 5 4)

PHQ (n 5 1); POMS (n 5 1)

PRISM (n 5 1)

Low: (n 5 6)5,53,54,57,65,110 Clinician diagnosis‡ (n 5 1)

Medical file (ICD-9, n 5 1; unclear, n 5 2)

Cross-sectional (SMD) (n 5 8; Fig. 5/text)

Cross-sectional (correlation) (n5 5; Fig. S9,


Posttraumatic stress (PTSS)

High: (n 5 1)79 HTQ (n 5 1) Prospective (ANCOVA) (n 5 1; text) PTSS associated with poorer pain outcomes

in one prospective and one high-quality

cross-sectional study.

Medium: (n 5 2)31,107 PCL-C (n 5 1) Cross-sectional (SMD and correlation)

(n 5 1; text)

Evidence level: some

PRISM (n 5 1) Cross-sectional (OR) (n 5 1; text)

Drug abuse

Medium: (n 5 9)47,52,60,67,76,92,104,114,119

CIDI (n 5 1); DIS-IV (n 5 1) Prospective (OR) (n 5 4; Fig. S10,


Pain and drug abuse consistently positively

associated in prospective and cross-

sectional studies.

Low: (n 5 5)39,53,54,81,110 GAIN (n 5 1); PRISM (n 5 1) Cross-sectional (OR) (n 5 8, includes 1

cross-sectional case-control; Fig. 6)

Evidence level: moderate

RBS (n 5 1) Cross-sectional (SMD) (n 5 1; text)

Medical file (ICD-9; n 5 1) Cross-sectional (correlation) (n 5 1; text)

Medical file (other; n 5 2)

Other self-report (n 5 6)

Alcohol abuse

Medium: (n 5 8)27,52,60,76,77,115,117,119

CIDI (n 5 2); DIS-IV (n 5 1) Prospective (HR) (n 5 1; text) Lack of association between pain and

alcohol abuse in cross-sectional studies;

inconsistent findings in prospective studies.

Low: (n 5 3)6,39,56 DSM-IV (n 5 1); NIAAA (n 5 2) Prospective (OR) (n 5 3; Fig. S11, http://

links.lww.com/PAIN/A643/text)

Evidence level: inconclusive

PRISM (n 5 1); SCID (n 5 1) Cross-sectional (OR) (n 5 7) (Fig. S12,


Other self-report (n 5 2)

Unclear (n 5 1)

Sleep disturbance

High: (n 5 1)84 GSDS (n 5 1) Cross-sectional (SMD) (n 5 2; Fig. S13,


Sleep disturbance and pain positively

associated in 2/3 cross-sectional studies.

Medium: (n 5 2)2,94 ISI (n 5 1) Evidence level: some

PSQI (n 5 1) Cross-sectional (correlation) (n 5 1; text)

(continued on next page)

2464 W. Scott et al.·159 (2018) 2461–2476 PAIN®

24

Table 1 (continued)

Study quality Psychosocial assessment Study design/analyses Main summary

Antiretroviral nonadherence

Medium: (n 5 2)66,68 ACTG-AQ (n 5 5) Prospective (OR) (n 5 1; text) Pain associated with reduced antiretroviral

adherence in one prospective study and

majority of cross-sectional studies.

Evidence: some

Low: (n 5 5)8,46,57,72,110 Other self-report (n 5 2) Cross-sectional (OR) (n 5 4; Fig. S14/15,


Cross-sectional (correlation/SEM; n 5 2;

text)

Health care use

Medium: (n 5 4)47,61,68,119 Medical records (GP/HIV/urgent care visits)

(n 5 4)

Prospective (OR) (n 5 1; text) Pain associated with greater health care use

in one large prospective study and 2 large

cross-sectional studies. No association in 3

relatively small cross-sectional studies.

Low: (n 5 2)39,53 Self-report (mental health treatment) (n5 2) Cross-sectional (OR) (n 5 2; Fig. S16,



Cross-sectional (SMD) (n 5 3; Fig. S17,


Missed HIV clinic visits

Medium: (n 5 2)68,92 Medical records Prospective (OR) (n 5 2; Fig. S18, http://

links.lww.com/PAIN/A643)

Pain associated with greater missed HIV

clinic visits based on pooled effect of 2 large

prospective studies.


Unemployment

High: (n 5 1)79 WPAIQ (n 5 1) Cross-sectional (OR) (n 5 8, includes 1

cross-sectional case-control; Fig. S19,


Pain consistently associated with

unemployment (cross-sectional studies).

Medium: (n 5 7)2,27,61,92,114,119,121 Other self-report (n 5 7) Evidence level: some

Protective factors

High: (n 5 1)79 ACTG-AQ self-efficacy item (n 5 1) Prospective (correlation) (n 5 1) Resilience (1 case-control), acceptance (1

prospective), and self-efficacy (2 cross-

sectional, including 1 high quality)

associated with improved pain outcomes.

Optimism examined in only one low-quality

cross-sectional study.


Medium: (n 5 2)86,121 CPAQ (n 5 1) Cross-sectional (OR) (n 5 1)

Low: (n 5 2)8,101 Life orientation test (n 5 1) Cross-sectional (SMD/correlation) (n 5 3,

includes 1 cross-sectional case-control)

The resilience scale (n 5 1) All discussed in text

SE-6 (n 5 1)

Social factors

Medium: (n 5 3)73,74,91,122 HASI-P (n 5 1) (stigma) Prospective (correlation) (n 5 1) Too few studies, lack of high-quality studies,

and some inconsistencies for social

processes (eg, negative support/support

quality)

Low: (n 5 2)72,98 RSS (n 5 1) (stigma) Cross-sectional (OR) (n 5 1) Evidence level: inconclusive

SSQ-SF (n 5 1) (support quality/number) Cross-sectional (SMD/correlation) (n 5 3)

Self-report: Negative support/support

reciprocity/patient–provider engagement

(not validated; n 5 1)

All discussed in text

As justified in the Methods, we chose to focus on interpreting bivariate data. However, studies reporting a multivariate model between psychosocial and pain variables are bolded in column 1 for ease of reference. Additional

references relating to the same or overlapping studies reported in Table 1: 11, 12, 13, 22, 30, 70, 90, 105.

* Where different measures of the same variable were used for between-groups and within-group analyses in the same study, both measures are reported.

† Reference for multivariate analyses related to bivariate data reported in Ref. 29 is Ref. 38.

‡ Patient report of diagnosis of mental illness by a clinician.

ACTG-AQ, AIDS Clinical Trials Group Adherence Questionnaire; ANCOVA, analysis of covariance; BDI, Beck Depression Inventory; BSI, Brief Symptom Inventory; CDC, Centers for Disease Control; CES-D, Centre for

Epidemiological Studies Depression Scale; CIDI, Composite International Diagnostic Interview; CPAQ, Chronic Pain Acceptance Questionnaire; CSQ, Coping Strategies Questionnaire; DAPOS, Depression, Anxiety, and Positive

Outlook Scale; DASS, Depression Anxiety Stress Scales; DIS-IV, Diagnostic Interview Schedule for DSM-IV (Diagnostic and Statistical Manual of Mental Disorders, 4th Edition); GAIN, Global Appraisal of Individual Needs; GSDS;

General Sleep Disturbance Scale; HADS, Hospital Anxiety and Depression Scale; HASI-P, HIV/AIDS Stigma Scale—People Living with AIDS; HR, hazard ratio; HSC, Hopkins Symptoms Checklist; HTQ, Harvard Trauma

Questionnaire; ICD-9, International Classification of Disease mental illness/substance abuse codes extracted from medical file; INTRP, Inventory of Negative Thoughts in Response to Pain; ISI, Insomnia Severity Index; NIAAA,

National Institute on Alcoholism and Alcohol Abuse Guidelines; OR, odds ratio; PASS, Pain Anxiety Symptom Scale; PCL-C: PTSD Checklist-Civilian; PCS, Pain Catastrophizing Scale; PHQ, Patient Health Questionnaire

Depression/Anxiety Module; POMS, Profile of Mood States—Tension/Anxiety; PRISM, Psychiatric Research Interview for Substance and Mental Disorders; PSQI, Pittsburgh Sleep Quality Index; PTSS, posttraumatic stress

symptom; RBS, Risk Behaviour Survey; RSS, Reece Stigma Scale; SCID, Structured Clinical Interview for DSM-IV; SE-6, Self-Efficacy for Managing Chronic Disease—6 Item Scale; SEM, structural equation model; SMD,

standardized mean difference; SSQ-SF, Social Support Questionnaire—Short Form; WPAIQ, Work Productivity and Activity Impairment Questionnaire.


25

overlapping measurement content. All analyses were con-ducted using random effects, given likely heterogeneity.43

Between-study heterogeneity (I2 statistic) was interpreted aslow (,25%), medium (25%-50%), and high (.50%).124 Forbetween-groups comparisons of continuous data, meanvalues, SDs, and sample sizes were extracted to compute

the pooled standardized mean difference (SMD). For between-groups comparisons of dichotomous data, events data andsample sizes were extracted. Where events data were notreported, odds or hazard ratios, 95% confidence intervals(95% CIs), and sample sizes were extracted. To aggregatestudies reporting a mixture of odds ratios (ORs) and events

Figure 2. Forest plot of cross-sectional odds ratios (ORs) for depression. Depression wasmore likely in participants with vs without pain, as reflected in the pooledOR of.1. Gray boxes show weighting of individual studies; the red dotted line indicates the pooled effect around which effects from individual studies vary; theblue diamond shows the 95% CI around the pooled effect. CI, confidence interval.

Figure 3. Forest plot of cross-sectional standardized mean differences (SMDs) for depression. Depression symptoms were more severe in participants with vswithout pain, as indicated by a positive pooled SMD. CI, confidence interval.

2466 W. Scott et al.·159 (2018) 2461–2476 PAIN®

26

data, ORs were first computed from studies reporting eventsdata and then pooled with ORs reported in other studies. Oddsratios and hazard ratios were analysed separately. Whereapplicable, correlation coefficients (Pearson r) were extractedwith sample sizes. We transformed r to Fisher z and computed95% CIs of z to compute the pooled estimate.18,100

Data extracted were from bivariate analyses. Multivariate data(eg, adjusted ORs) were only extracted where bivariate data werenot available. We focused on bivariate analyses because manystudies did not report multivariate analyses. Moreover, studiesthat reportedmultivariatemodels varied substantially with respectto control variables included and inconsistently used psychoso-cial variables as independent or dependent variables. Takentogether, these differences limit meaningful interpretation ofmultivariate analyses across studies.

Several studies presented data on more than 2 pain/no paingroups, often with idiosyncratic group definitions, which limitedour ability to compare studies. Where studies reported 3 or moregroups, we collapsed these into 2 to represent groups with andwithout pain (eg, frequent/moderate/severe vs infrequent/mild/none), and computed effects between these. This approachfacilitated more direct comparison across studies and thusenabled us to include a larger number of studies in the analyses.For studies comparing participants on the presence of neurop-athy, we prioritised extracting data from these comparisons in thefollowing order depending on the data reported: (1) painful vsnonpainful neuropathy; (2) painful neuropathy vs no neuropathy;and (3) neuropathy vs no neuropathy.Where there wasmore thanone measure of the same variable, we extracted data for themeasure with the widest usage or the longer measure to increasereliability.124 Our protocol specified that funnel plots would beinspected to assess for publication bias. However, due to therelatively small number of studies in each meta-analysis and thelikelihood of high heterogeneity, inspection of funnels plots wasnot appropriate109 and, therefore, was not undertaken.

We conducted sensitivity analyses to examine the influence ofthe following study and patient characteristics on the findings:certainty of pain chronicity, pain type, immune functioning andviral suppression, ART treatment era, and health care system.With the exception of the pain chronicity analysis, these sensitivityanalyses were prespecified. Given the large number of potentialanalyses, we restricted sensitivity analyses to the between-groups SMDs for depression because this was the analysis withthe largest number of studies.

3. Results

Forty-six studies were included in the review (13,480 participants)and 37 of these provided data for meta-analyses (12,493participants; Fig. 1). Most (83%) were conducted in the UnitedStates, with 4 studies fromSouth Africa,79,88,121,122 and one eachfrom the United Kingdom,84 Thailand,89 Uganda,95 and Rus-sia.114 Participants were primarily recruited from HIV clinics orusing multifaceted strategies that also included recruitment fromsubstance abuse clinics and community outreach. One studyrecruited exclusively from a methadone clinic,8 whereas 2 othersrecruited in high-poverty areas.39,110 The samples comprisedpredominantly men in 41 studies, with the proportion of men inthese studies ranging from 51%50 to 100%.28,104 Five studies (4from South Africa and 1 from the United States) recruited womenexclusively79,92 or predominantly (proportion of women rangingfrom 72% to 88%).88,121,122 The mean age ranged from 30.1 (SD5 5.2)114,115 to 51.0 (SD 5 9.3) years.119 HIV duration was notconsistently reported; however, of the studies providing data,

duration ranged from 2.09 (SD 5 1.22)102 to 16.95 years (SD 58.70).119 Eighteen studies (39%) reported on mixed HIV/AIDSsamples (reported proportion with AIDS ranged from 10% to74%). Four studies included only participants with AIDS, onestudy excluded patients with AIDS, whereas 23 studies did notclearly report the proportion (if any) with AIDS. SupplementalTable 1 shows further demographic characteristics of the studysamples (available at http://links.lww.com/PAIN/A643).

Table 1 provides a summary of study designs, quality, andevidence level for each psychosocial factor. The studies showedsubstantial variability in the measurement of pain and psychoso-cial variables. Most studies (63%) were of medium quality. Fifteenstudies were of low quality, and only 2 were of high quality.79,84

The most common limitations included unclear reporting ofresponse rates, no a priori sample size justification, and poorreporting of HIV and pain characteristics. There is no singleagreed upon strategy to best address low-quality studies withinmeta-analyses, an issue which is compounded by the arbitrarynature of study quality scoring and cutoff points.43 This can bedealt with by only including high-quality studies, performingsensitivity analyses, or including all studies irrespective of qualityand discussing risk of bias.43 Given that only 2 of 46 studies wererated as high quality, ameta-analysis of these cannot be regardedas reflecting most of the studies. Sensitivity analyses wouldlikewise not be meaningful. Including all studies is thus the mostjustifiable approach for the current data. Although we havechosen to focus on data from bivariate analyses for reasonsoutlined in the Methods, studies that reported a multivariatemodel of the association between psychosocial and painvariables are shown in bold in Table 1 for ease of reference.

3.1. Depression

Depression was the most frequently assessed psychologicalvariable, investigated in 29 studies. Two prospective studiesreported hazard ratios for baseline depression predicting time toonset of symptomatic neuropathy. The pooled hazard ratio wassignificant and indicated that baseline depression was moresevere in participants who developed symptomatic neuropathyat follow-up than those who did not: HR 5 1.04 (95% CI 1.02-1.07), z 5 3.23, P 5 0.001 (supplemental Figure 1, available athttp://links.lww.com/PAIN/A643). Heterogeneity was 0.0%.Two further prospective studies reported ORs. The pooled ORwas significant and indicted that higher baseline depressionsymptoms were associated with greater likelihood of follow-uppain: OR 5 2.26 (95% CI 1.47-3.47), z 5 3.72, P , 0.001(supplemental Figure 2, available at http://links.lww.com/PAIN/A643). Heterogeneity was medium (40.1%). Nine cross-sectional studies provided events data or ORs. The pooled ORwas significant such that depression was more likely inparticipants with vs without pain: OR 5 2.65 (95% CI 1.62-4.34), z 5 3.90, P , 0.001 (Fig. 2). Heterogeneity was high(83.0%). Twelve cross-sectional studies provided data tocompute SMDs (Fig. 3). The overall effect was significant andshowed moderately greater depression in participants with vswithout pain: SMD 5 0.68 (95% CI 0.42-0.93), z 5 5.22, P ,0.001. Heterogeneity was high (I2 5 89.2%). Another cross-sectional study that reported the median and interquartile rangefound no difference in depression between groups with (n5 125)and without pain (n 5 72).88

Six cross-sectional studies reported correlation coefficientsbetween depression and pain severity. The pooled correlationwas small, but significant: Fisher z5 0.26 (95%CI 0.18-0.33), z56.77, P , 0.001. Heterogeneity was 0.0%. One additional study


27

found a nonsignificant correlation, although the coefficient wasnot reported.27 Four cross-sectional studies reported correla-tions between depression and pain interference/disability. Thepooled correlation was moderate: Fisher z 5 0.48 (95% CI 0.41-0.56), z 5 12.48, P , 0.001. Heterogeneity was 0.0%. Threecross-sectional studies reported correlations between depres-sion and quality of life. The pooled correlation was large andsignificant: Fisher z520.52 (95% CI20.75 to20.30), z5 4.51,P , 0.001. Heterogeneity was high at 73.3% (all correlationanalyses, supplemental Figure 8, available at http://links.lww.com/PAIN/A643). One final cross-sectional study (n 5 120)reported a moderate correlation between pain presence anddepression.55

3.2. Depression sensitivity analyses

We conducted sensitivity analyses on the SMDs for depressionbetween pain and no pain groups (supplemental Figures 3–7,

available at http://links.lww.com/PAIN/A643). We excluded data

from the Kirkland study here because the SMD from this study

was substantial and seemed to be driving heterogeneity in the

primary meta-analysis. Excluding the Kirkland data reduced

heterogeneity from I2 5 89.2% to 52.2%. Thus, 11 studies were

included in sensitivity analyses.The pooled SMD for depression was medium for studies with

certain (0.61,95%CI0.11-1.12, z52.39,P50.02; I2575.1%) and

uncertain pain chronicity (0.53, 95% CI 0.41-0.64, z 5 9.04,

Figure 4. Forest plot of cross-sectional events data for psychological distress. Distress was more likely in participants with vs without pain, as reflected in thepooled odds ratio (OR) of .1. CI, confidence interval.

Figure 5. Forest plot of cross-sectional standardizedmean differences (SMDs) for psychological distress. Distress wasmore severe in participants with vs withoutpain, as reflected by a positive pooled SMD. CI, confidence interval.

2468 W. Scott et al.·159 (2018) 2461–2476 PAIN®

28

P,0.001; I2534.5%). In the analysis by pain type, thepooledSMDfor depression wasmoderate in studies withmixed pain types (0.75,95% CI 0.58-0.92, z5 8.51, P, 0.001; I2 5 31.7%) and for whichpain type was not reported (0.52, 95% CI 0.39-0.65; z5 7.91, P,0.001; I2 5 0.0%). By contrast, studies with neuropathic pain (0.31,95% CI 0.11-0.52; z 5 2.96, P 5 0.003; I2 5 0.0%) or headache(0.38, 95% CI 0.06-0.70, z 5 2.34, P 5 0.02) showed small butsignificant differences between groups on depression.

The pooled SMD for depression was moderate for studies inwhich participants had less than adequate immune functioning andviral suppression (0.56, 95%CI 0.42-0.70, z5 7.77,P, 0.001; I250.0%) and for studies in which these indicatorswere uncertain (0.53,95%CI 0.33-0.73, z5 5.28,P, 0.001; I25 69.1%). There were nostudies with “adequate” functioning on these indices in this analysis.Studies from the pre-cART (0.49, 95% CI 0.32-0.66, z5 5.75, P,0.001; I2 5 0.0%), cART (0.73, 95% CI 0.37-1.10, z 5 3.91, P ,0.001), and current-cART era (0.55, 95% CI 0.38-0.72, z5 6.21, P, 0.001; I2 5 62.5%) all had moderate or near-moderate pooledSMDs. Finally, pooled SMDs for depression were similar in studiesfrom the United States, which has amixed health care system (0.57,95%CI 0.43-0.72, z5 7.58, P, 0.001; I25 55.7%), and one studyfrom the United Kingdom, which has universal health care (0.68,95%CI 0.09-1.28; z52.25,P50.03). ThepooledSMDof 2 studiesconducted in lower- and middle-income countries was smaller, butstatistically significant (0.43, 95%CI 0.14-0.72, z5 2.90,P5 0.004;I2 5 54.2%).

3.3. Psychological distress

Eighteen studies examined variables representing psychologicaldistress, including anxiety-related constructs and the presence of“mental illness,” which generally described a combination ofanxiety and depression. Five cross-sectional studies providedevents data (Fig. 4). The pooled ORwas significant and indicatedthat participants with pain were more likely to have psychologicaldistress than those without pain: OR5 2.56 (95% CI 1.67-3.90),

z 5 4.34, P , 0.001. Heterogeneity was high (I2 5 68.3%). Oneprospective study (n5 127) found that baseline mental illness didnot predict presence of pain over follow-up.54 Seven cross-sectional studies provided mean values and SDs (Fig. 5). Thepooled SMD showed a large and statistically significant differencebetween groups such that distress was worse in participants withvs without pain (SMD 5 0.85, 95% CI 0.35-1.35); z 5 3.33, P 50.001). Heterogeneity was very high (I2 5 95.4%). One furtherstudy that reported the median and interquartile range found nodifference between groups.88

Four cross-sectional studies reported correlations betweendistress and pain severity (supplemental Figure 9, available athttp://links.lww.com/PAIN/A643). The pooled correlation wasmoderate: Fisher z 5 0.35 (95% CI 0.09-0.60), z 5 2.68, P 50.007. Heterogeneity was high (82.2%). Three cross-sectionalstudies reported correlations between distress and paininterference/disability. The pooled correlation was moderate:Fisher z 5 0.59 (95% CI 0.24-0.93), z 5 3.34, P 5 0.001.Heterogeneity was high (81.2%). One prospective study (n5 45-62) found a nonsignificant correlation between change in distressand pain severity after CBT, and a significant, moderatecorrelation between change in distress and pain interference.44,86

Finally, one cross-sectional study reported a nonsignificantcorrelation between pain intensity and distress (r not reported),and a small negative correlation between distress and quality oflife.88

3.4. Posttraumatic stress

Three studies investigated posttraumatic stress. These studiesare reported separate from studies measuring psychologicaldistress, given the specificity of posttraumatic stress as a variable.Different study designs and analyses precluded meta-analysis.One prospective study (n5 143) found that posttraumatic stresssymptoms (PTSSs) were associated with significantly higher painseverity and interference over time in a sample with HIV and

Figure 6. Forest plot of cross-sectional events data for drug abuse. Drug abuse wasmore likely in participants with vs without pain, as reflected in the pooled oddsratio (OR) of .1. CI, confidence interval.


29

persistent pain.107 One high-quality cross-sectional study foundthat participants with pain (n5 170) had significantly higher PTSSthan those without pain (n 5 59).79 Within the pain group in thisstudy, there was a nonsignificant correlation between PTSSs andpain severity, and small but significant correlations betweenPTSSs and pain interference (positive correlation) and quality oflife (negative correlation).79 Posttraumatic stress disorder did notdiffer between groups with (n 5 150) and without (n 5 128)neuropathy in another cross-sectional study.31

3.5. Drug abuse

Fourteen studies examined drug abuse. We prioritised extractingopioid abuse data when multiple drug abuse categories werereported, given the relevance of opioid use in chronic pain. Twoprospective studies reported ORs for pain predicting heroin useat the time of follow-up. The pooledOR indicated that participantswith pain at baseline were more likely at follow-up to be usingheroin: OR 5 1.70 (95% CI 1.22-2.38), z 5 3.13, P 5 0.002(supplemental Figure 10, available at http://links.lww.com/PAIN/A643). Heterogeneity was low (I2 5 14.0%). Conversely, anotherprospective study (n 5 493) reported that baseline opioid usedisorder history predicted new onset of neuropathic pain, OR 52.87 (1.31-6.28), P , 0.01.60 One low-quality prospective study(n 5 127) found that baseline drug abuse history did not predictthe presence of pain at follow-up, 0.55 (0.25-1.21).54 These 2studies could not be combined due to different coding of thedependent variable.

Eight cross-sectional studies reported events data (Fig. 6). Thepooled OR was significant such that participants with pain weremore likely to have comorbid drug abuse than those without pain:OR5 1.59 (95%CI 1.12-2.26), z5 2.58,P5 0.01. Heterogeneitywas high (I25 69.8%), mainly attributable to one study that foundthe opposite effect, such that participants with symptomaticdistal sensory polyneuropathy were less likely to have opioid usedisorder than those with asymptomatic distal sensory polyneur-opathy.76 One low-quality cross-sectional study (n 5 503) foundthat participants with “untreated” pain had greater dependencesymptoms than those with “treated” pain or without pain.110

Another low-quality cross-sectional study (n 5 73) found a smallpositive correlation between “aberrant drug behaviours” and paininterference, but not quality of life.81

3.6. Alcohol abuse

Eleven studies investigated alcohol abuse. Two prospectivestudies reported ORs for baseline pain predicting subsequentalcohol abuse. The pooled OR was not significant: OR 5 0.94(95% CI 0.39-2.26), z5 0.13, P5 0.90 (supplemental Figure 11,available at http://links.lww.com/PAIN/A643). Heterogeneity washigh (84.1%). Two additional prospective studies examinedbaseline alcohol abuse as a predictor of developing pain/neuropathy but could not be combined due to different analyses.Both studies reported a nonsignificant association between thesevariables.60,77 Seven cross-sectional studies provided eventsdata or ORs. The pooled ORwas not significant: OR5 1.22 (95%CI 0.92-1.62), z 5 1.36, P 5 0.17 (supplemental Figure 12,available at http://links.lww.com/PAIN/A643). Heterogeneity wasmedium (I2 5 39.0%).

3.7. Sleep disturbance

Three studies investigated sleep disturbance. Two cross-sectional studies reported mean values and SDs. The pooled

effect was significant and showed moderately greater sleepproblems in participants with vs without pain: SMD 5 0.66 (95%CI 0.45-0.87), z 5 6.12, P , 0.001. Heterogeneity was 0.0%(supplemental Figure 13, available at http://links.lww.com/PAIN/A643). Another cross-sectional (n 5 45) study reported a signif-icant correlation between pain severity and sleep disturbance,and a nonsignificant correlation between sleep and functioning.94

3.8. Antiretroviral nonadherence

Seven studies investigated the association between pain andsuboptimal ART adherence. Data were analysed separatelyaccording to whether the adherence variable was coded in thedirection of nonadherence or adherence. One prospective study(n5 258) reported that severe pain at baseline predicted higherodds (OR 5 1.37, 95% CI 1.02-1.85) of follow-up ARTnonadherence.46 One cross-sectional study provided eventsdata, whereas another provided an OR. The pooled OR wassignificant and indicated that participants with pain were morelikely to report nonadherence: OR 5 1.40 (95% CI 1.07-1.82), z5 2.50, P 5 0.01 (supplemental Figure 14, available at http://links.lww.com/PAIN/A643). Heterogeneity was 0.00%. Onecross-sectional study (n 5 42) found significant positivecorrelations between pain severity and adherence forgetfulnessand fears.57

Two cross-sectional studies reported data for the associationbetween pain and adherence (events data or OR). The pooledOR was less than one, indicating the likelihood of adherencewas lower in participants with pain, but this was not statisticallysignificant: OR 5 0.32 (95% CI 0.08-1.32), z 5 1.57, P 5 0.12(supplemental Figure 15, available at http://links.lww.com/PAIN/A643). Heterogeneity was high (74.3%). Finally, one low-quality cross-sectional study (n5 377) found that pain presencewas not associated with adherence in a structural equationmodel.72

3.9. Health care use

Six studies examined health care use. One prospective study (n51521) found that baseline pain predicted significantly higher odds(OR 5 1.6, 95% CI 1.2-2.0) of urgent care visits.68 Two cross-sectional studies reported events data. The pooled OR was notsignificant: OR 5 0.98 (95% CI 0.58-1.66, z 5 0.07, P 5 0.94)(supplemental Figure 16, available at http://links.lww.com/PAIN/A643). Heterogeneity was 0.0%. Two further cross-sectionalstudies reported mean values and SDs. The pooled effect wassmall but significant, such that participants with pain had greaterhealth care use than those without pain: SMD 5 0.36 (95% CI0.21-0.51, z 5 4.66, P , 0.001). Heterogeneity was 0.0%(supplemental Figure 17, available at http://links.lww.com/PAIN/A643). One further cross-sectional study (n 5 1120) found thatparticipants with pain and daily opioid use had more clinic visitsthan those with pain without daily opioid use and those withoutpain (SD not reported).53

3.10. Missed HIV clinic visits

Two prospective studies reportedORs for baseline presence of painpredicting missed HIV clinic visits over 1-year follow-up. The pooledORwas significant, such that those with pain at baseline had higherodds of a missed HIV clinic visit: OR5 1.42 (95%CI 1.13-1.79), z52.98, P 5 0.003 (supplemental Figure 18, available at http://links.lww.com/PAIN/A643). Heterogeneity was 0.0%.

2470 W. Scott et al.·159 (2018) 2461–2476 PAIN®

30

3.11. Unemployment

Seven cross-sectional studies provided events data or ORs forthe association between unemployment and pain. The pooledOR was significant and indicated that participants with pain hadhigher odds of being unemployed than those without pain: OR52.09 (95% CI 1.59-2.76, z 5 5.25, P , 0.001) (supplementalFigure 19, available at http://links.lww.com/PAIN/A643). Hetero-geneity wasmoderate (48.6%). One further cross-sectional study(n 5 229) that did not have data available for meta-analysislikewise found that participants with pain were significantly morelikely to be unemployed than those without pain.79

3.12. Protective factors

Five studies examined protective psychological factors. Oneprospective study (n 5 62) found significant small and mediumcorrelations between change in self-reported pain acceptanceduring CBT and posttreatment pain severity and interference,respectively.86 One cross-sectional case-control study observedlower resilience in participants with (n 5 99) vs without pain (n 598; medium effect); however, this study found nonsignificantcorrelations between resilience and pain severity and interferencein the pain group.121

One high-quality cross-sectional study found that participantswith pain (n 5 170) reported lower disease management self-efficacy than did those without pain (n 5 59) (small effect).79

Within the pain group in this study, there were nonsignificantcorrelations between self-efficacy and pain severity and in-terference, and a small positive correlation between self-efficacyand quality of life.79 One low-quality cross-sectional study foundthat those with greater adherence self-efficacy were less likely toreport pain (n 5 70).8 Finally, one low-quality cross-sectionalstudy found lower mean self-reported optimism in participantswith (n 5 50) vs without pain (n 5 46) (small effect).101

3.13. Social factors

Four studies investigated social factors. The BEACON study (n5377) explored social processes across 3 papers, 2 of whichdescribe prospective data (medium quality), whereas the thirdreported cross-sectional data (low quality). Baseline chronic painpredicted “negative social support” (ie, overly intrusive orinsensitive responses from others and a lack of support) at 12months, controlling for baseline social support.73 Anotherprospective analysis showed that no chronic pain at baselinepredicted greater support reciprocity at follow-up.74 Chronic painwas associated with significantly poorer ratings of patient–provider engagement in cross-sectional analyses.72

Two studies examined self-reported stigma, but could not becombined. One medium-quality cross-sectional study (n 5 50)found a moderate positive correlation between stigma and painseverity.122 One low-quality cross-sectional study (n 5 201)found that participants with “pain disorder” reported higherstigma scores than those without “pain disorder.”98 Onemedium-quality cross-sectional study found no difference inmean number or quality of self-reported social supports betweenparticipants with (n 5 274) and without pain (n 5 164).91

4. Discussion

This review including over 13,000 participants found “some” or“moderate” evidence supporting an association between painoutcomes and depression, psychological distress, posttraumatic

stress, drug abuse, sleep disturbance, health care use, missedHIV clinic visits, ART adherence, unemployment, and protectivepsychological factors in people with HIV. Surprisingly few studieshave examined protective psychological factors or social pro-cesses. There is a lack of high-quality research on psychosocialfactors related to chronic pain in people with HIV. These findingscan inform future research and treatment development in thisarea.

The association between depression and poorer self-reportedpain outcomes in HIV is consistent with the wider painliterature.4,64 Data from prospective studies suggest depressionis a risk factor for pain. However, caution is warranted in thisinterpretation, given the observational nature of studies. There islikely a bidirectional relationship, with shared neurobiologicalpathways, cognitive appraisals, and behavioural disengagementunderpinning this association.4,7,14 Evidence supporting theassociation between pain and sleep disturbance is consistentwith the wider literature that reports reciprocal associationsbetween pain, sleep, and depression.106

There was substantial variability in the assessment of “psycho-logical distress,” which may have contributed to the statisticalheterogeneity observed. Although different measures were used toassess variables such as pain catastrophizing, pain-related fear,stress, and general anxiety, these measures overlap conceptuallyand in item content. The consistency of results within thepsychological distress category suggests the findings are robustacross different assessment methods. Several studies assessed“mental illness” based on a range of diagnoses in participants’medical filewithout clear diagnostic criteria. Studies exploringmentalhealth diagnoses should use valid and reliable criteria and, ideally,semi-structured clinician-administered interviews as the goldstandard.128 In light of high rates of posttraumatic stress disorder(PTSD) inHIV,99 further research isparticularly needed tounderstandthe role of PTSD in pain in this context. Alternately, rather thanfocusing on specificmental health diagnoses, research investigatingpsychosocial processes that explain the impact of a range ofpsychological difficulties may prove useful moving forward.42

Few studies investigated fear-avoidancemodel variables, suchas pain catastrophizing and pain-related fear, which havedominated the musculoskeletal pain literature. Fear of movementis strongly associated with musculoskeletal pain disability.20

However, neuropathic pain is often spontaneous and not clearlyprovoked by movement, although it may inhibit movement.Therefore, research is needed to determine the relevance of fear-avoidance model constructs in neuropathic pain that is commonin HIV.

A bidirectional association between drug abuse and pain in HIVis suggested by prospective data showing opioid abuse as bothpredictor60 and outcome114 of pain. In a population where thereare concerns about analgesic prescribing,58 poorly managedpain may contribute to increased abuse of nonprescribedopioids, which may be exacerbated by depression.114 Alter-nately, prolonged opioid abuse may disrupt descending paininhibition, exacerbating pain.49 Differing definitions of drug andalcohol abuse across studies may account for variability ineffects. Future research on substance abuse in this contextshould use validated assessments, either screening tools ordiagnostic interviews that capture key features of abuse, such ascontinued use despite harm.1,41,128

Adherence to ART and retention in care are psychosocialfactors unique to the HIV context and are of vital importance,given their associations with mortality, morbidity, and drugresistance.112 The finding that pain was associated with reducedART adherence and missed HIV clinic visits highlights the


31

necessity of adequate pain management in HIV. Understandingthe links between pain, ART adherence, and retention in carelikely requires consideration of other psychosocial factors, suchas substance abuse and depression, which may mediate ormoderate this association.68,92

Findings that pain was associated with greater unemploymentand health care use highlight the individual, societal, andeconomic costs of pain in HIV. This is consistent with the broaderliterature, although health care use is typically underassessed intrials of psychotherapy for chronic pain.87 Studies assessinghealth care use were restricted to the United States, whereas theunemployment–pain link was consistent in studies from theUnited States, Russia, and South Africa. The associationbetween pain and health care use differed across studies onthe basis of the type of health care assessed. Assessment of themost frequently accessed services (eg, general practitioner visits),rather than relatively infrequent events (eg, hospitalisations), mayincrease the interpretability of future health care data.

Surprisingly, only 5 studies assessed protective psychologicalfactors. The lack of studies on protective factors mirrors historicaltrends in the general pain literature, although there has beengreater focus on protective factors more recently. The focus on“maladaptive” responses to pain is problematic because suchresponses can be understood as a function of their short-termutility.126 Moreover, abnormal conceptualizations often fail tospecify pathways through which recovery and successfulfunctioning occur when pain is present. The psychologicalflexibility model, within which pain acceptance has beenconceptualised, might prove useful for future research.63

A recent proposal for updating the definition of pain highlightsthe central role of social factors.127 However, our review identifiedonly 4 studies exploring interpersonal variables. The lack ofresearch on stigma in relation to pain is particularly surprisingbecause managing stigma is key to the success of the HIV/AIDSresponse.108 Stigma has recently been highlighted as importantfor the well-being of patients with chronic pain in general.23,125

Future research is needed to determine the function of stigma inchronic pain in people with and without HIV.

The study samples included in our review variedwidely in termsof the proportion of men and women, participant age, ethnicity,and duration and severity of HIV. Our sensitivity analyses supportthe potential applicability of findings across pain types, ARTtreatment eras, and health care systems. Due to poor reporting ofviral loads and CD41 counts, our analysis stratifying by theseindicators is difficult to interpret. Caution is also warrantedregarding the cross-cultural applicability of the findings becausemost studies were from the United States. One South Africanstudy with a predominantly female sample found patients withand without pain did not differ on depression or anxiety, likely dueto high scores across the sample.88 Socioeconomic factors, suchas poverty and gender, may thus alter the relationships betweenpain, functioning, and mental health.88,121 Care is needed inapplying Western psychological concepts in non-Westerncultures.51,82,83 Research must also determine unique culturalfeatures that influence the experience and expression of painin HIV.

A guiding theoretical model is needed to integrate psychosocialprocesses relevant to pain and HIV. Such a model should makespecific predictions about the relative contributions of cognitive,affective, behavioural, and sociocultural processes in relation tospecific pain outcomes. This review identified a number of closelyrelated psychosocial constructs. Therefore, a theoretical model maybenefit from identifying a key set of nonoverlapping variables.63 This

may draw on prominent models within the field of pain, such as thefear-avoidance19 and psychological flexibility models,63 and thosewithin the HIV literature that focus heavily on socioculturalperspectives to understand the impact of processes, such asstigma, on well-being.78

The current findings suggest the relevance of psychosocialtreatments to manage persistent pain in HIV. To the best of ourknowledge, only 3 small RCTs have evaluated CBT andmindfulness-based treatment.29,35,118 Nonrandomized trials ofCBT113 and hypnosis25 have also been conducted. Furtherevaluation of psychosocial treatments for HIV and chronic pain isthus needed. The development of treatments that specificallytarget psychosocial factors identified in this review with “some” or“moderate” evidence may prove fruitful.

Several limitationswarrant consideration.We used a comprehen-sive search strategy that included efforts to identify gray literature tolimit publication bias; however, relevant studies may have beenmissed, given the broad nature of the search. We used an adaptedquality assessment tool. Although we based this on previouslyvalidated tools, the adaptations may have limited the reliability andvalidity of our quality assessment. Assessment of pain wasinadequate in many studies. Future research should assessinformation regarding pain duration, intensity, location, and type.Studies investigating chronic pain should specify eligibility criteria inline with recognized definitions: the presence of daily, clinicallymeaningful pain intensity and functional interference for at least 3months.10,26,131 Given the relevance of neuropathic pain in thispopulation, the use of well-validated screening tools of neuropathysigns and symptoms is important.16,33,129

This review identified a large number of psychosocial factors.As evidence on specific psychosocial factors develops in thisarea, it may be useful for a future review to use a more targetedapproach to synthesize data on a smaller number of prespecifiedvariables. We focused on bivariate analyses and dichotomizedmultiple between-groups analyses to facilitate comparisonacross studies and minimise pairwise comparisons. However,this may have limited an in-depth understanding of psychosocialfactors from multivariate models and more subtle subgroupanalyses. Future research examining the association betweenpsychosocial factors and pain outcomes in HIV should considercontrolling for such variables as age, sex/gender, race/ethnicity,socioeconomic status, HIV duration, current and nadir CD41

count and current and peak viral load, current and past ARTregimens, and other medical comorbidities (eg, hepatitis C,diabetes, and tuberculosis). Where multiple psychosocial varia-bles are included, sufficient rationale for each variable should beprovided and care should be taken to minimize overlap inassessment content between variables.

Despite these limitations, this is the first systematic review tospecifically explore psychosocial variables associated withpersistent pain in HIV. From this review, it is recommended thatresearchers (1) focus greater attention on protective psycholog-ical factors and social processes, such as stigma and processesto undermine stigma; (2) use higher-quality assessment tools;and (3) develop and test treatments to target key psychosocialfactors to improve pain outcomes in HIV. Improving quality of lifeis a priority as people with HIV live longer. Adequate, whole-person pain management is vital to achieve this goal.


The authors have no conflict of interest to declare.

2472 W. Scott et al.·159 (2018) 2461–2476 PAIN®

32

This research is an independent work supported by theNational Institute for Health Research (NIHR PostdoctoralFellowship, W. Scott, PDF-2015-08-059). L.M. McCracken ispartly funded through the NIHR Biomedical Research Centre atSouth London and Maudsley NHS Foundation Trust and King’sCollege London. The views expressed in this publication are thoseof the authors and not necessarily those of the NHS, the NationalInstitute for Health Research, or the Department of Health. Dr.Kemp reports grants from the European Commission (FP7Neuropain #HEALTH F2-2013-602891) during the conduct ofthe study. Dr. Rice reports grants from Orion Pharma, other fromSpinifex/Novartis, personal fees from Imperial CollegeConsultants,outside the submitted work. In addition, Dr. Rice has a patent, RiceA.S.C., Vandevoorde S. and Lambert D.M Methods using N-(2-propenyl)hexadecanamide and related amides to relieve pain. WO2005/079771 pending, and a patent, Okuse K. et al., Methods oftreating pain by inhibition of vgf activity EP13702262.0/ WO2013110945, pending.

Appendix A. Supplemental digital content

Supplemental digital content associated with this article can befound online at http://links.lww.com/PAIN/A643.

Supplemental video content

Video content associated with this article can be found at http://links.lww.com/PAIN/A645.

Article history:Received 27 March 2018Accepted 26 June 2018Available online 16 August 2018

References

[1] Ali R, Meena S, Eastwood B, Richards I, Marsden J. Ultra-rapidscreening for substance-use disorders: the Alcohol, Smoking andSubstance Involvement Screening Test (ASSIST-Lite). Drug AlcoholDepend 2013;132:352–61.

[2] Aouizerat BE, Miaskowski CA, Gay C, Portillo CJ, Coggins T, Davis H,Pullinger CR, Lee KA. Risk factors and symptoms associatedwith pain inHIV-infected adults. J Assoc Nurses AIDS Care 2010;21:125–33.

[3] Ariens GA, Mechelen Wv, Bongers PM, Bouter LM, van der Wal G.Psychosocial risk factors for neck pain: a systematic review. Am J IndMed 2001;39:180–93.

[4] Bair MJ, Robinson RL, Katon W, Kroenke K. Depression and paincomorbidity—a literature review. Arch Intern Med 2003;163:2433–45.

[5] Bakka JC. The relationships of physical functioning, psychologicaldistress, and negative thoughts to pain in people with AIDS [doctoralthesis]. Ann Arbor: UMI Company, 1995.

[6] Banerjee S, McCutchan JA, Ances BM, Deutsch R, Riggs PK, Way L,Ellis RJ. Hypertriglyceridemia in combination antiretroviral-treated HIV-positive individuals: potential impact on HIV sensory polyneuropathy.AIDS 2011;25:F1–6.

[7] Banks SM, Kerns RD. Explaining high rates of depression in chronicpain: a diathesis-stress framework. Psychol Bull 1996;119:95–110.

[8] Berg KM, Cooperman NA, Newville H, Arnsten JH. Self-efficacy anddepression as mediators of the relationship between pain andantiretroviral adherence. AIDS Care 2009;21:244–8.

[9] Bor J, Herbst AJ, Newell ML, Barnighausen T. Increases in adult lifeexpectancy in rural South Africa: valuing the scale-up of HIV treatment.Science 2013;339:961–5.

[10] Bouhassira D, Lanteri-Minet M, Attal N, Laurent B, Touboul C.Prevalence of chronic pain with neuropathic characteristics in thegeneral population. PAIN 2008;136:380–7.

[11] Breitbart W, McDonald MV, Rosenfeld B, Passik SD, Hewitt D, Thaler H,Portenoy RK. Pain in ambulatory AIDS patients. I: pain characteristicsand medical correlates. PAIN 1996;68:315–21.

[12] Breitbart W, Passik S, McDonald MV, Rosenfeld B, Smith M, Kaim M,Funesti-Esch J. Patient-related barriers to pain management inambulatory AIDS patients. PAIN 1998;76:9–16.

[13] Breitbart W, Rosenfeld B, Passik S, Kaim M, Funesti-Esch J, Stein K. Acomparison of pain report and adequacy of analgesic therapy inambulatory AIDS patients with andwithout a history of substance abuse.PAIN 1997;72:235–43.

[14] Campbell LC, Clauw DJ, Keefe FJ. Persistent pain and depression:a biopsychosocial perspective. Biol Psychiatry 2003;54:399–409.

[15] Cherry CL, Wadley AL, Kamerman PR. Painful HIV-associated sensoryneuropathy. Pain Manage 2012;2:543–52.

[16] Cherry CL, Wesselingh SL, Lal L, McArthur JC. Evaluation of a clinicalscreening tool for HIV-associated sensory neuropathies. Neurology2005;65:1778–81.

[17] Clifford DB, Simpson DM, Brown S, Moyle G, Brew BJ, Conway B,Tobias JK, VanhoveGF. A randomized, double-blind, controlled study ofNGX-4010, a capsaicin 8% dermal patch, for the treatment of painfulHIV-associated distal sensory polyneuropathy. J Acquir Immune DeficSyndr 2012;59:126–33.

[18] Corey DM, Dunlap WP, Burke MJ. Averaging correlations: expectedvalues and bias in combined Pearson rs and Fisher’s z transformations.J Gen Psychol 1998;125:245–61.

[19] Crombez G, Eccleston C, Van Damme S, Vlaeyen JWS, Karoly P. Fear-avoidance model of chronic pain: the next generation. Clin J Pain 2012;28:475–83.

[20] Crombez G, Vlaeyen JWS, Heuts PHTG, Lysens R. Pain-related fear ismore disabling than pain itself: evidence on the role of pain-related fear inchronic back pain disability. PAIN 1999;80:329–39.

[21] Cucciare MA, Sorrell JT, Trafton JA. Predicting response to cognitive-behavioral therapy in a sample of HIV-positive patients with chronic pain.J Behav Med 2009;32:340–8.

[22] Davis L, Evans S, Fishman B, Haley A, Spielman L. Predictors of attritionin HIV-positive subjects with peripheral neuropathic pain. AIDS Care2004;16:395–402.

[23] De Ruddere L, Craig KD. Understanding stigma and chronic pain:a state of the art review. PAIN 2016;157:1607–10.

[24] Dinat N, Marinda E, Moch S, Rice AS, Kamerman PR. Randomized,double-blind, crossover trial of amitriptyline for analgesia in painful HIV-associated sensory neuropathy. PLoS One 2015;10:e0126297.

[25] Dorfman D, George MC, Schnur J, Simpson DM, Davidson G,Montgomery G. Hypnosis for treatment of HIV neuropathic pain:a preliminary report. Pain Med 2013;14:1048–56.

[26] Dworkin RH, Turk DC, Farrar JT, Haythornthwaite JA, Jensen MP, KatzNP, Kerns RD, Stucki G, Allen RR, Bellamy N. Core outcome measuresfor chronic pain clinical trials: IMMPACT recommendations. PAIN 2005;113:9–19.

[27] Ellis RJ, Rosario D, Clifford DB, McArthur JC, Simpson D, Alexander T,Gelman BB, Vaida F, Collier A, Marra CM. Continued high prevalenceand adverse clinical impact of human immunodeficiencyvirus–associated sensory neuropathy in the era of combinationantiretroviral therapy: the CHARTER Study. Arch Neurol 2010;67:552–8.

[28] Evans S, Ferrando S, Sewell M, Goggin K, Fishman B, Rabkin J. Painand depression in HIV illness. Psychosomatics 1998;39:528–35.

[29] Evans S, Fishman B, Spielman L, Haley A. Randomized trial of cognitivebehavior therapy versus supportive psychotherapy for HIV-relatedperipheral neuropathic pain. Psychosomatics 2003;44:44–50.

[30] Evans S, Weinberg BA, Spielman L, Fishman B. Assessing negativethoughts in response to pain among people with HIV. PAIN 2003;105:239–45.

[31] Fellows RP, Byrd DA, Elliott K, Robinson-Papp J, Mindt MR, Morgello S.Distal sensory polyneuropathy is associated with neuropsychologicaltest performance among persons with HIV. J Int Neuropsychol Soc2012;18:898–907.

[32] Finnerup NB, Attal N, Haroutounian S,McNicol E, Baron R, Dworkin RH,Gilron I, Haanpaa M, Hansson P, Jensen TS. Pharmacotherapy forneuropathic pain in adults: a systematic review and meta-analysis.Lancet Neurol 2015;14:162–73.

[33] Finnerup NB, Haroutounian S, Kamerman P, Baron R, Bennett DL,Bouhassira D, Cruccu G, Freeman R, Hansson P, Nurmikko T.Neuropathic pain: an updated grading system for research and clinicalpractice. PAIN 2016;157:1599.

[34] Gatchel RJ, Peng YB, Peters ML, Fuchs PN, Turk DC. Thebiopsychosocial approach to chronic pain: scientific advances andfuture directions. Psychol Bull 2007;133:581.

[35] George MC, Wongmek A, Kaku M, Nmashie A, Robinson-Papp J. Amixed-methods pilot study of mindfulness-based stress reduction forHIV-associated chronic pain. Behav Med 2017;43:108–19.


33

[36] Golin C, Isasi F, Bontempi JB, Eng E. Secret pills: HIV-positive patients’experiences taking antiretroviral therapy in North Carolina. AIDS EducPrev 2002;14:318–29.

[37] Gøtzsche PC. Why we need a broad perspective on meta-analysis: itmay be crucially important for patients. BMJ 2000;321:585.

[38] Griswold GA, Evans S, Spielman L, Fishman B. Coping strategies of HIVpatients with peripheral neuropathy. AIDS Care 2005;17:711–20.

[39] Hansen L, Penko J, Guzman D, Bangsberg DR, Miaskowski C, KushelMB. Aberrant behaviors with prescription opioids and problem drug usehistory in a community-based cohort of HIV-infected individuals. J PainSymptom Manage 2011;42:893–902.

[40] Harrison AM, McCracken LM, Bogosian A, Moss-Morris R. Towardsa better understanding of MS pain: a systematic review of potentiallymodifiable psychosocial factors. J Psychosom Res 2015;78:12–24.

[41] Hasin DS, O’Brien CP, Auriacombe M, Borges G, Bucholz K, Budney A,Compton WM, Crowley T, Ling W, Petry NM. DSM-5 criteria forsubstance use disorders: recommendations and rationale. Am JPsychiatry 2013;170:834–51.

[42] Hayes SC, Barnes-Holmes D, Wilson KG. Contextual behavioralscience: creating a science more adequate to the challenge of thehuman condition. J Context Behav Sci 2012;1:1–16.

[43] Higgins JP, Green S. Cochrane handbook for systematic reviews ofinterventions. Chichester: Wiley-Blackwell, 2011.

[44] Huggins JL, Bonn-Miller MO, Oser ML, Sorrell JT, Trafton JA. Painanxiety, acceptance, and outcomes among individuals with HIV andchronic pain: a preliminary investigation. Behav Res Ther 2012;50:72–8.

[45] Ioannidis JP, Patsopoulos NA, Rothstein HR. Research methodology:reasons or excuses for avoidingmeta-analysis in forest plots. BMJ 2008;336:1413.

[46] Jeevanjee S, Penko J, Guzman D, Miaskowski C, Bangsberg DR,Kushel MB. Opioid analgesic misuse is associated with incompleteantiretroviral adherence in a cohort of HIV-infected indigent adults in SanFrancisco. AIDS Behav 2014;18:1352–8.

[47] Jiao JM, So E, Jebakumar J, GeorgeMC, Simpson DM, Robinson-PappJ. Chronic pain disorders in HIV primary care: clinical characteristics andassociation with healthcare utilization. PAIN 2016;157:931–7.

[48] Keltner JR, Vaida F, Ellis RJ, Moeller-Bertram T, Fitzsimmons C, DuarteNA, Robinson-Papp J, Dworkin RH, Clifford DB, McArthur JC, SimpsonDM, Collier AC, Marra CM, Atkinson J, Grant I. Health-related quality oflife “well-being” in HIV distal neuropathic pain is more stronglyassociated with depression severity than with pain intensity.Psychosomatics 2012;53:380–6.

[49] King T, Ossipov MH, Vanderah TW, Porreca F, Lai J. Is paradoxical paininduced by sustained opioid exposure an underlying mechanism ofopioid antinociceptive tolerance? Neurosignals 2005;14:194–205.

[50] Kirkland KE. Psychological factors in HIV-related headaches [doctoralthesis]. Ann Arbor: UMI Company, 2012.

[51] Kirmayer LJ. Beyond the “new cross-cultural psychiatry”: culturalbiology, discursive psychology and the ironies of globalization.Transcult Psychiatry 2006;43:126–44.

[52] Knowlton AR, Nguyen TQ, Robinson AC, Harrell PT, Mitchell MM. Painsymptoms associated with opioid use among vulnerable persons withHIV: an exploratory study with implications for palliative care and opioidabuse prevention. J Palliat Care 2015;31:228–33.

[53] Koeppe J, Armon C, Lyda K, Nielsen C, Johnson S. Ongoing paindespite aggressive opioid pain management among persons with HIV.Clin J Pain 2010;26:190–8.

[54] Koeppe J, Lyda K, Johnson S, Armon C. Variables associated withdecreasing pain among persons living with human immunodeficiencyvirus: a longitudinal follow-up study. Clin J Pain 2012;28:32–8.

[55] Lagana L, Chen X, Koopman C, Classen C, Kimerling R, Spiegel D.Depressive symptomatology in relation to emotional control and chronicpain in persons who are HIV positive. Rehab Psychol 2002;47:402–14.

[56] Lopez O, Becker J, Dew MA, Caldararo R. Risk modifiers for peripheralsensory neuropathy in HIV infection/AIDS. Eur J Neurol 2004;11:97–102.

[57] Lucey BP, Clifford DB, Creighton J, Edwards RR, McArthur JC,Haythornthwaite J. Relationship of depression and catastrophizing topain, disability, and medication adherence in patients with HIV-associated sensory neuropathy. AIDS Care 2011;23:921–8.

[58] Lum PJ, Little S, Botsko M, Hersh D, Thawley RE, Egan JE, Mitty J,Boverman J, Fiellin DA, Collaborative B. Opioid-prescribingpractices and provider confidence recognizing opioid analgesicabuse in HIV primary care settings. J Acquir Immune Defic Syndr2011;56:S91–7.

[59] Mahungu TW, Rodger AJ, Johnson MA. HIV as a chronic disease. ClinMed (Lond) 2009;9:125–8.

[60] Malvar J, Vaida F, Sanders CF, Atkinson JH, Bohannon W, Keltner J,Robinson-Papp J, Simpson DM, Marra CM, Clifford DB, Gelman B, FanJ, Grant I, Ellis RJ, Group C. Predictors of new-onset distal neuropathicpain in HIV-infected individuals in the era of combination antiretroviraltherapy. PAIN 2015;156:731–9.

[61] Mann R, Sadosky A, Schaefer C, Baik R, Parsons B, Nieshoff E, StaceyBR, Tuchman M, Nalamachu S. Burden of HIV-related neuropathicpain in the United States. J Int Assoc Provid AIDS Care 2016;15:114–25.

[62] McAuley L, Tugwell P, Moher D. Does the inclusion of grey literatureinfluence estimates of intervention effectiveness reported in meta-analyses? Lancet 2000;356:1228–31.

[63] McCracken LM, Morley S. The psychological flexibility model: a basis forintegration and progress in psychological approaches to chronic painmanagement. J Pain 2014;15:221–34.

[64] McWilliams LA, Cox BJ, Enns MW. Mood and anxiety disordersassociated with chronic pain: an examination in a nationallyrepresentative sample. PAIN 2003;106:127–33.

[65] Merlin JS, Cen L, Praestgaard A, Turner M, Obando A, Alpert C,Woolston S, Casarett D, Kostman J, Gross R, Frank I. Pain and physicaland psychological symptoms in ambulatory HIV patients in the currenttreatment era. J Pain Symptom Manage 2012;43:638–45.

[66] Merlin JS, Westfall AO, Chamot E, Saag M, Walcott M, Ritchie C,Kertesz S. Quantitative evaluation of an instrument to identify chronicpain in HIV-infected individuals. AIDS Res Hum Retroviruses 2015;31:623–7.

[67] Merlin JS, Westfall AO, Heath SL, Goodin BR, Stewart JC, Sorge RE,Younger J. Brief report: IL-1beta levels are associated with chronicmultisite pain in people living with HIV. J Acquir Immune Defic Syndr2017;75:e99–e103.

[68] Merlin JS, Westfall AO, Raper JL, Zinski A, Norton WE, Willig JH, GrossR, Ritchie CS, Saag MS, Mugavero MJ. Pain, mood, and substanceabuse in HIV: implications for clinic visit utilization, ART adherence, andvirologic failure. J Acquir Immune Defic Syndr 2012;61:164.

[69] Merlin JS, Zinski A, Norton WE, Ritchie CS, Saag MS, Mugavero MJ,Treisman G, Hooten WM. A conceptual framework for understandingchronic pain in patients with HIV. Pain Pract 2014;14:207–16.

[70] Miaskowski C, Penko JM, Guzman D, Mattson JE, Bangsberg DR,Kushel MB. Occurrence and characteristics of chronic pain ina community-based cohort of indigent adults living with HIV infection.J Pain 2011;12:1004–16.

[71] Miles MS, Isler MR, Banks BB, Sengupta S, Corbie-Smith G. Silentendurance and profound loneliness: socioemotional suffering in AfricanAmericans living with HIV in the rural south. Qual Health Res 2011;21:489–501.

[72] Mitchell M, Nguyen T, Maragh-Bass A, Isenberg S, Beach M, KnowltonA. Patient-provider engagement and chronic pain in drug-using,primarily African American persons living with HIV/AIDS. AIDS Behav2017;21:1768–74.

[73] Mitchell MM, Maragh-Bass AC, Nguyen TQ, Isenberg S, Knowlton AR.The role of chronic pain and current substance use in predicting negativesocial support among disadvantaged persons living with HIV/AIDS.AIDS Care 2016;28:1280–6.

[74] Mitchell MM, Isenberg SR, Maragh-Bass AC, Knowlton AR. Chronicpain predicting reciprocity of support among vulnerable, predominantlyAfrican-American persons living with HIV/AIDS. AIDS Behav 2018;22:2002–7.

[75] Moher D, Liberati A, Tetzlaff J, Altman DG; Group P. Preferred reportingitems for systematic reviews and meta-analyses: the PRISMAstatement. PLoS Med 2009;6:e1000097.

[76] Morgello S, Estanislao L, Simpson D, Geraci A, DiRocco A, Gerits P,Ryan E, Yakoushina T, Khan S, Mahboob R, Naseer M, Dorfman D,Sharp V. HIV-associated distal sensory polyneuropathy in the era ofhighly active antiretroviral therapy: the Manhattan HIV Brain Bank. ArchNeurol 2004;61:546–51.

[77] Nakamoto BK, McMurtray A, Davis J, Valcour V, Watters MR, ShiramizuB, Chow DC, Kallianpur K, Shikuma CM. Incident neuropathy in HIV-infected patients on HAART. AIDS Res Hum Retroviruses 2010;26:759–65.

[78] Parker R, Aggleton P. HIV and AIDS-related stigma and discrimination:a conceptual framework and implications for action. Soc Sci Med 2003;57:13–24.

[79] Parker R, Jelsma J, Stein DJ. Pain in amaXhosa women living with HIV/AIDS: a cross-sectional study of ambulant outpatients. BMC WomensHealth 2017;17:126–32.

[80] Parker R, Stein DJ, Jelsma J. Pain in people living with HIV/AIDS:a systematic review. J Int AIDS Soc 2014;17:18719.

2474 W. Scott et al.·159 (2018) 2461–2476 PAIN®

34

[81] Passik SD, Kirsh KL, Donaghy KB, Portenoy RK. Pain and aberrantdrug-related behaviors in medically ill patients with and without historiesof substance abuse. Clin J Pain 2006;22:173–81.

[82] Patel V. Cultural factors and international epidemiology: depression andpublic health. Br Med Bull 2001;57:33–45.

[83] Patel V, Abas M, Broadhead J, Todd C, Reeler A. Depression indeveloping countries: lessons from Zimbabwe. BMJ 2001;322:482–4.

[84] Phillips TJ, Brown M, Ramirez JD, Perkins J, Woldeamanuel YW,Williams ACdC, Orengo C, Bennett DL, Bodi I, Cox S, Maier C, KrumovaEK, Rice AS. Sensory, psychological, and metabolic dysfunction in HIV-associated peripheral neuropathy: a cross-sectional deep profilingstudy. PAIN 2014;155:1846–60.

[85] Phillips TJ, Cherry CL, Cox S, Marshall SJ, Rice AS. Pharmacologicaltreatment of painful HIV-associated sensory neuropathy: a systematicreview and meta-analysis of randomised controlled trials. PLoS One2010;5:e14433.

[86] Pierson HM. The role of acceptance in cognitive behavioral treatment forchronic pain in an HIV-positive community sample [doctoral thesis]. AnnArbor: UMI Company, 2009.

[87] Pike A, Hearn L, Williams ACDC. Effectiveness of psychologicalinterventions for chronic pain on health care use and work absence:systematic review and meta-analysis. PAIN 2016;157:777–85.

[88] Pillay P, Wadley AL, Cherry CL, Karstaedt AS, Kamerman PR.Psychological factors associated with painful versus non-painful HIV-associated sensory neuropathy. AIDS Behav 2018;22:1584–95.

[89] Robbins NM, Chaiklang K, Supparatpinyo K. Undertreatment of pain inHIV plus adults in Thailand. J Pain SymptomManage 2013;45:1061–72.

[90] Robbins NM, Chaiklang K, Supparatpinyo K. Better antiretroviral centralnervous system penetration is not associated with reduced chronic painin people living with human immunodeficiency virus. Antiinfect Agents2016;14:38–46.

[91] Rosenfeld B, Breitbart W, McDonald MV, Passik SD, Thaler H, PortenoyRK. Pain in ambulatory AIDS patients. II: impact of pain on psychologicalfunctioning and quality of life. PAIN 1996;68:323–8.

[92] Safo S, Blank AE, Cunningham C, Quinlivan EB, Lincoln T, BlackstockOJ. Pain is associated with missed clinic visits among HIV-positivewomen. AIDS Behav 2017;21:1782–90.

[93] Samji H, Cescon A, Hogg RS, Modur SP, Althoff KN, Buchacz K,Burchell AN, Cohen M, Gebo KA, Gill MJ. Closing the gap: increases inlife expectancy among treated HIV-positive individuals in the UnitedStates and Canada. PLoS One 2013;8:e81355.

[94] Sandoval R, Roddey T, Giordano TP, Mitchell K, Kelley C. Pain, sleepdisturbances, and functional limitations in people living with HIV/AIDS-associated distal sensory peripheral neuropathy. J Int Assoc ProvidAIDS Care 2014;13:328–34.

[95] Saylor D, Nakigozi G, Nakasujja N, Robertson K, Gray RH, Wawer MJ,Sacktor N. Peripheral neuropathy in HIV-infected and uninfectedpatients in Rakai, Uganda. Neurology 2017;89:485–91.

[96] Schifitto G, McDermott M, McArthur J, Marder K, Sacktor N, Epstein L,Kieburtz K. Incidence of and risk factors for HIV-associated distalsensory polyneuropathy. Neurology 2002;58:1764–8.

[97] Schifitto G, McDermott M, McArthur J, Marder K, Sacktor N, McClernonD, Conant K, Cohen B, Epstein L, Kieburtz K. Markers of immuneactivation and viral load in HIV-associated sensory neuropathy.Neurology 2005;64:842–8.

[98] Shacham E, Rosenburg N, Onen NF, Donovan MF, Overton ET.Persistent HIV-related stigma among an outpatient US clinic population.Int J STD AIDS 2015;26:243–50.

[99] Sherr L, Nagra N, Kulubya G, Catalan J, Clucas C, Harding R. HIVinfection associated post-traumatic stress disorder and post-traumaticgrowth–a systematic review. Psychol Health Med 2011;16:612–29.

[100] Silver NC, Dunlap WP. Averaging correlation coefficients: shouldFisher’s z transformation be used? J App Psychol 1987;72:146.

[101] Simmonds MJ, Novy D, Sandoval R. The differential influence of painand fatigue on physical performance and health status in ambulatorypatients with human immunodeficiency virus. Clin J Pain 2005;21:200–6.

[102] Simms RW, Zerbini CAF, Ferrante N, Anthony J, Felson DT, Craven DE.Fibromyalgia syndrome in patients infected with humanimmunodeficiency virus. Am J Med 1992;92:368–74.

[103] SimpsonDM,Rice AS, Emir B, Landen J, Semel D, ChewML, Sporn J. Arandomized, double-blind, placebo-controlled trial and open-labelextension study to evaluate the efficacy and safety of pregabalin in thetreatment of neuropathic pain associated with humanimmunodeficiency virus neuropathy. PAIN 2014;155:1943–54.

[104] Singer EJ, Kim J, Fahy-Chandon B, Datt A, Tourtellotte W. Headache inambulatory HIV-1-infected men enrolled in a longitudinal study.Neurology 1996;47:487–94.

[105] Singer EJ, Zorilla C, Fahy-Chandon B, Chi S, Syndulko K, TourtellotteWW. Painful symptoms reported by ambulatory HIV-infected men ina longitudinal study. PAIN 1993;54:15–19.

[106] Smith MT, Haythornthwaite JA. How do sleep disturbance and chronicpain inter-relate? Insights from the longitudinal and cognitive-behavioralclinical trials literature. Sleep Med Rev 2004;8:119–32.

[107] Smith MY, Egert J, Winkel G, Jacobsen J. The impact of PTSD on painexperience in persons with HIV/AIDS. PAIN 2002;98:9–17.

[108] Stangl AL, Lloyd JK, Brady LM, Holland CE, Baral S. A systematicreview of interventions to reduce HIV-related stigma and discriminationfrom 2002 to 2013: how far have we come? J Int AIDS Soc 2013;16:18374.

[109] Sterne JA, Sutton AJ, Ioannidis JP, Terrin N, Jones DR, Lau J, CarpenterJ, Rucker G, Harbord RM, Schmid CH. Recommendations forexamining and interpreting funnel plot asymmetry in meta-analyses ofrandomised controlled trials. BMJ 2011;343:d4002.

[110] Surratt HL, Kurtz SP, Levi-MinziMA,Cicero TJ, Tsuyuki K,O’GradyCL. Paintreatment and antiretroviral medication adherence among vulnerable HIV-positive patients. AIDS Patient Care STDS 2015;29:186–92.

[111] Teti M, Bowleg L, Lloyd L. “Pain on top of pain, hurtness on top ofhurtness”: social discrimination, psychological well-being, and sexualrisk among women living with HIV/AIDS. Int J Sex Health 2010;22:205–18.

[112] Thompson MA, Mugavero MJ, Amico KR, Cargill VA, Chang LW, GrossR, Orrell C, Altice FL, Bangsberg DR, Bartlett JG. Guidelines forimproving entry into and retention in care and antiretroviral adherence forpersons with HIV: evidence-based recommendations from anInternational Association of Physicians in AIDS Care panel. Ann InternMed 2012;156:817–33.

[113] Trafton JA, Sorrell JT, Holodniy M, Pierson H, Link P, Combs A, IsraelskiD. Outcomes associated with a cognitive-behavioral chronic painmanagement program implemented in three public HIV primary careclinics. J Behav Health Serv Res 2012;39:158–73.

[114] Tsui JI, Cheng DM, Coleman SM, Blokhina E, Bridden C, Krupitsky E,Samet JH. Pain is associated with heroin use over time in HIV-infectedRussian drinkers. Addiction 2013;108:1779–87.

[115] Tsui JI, Cheng DM, Coleman SM, Lira MC, Blokhina E, Bridden C,Krupitsky E, Samet JH. Pain is associated with risky drinking over timeamong HIV-infected persons in St. Petersburg, Russia. Drug AlcoholDepend 2014;144:87–92.

[116] Tsui JI, Cheng DM, Libman H, Bridden C, Samet J. Hepatitis C virusinfection is associated with painful symptoms in HIV-infected adults.AIDS Care 2012;24:820–7.

[117] Tsui JI, Cheng DM, Quinn E, Bridden C, Merlin JS, Saitz R, Samet JH.Pain and mortality risk in a cohort of HIV-infected persons with alcoholuse disorders. AIDS Behav 2016;20:583–9.

[118] Uebelacker LA, Weisberg RB, Herman DS, Bailey GL, Pinkston-CampMM, Garnaat SL, Stein MD. Pilot randomized trial of collaborativebehavioral treatment for chronic pain and depression in persons livingwith HIV/AIDS. AIDS Behav 2016;20:1675–81.

[119] Uebelacker LA, Weisberg RB, Herman DS, Bailey GL, Pinkston-CampMM, Stein MD. Chronic pain in HIV-infected patients: relationship todepression, substance use, and mental health and pain treatment. PainMed 2015;16:1870–81.

[120] van Sighem A, Gras L, Reiss P, Brinkman K, de Wolf F. Lifeexpectancy of recently diagnosed asymptomatic HIV-infectedpatients approaches that of uninfected individuals. AIDS 2010;24:1527–35.

[121] Wadley AL, Mitchell D, Kamerman PR. Resilience does not explain thedissociation between chronic pain and physical activity in South Africansliving with HIV. PeerJ 2016;4:e2464.

[122] Wadley AL, Pincus T, Evangeli M. A preliminary analysis of theassociation between perceived stigma and HIV-related pain.Unpublished. Accessed September 27, 2017. doi: http://dx.doi.org/10.1101/194191.

[123] Ware NC, Wyatt MA, Tugenberg T. Social relationships, stigma andadherence to antiretroviral therapy for HIV/AIDS. AIDS Care 2006;18:904–10.

[124] Williams A, Eccleston C, Morley S. Psychological therapies for themanagement of chronic pain (excluding headache) in adults. CochraneDatabase Syst Rev 2012;11:CD007407.

[125] Williams ACdC. Defeating the stigma of chronic pain. PAIN 2016;157:1581–2.

[126] Williams ACdC.What can evolutionary theory tell us about chronic pain?PAIN 2016;157:788–90.

[127] Williams ACdC, Craig KD. Updating the definition of pain. PAIN 2016;157:2420–3.


35

[128] Wittchen HU, Robins LN, Cottler LB, Sartorius N, Burke JD, Regier D.Cross-cultural feasibility, reliability and sources of variance of theComposite International Diagnostic Interview (CIDI). The multicentreWHO/ADAMHA field trials. Br J Psychiatry 1991;159:645–53.

[129] Woldeamanuel YW, Kamerman PR, Veliotes DG, Phillips TJ, Asboe D,Boffito M, Rice AS. Development, validation, and field-testing of aninstrument for clinical assessment of HIV-associated neuropathy andneuropathic pain in resource-restricted and large population studysettings. PLoS One 2016;11:e0164994.

[130] World Health Organization. Global health observatory data. Available at:

http://www.who.int/gho/hiv/en/. Accessed March 19, 2018.

[131] Zelman DC, Dukes E, Brandenburg N, Bostrom A, Gore M. Identification

of cut-points for mild, moderate and severe pain due to diabetic

peripheral neuropathy. PAIN 2005;115:29–36.

[132] Zukoski AP, Thorburn S. Experiences of stigma and discrimination

among adults livingwith HIV in a lowHIV-prevalence context: a qualitative

analysis. AIDS Patient Care STDS 2009;23:267–76.

2476 W. Scott et al.·159 (2018) 2461–2476 PAIN®

36

Postoperative opioids, endocrine changes,and immunosuppressionSimon Haroutounian

1. Introduction

“The pathetically grim and perspiring patient, fearful ofmoving or breathing, has become a constant fixture onpostsurgical wards. His suffering was anticipated by hisphysician and is accepted in the knowledge that it willdisappear in time.”22

Acute postoperative pain is considered an unavoidable conse-quence of a surgery, and is somewhat anticipated, given the(typically) massive tissue injury, inflammation, and stress responseas a result of a surgical procedure. However, it is well documentedthat the “perspiring patient, who is fearful of moving or breathing,”ie, the patient who suffers from severe postoperative pain, is moreprone to adverse outcomes. Patients, in whom severe pain impairsbreathing, coughing, physical therapy, and early mobilization, aremore likely to develop complications such as respiratory infectionsand thromboembolic events. The pain is also a significant stressorboth from the psychological and physiological points of view.

The responses of the central nervous system and thehypothalamus-pituitary-adrenal (HPA) axis to a perceived stressinvolve a complex network of signaling molecules, includingendorphins, catecholamines, and cortisol. The stress response tosurgery is characterized by activation of HPA axis, as reflected byincreased secretion of hypothalamic corticotropin-releasinghormone (CRH), and subsequently increased levels of adreno-corticotropic hormone (ACTH) and cortisol. In addition, surgicalstress enhances the secretion of other catabolically activehormones, mainly catecholamines and glucagon, but alsoprolactin, growth hormone, and b-endorphin.42

The enhanced postsurgical stress response, which leads tohigh cortisol levels, can result in immunosuppression,33 support-ing the notion that the relief of pain should be beneficial inpreventing, eg, postoperative infections.33 In the setting of cancersurgery, the immune function may have even further implications,

as animal studies have shown that surgery-induced stress isassociated with impaired natural killer (NK) lymphocyte activity,impairing the body’s ability to clear tumor cells.33

Although the description of a grim, perspiring, and sufferingpatient as a constant fixture may not have been surprising in the1950s, one would expect that the development of novel drugsand multimodal analgesia approaches in the past few decadeswould dramatically improve the quality of postoperative painmanagement, and reduce the undesired hormonal and in-flammatory consequences associated with the surgical stress.However, current data from inpatient and outpatient surgicalsettings indicate that between 30% and 50% of postoperativepatients consistently experience moderate to severe pain.19,35,36

2. The role of opioids in the management of acutepostoperative pain

Opioids are the most commonly used drugs for the managementof acute postoperative pain. Depending on the setting and type ofsurgery, opioids are delivered systemically either throughscheduled or pro re nata (as needed) dosing, or througha patient-controlled analgesia device. Alternatively, the post-operative opioid regimen may include neuraxial delivery throughan epidural catheter, as a part of patient-controlled epidural



Department of Anesthesiology andWashington University Pain Center, Washington

University School of Medicine, St. Louis, MO, USA


of The International Association for the Study of Pain. This is an open access article

distributed under the Creative Commons Attribution License 4.0 (CCBY), which

permits unrestricted use, distribution, and reproduction in anymedium, provided the

original work is properly cited.

PR9 3 (2018) e640

http://dx.doi.org/10.1097/PR9.0000000000000640

Key Points

1. Surgery is associated with a massive inflammatory andstress response. Untreated postoperative pain results inimmunosuppression, increases the risk of thromboem-bolic events, and delays recovery.

2. Chronic opioid therapy results in major endocrine changessuch as opioid-induced androgen deficiency and bonedemineralization. The clinical relevance of these phenom-ena with short-term opioid use for postoperative pain isunclear.

3. In the surgical setting, high-dose opioidsmay contribute toinhibition of immune responses and curbing of stressresponse (eg, cortisol rise), but the clinical consequencesof these observations are still unclear.

4. Multimodal postoperative approaches, especially thosecombining regional anesthesia with local anesthetics, helpimprove pain scores and reduce postoperative opioidrequirements; however, contradictory data exist on theireffect in reducing immune-mediated complications suchas infections or tumor dissemination.

3 (2018) e640 www.painreportsonline.com 1

37

analgesia (PCEA) technique. Mostly, opioids that are used foracute postoperative pain relief, include the short-acting mor-phine, hydromorphone, oxycodone (in countries where paren-teral formulation is available), and occasionally fentanyl ortramadol (which is a weak opioid agonist and a serotonin–norepinephrine reuptake inhibitor).

Opioids are very effective in treating acute postoperative pain;however, not without important side effects. Among their mostcommonly observed and reported adverse effects in this settingare nausea, vomiting, sedation, pruritus, and constipation. Themore severe undesired outcome is respiratory depression, whichis potentially life-threatening, and is the most feared side effect ofopioid medications. The brainstem control of respiratory rate andtidal volume is driven by afferent input of partial pressure of arterialO2 (through chemosensors in carotid and aortic bodies) and CO2

(through chemosensors in the brainstem). Opioids, throughm-opioid receptor-mediated depression of excitability of brain-stem chemosensory neurons, depress the ventilatory responseto increased CO2, thus depressing respiration.31

3. Literature search methodology and focus

This update will focus on a different subset of effects associatedwith the use of opioid analgesics in the acute postoperativesetting; these are endocrine changes and immunosuppression.This clinical update will discuss the evidence behind the role ofopioids in contributing to each of these phenomena in thepostoperative setting, discuss their clinical relevance, andsummarize the recent data on approaches that could beconsidered when treating pain in patients after surgery.

To provide systematicity to literature retrieval, PubMed searchwas performed in September 2017 with the following keywords:“opioids” (All Fields) AND “surgery” (All Fields) AND “pain” (All Fields)AND (“immune suppression” [All Fields] OR endocrine” [All Fields]).The search resulted in 125articles,whichwere screened for relevantinformation, including articles identified from their reference lists.

To obtain a comprehensive evaluation of opioid effects in theperioperative setting, information on both postoperative opioidsfor pain relief and data on intraoperative opioids for analgesia/anesthesia purposes were considered. As rapid recovery fromanesthesia is becoming an increasingly important outcome,opioids with faster offset such as remifentanil and sufentanil areused more commonly, but (especially remifentanil) may beassociated with certain adverse effects.

4. Major endocrine changes associated with opioids

The current evidence suggests that opioids cause endocrinechanges by 2 major mechanisms:(1) Opioids affect the hypothalamic-pituitary-gonadal (HPG) axis.(2) Opioids affect the HPA axis.

The impact on the HPG axis is a well-described dose-dependent effect of opioids, particularly related to treatment withdaily doses above 100 to 200mg of oral morphine equivalents formore than a few weeks. The HPG cascade is initiated by therelease of gonadotropin-releasing hormone (GnRH) from thehypothalamus, which stimulates the anterior pituitary to releaseluteinizing hormone (LH) and follicle-stimulating hormone (FSH),which stimulate the gonads to produce testosterone andestrogen. Opioids bind to m-opioid receptors in the hypothala-mus, inhibiting the release of GnRH, thus decreasing LH and FSHsecretion from the pituitary. Subsequently, this leads to decreasein testosterone levels and hypogonadism, a condition typicallyreferred to as opioid-induced androgen deficiency (OPIAD).29

Opioid-induced androgen deficiency is a significant observa-tion in the setting of chronic opioid therapy, affecting 53% to 90%of male patients on chronic opioid therapy.29 The long-termsymptoms of OPIAD in this setting include decreased attention,decreased libido, fatigue, erectile dysfunction, and osteoporosis.In women receiving chronic opioids, LH and FSH levels aremarkedly reduced and are associated with amenorrhea andimpaired adrenal androgen production.13

Some studies have questioned the suggestion that OPIAD isassociated only with chronic opioid therapy. Several animal studieshave shown that even acute opioid administration (eg, morphine,fentanyl, and buprenorphine), especially at high doses, results inreduced levels of testosterone,10 but the duration of this effect isreported to last between24hours toup to8weeksafter treatment.11

A significant drop in total testosterone level in humans is alsoobserved with a single $30 mg dose of methadone, or within 24hours of perioperative morphine administration.30,41

Unfortunately, well-designed studies aimed at understandingthe mechanisms of testosterone suppression, and the clinicalrelevance of short-term hypogonadism in the setting of acuteopioid administration are lacking. Moreover, surgical stress mayalso contribute to these phenomena, and patients receivingketamine (and not morphine) analgesia still display reducedtestosterone plasma levels.30 Therefore, more careful studies arewarranted to dissect themagnitude and the importance of OPIADin the perioperative setting.

The impact of the opioids on the HPA axis is less welldescribed. The HPA cascade is initiated by CRH release from thehypothalamus, which stimulates the pituitary to release ACTH tothe systemic circulation, stimulating, in turn, the adrenal glands toproduce 2 hormones—dehydroepiandrosterone (DHEA) andcortisol. Opioids seem to inhibit the functioning of the entireHPA axis. On one hand, they reduce the production/release ofCRH, and on the other, they decrease the responsiveness of theanterior pituitary to CRH. Both processes lead to reduced ACTHsecretion. Independently, opioids may also directly interfere withthe adrenal gland production of cortisol and DHEA. Cortisol isimportant for mounting stress responses, and DHEA is animportant precursor for testosterone (in men) and estradiol (inwomen).

The mechanistic data on opioid-associated changes in HPAaxis come primarily from rodent and healthy volunteer experi-ments. For example, in healthy volunteers, single-dose morphinesuppresses ACTH and cortisol levels both at baseline and afterCRH stimulation.2 However, the extrapolation of healthy volun-teer data to the surgical setting is not straightforward, as thesurgical stress, and the postoperative pain per se, can havea substantial effect on the functionality of the HPA axis. Thesurgery typically results in an increased stress response, whichsubsides after about 24 hours. Certain opioids (remifentanil,particularly) are reported to acutely suppress plasma cortisol ina dose-dependent manner.1 In the setting of elective C-section,remifentanil administered as a bolus followed by continuousintraoperative infusion (compared with fentanyl administrationafter delivery) partially obtunded the neuroendocrine response tosurgery with a decrease in ACTH rise (but not in norepinephrine,epinephrine, and growth hormone).14

The association between opioids and blunting ACTHor cortisolrise seems to be modest at best, and highly dependent on thetimeframe of the assessment. In addition, some studies failed tofind changes in cortisol levels after intraoperative remifentanilinfusion.3 As in the case of reduced testosterone concentrations,the clinical relevance of altered cortisol levels in the setting ofperioperative opioid therapy is not clear.

2 S. Haroutounian·3 (2018) e640 PAIN Reports®

38

5. Opioids and immune response—inflammationand infections

Opioids do not possess strong anti-inflammatory propertiessuch as nonsteroidal anti-inflammatory drugs or corticosteroiddrugs, and their potential effects on inflammatory responsesseem to be highly dependent on the setting. For example, inpatients undergoing coronary artery bypass graft procedure,administration of continuous remifentanil infusions (vs intermit-tent fentanyl dosing) resulted in lower levels of proinflammatorycytokines (such as TNF-a and IL-6) at some time points aftercardiopulmonary bypass. Eight hours after surgery, however, nodifferences were observed between the groups.40 On thecontrary, some studies have reported lower inflammatoryresponse (lower C-reactive protein level) with opioid-minimizing analgesia in the setting of colorectal surgery38 orenhanced proinflammatory cytokine release in the spinal cordas a response to opioid challenge. The clinical relevance ofthese short-term effects of opioids on inflammatory markersremains to be investigated.

The 2 main areas of research focused on opioid-associatedimmune effects deal with (1) the effects of opioids on immuneresponse to infections; and (2) opioid effects on tumor-specificimmune responses, affecting tumor growth and dissemination.

Major histocompatibility complex, class II (MHC-II) molecules,expressed on antigen-presenting cells, are important regulatorsof immune cell development and function. Morphine has beenshown to alter gene expression of the MHC-II in circulatingimmunocytes (Beagles 2004), and thus suspected in causingimmunosuppression. It is a matter of debate whether theseeffects aremediated by opioid-induced alteration in cortisol levels(as in adrenalectomized rats, morphine exposure does not affectMHC-II), or there is a direct immunosuppressive effect attribut-able to opioids. Interestingly, morphine withdrawal results inrenewed increase in circulating corticosterone levels anda renewed suppression of MHC-II in previously opioid-tolerantanimals.28

The risk of surgical site infection was higher after abdominalsurgery with remifentanil anesthesia vs fentanyl anesthesia.21 Thefindings could be attributable to more substantial immunosup-pression with remifentanil (although direct data are lacking), or toopioid withdrawal, which is more likely with the short-actingremifentanil. Additional data support this notion of immunosup-pression associated with opioid withdrawal; for example,remifentanil discontinuation increased the risk of intensive careunit–acquired infection,27 and morphine withdrawal in mice (after96-hour exposure) increased the risk of infection in an experi-mental model of septic shock. Interestingly, the most commonlydetected organisms in tissue of morphine-withdrawn mice werebacteria that are part of the normal gastrointestinal flora.18 Thereare additional rodent studies suggesting that morphine, byaltering gut microbiome, may increased the risk of sepsis bybacterial dissemination.24

A review examining whether opioids increase the risk ofinfections in the perioperative or intensive care setting, suggestedthat patients receiving higher doses of systemic opioids had anincreased risk of developing pneumonia perioperatively.32

However, these results are not universal and were observed onlyif laparoscopic vs open surgery (or epidural vs systemic opioidtherapy) was compared.

Overall, the data are both inconsistent and insufficient todetermine the extent of opioid-associated immune suppressionon infectious complications after surgery. Untreated pain itselfcan increase the risk of infections, eg, because of impaired

mobility; pain also enhances the body’s stress response, whichby enhancing circulating cortisol, can contribute to immunosup-pression. It needs to be taken into account that low-dose opioidcontrol groups in some studies (eg, epidural analgesia andlaparoscopic surgeries) could have better quality of analgesia andimproved ability to clear secretions. It is also possible thatincreased risk of infection is among the immunological sequelaeof opioid withdrawal, rather that opioid therapy, per se. Themicrobiome-related effects of opioids (and the potential dissem-ination of gut microorganisms) are another area that requiresadditional research. Although some studies suggest thatbuprenorphine (partial m-opioid agonist and a k-opioid antago-nist) may be devoid of immunosuppressive effects,33 whetherdifferent opioids have differential effects on immune function stillrequires detailed investigation.

6. Opioids and immune response—cancer

Opioids can suppress NK cell cytotoxicity. Both high-dose andlow-dose fentanyl suppressed NK cytotoxicity for 24 hours.5

However, rate of recovery of NK cell suppression was longer inthe high-dose fentanyl group. Rats, which were treated with 20mg/kg morphine (vs saline), developed a decrease in B-lymphocyte blood expression of MHC class II molecules within2 hours.4 The same group has previously reported that in rats,fentanyl suppresses NK cell cytotoxicity and increases the risk oftumor metastases.34 Although there might be a dose-dependenteffect, it is not clear whether different opioids affect the NKfunction differentially. In this regard, data suggest that endoge-nous opioids (b-endorphin) inhibit T-cell proliferation to a lesserextent than exogenous morphine.15

Despite these experimental findings, it is unclear whether thereis any long-term immunosuppression associated with similarchanges in antigen-presenting cells in humans.

A large retrospective analysis of a national registry data fromDenmark (n 5 34,188) showed no difference of breast cancerrecurrence as a function of opioid use.12 The researcherscategorized “strong immunosuppressive opioids (codeine, mor-phine, and fentanyl) vs weakly immunosuppressive opioids(oxycodone, tramadol, buprenorphine, and hydromorphone),”based on previous literature, but found no difference between thegroups.

Considering the immunosuppressive effects attributable topain and the enhanced stress response (and cortisol increase),which opioids may blunt, it is currently unclear whether theclinically relevant “net effect” of postoperative pain managementwith opioids tips the immune balance towards immune suppres-sion. There is also insufficient evidence to determine that someopioids produce strong immunosuppressive effects, whereasothers produce only weak or no such effects.

7. Mitigation approaches

7.1. Stress response and inflammation

Inflammatory reaction after surgery is a physiological responsethat helps the healing process. An excessive inflammatoryresponse can lead to complications, but immune suppressioncould negatively affect the healing process. With potential effectsof opioids on inflammation, and potential immunosuppressantactivity, the attempt in the recent years has been to usemultimodal analgesia approaches that provide adequate analge-sia, but avoid excessive opioid use, especially high-dose intra-operative remifentanil.


39

Patient-controlled epidural analgesia (with bupivacaine andfentanyl) after lower abdominal surgery have resulted in bluntedpostoperative elevation of cortisol and prolactin, and lower painscores, compared with other opioid-only systemic analgesiaregimens.42

In a setting of retropubic prostatectomy, PCEA (with ropiva-caine and sufentanil) resulted in lower pain intensity, and reducedthe postoperative stress response (plasma cortisol and glucose),but not the acute inflammatory response (TNF-a and IL-6levels).17

Another study in 24 women undergoing laparoscopic chole-cystectomy assessed the effect of a single-dose intrathecal(bupivacaine with fentanyl) vs epidural (ropivacaine with fentanyl)anesthesia, before general anesthesia in both groups.7 Intra-operative cortisol, noradrenaline, and total catecholamine levelswere significantly lower in the intrathecal (spinal) anesthesia group(and patients required less systemic fentanyl); ie, spinal anaes-thesia produced a more favourable endocrine response thanepidural anaesthesia.

In pediatric cardiac surgery, on the other hand, high-doseintraoperative fentanyl has demonstrated better (lower) profile ofstress markers such as ACTH, glucose, cortisol, and lactate,compared with low-dose fentanyl. High-dose intraoperativefentanyl also resulted in lower postoperative opioidrequirements.26

The available results demonstrate that neuraxial anesthesiaand analgesia (intrathecal route perhaps more advantageousthan epidural), result in either lower pain scores or less systemicopioid requirements, and can blunt the stress response, withminimal effect on acute inflammatory response. This suggeststhat multimodal anesthesia with neuraxial opioid and localanesthetic delivery, where possible, may provide advantage tosystemic (especially high-dose) opioid administration. Theadvantages in the pediatric surgery setting have not beenexplored sufficiently and merit more thorough investigation.

7.2. Cancer

The literature is divided on the topic of regional anesthesiaapproaches and their potential effect on cancer-related out-comes. Initial retrospective studies reported that perioperativeuse of regional (neuraxial or peripheral) anesthesia is associatedwith improved outcomes in terms of cancer recurrence andsurvival after breast and prostate surgery.6,16

However, a later retrospective study in colorectal cancersurgery,20 an ad hoc analysis of a prostatectomy study,37 anda retrospective study in lung cancer did not show any benefit ofregional anesthesia on cancer-related outcomes.9 A prospec-tive randomized controlled trial25 did not demonstrate pro-longed survival in major abdominal surgery, and a recentretrospective study in patients undergoing radical cystectomyfor bladder cancer demonstrated more than 50% reduction inperioperative opioid consumption with spinal anesthesia, butthis difference was not associated with changes in outcomessuch as all-cause mortality, bladder cancer mortality, or cancerrecurrence.39

A prospective study8 found that innate immunity (NK cells,CD41, and CD81 cells) was depressed in lung cancer patientsundergoing resection, but postoperative epidural analgesia didnot help preserve the immunity.

In addition, a recent systematic review of the literature (15studies) found inconclusive evidence to support or refute thesuggestion that paravertebral blocks in breast cancer can reducecancer recurrence or improve survival.

Currently, there is no conclusive evidence that regionalanesthesia, by either reducing opioid doses or by othermechanisms such as sympathetic blockade, can improve long-term cancer-related outcomes in cancer patients undergoingsurgery. With that said, there is no major disadvantage in usingregional techniques, and the opioid-sparing and stressresponse–blunting acute effects justify the widespread use ofsuch approaches for postoperative pain relief.

8. Critical questions to be addressed byfuture research

The undesired effects attributable to opioids, which werediscussed in this article, are often challenging to address, assome of these effects may be related to surgical stress orpostoperative pain, per se. For example, endogenous opioidssuch as beta-endorphins play an important role in the regulationof gonadotropins through modulation of GnRH pulse amplitudeand frequency. Therefore, with the question of OPIAD inmind, it iscritical to control for this variable, as some data suggest thattestosterone levels are lower in subjects with pain compared withcontrols, irrespective of opioid use.23 Are then patients, who areinmore pain (and therefore are likely to require higher opioid dose)more likely to develop endocrine adverse effects? What is thenthe contribution of pain vs opioid (type and dose)? Is it moreimportant to control the pain well, or avoid high-dose opioids,even at the expense of higher pain? These research questions areyet to be answered and need to be address accurately to movethe field forward, toward safer and more rational perioperativeopioid use.

In a similar scenario, severe pain is associated with enhancedstress response, accompanied by catecholamine and cortisolrelease. Numerous preclinical and clinical studies have focusedon the effect of opioid analgesics on the postoperative stressresponse. As enhanced stress response may cause immunesuppression and lead to postsurgical complications, bluntingthe stress response has been historically considered as a desir-able perioperative outcome. However, the advantages ofblunting the stress response with opioids should be weighedagainst potential immunosuppression, and more research isneeded to achieve an optimal balance that would positivelyaffect patient outcomes.

Disclosures

The author has no conflict of interest to declare.

Article history:Received 2 November 2017Accepted 2 December 2017

References

[1] Aceto P, Dello Russo C, Lai C, Perilli V, Fucci N, De Giovanni N, Piras A,Navarra P, Sollazzi L. Relationship between blood remifentanilconcentration and stress hormone levels during pneumoperitoneum inpatients undergoing laparoscopic cholecystectomy. Eur Rev MedPharmacol Sci 2017;21:4419–22.

[2] Allolio B, Schulte HM, Deuss U, Winkelmann W. Cyproheptadine inhibitsthe corticotropin releasing hormone (CRH)–induced hormone release innormal subjects. Horm Metab Res Suppl 1987;16:36–8.

[3] Barbieri A, Giuliani E, Genazzani A, Baraldi E, Ferrari A, D’Amico R,Coppi G. Analgesia and endocrine surgical stress: effect of twoanalgesia protocols on cortisol and prolactin levels during abdominalaortic aneurysm endovascular repair. Neuro Endocrinol Lett 2011;32:526–9.

4 S. Haroutounian·3 (2018) e640 PAIN Reports®

40

[4] Beagles K, Wellstein A, Bayer B. Systemic morphine administrationsuppresses genes involved in antigen presentation. Mol Pharmacol 2004;65:437–42.

[5] Beilin B, Shavit Y, Hart J, Mordashov B, Cohn S, Notti I, Bessler H. Effectsof anesthesia based on large versus small doses of fentanyl on naturalkiller cell cytotoxicity in the perioperative period. Anesth Analg 1996;82:492–7.

[6] Biki B, Mascha E, Moriarty DC, Fitzpatrick JM, Sessler DI, Buggy DJ.Anesthetic technique for radical prostatectomy surgery affects cancerrecurrence: a retrospective analysis. Anesthesiology 2008;109:180–7.

[7] Calvo-Soto P,Martinez-Contreras A, Hernandez BT, And FP, Vasquez C.Spinal-general anaesthesia decreases neuroendocrine stress responsein laparoscopic cholecystectomy. J Int Med Res 2012;40:657–65.

[8] Cata JP, Bauer M, Sokari T, Ramirez MF, Mason D, Plautz G, Kurz A.Effects of surgery, general anesthesia, and perioperative epiduralanalgesia on the immune function of patients with non-small cell lungcancer. J Clin Anesth 2013;25:255–62.

[9] Cata JP, Gottumukkala V, Thakar D, Keerty D, Gebhardt R, Liu DD.Effects of postoperative epidural analgesia on recurrence-free and overallsurvival in patients with nonsmall cell lung cancer. J Clin Anesth 2014;26:3–17.

[10] Ceccarelli I, De Padova AM, Fiorenzani P, Massafra C, Aloisi AM. Singleopioid administration modifies gonadal steroids in both the CNS andplasma of male rats. Neuroscience 2006;140:929–37.

[11] Chrastil J, Sampson C, Jones KB, Higgins TF. Evaluating the affect andreversibility of opioid-induced androgen deficiency in an orthopaedicanimal fracture model. Clin Orthop Relat Res 2014;472:1964–71.

[12] Cronin-Fenton DP, Heide-Jorgensen U, Ahern TP, Lash TL, ChristiansenPM, Ejlertsen B, Sjogren P, Kehlet H, Sorensen HT. Opioids and breastcancer recurrence: a Danish population-based cohort study. Cancer2015;121:3507–14.

[13] Daniell HW. Opioid endocrinopathy in women consuming prescribedsustained-action opioids for control of nonmalignant pain. J Pain 2008;9:28–36.

[14] Draisci G, Valente A, Suppa E, Frassanito L, Pinto R, Meo F, De Sole P,Bossu E, Zanfini BA. Remifentanil for cesarean section under generalanesthesia: effects on maternal stress hormone secretion and neonatalwell-being: a randomized trial. Int J Obstet Anesth 2008;17:130–6.

[15] Du JY, Liang Y, Fang JF, Jiang YL, Shao XM, He XF, Fang JQ. Effect ofsystemic injection of heterogenous and homogenous opioids onperipheral cellular immune response in rats with bone cancer pain:a comparative study. Exp Ther Med 2016;12:2568–76.

[16] Exadaktylos AK, Buggy DJ, Moriarty DC, Mascha E, Sessler DI. Cananesthetic technique for primary breast cancer surgery affect recurrenceor metastasis? Anesthesiology 2006;105:660–4.

[17] Fant F, Tina E, Sandblom D, Andersson SO, Magnuson A, Hultgren-Hornkvist E, Axelsson K, Gupta A. Thoracic epidural analgesia inhibits theneuro-hormonal but not the acute inflammatory stress response afterradical retropubic prostatectomy. Br J Anaesth 2013;110:747–57.

[18] Feng P, Truant AL, Meissler JJ Jr, Gaughan JP, Adler MW, Eisenstein TK.Morphine withdrawal lowers host defense to enteric bacteria:spontaneous sepsis and increased sensitivity to oral Salmonellaenterica serovar Typhimurium infection. Infect Immun 2006;74:5221–6.

[19] Gerbershagen HJ, Aduckathil S, van Wijck AJ, Peelen LM, Kalkman CJ,Meissner W. Pain intensity on the first day after surgery: a prospectivecohort study comparing 179 surgical procedures. Anesthesiology 2013;118:934–44.

[20] Gottschalk A, Ford JG, Regelin CC, You J, Mascha EJ, Sessler DI,Durieux ME, Nemergut EC. Association between epidural analgesia andcancer recurrence after colorectal cancer surgery. Anesthesiology 2010;113:27–34.

[21] Inagi T, Suzuki M, Osumi M, Bito H. Remifentanil-based anaesthesiaincreases the incidence of postoperative surgical site infection. J HospInfect 2015;89:61–8.

[22] Keats AS. Postoperative pain: research and treatment. J Chronic Dis1956;4:72–83.

[23] Lundh D, Hedelin H, Jonsson K, Gifford M, Larsson D. Assessing chronicpelvic pain syndrome patients: blood plasma factors and cortisol saliva.Scand J Urol 2013;47:521–8.

[24] Meng J, Banerjee S, Li D, Sindberg GM, Wang F, Ma J, Roy S. Opioidexacerbation of gram-positive sepsis, induced by gut microbialmodulation, is rescued by IL-17A neutralization. Sci Rep 2015;5:10918.

[25] MylesPS,PeytonP,Silbert B,Hunt J,Rigg JR,SesslerDI, InvestigatorsATG.Perioperative epidural analgesia for major abdominal surgery for cancer andrecurrence-free survival: randomised trial. BMJ 2011;342:d1491.

[26] Naguib AN, Tobias JD, Hall MW, Cismowski MJ,Miao Y, Barry N, PrestonT, Galantowicz M, Hoffman TM. The role of different anesthetictechniques in altering the stress response during cardiac surgery inchildren: a prospective, double-blinded, and randomized study. PediatrCrit Care Med 2013;14:481–90.

[27] Nseir S, Hoel J, Grailles G, Soury-Lavergne A, Di Pompeo C, Mathieu D,Durocher A. Remifentanil discontinuation and subsequent intensive careunit-acquired infection: a cohort study. Crit Care 2009;13:R60.

[28] Nugent AL, Houghtling RA, Bayer BM. Morphine suppresses MHC-IIexpression on circulating B lymphocytes via activation of the HPA.J Neuroimmune Pharmacol 2011;6:130–41.

[29] O’Rourke TK Jr, Wosnitzer MS. Opioid-induced androgen deficiency(OPIAD): diagnosis, management, and literature review. Curr Urol Rep2016;17:76.

[30] Oyama T, Toyota M, Shinozaki Y, Kudo T. Effects of morphine andketamine anaesthesia and surgery on plasma concentrations ofluteinizing hormone, testosterone and cortisol in man. Br J Anaesth1977;49:983–90.

[31] Pattinson KT. Opioids and the control of respiration. Br J Anaesth 2008;100:747–58.

[32] Rittner HL, Roewer N, Brack A. The clinical (ir)relevance of opioid-inducedimmune suppression. Curr Opin Anaesthesiol 2010;23:588–92.

[33] SacerdoteP.Opioids and the immune system.PalliatMed2006;20(suppl 1):s9–15.

[34] Shavit Y, Ben-Eliyahu S, Zeidel A, Beilin B. Effects of fentanyl on naturalkiller cell activity and on resistance to tumor metastasis in rats. Dose andtiming study. Neuroimmunomodulation 2004;11:255–60.

[35] Sommer M, de Rijke JM, van Kleef M, Kessels AG, Peters ML, Geurts JW,Gramke HF, Marcus MA. The prevalence of postoperative pain in a sampleof 1490 surgical inpatients. Eur J Anaesthesiol 2008;25:267–74.

[36] SussmanM,Goodier E, Fabri I, Borrowman J, Thomas S,Guest C, BantelC. Clinical benefits, referral practice and cost implications of an in-hospitalpain service: results of a service evaluation in a London teaching hospital.Br J Pain 2017;11:36–45.

[37] Tsui BC, Rashiq S, Schopflocher D, Murtha A, Broemling S, Pillay J,Finucane BT. Epidural anesthesia and cancer recurrence rates afterradical prostatectomy. Can J Anaesth 2010;57:107–12.

[38] Watt DG, Horgan PG, McMillan DC. Routine clinical markers of themagnitude of the systemic inflammatory response after electiveoperation: a systematic review. Surgery 2015;157:362–80.

[39] Weingarten TN, Taccolini AM, Ahle ST, Dietz KR, Dowd SS, Frank I,Boorjian SA, Thapa P, Hanson AC, Schroeder DR, Sprung J.Perioperative management and oncological outcomes following radicalcystectomy for bladder cancer: a matched retrospective cohort study.Can J Anaesth 2016;63:584–95.

[40] Winterhalter M, Brandl K, Rahe-Meyer N, Osthaus A, Hecker H, Hagl C,Adams HA, Piepenbrock S. Endocrine stress response and inflammatoryactivation during CABG surgery. A randomized trial comparingremifentanil infusion to intermittent fentanyl. Eur J Anaesthesiol 2008;25:326–35.

[41] Woody G, McLellan AT, O’Brien C, Persky H, Stevens G, Arndt I, CarroffS. Hormone secretion in methadone-dependent and abstinent patients.NIDA Res Monogr 1988;81:216–23.

[42] Yardeni IZ, Shavit Y, Bessler H, Mayburd E, Grinevich G, Beilin B.Comparison of postoperative painmanagement techniques on endocrineresponse to surgery: a randomised controlled trial. Int J Surg 2007;5:239–43.


41

journal of issn 1998-2062 - painsa · 2019. 3. 18. · journal of the south african chapter of the...

Documents