c published by blackwell publishing letters to the … in... · published by blackwell publishing...

American Journal of Gastroenterology ISSN 0002-9270C© 2008 by Am. Coll. of Gastroenterology doi: 10.1111/j.1572-0241.2008.01959.xPublished by Blackwell Publishing

LETTERS TO THE EDITOR

Refractory Autoimmune Pancreatitis:Azathioprine or Steroid Pulse Therapy?

TO THE EDITOR: We read with interest the article by Churchet al. (1) on autoimmune pancreatitis (AIP) in the first seriesfrom the United Kingdom. They diagnosed 11 patients withAIP and treated them with oral steroid. Altough initial re-sponse to steroid therapy was excellent in all patients, sixpatients (55%) relapsed on steroid reduction or withdrawal,and four of them responded to azathioprine and increasedsteroid. Because azathioprine therapy may cause acute pan-creatitis and requires frequent blood tests for early detectionof side effects (2–4), a safe and simple alternative for refrac-tory AIP is needed.

AIP is a recently described unique form of chronic pan-creatitis characterized by sausage-like diffuse swelling of thepancreas, a diffusely irregular narrowing of the main pancre-atic duct, a high serum IgG4 concentration, and a responseto oral steroid therapy (5, 6). In unresponsive patients withthe biliary stenosis caused by AIP, surgery may be necessaryfor the relief of symptoms and for differentiation from ma-lignancy (5). Because oral steroid therapy requires a long pe-riod for the drug tapering, the biliary stenosis suspected to becaused by AIP but cannot be distinguished from malignancyis not indicated for the therapy (5, 6). Steroid pulse therapyis a well-recognized alternative for refractory autoimmunedisorders without steroid tapering. We, therefore, applied thetherapy for refractory AIP, resulting in dramatic response fora short period (5, 6).

Azathioprine is an effective drug as a maintenance andsteroid-sparing agent in autoimmune disorders and organtransplants (2–4). Although azathioprine is widely used inCrohn’s disease, a major drawback is the frequent occur-rence of side effects, especially acute pancreatitis(2). A largepopulation-based case-control study found an eight-fold in-creased relative risk of acute pancreatitis in all users ofazathioprine (3). In animal models, azathioprine involvesactivation of circulating vasoactive mediators, formation ofmicrothrombi, or direct injury of the capillary endothelium,and then deteriorates pancreatic microcirculation, thereby in-creasing ischemia and acinar cell injury (4). Altough furthercomparative studies are needed, we believe that azathioprinetherapy should be avoided in pancreatic diseases such as AIP,whereas steroid pulse therapy is a more effective alternativefor refractory AIP.

Mitsunobu Matsushita, M.D.Tsukasa Ikeura, M.D.

Toshiro Fukui, M.D.Kazushige Uchida, M.D.Kazuichi Okazaki, M.D.

Third Department of Internal MedicineKansai Medical University, Osaka, Japan

REFERENCES

1. Church NI, Pereira SP, Deheragoda MG, et al. Autoim-mune pancreatitis: Clinical and radiological features andobjective response to steroid therapy in a US series. Am JGastroenterol 2007;102:2417–25.

2. Bajaj JS, Saeian K, Varma RR, et al. Increased rats ofearly adverse reaction to azathioprine in patients withCrohn’s disease compared to autoimmune hepatitis: Atertiary referral center experience. Am J Gastroenterol2005;100:1121–5.

3. Floyd A, Pedersen L, Nielsen GL. Risk of acute pancreatitisin uses of azathioprine: A population-based case-controlstudy. Am J Gastroenterol 2003;98:1305–8.

4. Foitzik T, Forgacs B, Ryschich E, et al. Effect of differentimmunosuppressive agents on acute pancreatitis: A com-parative study in an improved animal model. Transplanta-tion 1998;65:1030–6.

5. Matsushita M, Yamashina M, Ikeura T, et al. Effectivesteroid pulse therapy for the biliary stenosis caused byautoimmune pancreatitis. Am J Gastroenterol 2007;102:220–1.

6. Nakayama S, Matsushita M, Yamashina M, et al. Effectivesteroid mini-pulse therapy for refractory biliary stenosisdue to autoimmune pancreatitis. Pancreas 2007;35:83–4.

Autoimmune Pancreatitis: OptimalTherapy Has Not Been Determined

TO THE EDITOR: We would like to thank Matsushita et al.for their interest in our article. They raise specific points con-cerning the type of steroid therapy for acute autoimmune pan-creatitis (AIP) and the use of azathioprine in disease relapse.These points are part of the broader question: What is theoptimum therapy for autoimmune pancreatitis (AIP)? Thisquestion remains unanswered, as published data include onlysmall case series and case reports. No randomized controlleddata exist.

Treatment options for acute AIP include observation only,a tapering regimen of oral steroids without maintenance ther-apy, tapering oral steroids followed by low-dose maintenancetherapy, and intermittent steroid pulse therapy. Disease re-lapse may be treated by further steroid therapy with or with-out second line therapy in the form of immunosuppressivedrugs such as azathioprine or 6-mercaptopurine.

1834

Letters to the Editor 1835

Although a proportion of patients with AIP appear to im-prove spontaneously, it is increasingly apparent that steroidtherapy is effective in resolving symptoms, improving imag-ing, and possibly in reducing the longer-term consequences ofthe disease (1). The usual approach is to use oral prednisolone30–40 mg daily for 2–4 wk with a dose reduction of 5 mg perweek thereafter. This approach results in a rapid clinical andradiological improvement, usually evident within 2–4 wk, asreported by a number of authors (2–5) and also demonstratedin our series. Matsushita et al. report a single case of AIPrecurrence after surgery for biliary stricture (6). The authorsstate that in cases of biliary stricturing disease, which are dif-ficult to differentiate from malignancy, oral steroid therapyshould not be used. They treated their patient with two coursesof intravenous methylprednisolone over 3 days each, and as-sessed response after 2 wk. The biliary stricture improved andthe patient was then managed with a tapering course of oralprednisolone. From our own experience we would argue thata similar response would be expected with oral prednisolonefrom the outset. Therefore, we question the need for intra-venous therapy. We would agree that when there is a strongsuspicion that a patient has a resectable malignancy, surgeryshould not be delayed. However, in cases of AIP, there areusually features in the patient’s history, imaging, and serol-ogy which make AIP at least as likely, and in such cases atrial of steroids with regular review would be a reasonableapproach that would be likely to save these patients fromundergoing unnecessary surgery.

Our patient series currently includes 26 patients, of whom11 (42%) have experienced a disease relapse (unpublisheddata), a rate similar to that reported by other authors (1).Perhaps the lessons of autoimmune hepatitis will ultimatelyapply to AIP: that good responses to initial steroid therapyare seen, but disease relapse is common in the absence oflonger-term maintenance immunosuppression (7). Relapsesin AIP usually respond to reintroduction or an increased doseof steroids. In order to prevent further relapse, most authorshave reported either the use of low-dose maintenance steroids(prednisolone 2.5–10 mg daily), or immunomodulators (e.g.,azathioprine, or mycophenolate mofetil). The use of low-dosemaintenance steroids is associated with two main problems.First, there is a significant relapse rate: the study by Hiranoet al. reported a 32% relapse in patients on low-dose steroids(4). Second, there is a risk of developing steroid side effectsin this generally elderly population. Vertebral fractures, os-teonecrosis of the femoral head, and diabetes have all beenreported in AIP patients (4) with steroid effects perhaps ac-centuated by associated pancreaticobiliary disease. While theside effects of immunosuppressants may also be severe, noneof the eight azathioprine-treated patients in our series experi-enced severe side effects, although one patient required dosereduction due to nausea. Similar results have been reportedby other groups (1). The risk of acute pancreatitis secondaryto azathioprine is low, occurring in 2–3.8% of inflammatorybowel disease patients treated with the drug (8, 9). The actualrisk in AIP patients is unknown. The analogy with autoim-

mune hepatitis may again be appropriate in that azathioprinemay rarely cause significant liver function derangement buthas a proven role in maintaining disease remission (10).

We believe that AIP is a chronic, multisystem inflamma-tory condition that is likely to require long-term maintenancetherapy in a significant proportion of patients. We feel that thecurrent evidence supports the use of azathioprine as a steroid-sparing agent in those patients who relapse on withdrawal ofsteroid therapy. Further studies are required to define the op-timum dose and duration of therapy in these patients and alsoto obtain safety data.

Nicholas I. Church, M.D., M.R.C.P.George J.M. Webster, B.Sc., M.D., F.R.C.P.

Department of GastroenterologyUniversity College Hospital, London, UK

REFERENCES

1. Ghazale A, Chari ST. Optimising corticosteroid treatmentfor autoimmune pancreatitis. Gut 2007;56:1650–2.

2. Chari ST, Smyrk TC, Levy MJ, et al. Diagnosis of au-toimmune pancreatitis: The Mayo clinic experience. ClinGastroenterol Hepatol 2006;4:1010–16.

3. Kamisawa T, Yoshiike M, Egawa N, et al. Treating patientswith autoimmune pancreatitis: Results from a long-termfollow-up study. Pancreatology 2005;5:234–8.

4. Hirano K, Tada M, Isayama H, et al. Long-term prognosisof autoimmune pancreatitis without and with corticosteroidtreatment. Gut 2007;56:1719–24.

5. Kamisawa T, Egawa N, Nakajima H, et al. Morphologicalchanges after steroid therapy in autoimmune pancreatitis.Scand J Gastroenterol 2004;39:1154–8.

6. Matsushita M, Yamashina M, Ikeura T, et al. Effectivesteroid pulse therapy for the biliary stenosis caused by au-toimmune pancreatitis. Am J Gastroenterol 2007;102:220–1.

7. Hegarty JE, Nouri Aria KT, Portmann B, et al. Relapse fol-lowing treatment withdrawal in patients with autoimmunechronic active hepatitis. Hepatology 1983;3:685–9.

8. Timmer A, McDonald JW, Macdonald JK. Azathioprineand 6-mercaptopurine for maintenance of remission inulcerative colitis. Cochrane Database Syst Rev 2007;1:CD000478.

9. Bajaj JS, Saeian K, Varma RR, et al. Increased rates ofearly adverse reaction to azathioprine in patients withCrohn’s disease compared to autoimmune hepatitis: Atertiary referral center experience. Am J Gastroenterol2005;100:1121–5.

10. Johnson PJ, McFarlane IG, Williams R. Azathioprine forlong term maintenance of remission in autoimmune hep-atitis. New Engl J Med 1995;333:958–63.

Comparison of Seven Staging Systems inCirrhotic Patients with HepatocellularCarcinoma in a Cohort of Patients WhoUnderwent Radiofrequency Ablationwith Complete Response

TO THE EDITOR: I read with great interest the articleby Guglielmi et al. (1), where the authors compared seven

1836 Letters to the Editor

staging systems for patients with hepatocellular carcinoma(HCC) undergoing radiofrequency ablation (RFA). The au-thors reported that the Barcelona Clinic Liver Cancer (BCLC)system showed the best discrimination ability and monotonic-ity of gradients for the patients, as compared to the other sixstaging systems (1, 2). The investigators demonstrated thatthe BCLC system had the potential to be the most predictivemodel in nonsurgical patients, as well as in surgical patients.The investigators must be congratulated for their difficultwork. However, I have some comments on this article. First,an early censor rate was not mentioned in the text. Any cohortstudy with a rate of loss to follow-up greater than 20% is re-garded as low quality (3). Early censor rates can be differentbetween each stage, and I think it can affect the discrimina-tory ability of each staging system. I would like to know theauthor’s opinion concerning the effect of the early censor rateon the staging systems. Second, five patients with portal veinthrombosis were included in the study. I think that patientswith portal vein thrombosis are usually contraindicated forRFA. I am curious what the rationale was for the inclusion ofthese patients in the study.

Yun Ku Cho, M.D.

Department of Radiology, Seoul VeteransHospital, 6-2 Dunchon-dong, Gangdong-gu

134-060 Seoul, Korea

REFERENCES

1. Guglielmi A, Ruzzenente A, Pachera S, et al. Comparisonof seven staging systems in cirrhotic patients with hep-atocellular carcinoma in a cohort of patients who under-went radiofrequency ablation with complete response. AmJ Gastroenterol 2007;102:1–8.

2. Bruix J, Sherman M, Llovet JM, et al. EASL Panel ofExperts on HCC. Clinical management of hepatocellularcarcinoma. Conclusions of the Barcelona-2000 EASL con-ference. European Association for the Study of the Liver.J Hepatology 2001;35:421–30.

3. Maceneaney PM, Malone DE. Applying ‘Evidence-BasedMedicine’ theory to interventional radiology. Part 2: Aspreadsheet for swift assessment of procedural benefit andharm. Clin Radiol 2000;55:938–45.

Comparison of Seven Staging Systems inCirrhotic Patients with HepatocellularCarcinoma in a Cohort of Patients whoUnderwent Radiofrequency Ablationwith Complete Response—Response toDr. Yun Ku Cho

TO THE EDITOR: We appreciate the commentary on ourpaper “Comparison of seven staging systems in cirrhotic pa-tients with hepatocellular carcinoma in a cohort of patientswho underwent radiofrequency ablation with complete re-

sponse” by Dr. Yun Ku Cho (1). He underlined the problemof early censor rate. In our study we retrospectively analyzedpatients who underwent radiofrequency ablation (RFA) from1998 to 2005. The 6 and 12 months censor rates were 7 and16%, respectively, and we did not identify significant differ-ences of early censor rate among categories of different stag-ing systems. In particular in Barcelona Clinic Liver Cancer(BCLC) staging system the 12-month censor rate was 18% forclass A1, 14% for A2, 9% for A4, and 15% for B, respectively.The BCLC classes A3 and C (3 patients in each group) didnot have 12-months censored patients. RFA was introducedaround 1990s as a palliative treatment of hepatocellular car-cinoma (HCC) and the follow-up time of published studiesare still short (2). It should be underlined that the prognosisof HCC is dismal and the 5-yr survival after RFA is lowerthan 30% (2). For this reason in our experience, as othersin the literature, a mean follow-up time of almost 2 yr isconsidered adequate for survival analysis. Our mean follow-up time of 26 months is comparable to other studies in theliterature.

Dr. Yun Ku Cho also observed the inclusion in our study ofseven patients with portal vein thrombosis. The presence ofportal vein thrombosis is a major prognostic factor in HCCafter surgical and nonsurgical therapies. In our study, pa-tients with main portal vein thrombosis were not included,whereas a small group of patients (less than 10% of patients)with segmental portal vein involvement has been treated withRFA. The prognostic significance of portal vein involvementhas been demonstrated also in our study with 57% of pa-tients who died within 3 yr compared to a lower rate (39%)for patients without vascular involvement; however, it shouldbe noted that two out of seven patients survived more than3 yr. In this group of patients, RFA showed its safety with nopatients who experienced major complication. Also in otherstudies in literature, RFA has proved its safety in patients withportal vein thrombosis, and the authors stated that portal veinthrombosis should not be considered a contraindication forRFA (4, 5). Obviously, the long-term efficacy of RFA in thesepatients needs further validation.

Andrea Ruzzenente, M.D.Alfredo Guglielmi, M.D.

Calogero Iacono, M.D.

Department of Surgery and GastroenterologyUniversity of Verona Medical School, Verona, Italy

REFERENCES

1. Guglielmi A, Ruzzenente A, Pachera S, et al. Comparisonof seven staging systems in cirrhotic patients with hep-atocellular carcinoma in a cohort of patients who under-went radiofrequency ablation with complete response. AmJ Gastroenterol 2008;103:597–604.

2. Machi J, Bueno RS, Wong LL. Long-term follow-upoutcome of patients undergoing radiofrequency ablation


for unresectable hepatocellular carcinoma. World J Surg2005;29:1364–73.

3. Baldan A, Marino D, DE Giorgio M, et al. Gene–Gruppo Epatocarcinoma NORD-EST. Percutaneous ra-diofrequency thermal ablation for hepatocellular carci-noma. Aliment Pharmacol Ther 2006;24:1495–501.

4. Neeman Z, Libutti SK, Patti JW, et al. Percutaneousradiofrequency ablation of hepatocellular carcinoma inthe presence of portal vein thrombosis. Clin Imaging2003;27:417–20.

5. Poggi G, Gatti C, Teragni C, et al. Radiofrequency abla-tion combined with percutaneous ethanol injection in thetreatment of hepatocellular carcinoma and portal vein neo-plastic thrombosis. Anticancer Res 2004;24:2419–21.

Meta-Analyses on the Prophylactic Useof Antibiotics in Acute Pancreatitis:Many Are Called but Few Are Chosen

TO THE EDITOR: I read with interest the article by Bai et al.that was published in the recent issue of the Journal (1). Theauthors conducted a meta-analysis of seven randomized con-trolled trials (RCTs) on the prophylactic use of antibiotics ver-sus placebo/no antibiotics in acute pancreatitis patients withnecrosis proven by computed tomography (CT). Bai and col-leagues concluded that the prophylactic administration of an-tibiotics did not reduce the risk of infected pancreatic necrosisand mortality in patients with acute necrotizing pancreatitis,as well as encouraged other researchers to conduct furthertop-quality trials to investigate this question. However, I havesome comments with regard to this publication.

My first concern regards the selection of studies for thismeta-analysis. Although the authors aimed to derive a sum-mary estimate of antibiotics effect exclusively from patientswith CT-proven necrosis, this was not the case in some ofthe included trials. In particular, some patients were includedin the individual trials on the basis of ultrasound diagno-sis (2) or the level of serum C-reactive protein (3). On theother hand, the authors excluded the study by Spicak et al.(4) because of “no data of pancreatic necrosis.” Awkwardlyenough, their colleagues from the same research institution,who conducted a meta-analysis on the same topic less than2 yr ago, did find the required data and included that study intheir meta-analysis (5).

Table 1. Characteristics of the Meta-Analyses Published in 2006–2007.

Time of Number of Total ProphylacticOn-line Included Number of Effect of

Reference Publication Trials Patients Antibiotics

5 February 2006 6 338 Nonsignificant9 February 2006 6 390 Significant

10 June 2006 6 329 Nonsignificant11 October 2006 5 294 Significant12 April 2007 10 1,279 Significant13 September 2007 6 397 Nonsignificant

1 October 2007 7 467 Nonsignificant

The second point is that Bai et al. recommended enrollingonly patients with CT-verified pancreatic necrosis in a futurelarge-scale RCT. However, it is known that CT is of limiteduse within the first days of the disease and has a good diagnos-tic performance only by day 3 after admission (6). Thereby,this recommendation would lead to a delayed institution ofantibiotics, the possible potential of which as a prophylactictreatment might be exhausted by the time of diagnosis of pan-creatic necrosis by CT. Furthermore, one should be aware thata future adequately powered RCT should recruit 400 patientswith acute necrotizing pancreatitis to show a significant re-duction in infected pancreatic necrosis from 20% to 10% (7).As the recent RCT in 32 centers (many of which are highlydedicated to the studies on acute pancreatitis) in North Amer-ica and Europe was able to recruit only 100 patients over2 yr (8), the recommendation with regard to a new large-scale trial seems fairly formal and unrealistic at the presenttime.

Third, it is interesting to know that, apart from the twometa-analyses from China mentioned above (1, 5), eight moreattempts were undertaken to statistically aggregate the data onthe same research question. While only two new RCTs werepublished in 2006–2007 (3, 8), it seems paradoxical that 7of the 10 meta-analyses were published within the same timeperiod (1, 5, 9–13). Notably, a total of 13 different RCTs wereincluded in those seven meta-analyses, all of which purportedto define the role of antibiotic prophylaxis in acute pancre-atitis. However, as a result of employing different inclusioncriteria, the authors alternately came to ambivalent conclu-sions and provided opposite recommendations regarding theprophylactic effect of antibiotics on the risk of pancreaticinfectious complications (Table 1).

In the setting of this looming deadlock, some authoritiesrecommended to leave the decision on the use of prophy-lactic antibiotics to the discretion of single institutions (7).However, I believe that such an approach may contravene theessence of evidence-based medicine. To my mind, the contra-dictory findings of different authors indicate that the exploitedstandard methodology of meta-analysis is inapt in addressingsuch complex question as the prophylactic effect of antibi-otics in patients with acute pancreatitis because of the markedheterogeneity among the conducted RCTs. Thereby, an ap-plication of advanced statistical techniques (e.g., α-spendingfunction and stochastic curtailment (14)) should be used toprovide a sound and unbiased estimation of the antibiotics asa prophylactic modality in acute pancreatitis. With this aimin mind, a close collaboration among the physicians, statis-ticians, and clinical epidemiologists should be established toanswer one of the most controversial questions in the modernpancreatology.

Maxim S. Petrov, M.D., M.P.H.

Department of SurgeryNizhny Novgorod State Medical Academy

Nizhny Novgorod, Russia


REFERENCES

1. Bai Y, Gao J, Zou DW, et al. Prophylactic antibioticscannot reduce infected pancreatic necrosis and mortalityin acute necrotizing pancreatitis: Evidence from a meta-analysis of randomized controlled trials. Am J Gastroen-terol 2008;103:104–10.

2. Pederzoli P, Bassi C, Vesentini S, et al. A randomizedmulticenter clinical trial of antibiotic prophylaxis of sep-tic complications in acute necrotizing pancreatitis withimipenem. Surg Gynecol Obstet 1993;176:480–7.

3. Rokke O, Harbitz T, Liljedal J, et al. Early treatmentof severe pancreatitis with imipenem. A prospective ran-domised clinical trial. Scand J Gastroenterol 2007;41:771–6.

4. Spicak J, Hejtmainkovai S, Hubaczovai M, et al. Antibioticprophylaxis of infectious complications of acute pancreati-tis: The results of randomised study by meropenem. CeskaSlovenska Gastroenterol Hepatol 2003;57:222–7.

5. Xiong GS, Wu SM, Wang ZH. Role of prophylactic antibi-otic administration in severe acute pancreatitis: A meta-analysis. Med Princ Pract 2006;15:106–10.

6. Banks PA, Freeman ML. The Practice Parameters Com-mittee of the American College of Gastroenterology. Prac-tice guidelines in acute pancreatitis. Am J Gastroenterol2006;101:2379–400.

7. Werner J, Hartwig W, Buchler MW. Antibiotic prophy-laxis: An ongoing controversy in the treatment of severeacute pancreatitis. Scand J Gastroenterol 2007;42:667–72.

8. Dellinger E, Tellado J, Soto NE, et al. Early antibiotic treat-ment for severe acute necrotizing pancreatitis: Random-ized, double-blind, placebo-controlled study. Ann Surg2007;245:674–83.

9. Heinrich S, Schafer M, Rousson V, et al. Evidence-basedtreatment of acute pancreatitis: A look at establishedparadigms. Ann Surg 2006;243:154–68.

10. Mazaki T, Ishii Y, Takayama T. Meta-analysis of prophy-lactic antibiotic use in acute necrotizing pancreatitis. Br JSurg 2006;93:674–84.

11. Villatoro E, Bassi C, Larvin M. Antibiotic therapyfor prophylaxis against infection of pancreatic necro-sis in acute pancreatitis. Cochrane Database Syst Rev2006;18:CD002941.

12. Dambrauskas Z, Gulbinas A, Pundzius J, et al. Meta-analysis of prophylactic parenteral antibiotic use in acutenecrotizing pancreatitis. Medicina (Kaunas) 2007;43:291–300.

13. de Vries AC, Besselink MG, Buskens E, et al. Randomizedcontrolled trials of antibiotic prophylaxis in severe acutepancreatitis: Relationship between methodological qualityand outcome. Pancreatology 2007;7:531–8.

14. Pogue J, Yusuf S. Overcoming the limitations of cur-rent meta-analysis of randomised controlled trials. Lancet1998;351:47–52.

Prophylactic Antibiotics in AcutePancreatitis: Further High-QualityTrials Are Still Warranted

TO THE EDITOR: We wish to thank Dr. Maxim S. Petrovfor showing interest in our article (1); however, some issuesneed to be clarified.

First, we are not awkward at all because, if the addressesand affiliations of the authors of the two different articles

were reviewed carefully, it is quite clear that the institutionof the authors (Renji Hospital, Shanghai Second MedicalUniversity) mentioned (2) is NOT the same institution as ours(Changhai Hospital, Second Military Medical University).

Second, though the authors (2) included an article by Spi-cak et al. (3) in their meta-analysis, the focus of their re-search was on the role of prophylactic antibiotics in severeacute pancreatitis, and not computerized tomography (CT)-verified acute necrotizing pancreatitis; consequently, the sub-group patients with CT-verified acute necrotizing pancreatitiswere not analyzed.

During the process of selection of relevant trials, we foundthree, and not one, articles by Spicak et al. (3–5), but, un-fortunately, the data for the subgroup of patients with CT-verified pancreatic necrosis in the first two studies (3, 4) werenot available. Additionally, the third study (5) seems to be akind of combination of the previous two studies; therefore,we decided it would be more appropriate to wait for the full-text publication to evaluate the methodological quality of thestudy.

With regard to one of the included trials (6), due to ouroversight, we did not notice that not all the enrolled patientsreceived CT scan examination; nonetheless, after personalcommunication with Dr. Ola Røkke, we retrieved the correctsubgroup data of mortality and pancreatic necrosis infectionrate in patients with CT-documented necrotizing pancreatitis(imipenem group 12 patients, control group 16 patients), thatis, 2 patients in the imipenem group and 2 patients in the con-trol group died, while 2 patients in the imipenem group and4 patients in the control group developed pancreatic necro-sis infection. Therefore, we reanalyzed the seven trials, andthe overall effect of antibiotics on mortality (risk ratio [RR]0.73, 95% confidence interval [CI] 0.43–1.23, P = 0.24) andpancreatic necrosis infection (RR 0.85, 95% CI 0.58–1.26,P = 0.42) did not change much. We believe these analysesdefended the results of our study.

Dr. Maxim S. Petrov doubted why only 7 of 15 trials wereincluded in our study. The selection of trials was not cherry-picking; by contrast, as stated in the method section, the aimof our study was to evaluate the effect of prophylactic intra-venous antibiotics on infected pancreatic necrosis and mor-tality in acute necrotizing pancreatitis, and thus we choserelevant trials according to the predetermined protocol. Wehad comprehensively reviewed all the relevant trials, includ-ing the 15 trials mentioned by Dr. Maxim S. Petrov, beforeconducting this present meta-analysis, and revealed that therewas substantial clinical heterogeneity among all these trials.In view of this, it is impossible to combine the results ofthese trials, and based on these greatly heterogeneous trials,it is also impossible to produce meaningful conclusion. Wesuggest it may be appropriate to include all these trials in asystematic review but not in a meta-analysis.

From these trials that we analyzed, we may conclude thatwithin the past two decades significant progress has beenmade in the medical approach to pancreatic necrosis, but theevidence for the recommendations for or against the use of


antibiotics remains inconclusive. It is clear that the most im-portant limitation to an evidence-based approach of use ofantibiotics lies in the difficulty for any single medical cen-ter to enroll sufficient number of patients with pancreaticnecrosis to satisfy the requirements for statistical power. Webelieve future multicenter or multinational trials or individualpatient data analysis would be mandatory to answer currentand future critical problems in necrotizing pancreatitis. Al-though advanced statistical methods are important for dataanalysis, without well-designed and executed trials, no sta-tistical method can close the book on the prophylactic effectof antibiotics.

Finally, we sincerely appreciate the help from Dr. OlaRøkke for kindly providing the necessary data in their study(6).

Yu Bai, M.D.Jun Gao, M.D., Ph.D.

Duo-Wu Zou, M.D.Zhao-Shen Li, M.D.

Evidence-Based Medicine GroupDepartment of Gastroenterology, Changhai Hospital

Second Military Medical UniversityShanghai, China

REFERENCES

1. Bai Y, Gao J, Zou DW, et al. Prophylactic antibioticscan not reduce infected pancreatic necrosis and mortalityin acute necrotizing pancreatitis: Evidence from a meta-analysis of randomized controlled trials. Am J Gastroen-terol 2008;103:104–10.

2. Xiong GS, Wu SM, Wang ZH. Role of prophylactic antibi-otic administration in severe acute pancreatitis: A meta-analysis. Med Princ Pract 2006;15:106–10.

3. Spicak J, Hejtmankova S, Cech P, et al. Antibiotic prophy-laxis in severe acute pancreatitis: Randomized multicenterprospective trial with meropenem. Ceska a Slovenska Gas-troenterologie a Hepatologie 2003;57:228–32.

4. Spicak J, Hubaczova M, Antos F, et al. Antibioticsin the treatment of acute pancreatitis—findings froma randomized multi-centre prospective study. Ceska aSlovenska Gastroenterologie a Hepatologie 2002;56:183–9.

5. Spicak J, Hejtmankova S, Cech P, et al. Antibiotics prophy-laxis in large pancreatic necrosis: Multicenter randomizedtrial with ciprofloxacin and metronidazole or meropenem.Gastroenterology 2004;126(Suppl 2):A229.

6. Rokke O, Harbitz T, Liljedal J, et al. Early treatmentof severe pancreatitis with imipenem. A prospective ran-domised clinical trial. Scand J Gastroenterol 2007;41:771–6.

Disparities in Liver Transplantation forLocalized Hepatocellular Carcinoma:Race or Socioeconomic Class?

TO THE EDITOR: Race has been identified as a variable forliver transplant in localized hepatocellular cancer. However,

socio-economic class may adversely affect certain races. Thatis why the socioeconomic status of an individual may bethe predominant factor in variable transplant rates and willrequire different kinds of intervention.

Siegel et al. (1) studied the patients with nonmetastatichepatocellular cancer (HCC) using the population-basedcancer registries using the US National Cancer Insti-tute’s Surveillance, Epidemiology and End Results (SEER)database.

Forty-five percent of the final study population was Non-Hispanic Whites, 29% Asians, 17% Hispanics and 9%African American, showing disproportionately more Asiansaffected by HCC (Asians comprise only 3.6% of US popu-lation). Only 21% of the analyzed population received trans-plant despite having high survival benefit. Married, youngage group, whites were more likely to receive liver trans-plant. African Americans and Asians were half as likely toreceive liver transplant as the Non-Hispanic Whites (13% and17% vs. 25% whites).

The study did not address why, despite having significantsurvival advantage, only 21% (25% of whites) of the localizedHCC population received a liver transplant. Are there referralbias, insurance limitations, lack of social support, comorbidi-ties, substance/alcohol abuse, etc., issues? African Americans(even belonging to younger age groups) and Asians receivedproportionately lower transplants. If this is true, it is definitelynot a good sign more than four decades after the Civil RightsAct was passed in the U.S. Congress in 1964. How valid isthe above statement? Are we really giving preference to onerace over the other? Is there any other confounding variablethat is being projected as a racial factor?

Noteworthy, unlike other cancer surgeries such as eso-phageal cancer or lung cancer, liver transplant requires long-term follow-up, has strict selection criteria, requires strictpsycho-social scrutiny in terms of compliance potential-ity, family support for post transplant follow-up and care,assessment of personal habits, i.e., substance abuse, alco-hol abuse, and, of course, the financial factor with validhealth insurance. It is in that area that some of the minorityethnic groups may not do favorably well and may act as con-founding variables. Mentionable here, race/ethnicity and so-cial/economic class are very closely inter-linked to affect thelifestyle, mortality and morbidity of an individual (2). Lowerincome groups have higher morbidity (3) as well as mortality(4) even after the race factors are taken into consideration.Likewise, socioeconomic classes could have a stronger in-fluence on low proportion of liver transplant and also lowertransplant rates in African Americans and Asians.

Like all national data-based programs on health mortal-ity, the SEER program also does not have any data on theclass characteristics (i.e., income, education or profession).In fact, United States is one of the few developed countrieswhere mortality statistics are not collected by class. (2) As aconsequence, we may be overemphasizing the racial factors.This major flaw in the data was duly acknowledged by theauthors. A prospective population-based study will be able


to address this issue as well as the 21% overall low transplantrate and focus our attention to the problem of socio-economicdisparity contributing to variable liver transplant rates. Thistype of analysis in liver transplant and health care deliverywill have broader policy implications (5).

Mahmudul Haque, M.D., Ph.D.Asheen Zariat

Gastroenterology Section, Lakeland RegionalHospital, Lakeland, Florida

REFERENCES

1. Siegel AB, McBride RB, El-Serag HB, et al. Racial dispar-ities in utilization of liver transplantation of hepatocellularcarcinoma in the United States, 1998–2002. Am J Gast-troenterol 2008;103:120–7.

2. Navarro V. Race or class versus race and class: mortalitydifferential in the United States. Lancet 1990;336:1238–40.

3. PowellGriner E. Characteristics of persons dying fromcerebrovascular diseases. Washington DC: National Cen-ter for Health Statistics. Preliminary data from vital andhealth statistics. No 180, Feb 8, 1990:2–3).

4. Self assessment of health according to race and income.Table 45 in Health, United States 1987. Washington DC:US Department of Health and Human Services, 1988:91.

5. Fiscella K, Franks P, Gold M, et al. Inequality in quality:Addressing socioeconomic, racial and ethnic disparities inhealth care. JAMA 2000;283:2579–84.

Response to Haque

TO THE EDITOR: Haque and Zariat correctly point out thatrace/ethnicity is probably an indicator for other factors, in-cluding socioeconomic status. SEER does not allow us todistinguish issues related to class and education, as we ac-knowledge in our manuscript. SEER also does not give infor-mation related to underlying comorbidities. This may explainto some extent why the vast majority of those with localizedhepatocellular cancer (HCC) in our study did not receivea transplant for their HCC. However, we have also shownboth overuse and underuse of transplant for HCC in a cohortof older patients, pointing out that other biases may play arole (1).

While selection for receipt of a liver transplant for HCCmay be more involved than for some other types of cancersurgeries, underuse of potentially curative surgery has beennoted in many other cancer types, including esophageal andlung (2, 3). All of these issues support the need for furtherresearch in this area.

Abby Siegel, M.D.

Medical Oncology, Columbia UniversityCollege of Physicians and Surgeons

New York, NY

REFERENCES

1. El-Serag HB, Siegel AB, Davila JA, et al. Treatmentand outcomes of treating hepatocellular carcinoma amongMedicare recipients in the United States: A population-based study. J Hepatol 2006;44:158–66.

2. Steyerberg EW, Neville BA, Koppert LB, et al. Surgicalmortality in patients with esophageal cancer: Develop-ment and validation of a simple risk score. J Clin Oncol2006;24:4277–84.

3. Bach PB, Cramer LD, Warren JL, et al. Racial differencesin the treatment of early-stage lung cancer. N Engl J Med1999;341:1198–205.

The Cell Phone Camera Enters theGastroenterologist’s Office

TO THE EDITOR: Cell phones have become ubiquitous. Ar-ranging for appointments and sick calls have been handled bythe telephone for decades. Newer technology allows the auto-mated refilling of prescriptions and notification of laboratoryresults. Newer Internet-enabled phones permit research of thedisease pathophysiology, local and national support groups,and even a comparison of physicians’ qualifications. I re-cently had two experiences illustrating new ways in which pa-tients are using technology to change the practice of medicineand, particularly, gastroenterology.

A 30-yr-old man presented with an 8-yr history of bowelproblems. He often has 2–3 days between bowel movementsand then will have 2–5 stools over a few hours. The stool con-sistency can vary from hard pellets to loose brown water withmucus. Often there is an aching pain in the left lower quadrant,which is relieved after the bowel movement (BM). Problemsat work or with his family often precipitate these distressingepisodes. There is no bleeding, fevers, or weight loss. Eval-uation by his primary care physician, including laboratorywork, was unremarkable. Dietary changes with decreased fatand increased fiber intake improved his problem for a shortwhile. Coincident with visa issues, his symptoms worsened.At the patient’s insistence, he was referred to a gastroenterol-ogist. To be sure that his concerns were fully appreciated, heused his cell phone to take photos of each BM for 10 dayswith annotations of date and time. He also prepared picturesof his commode at home and work so that the place could beidentified. His cell phone model was capable of producing aslideshow presentation.

A 74-yr-old woman had a near syncopal episode at home.Her family brought her to the emergency department. Eval-uation, which included electrocardiogram, cardiac enzymes,and head computerized axial tomography scan, was unre-vealing. She was released despite objections by the family.After returning home, she had a bout of coffee ground eme-sis. She went to her bathroom and had a syncopal event on thecommode. Voluminous dark stools were noted. The patient’sgranddaughter used her cell phone to record the appearanceof emesis and stool.


These two cases illustrate patients’ increasing ability toprovide historical information with a visual impact that wasnot easily accessible prior to cell phone cameras. Of particularnote, both instances were initiated by the patients. Cell phonecameras have proven useful in providing visual informationto dermatologists (1) and primary care physicians (2). Thereis now also a niche for gastroenterology.

John T. Geneczko Jr., M.D., F.A.C.G.

Department of GastroenterologyClarian Arnett Health

Lafayette, Indiana

REFERENCES

1. Burdick A. Teledermatology extending specialty care be-yond borders. Arch Dermatol 2007;143:1581–2.

2. Parikh R, Wong R. Video phone diagnosis of “funnyturns.”Age Ageing 2007;36:234.

Rupture of Pancreaticoduodenal ArteryAneurysm Caused by SuperiorMesenteric Artery Embolism

TO THE EDITOR: True pancreaticoduodenal artery (PDA)aneurysms are highly associated with celiac artery (CA)stenosis or occlusion, about 60–75% (1), and pancreaticoduo-denal arcades ordinarily function as collateral pathways fromthe superior mesenteric artery (SMA) to the celiac branchesin this situation (1, 2). Sutton and Lawton postulated, with thephrase ”aneurysm of collateral supply,” that development ofPDA aneurysm was a direct consequence of increased bloodflow (3). In fact, the aneurysms disappear with recovery ofnormal blood flow by revascularization of the CA (4). Wereport an interesting case in which a high flow by an SMAembolism at the distal portion of the PDA inflow contributedto the rupture of the PDA aneurysm with CA stenosis; inthis case, angiographic findings clearly indicated the critical

Figure 1. Computed tomography (CT) scan of the abdomen showed a retroperitoneal hematoma (A, white arrow) between the duodenumand the pancreas with PDA aneurysm (white arrowhead) and the thickened wall of the terminal ileum and cecum (B, white arrow) consistentwith ischemic changes.

participation of a rapid rise in PDA flow in the aneurysmrupture.

A 65-yr-old man was referred to our hospital because ofepigastralgia and diarrhea for 2 days. His body temperaturewas 37.2◦C, blood pressure was 126/100 mmHg, and he hadan irregular pulse rate of 103 beats/min. Hemoglobin was 15.6g/dL, white blood cell count was 15,310/mm3, C-reactive pro-tein was 19.4 mg/dL, and serum amylase was 63 IU/L. Otherlaboratory tests were within normal limits. Electrocardiogra-phy showed atrial fibrillation.

Computed tomography (CT) scan of the abdomen showeda retroperitoneal hematoma (Fig. 1A, white arrow) be-tween the duodenum and the pancreas with PDA aneurysm(Fig. 1A, white arrowhead) and the thickened wall of theterminal ileum and cecum (Fig. 1B, white arrow) consistentwith ischemic changes (Figure 1B). Three-dimensional CTangiography (Fig. 2A) revealed the inferior PDA aneurysm(Fig. 2A, yellow arrow) with CA stenosis (Fig. 2A, red arrow)and the embolism at the distal portion of the PDA inflow inthe SMA (Fig. 2A, white arrow), respectively. Selective CAand SMA angiography also showed these findings (Fig. 2B).Selective microcoil embolization for aneurysm was under-taken and the patient has been on a good clinical course.Ischemic colitis due to SMA embolism was also confirmedby colonoscopy.

In the present case, it seems to be likely that the PDAaneurysm originally developed by increased PDA flow dueto CA stenosis and was followed by the embolism at the distalportion of the PDA inflow in the SMA due to atrial fibrilla-tion. These events brought ischemic colitis and a rapid rise inPDA blood flow and, finally, the rupture. To our knowledge,this is the first case of PDA aneurysm rupture by SMA em-bolism, and it also supports the hypothesis that hemodynamicchanges participate in both formation and rupture of PDAaneurysms.

The management of not only ruptured but also nonrup-tured PDA aneurysms is important because the size of PDAaneurysms is not related to the risk of rupture (1). In emboliza-tion, new aneurysm formation due to increased blood flow in


Figure 2. A: Three-dimensional CT angiography revealed the inferior PDA aneurysm (yellow arrow) with CA stenosis (red arrow) and theembolism at the distal portion of the PDA inflow in the SMA (white arrow), respectively. B: Selective SMA angiography also showed thesefindings (inferior PDA, yellow arrow; SMA embolism, white arrow, respectively).

the other PDA was reported (5). And thus, a revascularizationprocedure is recommended for either nonruptured or rupturedcases with CA stenosis. However, it was not performed at thepatient’s request. Careful follow-up for aneurysm recurrencewill be required.

Chiyo Sugiyama, M.D.1

Akifumi Akai, M.D.1

Noriyoshi Yamakita, M.D.1

Keigo Yasuda, M.D.1

Noboru Ihara, M.D.2

Departments of 1Internal Medicine and 2RadiologyMatsunami General Hospital

Dendai Kasamatsu-cho, Hashima-gunGifu, Japan

REFERENCES

1. Marc AM, Andrew TR, Glenn ML, et al. A 45-year-old manwith sudden onset of abdominal pain and hypotension. NEngl J Med 2008;358:178–86.

2. Song SY, Chung JW, Kwon JW, et al. Collateral pathway inpatients with celiac axis stenosis: Angiographic-spiral cor-relation. Radiographics 2002;22:881–93.

3. Sutton D, Lawton G. Coeliac stenosis or occlusion withaneurysm of the collateral supply. Clin Radiol 1973;24:49–53.

4. Mora JD. Coeliac axis artery stenosis with aneurismalcalcification of the collateral supply. Australas Radiol1976;20:252–4.

5. Kitagawa H, Ota T, Kayahara M, et al. A case of report—a newly developed arterial lesion associated with TAE foraneurysm of the pancreaticoduodenal artery (in Japanese).J Biliary Tract Pancreas 2000;21:147–51.

Hepatic Splenosis Diagnosed by SpleenScintigraphy

TO THE EDITOR: A 59-yr-old gentleman had a splenec-tomy in his childhood after trauma to the abdomen. He un-

derwent a physical check-up at our hospital and denied anyabdominal discomfort. There was no personal or family his-tory of neoplasm. Physical examination was unremarkable.Blood serological tests revealed positive hepatitis B virussurface antigen, but antibody against hepatitis C virus wasnegative. As well, both alpha-fetoprotein (4.8 ng/mL, nor-mal <20 ng/mL) and carcinoembryonic antigen (2.4 ng/mL,normal <6 ng/mL) levels were within normal limits. Ab-domen ultrasound (US) disclosed absence of spleen and a0.9-cm-sized hypoechoic nodule abutting the anterior surfaceof left lateral hepatic segment. Abdomen computed tomog-raphy (CT) without contrast medium administration revealedtwo hypodense nodules in hepatic segment VII, 2.2 cm and1.5 cm in size, respectively. There was also a 1.2-cm-sized hy-podense nodule at the left lateral hepatic segment. All of theselesions were located near the liver capsule. After contrast ad-ministration, the lesions became homogeneously hyperdensein the arterial phase, isodense in the portal venous phase, andslightly hypodense in the equilibrium phase. The main portaltrunk and its branches were patent. Magnetic resonance imag-ing (MRI) of the abdomen without contrast medium injectionshowed that the lesions were homogeneously hypointenseon T1-weighted imaging (Fig. 1A) and hyperintense onT2-weighted imaging (Fig. 1B). After intravenous admin-istration of gadolinium-diethylenetriaminepentaacetic acid(Gd-DTPA), global enhancement of the lesions was found inthe arterial phase (Fig. 1C) and the lesions became isointensein the portal phase (Fig. 1D). With reference to the history ofsplenectomy, hepatic splenosis was strongly suspected. How-ever, with only these US, CT, and MRI features, it was diffi-cult to distinguish splenosis from other hepatic masses suchas adenoma, focal nodular hyperplasia, and hepatocellularcarcinoma (HCC). We then arranged 99m technetium (Tc)-labeled heat-denatured red blood cell scintigraphy, which dis-closed uptake of the labeled cells in areas corresponding tothe nodules seen on other image studies (Fig. 2) and con-firmed the diagnosis of hepatic splenosis.


Figure 1. (A) T1-weighted MR image before the administration of Gd-DTPA delineated the lesions as hypointense nodules. (B) T2-weightedMR image showed the lesions as hypointense nodules. (C) T1-weighted MR image after the administration of Gd-DTPA revealed globalenhancement of the lesions in the arterial phase. (D) T1-weighted MR image after the administration of Gd-DTPA disclosed that the lesionsbecame isointense in the portal phase.

Figure 2. 99m Tc-labeled heat-denatured red blood cell scintigraphy showed uptake of the labeled cells in areas corresponding to the MRI.


In 1939, Buchbinder and Lipkoff introduced the term“splenosis” to describe the first case in a young woman(1). Splenosis represents heterotopic autotransplantation andseeding of splenic tissue, usually occurring after splenictrauma or operation. Splenic vein emboli or a hematogenousspread of splenic pulp has also been described as a possi-ble mechanism of seeding into the liver (2). The implantsare rarely clinically important and are incidental findings (3).Some lesions, however, can cause abdominal pain by twist-ing of a long pedicle attached to the splenic nodule (4). Insplenosis, US, CT, and MRI findings are not specific andalmost indistinguishable from those of other hepatic neo-plasms, especially HCC. As Taiwan is in an endemic areafor HCC, and the patient was a case of chronic hepatitis B,it was a difficult task to make an accurate diagnosis of hep-atic splenosis by abovementioned image modalities in suchhigh-risk patient. When splenosis is suspected, although his-tology is the gold standard for the diagnosis, scintigraphywith sensitive 99m Tc-labeled heat-denatured red blood cellsrepresents a diagnostic technique of high specificity (5).

This case report demonstrates that hepatic splenosis shouldbe considered in the differential diagnosis of hepatic nod-ules in a patient with splenectomy, particularly when the le-sions are located at the periphery of the liver. In addition, the99m Tc-labeled heat-denatured red blood cell scintigraphycan confirm the diagnosis of hepatic splenosis, and precludeunnecessary interventions such as biopsy, angiography, andoperation.

Hsueh-Chieh Lu, M.D.1

Chien-Wei Su, M.D.1,3,4

Ching-Liang Lu, M.D.1,4

Cheng-Pei Chang, M.D.2

Han-Chieh Lin, M.D.1,4

1Division of GastroenterologyDepartment of Medicine and

2Department of Nuclear MedicineTaipei Veterans GeneralHospital, Taipei, Taiwan

3Institutes of Clinical Medicine4Faculty of Medicine, National Yang-Ming

University, Taipei, Taiwan

REFERENCES

1. Buchbinder JH, Lipkoff CJ. Splenosis: Multiple peritonealsplenic implants following abdominal injury. A report of acase and review of the literature. Surgery 1939;6:927–34.

2. Fleming CR, Dickson ER, Harrison EG. Splenosis: Auto-transplantation of splenic tissue. Am J Med 1976;61:414–9.

3. Ferraioli G, Di Sarno A, Coppola C, et al. Contrast-enhancedlow-mechanical-index ultrasonography in hepatic splenosis.J Ultrasound Med 2006;25:133–6.

4. DeVuysere S, Van Steenbergen W, Aerts R, et al. In-trahepatic splenosis: Imaging features. Abdom Imaging2000;25:187–9.

5. Gunes I, Yilmazlar T, Sarikaya I, et al. Scintigraphic de-tection of splenosis: Superiority of tomographic selectivespleen scintigraphy. Clin Radiol 1994;49:115–7.

Mediastinal Mycobacteriumavium-intracellulare InfectionDiagnosed by TransesophagealEndoscopic Ultrasound-GuidedFine-Needle Biopsy

TO THE EDITOR: As a relatively recently implementedmodality for sampling mediastinal lesions, the diagnosticspectrum of transesophageal endoscopic ultrasound (EUS)-guided fine-needle aspiration/biopsy (FNAB) is yet to be fullydefined at this anatomic site.

While studies have documented the effectiveness of EUS-FNAB sampling of mediastinal lymph nodes for stag-ing of lung carcinoma, few reports have described theutility of EUS-FNAB in infectious processes of the me-diastinum. We report the case of a 46-yr-old African-American male patient who presented to our institutionwith the chief complaint of shortness of breath. His med-ical history was significant for stage C3 human immun-odeficiency virus (HIV) infection, and an immunodefi-ciency panel revealed a CD4 lymphocyte count of 24.A computed tomography scan of the chest showed amediastinal mass that was felt to represent adenopathy(Fig. 1). Due to the patient’s dyspnea, bronchial washing wasperformed.

Figure 1. Computed tomography scan showing the posterior medi-astinal mass.


Microscopic examination of the bronchial fluid revealedsparse squamous cells, and the specimen was considered in-sufficient for cytologic diagnosis. The remainder of the fluidwas submitted for bacterial, viral, and mycobacterial cultures.At that time, a transbronchial biopsy of the enlarged subcari-nal lymph nodes was attempted.

Microscopic review of the sampled tissue proved it to beinsufficient for diagnosis.

With culture results pending, the patient underwent trans-esophageal EUS-FNAB of the mediastinal mass. During theprocedure, a mass was identified in the area of the distal leftatrium and extended superiorly to the subcarina. By ultra-sonography, the mediastinal mass was felt to consist of mul-tiple enlarged lymph nodes, the largest measuring 3.5 × 1.7cm. Three core biopsies were obtained by EUS-FNAB, andas the differential diagnosis included lymphoma, one biopsywas submitted for flow cytometric immunophenotypic anlay-sis, but was insufficiently cellular for diagnosis. Microscopicexamination of the remaining core biopsies revealed sheets ofspindle cells that had relatively bland, elongated nuclei (Fig.2). Based upon the histologic appearance, the differential di-agnosis included a benign fibrous proliferative lesion, a low-grade sarcoma, and a proliferation of epithelioid histiocytes.Immunohistochemical stains for cytokeratin, smooth muscleactin, S-100 protein, and CD34 performed to investigate thehistogenesis of the lesion were negative. Special stains foracid-fast and fungal organisms, however, highlighted numer-ous elongated acid-fast organisms in the spindle cells, con-firming the spindle cells as histiocytes filled with mycobac-terial organisms. Subsequently, the pending respiratory andblood cultures speciated the organisms as Mycobacteriumavium-intracellulare (MAI).

Figure 2. Biopsy of the mediastinal lymph nodes showing sheets ofspindle cells with focal necrosis (Hematoxylin and eosin, ×200).

While EUS-FNAB was initially described for the diagnosisof gastrointestinal lesions, its diagnostic utility has been ex-panded for the sampling of mediastinal lesions, particularlyat lymph node stations 4 (lower paratracheal), 5 (subaortic/anteroposterior window), 7 (subcarinal), 8 (paraesophagealbelow the carina), and 9 (pulmonary ligament) in the Moun-tain/Dresler classification (1). The utility of EUS-FNAB forstaging of non-small cell carcinoma of the lung has been de-scribed (2), but its use in the diagnosis of infectious mediasti-nal lesions has rarely been documented. Sampling of the me-diastinal lymph nodes using this technology has documentedinfection with Nocardia spp., Histoplasma capsulatum, Can-dida spp., and the mycobacterial organisms M. tuberculosisand M. kansasii (3–5), but to our knowledge, this is the firstcase in which MAI has been isolated. While other samplingmodalities such as transbronchial ultrasound-guided needleaspiration have well-established utility in the diagnosis ofboth neoplastic and infectious mediastinal lesions at lymphnode stations 2 (upper paratracheal), 4 (prevascular and retro-tracheal), 7 (subcarinal), 10 (hilar), and 11 (interlobar), lowersubcarinal and paraesophageal lesions frequently exceed thepossible range of samplings available with these modalities.Even in the present case, a transbronchial biopsy of the sub-carinal lymph node was unsatisfactory. The current case isa valuable example of the utility of EUS-FNAB in samplinginfectious lesions of the mediastinum.

James Elliot Carter, M.D.1

Matthew Eves, M.D.2

Javier A. Laurini, M.D.1

1Department of Pathology, Universityof South Alabama and

2Diagnostic Medical ClinicMobile, Alabama

REFERENCES

1. Mountain CF, Dresler CM. Regional lymph node classifica-tion for lung cancer staging. Chest 1997;111:1718–23.

2. Annema JT, Versteegh MI, Veselic M, et al. Endoscopicultrasound-guided fine-needle aspiration in the diagnosisand staging of lung cancer and its impact on surgical staging.J Clin Oncol 2005;23:8357.

3. Chaya CT, Schnadig V, Gupta P, et al. Endoscopicultrasound-guided fine-needle aspiration for diagnosis of aninfectious mediastinal mass and/or lymphadenopathy. En-doscopy 2006;38(Suppl 2):E99–101.

4. Wiersema MJ, Chak A, Wiersema LM. Mediastinal histo-plasmosis: Evaluation with endosonography and endo-scopic fine-needle aspiration biopsy. Gastrointest Endosc1994;40:78–81.

5. Prasad VM, Erickson R, Contreras ED, et al. Spon-taneous Candida mediastinitis diagnosed by endoscopicultrasound-guided, fine-needle aspiration. Am J Gastroen-terol 2000;95:1072–5.

c published by blackwell publishing letters to the … in... · published by blackwell publishing...

Documents