7/27/2019 jvp130101

http://slidepdf.com/reader/full/jvp130101 1/2

Overdiagnosis and Overtreatment in Cancer An Opportunity for Improvement

Overthe past 30 years, awareness andscreening haveled to an emphasis on early diagnosis of cancer. Al-thoughthegoalsoftheseeffortsweretoreducetherateoflate-stagediseaseanddecreasecancermortality,secu-lartrendsandclinicaltrialssuggestthatthesegoalshavenotbeen met;nationaldatademonstrate significant in-creasesinearly-stagedisease,withoutaproportionalde-cline in later-stagedisease.Whathasemerged hasbeenanappreciationof thecomplexityof thepathologic con-ditioncalledcancer.Theword“cancer”often invokes thespecter of an inexorably lethal process; however, can-cers are heterogeneous and can follow multiple paths,not allof which progress to metastases and death,andinclude indolentdisease thatcauses noharm during thepatient’s lifetime.Better biology alone can explain bet-teroutcomes.Although thiscomplexitycomplicatesthegoal of early diagnosis, its recognition provides an op-portunitytoadaptcancerscreeningwithafocusoniden-tifyingand treating those conditions mostlikelyassoci-atedwith morbidityand mortality.

Changes in cancer incidence and mortality 1 reveal3 patterns that emerged after inception of screening(Table ). Screening for breast cancer and prostate can-cer appears to detect morecancers thatare potentiallyclinically insignificant. 4 Lungcancermayfollowthispat-tern if high-risk screeningis adopted. 5 Barrett esopha-gusand ductal carcinoma of thebreastare examplesforwhich thedetectionandremoval of lesions consideredprecancerous have not led to lower incidence of inva-

sive cancer. Incontrast,colonandcervicalcancerareex-amples of effective screening programs in which earlydetectionand removalof precancerous lesionshavere-duced incidence as well as late-stage disease. Thyroidcancers and melanomaare examples for which screen-inghasexpanded and,alongwith it,thedetection ofin-dolent disease.

Optimal screeningfrequency depends on thecan-cer’s growth rate. If a cancer is fast growing, screeningis rarely effective. If a cancer is slow growing but pro-gressive,with a long latency and a precancerouslesion(eg, colonic polyps or cervical intraepithelial neopla-sia), screening is ideal and less frequent screening (eg,

10 years forcolonoscopy) maybe effective. In thecaseofanindolent tumor,detection ispotentiallyharmfulbe-cause it can result in overtreatment. These observa-tionsprovidean opportunityto refocusscreeningon re-ducing disease morbidity and mortality and lower theburdenof cancer screeningand treatments.

In March 2012, the National Cancer Institute con-vened a meeting to evaluatethe problem of “overdiag-nosis,” which occurs when tumorsare detected that, if left unattended, would not become clinically apparentor cause death. Overdiagnosis, if not recognized, gen-

erally leads to overtreatment. This Viewpoint summa-rizes the recommendations from a working groupformedtodevelopastrategytoimprovethecurrentap-proach to cancer screening and prevention.

Periodic screeningprograms have the potential toidentify a reservoir of indolent tumors. 4 However, can-cer is still perceived as a diagnosis with lethal conse-quences if left untreated.

An ideal screening intervention focuses on detec-tion of disease that will ultimately cause harm, that ismore likely to be cured if detected early, and for whichcurativetreatmentsare moreeffectivein early-stagedis-ease.Goingforward, theabilityto design better screen-ing programs will depend on the ability to better char-acterize thebiology of thedisease detected andto usediseasedynamics(behaviorover time)and moleculardi-agnosticsthatdeterminewhether cancerwill beaggres-sive or indolent to avoid overtreatment. Understand-ing the biology of individual cancers is necessary tooptimize early detection programs and tailor treat-ments accordingly. The following recommendationswere made to theNational Cancer Institutefor consid-eration and dissemination.

Physicians,patients,and thegeneralpublicmust rec-ognizethatoverdiagnosisiscommonandoccursmorefre-quently with cancer screening . Overdiagnosis, or iden-tification of indolentcancer, is common in breast, lung,prostate, and thyroid cancer. Whenever screening isused, thefractionof tumors in this category increases.

By acknowledging this consequence of screening, ap-proaches that mitigate the problem canbe tested.

Changecancerterminology based oncompanion di-agnostics . Useof theterm “cancer”should be reservedfordescribinglesions witha reasonable likelihood of le-thal progression if left untreated. There are2 opportu-nities forchange.First,premalignantconditions (eg,duc-talcarcinomain situor high-gradeprostatic intraepithelialneoplasia) should not be labeled as cancers or neopla-sia, nor should the word “cancer” be in the name. Sec-ond,moleculardiagnostictoolsthatidentify indolentorlow-risk lesions need to be adopted andvalidated. An-other step is to reclassify such cancers as IDLE (indo-

lent lesions of epithelial origin) conditions.4

An ex-ample is the reclassification of grade 1 papilloma tourothelial neoplasia of low malignant potential. 6 Pre-sciently, the rationale for reclassifying papilloma andgrade1carcinomaas“papillaryurothelialneoplasiaoflowmalignant potential” was “to take the lowest grades of tumor,the mostbenign-appearinglesions, andremovethe wordcarcinoma.” 6 A multidisciplinary effort acrossthe pathology, imaging, surgical, advocate, and medi-cal communities could be convened by an indepen-dent group (eg, the Instituteof Medicine)to revisethe

LauraJ. Esserman,MD,MBAUniversityof California,San Francisco.

Ian M. Thompson Jr,MDUniversityof TexasHealthScienceCenterat SanAntonio.

BrianReid,MD,PhDFredHutchinsonCancer ResearchCenter,Seattle,Washington.

CorrespondingAuthor: LauraJ.Esserman, MD,MBA,MtZion Carol FrancBuck BreastCareCenter,University of California, SanFrancisco,1600DivisaderoSt, POBox1710,San Francisco, CA94115 (laura.esserman@ucsfmedctr.org).

VIEWPOINT

Opinion

jama.com JAMA Published onlineJuly 29,2013 E1

wnloaded From: http://jama.jamanetwork.com/ by a Universidade Federal do Espírito Santo User on 07/29/2013

7/27/2019 jvp130101

http://slidepdf.com/reader/full/jvp130101 2/2

taxonomy of lesions now called cancer and to create reclassifica-tioncriteria for IDLEconditions. 7

Create observational registries for low malignant potential le- sions. Providingpatientsand clinicianswith pathologicdiagnosisand

information related to disease prognosis is crucial to informed de-cision making, including comfort with alternate treatment strate-giessuch asactive surveillance.Prognosisfor precancerous lesions

includesthe riskof developmentofinvasive cancer, theperiodoverwhich such a tumor would develop, andthe prognosisof that typeof tumor should it occur.Prognosis forinvasive cancer includesriskand timing of development of metastatic disease and death. Largeregistriesforpotentiallyindolentconditionswouldprovidedata link-ing disease dynamics 8 (eg, tumor growthrate over time) and diag-nosticsneeded to providepatientsand physicians withconfidence

to selectless invasive interventions.Mitigateoverdiagnosis . Strategiesto reducedetection of indo-lent disease include reducing low-yield diagnostic evaluations ap-propriately, reducing frequency of screening examinations, focus-ing screeningon high-riskpopulations, raising thresholds for recallandbiopsy, andtesting thesafetyand efficacyof risk-based screen-ingapproaches to improve selectionof patients forcancer screen-ing.The ultimategoal is topreferentiallydetect consequentialcan-cer while avoidingdetection of inconsequential disease.

Expand theconceptof howto approachcancerprogression .Fu-ture research should include controlling the environment in whichprecancerous and cancerous conditions arise, as an alternative tosurgical excision.

ConclusionThe original intentof screeningwas todetectcancer at theearlieststagestoimproveoutcomes;however,detectionofcancerswithbet-ter biology contributes to better outcomes. Screening always re-sultsin identifyingmoreindolentdisease. Althoughnophysicianhastheintentiontoovertreatoroverdiagnosecancer,screeningand pa-tient awareness have increased the chance of identifying a spec-trumof cancers,someof which arenot lifethreatening.Policiesthatpreventor reduce thechanceof overdiagnosis andavoidovertreat-ment areneeded, while maintainingthose gains bywhichearlyde-tectionis a major contributor todecreasingmortalityandlocally ad-vanceddisease.The recommendationsof thetaskforceare intendedasinitial approaches.Physicians andpatientsshould engagein open

discussionaboutthesecomplex issues.The mediashouldbetterun-derstand andcommunicatethemessageso that asa communitytheapproach to screeningcan be improved.

ARTICLE INFORMATION

Published Online: July 29,2013.doi:10.1001/jama.2013.108415.

Conflict of Interest Disclosures: All authors havecompletedand submitted theICMJE Form forDisclosure of PotentialConflicts of Interest.DrThompsonreported serving as a board memberorconsultantfor, and receiving grants or grantspending, payment for lectures, patents, andhonorariafrom,a variety of sources. Nootherauthors reported disclosures.

Additional Contributions: We thank BarnettKramer,MD, MPH(Division of Cancer Prevention,National Cancer Institute[NCI]), for conveningthebrainstorming meeting heldin March2012. Theauthors chaired theNCIworking group,whichincluded DonaldBerry,PhD, Mina Bissel, PhD,WilliamBlack, MD,ShelleyHwang, MD,KennethKinzler,PhD, Peter Nelson, MD,David Ransahoff,PhD, Howard Parnes, PhD, Sudhir Srivastava,PhD,and Gilbert Welch, MD.Follow-up discussions andrecommendations form thebasis of thisarticle.

Nadarajen A. Vydelingum, PhD, FSB, FRSPH(Division of Cancer Prevention, NCI),providedadministrative supportand coordinationamong theauthors throughout thepreparationof this article.

REFERENCES

1. Howlader N, NooneAM, Krapcho M,et al,eds. SEER Cancer Statistics Review, 1975-2010 .http://seer.cancer.gov/csr/1975_2010/. April 2013.Accessed July 10,2013.

2. Berry DA,Cronin KA,PlevritisSK, etal. Effectof screeningand adjuvant therapy on mortality frombreast cancer. N EnglJ Med . 2005;353(17):1784-1792.

3. Kalager M,ZelenM, LangmarkF, Adami HO.Effect of screeningmammography onbreast-cancermortality in Norway. N EnglJ Med .2010;363(13):1203-1210.

4 . Esserman L, ShiehY, Thompson I. Rethinkingscreeningfor breast cancer and prostate cancer. JAMA. 2009;302(15):1685-1692.

5. Aberle DR,AdamsAM, Berg CD, etal. Reducedlung-cancer mortality withlow-dose computedtomographic screening. NEngl JMed .2011;365(5):395-409.

6 . EpsteinJI, Amin MB,Reuter VR,Mostofi FK.TheWorldHealth Organization/International Society of Urological Pathology consensus classificationof urothelial (transitional cell)neoplasmsof theurinarybladder. Am J Surg Pathol .1998;22(12):1435-1448.

7. Committeeon a Framework forDevelopmentof a New Taxonomy of Disease; National ResearchCouncil. TowardPrecisionMedicine: Building aKnowledgeNetwork for Biomedical Research and aNew Taxonomy of Disease. Washington, DC:National Academies Press;2011.

8 . Welch HG, BlackWC. Overdiagnosis in cancer. J Natl Cancer Inst . 2010;102(9):605-613.

Table.Changein Incidenceand Mortalityof Cancers Over TimeFrom1975 to 2010as Reportedin Surveillance, Epidemiology and End Results 1

Change a

Incidence Mortality

Per 100 000%

Change

Per 100 000%

Change1975 2010 b 1975 2010 b

Example 1

Breast c 105.07 126.02 20 31.45 21.92 −30

Prostate 94 145.12 54 30.97 21.81 −30

Lung and bronchus d 52.26 56.68 8 42.56 47.42 11

Example 2

Colon 41.35 28.72 −31 28.09 15.51 −45

Cervical 14.79 6.71 −55 5.55 2.26 −59

Example 3

Thyroid 4.85 13.83 185 0.55 0.51 −7

Melanoma 7.89 23.57 199 2.07 2.74 32

a Example 1: Indolent and consequential tumorsare identified withscreening,leading to an overall increase in incidence rates. Example 2: Prescreenedtumorpopulationis more homogeneous, slower-growing but consequential.Screening substantially decreases incidence (through detection and removalof precursor lesions) and mortality. Example 3: Screeningexpands thepopulationof indolent tumors,withlittle or no effecton thesmall populationof more aggressive tumors.

b Representsperiod in whichscreening (except for lungcancer) is prevalent.c At leasttwo-thirds of themortality reduction is believed attributableto

adjuvant therapy. 2,3

d The National Lung ScreeningTrial conductedamongindividualsat riskfor lungcancers showsthat theproportion of stageI detected tumorsis more than2-fold higher thanthe decrease in thehigher-stagetumors, accountingfor itsinclusion in example 1. 5

Opinion Viewpoint

E2 JAMA Published online July 29, 2013 jama.com

wnloaded From: http://jama.jamanetwork.com/ by a Universidade Federal do Espírito Santo User on 07/29/2013