Reducing Harms in Lung Cancer Screening
Bach to the Future
Michael ln cze, MD, MSEd: Rita F. Redberg, MD, MSc
TbeUS PreventativeServices Task Force cmrcntly recom mends si:;ree ning (grade Brecommendation)for lung canc er witha nnuallow-dose computed tomo graph}’ for high-risk in dividuals ages55 to 80 years, defined as those having greate r
gLblefor LCS using the Bach risk tool,11 a vaJidatcd risk model usingsex,age, smokingduration, durationof abstinence from smoking and number of cigarettes smoked per day as inpu ts.
The asto undingly high ratesof false-pos itiveresults in the low
=Related attid e
than a 30 pack-year cumula tivesmoking historyand h av• ing quit with in the past 15 years.1 The evide nce to sup
est risk quintiles (eg, 2221false-positive resul ts per lung ca n cer death averted and a NNS of nearly 5600 in quintile1), as well as extremelylow ratesoflungcancerincidencein the low est-risk groups, confirm trends illustrated in previous stud
port thisrecommendation overwhelminglycomes rrom the Na
tional Lung CancerScreenfngTrial(NL ST). While3 other large randomized clinical trials failed to show any mortality ben efit tolung cancer screening (LCS), the NLST demonstrateda 20% reduction in lungcan ce r mortality,a lo ng with a 6.7% re duction in .ill-ca use mortality, when compared with an an nual chest radiograph, witb a number needed toscreen (NNS} of256to prevent I lung-cancerassociated death over3years.-2 5 Real-worldapplication ofLCS has been particularly chal lenging, w ith evidence of inappropriate U.% in low-risk per sonsalong with very high rates.o fincidental lindings a.nd false positive resultsleadiag to a muchhigher harm o benefit ratio than what was seen in randomized clinical trials.·2 7 The most robust implementation data to date,conducted on alargehigh risk population through the Veterans Health Administration system, dcmonstr.itc-0 that 56% of those screened had nod ules requiring follow-up with repeated imaging and/or inva siveprocedures, and 40%ha.d incidentalfindings,such asem physema and coronary arte ry calcificatio, 11 with a relatively
!ow cm ce r detection rate of 1.5% (even lower for early-stage
-cancers tbat mostbenefit fro m sc;r,eening).GTncreasing l y, there has beenconcern aboutboth the cost-effectiven essofLCSand the harms associated with screening, including high rates of incidentalfindings (resulting in unnecessary i.nvasive proce dures and emotionalstress), as well as radiation ex posure (pos sibly leading tosecondar y malignant neop!asm.7 9 1 hese flild ings raise the question of whethe r a more targeted inclusion criteria maydecrease the total numberofpatientsscreened and the false-posi tivity ra te withoutsacrificingt.he mortality ben efit seen in the NLST.
It is against this background that Caverly ct al10 examine
the impact of riskstratifying a real-world cohort of patientseli·
ies and make lhe case for refined guidelines for LCS.1 .n This is es pecially salienti n light of recent evi dence suggesting t ha t bl,gh numbers oflow-riskindividuals arebeingscreenedin.real· world practice.7
Unfortunately, the useof LDCTscreeningevenin the high est -risk quintiles isstill associated with alarm ingly high rates of false-positive results-302 false-positive resqJts for every lung cancer death prevented (NNS, 552), in addition to high rates ofincidental findingspotentiaUy req uiringinterven lion as noted in theoriginalcohort study.6 Furthem1ore, while the author co rrectly poi.n,t out large differences in sc ree ning ef ficacy and benefit to risk ratio between the highest and low est risk quintiles, thedata are murkier in between, with greater than 500 false-positive res ults per lung cancer death pre vented foraII b ut the highest-risk quintile, mea.ning th at most participants are exposed to an unfavorable benefit to risk ra tio.
Caverly et a.110 have made the important contribution of applying a validated risk stratification tool toa real-world co hort to improve sc reening c rite ria. Other groups have retro spectively demonstrated reduced hamis in stu dy popula tions with novel1iskstratificationtoolsusingeasilyobtainable patient infomiation, such as em physema diag1l0s is, s moking history,age,sex,and family history.’2·’ 3 However,dearly more work is needed to minimizethe harms of radiation eiq,osure, invasive procedure,sand emotionalstressundercurrentguide Jines , while preservingbenefit for those whose livescould be saved by the early detection oflung cancer. Ifand howwe will get there has yet to be deterrni.necl, but one thing is d ear: the future of LCSdepends on our ability to reexamine and refine our approach to patient selection and clearly commun.icate risks and benefits of screening.
ARTICLE INF ORMATION
Author Afffliatlons: partmellt of Medicine, Uniot!rsityof C lifomia,S/ln Francisco. Schoolcl Medicine.SanFrancis.co (rncre,Redberg}:l:di tor,
JAMA lntemcl Medi6ne (Re<Jberg).
Corrcspondnl g Author: AltaF.Re<lbcrg, MO, MSc, Departm I of Medicine, Universrtyof Calirorni a, Sa11Fralldsm,5d1oolof Medicine, 505 f>amassus, M i l 80 , SanFrancisco, CA9414 3·0124-(r it a.redberg
f>ubllshedOnHnc : January22. 20 18 .
doi:10.1001/ Jan; ainternrni:d.20T7.8217
Conflict ofl nterest Di.sdosur : Nooe l'<!p<,rted.
Jamaintemalmedidne.com JAMA l llt .ern alMed[dnc Publishedonline Janua1y 22,2018 El
© 2018 American Medical Association. All rights reserve,d.