Challenges in evaluation of screening for gastric cancer among men based on nonrandomized design

Full Te ms & Condi ions o access and use can be ound a h p://www. and online.com/ac ion/jou nalIn o ma ion?jou nalCode=ionc20 Download by: [Tampe e Uni e si y] Da e: 11 June 2017, A : 22:06 Ac a Oncologica ISSN: 0284-186X (P in ) 1651-226X (Online) Jou nal homepage: h p://www. and online.com/loi/ionc20 Challenges in e alua ion o sc eening o gas ic cance among men based on non andomized design Ilkka Vohlonen, Ma i Hä könen, Nea Malila, Ee o Pukkala, Pen i Sipponen, Veli Kois inen & Ma i Hakama To ci e his a icle: Ilkka Vohlonen, Ma i Hä könen, Nea Malila, Ee o Pukkala, Pen i Sipponen, Veli Kois inen & Ma i Hakama (2017) Challenges in e alua ion o sc eening o gas ic cance among men based on non andomized design , Ac a Oncologica, 56:7, 917-922, DOI: 10.1080/0284186X.2016.1278304 To link o his a icle: h p://dx.doi.o g/10.1080/0284186X.2016.1278304 © 2017 The Au ho (s). Published by In o ma UK Limi ed, ading as Taylo & F ancis G oup Published online: 23 Feb 2017. Submi you a icle o his jou nal A icle iews: 118 View ela ed a icles View C ossma k da a ORIGINAL ARTICLE Challenges in e alua ion o sc eening o gas ic cance among men based on non andomized design Ilkka Vohlonen a , Ma i H€ a k€ onen b , Nea Malila c,d , Ee o Pukkala e, , Pen i Sipponen g , Veli Kois inen h and Ma i Hakama i a Depa men o Public Heal h, Heal h Policy, Uni e si y o Eas e n Finland, Kuopio, Finland; b Depa men o Clinical Chemis y, Clinical Chemis y, Uni e si y o Helsinki, Helsinki, Finland; c Cance Epidemiology, Finnish Cance Regis y, Helsinki, Finland; d School o Heal h Sciences, Uni e si y o Tampe e, Tampe e, Finland; e Epidemiology, Finnish Cance Regis y, Helsinki, Finland; School o Heal h Sciences, Uni e si y o Tampe e, Tampe e, Finland; g Depa men o Pa hology, Pa olab Oy, Espoo, Finland; h Depa men o Bios a is ics, Finnish Consul ing G oup, Helsinki, Finland; i Depa men o Epidemiology, Finnish Cance Regis y, Helsinki, Finland ABSTRACT Backg ound: Objec i e was o quan i y biases in sc eening o gas ic cance when compa ing a end- e s o nona ende s using se um pepsinogen I (SPGI) le el as p ima y es . Me hods: In mid 1990s, all men aged 51–65 yea s om wo Finnish ci ies we e in i ed o SPGI sc een- ing. Mo ali y and p ema u e mo ali y in a ende s we e compa ed o nona ende s. E icacy o sc een- ing was s udied by 15 yea s’ ollow-up o s anda dized mo ali y a io (SMR) and po en ial yea s o li e los (PYLL) due o gas ic cance . Bias due o selec i e a endance was quan i ied using co ec i e coe - icien s based on o al cance incidence and mo ali y, and gas ic cance -speci ic incidence and mo al- i y o o al popula ion and nona ende s. Resul s: In 1994–1996, men aged 51–65 yea s (16,872) we e in i ed o SPGI assay and 12,175 men (72%) a ended. SPGI was 25 mic og/l o less in 610 (5%) men, indica ing se e e a ophic gas i is (AG). Pos -sc eening gas oscopy was pe o med o 435 men wi h low SPGI. O hese, 168 men we e e e ed o ea men due o abno mal ocal lesions. A ibu able p opo ions in educ ions o SMR and PYLL om gas ic cance due o sc eening we e 59% and 67%. A e co ec ing o selec i e pa icipa ion, a ibu able p opo ions we e educed o 23% and 39%. Conclusions: Bioma ke sc eening by low SPGI among middle-aged men ollowed by uppe gas o- in es inal endoscopy dec eased long- e m and p ema u e mo ali y due o gas ic cance . Howe e , in spi e o me hodological co ec ions done, he esul s do no jus i y any i m conclusions o ecommend gene al sc eening p og ams. Randomized ials a e wa an ed o his pu pose. Abb e ia ions: SPGI: se um pepsinogen I; AG: a ophic gas i is; SMR: s anda dized mo ali y a io; SIR: s anda dized incidence a io; PYLL: po en ial yea s o li e los ARTICLE HISTORY Recei ed 9 Oc obe 2016 Accep ed 28 Decembe 2016 In oduc ion In Finland, a popula ion-based sample o middle-aged men om wo ci ies was in i ed in 1994–1996, o in es iga e whe he a se um pepsinogen I (SPGI) sc eening, ollowed by uppe gas oin es inal (GI) endoscopy in hose wi h low SPGI, can be applied as a public heal h p og am. I was known ha SPGI le els o 25 mic og/l o less indica e ad anced (mode a e o se e e) a ophic gas i is (AG) in he gas ic co - pus and undus wi h high eliabili y [1–4]. Howe e , he indi- ca o s o s uc u e, p ocess, and ou come ela ed o sc eening we e comple ely unknown. The objec i e o his s udy was o ind ou whe he he ecen labo a o y e idence o he e iological ole o SPGI was su icien o s a sc eening and o quan i y challenges due o bias in he e alua i e design. The sc eening was based on pe sonal in i a ion o he a ge popula ion bu i did no con ain any con ols. We s udied gas ic cance mo ali y by compa ing he a ende s in sc eening wi h he nona ende s o e mo e han 15 yea s ( om 1994 o 2011) a e sc eening in 1994–1996. Since he design did no allow o a non- biased assessmen o e icacy, s a is ical co ec ions we e pe - o med o adjus o he e ec s o selec i e pa icipa ion. Ma e ial and me hods S udy popula ion and s udy coho s All 16,872 men bo n in 1929–1943 and li ing in 1994–1996 in wo Finnish ci ies (Ko ka and Van aa) we e iden i ied by popula ion egis y, and we e in i ed o gi e a blood sample (se um) o he SPGI es : hal o hem in au umn 1994 and he emainde in au umn 1996 (Table 1). Al oge he , 12,175 CONTACT Ilkka Vohlonen [email p o ec ed] Depa men o Public Heal h, Uni e si y o Eas e n Finland, B.O.X. 1627, 70100 Kuopio, Finland Depa men o Bios a is ics, Finnish Consul ing G oup, Helsinki, Finland. ß2017 The Au ho (s). Published by In o ma UK Limi ed, ading as Taylo & F ancis G oup This is an Open Access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion-NonComme cial-NoDe i a i es License (h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/), which pe mi s non-comme cial e-use, dis ibu ion, and ep oduc ion in any medium, p o ided he o iginal wo k is p ope ly ci ed, and is no al e ed, ans o med, o buil upon in any way. ACTA ONCOLOGICA, 2017 VOL. 56, NO. 7, 917–922 h p://dx.doi.o g/10.1080/0284186X.2016.1278304 men (72%) a ended he SPGI sc eening and 4697 did no . They o m he a ende s (sc eened) and nona ende s (no sc eened) in he p esen in es iga ion, espec i ely (Figu e 1). Blood sampling and SPGI es The men we e in i ed by mail o isi municipal heal h cen- e s o blood sampling d awn by expe ienced labo a o y nu ses. Fas ing se a we e collec ed in ap o inin (T asylol, Baye Ge many, 200 KIE/ml) con aining Venojec ubes and s o ed a –70 C un il analyzed. SPGI was analyzed using he speci ic ELISA es s p o ided by Biohi Oyj, Helsinki, Finland. The assay has been calib a ed o co espond o esul s ob ained by adioimmunoassay (RIA) used by Samlo in 1982 using 238 se um samples wi h se um pepsinogen concen a ions be ween 1.5 and 120 mic og/L. The clinical sensi i i y and speci ici y o ad anced AG a e 92% and 90% o he es , espec i ely, acco ding o he man- u ac u e ’s ki ins uc ions o use. The sc eening was app o ed by he E hics Commi ee o he Uni e si y Hospi al o Kuopio in 1994. Low SPGI le els (25 mic og/l o lowe ), indica ing he p es- ence o ad anced (mode a e o se e e) AG in gas ic co pus and undus, we e ound in 610 men (5% o he sc eened men). Men wi h low SPGI le els we e con ac ed by elephone and in i ed o gas oscopy unless hey had medical con aindica ions. Sc eening endoscopy in 1994–1996 A diagnos ic uppe GI endoscopy (sc eening gas oscopy) was pe o med in 435 SPGI- es posi i e men; i.e., 71% o he men wi h a low SPGI (Figu e 1). All hese endoscopies we e pe o med by one expe ienced clinician. In endoscopy, biop- sies om he s omach we e collec ed: wo om he an um and wo om he co pus, acco ding o he guidelines o he Sydney Sys em. Specimens o his opa hology we e also aken om all ocal abno mali ies (e.g., e osions, ulce s, nod- ules, polyps, masses, unusual colo spo s, e c.) seen by an endoscopis in he s omach mucosa. In all, 168 men (1.4% o all men in he sc eened coho ) we e e e ed o u he clin- ical ea men o su eillance. O hese, 56 men had ocal gas ic lesions wi h abno mal and a ypical mo phology in biopsy pa hology, and all hese we e conside ed po en ially cance ous o p ecance ous lesions (Figu e 1). Follow-up and collec ion o mo ali y da a in 1994–2011 Long- e m ollow-up o all men in i ed o sc eening in 1994–1996 was un o cance incidence, dea hs, o emig a- ion. These men o med he a ende s (sc eened) and nona - ende s (nonsc eened) coho s as explained abo e. The ollow-up s a ed he mon h a e s a o sc eening (on 1 Decembe 1994 o he hal o he men and on 1 No embe Figu e 1. Design o he easibili y s udy on sc eening o gas ic cance in middle-aged men in wo Finnish ci ies in 1994–1996 using a bioma ke . Table 1. Numbe and pe cen age o men, numbe o accumula ed pe son yea s, and mean ime pe pe son in ollow-up in 1994–2011 by sc eening s a us. Men Pe son yea s Sc eening s a us Numbe Pe cen Numbe Mean/pe son Non-sc eened 4697 27.8 54,955 11.7 Sc eened 12,175 72.2 166,798 13.7 To al in i ed 16,872 100.0 221,023 13.1 918 I. VOHLONEN ET AL. 1996 o he o he hal ) and ended a dea h, emig a ion o on 31 Decembe 2011 (Table 1). In o ma ion on gas ic can- ce cases and dea hs in he coho s was ecei ed om he Finnish Cance Regis y (FCR) and da a on emig a ion and da e o dea h was om he popula ion egis e . The cause o dea h in o ma ion o igina es om S a is ics Finland [5]. The indi iduals in he s udy we e linked o egis e s using he na ional pe sonal iden i y code o all Finnish esiden s. Due o manda o y epo ing o all cance diagnoses in Finland, he FCR has a good na ional co e age o o e 99% o solid cance cases in Finland [6–8]. The egis y includes da a o all gas ic cance cases ( opog aphy code C16 in ICD- O-3) bu i does no include da a o cases classi ied as nonin- asi e p emalignan lesions (dysplasia o in amucosal neo- plasia). The cance in o ma ion used in his s udy included he da e o diagnosis, opog aphy, mo phology, and da a indica ing whe he he cance pa ien died om he cance in ques ion o om any o he cause. S a is ical me hods The s anda dized mo ali y a ios (SMRs) and po en ial yea s o li e los (PYLL) we e calcula ed o gas ic cance and o all cance s combined (cance a any si e) [6,9] o a ende s and nona ende s sepa a ely. In o de o calcula e he SMR, he obse ed numbe s o dea h we e compa ed wi h he expec ed numbe s among he a ende s and nona ende s. The expec ed numbe s o each age g oup (5-yea age ca e- go ies) and 4-yea calenda pe iod we e es ima ed by mul i- plying he numbe o pe son yea s in he ca ego y accumula ed among he a ende s and he nona ende s coho s wi h he espec i e mo ali y a e in he o al male popula ion in Finland. The 95% con idence in e als o SMRs we e es ima ed assuming a Poisson dis ibu ion o he numbe s o obse ed dea hs. S anda dized Incidence Ra ios (SIRs) we e calcula ed acco ding o he same p inci- ples as SMRs. PYLL is an indica o o p ema u e mo ali y. I ep esen s he o al numbe o li e yea s los (no li ed) by an indi idual who was no assumed o die by a gi en age. The PYLL- alue was calcula ed o each dead pe son as he di e ence be ween he obse ed da e o dea h and he expec ed leng h o li e. In he calcula ion o PYLL- alues, he expec ed leng h o li e was se a he Finnish a e age s anda d (80 yea s). Those men who died a e hei 80-yea -old bi hday did no con ibu e o he PYLL alue in any way. PYLL- alues we e classi ied acco ding o he cause o dea h (s omach cance o all cance s) and epo ed pe 1000 pe sons and pe dea h among he sc eened and he non-sc eened. S a is ical analyses Due o selec i e a endance, he e ec o sc eening, i.e., e i- cacy, was measu ed by wo ypes o a ibu able p opo ions in educ ions o SMR and PYLL, he c ude a es and he co - ec ed a es. In analyses o he a ibu able p opo ions in educ ions o SMR and PYLL, he di e ences in he geo- g aphic (e.g., supply and use o heal h se ices), social (e.g., demog aphic and e iological), and heal h (e.g., ea lie diagno- ses) cha ac e is ics o he men be ween he wo coho s we e co ec ed o , since he po en ial sel -selec ion co e- la es wi h he p obabili y o dea h om gas ic cance . The egional di e ences (heal h ca e supply) and he gene al isks o any cance we e co ec ed o by aking in o accoun he s anda dized incidence a ios and SMRs o all cance s o he o al popula ion and o he nona ende s, as desc ibed below. In o de o adjus o sel -selec ion, we deno e: M¼SMR o PYLL due o gas ic cance , I¼SIR o gas ic cance , a¼ hose sc eened, an¼ hose no sc eened, m¼SMR o PYLL due o all cance s, i¼SIR o all cance s, E1 ¼c ude e ec (a ibu able p opo ion), E2 ¼e ec wi h co ec ion o selec ion and con ounding ac o s (co ec ed a ibu able p opo ion). The c ude es ima e o he e ec (E1) was es ima ed as: E1 ¼1–(M an /M n ). The co ec ed es ima e o he e ec (E2) was es ima ed as: E2 ¼1–(I an /I a )(M an /M n ) (ma an /m a )(i an /i n ). Selec ion bias was co ec ed by he ela i e isk o back- g ound incidence o gas ic cance in sc eened and non- sc eened coho s (I an /I a ). Fu he mo e, hose a ending sc eening may ha e a be e gene al heal h and a mo e a o able p ognosis han hose no a ending. We assumed ha hese biases a e seen in he su i al and in he a io o cance mo ali y o cance incidence (m/i) in he coho s, jus- i ying he co ec ion ac o (m an i a )/(m a i an ). The same o - mulae we e used o co ec ing he di e ences in educ ions o PYLL alues due o s omach cance be ween he a ende s and nona ende s coho s by adjus ing o PYLL alues o all cance s. The mo ali y analyses o gas ic and all cance s we e i s pe o med wi h e ined SMRs, i.e., excluding dea hs om gas- ic cance diagnosed be o e he beginning o he ollow-up. The e we e only ou such dea hs in he p esen se ies. The e ined and he c ude SMR a es o all gas ic cance dea hs u ned ou o be simila independen ly o ime o he diag- nosis. Only he c ude SMRs and PYLL alues we e p esen ed including all gas ic cance s wi hou conside ing he iming o diagnosis. Resul s The SMRs and SIRs o all cance s combined (cance s a all si es) and o gas ic cance sepa a ely in bo h s udy coho s a e p esen ed in Table 2. SMR o gas ic cance (0.53) was signi ican ly below uni y in he a ende coho and i was highe in he nona ende s coho (SMR 1.28). The SMR o all cance s in he a ende coho was 0.91 and in he nona - ende s coho 1.45. The SIRs o all cance s we e oughly simila in bo h s udy coho s. Howe e , in he a ende coho , he SIR o s omach cance was lowe han expec ed (0.75, 95% CI 0.57–0.95) and ACTA ONCOLOGICA 919 i was lowe in he a ende s han in he nona ende s coho (0.88, 95% CI 0.56–1.32) (Table 2). The PYLL alue o gas ic cance was signi ican ly lowe among he a ende s (27.7, 95% CI 15.5–36.0) han among he nona ende s (77.3 95% CI 43.1–111.4) coho (Table 3). Again, he PYLL alue o all cance s was lowe in he a ende coho han in he nona ende s coho bu his di - e ence was smalle han he di e ence in PYLL o gas ic cance . The c ude and co ec ed e ec s o sc eening on mo ali y and p olonga ion o li e a e gi en in Table 4. The ela i e dec ease in SMR a ibu able o SPGI sc eening o gas ic cance esul ed in a alue 0.59 o he c ude p opo ion (E1). The co esponding c ude a ibu able p opo ion o PYLL was 0.67. A e co ec ing o selec ion bias, he co ec ed a ibu able p opo ions (E2) o SMR and PYLL we e 0.23 and 0.39, espec i ely (Table 4). Discussion Gas ic cance is one o he common cance s wi h high mo - ali y wo ldwide. The age-adjus ed incidence a e (wo ld s anda d) in 2013 among Finnish men was 6 pe 100 000 pe son yea s, which is one- en h o he espec i e a e in he 1950s [10]. The li e expec ancy o pa ien s wi h gas ic cance is no long, unless he cance is diagnosed in i s ea ly s age. In 2009–2013, he i e-yea ela i e su i al a io in males wi h s omach cance in Finland was 23% [10]. Abou a hal o gas ic cance s, being mos o en o he so-called in es inal ype, a e conside ed o de elop in AG and an acid- ee s om- ach ia he ‘Co ea cascade’[11–14]. The e o e, ea ly iden i i- ca ion and endoscopy o subjec s wi h AG may acili a e he diagnosis o gas ic cance a an ea ly s age, and may enable ea men o p emalignan gas ic lesions (in amucosal neo- plasias) in an asymp oma ic phase [6–9]. A low se um le el o pepsinogen I (SPGI) is a eliable bioma ke o AG and is, he e o e, a ool o nonin asi ely delinea e subjec s wi h ad anced AG who need a p omp diagnos ic uppe GI endos- copy because o inc eased cance isk [1–4]. Co espondingly, i is concei able ha an ea ly diagnosis o AG wi h a bio- ma ke es ollowed wi h a diagnos ic uppe GI endoscopy will imp o e he cance su i al, esul ing also in dec ease o p ema u e mo ali y. In o de o de elop a p og am o sc eening o s omach cance by a bioma ke , his s udy comp ised a la ge popula- ion-based sample o men bo n in 1929–1943 who we e in i ed o sc eening by SPGI/endoscopy in 1994–1996 in wo Finnish ci ies. This sc eening p og am applying a simple SPGI bioma ke blood es was well accep ed wi h a high pa ici- pa ion pe cen age (72%) and he sc eening p ocess was adequa e. He e we e alua e whe he he design was adequa e o e alua ion o he e icacy o he sc eening p o- g am by gas ic cance mo ali y. Biases be ween he a ende s and nona ende s coho s cons i u e a dilemma in assessmen o he e icacy o he sc eening. Because he sc eening ial could no be imple- men ed o allow compa ison o in i ed and non-in i ed g oups o men, he only op ion was o y o co ec o he e ec s o selec i e a endance a e wa ds using gene al indi- ca o s o cance equency and mo ali y in he ca ego ies o be compa ed. In gene al, he epo s o sc eening p og ams on cance s do no include da a o a emp s o co ec o selec ion biases e en hough hey a e ce ainly no ewo hy in all sc eening p ojec s [8,15]. The men who did no wan o pa - icipa e in he SPGI es may ha e been less in e es ed in heal h issues in gene al han hose who pa icipa ed and we e, he e o e, mo e liable o ea ly dea h om gas ic can- ce and o high mo ali y om any cause o dea h [16–20]. The men in he p esen nona ende coho had a clea ly highe incidence (SIR) o gas ic cance and o all cance s han hose in he a ende coho . In spi e o he co ec ions a e wa ds, he e may s ill emain biases ha could no be aken in o accoun . Table 3. Po en ial yea s o li e los (PYLL alues) pe 1000 pe sons and mean PYLL alues pe dea h be o e age o 80 among he sc eened and non-sc eened men by si e o cance 1994–2011. Sc eened (n¼12,175) Non-sc eened (n¼4697) Cance si e PYLL 95%CI PYLL/dea h PYLL 95%CI PYLL/dea h Gas ic cance 25.7 15.5–36.0 10.4 77.3 43.1–111.4 15.1 All cance s 875 815–935 11.0 1388 1259–1518 13.2 Table 4. C ude and co ec ed p opo ions (%) in educ ion o mo ali y and o po en ial yea s o li e los (PYLL) among men a ibu able o SPGI sc eening. A ibu able p opo ion (%) Ou come indica o C ude (E1) Co ec ed (E2) Mo ali y 59 23 PYLL 67 39 Table 2. S anda dized mo ali y a ios (SMR) and s anda dized incidence a ios (SIR) among he sc eened and non-sc eened men by he si e o cance 1994–2011. Sc eened (N¼12,175) Non-sc eened (n¼4697) Cance si e E en OBS EXP SMR o SIR 95%CI OBS EXP SMR o SIR 95%CI S omach cance Dea hs 31 58.1 0.53 a 0.36–0.75 24 18.7 1.28 0.82–1.90 Cases 60 80.5 0.75 0.57–0.95 23 26.1 0.88 0.56–1.32 All cance s Dea hs 1 020 1 116.7 0.91 0.86–0.97 514 354.8 1.45 1.33–1.57 Cases 3 088 2 747.6 1.12 1.08–1.16 976 881.4 1.11 1.04–1.17 a p<0.001. 920 I. VOHLONEN ET AL. The p esen s udy sugges s ha bioma ke sc eening by SPGI may educe mo ali y om gas ic cance by one- i h o e 15 yea s. Co espondingly, sc eening was es ima ed o educe he po en ial yea s o li e-los (PYLL) due o gas ic cance by almos 40%. These es ima es we e, howe e , incon- clusi e due o me hodological issues as desc ibed abo e. In ou s udy, wo cases o gas ic cance we e ound by sc eening in men wi h low SPGI in 1994-1996, and bo h pa ien s died om gas ic cance wi hin 5 yea s. The e o e, he long e m dec ease in mo ali y be ween 1994 and 2011 was likely due o ea men o p e-cance ous lesions obse ed in sc eening endoscopy o we e due o e adica ion o he on-going Helicobac e pylo i in ec ion [16–20]. I can be assumed ha he p ecance ous lesions in an a ophic s omach mucosa de elop o in asi e cance s du ing li e- ime in up o one hi d o cases i he lesions a e no p ope ly ea ed [20]. In ou s udy, he expec ed numbe o dea hs due o gas- ic cance was 58 and his numbe was based on a o ecas made on he basis o he Finnish Cance Regis y. In he sc eening o gas ic cance by a bioma ke , we ound 56 lesions and 2 cases o gas ic cance . In his iew, expec ed and obse ed cases o gas ic cance we e simila . This obse a ion indica es, ha a leas some 20 men would be a isk o gas ic cance among he 56 men who we e diagnosed wi h a p ecance ous lesion. Based on he obse ed gas ic cance dea hs, he c ude a ibu able p o- po ion in educ ion o dea hs due o gas ic cance was 59% and he e o e 16 dea hs due o gas ic cance we e p e- en ed i a endance was no selec i e. Because o selec ion, we co ec ed i in he analyzes and he co ec ed a ibu able p opo ion was 23% wi h he obse ed educ ion o 6 dea hs due o gas ic cance . We canno ega d his igu e as an unbiased one, because he unknown po en ial o esidual con ounding a e he me hodological co ec ions. Hence we do no know whe he he sc eening o gas ic cance was e icacious o no . Analyzes o he e ec s o SPGI sc eening on p ema u e gas ic cance mo ali y using PYLL ga e esul s consis en wi h hose ob ained by cance mo ali y. The co ec ed a ib- u able p opo ion o SPGI sc eening in educ ion o PYLL o gas ic cance was 39%. PYLL is dependen on he alue o li e expec ancy selec ed o he calcula ions. In he p esen in es iga ion, a li e expec ancy o 80 yea s was selec ed. Dea hs be o e his age a e conside ed p ema u e and mos dea hs due o gas ic cance in Finland occu be o e age o 80. The g ea e educ ion in co ec ed a ibu able p opo ion in educ ion o PYLL han o SMR o gas ic cance (39% s. 23%) may indica e ha some cance ous lesions among he a ending we e obse ed and ea ed in ea ly and cu able s ages. In spi e o he associa ion be ween SPGI sc eening and dec eased gas ic cance mo ali y and PYLL, he p esen esul s may s ill be biased al hough me hodological co ec- ions we e done o co ec o con ounding. The e o e, he p esen esul s do no jus i y de ini i e conclusions o imple- men a ion o gene al sc eening p og ams. In o ma ion on e icacy o a new and po en ial sc eening echnology o cance needs o be based on well-designed s udy p o ocols which esul in unbiased e idence. The e o e, con olled and andomized sc eening s udies a e an ul ima e p e equisi e. In he de eloped coun ies, incidence o gas ic cance is dec easing along wi h H. pylo i in ec ion and AG, bu in he de eloping coun ies hese a e s ill majo public heal h p oblems. Acknowledgmen s P o esso Ma i H€ a k€ onen and P o esso Pen i Sipponen a e sha eholde s o Biohi Oyj, a company which de elops and ma ke s labo a o y es s, including bioma ke es s o s omach diseases. Disclosu e s a emen The au ho s epo no con lic s o in e es . Re e ences [1] S o sk ubb T, A o P, Ronkainen J, e al. Se um bioma ke s p o ide an accu a e me hod o diagnosis o a ophic gas i is in a gen- e al popula ion: he Kalixanda s udy. Scand J Gas oen e ol. 2008;43:1448–1455. [2] Miki K, Mo i a M, Sasajima M, e al. Use ulness o gas ic cance sc eening using he se um pepsinogen es me hod. Am J Gas oen e ol. 2003;98:735–739. [3] Dinis-Ribei o M, Yamaki G, Miki K, e al. Me a-analysis on he al- idi y o pepsinogen es o gas ic ca cinoma, dysplasia o ch onic a ophic gas i is sc eening. J Med Sc een. 2004;11: 141–147. [4] Sacke DL, Holland WW. Con o e sy in he de ec ion o disease. Lance . 1975;2:357–359. [5] Teppo L, Pukkala E, Leh onen M. Da a quali y and quali y con ol o a popula ion-based cance egis y. Expe ience in Finland. Ac a Oncol. 1994;33:365–369. [6] Ah ens W, Pigeo I, edi o s. Handbook o epidemiology. Be lin: Sp inge ; 2005. [7] Cambell DT, S anley JC. Expe imen al and quasi-expe imen al designs o esea ch. Chicago (IL): Rand McNally; 1966. [8] Coch ane AL, Holland WW. Valida ion o sc eening p ocedu es. B Med Bull. 1971;27:3. [9] Holland WW, S ewa d S. Sc eening in heal h ca e. London: The Nu ield P o incial Hospi als T us ; 1990. [10] Engholm G, Fe lay J, Ch is ensen N, e al. Gas ic p ecance ous p ocess in a high isk popula ion: coho ollow-up. Cance Res. 1990;50:4737–4740. [11] Co ea P, Piazuelo MB. The gas ic p ecance ous cascade. J Dig Dis. 2012;13:2–9. [12] Haenszel W, Co ea P, Cuello C, e al. Gas ic cance in Colombia. II. Case-con ol epidemiologic s udy o p ecu so lesions. J Na l Cance Ins . 1976;57:1021–1026. [13] Sipponen P, Kekki M, Haapakoski J, e al. Gas ic cance isk in ch onic a ophic gas i is: s a is ical calcula ions o c oss-sec ional da a. In J Cance . 1985;35:173–177. [14] Va is K, Sipponen P, Laxen F, e al. The Helsinki Gas i is S udy G oup. Implica ions o se um pepsinogen I in ea ly endoscopic diagnosis o gas ic cance and dysplasia. Scand J Gas oen e ol. 2000;35:950–956. [15] Jang JS, Choi SR, Qu eshi W, e al. Long- e m ou comes o endo- scopic submucosal dissec ion in gas ic neoplas ic lesions a a single ins i u ion in Sou h Ko ea. Scand J Gas oen e ol. 2009;44:1315–1322. [16] De V ies AC, Kuipe s EJ. Epidemiology o p emalignan gas ic lesions: implica ions o he de elopmen o sc eening and su - eillance s a egies. Helicobac e . 2007;12:22–31. ACTA ONCOLOGICA 921 [17] C umley AB, Going JJ, McEwan K, e al. Endoscopic mucosal esec- ion o gas oesophageal cance in a U.K. popula ion. Long- e m ollow-up o a consecu i e se ies. Su g Endosc. 2011;25:543–548. [18] Ichinose M, Yahagi N, Oka M, e al. Sc eening o gas ic cance in Japan. In: Wu GY, Aziz K, edi o s. Cance sc eening. A p ac ical guide o physicians. Po owa (NJ): Humana P ess; 2001. p. 87–102. [19] Benichou J, Pal a M. Ra es, isks, measu es o associa ion and impac . In: Ah ens W, Pigeo I, edi o s. Handbook o epidemi- ology. Be lin: Sp inge ; 2005. p. 8–146. [20] Wilson JMG, Junge G. P inciples and p ac ice o sc eening o disease. Public Heal h Pape Numbe 34. Gene a: WHO; 1968. 922 I. VOHLONEN ET AL.