Association of symptoms and interval breast cancers in the mammography-screening programme: population-based matched cohort study

ARTICLE Epidemiology Associa ion o symp oms and in e al b eas cance s in he mammog aphy-sc eening p og amme: popula ion-based ma ched coho s udy Deepend a Singh 1,2 , Joonas Mie inen 1 , S ephen Du y 3 , Nea Malila 1,2 , Janne Pi käniemi 1 and Ah i An ila 1 BACKGROUND: We assessed he associa ion be ween symp oms epo ed a b eas cance sc eening isi s and in e al cance s (ICs) in a p ospec i e manne . METHODS: This popula ion-based ma ched coho s udy uses da a o he Finnish Na ional B eas Cance Sc eening P og amme ha in i es women aged 50–69 yea s old du ing 1992–2012. Subjec s who a ended sc eening wi h symp oms we e ma ched wi h asymp oma ic e e ence coho s based on age a sc eening isi , yea o in i a ion, numbe o in i ed isi s and municipali y o in i a ion. The p ima y ou come was ICs. RESULTS: Women wi h a lump had a h ee old (haza d a io 3.7, 95% confidence in e al (CI) 3.0–4.6) isk o ICs and a highe isk (haza d a io 1.7, 95% CI 1.4 o 2.0) a he subsequen isi compa ed wi h hose wi hou a lump. The a al in e al cance isk inc eased by 0.39 pe 1000 sc eens wi h a lump. The cumula i e incidences o in e al cance inc eased wi hin a mon h o a mammog aphy-nega i e isi wi h a lump and a e abou 6 mon hs o he isi wi h e ac ion o nipple discha ge. CONCLUSION: Women wi h b eas symp oms ha e a clea ly inc eased isk o in e al b eas cance a e he sc eening isi . Ou findings indica e he need o di e en sc eening s a egies in symp oma ic women. B i ish Jou nal o Cance (2018) 119:1428–1435; h ps://doi.o g/10.1038/s41416-018-0308-2 BACKGROUND Beyond he andomised ial en i onmen , he e is some unce ain y as o he unde lying incidence o b eas cance and he a e o o e diagnosis in he sc eening popula ion. 1,2 Along wi h sc eening pe o mance, 3–5 a es and p opo ions o in e al cance s (ICs) a e impo an indica o s o assessing he e ec i e- ness and quali y o sc eening. 2,6–8 The subs an ial p opo ion (abou a hi d) o inciden b eas cance diagnosed ou side he mammog aphy-sc eening p og amme 9,10 likely indica es ha he e is oom o imp o emen in he de ec ion capabili y o he mammog aphy-sc eening p og amme. Ea lie obse a ional s u- dies 7,11–15 ha e highligh ed se e al easons o he inc eased p opo ion o ICs in he sc eening p og amme. Howe e , in e ms o equi y wi hin he sc eening popula ion, i is easonable o aim o simila in e al cance a es o a leas simila p opo ions o cance s a ising as in e al cases o he a ious he e ogeneous g oups pa icipa ing in he sc eening. Fu he mo e, op ions o modi y sc eening policies should be conside ed o high- isk g oups. A sho e sc eening in e al may be jus ified, o ins ance, i he in e al cance a e is significan ly high. Based on he Eu opean Union (EU) guidelines, sc eening is mean o unselec ed a ge popula ion. 7 Ea lie s udies om Finland ha e indica ed ha a no iceable p opo ion (~ 2–3%) o women ha e clinically significan symp oms when hey pa icipa e in b eas cance sc eening. 16,17 Mos bu no all symp oma ic women will ha e u he assessmen s wi h ul asound, addi ional mammog ams o o he me hods; i hese e u n nega i e esul s, he women e u n o he no mal, biennial sc eening in e al. S udies on b eas symp oms (such as a lump, e ac ion o nipple discha ge) indica e an inc eased isk o b eas cance 16–19 a he cos o ise in alse-posi i e findings. The ela ions o symp oms wi h in e al b eas cance s and sc een-de ec ed cance s (SDCs) a he subsequen isi ha e ne e been s udied. We in es iga ed whe he women epo ing b eas symp oms a sc een a e a a highe isk o de eloping subsequen b eas cance s (ICs and cance s diagnosed a he nex sc een) han hose wi hou symp oms. To c ea e ounda ions o modi y he sc eening policies in high- isk g oups, we es ima ed he cumula i e incidence o ICs and a al in e al b eas cance s, and compa ed he espec i e incidences in women wi h and wi hou symp oms. The quali y measu es o sc eening mammog aphy we e compa ed be ween isi s among subjec s wi h and wi hou symp oms o ga he e idence o imp o ing p og amme pe o mance. METHODS S udy design, da a sou ce and s udy popula ion Ou ma ched coho s udy design was based on he ollow-up o he ongoing Finnish Na ional B eas Cance Sc eening P og amme ha began in 1987. Biennial sc eening isi s made by women www.na u e.com/bjc Recei ed: 10 July 2018 Re ised: 25 Sep embe 2018 Accep ed: 26 Sep embe 2018 Published online: 7 No embe 2018 1 Mass Sc eening Regis y, Finnish Cance Regis y, FI-00130 Helsinki, Finland; 2 Epidemiology g oup, Depa men o Heal h Sciences, Uni e si y o Tampe e, FI-33520 Tampe e, Finland and 3 Wol son Ins i u e o P e en i e Medicine, Queen Ma y Uni e si y o London, Cha e house Squa e, London, England Co espondence: Deepend a Singh (deepend a.singh@cance .fi) ©The Au ho (s) 2018 Published by Sp inge Na u e on behal o Cance Resea ch UK aged 50–69 yea s be ween 1991 and 2012 we e selec ed. Th ee egis ies, he Finnish Cance Regis y (1953–2014), he Mass Sc eening Regis y (1992–2012) and he Cen al Popula ion Regis y (1992–2014) we e used o ex ac in o ma ion on he s udy pa icipan s a he indi idual le el. The Mass Sc eening Regis y was used o ex ac in o ma ion on demog aphic, symp oma ic and sc eening p ocedu e ac o s, including ecalls and e e al da a ha ha e been shown o be alid and o high quali y. All indi idual isi s we e linked o he Finnish Cance Regis y da abase o e ie e in o ma ion on b eas cance s (sc een-de ec ed and ICs). This included his ological findings and po en ial dea h om inciden b eas cance . The Popula ion Regis y was used o iden i y possible da es o dea h o emig a ion, and whe e applicable, he cause o dea h was e ie ed om S a is ics Finland. Fig. 1shows he flow diag am o he s udy design. Exposu e g oup (sc eening isi s wi h symp oms) The exposed g oup is defined as isi s by women wi h b eas symp oms (lump, e ac ion and nipple discha ge) epo ed a a gi en sc eening ound. This g oup con ained all isi s wi h a leas one symp om epo ed. Di e en symp oms (lump, e ac ion, nipple discha ge) we e analysed sepa a ely, which in echnical e ms assumes ha he fi s occu ence o any symp om was independen gi en he co a ia es. Fo example, i mo e han one symp om was epo ed a a single isi hen each symp om was analysed sepa a ely. He e, he index isi mean any sc eening isi wi h any gi en symp om. Re e ence g oup (sc eening isi s wi hou symp oms) The e e ence g oup is defined as isi s by women wi h no epo ed b eas symp oms in he sc eening his o y be o e he index isi . The women om he e e ence g oup can la e be he pa o he exposed g oup i symp oms a e epo ed a u u e sc eening isi s. Thus, symp oms a e ime-dependen co a ia es. Indi idual se s o e e ence isi s o each symp om—al oge he h ee se s—we e o med by ma ching. Ma ching The h ee exposed g oups we e equency ma ched o he e e ence g oups by age a he sc eening isi (wi hin 2 yea s), yea o in i a ion (2-yea band), numbe o isi s in he pas and municipali y o in i a ion. Visi s wi h symp oms we e hen agg ega ed based on ma ching a iables (and o he co a ia es). Each symp om s a um was ma ched o he iable con ols ( e e ence isi s) by andom sampling. Random con ols we e selec ed, based on he ma ching a iables abo e, as many imes as he numbe o isi s in each s a um o symp oms by he eplacemen sampling me hod. Hence, a single con ol had he possibili y o be andomly selec ed mo e han once o he same s a um. Based on ou assump ion o an e ec size and equi ed powe o 0.80, he exposed- o- e e ence- isi s a io was 1:1 o lump and e ac ion, and 1:2 o nipple discha ge. Ou come assessmen ICs we e defined as b eas cance s diagnosed in sc eened women be o e he nex sc eening isi o wi hin a pe iod equal o a sc eening in e al wi h (i) nega i e mammog aphy a he index isi (i.e., es nega i e); (ii) posi i e mammog aphy a he index isi , bu nega i e u he assessmen (i.e., episode nega i e); and (iii) posi i e u he assessmen bu a da e o diagnosis > 6 mon hs a e mammog aphy. 2 SDCs we e defined as p ima y b eas cance diagnosed among he sc eening a endees wi hin 6 mon hs ollowing an abno mal mammog am ( es posi i e). The subse- quen ound SDCs we e analysed ollowing he index isi s wi h o wi hou symp oms i he women a ended he subsequen ound. Finnish na ional b eas cance sc eening p og am (FNBCSP) Mass sc eening egis y (1992-2012) In o ma ion on he mammog aphy sc eening p ocess, includes symp oms in o ma ion In o ma ion on cance s, sc een- de ec ed and in e al cance s, and a al b eas cance s 222848 isi s excluded because o missing in o ma ion on symp oms s a us Visi s wi h symp oms (exposu e g oup) Lump = 51333 Re ac ion = 40917 Nipple discha ge = 9083 Visi s wi hou symp oms ( e e ence g oup) No lump = 51332 No e ac ion = 40917 No nipple discha ge = 18166 Ma ching* Exposu e g oup we e equency ma ched o e e ence g oup ( isi s wi hou symp oms) by co a ia es, selec ed by andom sampling using eplacemen me hod Popula ion egis y (1992-2014) In i ed o sc eening (4,594,335) A ended (3,958,312) *1:1 ma ching a io o lump and e ac ion, 1:2 a io o nipple discha ge S udy popula ion (3,735,464) Cance egis y (1992-2014) Fig. 1 Flow diag am o s udy se ings Associa ion o symp oms and in e al b eas cance s in he. . . D Singh e al. 1429 1234567890();,: In addi ion, cance s we e sub-g ouped in o in si u ca cinomas and non-localised b eas cance s. Fa al cance s we e defined as hose b eas cance s ha esul ed in dea h du ing ollow-up. Follow-up The ollow-up ime s a ed om he index isi in 1 Janua y 1992 o 31 Decembe 2012 and ended a he da e o emig a ion o dea h, upon diagnosis o in e al cance o a he end o he ollow-up—i.e., 31 Decembe 2014—whiche e occu ed fi s . Cance cases diagnosed among hose sc eened up o 31 Decembe 2012 and ollowed up o 31 Decembe 2014 ( o hose sc eened in 2011 and 2012) we e di ided in o ICs and subsequen SDCs using Finnish Cance Regis y da a. Conside ing possible delays in he diagnosis da e a e posi i e mammog aphy findings, a sc eening episode o 6-mon h in e als was used in he defini ion o de ec ion mode. Thus, he ollow-up ime o ICs s a ed a 7 mon hs o episode nega i e isi s and a 1 mon h o es nega i e isi s and ended a he da e o he subsequen sc eening isi a 23 mon hs. S a is ical analysis We compa ed b eas cance isk and he isk o b eas cance dea h using Cox p opo ional haza d eg ession among women wi h and wi hou epo ed symp oms a he index sc een. Confidence in e als we e compu ed exac ly om pa ame e likelihoods. The analyses we e adjus ed o age a he sc eening isi . We calcula ed he incidence a e o in e al cance be ween he sc eens ( om he index sc een and he 6-mon h episode o he subsequen sc eening isi ) sepa a ely o es nega i es and episode nega i es. We also e alua ed he p og amme cha ac e is ics using basic s a is ics. A es sensi i i y was es ima ed as he numbe o isi s wi h a posi i e mammog aphy es and diagnosis o cance a sc een di ided by he sum o SDCs plus ICs diagnosed a e nega i e es esul s. Episode sensi i i y was calcula ed as he numbe o isi s wi h a diagnosis o cance in a ull diagnos ic p ocess in he sc eened popula ion di ided by all cance s de ec ed in a sc eening ound among a ende s. Simila ly, he posi i e p edic i e alue (PPV) was assessed as he numbe o isi s wi h a posi i e mammog aphy es and diagnosis o cance di ided by he numbe o es posi i es. The nega i e p edic i e alue (NPV) was es ima ed as he numbe o isi s wi h a nega i e es esul and no cance diagnosed di ided by he numbe o es nega i es. All s a is ical analyses we e pe o med using R-3.4.0. RESULTS O e he s udy pe iod o 21 yea s, a lump was epo ed a 51,333 isi s and e ac ion a 40,917 isi s. These isi s we e ma ched o an equal numbe o asymp oma ic isi s. The e we e 9083 isi s wi h nipple discha ge, and hey we e ma ched wi h double he numbe o e e ence isi s (i.e., 18,166 isi s) wi hou nipple discha ge. De ailed numbe s o he po en ially eligible and he confi med eligible popula ion included in he s udy a e shown in Fig. 1. The mean age a a sc eening isi did no di e be ween isi s wi h and wi hou symp oms (mean age 55.7 s 56.1 yea s o isi s wi h and wi hou a lump, espec i ely). Table 1shows he cha ac e is ics o he final s udy coho . Abou one in h ee women who epo ed a lump o nipple discha ge and one in eigh women who epo ed e ac ion we e fi s - ime a endees. The fi s and subsequen a endee’s p opo ions we e simila be ween isi s wi h and wi hou symp oms. Mo e han 80% o ICs and he subsequen ound’s SDCs, i espec i e o epo ed symp oms s a us, we e no ecalled o u he assessmen a he index isi . Bo h he es and episode sensi i i y o he mammog aphy was highe o isi s wi h a lump o e ac ion compa ed wi h hose wi hou such symp oms (82 and 75% s 64 and 63% o a lump s no lump; 77 and 76% s 67 and 66% o e ac ion s no e ac ion; Table 2). Likewise, he PPV o mammog aphy was highe o e ac ion and a lump, compa ed wi h hose wi hou hese symp oms. Howe e , he specifici y o mammog aphy was clea ly lowe o isi s wi h a lump (88%) han hose wi hou (98%). Some dec ease in specifici y was also seen o e ac ion and nipple discha ge. Incidence o sc een-de ec ed and ICs, subsequen SDCs and a al ICs In o al, 1440 (2.8%) SDCs and 387 (0.7%) ICs (ICs) we e diagnosed in hose who epo ed a lump compa ed wi h 174 (0.3%) SDCs and 103 (0.2%) ICs in hose wi hou a lump, espec i ely (Table 3). The p opo ions o SDCs and ICs we e highe also o e ac ion and nipple discha ge compa ed wi h hose wi hou hese symp oms. The age-adjus ed isk o SDCs was significan ly highe in hose who epo ed a lump (adjus ed haza d a io 8.2, 95% CI 7.0–9.7), e ac ion (adjus ed haza d a io 2.3, 95% CI 2.0–2.8) o nipple discha ge (adjus ed haza d a io 1.5, 95% CI 1.1–2.3) compa ed wi h hose wi hou symp oms. In addi ion, he age-adjus ed isk o ICs was significan ly highe o a lump (adjus ed haza d a io 3.7, 95% CI 3.0–4.6), e ac ion (adjus ed haza d a io 1.5, 95% CI 1.1–1.9), and nipple discha ge (adjus ed haza d a io 2.4, 95% CI 1.6–3.7) compa ed wi h hose wi hou hese symp oms. The isk o SDCs in he subsequen ound was significan ly highe only a e isi s wi h a lump compa ed wi h hose wi hou a lump (adjus ed haza d a io 1.6, 95% CI 1.3–2.0). The isk o in si u in e al ca cinomas o subsequen sc een-de ec ed ca cinomas was also highe in isi s wi h a lump and nipple discha ge compa ed wi h a isi wi h no symp oms. The isk o non-localised in e al b eas cance as well as SDCs in he subsequen ound we e also g ea e o all h ee symp oms in compa ison wi h isi s wi hou symp oms. The age-adjus ed isk o dying om b eas cance was significan ly highe in hose who epo ed a lump and we e diagnosed wi h in asi e cance s (SDCs =adjus ed haza d a io 19, 95% CI 11–38; ICs =adjus ed haza d a io 2.0, 95% CI 1.1–3.4; subsequen ound cance s =adjus ed haza d a io 2.7, 95% CI 1.6–4.1) compa ed wi h hose wi hou a lump (Table 3). In addi ion, he isk o dying was highe in hose who epo ed e ac ion and we e diagnosed wi h SDCs (adjus ed haza d a io 6.3, 95% CI 2.8–16) compa ed wi h hose wi hou e ac ion. Only a ew dea hs occu ed du ing he ollow-up in hose who epo ed nipple discha ge. Cumula i e incidence o b eas cance s du ing he sc eening in e al The incidence o ICs a e a isi wi h a lump o nipple discha ge inc eased a he apidly a e he isi in all he s udied p og ession ypes (Fig. 2a–d). When using he cumula i e incidence o he asymp oma ic o e he whole in e al as a e e ence, he same le el o ICs was eached in only abou six mon hs a e isi s wi h a lump and 12 mon hs a e isi s wi h nipple discha ge. Rema kably, in mammog aphy-nega i e isi s wi h a lump, he cumula i e incidence cu e de ached om he no lump cu e immedia ely a e he fi s mon h o isi , whe eas such a di e ence was no obse ed o he o he symp om ypes. DISCUSSION In his popula ion-based s udy wi h 21 yea s o sc eening ( he ollow-up is es ic ed o 23 mon hs, al hough he s udy pe iod was 21 yea s), we obse ed s ong associa ions be ween symp oms and b eas cance isks. Women epo ing a lump a a sc eening isi had a h ee old isk o ICs compa ed wi h hose wi h no symp oms, also including subsequen SDCs. The (cumula i e) incidence o in e al cance was highe in hose who epo ed a lump i espec i e o he mammog aphy findings Associa ion o symp oms and in e al b eas cance s in he. . . D Singh e al. 1430 ( es nega i es o episode nega i es) compa ed wi h hose wi hou symp oms a he index isi . Likewise, e ac ion and nipple discha ge we e significan ly associa ed wi h inc eased isks o in e al b eas cance s. In absolu e e ms, pe 1000 women who a ended and epo ed a lump, se en women we e diagnosed wi h in asi e in e al b eas cance s, i.e., wi hin 24 mon hs compa ed wi h abou wo cance s diagnosed wi hou symp oms. The a al in e al cance isk inc eased by 0.39 pe 1000 sc eens wi h a lump. Women wi h symp oms had a clea ly inc eased ‘backg ound’ isk o b eas cance ; he con en ional sc eening pe o mance measu es—such as sensi i i y, PPV and specifici y—did no ully assess his aspec . The cumula i e incidence pa e ns as well as he de ec ion o cance s du ing he subsequen ound p o ided di ec e idence o he need o isk-adjus ed sc eening and a be e diagnos ic wo k-up in he symp oma ic women. The diagnos ic wo k-up could include de eloping be e eading and ecall c i e ia in sc eening mammog ams, a mo e de ailed Table 2 Pe o mance quali y measu es o sc eening mammog aphy in ela ion o symp oms s a us Symp oms Sc eening episode Mammog aphy es Sensi i i y % (95% CI) Sensi i i y % (95% CI) PPV % (95% CI) NPV % (95% CI) Specifici y % (95%CI) Lump Yes 74 (71–78) 81 (80–83) 19 (19–20) 99 (99–99) 88 (88–88) No 62 (56–68) 64 (58–70) 12 (11–13) 99 (99–99) 97 (97–97) Re ac ion Yes 75 (72–79) 76 (73–80) 28 (26–29) 99 (99–99) 97 (96–97) No 65 (59–70) 66 (60–72) 17 (16–19) 99 (99–99) 97 (97–98) Nipple discha ge Yes 55 (46–65) 62 (52–71) 8.1 (6.9–9.4) 99 (99–99) 91 (91–92) No 66 (57–75) 70 (61–78) 15 (13–17) 99 (99–99) 97 (97–97) PPV posi i e p edic i e alue, NPV nega i e p edic i e alue, CI confidence in e al Table 1 Coho cha ac e is ics Cha ac e is ics Lump (n=51,333) Re ac ion (n=40,917) Nipple discha ge* Yes (%) No (%) Yes (%) No (%) Yes (%) (n=9083) No (%) (n=18,166) Age a index isi (mean, SD) 55.7 (4.6) 56.1 (4.5) 57.7 (4.9) 56.6 (4.8) 54.9 (4.6) 56.1 (4.7) A endance Fi s a endance 18,305 (35) 17,954 (35) 5012 (12) 4959 (12) 2904 (32) 5643 (31) Subsequen a endance 33,028 (64) 33,379 (65) 35,905 (87) 35,958 (87) 6179 (68) 12,523 (68) P e ious ound a endance Yes 31,571 (61) 32,228 (62) 34,168 (83) 34,263 (83) 5842 (64) 12,069 (66) No 19,762 (38) 19,105 (37) 6749 (16) 6654 (16) 3241 (35) 6097 (33) Pe iod o isi 1992–1997 15,478 (30) 15,520 (30) 0 0 1134 (12) 2298 (12) 1998–2002 7535 (14) 7493 (14) 3552 (8.7) 3552 (8.7) 1673 (18) 3316 (18) 2003–2007 11,791 (23) 11,804 (23) 12,996 (31) 13,190 (32) 2872 (31) 5671 (31) 2008–2012 16529 (32) 16,516 (32) 24,369 (59) 24,175 (59) 3404 (37) 6881 (37) Recall ( es posi i es) In asi e cance s Sc een-de ec ed cance s 1440 (2.8) 174 (0.34) 461 (1.1) 192 (0.47) 66 (0.73) 83 (0.46) In e al cance s 70 (0.14) 7 (0.01) 16 (0.04) 5 (0.01) 12 (0.13) 6 (0.03) Subsequen sc een a nex ound 34 (0.07) 6 (0.01) 11 (0.03) 7 (0.02) 4 (0.04) 3 (0.01) Fa al b eas cance s Sc een-de ec ed cance s 215 (0.41) 11 (0.02) 38 (0.09) 6 (0.01) 4 (0.04) 2 (0.01) In e al cance s 8 (0.01) 2 (0.01) 2 2 0 0 Subsequen sc een a nex ound 1 (0.01) 0 1 (0.01) 0 0 0 No ecall ( es nega i es) In asi e cance s Sc een-de ec ed cance s 0 0 0 0 0 0 In e al cance s 317 (0.62) 96 (0.19) 138 (0.34) 96 (0.23) 40 (0.44) 35 (0.19) Subsequen sc een a nex ound 230 (0.45) 149 (0.29) 144 (0.35) 135 (0.33) 28 (0.31) 57 (0.31) Fa al b eas cance s Sc een-de ec ed cance s 0 0 0 0 0 0 In e al cance s 32 (0.06) 18 (0.03) 13 (0.03) 5 (0.01) 5 (0.05) 2 (0.01) Subsequen sc een a nex ound 18 (0.03) 7 (0.01) 4 (0.01) 4 (0.01) 0 1 (0.01) Values a e numbe s (pe cen age) unless s a ed o he wise *Fo each isi s wi h nipple discha ge we e ma ched wi h wo isi s wi hou nipple discha ge Associa ion o symp oms and in e al b eas cance s in he. . . D Singh e al. 1431 Table 3 F equency (pe 1000 sc eening isi s) and age-adjus ed isk o cance ou comes in hose who epo ed symp oms compa ed wi h hose wi hou symp oms a sc een Ou comes Lump (n=51,333) Re ac ion (n= 40,917) Nipple discha ge* Yes (pe 1000) No (pe 1000) Age-adjus ed haza d a io (95% CI) wi h e e ence o no lump Yes (pe 1000) No (pe 1000) Age-adjus ed haza d a io (95% CI) wi h e e ence o no e ac ion Yes (pe 1000) (n= 9083) No (pe 1000) (n=18,166) Age-adjus ed haza d a io (95% CI) wi h e e ence o no nipple discha ge In asi e cance s Sc een-de ec ed cance s 1440 (28) 174 (3.4) 8.3 (7.1–9.7) 461 (11) 192 (4.7) 2.4 (2–2.8) 66 (7.3) 83 (4.6) 1.7 (1.2–2.3) In e al cance s 387 (7.5) 103 (2.0) 3.8 (3–4.7) 154 (3.8) 101 (2.5) 1.5 (1.2–1.9) 52 (5.7) 41 (2.3) 2.5 (1.7–3.8) Subsequen sc een a nex ound 264 (5.1) 157 (3.1) 1.7 (1.4–2.1) 156 (3.8) 158 (3.9) 1.1 (0.87–1.4) 32 (3.5) 72 (3.9) 1.0 (0.64–1.5) In si u ca cinomas Sc een-de ec ed cance s 61 (1.2) 38 (0.74) 1.6 (1.1–2.4) 35 (0.86) 31 (0.76) 1.1 (0.70–1.8) 15 (1.7) 16 (0.88) 1.9 (0.92–3.8) In e al cance s 35 (0.68) 9 (0.18) 3.9 (1.9–8.6) 7 (0.17) 6 (0.15) 1.2 (0.39–3.6) 9 (0.99) 4 (0.22) 4.0 (1.3–14) Subsequen sc een a nex ound 159 (3.1) 102 (1.9) 1.5 (1.2–1.9) 76 (1.9) 66 (1.6) 1.2 (0.83–1.6) 32 (3.5) 32 (1.8) 2.0 (1.2–3.3) Non-localised cance s Sc een-de ec ed cance s 693 (13) 57 (1.1) 12 (9.2–16.0) 230 (5.6) 61 (1.5) 3.5 (2.6–4.7) 25 (2.8) 27 (1.5) 1.9 (1.1–3.2) In e al cance s 178 (3.5) 57 (1.1) 3.4 (2.5–4.6) 79 (1.9) 48 (1.2) 1.7 (1.2–2.4) 20 (2.2) 18 (0.99) 2.2 (1.2–4.2) Subsequen sc een a nex ound 649 (12) 438 (8.5) 1.5 (1.3–1.7) 244 (5.9) 182 (4.5) 1.4 (1.1–1.7) 110 (12) 181 (9.9) 1.2 (0.96–1.6) Fa al cance s Sc een-de ec ed cance s 215 (4.2) 11 (0.21) 19 (11–38) 38 (0.93) 6 (0.15) 6.3 (2.9–16) 4 (0.44) 2 (0.11) 4.0 (0.78–28) In e al cance s 40 (0.78) 20 (0.39) 2 (1.2–3.5) 15 (0.36) 7 (0.17) 2.1 (0.90–5.6) 5 (0.55) 2 (0.11) 5.0 (1.1–34) Subsequen sc een a nex ound 19 (0.37) 7 (0.14) 2.7 (1.7–4.1) 5 (0.12) 4 (0.07) 1.3 (0.45–2.7) 0 1 (0.05) NA *Fo each isi s wi h nipple discha ge we e ma ched wi h wo isi s wi hou nipple discha ge; CI confidence in e al, NA no a ailable Associa ion o symp oms and in e al b eas cance s in he. . . D Singh e al. 1432 indica ion o biopsies in he u he assessmen , and ea ly ecall o hose wi h high- isk symp oms o whom he sc eening es o u he assessmen p o es nega i e. S eng hs and limi a ions o he s udy To ou knowledge, his is he fi s popula ion-based s udy o analyse he associa ion be ween symp oms and in e al and a al b eas cance s. Ou s udy has se e al s eng hs. Fi s , we included all sc eening isi s wi h symp oms om he s a o he mammog aphy-sc eening p og amme (o e 21 yea s) and com- pa ed hem wi h isi s wi hou symp oms. Since his was a popula ion-based se ice-sc eening p og amme, he selec ion bias was minimal. Second, we could use alida ed p ospec i ely collec ed egis e da a wi h no ecall bias. 20 Valida ion o he cance diagnosis (sc een-de ec ed and ICs) and dea h om cance using na ional and co e ing da a sou ces (Finnish Cance Regis y and S a is ics Finland) and he use o unique pe sonal iden ifie o indi idual-le el linkage elimina es he possibili y o selec i e misclassifica ion. Thi d, we ma ched he symp oma ic isi s o asymp oma ic isi s by possible con ounding baseline cha ac e - is ics o minimise bias in he isk es ima es. As a esul , we ound no significan di e ence in he backg ound a iables be ween isi s wi h and wi hou symp oms. Finally, he p og amme p ocess and ou come indica o defini ions used in his s udy adhe e o hose defined by WHO-IARC and men ioned in EU guidelines on b eas cance sc eening and diagnosis, 7,21 including he p o ision o ele an in o ma ion desc ibing he pe o mance and also ailu es du ing he a ious s eps o he sc eening p ocess. Ou s udy also has po en ial limi a ions. The symp om in o ma- ion was based on he women’s epo ing in he pas 2 o 6 mon hs and a check by a adiog aphe a he isi . The adiog aphe ’s inspec ion is likely o be less comp ehensi e han a ull clinical examina ion. The collec ion o symp om in o ma ion is mainly done in o de o suppo he in e p e a ion o he mammog ams. Howe e , in almos e e y case—i no all— he adiog aphe o nu se examines he b eas o confi m he p esence o symp oms (mainly a lump and e ac ion) be o e he mammog aphy is pe o med. Thus, he e a e easons o conside he collec ed symp om in o ma ion o be alid, albei no pe ec . A second po en ial limi a ion is ha ou es ima es could ha e been con ounded, because symp oma ic women a e mo e likely o a end han asymp oma ic women. Howe e , because o he high a endance a e in he Finnish mammog aphy-sc eening p og amme (84% among he in i ed, he highes among any 2.0 1.5 1.0 Cumula i e p obabili y (pe 10,000) Cumula i e p obabili y (pe 10,000) Cumula i e p obabili y (pe 10,000)Cumula i e p obabili y (pe 10,000) Cumula i e p obabili y (pe 10,000) Cumula i e p obabili y (pe 10,000) Cumula i e p obabili y (pe 10,000) Cumula i e p obabili y (pe 10,000) Cumula i e p obabili y (pe 10,000)Cumula i e p obabili y (pe 10,000)Cumula i e p obabili y (pe 10,000) Cumula i e p obabili y (pe 10,000) 0.5 0.0 2.0 1.5 1.0 0.5 0.0 2.0 1.5 1.0 0.5 0.0 8 6 4 2 0 8 6 4 2 0 8 6 4 2 0 4 3 2 1 0 4 3 2 1 0 4 3 2 1 0 1.5 1.0 0.5 0.0 1.5 1.0 0.5 0.0 1.5 1.0 0.5 0.0 05 Follow up (mon hs) Follow up (mon hs) Follow up (mon hs) Follow up (mon hs) Follow up (mon hs) Follow up (mon hs) Follow up (mon hs) Follow up (mon hs) Follow up (mon hs) Follow up (mon hs) Follow up (mon hs) Follow up (mon hs) 10 15 20 0 510 15 20 0 510 15 20 0 510 15 20 0 5 10 15 20 0 5 10 15 20 0 5 10 15 20 0 5 10 15 20 0510 15 20 0510 15 20 0510 15 20 0510 15 20 abc d Fig. 2 a–dCumula i e incidence o in asi e (pe 1000): a ecalled ICs; bno ecalled ICs; cnon-localised ICs; d a al in e al cance s. No e: he confidence in e als lines o cumula i e incidence a e indica ed by ligh do ed lines in symp oma ic and asymp oma ic g oups Associa ion o symp oms and in e al b eas cance s in he. . . D Singh e al. 1433 exis ing mammog aphy-sc eening p og amme), and as 97% o a endees a e asymp oma ic, he a endance bias caused by symp oms is likely o be small. Thi d, we did no use he backg ound incidence o b eas cance o es ima e he in e al cance a e in he absence o sc eening, bu ins ead used a de ec ion me hod 22 ha akes in o accoun cance s om he sc eening p og amme; hus, es ima es a e sensi i e o o e - diagnosis and lead ime bias. Gi en he high co e age and a endance a e in he Finnish sc eening p og amme, i was no possible o find a compa able non-sc eened g oup o es ima e he backg ound incidence o b eas cance . In addi ion, he e was no possibili y o es ima e backg ound incidence o b eas cance in women wi h symp oms. We did no find any di e ence in he PPV o mammog aphy, wi h and wi hou symp oms, be ween he fi s and subsequen sc eening ounds (no shown in esul s); hus, he lead ime bias because o p e alen sc eens is negligible. Fu he mo e, we used in asi e b eas cance s—and also ad anced and a al b eas cance s, which would be less a ec ed by o e diagnosis— o es ima e he incidence a es and haza d a ios. Only 5% o all ca cinomas in hose wi h symp oms we e in si u ca cinomas. P e ious s udies on he associa ion be ween symp oms and he isk o b eas cance a e limi ed because o ac o s such as s udy design, size, ollow-up ime and assessmen o he possible ou come measu es o b eas cance . None heless, a ew s udies ha e ound ha he p esence o a palpable lump is associa ed wi h a highe isk o SDCs. 18,19 We a e no awa e o any s udies ha ha e assessed he ela ionship o symp oms wi h o he ou come measu es. Clinical and public heal h implica ions Bo h he sc eening es and episode sensi i i ies ended o be highe in symp oma ic women compa ed wi h he asymp oma ic. Co espondingly, he sc eening specifici y was lowe in he symp oma ic women. O no e, pa icula ly in women wi h a lump, we e he (episode) sensi i i y losses in he u he assessmen du ing he index isi s. Mo eo e , highe in e al cance incidence wi hin six mon hs a e a nega i e mammog aphy wi h a lump is a clea conce n o he p og amme. This indica es ha u he assessmen is needed mo e equen ly (albei wi h a po en ial loss o specifici y), and he e needs o be highly s ingen diagnos ic e idence o a decision no o ca y ou a ull u he assessmen including a co e biopsy in hese cases. As mos isi s (~ 97%) we e asymp oma ic, he numbe o addi ional se ices would be a he small as well as he imp o emen o p og amme’s o e all pe o mance and ou come. Bu s ill, high-quali y and clinically app op ia e se ices a e impo an o women ha ing symp oms a a sc eening isi . One op ion is o ecall all women wi h symp oms e en i he mammog aphy esul is nega i e, as p ac ised in No way (< 0.3% o all hose sc eened we e ecalled wi h symp oms). 23 Doing his in Finnish p og amme would significan ly lowe he PPV o ecall, as 2.5% o all sc eens had symp oms and only ~ 1 ou o 10 symp o- ma ic isi s ha e been ecalled. 17 Taking in o accoun he high incidence o ad anced and a al in e al b eas cance s in symp oma ic women, i is likely ha p o ec ion by biennial sc eening isi would clea ly no be su ficien e en a e po en ial imp o emen s in u he assess- men s. We a e no awa e o ecommenda ions o su eillance o ollow-up o symp oma ic women in he p og amme. Hence, we ecommend a sho e sc eening in e al o he symp oma ic g oup so ha he cumula i e incidences o in e al and a al in e al cance would possibly become mo e equi able. Fo wo o he s udied symp oms (lump and nipple discha ge), he in e al cance incidence inc eased so apidly ha he fi s ollow-up isi could ake place e y sho ly a e he index isi . Finally, aking in o accoun he p obabili y o a al sc een-de ec ed b eas cance is highe in women wi h a symp om al eady a he index isi , women need o be be e in o med abou symp oms and made awa e ha i a symp om occu s, i is no a good idea o wai un il he nex in i a ion o he p og amme. Guidelines o u he assessmen in pa ien s p esen ing wi h symp oms be o e ha ing a scheduled in i a ion acco ding o he p og amme—as de eloped by Na ional Heal h Se ice in he UK 24 —could be use ul also in o he se ings. CONCLUDING REMARKS Women wi h b eas symp oms a isi s wi hin he popula ion- based b eas cance sc eening p og amme ha e a clea ly inc eased isk o b eas cance . The cumula i e incidence o in asi e, ad anced and a al b eas cance s, as well as he de ec ion o hem a he nex sc een, p o ide di ec e idence o he need o isk-adjus ed sc eening in symp oma ic women— e.g., ailo ing he managemen p ocedu es a index isi s and sho ening he sc eening in e als o hese women. This s udy p o ides clea e idence o upda e and suppo he EU guidelines 7 ecommenda ion ha ensu es su ficien a en ion being paid o symp oma ic de ails p o ided by women. Ou findings he e o e ha e impo an implica ions o sc eening-aged women, adiolo- gis s, nu ses and mammog aphy-sc eening p og ammes o e all. ACKNOWLEDGEMENTS This wo k was suppo ed by he Cance Founda ion G an o Finland. The unde had no ole in he design, he conduc o he s udy and finalisa ion o he manusc ip . AUTHOR CONTRIBUTIONS DS and AA designed he s udy. JM and DS did he s a is ical analysis and ake esponsibili y o he accu acy o he da a analysis. DS, AA, JM, SD, NM and JP in e p e ed he da a and edi ed he manusc ip . ADDITIONAL INFORMATION Compe ing in e es s: The au ho s ha e no conflic o in e es o disclosu es. All au ho s decla e no financial o o he ela ionships o ac i i ies ha could appea o ha e influenced he submi ed wo k. E hical app o al: This s udy comply wi h e hical s anda ds. REFERENCES 1. Taylo , R., Sup amaniam, R., Ricka d, M., Eos oes a, J. & Mo ei a, C. In e al b eas cance s in New Sou h Wales, Aus alia, and compa isons wi h ials and o he mammog aphic sc eening p og ammes. J. Med. Sc een. 9,20–25 (2002). 2. Houssami, N., I wig, L. & Cia o, S. Radiological su eillance o in e al b eas cance s in sc eening p og ammes. Lance Oncol. 7, 259–265 (2006). 3. Day, N. e al. Moni o ing in e al cance s in b eas sc eening p og ammes: he Eas Anglian expe ience. J. Med. Sc een. 2, 180–185 (1995). 4. Day, N. E. The NHS b eas cance sc eening p og amme. B . J. Cance 89, 1 (2003). 5. Wa en, R. & Du y, S. In e al cance s as an indica o o pe o mance in b eas sc eening. B eas Cance 7,9–18 (2000). 6. Nys om, L. e al. Long- e m e ec s o mammog aphy sc eening: upda ed o e iew o he Swedish andomised ials. Lance 359, 909–919 (2002). 7. Eu opean Commission. Eu opean Guidelines o Quali y Assu ance in B eas Cance Sc eening and Diagnosis. 4 h edn. In Pe y N. e al. (eds). O fice o O ficial Publica ions o he Eu opean Communi ies, Luxembou g, 2006. 8. Lauby-Sec e an, B. e al. B eas -cance sc eening - iewpoin o he IARC wo king g oup. N. Engl. J. Med. 372, 2353–2358 (2015). 9. Sa keala, T. e al. Episode sensi i i y in associa ion wi h p ocess indica o s in he Finnish b eas cance sc eening p og am. In . J. Cance 118, 174–179 (2006). 10. Benne , R. L., Sella s, S. J. & Moss, S. M. In e al cance s in he NHS b eas cance sc eening p og amme in England, Wales and No he n I eland. B . J. Cance 104, 571–577 (2011). 11. Gio dano, L. e al. Mammog aphic sc eening p og ammes in Eu ope: o ganiza- ion, co e age and pa icipa ion. J. Med. Sc een. 19,72–82 (2012). 12. Finnish Cance Regis y. B eas cance sc eening; Mass sc eening p og amme, 2018. A ailable a h p://www.cance .fi/syopa ekis e i/en/mass-sc eening- egis y/ b eas _cance _sc eening/sc eening_p og amme/. Associa ion o symp oms and in e al b eas cance s in he. . . D Singh e al. 1434 13. Walle , M., Moss, S., Wa son, J. & Mølle , H. The e ec o mammog aphic sc eening and ho mone eplacemen he apy use on b eas cance incidence in England and Wales. Cance Epidemiol. Bioma k. P e . 16, 2257–2261 (2007). 14. Cia o, S., Visioli, C., Paci, E. & Zappa, M. B eas densi y as a de e minan o in e al cance a mammog aphic sc eening. B . J. Cance 90, 393–396 (2004). 15. Wang, H. e al. In e al cance s in he No wegian b eas cance sc eening p og am: equency, cha ac e is ics and use o HRT. In . J. Cance 94, 594–598 (2001). 16. Singh, D., Malila, N., Pokh el, A. & An ila, A. Associa ion o symp oms and b eas cance in popula ion-based mammog aphy sc eening in finland. In . J. Cance 136, E630–E637 (2015). 17. Singh, D., Pi kaniemi, J., Malila, N. & An ila, A. Cumula i e isk o alse posi i e es in ela ion o b eas symp oms in mammog aphy sc eening: a his o ical p o- spec i e coho s udy. B eas Cance Res. T ea . 159, 305–313 (2016). 18. B ouckae , O. e al. B eas cance pheno ype, nodal s a us and palpabili y may be use ul in he de ec ion o o e diagnosed sc eening-de ec ed b eas cance s. Ann. Oncol. 24, 1847–1852 (2013). 19. Gill, G., Luke, C. & Rode , D. P ognos ic impo ance o palpabili y as a ea u e o sc een-de ec ed b eas cance s. J. Med. Sc een. 13,98–101 (2006). 20. Leinonen, M. K., Mie inen, J., Heikkinen, S., Pi kaniemi, J. & Malila, N. Quali y measu es o he popula ion-based Finnish cance egis y indica e sound da a quali y o solid malignan umou s. Eu . J. Cance 77,31–39 (2017). 21. IARC Handbook o Cance P e en ion. B eas Cance Sc eening. IARC, F ance, 2016. A ailable a h p://publica ions.ia c. /Book-And-Repo -Se ies/Ia c-Handbooks-O - Cance -P e en ion/B eas -Cance -Sc eening-2016. 22. Hakama, M., Pokh el, A., Malila, N. & Hakulinen, T. Sensi i i y, e ec and o e - diagnosis in sc eening o cance s wi h de ec able p e-in asi e phase. In . J. Cance 136, 928–935 (2015). 23. Ho ind S. e al. The No wegian B eas Cance Sc eening P og am, 1996–2016: celeb a ing 20 yea s o o ganised mammog aphic sc eening. In: Cance in No way 2016–Cance incidence, mo ali y, su i al and p e alence in No way. Oslo, 2017. 24. Wille A., Michell M., Lee M. Bes P ac ice Diagnos ic Guidelines o pa ien s p esen ing wi h B eas Symp oms. No embe 2010. A ailable a h ps://www. sc ibd.com/documen /140938243/Bes -P ac ice-Diagnos ic-Guidelines- o - Pa ien s-P esen ing-Wi h-B eas -Symp oms. Open Access This a icle is licensed unde a C ea i e Commons A ibu ion 4.0 In e na ional License, which pe mi s use, sha ing, adap a ion, dis ibu ion and ep oduc ion in any medium o o ma , as long as you gi e app op ia e c edi o he o iginal au ho (s) and he sou ce, p o ide a link o he C ea i e Commons license, and indica e i changes we e made. The images o o he hi d pa y ma e ial in his a icle a e included in he a icle’s C ea i e Commons license, unless indica ed o he wise in a c edi line o he ma e ial. I ma e ial is no included in he a icle’s C ea i e Commons license and you in ended use is no pe mi ed by s a u o y egula ion o exceeds he pe mi ed use, you will need o ob ain pe mission di ec ly om he copy igh holde . To iew a copy o his license, isi h p://c ea i ecommons. o g/licenses/by/4.0/. © The Au ho (s) 2018 Associa ion o symp oms and in e al b eas cance s in he. . . D Singh e al. 1435