Biological subtyping of early breast cancer: a study comparing RT-qPCR with immunohistochemistry

PRECLINICAL STUDY Biological sub yping o ea ly b eas cance : a s udy compa ing RT-qPCR wi h immunohis ochemis y Ralph M. Wi z 1 •Ha i Sih o 2 •Jo ma Isola 3 •Pa ¨i i Heikkila ¨ 4 • Pi kko-Liisa Kellokumpu-Leh inen 5 •Pa ¨i i Au inen 6 •Taina Tu peenniemi-Hujanen 7 • Si kku Jy kkio ¨ 8 •So i is Lakis 1 •Ko nelia Schlombs 9 •Ma k Laible 9 • S e an Webe 10 •Sebas ian Eid 11 •Ugu Sahin 9 •Heikki Joensuu 12 Recei ed: 8 Ap il 2016 / Accep ed: 13 May 2016 / Published online: 24 May 2016 ÓThe Au ho (s) 2016. This a icle is published wi h open access a Sp inge link.com Abs ac The biological sub ype o b eas cance in lu- ences he selec ion o sys emic he apy. Dis inc ion be ween luminal A and B cance s depends on consis en assessmen o Ki-67, bu subs an ial in a-obse e and in e -obse e a iabili y exis s when immunohis ochem- is y (IHC) is used. We compa ed RT-qPCR wi h IHC in he assessmen o Ki-67 and o he s anda d ac o s used in b eas cance sub yping. RNA was ex ac ed om a chi al b eas umou issue o 769 women andomly assigned o he FinHe ial. Cance ESR1,PGR,ERBB2 and MKI67 mRNA con en was quan i a ed wi h an RT-qPCR assay. Local pa hologis s assessed ER, PgR and Ki-67 exp ession using IHC. HER2 ampli ica ion was iden i ied wi h ch omogenic in si u hyb idiza ion (CISH) cen ally. The esul s we e co ela ed wi h dis an disease- ee su i al (DDFS) and o e all su i al (OS). qPCR-based and IHC- based assessmen s o ER and PgR showed good conco - dance. Bo h low umou MKI67 mRNA (RT-qPCR) and Ki-67 p o ein (IHC) le els we e p ognos ic o a ou able DDFS [haza d a io (HR) 0.42, 95 % CI 0.25–0.71, P=0.001; and HR 0.56, 0.37–0.84, P=0.005, espec- i ely] and OS. In mul i a iable analyses, cance MKI67 mRNA con en had independen in luence on DDFS (ad- jus ed HR 0.51, 95 % CI 0.29–0.89, P=0.019) while Ki- 67 p o ein exp ession had no any in luence (P=0.266) whe eas bo h assessmen s in luenced independen ly OS. Luminal B pa ien s ea ed wi h doce axel-FEC had mo e a ou able DDFS and OS han hose ea ed wi h ino el- bine-FEC when he sub ype was de ined by RT-qPCR ( o Elec onic supplemen a y ma e ial The online e sion o his a icle (doi:10.1007/s10549-016-3835-7) con ains supplemen a y ma e ial, which is a ailable o au ho ized use s. &Ralph M. Wi z [email p o ec ed] Ha i Sih o [email p o ec ed] Jo ma Isola [email p o ec ed] Pa ¨i i Heikkila ¨ [email p o ec ed] Pi kko-Liisa Kellokumpu-Leh inen [email p o ec ed] Pa ¨i i Au inen [email p o ec ed] Taina Tu peenniemi-Hujanen [email p o ec ed] Si kku Jy kkio ¨ [email p o ec ed] So i is Lakis [email p o ec ed] Ko nelia Schlombs [email p o ec ed] Ma k Laible [email p o ec ed] S e an Webe [email p o ec ed] Sebas ian Eid [email p o ec ed] Ugu Sahin [email p o ec ed] Heikki Joensuu [email p o ec ed] 1 STRATIFYER Molecula Pa hology GmbH, We hmanns . 1c, 50935 Cologne, Ge many 2 Labo a o y o Molecula Oncology, T ansla ional Cance Biology P og am, Uni e si y o Helsinki, Helsinki, Finland 123 B eas Cance Res T ea (2016) 157:437–446 DOI 10.1007/s10549-016-3835-7 DDFS, HR 0.52, 95 % CI 0.29–0.94, P=0.031), bu no when de ined using IHC. B eas cance sub ypes app oxi- ma ed wi h RT-qPCR and IHC show good conco dance, bu cance MKI67 mRNA con en co ela ed sligh ly be e wi h DDFS han Ki-67 exp ession. The indings based on MKI67 mRNA con en sugges ha pa ien s wi h luminal B cance bene i mo e om doce axel-FEC han om ino elbine-FEC. Keywo ds B eas cance Molecula sub ypes Ki-67  P edic ion Immunohis ochemis y RT-qPCR Abb e ia ions ESR1/ER Oes ogen Recep o alpha CISH Ch omogenic in si u hyb idiza ion Cq Quan i ica ion cycle DDFS Dis an disease- ee su i al FEC Fluo ou acil & epi ubicin, cyclophosphamide chemo he apy FFPE Fo malin ixed pa a in embedded GOI Gene o in e es HR Haza d a io IHC Immunohis ochemis y mRNA Messenge ibonucleic acid MKI67/ Ki-67 Ma ke o p oli e a ion Ki-67 OS O e all su i al PGR/PgR P oges e one ecep o REF Re e ence gene RT-qPCR Re e se ansc ip ion quan i a i e eal- ime polyme ase chain eac ion ERBB2/ HER2 Human epide mal g ow h ac o ecep o 2 In oduc ion Biological sub yping o b eas cance is an in eg al pa o he s anda d e alua ion o pa ien s diagnosed wi h b eas cance . Sub yping can be done wi h gene exp ession a ays [1], bu he molecula sub ypes a e equen ly app oxi- ma ed wi h immunohis ochemis y (IHC) due o i s wide a ailabili y and low cos . Howe e , assays o cance oes ogen ecep o (ER), p oges e one ecep o (PgR) and human epide mal g ow h ac o ecep o -2 (HER2) exp ession by IHC ha e an up o 20 % isk o disco dan o e oneous esul s [2,3], and making a dis inc ion be ween luminal A and luminal B b eas cance equi es assessmen wi h he p oli e a ion ma ke Ki-67, which is p one o high in a- and in e -obse e assessmen a i- abili y [4,5]. In his s udy, we compa ed assessmen o b eas cance key bioma ke s, ER, PgR, HER2 and Ki-67 quan i a i ely using RT-qPCR wi h hei assessmen using IHC o in si u hyb idiza ion as a pa o he clinical ou ine in b eas cance sub yping and p edic ion o pa ien ou come. We hypo hesized ha quan i ying Ki-67 wi h RT-qPCR migh esul in mo e obus ou come p edic ions. To ou knowl- edge, ew such compa a i e da a a e a ailable. Me hods Pa ien s The clinical da a and b eas umou issue samples we e col- lec ed wi hin he FinHe ial (iden i ie ISRCTN76560285), whe e 1010 women wi h axilla y node–posi i e o high- isk axilla y node-nega i e b eas cance we e andomly assigned be ween Oc obe 2000 and Sep embe 2003 o ecei e ei he h ee cycles o doce axel ollowed by h ee cycles o luo- ou acil, epi ubicin and cyclophosphamide (FEC) o h ee cycles o ino elbine ollowed by h ee cycles o FEC [6,7]. B eas umou e bB2 (HER2) copy numbe s we e de e - mined cen ally by ch omogenic in si u hyb idiza ion (CISH), and women wi h HER2-posi i e cance (n=232) had a second andomisa ion be ween nine weekly in usions o as uzumab, gi en concomi an ly wi h ei he doce axel o ino elbine, and simila chemo he apy wi hou as uzu- mab. A e a median ollow–up ime o 62 mon hs since andomisa ion, women assigned o doce axel had be e dis an disease- ee su i al (DDFS, he p ima y objec i e) han hose assigned o ino elbine (HR 0.66, 95 % CI 0.49–0.91; P=0.010) [6]. The absolu e bene i in 5-yea DDFS in a ou o he doce axel plus FEC egimen was 5.2 % (86.8 s 81.6 %), and 3.3 % (92.6 s 89.3 %) o o e all su i al (OS) ac oss all biological sub ypes [6]. 3 Labo a o y o Cance Biology, Ins i u e o Medical Technology, Tampe e, Finland 4 Depa men o Pa hology, HUSLAB, Helsinki Uni e si y Hospi al and Helsinki Uni e si y, Helsinki, Finland 5 Depa men o Oncology, Tampe e Uni e si y Hospi al and Uni e si y o Tampe e, Tampe e, Finland 6 Cance Cen e , Kuopio Uni e si y Hospi al, Kuopio, Finland 7 Depa men o Oncology and Radio he apy, Oulu Uni e si y Hospi al, Oulu, Finland 8 Depa men o Oncology, Tu ku Uni e si y Hospi al, Tu ku, Finland 9 BioNTech Diagnos ics GmbH, Mainz, Ge many 10 Acomed S a is ik, Leipzig, Ge many 11 Ins i u e o Pa hology a he S -Elisabe h-Hospi al, Cologne, Ge many 12 Depa men o Oncology, Helsinki Uni e si y Hospi al and Uni e si y o Helsinki, Helsinki, Finland 438 B eas Cance Res T ea (2016) 157:437–446 123 Immunohis ochemis y Immunos aining o ER, PgR, HER2 and Ki-67 was pe o med on issue sec ions cu om o malin- ixed, pa a in-embedded (FFPE) umou issue a he local pa hology labo a o ies o he 17 s udy si es (all loca ed in Finland) acco ding o each labo a o y’s s anda d p ocedu es. ER and PgR we e conside ed posi i e when 10 % o mo e o he cance cells s ained posi i ely. Ki-67 assays we e analysed by es ima ing he p opo ion o posi i ely s aining cance cell nuclei ou o all cance cell nuclei in he issue sec ion, and he esul was p o ided as a pe - cen age anging om 0 o 100 %. Fo he p esen s udy, Ki-67 exp ession was conside ed posi i e when C20 % o cance cell nuclei s ained posi i ely. Local pa hologis s in e p e ed he ER, PgR and Ki-67 immunos aining esul s, as pe each ins i u e’s s anda d p ac ice. Ch omogenic in si u hyb idiza ion (CISH) Tumou s wi h a sco e o 2?o 3?(on a scale o 0 o 3?) o HER2 exp ession in IHC we e u he analysed o HER2 gene ampli ica ion by CISH in one o wo cen al labo a o ies. The HER2 s a us was conside ed posi i e when six o mo e gene copies pe nucleus we e p esen . As in he o iginal ial [6,7], in he p esen s udy, cance HER2 s a us was conside ed posi i e whene e CISH o HER2 was posi i e, and nega i e whene e CISH was nega i e, ega dless o he deg ee o HER2 p o ein exp ession in IHC. RT-qPCR A e pa hologic con i ma ion o ep esen a i eness o he issue sec ions o p esence o cance , a single whole- ace 10-lm- hick slice om each FFPE umou block was p ocessed wi h he RNX ac Ò RNA ex ac ion ki (BioN- Tech Diagnos ics GmbH, Mainz) using a magne ic pa i- cle-based assay (Supplemen al ile 1A). RT-qPCR was done wi h he MammaType Ò ki (BioNTech Diagnos ics GmbH, Mainz) o ESR1, PGR, ERBB2 and MKI67, and he wo e e ence genes B2 M and CALM2 on a Ve san kPCR sys em (Siemens, E langen, Ge many) by applying one cycle o p ime -speci ic e e se ansc ip ion ollowed by 40 cycles o nucleic acid ampli ica ion (Supplemen al ile 1B). The median quan i ica ion cycle (Cq) o each o he ou genes o in e es (GOI) we e no malized agains he wo e e ence genes (REF) and p esen ed as DDCq alues ela i e o he posi i e con ol, ob ained a e sub- ac ing he DCq alue o he posi i e con ol (pc) om he DCq o he sample (s) by he o mula 40 DDCq GOIðÞ S¼40 Cq GOI½ S- meanCq REF½ S  Cq GOI½ pc- meanCq REF½ pc  : To exclude a majo in luence o a a ying umou cell con en o he assay esul s, sensi i i y s udies we e unde aken simila ly as p e iously epo ed [8]. A se ies o ex eme cases wi h low con en o in asi e ca cinoma and a ying amoun o DCIS we e analysed be o e and a e mac odissec ion and i could be con i med ha he TCC did no in luence he inal es esul [9, Laible e al. submi ed]. The e o e, a majo in luence o TCC on MKI67 mRNA exp ession can be excluded. Cu -o s o he ma ke s ERBB2,ESR1 and PGR we e de ined in an independen echnical coho based on e e ence pa hology IHC esul s. P ognos ic and p edic i e alue o MKI67 cu -o s had p e iously been analysed by es ing objec i e cu -o s in 562 A yme ix U133 A da ase s om b eas cance pa ien coho s ha ing ecei ed ei he no sys emic he apy, only endoc ine ea men o chemo-endoc ine egimen [10]. In iew o hese analyses, he MKI67 cu -o was se a he 3 d qua ile o he no mally dis ibu ed MKI67 exp ession da a om 90 FFPE b eas cance e e ence umou samples and hus ough o e lec a co ela e o he s anda d Ki-67 cu - o a 20 % posi i ely s ained nuclei. De ini ion o b eas cance biological sub ypes A e de ining each o he ou bioma ke s ei he posi i e o nega i e, he molecula sub ype o each umou was de e - mined using a sligh ly modi ied e sion o he cu en ly p oposed IHC-based b eas cance molecula sub yping algo i hm [1] (Supplemen al File 1C). In b ie , luminal A cance s we e de ined as ha ing high ESR1 and/o PGR mRNA con en and low ERBB2 and MKI67 con en . Luminal B cance s we e de ined as ha ing high cance ESR1 and MKI67 con en , o high ESR1 con en bu low PGR and ERBB2 con en . Cance s wi h a high ERBB2 mRNA con en we e conside ed as HER2-posi i e cance s and we e no u he ca ego ized in o luminal and non-luminal (‘‘en- iched’’) lesions. T iple-nega i e cance s consis ed o can- ce s ha had low ESR1,PGR and ERBB2 mRNA con en i espec i e o cance MKI67 mRNA con en . The same scheme was used o ca ego ize he cance s acco ding o he IHC and CISH esul s, bu using p o ein exp ession (a IHC) and he numbe o HER2 gene copies (a CISH) in place o cance mRNA con en . Fo example, cance s ha we e posi i e o ER and PgR (wi h C10 % o he nuclei ha we e posi i e in each s aining), HER2 nega i e (by CISH) and had low Ki-67 ( 20 % o nuclei s ained posi i ely a IHC) we e conside ed luminal A cance s. B eas Cance Res T ea (2016) 157:437–446 439 123 S a is ical me hods The esul s we e analysed acco ding o a s a is ical analysis plan w i en and app o ed p io o he ini ia ion o he s udy, and he RT-qPCR esul s we e in e p e ed blinded o he clinical in o ma ion. Kappa (j) s a is ic nume ic alues a e ca ego ized in o poo (B0.2), ai ([0.2–0.4), mode a e ([0.4–0.6), good ([0.6–0.8) and e y good ([0.8) associ- a ions, and we e used as a measu e o posi i e pe cen ag eemen (PPA), nega i e pe cen ag eemen (NPA) and o e all pe cen ag eemen (OPA). The es s a e accompa- nied by hei espec i e 95 % con idence in e als (95 % CI). A wo-sided P alue 0.05 was conside ed signi ican . The p ima y clinical endpoin was DDFS, de ined as he ime pe iod be ween he da e o andomisa ion and he da e o i s dis an me as asis o he da e o dea h when dea h p eceded de ec ion o dis an ecu ence. O e all su i al (OS) was de ined as he ime pe iod be ween he da e o andomisa ion and he da e o dea h. Su i al was analysed using he Kaplan–Meie me hod. Uni a iable and mul i a iable Cox p opo ional haza ds models we e cons uc ed o compa e p ognosis be ween g oups and o s udy he in e ac ions be ween a iables. Haza d a ios (HRs) we e calcula ed using a uni a iable Cox model. In mul i a iable Cox models, a backwa d selec ion p ocedu e was used o adjus o he co a iables. Resul s Pa ien s An RT-qPCR assay o ESR1,PGR,ERBB2 and MKI67 was success ully pe o med om b eas cance issues o 769 (76 %) ou o he 1010 pa ien s en e ed o he FinHe ial. In he emaining 241 cases, cance issues we e no a ail- able, and he issue block did no consis mos ly o cance cells, o RNA ex ac ion did no yield good-quali y mRNA. We included in his s udy all 719 (71 % ou o 1010) cases wi h success ul RT-qPCR assay o he ou genes and wi h IHC da a a ailable o sub yping. The inaccessibili y a e o he issue samples was simila ac oss he s udy ea men a ms (a modi ied CONSORT diag am shown in Supple- men al ile 1D). The cha ac e is ics o he pa ien s and umou s included in he p esen s udy (Table 1) we e simila o hose o he en i e FinHe ial coho [7]. The median age o he pa ien s a s udy en y was 50.9 yea s ( ange, 25.5–65.8). Tumou s had a mean diame e o 26 mm ±16 mm (6–150 mm), and he majo i y (n=637, 88.6 %) had gi en ise o egional lymph node me as ases a he ime o he diagnosis. The e we e 511 (71.1 %) ER- posi i e, 395 (54.9 %) PgR-posi i e and 163 (22.7 %) HER2-posi i e cance s. A e andom alloca ion, 357 (49.7 %) pa ien s we e ea ed wi h doce axel plus FEC, 362 (50.4 %) wi h ino elbine plus FEC and 83 (50.9 %) o he 163 pa ien s wi h HER2-posi i e cance ecei ed as uzumab. The median ollow-up ime a e andomisa- ion was 62 mon hs, du ing which ime pe iod 112 pa ien s had dis an cance ecu ence and 62 died. Conco dance be ween mRNA and IHC assays Tumou ESR1,PGR and ERBB2 mRNA con en assessed by RT-qPCR and he co esponding p o ein exp essions Table 1 Pa ien demog aphics, clinicopa hological da a and e- quencies o ma ke bina y ca ego ies N(%) pT N=719 1 290 (40.33) 2 369 (51.32) 3 47 (6.54) 4 13 (1.81) pN N=718 0 80 (11.14) 1 621 (86.49) 2 16 (2.23) 3 1 (0.14) His ological ype N=719 Duc al 575 (79.97) Lobula 131 (18.22) Papilla y 2 (0.28) Mucinous 2 (0.28) Medulla y 9 (1.25) His ological g ade N=694 I 102 (14.70) II 287 (41.35) III 305 (43.95) Adju an chemo he apy N=719 Doce axel 357 (49.65) Vino elbine 362 (50.35) HER2-pos N=163 T as uzumab 83 (50.92) No T as uzumab 80 (49.08) Type o Su ge y N=719 To al Mas ec omy 430 (59.81) B eas Conse ing 289 (40.19) 440 B eas Cance Res T ea (2016) 157:437–446 123 de e mined by IHC o ER and PgR and DNA ampli ica ion s a us by CISH o HER2 showed good conco dance, whe eas cance MKI67 mRNA con en and p o ein exp es- sion co ela ed only mode a ely well (Table 2). Many o he disco dan cases be ween IHC and he mRNA assay had a high cance MKI67 mRNA con en , bu despi e his, 20 % o cance cell nuclei s ained posi i ely wi h IHC (Fig. 1). P ognos ic alue o cance MKI67 mRNA con en and Ki-67 exp ession Pa ien s wi h low b eas umou MKI67 mRNA con en o low ( 20 %) Ki-67 exp ession had mo e a ou able DDFS and OS as compa ed o hose wi h high MKI67 mRNA con en o Ki-67 exp ession. Each me hod p oduced oughly simila haza d a ios o DDFS and OS (Fig. 2). In a mul i a ia e Cox eg ession analysis whe e he ype o chemo he apy ( ino elbine-FEC o doce axel-FEC), he axilla y nodal s a us (pN0, pN1, pN2 o pN3), umou size (as a con inuous a iable), his ological g ade (as a con- inuous a iable) and cance MKI67 mRNA con en (as a con inuous a iable) we e en e ed as co a iables, low umou MKI67 mRNA con en was independen ly associ- a ed wi h a ou able DDFS (adjus ed HR 0.51; 95 % CI, 0.29–0.90; P=0.019) oge he wi h a nega i e axilla y nodal s a us (P 0.0001) and small cance size (P=0.006). A low cance MKI67 mRNA con en was also independen ly associa ed wi h a ou able OS (adjus ed HR 0.44; 95 % CI, 0.23-0.87; P=0.018) in addi ion o he axilla y nodal s a us (P=0.003) and umou size (P=0.006). When Ki-67 p o ein exp ession was en e ed in o he same models in place o cance mRNA con en , Ki-67 was no signi ican ly associa ed wi h DDFS (P=0.266), bu when OS was selec ed as he endpoin , low cance Ki-67 exp ession was associa ed wi h a ou - able su i al (adjus ed HR 0.43; 95 % CI, 0.24–0.77; P=0.005) oge he wi h he axilla y nodal s a us (P=0.002) and small umou size (P=0.006). Conco dance o molecula sub yping wi h IHC and RT-qPCR The me hod o Ki-67 assessmen had subs an ial impac on making he dis inc ion be ween luminal A and B cance s. O he 189 cance s ha we e classi ied as luminal A by IHC/CISH, only 102 (54.0 %) we e simila ly classi ied, when MKI67 mRNA exp ession was used in place o Ki-67 p o ein s aining wi h he 87 disco dan cases being classi- ied as ei he luminal B (n=75, 39.7 %) o HER2 posi i e (n=12, 6.4 %, Table 3). O he 251 cance s ha we e classi ied as luminal B by IHC/CISH, 180 (71.7 %) we e simila ly classi ied using MKI67 mRNA exp ession, 48 (19.1 %) we e classi ied as luminal A, 17 (6.8 %) as HER2 posi i e and 6 (2.4 %) as iple nega i e. O he 156 and 294 umou s classi ied as luminal A and luminal B by RT- Table 2 Ag eemen be ween RT-qPCR-based and IHC-based bioma ke assessmen s ESR1 PGR ERBB2 MKI67 Conco dance 660/719 (91.8 %) 593/719 (82.5 %) 660/719 (91.8 %) 516/688 (75.0 %) PPA 490/511 (95.9 %) 368/395 (93.2 %) 140/163 (85.9 %) 369/414 (89.1 %) NPA 170/208 (81.7 %) 225/324(69.4 %) 520/556 (93.5 %) 147/274 (53.7 %) Kappa s a is ic 0.80 (0.75–0.85) 0.64 (0.58–0.70) 0.77 (0.72–0.83) 0.45 (0.38–0.52) P 0.0001 P 0.0001 P 0.0001 P 0.0001 PPA Posi i e pe cen ag eemen , NPA Nega i e pe cen ag eemen Fig. 1 A sca e plo depic ing he ela ion be ween umou MKI67 mRNA con en measu ed wi h RT-qPCR, and Ki-67 exp ession de e mined by immunohis ochemis y (IHC). Ve ical axis, umou Ki-67 exp ession (IHC, %); ho izon al axis, umou ela i e MKI67 mRNA exp ession. The cu -o o posi i i y was 20 % in he Ki-67 p o ein assays ( he ho izon al line) and 34.8 in he MKI67 mRNA (qPCR) assays ( he e ical line). Sec ions Aand Ddepic he disco dan cases, and sec ions Band Cdepic he conco dan cases B eas Cance Res T ea (2016) 157:437–446 441 123 qPCR, espec i ely, 102 (65.4 %) and 180 (61.2 %) we e classi ied as luminal A o B also wi h IHC/CISH. In luence o cance MKI67 mRNA exp ession-based and Ki-67 p o ein exp ession-based sub ypes on ou come The e was no signi ican di e ence in DDFS o OS be ween pa ien s ea ed wi h adju an doce axel plus FEC and hose ea ed wi h ino elbine and FEC in he subse s wi h luminal A, HER2-posi i e o iple-nega i e b eas cance when each sub ype was de ined ei he wi h IHC/ CISH o wi h RT-qPCR (DDFS and OS s a is ics o each sub ype acco ding o chemo he apy agen shown in Sup- plemen al ile 1E). In e es ingly, when luminal B sub ype was de ined by MKI67 mRNA exp ession, pa ien s ea ed wi h doce axel plus FEC had signi ican ly mo e a ou able DDFS and OS as compa ed wi h hose ea ed wi h ino elbine plus FEC ( o DDFS, HR 0.52, 95 % CI 0.29–0.94, P=0.031; OS, HR 0.24, 95 % CI 0.09–0.65, P=0.005). In con as no signi ican di e ence in DDFS o OS was ound, when he luminal B sub ype was de ined Fig. 2 In luence o cance MKI67 mRNA exp ession and Ki-67 p o ein exp ession on DDFS (panels aand c) and su i al (panels band d). Resul s ob ained by measu ing MKI67 mRNA exp ession a e shown in panels aand b, and hose ob ained by assessing Ki-67 p o ein exp ession in panels cand d Table 3 Conco dance o b eas cance sub ypes when cance Ki-67 exp ession is assessed wi h IHC and MKI67 mRNA exp ession wi h RT-qPCR RT-qPCR-based Luminal A Luminal B HER2 pos TNBC To al N(%) N(%) N(%) N(%) N(%) IHC-based Luminal A 102 65.4 75 25.5 12 6.8 0 0.0 189 26.3 Luminal B 48 30.8 180 61.2 17 9.7 6 6.5 251 34.9 HER2 pos 5 3.2 12 4.1 140 79.6 6 6.5 163 22.7 TNBC 1 0.6 27 9.2 7 4.0 81 87.1 116 16.1 To al 156 100.0 294 100.0 176 100.0 93 100.0 719 100.0 442 B eas Cance Res T ea (2016) 157:437–446 123 wi h Ki-67 p o ein exp ession (P[0.10 o bo h analyses; Fig. 3). The ype o adju an chemo he apy ( es ed doce axel plus FEC s ino elbine plus FEC) had an independen in luence on DDFS in he subse o pa ien s who had luminal B cance de ined by cance MKI67 mRNA con en in a mul i a iable analysis (HR 0.44; 95 % CI 0.23–0.84, P=0.013), oge he wi h cance his ological g ade ( es ed as a con inuous a iable; HR 1.67, 95 % CI 1.03–2.72, P=0.039) and umou size ( es ed as a con inuous ac o ; HR 1.02, 95 % CI 1.00–1.04, P=0.026). Simila ly, when OS was used as he end poin in place o DDFS and he luminal B sub ype was de ined by cance MKI67 mRNA con en , doce axel-con aining chemo he apy was indepen- den ly associa ed wi h a ou able su i al (HR 0.22; 95 % CI 0.08–0.60, P=0.003) oge he wi h his ological g ade (HR 2.29, 95 % CI 1.15–4.57; P=0.019), while umou size los i s signi icance. Unlike MKI67 mRNA con en , Ki- 67 p o ein exp ession did no ha e independen in luence on DDFS o OS in hese models. When he luminal B sub ype was de ined wi h umou MKI67 mRNA con en , he in e ac ion wi h he ype o adju an chemo he apy gi en was signi ican (P=0.040) when OS was selec ed as he end poin , bu no when DDFS was conside ed (P=0.352). No in e ac ion wi h ei he OS o DDFS and he ype o adju an chemo he apy was p esen when he luminal sub ype was de ined wi h Ki-67 p o ein exp ession (P=0.658 and 0.699, espec i ely). Discussion We app oxima ed commonly used b eas cance biologi- cal sub ypes using RT-qPCR and compa ed he esul s wi h he sub ypes de ined by IHC (and wi h CISH o de ec HER2 ampli ica ion) wi hin he amewo k o a la ge andomized clinical ial. The sub ypes de ined wi h each me hod ag eed mode a ely well wi h mos disc ep- ancy occu ing in he luminal B sub ype. Bo h high cance Ki-67 p o ein exp ession and high MKI67 mRNA con en we e associa ed wi h un a ou able DDFS and OS in a uni a iable analysis wi h app oxima ely simila haza d a ios, bu only umou MKI67 mRNA con en emained signi ican in a mul i a iable model o DDFS when bo h pa ame e s we e en e ed in o he same model a e a s epwise selec ion p ocess o he co a iables such as umou size, nodal s a us, his ological g ade and he ype o ea men gi en. Fig. 3 Dis an me as asis- ee su i al (panels aand b) and o e all su i al (panels cand d) o pa ien s ea ed wi h adju an doce axel plus FEC and hose ea ed wi h ino elbine plus FEC in he subse o pa ien s wi h luminal B b eas cance . Panels aand c, he luminal B sub ype was de ined wi h MKI67 mRNA exp ession; panels band d, he luminal B sub ype was de ined wi h Ki-67 p o ein exp ession B eas Cance Res T ea (2016) 157:437–446 443 123 A key di e ence be ween luminal A and luminal B sub ypes is a highe cell p oli e a ion a e in he la e , which is o en assessed by es ima ing he p opo ion o cance cells ha s ain posi i ely o Ki-67 a e immuno- his ochemical s aining. In e es ingly, when he luminal B ype was de ined using cance MKI67 mRNA con en in place o Ki-67 exp ession assessed wi h immunohis o- chemis y, pa ien s wi h luminal B b eas cance we e ound o bene i mo e om adju an doce axel plus FEC han om adju an ino elbine plus FEC, which associa- ion could no be de ec ed when he luminal B b eas cance sub ype was de ined by Ki-67 p o ein exp ession wi h immunohis ochemical s aining. Biological sub yping o b eas cance is he basis o selec ion o sys emic cance ea men [1]. O he ou bioma ke s commonly used o his pu pose, i.e. ER, PgR, HER2 and Ki-67, he assays o Ki-67 ha e u ned ou he mos challenging ones o s anda dize and o make ep o- ducible. Fo example, in a s udy ca ied ou in a ew leading pa hology labo a o ies, he e was subs an ial a iabili y be ween he labo a o ies in sco ing o Ki-67 exp ession om sha ed b eas cance issue slides s ained wi h IHC, and a emp s o educe he in e labo a o y a iabili y we e only pa ially success ul [4]. In he p esen s udy, IHC s aining o Ki-67 was done locally in many pa hology labo a o ies using he ins i u ional s aining p o ocols and was assessed by many pa hologis s, whe eas cance MKI67 mRNA con en was de e mined cen ally in one labo a o y. To educe he po en ial a iabili y in Ki-67 s aining and sco ing, we con- side ed ca ying ou s aining o Ki-67 also cen ally, bu due o he di icul ies o s anda dize Ki-67 immunos aining e en in leading labo a o ies and o es ablish a e e ence p ocedu e [4], we p e e ed o use he Ki-67 s aining esul s epo ed o iginally by he local labo a o ies om whole umou issue sec ions as he compa a o o he MKI67 mRNA assay. Image analysis o Ki-67 om IHC s ained slides is a p omising me hod o imp o e he ep oducibili y o Ki-67 sco ing om immunos ained slides, bu , o ou knowledge, no s anda d pa ame e alues o sco ing o he nuclei as ei he posi i e o nega i e a e a ailable. To es ima e how well he locally assessed Ki-67 assays done om whole umou issue sec ions migh co ela e wi h a cen ally done Ki-67 assay, we analysed cance Ki-67 exp ession om TMAs (as whole umou sec ions we e no a ailable) con- aining issue om 745 b eas cance s using image analysis [11]. The median cance Ki-67 exp ession u ned ou o be simila wi h image analysis and locally done IHC (19.7 and 20.0 %, espec i ely), and he wo assays showed s ong co ela ion (P 0.0001, Spea man’s ho 0.633). These obse a ions sugges ha cen ally done image analysis o Ki-67 migh ha e esul ed in simila conclusions had i been selec ed as he compa a o assay in place o he local Ki-67 IHC assays. The sub ypes de ined wi h MKI67 mRNA we e associ- a ed wi h su i al ou comes ha ag ee well wi h he esul s ob ained wi h IHC om o he clinical ials [9,10,12–14]. Pa ien s wi h he luminal A sub ype had he bes 5-yea DDFS, pa ien s wi h HER2 posi i e and iple-nega i e cance had he leas a ou able ou comes, while pa ien s wi h luminal B cance had an ou come in e media e o hese sub ypes (see Supplemen File 1E). These esul s a e well in ag eemen despi e sligh dissimila i ies in he de - ini ion o luminal B and HER2-posi i e sub ypes be ween he ials. Taxane-con aining adju an egimens a e e ec i e in he ea men o ea ly b eas cance bu a e associa ed wi h side e ec s, and he e o e, me hods o op imize pa ien selec ion o egimens ha con ain a axane a e needed. The cu en inding ha pa ien s wi h luminal B cance ha e longe DDFS and OS when ea ed wi h doce axel plus FEC as compa ed wi h ino elbine plus FEC is sup- po ed by obse a ions made by Jacquemie e al. and Ni z e al. who ound ha chemo he apy con aining doce axel was associa ed wi h a signi ican educ ion in he isk o elapse in he subse o pa ien s wi h luminal B b eas cance in he PACS 01 ial [13] and WSG-AGO EC-Doc ial [12], espec i ely. Bo h o hese ials compa ed doce axel-con aining egimens wi h s anda d an h acy- cline-con aining ea men s. In he BCIRG 001 ial ha compa ed doce axel, doxo ubicin and cyclophosphamide (TAC) e sus luo ou acil, doxo ubicin and cyclophos- phamide (FAC) in he ea men o ope able node-posi i e b eas cance , only pa ien s wi h ER-posi i e umou s wi h ei he high Ki-67 exp ession o HER2 o e exp ession had a s a is ically signi ican imp o emen in disease- ee su - i al when ea ed wi h TAC [14]. Howe e , unlike hese s udies, we did no ind a su i al bene i om he doc- e axel-con aining egimen in he subse o women wi h HER2 posi i e cance . In FinHe , hal o he pa ien s wi h HER2-ampli ied cance we e andomly assigned o ecei e adju an as uzumab, which may ha e masked he po en ial doce axel bene i in his sub ype and may ha e educed he s a is ical powe o de ec he associa ion. The PAM50 gene exp ession a ay has also been e al- ua ed in p edic ing he po en ial bene i o adding a axane o an h acycline-based chemo he apy, bu none o he PAM50-de i ed sub ypes including he luminal B sub ype we e p edic i e o a axane bene i in he GEICAM/9966 and he NCIC CTG MA.21 andomized phase III ials [15, 16]. Simila ly he Endop edic gene exp ession assay did no p edic axane bene i in he GEICAM/9966 s udy popula ion [17]. The limi a ions o he s udy include he e ospec i e na u e o he s udy, al hough we de e mined umou MKI67 mRNA wi hou knowledge o he clinical da a and planned he s a is ical analyses p ospec i ely. We es ed he 444 B eas Cance Res T ea (2016) 157:437–446 123 me hods wi hin he con ex o a ela i ely la ge andomized ial bu lacked a alida ion se ies, and some subg oup analyses ha e limi ed powe . Howe e , he PCR me hod used u ned ou o be ep oducible ac oss mul iple es ing si es o all ou bioma ke s including MKI67 mRNA (Laible e al., manusc ip submi ed o publica ion). The de ails o he IHC me hods used o assaying Ki-67 in he local pa hology labo a o ies we e no cap u ed du ing he FinHe ial, as Ki-67 was no a p o ocol-manda ed assay, bu mos pa hology labo a o ies in Finland assess Ki-67 om he umou ho spo a eas. The ecommended cu -o o ER and PgR posi i i y is now 1 % and no longe 10 % as i was a he ime when he FinHe ial acc ued pa ien s, bu he p opo ion o b eas cance s whe e ER o PgR a e exp essed in 1 % o 10 % o nuclei is small [18]. Conclusions Measu ing o cance ESR1,PGR and ERBB2 mRNA co - ela ed well wi h he esul s ob ained wi h IHC and CISH in clinical pa hology labo a o ies. Tumou MKI67 mRNA con en quan i a ed wi h RT-qPCR is associa ed wi h DDFS and OS o pa ien s ea ed wi h mode n adju an egimens. The esul s sugges ha assessmen o umou MKI67 mRNA con en may be aluable o selec ion o pa ien s o doce axel-con aining adju an he apy. Since he immunohis ochemical assay esul s o Ki-67 exp es- sion a e challenging o ans e be ween labo a o ies, and he assay o measu ing cance MKI67 mRNA con en wi h RT-qPCR migh be less challenging o s anda dize han IHC s ainings, pe o ming s udies ha e alua e in e labo- a o y compa isons o cance ESR1,PGR,ERBB2 and MKI67 mRNA con en using RT-qPCR a e wa an ed. Acknowledgmen s We hank Elke Vel up, Susanne Scha , Silke Claas and To s en Ach o excellen echnical suppo in de eloping molecula sub yping echnologies, and. D s. Thomas Kelle and S e an Webe o pe o ming s a is ical analyses. The s udy was suppo ed inancially by he Academy o Finland, Cance Socie y o Finland, Jane and Aa os E kko Founda ion, he Sig id Juselius Founda ion, and he Resea ch Funds o he Helsinki Uni e si y Hospi al. Au ho con ibu ions HJ, RMW and US we e in ol ed in he concep ion o he s udy. RMW pe o med he RT-qPCR assays; HS, JI, PH, P-LK-L, PA, TT-H, SJ and HJ p o ided s udy da a and ma e ials; SW pe o med he s a is ical analysis and w o e he s a- is ical plan; RMW, HJ, SL, KS, ML, SE and US in e p e ed he da a; RMW, SL and HJ d a ed he manusc ip ; all au ho s ead and app o ed he inal manusc ip . Compliance wi h e hical s anda ds Con lic o in e es s RM Wi z is he ounde and CEO o STRA- TIFYER Molecula Pa hology GmbH. S Eid has s ock op ions in STRATIFYER Molecula Pa hology GmbH. S Lakis is an employee o STRATIFYER Molecula Pa hology GmbH. U Sahin is he oun- de and CEO o BioNTech Diagnos ics GmbH. K Schlombs and M Laible a e employees o BioNTech Diagnos ics GmbH. No po en ial con lic s o in e es s we e disclosed by he o he au ho s. E hical s anda ds Each s udy pa icipan p o ided signed in o med consen o he clinical ial, and a sepa a e consen o use umou issue o esea ch ela ed o he FinHe s udy. An E hics Commi ee o he Helsinki Uni e si y Hospi al app o ed he p o ocol o he cu en s udy (HUS 124/13/03/02/2014, pe mission g an ed on Ap 9, 2014). Open Access This a icle is dis ibu ed unde he e ms o he C ea i e Commons A ibu ion-NonComme cial 4.0 In e na ional License (h p://c ea i ecommons.o g/licenses/by-nc/4.0/), which pe - mi s any noncomme cial use, dis ibu ion, and ep oduc ion in any medium, p o ided 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. Re e ences 1. Goldhi sch A, Wine EP, Coa es AS, Gelbe RD, Picca -Gebha M, Thu ¨ limann B e al (2013) Pe sonalizing he ea men o women wi h ea ly b eas cance : highligh s o he S Gallen In e na ional Expe Consensus on he P ima y The apy o Ea ly B eas Cance 2013. Ann Oncol 24:2206–2223 2. Hammond ME, Hayes DF, Dowse M, All ed DC, Hage y KL, Bad e S e al (2010) Ame ican Socie y o Clinical Oncology/ College o Ame ican Pa hologis s guideline ecommenda ions o immunohis ochemical es ing o es ogen and p oges e one ecep o s in b eas cance . A ch Pa hol Lab Med 134:907–922 3. Wol AC, Hammond ME, Schwa z JN, Hage y KL, All ed DC, Co e RJ e al (2007) Ame ican Socie y o Clinical Oncology/ College o Ame ican Pa hologis s guideline ecommenda ions o human epide mal g ow h ac o ecep o 2 es ing in b eas can- ce . J Clin Oncol 25:118–145 4. Polley MY, Leung SC, McShane LM, Gao D, Hugh JC, Mas- opasqua MG e al (2013) An in e na ional Ki67 ep oducibili y s udy. J Na l Cance Ins 105:1897–1906 5. Va ga Z, Diebold J, Dommann-Sche e C, F ick H, Kaup D, Noske A e al (2012) How eliable is Ki-67 immunohis ochem- is y in g ade 2 b eas ca cinomas? A QA s udy o he Swiss Wo king G oup o B eas - and Gynecopa hologis s. PLoS One 7:e37379 6. Joensuu H, Bono P, Ka aja V, Alanko T, Kokko R, Asola R e al (2009) Fluo ou acil, epi ubicin, and cyclophosphamide wi h ei he doce axel o ino elbine, wi h o wi hou as uzumab, as adju an ea men s o b eas cance : inal esul s o he FinHe T ial. J Clin Oncol 27:5685–5692 7. Joensuu H, Kellokumpu-Leh inen PL, Bono P, Alanko T, Ka aja V, Asola R e al (2006) Adju an doce axel o ino elbine wi h o wi hou as uzumab o b eas cance . N Engl J Med 354:809–820 8. Ko oula V, Kaloge as KT, Kou a seas G, Tele an ou D, K o- nenwe R, Wi z RM, Foun zilas G (2013) Sample pa ame e s a ec ing he clinical ele ance o RNA bioma ke s in ansla- ional b eas cance esea ch. Vi chows A ch 462(2):141–154 9. Wi z RM, Rau T, Laible M, Schlombs K, Lakis S, Wach e D, Sahin U, Ha mann A (2015) Low in luence o umo cell con en on mRNA exp ession le els o ERBB2, ESR1, PGR and KI67 when pe o ming he MammaType Ò RT-PCR ki . Cance Res. doi:10.1158/1538-7445.SABCS14-P5-10-13 B eas Cance Res T ea (2016) 157:437–446 445 123