Insulin glargine use and breast cancer risk: Associations with cumulative exposure

ORIGINAL ARTICLE Insulin gla gine use and b eas cance isk: Associa ions wi h cumula i e exposu e Paul J. H. L. Pee e s a ,Ma loes T. Bazelie a ,Hube G. M. Leu kens a ,Anssi Au inen b ,Tjee d P. an S aa a,c , F ank de V ies a,d,e, and Ma ie L. De B uin a a U ech Ins i u e o Pha maceu ical Sciences, U ech Uni e si y, The Ne he lands; b School o Heal h Sciences, Uni e si y o Tampe e, Tampe e, Finland; c Heal h eResea ch Cen e, Fa Ins i u e o Heal h In o ma ics Resea ch, Uni e si y o Manches e , Manches e , UK; d Medical Cen e Maas ich and School o Public Heal h and P ima y Ca e (CAPHRI), Maas ich Uni e si y, Maas ich , The Ne he lands; e MRC Li ecou se Epidemiology Uni , Uni e si y o Sou hamp on, Sou hamp on, UK; Depa men o Clinical Pha macy and Toxicology, Maas ich Uni e si y Medical Cen e , Maas ich , The Ne he lands. ABSTRACT Backg ound: This s udy was aimed o assess he isk o b eas cance associa ed wi h exposu e o insu- lin gla gine in women wi h ype 2 diabe es and e alua e whe he he pa e n o isk concu s wi h he hypo hesized end o an inc ease in isk wi h longe du a ion o use, aking in o accoun p e ious cumula i e exposu e o o he ypes o insulin. Me hods: We pe o med a es ospec i e coho s udy (2002–2013) in he Clinical P ac ice Resea ch Da alink among adul emale pa ien s wi h a i s e e insulin p esc ip ion (n¼12 468). Time-dependen exposu e measu es we e used o assess associa ions wi h du a ion o use o : (1) o he insulin ypes be o e gla gine was i s p esc ibed (i.e. among swi che s); and (2) o gla gine du ing ollow-up. Analyses we e pe o med sepa a ely o insulin-na € ı e gla gine use s and pa ien s swi ched o gla gine. Cox p opo ional haza ds models we e used o de i e p- ends, haza d a ios (HR) and 95% con idence in e als (CI) o b eas cance associa ed wi h gla gine use. Resul s: Du ing 66 151 pe son yea s, 186 b eas cance cases occu ed; 76 in gla gine use s (3.0/1000 yea s) and 110 in use s o o he insulins (2.7/1000 yea s). Among insulin-na € ı e women, no associa ion wi h cumula i e gla gine use was obse ed (p- end ¼0.91), e en a e 5 yea s (HR ¼1.06, 95% CI 0.48–2.33). Among swi che s, a linea end wi h yea s o p io exposu e o o he insulins was ound (p- end ¼0.02). An inc eased isk was obse ed in gla gine use s wi h ex ensi e (>3 yea s) pas expos- u e o o he insulins (HR ¼3.17, 95% CI 1.28–7.84). A non-signi ican end wi h cumula i e gla gine exposu e was ound among swi che s (p- end ¼0.24). Conclusions: Exposu e o gla gine was no associa ed wi h an inc eased b eas cance isk in insulin- na € ı e pa ien s. Exposu e o o he insulins p io o he s a o gla gine appea s o be ele an when s udying b eas cance isk associa ed wi h gla gine use. ARTICLE HISTORY Recei ed 9 Oc obe 2015 Re ised 9 Feb ua y 2016 Accep ed 10 Feb ua y 2016 In women, ype 2 diabe es is associa ed wi h an inc eased isk o b eas cance [1,2]. In 2009, a numbe o obse a ional s udies eme ged ha linked he use o long-ac ing insulin gla gine o an inc eased cance incidence among women wi h ype 2 diabe es, in pa icula b eas cance [3–5]. Since hen, ea men wi h insulin gla gine has been s udied in en- si ely o i s possible associa ion wi h an inc eased b eas cance isk [6,7]. Howe e , obse a ional s udies among insu- lin use s a e complica ed by he ac ha all insulins o some ex en ac as a g ow h s imula ing ac o [8]. Human insulin ac s as a g ow h p omo ing agen , s imula - ing b eas cance cell g ow h and inhibi ing apop osis in i o [9]. Di e ences in mi ogenic po ency be ween human insulin and insulin analogs on b eas cance cells ha e been shown in i o [8,10]. Insulin gla gine speci ically appea s o ha e an inc eased mi ogenic po ency [8,10], possibly ela ed o i s inc eased a ini y o he insulin-like g ow h ac o -1 ecep- o [11]. O e all, esul s om cell s udies indica e a cell g ow h s imula ing e ec , a he han a ca cinogenic e ec [12]. Consequen ly, he isk o b eas cance is expec ed o inc ease wi h longe du a ion o exposu e. Howe e , he majo i y o epidemiological s udies con- duc ed did no assess ends in b eas cance isk wi h du - a ion o gla gine use. O he mo e de ailed second gene a ion s udies, only i e assessed cance isk in ela ion o cumula- i e exposu e measu es [13–17]. All o hem e alua ed du - a ion o gla gine use among insulin-na€ ı e use s sepa a ely and only one o he i e s udies obse ed a signi ican inc eased b eas cance isk associa ed wi h high cumula i e exposu e o gla gine [14]. Howe e , all s udies lacked su i- cien ollow-up o obus ly es ima e cance isk beyond h ee yea s o cumula i e du a ion o use [13–17]. The e o e, u he CONTACT M. L. De B uin [email p o ec ed] Depa men o Pha macoepidemiology and Clinical Pha macology, U ech Ins i u e o Pha maceu ical Sciences, U ech Uni e si y, Uni e si ei sweg 99, 3584 CG U ech , The Ne he lands Supplemen al da a o his a icle can be accessed he e. ß2016 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, 2016 VOL. 55, NO. 7, 851–858 h p://dx.doi.o g/10.3109/0284186X.2016.1155736 s udy o b eas cance isk wi h long- e m gla gine use (>3 yea s) emains necessa y. O he wo s udies ha included p e alen insulin use s who swi ched o gla gine, one obse ed an inc eased b eas cance isk a e a leas i e yea s o cumula i e gla gine use [13]. The o he did no ind an associa ion wi h du a ion o use, bu was unable o s udy e ec s o long- e m gla gine use due o a median du a ion o gla gine use o 1.2 yea s [14]. Bo h s udies de e mined exposu e o gla gine a baseline (in en ion- o- ea ) and we e limi ed by le - unca ed da a, which esul ed in misclassi ica ion o du a ion o exposu e and po en ial unde es ima ion o pas exposu e o o he insu- lins p io o he swi ch o gla gine. Mo eo e , nei he s udy was able o de e mine how he du a ion o non-gla gine insu- lin use be o e coho en y modi ies he e ec o gla gine use. The aim o ou s udy was o assess he isk o b eas can- ce associa ed wi h exposu e o insulin gla gine in women wi h ype 2 diabe es and e alua e whe he he pa e n o isk concu s wi h he hypo hesized end o an inc ease in isk wi h longe du a ion o use, aking in o accoun p e ious cumula i e exposu e o o he ypes o insulin. Me hods Sou ce o da a Da a we e ob ained om he Clinical P ac ice Resea ch Da alink (CPRD), which comp ises elec onic medical eco ds om B i ish gene al p ac i ione s since 1987 [18]. The accu - acy and comple eness o CPRD da a ha e been well alida ed in p e ious s udies [19]. Cu en ly, CPRD includes app oxi- ma ely 7% o he o al UK popula ion [18]. The p o ocol o his s udy was app o ed by CPRD’s Independen Scien i ic Ad iso y Commi ee. S udy popula ion Fo his e ospec i e coho s udy, we used a ‘new use ’ design wi h inciden insulin use s. All women (18 yea s) wi h a leas one p esc ip ion o any ype o insulin in CPRD du ing he inclusion pe iod we e eligible. To ensu e a minimal ollow-up pe iod o app oxima ely h ee yea s be ween coho en y and he end o da a collec ion (1 Oc obe 2013), he inclusion pe iod s e ched om 1 Sep embe 2002 – he ma ke ing da e o gla - gine in he UK – o 31 Decembe 2010. The index da e was de ined as he da e o he i s eco ded p esc ip ion o any ype o insulin wi hin he inclusion pe iod. On he index da e, all subjec s we e equi ed o ha e a leas one yea o up- o-s and- a d pa ien his o y in CPRD wi hou any eco ded his o y o insulin use o imp o e he alidi y o he ‘new use ’ design. Pa ien s conside ed o ha e ype 1 diabe es we e excluded. These we e pa ien s wi hou any use o non-insulin an idiabe ic d ugs (NIADs) in he yea p io o coho en y who: 1) had a eco ded diagnosis o ype 1 diabe es be o e coho en y; o 2) we e unde 30 yea s o age a coho en y. Subjec s wi h a his o y o b eas cance a baseline we e also excluded. All sub- jec s we e ollowed om he index da e un il he ou come o in e es , end o da a collec ion, da e o mig a ion ou o he CPRD popula ion, o dea h, whiche e came i s . Exposu e o insulins We used a ime-dependen design (Figu e 1) o de ine expos- u e. Fo all pa ien s, he ollow-up pe iod a e he index da e was di ided in o disc e e 30-day in e als. Exposu e o gla - gine (‘any use’) was hen de ined as a p esc ip ion o gla - gine on he s a da e o a any ime be o e he s a o each in e al. Pa ien s wi h a p esc ip ion o non-gla gine insulin (i.e. any insulin ype excep gla gine) a coho en y could become exposed du ing ollow-up i a p esc ip ion o gla - gine was eco ded (‘swi che s’). Cu en exposu e o gla gine and non-gla gine insulins was de ined as a p esc ip ion on he s a da e o in he h ee mon hs p io o he s a o each 30-day in e al. In a s epwise manne , we added ime-dependen cumula- i e measu es o du a ion o use o : (1) gla gine du ing ol- low-up; and (2) non-gla gine insulins be o e he ini ia ion o gla gine he apy. In a inal model (3), we s udied b eas can- ce isk associa ed wi h cumula i e du a ion o gla gine use sepa a ely among insulin-na€ ı e gla gine use s and use s o gla gine wi h p io use o o he insulin ypes (Figu e 1). Cumula i e du a ion o use calcula ions we e based on he numbe o days o ‘cu en exposu e’. Du a ion o use o gla gine du ing ollow-up was de e - mined a he s a o each 30-day in e al and classi ied as ‘0–1 yea s’, ‘1–3 yea s’, o ‘>3 yea s’ (Figu e 1a), based on he o al numbe o days o cu en exposu e o gla gine. Consequen ly, cumula i e exposu e o gla gine could only inc ease o emain s able o e ime. In swi che s, cumula i e numbe o days o pas exposu e o non-gla gine insulins was calcula ed a he s a da e o gla gine ea men . He e, we di e en ia ed be ween insulin- na€ ı e pa ien s a he s a o gla gine ea men , and swi ch- e s wi h a cumula i e du a ion o pas exposu e o non-gla - gine insulins o ‘0–3 yea s’ and ‘>3 yea s’ (Figu e 1b). In ou inal model, we pe o med sepa a e analyses ega d- ing associa ions wi h du a ion o gla gine exposu e du ing ol- low-up among insulin-na€ ı e pa ien s and p e alen insulin use s swi ched o gla gine (Figu e 1c). In all models, we quan- i ied he isk o b eas cance associa ed wi h gla gine use as compa ed o ‘ne e use’ o gla gine; as he s udy was pe - o med among insulin use s, pe son ime on gla gine was compa ed o pe son ime on o he insulins. S udy ou come All subjec s we e ollowed up o he occu ence o a i s medical diagnosis o b eas cance in CPRD (see Supplemen al Table A o he used medical codes, a ailable online a h p://www.in o maheal hca e.com). Comple eness o case asce ainmen o b eas cance in CPRD is high as compa ed o he na ional cance egis y da a [20]. A ecen s udy ound a conco dance a e o 89.8% wi h cance egis- ies and a subsequen 6.4% o he eco ds we e in ag ee- men wi h hospi al eco ds o dea h ce i ica es [20]. Co a ia es Models we e adjus ed o po en ial con ounde s in a ime- dependen manne . Age was de e mined as he yea 852 P. J. H. L. PEETERS ET AL. di e ence be ween calenda yea and yea o bi h a he s a o each 30-day in e al. A his o y o cance o he han b eas cance (o non-melanoma skin cance ) and oopho ec- omy was de e mined as a medical diagnosis a any ime be o e he s a o each in e al. Smoking s a us (yes o no) and alcohol use (yes o no) we e de e mined a coho en y and subsequen ly upda ed du ing ollow-up a he s a o each in e al. Cu en use o comedica ion (i.e. ho mone eplacemen he apy, s a ins, me o min, sul onylu eas, and gli azones) was de e mined as a p esc ip ion in he pas 180 days p io o he s a o each in e al. Fo body mass index (BMI) and HbA 1c , he mos ecen eco d be o e he s a o ollow-up was used o classi y pa ien s a baseline. Subsequen ly, obesi y (BMI 30 kg/m 2 ) and inc eased HbA 1c le el (>75 mmol/mol) we e de e mined based on he mos ecen measu emen a he s a o each in e al. We used s epwise model building o adjus men o po en ial con- ounde s. In Model 1, we adjus ed o all po en ial con ound- e s, while in Model 2 we pe o med addi ional adjus men o numbe o yea s o pas exposu e o non-gla gine insulins be o e he s a o gla gine as a con inuous a iable, o adjus o he po en ial e ec o pas exposu e o non-gla gine insulins on he associa ion be ween gla gine use and b eas cance isk among swi che s. We e alua ed he linea i y assump ion by adding a squa ed e m o he model, oge he wi h he con inuous a iable. S a is ical analysis Mul i a ia e Cox p opo ional haza ds models we e used o es ima e he haza d a io (HR) and 95% con idence in e als (CI) o b eas cance associa ed wi h he use o gla gine, wi h su i al ime in 30-day in e als as he ime a iable. Fo all models, ‘ne e use’ o gla gine was used as he e e ence ca - ego y. In addi ion o he analyses s a i ied by ca ego ies o cumula i e du a ion o use, we pe o med p- end analyses, whe e cumula i e exposu e o insulins was included as a con- inuous a iable. In a sensi i i y analysis, an ex ended ca ego y o cumula- i e du a ion o gla gine exposu e was added – ‘0–1 yea s’, ‘1–3 yea s’, o ‘3–5 yea s’, and ‘>5 yea s’ – when s udy powe Figu e 1. Schema ic ep esen a ion o exposu e measu es used. The basic ime-dependen design conside s a pa ien exposed om he i s p esc ip ion o gla gine onwa d. Cumula i e exposu e measu es we e included as a e inemen , whe e pa ien s a e s a i ied by (a) cumula i e exposu e o insulin gla gine du ing ollow-up, (b) du a ion o exposu e o o he insulin ypes be o e he s a o gla gine ea men , and (c) s a i ied o insulin-na € ı e s a e s o gla gine and pa ien s swi ched o gla gine. ACTA ONCOLOGICA 853 was su icien . In a sepa a e analysis, he cumula i e du a ion- esponse e ec was s udied among gla gine use s independ- en ly. He e, pa ien s s a ed on gla gine we e censo ed a he ime a di e en insulin ype was p esc ibed. In his sensi i i y analysis, he lowes ca ego y o cumula i e exposu e was aken as he e e ence. All da a managemen and s a is ical analyses (PROC PHREG) we e conduc ed using SAS 9.2 (SAS Ins i u e Inc, Ca y, NC, USA). Resul s Fo his s udy, we selec ed 12 468 emale inciden insulin use s o ou inal s udy coho (Supplemen al Figu e A, a ail- able online a h p://www.in o maheal hca e.com). Baseline cha ac e is ics o new use s o insulin who ecei ed a i s p esc ip ion o gla gine (n¼3858) o non-gla gine insulins (n¼8610) a e shown in Table 1. The median du a ion o ollow-up was compa able be ween he gla gine and non- gla gine s a e s (5.0 s. 5.1 yea s), as was he median du - a ion o insulin exposu e du ing ollow-up; 2.6 yea s o gla gine use among gla gine s a e s, and 3.0 yea s o non- gla gine insulin use among non-gla gine s a e s. Gla gine s a e s we e in gene al olde a baseline (median age o 66 s. 61 yea s). O he non-gla gine s a e s, he majo i y ecei ed a i s p esc ip ion o insulin aspa (44.6%) o human insulin (39.4%). Gla gine s a e s in gene al ecei ed NIADs in he yea p io o baseline (95.3%). Risk o b eas cance Du ing a o al ollow-up o 66 151 pe son yea s, 186 b eas cance cases occu ed. O hese, 76 occu ed in pa ien s a e exposu e o gla gine (3.0 pe 1000 pe son yea s), and 110 in pa ien s ne e exposed o gla gine (2.7 pe 1000 pe son yea s). In ou model adjus ed o po en ial con- ounde s (Model 1), no disce nible inc ease in b eas cance isk was associa ed wi h ‘e e use’ o gla gine (HR 1.06, 95% CI 0.79–1.44), as compa ed o ‘ne e use’ (Table 2). When adjus ed o yea s o exposu e o o he insulins be o e he s a o gla gine ea men (Model 2), no isk di e ence was obse ed (HR 0.98, 95% CI 0.72–1.35). Cumula i e exposu e measu es S a i ica ion by cumula i e du a ion o exposu e o gla gine du ing ollow-up (Figu e 1a,Table 2) did no show an associ- a ion wi h b eas cance isk (p- end ¼0.83 in Model 1). E en when cumula i e gla gine use o o e i e yea s was modeled as a sepa a e ca ego y (sensi i i y analysis), no signi ican di - e ence in isk was obse ed (HR 1.26, 95% CI 0.64–2.47 in Model 1). Among swi che s, s a i ica ion by p io exposu e o non- gla gine insulins a he s a o gla gine ea men (Figu e 1b, Table 2) showed a linea end wi h inc easing yea s o p io exposu e o non-gla gine insulins (p- end ¼0.02 in Model 1). He e, a signi ican h ee- old inc ease in b eas cance isk was obse ed in swi che s wi h a his o y o non-gla gine insulin use o mo e han h ee yea s, as compa ed o women ne e exposed o gla gine (HR 3.17, 95% CI 1.28–7.84). Among insulin-na€ ı e women (Figu e 1c,Table 3), no inc eased b eas cance isk was associa ed wi h ‘e e use’ o gla gine (HR 0.99, 95% CI 0.71–1.37). Also, no end wi h cumula i e du a ion o gla gine exposu e was obse ed (p- end ¼0.91, HR ¼0.99, 95% CI 0.90–1.10 pe addi ional yea o exposu e). A e addi ional s a i ica ion (sensi i i y analysis), no inc eased b eas cance isk was associa ed wi h 5 yea s o cumula i e exposu e (HR ¼1.06, 95% CI 0.48–2.33). Among swi che s, a sligh , non-signi ican ly inc eased b eas cance isk was associa ed wi h ‘e e use’ o gla gine Table 1. Baseline cha ac e is ics o inciden insulin use s s a ed on insulin gla - gine, non-gla gine insulin, o bo h. Gla gine coho Non-gla gine coho (n¼3858) (n¼8610) Follow-up (median, IQR) 5.0 (3.0–7.5) 5.1 (3.1–7.9) Maximum 11.0 11.2 Age (median, IQR) 66 (55–76) 61 (45–72) <30 43 (1.1) 165 (1.9) 30–40 183 (4.7) 1387 (16.1) 40–50 381 (9.9) 1025 (11.9) 50–60 736 (19.1) 1476 (17.1) 60–70 923 (23.9) 1894 (22.0) 70–80 937 (24.3) 1791 (20.8) >80 655 (17.0) 872 (10.1) BMI (median, IQR) 30.4 (26.0–35.3) 30.4 (25.9–35.4) <20 125 (3.2) 337 (3.9) 20–25 589 (15.3) 1326 (15.4) 25–30 1085 (28.1) 2137 (24.8) 30–35 962 (24.9) 2054 (23.9) >35 962 (24.9) 2086 (24.2) Missing 135 (3.5) 670 (7.8) HbA 1c (median, IQR) 79.1 (67.1–95.5) 78.0 (63.8–94.4) <32 mmol/mol 8 (0.2) 39 (0.5) 32–64 mmol/mol 689 (17.9) 1756 (20.4) 64–75 mmol/mol 815 (21.1) 1250 (14.5) >75 mmol/mol 2104 (54.5) 3796 (44.1) Missing 242 (6.3) 1769 (20.5) Smoking habi Non-smoke 1900 (49.2) 4187 (48.6) Cu en smoke 550 (14.3) 1398 (16.2) Ex-smoke 983 (25.5) 2051 (23.8) Missing 425 (11.0) 974 (11.3) Medical diagnosis (e e be o e) O he cance 255 (6.6) 583 (6.8) Oopho ec omy 6 (0.2) 25 (0.3) Type o insulin (a baseline) Gla gine 3858 (100.0) 0 (0.0) Non-gla gine 2888 (7.5) 8610 (100.0) Human insulin 12 (0.3) 3394 (39.4) Aspa 207 (5.4) 3841 (44.6) De emi 1 (0.0) 1186 (13.8) Glulisine 32 (0.8) 16 (0.2) Lisp o 38 (1.0) 656 (7.6) O he insulins 0 (0.0) 27 (0.3) Medica ion use (yea p io o index) NIADs None 220 (5.7) 2093 (24.3) Me o min 3014 (78.1) 5233 (60.8) Sul onylu eas 3158 (81.9) 5174 (60.1) Thiazolidinediones 1331 (34.5) 2101 (24.4) O he * 518 (13.4) 826 (9.6) HRT 286 (7.4) 594 (6.9) S a ins 2701 (70.0) 4722 (54.8) BMI: body mass index; HbA 1c : glyca ed hemoglobin; HRT: ho mone eplacemen he apy; IQR: in e qua ile ange; NIAD: non-insulin an idiabe ic d ug. *Glinides, alpha-glucosidase inhibi o s, glucogon-like pep ide 1 agonis s, dipep- idyl pep idase-4 inhibi o s, gua gum. 854 P. J. H. L. PEETERS ET AL. (HR ¼1.40, 95% CI 0.83–2.34). Mo eo e , a non-signi ican end was obse ed wi h cumula i e numbe o yea s o gla - gine exposu e (p- end ¼0.24, HR ¼1.10, 95% CI 0.94–1.29 pe addi ional yea o exposu e in Model 1). Adjus men o numbe o yea s o exposu e o non-gla gine insulins be o e he s a o gla gine ea men (Model 2) esul ed in no ice- able educ ions in isk es ima es. The HR o ‘e e use’ o gla - gine in swi che s was educed o 0.97 (95% CI 0.47–1.99), while he HR o he highes ca ego y o cumula i e gla gine exposu e was educed om 1.58 (95% CI 0.63–3.94) o 1.14 (95% CI 0.42–3.09) (Table 3). In addi ion, he non-signi ican end wi h cumula i e numbe o yea s o gla gine exposu e disappea ed a e adjus men o cumula i e numbe o yea s o exposu e o non-gla gine insulins (p- end ¼0.99, HR ¼1.00, 95% CI 0.91–1.10 pe addi ional yea o exposu e, Model 2 (no shown)). In his model, a signi ican end was obse ed wi h numbe o yea s o p io non-gla gine insulin exposu e (p- end ¼0.02, HR ¼1.24, 95% CI 1.03–1.50 pe addi ional yea o exposu e, Model 2) (no shown). Resul s om ou sensi i i y analysis ha censo ed gla gine insulin use s i any o he ype o insulin was ini ia ed, also showed no end wi h cumula i e du a ion o use (Supplemen al Table A). Discussion In his coho s udy among women wi h ype 2 diabe es newly s a ed on insulin, gla gine use was no associa ed wi h an inc eased isk o b eas cance a e a median ollow-up o Table 2. Haza d a ios o b eas cance associa ed wi h he use o insulin gla gine, s a i ied by cumula i e du a ion o gla gine use du ing ollow-up and by cumula i e exposu e o o he insulins be o e he ini ia ion o gla gine he apy. E en s Pe son yea s IR Age adj. HR (95% CI) Model 1 HR (95% CI)* Model 2 HR (95% CI) † Use o non-gla gine insulins‡110 40912.6 2.7 1 Re e ence 1 Re e ence 1 Re e ence E e use o gla gine 76 25238.5 3.0 1.08 0.81–1.45 1.06 0.79–1.44 0.98 0.72–1.35 Cumula i e gla gine use§,k <1 yea 29 8621.0 3.4 1.11 0.74–1.68 1.15 0.76–1.76 1.05 0.68–1.62 1–3 yea s 26 8994.3 2.9 1.02 0.66–1.56 0.99 0.64–1.53 0.91 0.58–1.42 >3 yea s 21 7623.1 2.8 1.12 0.69–1.81 1.05 0.64–1.70 0.99 0.61–1.62 3–5 yea s 11 4550.0 2.4 0.97 0.52–1.81 0.91 0.48–1.71 0.86 0.45–1.61 >5 yea s 10 3073.1 3.3 1.37 0.70–2.67 1.26 0.64–2.47 1.22 0.62–2.39 Cumula i e p io non-gla gine use¶,# None 59 19923.3 3.0 1.03 0.75–1.41 0.99 0.71–1.37 0–3 yea s 12 4628.8 2.6 1.08 0.59–1.96 1.14 0.62–2.07 >3 yea s** 5 686.4 7.3 3.11 1.26–7.67 3.17 1.28–7.84 adj: adjus ed; CI: con idence in e al; HR: haza d a io; IR: incidence a e in e en s/1000 pe son yea s. *Model 1, adjus ed o po en ial con ounde s; i.e. obesi y, smoking, alcohol use, HbA 1c >75 mmol/mol, his o y o oopho ec omy o o he cance ypes, and use o me o min, sul onylu eas, gli azones, ho mone eplacemen he apy, o s a ins; †Model 2, addi ional adjus men o numbe o yea s o exposu e o o he non-gla gine insulin(s) a he s a o gla gine ea men as a con inuous a iable; ‡Pe son- ime wi hou any his o y o insulin gla gine use; §Cumula i e numbe o yea s o cu en exposu e o insulin gla gine du ing ollow-up; ¶Cumula i e numbe o yea s o cu en exposu e o non-gla gine insulins be o e he s a o gla gine ea men ; **Median pas exposu e 4.4 (in e qua ile ange, 3.6–5.6 yea s; maximum, 10.3 yea s). Linea ends based on he slope o con inuous cumula i e numbe o yea s o cu en exposu e in Model 1: kp- end ¼0.83, (HR ¼1.01, 95% CI 0.92–1.10), o cumula i e cu en exposu e o gla gine; #p- end ¼0.02, (HR ¼1.24, 95% CI 1.03–1.49), o cumula i e cu en exposu e o non-gla gine insulin(s). Table 3. Haza d a ios o b eas cance associa ed wi h he use o insulin gla gine, among insulin-na € ı e gla gine use s and gla gine use s wi h p io exposu e o o he insulins, s a i ied by ca ego ies o cumula i e gla gine use and adjus ed o use o o he insulins be o e he s a o gla gine. E en s Pe son yea s IR Age adj. HR (95% CI) Model 1 HR (95% CI)* Model 2 HR (95% CI)† Use o non-gla gine insulin‡110 40912.6 2.7 1 Re e ence 1 Re e ence 1 Re e ence E e use o gla gine 76 25238.5 3.0 1.08 0.81–1.45 1.06 0.79–1.44 0.98 0.72–1.35 Insulin-na € ı e gla gine use s§59 19923.3 3.0 1.02 0.74–1.41 0.99 0.71–1.37 Cumula i e gla gine usek <1 yea 23 6759.8 3.4 1.09 0.69–1.71 1.11 0.69–1.77 1–3 yea s 20 7119.9 2.8 0.96 0.59–1.54 0.91 0.56–1.49 >3 yea s 16 6043.6 2.6 1.05 0.61–1.80 0.96 0.56–1.66 3–5 yea s 9 3610.5 2.5 0.97 0.49–1.93 0.90 0.45–1.80 >5 yea s 7 2433.0 2.9 1.18 0.54–2.57 1.06 0.48–2.33 Pa ien s swi ched o gla gine¶17 5315.2 3.2 1.33 0.80–2.22 1.40 0.83–2.34 0.97 0.47–1.99 Cumula i e gla gine use# <1 yea 6 1861.2 3.2 1.24 0.55–2.83 1.32 0.58–3.03 0.84 0.31–2.31 1–3 yea s 6 1874.4 3.2 1.31 0.57–2.98 1.36 0.59–3.10 0.89 0.33–2.37 >3 yea s 5 1579.6 3.2 1.59 0.64–3.95 1.58 0.63–3.94 1.14 0.42–3.09 adj: adjus ed; CI: con idence in e al; HR: haza d a io; IR: incidence a e in e en s/1000 pe son yea s. *Model 1, adjus ed o po en ial con ounde s; i.e. obesi y, smoking, alcohol use, HbA 1c >75 mmol/mol, his o y o oopho ec omy o o he cance ypes, and use o me o min, sul onylu eas, gli azones, ho mone eplacemen he apy, o s a ins; †Model 2, addi ional adjus men o numbe o yea s o exposu e o o he non-gla gine insulin(s) a he s a o gla gine ea men as a con inuous a iable; ‡Pe son- ime wi hou any his o y o insulin gla gine use; §Pa ien s wi hou pas cumula i e exposu e o o he insulins a he s a o gla gine he apy; ¶Pa ien s wi h pas cumula i e exposu e o o he insulins a he s a o gla gine he apy. Linea ends based on he slope o con inuous cumula i e numbe o yea s o cu en exposu e in Model 1: kp- end ¼0.91, (HR ¼0.99, 95% CI 0.90–1.10), o cumula i e yea s exposed o gla gine; #p- end ¼0.24, (HR ¼1.10, 95% CI 0.94–1.29), o cumula i e yea s exposed o gla gine. ACTA ONCOLOGICA 855 i e yea s as compa ed o use o o he insulins. Howe e , a di e ence in b eas cance isk be ween insulin-na€ ı e new use s o gla gine and women who swi ched o gla gine a e ha ing used o he ypes o insulin was obse ed. Mo e speci - ically, no associa ion be ween gla gine use (ei he in gene al o wi h cumula i e use) and b eas cance isk was seen among insulin-na€ ı e new use s o gla gine, e en a e i e cumula i e yea s o gla gine exposu e. In con as , a non-sig- ni ican inc ease in b eas cance isk was ound among pa ien s who swi ched o gla gine, depending on he numbe o yea s o pas insulin use. Tha is, a signi ican end was obse ed o each addi ional yea o non-gla gine exposu e be o e he s a o gla gine ea men . Ou esul s ega ding insulin-na€ ı e pa ien s (i.e. wi hou p io exposu e o o he insulins) a e in ag eemen wi h hose om mos obse a ional s udies ha used cumula i e expos- u e measu es. In he p e ious s udies, ou ou o i e did no show an associa ion be ween cumula i e du a ion o gla gine use and b eas cance isk among insulin-na€ ı e s a e s o gla gine [13,15–17]. The s udy by Habel e al. (2013) did epo an inc eased b eas cance isk associa ed wi h ex ended du a ion (2 yea s) o use (HR ¼1.6, 95% CI 1.0–2.8) [14]. This esul migh be a chance inding, as wi h a median du a ion o gla gine use o 1.2 yea s hey we e unable o assess pa e ns o isk wi h longe du a ion o use. In ac , all p e ious s udies among insulin-na€ ı e gla gine use s we e limi ed by insu icien s udy powe o obus ly es ima e e ec s o long-s anding (>3 yea s) gla gine exposu e. Mo eo e , ou esul s a e also in line wi h hose om clinical ials among new use s o insulin ha consis en ly showed no inc eased cance isk associa ed wi h gla gine exposu e [21–23]. The ORIGIN ial, wi h a median ollow-up o 6.2 yea s, ound no inc eased isk o b eas cance among pa ien s assigned o gla gine e sus s anda d ca e. Howe e , wi h only 28 b eas cance cases in bo h ea men a ms, s udy powe was limi ed [22]. Among pa ien s swi ched o gla gine a e ha ing used o he insulins, we obse ed a non-signi ican inc ease in isk associa ed wi h gla gine use. This esul is in line wi h ha o Suissa e al. (2011) who epo ed a signi ican isk inc ease associa ed wi h gla gine use a e i e yea s o mo e among swi che s [13]. Howe e , he 2.7- old isk inc ease (95% CI 1.1–6.5) ound in hei s udy is much la ge han he one obse ed in ou s udy. Con e sely, he only o he s udy ha included swi che s, ound no inc eased isk associa ed wi h du a ion o gla gine use, bu was limi ed by a ela i ely sho ollow-up o gla gine use s (median o 2.3 yea s) [14]. In ou s udy, we obse ed ha b eas cance isk inc eased wi h each added yea o non-gla gine insulin use be o e he s a o gla gine (i.e. e ec modi ica ion). Nei he o he p e ious s udies assessed he e ec o pas insulin exposu e on he associa ion be ween gla gine use and b eas cance isk. Suissa e al. a e hus a he only ones o acknowledge ha du a ion o insulin use be o e he s a o gla gine use should be aken in o accoun . Howe e , by ma ching on du - a ion o pas use, he amoun o a po en ial e ec was no measu ed. Ex ensi e pas use (3 yea s) o non-gla gine insulins was associa ed wi h a h ee- old inc ease in b eas cance isk among pa ien s swi ched o gla gine. Howe e , when consid- e ing la ency pe iods and he longe du a ion o ype 2 dia- be es and ea men he eo [24], i is impossible o a ibu e his excess isk among swi che s o a single ac o (i.e. gla - gine use). None heless, his esul sheds some ligh on he dynamics linked o he appa en di e ence in b eas cance isk seen in insulin-na€ ı e pa ien s s a ed on gla gine and swi che s wi h pas exposu e o o he insulins. The impo - ance o aken p io use o o he insulins in o accoun was demons a ed by he no iceable educ ion in all isk es ima es o b eas cance associa ed wi h gla gine use a e adjus - men o he numbe o yea s p e iously exposed o o he insulins. Al e na i e explana ions o he di e ence in b eas cance isk associa ed wi h gla gine use be ween insulin-na€ ı e s a - e s o gla gine and swi che s may en ail ha o al du a ion o insulin use (o diabe es du a ion), a he han exposu e o any pa icula insulin ype, is associa ed wi h an inc eased isk o b eas cance . In addi ion, pa ien s swi ched o gla gine may di e om insulin-na€ ı e pa ien s s a ed on gla gine. I gla - gine is used as an add-on in p e alen insulin use s wi h poo ly con olled blood glucose le els, a possible a ia ion in backg ound isk migh be in oduced. Such a dynamic could lead o channeling and po en ial p o opa hic bias among swi che s. This al e na i e hypo hesis migh be e alua ed in u u e s udies. Majo s eng hs o ou s udy include he use o ime- dependen exposu e measu es based on p esc ip ion da a o de e mine insulin exposu e. This de ini ion minimizes expos- u e misclassi ica ion and mo e accu a ely e lec s eal expos- u e han ime since he s a o ollow-up, as used in p e ious s udies. In addi ion, as only inciden insulin use s (1 yea wi hou any insulin use a baseline) we e included, we had comp ehensi e in o ma ion on insulin use o pa ien s included in he coho . In pa ien s who swi ched o gla gine du ing ollow-up, we we e able o de e mine he e ec o pas insulin use on he associa ion be ween gla gine use and b eas cance isk. To ou knowledge, we a e he i s o assess his e ec . Fu he mo e, we had h ee addi ional yea s o ollow-up as compa ed o he s udy by Suissa e al. and we e he eby able o de e mine b eas cance isk es ima es o long-s anding gla gine use in insulin-na€ ı e pa ien s. Se e al limi a ions o ou s udy should be no ed as well. Fi s o all, he compa a o consis ed o all o he (non- gla gine) insulins. This ca ego y included bo h sho - and long-ac ing insulins, esul ing in a he e ogeneous e e ence g oup. Howe e , i can be ega ded as a ele an e e ence g oup when you wan o assess gla gine associa ed isks e - sus he o he ea men op ions a ailable. Con e sely, his app oach did no allow o di ec compa isons be ween long- ac ing insulin ypes. Second, ou models did no accoun o any addi ional e ec s o combined use o bo h gla gine and non-gla gine insulins. Ideally, combined use should be con- side ed as a sepa a e ca ego y. Also, we we e unable o make di ec compa isons be ween gla gine and non-gla gine insulin use s wi h he same du a ion o exposu e. Thi d, we did no ake la ency in o accoun , as ollow-up ime was insu icien o inco po a e a sensible la ency pe iod o b eas cance . Fou h, as he coho was es ic ed o new use s o 856 P. J. H. L. PEETERS ET AL. insulin a coho en y, s udy powe o analyze e ec s wi h cumula i e du a ion o use was limi ed o he g oup o swi che s, hinde ing u he s a i ica ion by s a a o cumula- i e yea s o pas non-gla gine insulin use. Fi h, as CPRD co - e s a dynamic pa ien popula ion (i.e. 7% o he o al UK popula ion), pa ien s who ans e o a gene al p ac i ione who does no p o ide da a o CPRD a e los o ollow-up. Finally, when i ing se e al exposu e models on he same da a, ocus should no be on indi idual signi ican esul s, bu he o al o analyses should be seen in pe spec i e and in e - p e ed oge he [25]. In conclusion, exposu e o insulin gla gine did no appea o be associa ed wi h an inc eased b eas cance isk in insu- lin-na€ ı e pa ien s. Ou esul s, howe e , do indica e an associ- a ion be ween gla gine use and b eas cance among pa ien s p e iously ea ed wi h o he insulins be o e he s a o gla gine. Gla gine use in pa ien s wi h ex ensi e pas exposu e o o he insulin ypes was associa ed wi h a h ee- old inc eased isk. The e o e, obse a ional s udies need o ake pas exposu e o o he insulins in o accoun when s udy- ing b eas cance isk associa ed wi h gla gine use. Fu u e s udies should conside whe he his excess isk o b eas can- ce obse ed in pa ien s swi ched o gla gine is caused by p o opa hic bias. Disclosu e s a emen The esea ch leading o he esul s o his s udy has ecei ed unding om he Eu opean Communi y’s Se en h F amewo k P og am (FP-7) unde g an ag eemen numbe 282526, he CARING p ojec . The unding sou ce had no ole in s udy design, da a collec ion, da a analysis, da a in e p e a ion o w i ing o he epo . The Depa men o Pha macoepidemiology and Clinical Pha macology employing au ho s Paul J. H. L. Pee e s, Ma oes T. Bazelie and F ank de V ies ecei ed un es ic ed unding o pha macoepidemiological esea ch om GlaxoSmi hKline, No o No disk, he p i a e–public- unded Top Ins i u e Pha ma (h p://www. ipha ma.nl; includes co- unding om uni- e si ies, go e nmen and indus y), he Du ch Medicines E alua ion Boa d and he Du ch Minis y o Heal h. Hube G. M. Leu kens is employed by U ech Uni e si y as p o esso conduc ing esea ch unde he umb ella o he WHO Collabo a ing Cen e o pha maceu ical policy and egula ion. This Cen e ecei es no di ec unding o dona ions om p i a e pa ies, including pha ma indus- y. Resea ch unding om public-p i a e pa ne ships, e.g. IMI, TI Pha ma (www. ipha ma.nl) is accep ed unde he condi ion ha no company- speci ic p oduc o company ela ed s udy is conduc ed. The Cen e has ecei ed un es ic ed esea ch unding om public sou ces, e.g. Ne he lands O ganiza ion o Heal h Resea ch and De elopmen (ZonMW), he Du ch Heal h Ca e Insu ance Boa d (CVZ), EU 7 h F amewo k P og am (FP7), Du ch Medicines E alua ion Boa d (MEB), and Du ch Minis y o Heal h. Anssi Au inen is employed by he Uni e si y o Tampe e as p o esso and has no pe sonal o inancial con lic o in e es o epo . Tjee d P. an S aa has pa icipa ed in expe mee ings wi h Glaxo Smi h Kline and Boeh inge and has p o ided me hodological ad ice o Lase (all un ela ed o he subjec o his manusc ip ). Ma ie L. De B uin is employed by U ech Uni e si y as a senio esea che conduc ing esea ch in collabo a ion wi h he WHO Collabo a ing Cen e o pha maceu ical policy and egula ion. This Cen e ecei es no di ec unding o dona ions om p i a e pa ies, including he pha ma indus y. Resea ch unding om public-p i a e pa - ne ships, e.g. IMI and TI Pha ma (h p://www. ipha ma.nl) is accep ed unde he condi ion ha no company-speci ic p oduc o company ela ed s udy is conduc ed. The Cen e has ecei ed un es ic ed esea ch unding om public sou ces, e.g. Ne he lands O ganiza ion o Heal h Resea ch and De elopmen (ZonMW), he Du ch Heal h Ca e Insu ance Boa d (CVZ), EU 7 h F amewo k P og am (FP7), Du ch Medicines E alua ion Boa d (MEB) and Du ch Minis y o Heal h. None o he abo emen ioned unding sou ces had a ole in he design, conduc , analysis, o epo ing o his s udy. Re e ences 1. Tsilidis KK, Kasimis JC, Lopez DS, N zani EE, Ioannidis JP. Type 2 diabe es and cance : umb ella e iew o me a-analyses o obse a- ional s udies BMJ 2015;350:g7607. 2. La sson SC, Man zo os CS, Wolk A. Diabe es melli us and isk o b eas cance : a me a-analysis. In J Cance 2007;121:856–62. 3. Colhoun HM. SDRN Epidemiology G oup. Use o insulin gla gine and cance incidence in sco land: a s udy om he sco ish dia- be es esea ch ne wo k epidemiology g oup. Diabe ologia 2009;52: 1755–65. 4. Jonasson JM, Ljung R, Talback M, Haglund B, Gudbjo nsdo i S, S eineck G. Insulin gla gine use and sho - e m incidence o malig- nancies-a popula ion-based ollow-up s udy in sweden. Diabe ologia 2009;52:1745–54. 5. Hemkens LG, G ou en U, Bende R, Guns e C, Gu schmid S, Selke GW, e al. Risk o malignancies in pa ien s wi h diabe es ea ed wi h human insulin o insulin analogues: a coho s udy. Diabe ologia 2009;52:1732–44. 6. Ka ls ad O, S a up-Linde J, Ves e gaa d P, Hjell ik V, Bazelie MT, Schmid MK, e al. Use o insulin and insulin analogs and isk o cance - sys ema ic e iew and me a-analysis o obse a ional s ud- ies. Cu D ug Sa 2013;8:333–48. 7. B ons eld HK, Te B aak B, Ka ls ad O, Ves e gaa d P, S a up- Linde J, Bazelie MT, e al. T ea men wi h insulin (analogues) and b eas cance isk in diabe ics; a sys ema ic e iew and me a-ana- lysis o in i o, animal and human e idence. B eas Cance Res 2015;17:100. DOI: 10.1186/s13058-015-0611-2. 8. Ku zhals P, Scha e L, So ensen A, K is ensen C, Jonassen I, Schmid C, e al. Co ela ions o ecep o binding and me abolic and mi ogenic po encies o insulin analogs designed o clinical use. Diabe es 2000;49:999–1005. 9. Osbo ne CK, Bolan G, Monaco ME, Lippman ME. Ho mone espon- si e human b eas cance in long- e m issue cul u e: E ec o insu- lin. P oc Na l Acad Sci USA1976;73:4536–40. 10. S aige K, Hennige AM, S aige H, Ha ing HU, Kelle e M. Compa ison o he mi ogenic po ency o egula human insulin and i s analogue gla gine in no mal and ans o med human b eas epi helial cells. Ho m Me ab Res 2007;39:65–7. 11. Weins ein D, Simon M, Yehezkel E, La on Z, We ne H. Insulin analogues display IGF-I-like mi ogenic and an i-apop o ic ac i i ies in cul u ed cance cells. Diabe es Me ab Res Re 2009;25:41–9. 12. Home P. Insulin he apy and cance . Diabe es Ca e 2013;36Suppl: S240–4. 13. Suissa S, Azoulay L, Dell’aniello S, E ans M, Vo a J, Pollak M. Long- e m e ec s o insulin gla gine on he isk o b eas cance . Diabe ologia 2011;54:2254–62. 14. Habel LA, Dan o h KN, Quesenbe y CP, Cap a A, Van Den Eeden SK, Weiss NS, e al. Coho s udy o insulin gla gine and isk o b eas , p os a e, and colo ec al cance among pa ien s wi h dia- be es. Diabe es Ca e 2013;36:3953–60. 15. S u me T, Ma quis MA, Zhou H, Meigs JB, Lim S, Blonde L, e al. Cance incidence among hose ini ia ing insulin he apy wi h gla gine e sus human NPH insulin. Diabe es Ca e 2013;36: 3517–25. 16. Fago JP, Blo ie e PO, Rico deau P, Weill A, Alla F, Allemand H. Does insulin gla gine inc ease he isk o cance compa ed wi h o he basal insulins?: A ench na ionwide coho s udy based on na ional adminis a i e da abases. Diabe es Ca e 2013;36: 294–301. ACTA ONCOLOGICA 857 17. Lim S, S embe KG, He W, Bianca PC, Yelibi C, Ma quis A, e al. Elec onic medical eco d cance incidence o e six yea s compa - ing new use s o gla gine wi h new use s o NPH insulin. PLoS One 2014;9:e109433. 18. He e E, Gallaghe AM, Bhaska an K, Fo bes H, Ma hu R, an S aa T, e al. Da a esou ce p o ile: Clinical p ac ice esea ch da a- link (CPRD). In J Epidemiol 2015;44:827–36. 19. He e E, Thomas SL, Schoonen WM, Smee h L, Hall AJ. Valida ion and alidi y o diagnoses in he gene al p ac ice esea ch da abase: a sys ema ic e iew. B J Clin Pha macol 2010; Jan;69:4–14. 20. Boggon R, an S aa TP, Chapman M, Gallaghe AM, Hammad TA, Richa ds MA. Cance eco ding and mo ali y in he gene al p ac ice esea ch da abase and linked cance egis ies. Pha macoepidemiol D ug Sa 2013;22:168–75. 21. Home PD, Laga enne P. Combined andomised con olled ial expe ience o malignancies in s udies using insulin gla gine. Diabe ologia 2009;52:2499–506. 22. ORIGIN T ial In es iga o s, Ge s ein HC, Bosch J, Dagenais GR, Diaz R, Jung H, e al. Basal insulin and ca dio ascula and o he ou comes in dysglycemia. N Engl J Med 2012;367:319–28. 23. Rosens ock J, Fonseca V, McGill JB, Riddle M, Halle JP, H amiak I, e al. Simila isk o malignancy wi h insulin gla gine and neu al p o amine hagedo n (NPH) insulin in pa ien s wi h ype 2 diabe es: indings om a 5 yea andomised, open-label s udy. Diabe ologia 2009;52:1971–3. 24. Suissa S, Azoulay L. Me o min and he isk o cance : Time- ela ed biases in obse a ional s udies. Diabe es Ca e 2012;35: 2665–73. 25. Pa el CJ, Ioannidis JP. Placing epidemiological esul s in he con ex o mul iplici y and ypical co ela ions o exposu es. J Epidemiol Communi y Heal h 2014;68:1096–100. 858 P. J. H. L. PEETERS ET AL.