ORIGINAL ARTICLE METFORMIN AND BREAST CANCER STAGE AT DIAGNOSIS, Leg et l. Metformin nd brest cncer stge t dignosis: popultion-bsed study I.C. Leg md msc,* K. Fung msc,* P.C. Austin phd, nd L.L. Lipscombe md msc* ABSTRACT Purpose The objective of the present study ws to use lrge, popultion-bsed cohort to exmine the ssocition between metformin nd brest cncer stge t dignosis while ccounting for mmmogrphy differences. Methods We used dt from Ontrio dministrtive helth dtbses to identify women 68 yers of ge or older with dibetes nd invsive brest cncer dignosed from 1 Jnury 2007 to 31 December 2012. Adjusted logistic regression models were used to compre brest cncer stge t dignosis (stges i nd ii vs. iii nd iv) between the women exposed nd not exposed to metformin. We lso exmined the ssocition between metformin use nd estrogen receptor sttus, tumour size, nd lymph node sttus in the subset of women for whom those dt were vilble. Results We identified 3125 women with dibetes nd brest cncer; 1519 (48.6%) hd been exposed to metformin before their cncer dignosis. Medin ge t brest cncer dignosis ws 76 yers (interqurtile rnge: 72 82 yers), nd men durtion of dibetes ws 8.8 ± 5.9 yers. In multivrible nlyses, metformin exposure ws not ssocited with n erlier stge of brest cncer (odds rtio: 0.98; 95% confidence intervl: 0.81 to 1.19). In secondry nlyses, metformin exposure ws not ssocited with estrogen receptor positive brest cncer, tumours lrger thn 2 cm, or positive lymph nodes. Conclusions This popultion-bsed study did not show n ssocition between metformin use nd brest cncer stge or tumour chrcteristics t dignosis. Our study considered older women with long-stnding dibetes, nd therefore further studies in younger ptients could be wrrnted. Key Words Brest cncer stge, popultion-bsed studies Curr Oncol. 2017 Apr;24(2):e85-e91 www.current-oncology.com INTRODUCTION Dibetes is ssocited with n incresed risk of brest cncer nd 40% higher risk of mortlity fter cncer dignosis 1. Insulin resistnce nd hyperinsulinemi might predispose women with type 2 dibetes to more ggressive tumours, contributing to their higher brest cncer mortlity 2. Indeed, we previously showed tht women with dibetes re more likely to present with lter-stge brest cncer, even fter djusting for prior screening mmmogrphy 3. Interventions tht reduce circulting insulin might influence cncer growth nd stge t dignosis in women with dibetes who develop brest cncer. Metformin, n insulin-sensitizing mediction used to tret dibetes, hs been ssocited with reduced tumour growth 4 7 nd improved pthologic complete response in brest cncer ptients receiving neodjuvnt chemotherpy 8. Furthermore, observtionl studies hve shown lower risk of brest 9,10 nd other cncers 11,12 nd lower brest cncer mortlity 13 in women with dibetes treted with metformin. It remins unknown whether tretment with metformin hs n effect on tumour growth nd subtype before brest cncer is dignosed cliniclly. To dte, no studies hve exmined the ssocition between metformin nd brest cncer stge. Becuse metformin is common therpy for type 2 dibetes, it is importnt to determine whether it modifies not only the risk of brest cncer but lso the stge t which the disese presents in women with dibetes. A smll study reported tht, compred with non-users of metformin, users presented with less loclly invsive brest cncer 14. Similrly, study from the U.S. Women s Helth Inititive found reduction in invsive brest cncer mong metformin users 15. Whether metformin ffects moleculr brest cncer subtype is controversil. Correspondence to: Ilin C. Leg, Women s College Reserch Institute, Women s College Hospitl, 76 Grenville Street, Toronto, Ontrio M5S 1B2. E-mil: Ilin.Leg@wchospitl.c n DOI: https://doi.org/10.3747/co.24.3380 e85
One study reported reduction in hormone-sensitive tumours with metformin exposure 15 ; others reported either n increse 16,17 or no difference in hormone-positive tumours 14. Those prior studies hve been limited by smll smple sizes, incomplete drug dt, nd inconsistent comprtor groups nd sttisticl methods. Furthermore, ll but one ccounted for screening ptterns, n importnt predictor of stge 18. Screening mmmogrphy rtes re lso influenced by comorbidities such s dibetes, which could modify the reltionship between dibetes tretment nd cncer stge 19. The objective of the present study ws to use lrge, popultion-bsed cohort to exmine the ssocition between metformin use nd brest cncer stge t dignosis while ccounting for mmmogrphy differences. We lso explored the ssocition between metformin use nd tumour size, lymph node sttus, nd hormone receptor sttus. METHODS This retrospective popultion-bsed cohort study used dt from dministrtive helth cre dtbses in Ontrio, which include records for ll individuls eligible for coverge under the province s universl helth insurnce pln. The vrious dtsets were linked using unique encoded identifiers nd were nlyzed t the Institute for Clinicl Evlutive Sciences. Dt Sources nd Popultion The study cohort ws identified from mong the popultion of Ontrio women with dibetes, 68 yers of ge or older, who were dignosed with invsive brest cncer from 1 Jnury 2007 to 31 December 2012. The cohort ws restricted to older women so s to cpture t lest 3 yers of prescription drug records before the cncer dignosis. Those records re vilble through the Ontrio Drug Benefit pln for ll individuls 65 yers of ge nd older. We used the Ontrio Cncer Registry to identify women dignosed with invsive brest cncer. The Ontrio Cncer Registry contins dt for ll Ontrio residents who hve been dignosed with or died of cncer since 1964 20. Individuls with ny prior dignosis of cncer (except non-melnom skin cncer) in the Ontrio Cncer Registry were excluded. We included only brest cncers dignosed fter 2007, becuse tht ws the yer in which stge dt becme vilble for most brest cncers in the relevnt dtbses. We then used the vlidted Ontrio Dibetes Dtbse to identify women with dibetes. The Ontrio Dibetes Dtbse hs been vlidted ginst primry cre records nd hs high sensitivity nd specificity for identifying individuls with dibetes 21. To be included in the study cohort, womn hd to hve been dignosed with dibetes t ny time before her brest cncer dignosis. We excluded ptients dignosed with dibetes before ge 30, becuse those women might hve been ffected by type 1 dibetes. The Dischrge Abstrct Dtbse mintined by the Cndin Institute for Helth Informtion nd the Ontrio Helth Insurnce Pln dtbse provided informtion bout physicin service clims. We used the Ontrio Brest Screening Progrm dtbse to look for mmmogrphy history. The Ntionl Ambultory Cre Reporting System ws used to identify ptients on dilysis nd those with chronic renl filure. We obtined informtion on demogrphics nd deths from the Registered Persons Dtbse. Records from these dministrtive helth cre dtbses were linked nonymously using encrypted helth crd numbers. Outcomes Our primry outcome ws brest cncer stge t time of dignosis bsed on Cncer Cre Ontrio s cncer stge dt. Ptients were clssified s hving stge i, ii, iii, or iv brest cncer. Collbortive stge ws determined if vilble; otherwise, we used stge dt supplied by regionl cncer centres. Our secondry outcomes were estrogen receptor sttus (positive, negtive, unknown), tumour size (<2 cm or 2 cm), nd lymph node sttus (positive, negtive, unknown). Primry Exposure Metformin use ws the min exposure of interest. For womn to be ctegorized s metformin user, she hd to fill t lest 2 consecutive scrips for metformin t ny point in the 3-yer period preceding her brest cncer dignosis. Two scrips were considered to be consecutive if the second scrip ws filled within n intervl no longer thn the durtion of the previous prescription, plus grce period of up to 50%. For the primry nlysis, metformin use ws ctegorized s binry exposure, where the comprtor group consisted of non-users of metformin. Non-users of metformin included women prescribed either nother orl glucose-lowering gent (sulfonylure, thizolidinedione, glucgon-like peptide 1, other) or insulin, or women receiving no phrmcotherpy. We lso clculted ech metformin user s cumultive exposure to metformin from ge 65 by summing the number of dys during which they were prescribed metformin (even if the dys were non-consecutive) before their brest cncer dignosis dte. We ctegorized exposure into less thn 1 yer, 1 3 yers, nd more thn 3 yers of cumultive metformin exposure. Other Covrites Income sttus ws bsed on neighborhood income quintile, derived from census dt linked to postl codes in the Registered Persons Dtbse. Rurl compred with urbn sttus ws determined by linking postl codes to census dt. Specific comorbidities were determined t bseline (derived from the Dischrge Abstrct Dtbse nd the Ntionl Ambultory Cre Reporting System). An overll comorbidity score ws estimted using the Johns Hopkins Adjusted Clinicl Group cse mix. The Adjusted Clinicl Group weighted cse-mix score hs been shown to predict mortlity in mbultory settings for ptients with dibetes in Ontrio 22. We lso recorded helth cre vribles. Number of visits to the primry cre physicin in the 2 yers preceding dignosis ws obtined from Ontrio Helth Insurnce Pln records. Receipt of mmmogrphy within 3 yers before, but no less thn 60 dys before, the brest cncer dignosis ws determined using Ontrio Helth Insurnce e86
Pln billing clims nd the Ontrio Brest Screening Progrm dtbse. We limited the period to no less thn 60 dys before the brest cncer dignosis to cpture mmmogrphy tht would most likely hve been performed for screening rther thn for dignostic purposes. Sttisticl Anlyses Bseline vribles re described using summry sttistics. Univrite nd djusted logistic regression models were used to compre brest cncer stge t dignosis for women exposed nd not exposed to metformin. The nlyses compred the likelihood of presenting with stge i or ii or with stge iii or iv disese. Covrites tht were sttisticlly significnt in the univrite models or those known to be cliniclly relevnt were included in the multivrible model. The multivrible model ws djusted for ge t brest cncer dignosis, neighbourhood income quintile, urbn residence (yes or no), number of outptient visits in the 2 yers preceding the brest cncer dignosis, the Johns Hopkins Aggregted Dignosis Groups weighted score, dibetes durtion (yers), use of other orl hypoglycemic gents (yes or no), use of insulin (yes or no), nd mmmogrphy within 3 yers of the brest cncer dignosis. Secondry Anlyses For secondry outcomes, we exmined the ssocition between metformin use nd estrogen receptor sttus (positive vs. negtive or unknown), tumour size (<2 cm vs. 2 cm), nd positive lymph node sttus in the subset of women dignosed with brest cncer during 2010 2012 for whom those dt were vilble. To exmine the effect of vrious ptterns of metformin use on the outcomes of interest, we ctegorized metformin use into current nd pst use. Current use ws ccepted if the two consecutive scrips overlpped with the 180 dys before the brest cncer dignosis. Pst use ws ccepted if 2 or more consecutive scrips for metformin were filled more thn 180 dys before the brest cncer dignosis. To isolte metformin s effect from tht of insulin or other glucose-lowering gents, we performed these dditionl subgroup nlyses: insulin users excluded; metformin monotherpy users compred with non-users of metformin; nd metformin monotherpy users compred with women tking no glucose-lowering gents. We lso conducted two sets of pre-specified subgroup nlyses. First, we strtified the cohort by dibetes durtion (<2 yers, 2 5 yers, >5 yers). Second, to ccount for totl drug exposure, we strtified the cohort by ge of dibetes dignosis (<65 yers, 65 yers), thus cpturing complete drug exposure for individuls more thn 65 yers of ge t time of dibetes dignosis. This study ws pproved by the institutionl review bord t Sunnybrook Helth Sciences Centre, Toronto, Ontrio. RESULTS The study popultion consisted of 3125 women with dibetes nd brest cncer, mong whom 1519 (48.6%) were exposed to metformin before their brest cncer dignosis. Medin ge t brest cncer dignosis ws 76 yers (interqurtile rnge: 72 82 yers), nd men durtion of dibetes before brest cncer dignosis ws 8.8 ± 5.9 yers. Tble i presents bseline vribles by exposure to metformin. Users of metformin were slightly younger t brest cncer dignosis nd hd longer durtion of dibetes. Non-users hd higher prevlence of chronic renl filure, but the two groups hd similr proportions of other comorbidities. The men durtion of metformin use ws 2.3 ± 0.9 yers. Among users of metformin, 636 (41.9%) were exposed to sulfonylures, 189 (12.4%) to thizolidinediones, 216 (14.2%) to insulin, 53 (3.5%) to DPP-4 inhibitors, nd 25 (1.7%) to crbose during the 3 yers before their cncer dignosis. In multivrible nlyses, metformin exposure before brest cncer, compred with no exposure, ws not ssocited with n erlier stge of brest cncer t dignosis [odds rtio (or): 0.98; 95% confidence intervl (ci): 0.81 to 1.19]. Similrly, no ssocition between cumultive metformin dose nd stge t dignosis ws observed (Tble ii). Secondry Outcomes For the subset of women with vilble dt bout receptor sttus, tumour size, nd lymph node sttus, we compred the likelihood of those outcomes in women using nd not using metformin. Compring users of metformin with non-users, we observed no differences in the likelihood of presenting with estrogen receptor positive brest cncer (or: 0.97; 95% ci: 0.73 to 1.29), with tumour 2 cm or lrger (or: 1.15; 95% ci: 0.92 to 1.44), or with positive lymph nodes (or: 0.95; 95% ci: 0.75 to 1.19; Tble iii). When women with metformin exposure were clssified into current nd pst users, no difference in stge t presenttion ws observed for either group compred with the non-users of metformin (current-use or: 1.06; 95% ci: 0.86 to 1.31; pst-use or: 0.82; 95% ci: 0.62 to 1.07). Similrly, when women with metformin exposure were further ctegorized into metformin monotherpy users, stge t presenttion ws not difference for those women compred with non-users of metformin or with women not on ny phrmcotherpy (Tble iv). In the subgroup nlyses, we observed no ssocitions of metformin use with brest cncer stge by ctegory of dibetes durtion or for women more thn 65 yers of ge t the time of brest cncer dignosis [ group for whom ll metformin exposure could be ccounted for (results not shown)]. DISCUSSION Our lrge popultion-bsed study filed to show n effect of metformin tretment on the stge or type of tumours in older women with dibetes who present with brest cncer. After djustment for prior mmmogrphy screening, ge, helth cre visits, comorbidity, nd income, no ssocition between metformin exposure nd brest cncer stge, hormone receptor sttus, tumour size, or lymph node sttus t dignosis ws evident. Those findings suggest tht metformin, when used to tret dibetes, does not influence brest tumour chrcteristics in the preclinicl period for older women dignosed with brest cncer. e87
TABLE I Bseline chrcteristics of women with dibetes exposed nd not exposed to metformin before brest cncer dignosis Chrcteristic Metformin exposure Overll Yes No Ptients (n) 3125 1519 1606 Age t cncer dignosis (yers) Medin 76 75 77 Interqurtile rnge 72 82 71 81 72 83 Men durtion (yers) Of dibetes 8.8±5.9 10.0±5.6 7.5±5.9 Of metformin use 2.3±0.9 2.3±0.9 Metformin durtion group [n (%)] <1 Yer 190 (12.5) 1 3 Yers 1027 (67.6) >3 Yers 302 (19.9) Exposure to other glucose-lowering medictions [n (%)] 1185 (37.9) 803 (52.9) 382 (23.8) Sulfonylure 830 (26.6) 636 (41.9) 194 (12.1) Thizolidinedione 228 (7.3) 189 (12.4) 39 (2.4) DDP4 inhibitors 59 (1.9) 53 (3.5) 6 (0.4) Insulin 407 (13.0) 216 (14.2) 191 (11.9) Comorbidities [n (%)] Stroke 58 (1.9) 33 (2.2) 25 (1.6) Myocrdil infrction 149 (4.8) 69 (4.5) 80 (5.0) Congestive hert filure 197 (6.3) 98 (6.5) 99 (6.2) Chronic renl filure 252 (8.1) 102 (6.7) 150 (9.3) Dibetic complictions 370 (11.8) 175 (11.5) 195 (12.1) Men ADG score 19.4±13.1 19.4±13.2 19.5±13.0 Screening mmmogrphy [n (%)] 1495 (47.8) 726 (47.8) 769 (47.9) Neighbourhood income quintile [n (%)] 1 716 (22.9) 375 (24.7) 341 (21.2) 2 704 (22.5) 364 (24.0) 340 (21.2) 3 579 (18.5) 264 (17.4) 315 (19.6) 4 591 (18.9) 282 (18.6) 309 (19.2) 5 524 (16.8) 229 (15.1) 295 (18.4) Defined s filling t lest 2 consecutive scrips for metformin t ny point in the 3-yer period preceding brest cncer dignosis. DDP4 = dipeptidyl peptidse-4; ADG = Johns Hopkins Aggregted Dignosis Groups. TABLE II Undjusted nd djusted logistic regression models estimting the risk of dvnced stge of brest cncer (I, II vs. III, IV) in women with nd without metformin exposure Vrible Undjusted nlysis Adjusted nlysis OR 95% CI OR 95% CI Metformin use 0.94 0.79 to 1.12 0.98 0.81 to 1.19 Cumultive metformin durtion <1 Yer 1.06 0.72 to 1.54 1.13 0.76 to 1.67 1 to 3 Yers 0.88 0.73 to 1.06 0.92 0.75 to 1.13 >3 Yers 1.13 0.83 to 1.55 1.15 0.82 to 1.61 Model djusted for ge t brest cncer dignosis, socioeconomic sttus, urbn residence (yes or no), number of outptient visits in the 2 yers preceding the brest cncer dignosis, ggregted dignosis group weighted score, dibetes durtion (yers), use of other orl hypoglycemic gents (yes or no), use of insulin (yes or no), mmmogrm within the 3 yers preceding the brest cncer dignosis. OR = odds rtio; CI = confidence intervl. e88
TABLE III Adjusted logistic regression model estimting the effect of metformin use on estrogen receptor sttus, tumour size, nd lymph node sttus Vrible Adjusted nlysis, metformin users To our knowledge, the present study is the first to exmine the ssocition between metformin nd brest cncer stge in women with dibetes. Furthermore, it is the lrgest study to dte to exmine the reltionship between metformin nd tumour chrcteristics nd subtypes. Erlier studies were smll, nd most did not ccount for dibetes durtion, comorbidities, or prior screening mmmogrphy. Furthermore, outcomes nd methods vried from study to study. In n nlysis of 253 women, metformin users hd lower proportion of loclly invsive disese; however, no difference ws reported in tumour size, hormone receptor sttus, or lymph node positivity 14. A smller study of 90 OR 95% CI Estrogen receptor positivity 0.97 0.73 to 1.29 Tumour size greter thn 2 cm 1.15 0.92 to 1.44 Lymph node positivity 0.95 0.75 to 1.19 Model djusted for ge t brest cncer dignosis, socioeconomic sttus, urbn residence (yes or no), number of outptient visits in the 2 yers preceding the brest cncer dignosis, ggregted dignosis group weighted score, dibetes durtion (yers), use of other orl hypoglycemic gents (yes or no), use of insulin (yes or no), mmmogrm within the 3 yers preceding the brest cncer dignosis. OR = odds rtio; CI = confidence intervl. TABLE IV Adjusted logistic regression models estimting the effect of metformin on brest cncer stge (I, II vs. III, IV), subgroup nlyses by metformin exposure Vrible Adjusted nlysis, metformin users OR 95% CI Any metformin use Current 1.06 0.86 to 1.31 Pst 0.82 0.62 to 1.07 Excluding insulin users b 0.96 0.78 to 1.18 Metformin monotherpy Versus no metformin 1.00 0.78 to 1.27 Versus other orl gents c 1.05 0.70 to 1.57 b c Adjusted for ge t brest cncer dignosis, socioeconomic sttus, urbn residence (yes or no), number of outptient visits in the 2 yers preceding the brest cncer dignosis, ggregted dignosis group weighted score, dibetes durtion (yers), use of other orl hypoglycemic gents (yes or no), use of insulin (yes or no), mmmogrm within the 3 yers preceding the brest cncer dignosis. Adjusted for ge t brest cncer dignosis, socioeconomic sttus, urbn residence (yes or no), number of outptient visits in the 2 yers preceding the brest cncer dignosis, ggregted dignosis group weighted score, dibetes durtion (yers), use of other orl hypoglycemic gents (yes or no), mmmogrm within the 3 yers preceding the brest cncer dignosis. Adjusted for ge t brest cncer dignosis, socioeconomic sttus, urbn residence (yes or no), number of outptient visits in the 2 yers preceding the brest cncer dignosis, ggregted dignosis group weighted score, dibetes durtion (yers), mmmogrm within the 3 yers preceding the brest cncer dignosis. women with dibetes nd brest cncer reported n incresed frequency of progesterone receptor positive brest tumours mong metformin users 16. Finlly, study from the U.S. Women s Helth Inititive found tht metformin users were less likely to present with invsive brest cncer; however, the uthors did not explore the reltionship between metformin nd brest cncer stge 15. There is evidence tht metformin hs ntitumour properties. In preclinicl studies, metformin ws found to inhibit tumour spred nd to reduce tumour growth 23,24. In short-term window-of-opportunity studies, metformin ws shown to ffect tumour growth both through indirect, insulin-dependent pthwys 6,7,25 nd by ctivting mpk pthwys to directly ffect cncer cells 26,27. However, in lrger epidemiologic studies, those ctions hve not trnslted into n improvement in tumour stge or cncer burden for either brest cncer or other cncers. A study exmining the effect of metformin exposure on colorectl cncer t presenttion found tht there ws no difference in the risk of disseminted disese between users nd non-users of metformin 28. Tht result might be due to the fct tht, in epidemiologic studies, metformin users often includes ptients on insulin, the presence of which could be negting ny effect of metformin. However, our study filed to support tht possibility: results were similr when insulin users were excluded. Furthermore, there is evidence tht metformin is ssocited with greter reduction in tumour growth in individuls who ttin the lrgest reduction in insulin level 29. Thus, metformin s effect might be lost when not ccounting for vrying levels of insulin resistnce. In previous work, we showed tht metformin use fter brest cncer dignosis is not ssocited with decresed ll-cuse mortlity 13. Although metformin use ws ssocited with lower brest cncer specific mortlity, our results did not rech sttisticl significnce. In the present study, we explored whether metformin might ffect tumour growth during the preclinicl stges of brest cncer. Although our results re null, our study popultion ws limited to older women dignosed with brest cncer fter the ge of 68. Metformin might hve different effect in women who re younger t time of their cncer nd dibetes dignoses, nd thus further studies could be wrrnted in those popultions. There re mny strengths to the present study. It is the first study to use lrge popultion-bsed cohort to determine the effect of metformin on brest cncer stge. We hd ccess to vlidted dtbses for defining brest cncer nd dibetes ptients, nd comprehensive stge dt on lrge proportion of the cohort. We lso hd universl drug dt nd were ble to identify minimum of 3 yers of drug exposure in the study popultion. Moreover, we were ble to ccount for mmmogrphy use. There re, however, limittions to the study. Clinicl dt tht is, body mss index, HbA1c, fmily history, nd so on were not vilble, nd thus we could not djust for those vribles. We hd to exclude 775 brest cncer ptients becuse of missing stge dt, which might hve ffected the generlizbility of our results. Finlly, we were limited to drug exposure tht occurred fter ge 65, nd thus the results might not be generlizble to younger ptients. e89
CONCLUSIONS This popultion-bsed study did not show n ssocition between metformin exposure nd brest cncer stge or tumour chrcteristics t dignosis. The null ssocition remined even when ccounting for long-term metformin use, s well s for differing durtions of dibetes. Those findings fil to support role for metformin in brest tumour growth in the preclinicl stge. Our study considered older women with long-stnding dibetes, nd therefore further studies in younger ptients could be wrrnted. ACKNOWLEDGMENTS This study ws conducted with the support of the Ontrio Institute for Cncer Reserch nd Cncer Cre Ontrio (cco) through funding provided by the Government of Ontrio. The study ws supported by the Institute for Clinicl Evlutive Sciences (ices), which is funded by n nnul grnt from the Ontrio Ministry of Helth nd Long-Term Cre (mohltc). The opinions, results, nd conclusions reported here re those of the uthors nd re independent from the funding sources. No endorsement by ices or the Ontrio mohltc is intended or should be inferred. Prts of this mteril re bsed on dt nd informtion compiled nd provided by the Cndin Institute for Helth Informtion (cihi). However, the nlyses, conclusions, opinions, nd sttements expressed herein re those of the uthors nd not necessrily those of cihi. Prts of this mteril re bsed on dt nd informtion provided by cco. The opinions, results, views, nd conclusions reported in this pper re those of the uthors nd do not necessrily reflect those of cco. No endorsement by cco is intended or should be inferred. We thnk ims Brogn Inc. for use of their Drug Informtion Dtbse. An overll comorbidity score ws estimted using the Johns Hopkins Adjusted Clinicl Group cse mix (softwre version 10.0.1). ICL ws supported by fellowship from the Cndin Brest Cncer Reserch Foundtion. LLL ws supported by Cndin Dibetes Assocition nd Cndin Institutes of Helth Reserch (cihr) Clinicin Scientist Awrd nd is currently supported by cihr New Investigtor Awrd. PCA is supported in prt by Creer Investigtor Awrd from the Hert nd Stroke Foundtion of Cnd (Ontrio Office). The uthors thnk Hds Fischer, ices epidemiologist, for help with the nlysis nd mnuscript preprtion. CONFLICT OF INTEREST DISCLOSURES We hve red nd understood Current Oncology s policy on disclosing conflicts of interest, nd we declre tht we hve none. AUTHOR AFFILIATIONS *Women s College Reserch Institute, Women s College Hospitl; Deprtment of Medicine, University of Toronto; Institute for Clinicl Evlutive Sciences; nd Institute of Helth, Policy, Mngement nd Evlution, University of Toronto, Toronto, ON. REFERENCES 1. Brone BB, Yeh HC, Snyder CF, et l. Long-term ll-cuse mortlity in cncer ptients with preexisting dibetes mellitus: systemtic review nd met-nlysis. JAMA 2008;300:2754 64. 2. Goodwin PJ, Ennis M, Pritchrd KI, et l. Fsting insulin nd outcome in erly-stge brest cncer: results of prospective cohort study. J Clin Oncol 2002;20:42 51. 3. Lipscombe LL, Fischer HD, Austin PC, et l. The ssocition between dibetes nd brest cncer stge t dignosis: popultionbsed study. Brest Cncer Res Tret 2015;150:613 20. 4. Dowling RJ, Zkikhni M, Fntus IG, Pollk M, Sonenberg N. Metformin inhibits mmmlin trget of rpmycin dependent trnsltion initition in brest cncer cells. Cncer Res 2007;67:10804 12. 5. Zkikhni M, Dowling R, Fntus IG, Sonenberg N, Pollk M. Metformin is n mp kinse dependent growth inhibitor for brest cncer cells. Cncer Res 2006;66:10269 73. 6. Bonnni B, Puntoni M, Czznig M, et l. Dul effect of metformin on brest cncer prolifertion in rndomized presurgicl tril. J Clin Oncol 2012;30:2593 600. 7. Nirul S, Dowling RJ, Ennis M, et l. Metformin in erly brest cncer: prospective window of opportunity neodjuvnt study. Brest Cncer Res Tret 2012;135:821 30. 8. Jirlerspong S, Pll SL, Giordno SH, et l. Metformin nd pthologic complete responses to neodjuvnt chemotherpy in dibetic ptients with brest cncer. J Clin Oncol 2009;27:3297 302. 9. Col NF, Ochs L, Springmnn V, Argki AK, Chlebowski RT. Metformin nd brest cncer risk: met-nlysis nd criticl literture review. Brest Cncer Res Tret 2012;135:639 46. 10. Yng T, Yng Y, Liu S. Assocition between metformin therpy nd brest cncer incidence nd mortlity: evidence from met-nlysis. J Brest Cncer 2015;18:264 70. 11. Decensi A, Puntoni M, Goodwin P, et l. Metformin nd cncer risk in dibetic ptients: systemtic review nd met-nlysis. Cncer Prev Res (Phil) 2010;3:1451 61. 12. Noto H, Goto A, Tsujimoto T, Nod M. Cncer risk in dibetic ptients treted with metformin: systemtic review nd met-nlysis. PloS One 2012;7:e33411. 13. Leg IC, Austin PC, Gruneir A, Goodwin PJ, Rochon PA, Lipscombe LL. Assocition between metformin therpy nd mortlity fter brest cncer: popultion-bsed study. Dibetes Cre 2013;36:3018 26. 14. Besic N, Stej N, Rtos I, et l. Long-term use of metformin nd the moleculr subtype in invsive brest crcinom ptients retrospective study of clinicl nd tumor chrcteristics. BMC Cncer 2014;14:298. 15. Chlebowski RT, McTiernn A, Wctwski-Wende J, et l. Dibetes, metformin, nd brest cncer in postmenopusl women. J Clin Oncol 2012;30:2844 52. 16. Berstein LM, Boyrkin MP, Tsyrlin EV, Turkevich EA, Semiglzov VF. More fvorble progesterone receptor phenotype of brest cncer in dibetics treted with metformin. Med Oncol 2011;28:1260 3. 17. Aksoy S, Sendur MA, Altundg K. Demogrphic nd clinicopthologicl chrcteristics in ptients with invsive brest cncer receiving metformin. Med Oncol 2013;30:590. 18. Tplin SH, Ichikw L, Yood MU, et l. Reson for lte-stge brest cncer: bsence of screening or detection, or brekdown in follow-up? J Ntl Cncer Inst 2004;96:1518 27. 19. Lipscombe LL, Hux JE, Booth GL. Reduced screening mmmogrphy mong women with dibetes. Arch Intern Med 2005;165:2090 5. 20. Brenner DR, Tmmemgi MC, Bull SB, Pinnduwje D, Andrulis IL. Using cncer registry dt: greement in cuseof-deth dt between the Ontrio Cncer Registry nd longitudinl study of brest cncer ptients. Chronic Dis Cn 2009;30:16 19. 21. Hux JE, Ivis F, Flintoft V, Bic A. Dibetes in Ontrio: determintion of prevlence nd incidence using vlidted dministrtive dt lgorithm. Dibetes Cre 2002;25:512 16. 22. Austin PC, Shh BR, Newmn A, Anderson GM. Using the Johns Hopkins Aggregted Dignosis Groups (dgs) to predict 1-yer mortlity in popultion-bsed cohorts of e90
ptients with dibetes in Ontrio, Cnd. Dibet Med 2012;29:1134 41. 23. Alimov IN, Liu B, Fn Z, et l. Metformin inhibits brest cncer cell growth, colony formtion nd induces cell cycle rrest in vitro. Cell Cycle 2009;8:909 15. 24. Zhung Y, Miskimins WK. Cell cycle rrest in metformin treted brest cncer cells involves ctivtion of mpk, downregultion of cyclin D1, nd requires p27kip1 or p21cip1. J Mol Signl 2008;3:18. 25. Dowling RJ, Nirul S, Chng MC, et l. Chnges in insulin receptor signling underlie neodjuvnt metformin dministrtion in brest cncer: prospective window of opportunity neodjuvnt study. Brest Cncer Res 2015;17:32. 26. Hdd S, Iwmoto T, Jordn L, et l. Evidence for biologicl effects of metformin in operble brest cncer: pre-opertive, window-of-opportunity, rndomized tril. Brest Cncer Res Tret 2011;128:783 94. 27. Schuler KM, Rmblly BS, DiFurio MJ, et l. Antiprolifertive nd metbolic effects of metformin in preopertive window clinicl tril for endometril cncer. Cncer Med 2015;4:161 73. 28. Spillne S, Bennett K, Shrp L, Brron TI. A cohort study of metformin exposure nd survivl in ptients with stge i iii colorectl cncer. Cncer Epidemiol Biomrkers Prev 2013;22:1364 73. 29. DeCensi A, Puntoni M, Gndini S, et l. Differentil effects of metformin on brest cncer prolifertion ccording to mrkers of insulin resistnce nd tumor subtype in rndomized presurgicl tril. Brest Cncer Res Tret 2014;148:81 90. e91