Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Diabetes and Breast Cancer Subtypes

  • Heleen K. Bronsveld,

    Affiliations Division of Molecular Pathology, The Netherlands Cancer Institute, Amsterdam, Netherlands, Division of Psychosocial Research and Epidemiology, The Netherlands Cancer Institute, Amsterdam, Netherlands

  • Vibeke Jensen,

    Affiliation Department of Pathology, Aarhus University Hospital THG, Aarhus, Denmark

  • Pernille Vahl,

    Affiliation Department of Pathology, Aarhus University Hospital THG, Aarhus, Denmark

  • Marie L. De Bruin,

    Affiliation Division of Pharmacoepidemiology & Clinical Pharmacology, Utrecht University, Utrecht, Netherlands

  • Sten Cornelissen,

    Affiliation Division of Molecular Pathology, The Netherlands Cancer Institute, Amsterdam, Netherlands

  • Joyce Sanders,

    Affiliation Division of Pathology, The Netherlands Cancer Institute, Amsterdam, Netherlands

  • Anssi Auvinen,

    Affiliation School of Public Health, University of Tampere, Tampere, Finland

  • Jari Haukka,

    Affiliation Department of Public Health, University of Helsinki, Helsinki, Finland

  • Morten Andersen,

    Affiliation Department of Medicine, Karolinska Institute, Stockholm, Sweden

  • Peter Vestergaard,

    Affiliation Clinical Institute, Aalborg University, Aalborg, Denmark

  • Marjanka K. Schmidt

    mk.schmidt@nki.nl

    Affiliations Division of Molecular Pathology, The Netherlands Cancer Institute, Amsterdam, Netherlands, Division of Psychosocial Research and Epidemiology, The Netherlands Cancer Institute, Amsterdam, Netherlands

Abstract

Background

Women with diabetes have a worse survival after breast cancer diagnosis compared to women without diabetes. This may be due to a different etiological profile, leading to the development of more aggressive breast cancer subtypes. Our aim was to investigate whether insulin and non-insulin treated women with diabetes develop specific clinicopathological breast cancer subtypes compared to women without diabetes.

Methods and Findings

This cross-sectional study included randomly selected patients with invasive breast cancer diagnosed in 2000–2010. Stratified by age at breast cancer diagnosis (≤50 and >50 years), women with diabetes were 2:1 frequency-matched on year of birth and age at breast cancer diagnosis (both in 10-year categories) to women without diabetes, to select ~300 patients with tumor tissue available. Tumor MicroArrays were stained by immunohistochemistry for estrogen and progesterone receptor (ER, PR), HER2, Ki67, CK5/6, CK14, and p63. A pathologist scored all stains and revised morphology and grade. Associations between diabetes/insulin treatment and clinicopathological subtypes were analyzed using multivariable logistic regression. Morphology and grade were not significantly different between women with diabetes (n = 211) and women without diabetes (n = 101), irrespective of menopausal status. Premenopausal women with diabetes tended to have more often PR-negative (OR = 2.44(95%CI:1.07–5.55)), HER2-negative (OR = 2.84(95%CI:1.11–7.22)), and basal-like (OR = 3.14(95%CI:1.03–9.60) tumors than the women without diabetes, with non-significantly increased frequencies of ER-negative (OR = 2.48(95%CI:0.95–6.45)) and triple negative (OR = 2.60(95%CI:0.88–7.67) tumors. After adjustment for age and BMI, the associations remained similar in size but less significant. We observed no evidence for associations of clinicopathological subtypes with diabetes in postmenopausal women, or with insulin treatment in general.

Conclusions

We found no compelling evidence that women with diabetes, treated with or without insulin, develop different breast cancer subtypes than women without diabetes. However, premenopausal women with diabetes tended to develop breast tumors that do not express hormonal receptors, which are typically associated with poor prognosis.

Introduction

Diabetes mellitus and breast cancer are chronic diseases with increasing incidence in many countries [1,2]. Recent estimates indicate that diabetes prevalence is 9.1% among women in Europe [1] and life-time risk for breast cancer is 9.7% [3]. Most patients with diabetes (~90%) have type 2 disease, characterized by reduced insulin secretion and insulin resistance with diagnosis in late adulthood, while patients with type 1 diabetes are insulin deficient [4].

Several studies have investigated whether diabetes and/or insulin (analogue) treatment increase breast cancer risk [510] or affect prognosis [1118], because of their potential impact on tumor progression through e.g. the insulin-like growth receptor pathway [5,19]. Women with diabetes have a 15–20% increased risk of breast cancer compared to women without diabetes [69], but no impact of insulin analogue treatment has been shown [5]. Breast cancer in women with diabetes is often diagnosed at an advanced stage compared to women without diabetes [13,14,2022]. Moreover, overall mortality after breast cancer diagnosis has been shown to be 30–60% higher in women with diabetes compared to women without diabetes [1116], even after adjustment for tumor stage [13,14,16]. However, studies that investigated the association between breast cancer-specific mortality and diabetes show inconsistent results [11,2327].

Diabetes itself might have a direct effect on breast cancer prognosis due to physiological effects of hyperglycemia, or hyperinsulinemia, which is a hallmark of insulin resistance commonly observed in patients with type 2 diabetes [28,29]. It has been shown that cancer-specific survival was decreased for women with abnormal glycemic status [25,27] and that fasting insulin levels are associated with worse outcome (distant recurrence and death), independent of Body Mass Index (BMI) [30]. However, diabetes itself and its complications may also increase risk of overall mortality [4] and shared cancer-promoting factors in patients with diabetes, such as obesity and a sedentary lifestyle, increases also the risk of death from competing causes (metabolic/cardiovascular diseases).

Another reason for the worse breast cancer survival may be that women with diabetes develop a more aggressive or less treatment-responsive tumor subtype. It has already been shown that hormone-related breast cancer and diabetes risk factors, such as obesity, are associated with the development of ER-negative breast cancer subtypes [31,32]. Insulin interacts with estrogens; there is experimental support that insulin may enhance estrogen production, stimulating the development of ER-positive breast cancer [19]. Furthermore, the promotion of tumor cell growth upon insulin exposure may differ by breast cancer subtype; we know from in vitro studies that mitogenic potential of insulins depends on the type of breast cancer cell line [5,33]. Although breast cancer subtypes have been extensively studied in the general population [31], few studies have assessed breast cancer subtypes in women with diabetes.

The aim of this study is to determine whether breast cancer patients with diabetes have a specific clinicopathological tumor subtype compared to those without diabetes, and whether the use of insulin is related to this.

Methods

The study protocol was approved by the Science Ethics Committee of the Region Midtjylland in Denmark. The Science Ethics Committee of the Region Midtjylland in Denmark approved that informed consent for this study was not obtained; however, all women had the possibility to opt-out from research through the nation-wide registry. Tumor tissue of the women had been collected for diagnostic or therapeutic purposes around the time of breast cancer diagnosis. This tissue is stored in biobanks and may be used for research (‘secondary use’) as long as coded and anonymous to the researcher. No tissue was used against the will of the patients (women who opt-out with regard to tissue use for future scientific purposes were excluded (http://sundhedsdatastyrelsen.dk/da/registre-og-services/vaevsanvendelsesregisteret); no risk was posed to the women as the tissue had already been removed; and tumor tissue and data were anonymous for the researcher.

Study design and patient selection

The study population consists of Caucasian women with and without diabetes, diagnosed with primary breast cancer between 2000 and 2010. The breast cancer patients were selected from a previously established nation-wide hospital-based cohort, by the Danish Breast Cancer Cooperative Group (DBCG) [34]. This cohort was linked to the National Patient Register in Denmark to identify women with and without diabetes, covering the years since 1977. In total, 43,701 women were diagnosed with incident breast cancer in 2000–2010 in the DBCG, of whom 3,047 had diabetes (7.0%). We used a cross-sectional study design with a randomly selected target population of 300 breast cancer patients. The selected women included breast cancer patients with diabetes (exposed) and without diabetes (non-exposed) sampled as follows: a random sample of women with diabetes in strata of age ≤50 and >50 years (1:1) at breast cancer diagnosis (stratification by age to increase the number of young women) frequency matched with women without diabetes from the same database (1:2) by year of birth and age at diagnosis (both in 10-year categories) (Fig 1). Twice as many women with diabetes were included as women without diabetes to allow analyses by insulin treatment. Patients with a history of other cancers, non-invasive or metastasized breast cancer, those treated with neo-adjuvant therapy, patients with diabetes diagnosed ≤1 year prior to their breast cancer diagnosis, and patients with no or insufficient tumor tissue were excluded.

thumbnail
Fig 1. Flow chart of patient identification and selection.

Stratified by age at breast cancer diagnosis (≤50 and >50 years), women with diabetes were 2:1 frequency-matched on year of birth and age at breast cancer diagnosis (both in 10-year categories) to women without diabetes, to select ~300 patients with tumor tissue available. ǂ Exact numbers <5 cannot be shown according to regulations of Statistics Denmark.

https://doi.org/10.1371/journal.pone.0170084.g001

Data collection

Age, menopausal status, year of breast cancer diagnosis and information on tumor and tumor treatment were obtained from the DBCG databank and the pathology register of the women. Only age, year of breast cancer diagnosis, and diabetes status were available at the time of patient identification. Diabetes status, diabetes type (1 or 2), and age at diabetes diagnosis, as well as data on socioeconomic status were collected by linkage with the National Patient Register (which included all medical diagnoses from 1977 onwards) in Denmark. Data on medication use, available from 1995 onwards, was obtained by linkage with the Danish Register of Medicinal Products Statistics. All linkages were done using codes which render the data anonymous to the researchers who do not have direct access to these source databases. Women were defined as oral contraceptive or hormone replacement users if at least 2 prescriptions of the drug were prescribed cumulatively in the period up to one year prior to breast cancer diagnosis. Additional information on height, weight, Body Mass Index (BMI), smoking, alcohol use, and HbA1C levels (measure of average glucose levels) prior to breast cancer diagnosis were retrieved from electronic patient files and anonymized before inclusion in the database for the researchers. Formalin-fixed, paraffin-embedded tissue samples of the primary tumors were retrieved from different Departments of Pathology in Denmark, for central pathology review and immunohistochemical (IHC) analyses.

Tumor review and IHC analyses

All formalin-fixed, paraffin-embedded tumors blocks of the primary tumor of each patient were collected and whole slides were stained with Hematoxylin and Eosin. The most representative tumor block was selected for the analyses. Hematoxylin and Eosin slides were reviewed by a breast pathologist for morphology and grade (VJ). Grade was scored following the modified Bloom-Richardson system.

For the IHC analysis, tissue microarrays with 2 cores of 2 mm per tissue block were constructed. Tissue microarrays 3μ slices were placed on superfrost+ glass slides, and stained and scored for ER, PR, HER2, Ki67, CK5/6, CK14, and p63. HER2 2+ tumors were evaluated using SISH (Silver In Situ Hybridization). Scoring of the IHC staining was performed by a breast pathologist (VJ). A 10% cut-off was used to define a positive staining for all markers, except Ki67: low if ≤14% and high if >14% according to the St Gallen guidelines of 2013 [35], and HER2: negative if 0/1+ and positive if 2+(SISH confirmed)/3+. Tumors were defined as basal-like if at least one out of three basal markers (CK14, CK5/6, P63) were positive. We also classified the tumors using the St Gallen guidelines of 2013 using ER, PR, HER2, and Ki67 [35,36].

Diabetes treatment classification

Diabetes status was determined based on medical diagnosis from the National Patient Register. Diabetes duration was defined as time from age of diabetes diagnosis till age of breast cancer diagnosis. Women with diabetes were classified as insulin users if at least 2 prescriptions of insulin were prescribed cumulatively in the period up to one year prior to breast cancer diagnosis. Exposure time was defined as time from age of start of insulin till age of breast cancer diagnosis. For women treated with other non-insulin antidiabetic drugs, the same method was used. Women with diabetes treated with insulin only were considered patients with type 1 diabetes, if they had a recorded diagnosis of type 1 diabetes (n = 21), or a medical code was missing but they were under age 30 years at diabetes diagnosis (n = 4). All other women with diabetes were considered type 2.

Imputation

For women with unknown menopausal status (n = 5), age over 52 years [37] was used as a proxy for postmenopausal status. Missing values for BMI (n = 51 in women with diabetes, n = 42 in women without diabetes) were imputed using Multivariate Imputations by Chained Equations [38] in R studio with a predictive mean matching regression model for each analyzed dataset, imputing variables with ascending number of missing values; number of imputations = 10, number of iterations = 25; see (S1 Table). We assumed that data was missing at random and could be imputed because of correlations with other variables (S2 and S3 Tables). Variables derived from the DBCG, i.e., age of breast cancer diagnosis, year of breast cancer diagnosis, menopausal status (for analyses in all women), breast cancer treatment; the electronic patient files, i.e., smoking, alcohol, height, weight, HbA1C levels; the National Patient Registry, i.e., diabetes type, diabetes duration, cardiovascular disease, microvascular disease, income, education; the Danish Register of Medicinal Products Statistics, i.e., diabetes medication, hormone replacement treatment and oral contraception use; and data on breast cancer characteristics and clinicopathological subtypes. In the subsequent analyses, we only included the variables relevant for the prediction of clinicopathological subtype, i.e. age, menopausal status, smoking, alcohol, BMI, HbA1C, diabetes duration, oral contraception use and hormone replacement treatment.

Statistical analyses

Patient and breast cancer characteristics at diagnosis were compared between breast cancer patients with and without diabetes using chi-square tests. Multivariable logistic regression models were used to estimate the association between diabetes status or insulin treatment with primary breast cancer clinicopathological subtypes. We constructed separate logistic regression models for each exposure (diabetes or insulin) to evaluate tumor subtype (various definitions) as model-specific outcomes. Multinomial logistic regression models were used for tumor subtypes which consisted of >2 categories. We tested for heterogeneity between insulin and non-insulin users in analysis restricted to diabetes patients only. In the analyses comparing women with and without diabetes, potential covariates were added in a one by one-stepwise manner; however, none of the covariates changed the beta-estimate for diabetes with >10% for any of the subtype classifications, except for BMI in the analysis of PR status and ER-/PR- in premenopausal women. Nonetheless, we are also showing adjusted models with breast cancer subtypes for age and BMI, because previous literature has shown associations between age, BMI and breast cancer subtypes [31]. Models for grade were adjusted for age only.

Modifications of the associations between diabetes status and breast cancer subtypes by menopausal status, BMI, and diabetes type were assessed using interactions terms. Although we found no statistically significant interactions between menopausal status and diabetes status (the lowest p-value was 0.07 in the analyses of PR), we show results for pre- and postmenopausal women separately based on previous evidence for different risk profiles [31]. To exclude potential bias by the inclusions of women with type 1 diabetes we performed a sensitivity analysis excluding women with type 1 diabetes. Moreover, explorative analyses were performed within women with type 1 and type 2 diabetes. SAS Enterprise guide 4.2 for Windows was used for statistical analyses.

Results

This cross-sectional study consisted of 211 women with diabetes and 101 women without diabetes, all diagnosed with breast cancer and with tumor tissue available (Fig 1). Breast cancer patients with diabetes had a similar distribution of menopausal status (as a result of the age-stratified selection), but were more often obese (BMI ≥30) (p <0.0001), compared to those without diabetes (Table 1). The majority of women with diabetes (88.2%) were diagnosed with type 2 diabetes and the mean diabetes duration was 8.9 years (S3 Table). Twenty-five percent (n = 53) of the women with diabetes were treated with insulin; including 18 combined with non-insulin antidiabetic drugs. The non-insulin users were treated with non-insulin antidiabetic drugs (35%) or diabetes was controlled by diet and exercise only (40%) (S3 Table). The mean duration of insulin use was 8.4 years (S3 Table). Insulin users (47% type 1 diabetes women) were more often premenopausal compared to non-insulin users (p = 0.04); and insulin users with premenopausal breast cancer had lower BMI compared to those not treated with insulin (p = 0.0003) (S4 Table).

thumbnail
Table 1. Characteristics of breast cancer patients with and without diabetes.

https://doi.org/10.1371/journal.pone.0170084.t001

Association between diabetes and clinicopathological breast cancer subtypes

Breast cancer patients with diabetes had a similar distribution of morphology, tumor size, and number of positive lymph nodes compared to those without diabetes (Table 1); also if stratified for menopausal status (S5 Table).

Premenopausal breast cancer patients with diabetes had more often PR-negative (OR = 2.44(95%CI:1.07–5.55), p = 0.03), HER2-negative (OR = 2.84(95%CI:1.11–7.21), p = 0.03), and basal-like (OR = 3.14(95%CI:1.03–9.60), p = 0.05) tumors than those without diabetes, with non-statically significant increased frequencies of ER-negative (OR = 2.48(95%CI:0.95–6.45)) and triple negative (OR = 2.60(95%CI:0.88–7.67) tumors (Table 2 and S6 Table). After adjustment for age and BMI, the associations remained similar in size but less statistically significant. We found no statistically significant associations between diabetes status and grade or Ki67, nor using the more refined St. Gallen subtyping (Table 2 and S6 Table). We found no modification of breast cancer subtype by BMI or diabetes type. Sensitivity analyses, in which women with type 1 diabetes were excluded, resulted in hazard ratios of the same direction and similar size (S7 Table). We did not find an association between any of the clinicopathological breast cancer subtypes and diabetes in postmenopausal women (Table 2). In analyses including all women, we only found statistically significant more basal-like tumors in women with diabetes compared to those without (OR = 2.39(95%CI:1.07–5.35), p = 0.03).

thumbnail
Table 2. Crude and adjusted odds ratios for breast cancer clinicopathological subtypes of women with diabetes compared to women without diabetes in subgroups of menopausal status using (multinomial) logistic regression.

https://doi.org/10.1371/journal.pone.0170084.t002

Association between insulin treatment and clinicopathological breast cancer subtypes

Tumor morphology, tumor size and number of positive lymph nodes did not differ between women with diabetes treated with or without insulin (S4 Table); similar results were found in analyses stratified for menopausal status (data not shown).

We observed no statistically significant evidence for the development of poor prognosis tumors among insulin users (Table 3 and S8 Table). Premenopausal women with diabetes not using insulin were more likely to develop ER-negative (OR = 3.06(95%CI:1.30–7.20), p = 0.01) and PR-negative (OR = 2.98(95%CI:1.11–8.00), p = 0.03) compared to women without diabetes, while ORs for ER and PR-negative tumors in insulin users compared to women without diabetes were only slightly increased (Table 3 and S8 Table). We performed explorative analyses separately in type 1 and type 2 insulin-treated premenopausal women with diabetes trying to understand these differences between insulin and non-insulin users. The associations between diabetes and tumor subtypes among type 1 diabetes insulin users were more in line with the findings in the non-insulin users (e.g. poor prognosis tumors), while we observed a suggestion that type 2 diabetes insulin users had better prognosis tumors (S8 and S9 Tables). However, overall, there was no evidence for a statistically significant heterogeneity between insulin and non-insulin users for any of the clinicopathological subtypes in the analyses restricted to breast cancer patients with diabetes (Table 3). In addition, adjustment for age and BMI did not materially change the effect estimates or their 95% confidence intervals (S8 and S10 Tables). In postmenopausal women, we observed no association of insulin, with breast cancer subtypes (Table 3). We did not have enough power to include subtypes using the more refined St Gallen criteria in the analyses stratified by menopausal status. In analyses including all women, we found significantly more basal-like tumors (OR = 2.5(95%CI:1.09–5.74), p = 0.03) and ER-/PR-negative tumors (OR = 1.99(95%CI:1.00–3.95), p = 0.05) in non-insulin users compared to women without diabetes.

thumbnail
Table 3. Crude and adjusted odds ratios for breast cancer clinicopathological subtypes of women with diabetes treated with or without insulin compared to women without diabetes in subgroups of menopausal status using (multinomial) logistic regression.

https://doi.org/10.1371/journal.pone.0170084.t003

Discussion

We found no compelling evidence that women with diabetes develop different clinicopathological subtypes compared to women without diabetes. However, premenopausal breast cancer patients with diabetes tend to develop breast tumors that do not express hormonal receptors and basal-like tumors, which are typically associated with poor prognosis. The majority of the women in our population had type 2 diabetes mellitus, so the results are most applicable for these patients. We also found no strong evidence that insulin treatment is associated with clinicopathological subtypes; though the poor-prognosis tumors were more often occurring in premenopausal women with diabetes not using insulin and in type 1 diabetes insulin users.

Only a few studies have investigated breast cancer characteristics among women with diabetes [20,22,39,40]. Two previous studies stratified the results for menopausal status and they also found that premenopausal women developed more often tumors that were hormone receptor negative [22,39], after multivariable adjustment [39]. Overall results were consistent with ours, showing more ER-negative, PR-negative and HER2-negative tumors in women with diabetes, with relative frequencies of 1.5 to 2.5, but most differences were not statistically significant, except for PR [20,22] and ER, even after adjustment for BMI [40]. A few studies that reported tumor markers (ER, PR, and some HER2 status) among women with diabetes [1113,26,41] compared (breast cancer) mortality or disease-free survival among women with and without diabetes as their primary objective. Therefore, only crude estimates of associations between diabetes and tumor subtype were reported and not stratified for menopausal status. Women included in these studies were mainly postmenopausal and no significant associations were found between tumor markers and diabetes status.

Studies on the association between diabetes treatment and breast cancer subtype are even more scarce. No difference in tumor stage and tumor subtype among glargine versus non-glargine users was previously described [42,43]. Studies that compared metformin users to women with diabetes treated with sulphonylurea or insulin (non-metformin) showed no difference in ER status [20,44], but sulphonylurea or insulin users presented more PR-negative tumors (63.0% versus 26.7%, p = 0.041) [44] and more HER-2 positive (29.5% versus 21%, p = 0.002) [20] than in the metformin-treated subgroup.

Our study was based on the comprehensive biobanks (archival tumor tissue from a randomly selected group of women), and databases available in Denmark, and included medication history at least five years prior to breast cancer diagnosis from prescription records, resulting in a patient selection minimally affected by survival, selection or ascertainment bias. Due to oversampling of young breast cancer patients, we could examine the association between diabetes and clinicopathological subtypes in both pre- and post-menopausal women. An experienced breast pathologist reviewed all tumor samples and we had complete data on IHC markers (including basal markers). All IHC stainings were validated and performed in one center and scored by the same breast pathologist, to prevent inter-laboratory and inter-observer variability [45,46] and to assure quality of the data. Additionally, data on risk factors such as BMI were obtained and effect estimates were adjusted for potential confounders and we followed the STROBE recommendations in reporting scientific research (S11 Table).

Our study was only sufficiently powered (around 80%; likelihood-ratio test with a two-sided p-value of 0.05) to detect large differences between breast cancer subtypes, e.g. 80% versus 60% ER-positive tumors, in women with and without diabetes and therefore, subtle differences may not have been detected. Furthermore, given the design of our study, in which odds ratios may represent on overestimation of the real risks, validation using prospective cohort analyses is recommended. Unfortunately, we had insufficient power for separate analyses of diabetes type 1 and different insulin analogues. We had also limited power to investigate the duration/dose of insulin exposure and the effect on breast cancer subtype. However, the majority of insulin users had prescriptions of insulin over several years.

We had no information on whether breast cancer patients were mammography screen-detected or not. Breast cancer subtype of screen-detected tumors differs from tumors found outside of screening [47] and there may be a higher non-participation for screening among postmenopausal women with diabetes compared to women without diabetes [48]. However, Danish national screening programs started only in 2007 for women aged 50–69 [49]. All statistically significant differences in our study were found in premenopausal women <52 years, which were mostly not screened, because the use of opportunistic screening in Denmark is low [49].

BMI, HbA1C and other risk factors such as alcohol and smoking were collected from the medical records of patients and were incomplete. However, since we had extensive data on variables associated with e.g. BMI, we were able to impute missing values using multiple imputations. Although the ratios for observed and imputed BMI were similar, BMI could still be misclassified for some patients. However, we think that misclassification of BMI is unlikely to influence our results, since BMI did not affect the association between diabetes and breast cancer subtype, except for PR status in premenopausal women. Nevertheless, we have to interpret both our positive and null results with caution.

There may be several reasons why we found a stronger and significant association between hormone receptor negative tumors and diabetes in premenopausal compared to postmenopausal women. Differences in levels of BMI-related and reproductive hormones, i.e., factors related to menopausal status, such as insulin, estrogen and adipokine, may play a role in tumor subtype formation. However, in contrast to what we have observed in postmenopausal women, a previous study showed increased estrogen levels in women with diabetes [19], which would imply that postmenopausal women would more often develop ER-positive tumors.

For the interpretation of the results, it is important to realize that diabetes and BMI are strongly associated. Women with diabetes are more likely to be obese, and premenopausal obese women tend to develop hormone receptor negative tumors [50]. Such an association between BMI and hormone receptor negative breast cancer has not been observed in postmenopausal women. Our results on the association between diabetes and breast cancer subtypes are in line with these findings, even after adjustment for BMI. The same has been reported by two other studies [39,40], which might indicate that diabetes itself contributes to higher rates of hormone receptor negative breast cancer in obese women. Our observation that poor prognosis tumors are unlikely to occur more often in premenopausal women using insulin, is in line with the earlier reports that insulin (analogues) do not increase the risk of breast cancer overall [10]. However, more research is needed for type 1 diabetes.

Diabetes medication depends on the type of diabetes, as well as the severity (insulin dependent, no endogenous insulin versus insulin resistant, high levels of endogenous insulin) and duration of diabetes. Not much is known about the mechanism, by which insulin treatment would possibly influence the receptor phenotype of breast cancer. It has been shown that insulin can induce ER and PR expression, which leads to increased binding capacity of ER in MCF-7 breast cancer cell line [51]. This may suggest that women with diabetes treated with insulin would develop more ER and PR-positive tumors, which we did not observe. Moreover, the interpretation and translation of in vivo and in vitro studies to the human setting is difficult [5].

In summary, our findings suggest that premenopausal women with diabetes tend to develop triple negative and basal tumors, which are typically associated with poor prognosis. Though our study had limited power, our results warrant further investigation and future studies should stratify their analyses by menopausal status.

Supporting Information

S1 Table. Average Body Mass Index of breast cancer patients in subgroups of menopausal status, in the ten imputed datasets (% (n)).

https://doi.org/10.1371/journal.pone.0170084.s001

(DOCX)

S2 Table. Characteristics of breast cancer patients with and without diabetes used for imputation of Body Mass Index.

https://doi.org/10.1371/journal.pone.0170084.s002

(DOCX)

S3 Table. Patient characteristics and medication use among women with type 1 and type 2 diabetes *.

https://doi.org/10.1371/journal.pone.0170084.s003

(DOCX)

S4 Table. Characteristics of breast cancer patients with diabetes treated with and without insulin.

https://doi.org/10.1371/journal.pone.0170084.s004

(DOCX)

S5 Table. Tumor characteristics of breast cancer patients with and without diabetes in subgroups of menopausal status.

https://doi.org/10.1371/journal.pone.0170084.s005

(DOCX)

S6 Table. Numbers and proportions of breast cancer clinicopathological subtypes of breast cancer patients with diabetes, type 2 diabetes and without diabetes in subgroups of menopausal status.

https://doi.org/10.1371/journal.pone.0170084.s006

(DOCX)

S7 Table. Crude and adjusted odds ratios for breast cancer clinicopathological subtypes of women with type 2 diabetes compared to women without diabetes in subgroups of menopausal status using (multinomial) logistic regression.

https://doi.org/10.1371/journal.pone.0170084.s007

(DOCX)

S8 Table. Number and proportion of breast cancer clinicopathological subtypes of breast cancer patients with diabetes (type 1 and type 2) treated with or without insulin in subgroups of menopausal status.

https://doi.org/10.1371/journal.pone.0170084.s008

(DOCX)

S9 Table. Crude odds ratios for breast cancer clinicopathological subtypes of premenopausal women with type 1 or type 2 diabetes treated with insulin compared to women without diabetes using (multinomial) logistic regression.

https://doi.org/10.1371/journal.pone.0170084.s009

(DOCX)

S10 Table. Adjusted odds ratios for breast cancer clinicopathological subtypes of women with diabetes treated with or without insulin compared to women without diabetes in subgroups of menopausal status using (multinomial) logistic regression.

https://doi.org/10.1371/journal.pone.0170084.s010

(DOCX)

Acknowledgments

We acknowledge the Departments of Pathology in Denmark for retrieving the tumor tissue and the Danish Breast Cancer Cooperative Group and Statistics Denmark for access to their databases. We acknowledge Jelle Wesseling, breast cancer pathologist at the Netherlands Cancer Institute, for his advice. We thank Jacob Starup Linde for his contribution to the data collection from the electronic medical records. We acknowledge the NKI-AVL Core Facility Molecular Pathology & Biobanking (CFMPB) for lab support.

Author Contributions

  1. Conceptualization: MKS MLB P. Vestergaard.
  2. Data curation: P. Vestergaard MLB MKS HKB AA JH MA.
  3. Formal analysis: HKB MKS.
  4. Funding acquisition: MLB MKS P. Vestergaard AA JH MA.
  5. Investigation: HKB VJ P. Vahl MLB SC JS AA JH MA P. Vestergaard MKS.
  6. Methodology: MKS MLB P. Vestergaard.
  7. Project administration: MKS P. Vestergaard HKB.
  8. Resources: VJ P. Vahl SC P. Vestergaard MKS.
  9. Supervision: MKS MLB VJ.
  10. Validation: HKB MKS.
  11. Visualization: HB.
  12. Writing – original draft: HKB MKS VJ.
  13. Writing – review & editing: HKB VJ P. Vahl MLB SC JS AA JH MA P. Vestergaard MKS.

References

  1. 1. Diabetes Atlas. Internal Diabetes Federation. 7th ed; 2015.
  2. 2. GLOBOCAN 2012 (IARC), Section of Cancer Surveillance. http://globocan.iarc.fr/Pages/fact_sheets_cancer.aspx. Accessed 28 June 2016.
  3. 3. Forouzanfar MH, Foreman KJ, Delossantos AM, Lozano R, Lopez AD, Murray CJ, et al. Breast and cervical cancer in 187 countries between 1980 and 2010: a systematic analysis. Lancet 2011;378: 1461–1484. pmid:21924486
  4. 4. Diabetes Atlas. Internal Diabetes Federation. 6th ed; 2013.
  5. 5. Bronsveld HK, ter Braak B, Karlstad O, Vestergaard P, Starup-Linde J, Bazelier MT, et al. Treatment with insulin (analogues) and breast cancer risk in diabetics; a systematic review and meta-analysis of in vitro, animal and human evidence. Breast Cancer Res 2015;17: 100. pmid:26242987
  6. 6. Starup-Linde J, Karlstad O, Eriksen SA, Vestergaard P, Bronsveld HK, de Vries F, et al. CARING (CAncer Risk and INsulin analoGues): the association of diabetes mellitus and cancer risk with focus on possible determinants—a systematic review and a meta-analysis. Curr Drug Saf 2013;8: 296–332. pmid:24215312
  7. 7. Larsson SC, Mantzoros CS, Wolk A. Diabetes mellitus and risk of breast cancer: a meta-analysis. Int J Cancer 2007;121: 856–862. pmid:17397032
  8. 8. Liao S, Li J, Wei W, Wang L, Zhang Y, Li J, et al. Association between diabetes mellitus and breast cancer risk: a meta-analysis of the literature. Asian Pac J Cancer Prev 2011;12: 1061–1065. pmid:21790252
  9. 9. Xue F, Michels KB. Diabetes, metabolic syndrome, and breast cancer: a review of the current evidence. Am J Clin Nutr 2007;86: s823–835. pmid:18265476
  10. 10. Karlstad O, Starup-Linde J, Vestergaard P, Hjellvik V, Bazelier MT, Schmidt MK, et al. Use of insulin and insulin analogs and risk of cancer—systematic review and meta-analysis of observational studies. Curr Drug Saf 2013;8: 333–348. pmid:24215311
  11. 11. Cleveland RJ, North KE, Stevens J, Teitelbaum SL, Neugut AI, Gammon MD. The association of diabetes with breast cancer incidence and mortality in the Long Island Breast Cancer Study Project. Cancer Causes Control 2012;23: 1193–1203. pmid:22674293
  12. 12. Erickson K, Patterson RE, Flatt SW, Natarajan L, Parker BA, Heath DD, et al. Clinically defined type 2 diabetes mellitus and prognosis in early-stage breast cancer. J Clin Oncol 2011;29: 54–60. pmid:21115861
  13. 13. Luo J, Virnig B, Hendryx M, Wen S, Chelebowski R, Chen C, et al. Diabetes, diabetes treatment and breast cancer prognosis. Breast Cancer Res Treat 2014;148: 153–162. pmid:25261292
  14. 14. Peairs KS, Barone BB, Snyder CF, Yeh HC, Stein KB, Derr RL, et al. Diabetes mellitus and breast cancer outcomes: a systematic review and meta-analysis. J Clin Oncol 2011;29: 40–46.
  15. 15. Redaniel MT, Jeffreys M, May MT, Ben-Shlomo Y, Martin RM. Associations of type 2 diabetes and diabetes treatment with breast cancer risk and mortality: a population-based cohort study among British women. Cancer Causes Control 2012;23: 1785–1795. pmid:22971998
  16. 16. Schrauder MG, Fasching PA, Haberle L, Lux MP, Rauh C, Hein A, et al. Diabetes and prognosis in a breast cancer cohort. J Cancer Res Clin Oncol 2011;137: 975–983. pmid:21132511
  17. 17. Bowker SL, Majumdar SR, Veugelers P, Johnson JA. Increased cancer-related mortality for patients with type 2 diabetes who use sulfonylureas or insulin. Diabetes Care 2006;29: 254–258. pmid:16443869
  18. 18. Currie CJ, Poole CD, Evans M, Peters JR, Morgan CL. Mortality and other important diabetes-related outcomes with insulin vs other antihyperglycemic therapies in type 2 diabetes. J Clin Endocrinol Metab 2013;98: 668–677. pmid:23372169
  19. 19. Rose DP, Vona-Davis L. The cellular and molecular mechanisms by which insulin influences breast cancer risk and progression. Endocr Relat Cancer 2012;19: R225–241. pmid:22936542
  20. 20. Hou G, Zhang S, Zhang X, Wang P, Hao X, Zhang J. Clinical pathological characteristics and prognostic analysis of 1,013 breast cancer patients with diabetes. Breast Cancer Res Treat 2013;137: 807–816. pmid:23292119
  21. 21. Renehan AG, Yeh HC, Johnson JA, Wild SH, Gale EAM, Moller H, et al. Diabetes and cancer (2): evaluating the impact of diabetes on mortality in patients with cancer. Diabetologia 2012;55: 1619–1632. pmid:22476948
  22. 22. Liao S, Li J, Wang L, Zhang Y, Wang C, Hu M, et al. Type 2 diabetes mellitus and characteristics of breast cancer in China. Asian Pac J Cancer Prev 2010;11: 933–937. pmid:21133604
  23. 23. Fleming ST, Rastogi A, Dmitrienko A, Johnson KD. A comprehensive prognostic index to predict survival based on multiple comorbidities: a focus on breast cancer. Med Care 1999;37: 601–614. pmid:10386572
  24. 24. Liu X, Ji J, Sundquist K, Sundquist J, Hemminki K. The impact of type 2 diabetes mellitus on cancer-specific survival: a follow-up study in Sweden. Cancer 2012;118: 1353–1361. pmid:21800292
  25. 25. Seshasai SR, Kaptoge S, Thompson A, Di Angelantonio E, Gao P, Sarwar N, et al. Diabetes mellitus, fasting glucose, and risk of cause-specific death. N Engl J Med 2011;364: 829–841. pmid:21366474
  26. 26. Srokowski TP, Fang S, Hortobagyi GN, Giordano SH. Impact of diabetes mellitus on complications and outcomes of adjuvant chemotherapy in older patients with breast cancer. J Clin Oncol 2009;27: 2170–2176. pmid:19307509
  27. 27. Zhou XH, Qiao Q, Zethelius B, Pyorala K, Soderberg S, Pajak A, et al. Diabetes, prediabetes and cancer mortality. Diabetologia 2010;53: 1867–1876. pmid:20490448
  28. 28. Westley RL, May FE. A twenty-first century cancer epidemic caused by obesity: the involvement of insulin, diabetes, and insulin-like growth factors. Int J Endocrinol 2013;2013: 632461. pmid:23983688
  29. 29. Flores-Lopez LA, Martinez-Hernandez MG, Viedma-Rodriguez R, Diaz-Flores M, Baiza-Gutman LA. High glucose and insulin enhance uPA expression, ROS formation and invasiveness in breast cancer-derived cells. Cell Oncol (Dordr) 2016.
  30. 30. Goodwin PJ, Ennis M, Pritchard KI, Trudeau ME, Koo J, Madarnas Y, et al. Fasting insulin and outcome in early-stage breast cancer: results of a prospective cohort study. J Clin Oncol 2002;20: 42–51. pmid:11773152
  31. 31. Yang XR, Chang-Claude J, Goode EL, Couch FJ, Nevanlinna H, Milne RL, et al. Associations of breast cancer risk factors with tumor subtypes: a pooled analysis from the Breast Cancer Association Consortium studies. J Natl Cancer Inst 2011;103: 250–263. pmid:21191117
  32. 32. Phipps AI, Buist DS, Malone KE, Barlow WE, Porter PL, Kerlikowske K, et al. Breast density, body mass index, and risk of tumor marker-defined subtypes of breast cancer. Ann Epidemiol 2012;22: 340–348. pmid:22366170
  33. 33. Mauro L, Salerno M, Panno ML, Bellizzi D, Sisci D, Miglietta A, et al. Estradiol increases IRS-1 gene expression and insulin signaling in breast cancer cells. Biochem Biophys Res Commun 2001;288: 685–689. pmid:11676497
  34. 34. Moller S, Jensen MB, Ejlertsen B, Bjerre KD, Larsen M, Hansen HB, et al. The clinical database and the treatment guidelines of the Danish Breast Cancer Cooperative Group (DBCG); its 30-years experience and future promise. Acta Oncol 2008;47: 506–524. pmid:18465317
  35. 35. Goldhirsch A, Winer EP, Coates AS, Gelber RD, Piccart-Gebhart M, Thurlimann B, et al. Personalizing the treatment of women with early breast cancer: highlights of the St Gallen International Expert Consensus on the Primary Therapy of Early Breast Cancer 2013. Ann Oncol 2013;24: 2206–2223. pmid:23917950
  36. 36. Sorlie T, Perou CM, Tibshirani R, Aas T, Geisler S, Johnsen H, et al. Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications. Proc Natl Acad Sci U S A 2001;98: 10869–10874. pmid:11553815
  37. 37. Brand JS, Onland-Moret NC, Eijkemans MJ, Tjonneland A, Roswall N, Overvad K, et al. Diabetes and onset of natural menopause: results from the European Prospective Investigation into Cancer and Nutrition. Hum Reprod 2015;30: 1491–1498. pmid:25779698
  38. 38. van Buuren S, Groothuis-Oudshoorn K. mice: Multivariate Imputation by Chained Equations in R. 2011 2011;45: 67.
  39. 39. Gillespie EF, Sorbero ME, Hanauer DA, Sabel MS, Herrmann EJ, Weiser LJ, et al. Obesity and angiolymphatic invasion in primary breast cancer. Ann Surg Oncol 2010;17: 752–759. pmid:19898898
  40. 40. Wolf I, Sadetzki S, Gluck I, Oberman B, Ben-David M, Papa MZ, et al. Association between diabetes mellitus and adverse characteristics of breast cancer at presentation. Eur J Cancer 2006;42: 1077–1082. pmid:16574404
  41. 41. Kaplan MA, Pekkolay Z, Kucukoner M, Inal A, Urakci Z, Ertugrul H, et al. Type 2 diabetes mellitus and prognosis in early stage breast cancer women. Med Oncol 2012;29: 1576–1580. pmid:22083554
  42. 42. Besic N, Satej N. Insulin glargine versus other types of basal insulin-clinical and tumor characteristics in patients with breast carcinoma. BMC Res Notes 2013;6: 416. pmid:24131755
  43. 43. Grimaldi-Bensouda L, Cameron D, Marty M, Barnett AH, Penault-Llorca F, Pollak M, et al. Risk of breast cancer by individual insulin use: an international multicenter study. Diabetes Care 2014;37: 134–143. pmid:23949559
  44. 44. Berstein LM, Boyarkina MP, Tsyrlina EV, Turkevich EA, Semiglazov VF. More favorable progesterone receptor phenotype of breast cancer in diabetics treated with metformin. Med Oncol 2011;28: 1260–1263. pmid:20490716
  45. 45. Mengel M, von Wasielewski R, Wiese B, Rudiger T, Muller-Hermelink HK, Kreipe H. Inter-laboratory and inter-observer reproducibility of immunohistochemical assessment of the Ki-67 labelling index in a large multi-centre trial. J Pathol 2002;198: 292–299. pmid:12375261
  46. 46. O'Leary TJ. Standardization in immunohistochemistry. Appl Immunohistochem Mol Morphol 2001;9: 3–8. pmid:11277412
  47. 47. Sihto H, Lundin J, Lehtimaki T, Sarlomo-Rikala M, Butzow R, Holli K, et al. Molecular subtypes of breast cancers detected in mammography screening and outside of screening. Clin Cancer Res 2008;14: 4103–4110. pmid:18593987
  48. 48. Hellmann SS, Njor SH, Lynge E, von Euler-Chelpin M, Olsen A, Tjonneland A, et al. Body mass index and participation in organized mammographic screening: a prospective cohort study. BMC Cancer 2015;15: 294. pmid:25880028
  49. 49. Njor SH, Olsen AH, Blichert-Toft M, Schwartz W, Vejborg I, Lynge E. Overdiagnosis in screening mammography in Denmark: population based cohort study. Bmj 2013;346: f1064. pmid:23444414
  50. 50. Yeh HC, Platz EA, Wang NY, Visvanathan K, Helzlsouer KJ, Brancati FL. A prospective study of the associations between treated diabetes and cancer outcomes. Diabetes Care 2012;35: 113–118. pmid:22100961
  51. 51. Panno ML, Salerno M, Pezzi V, Sisci D, Maggiolini M, Mauro L, et al. Effect of oestradiol and insulin on the proliferative pattern and on oestrogen and progesterone receptor contents in MCF-7 cells. J Cancer Res Clin Oncol 1996;122: 745–749. pmid:8954172