SEARCH

SEARCH BY CITATION

Keywords:

  • adjuvant hormonal therapy;
  • breast cancer;
  • diabetes;
  • estrogen deficiency;
  • tamoxifen

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. FUNDING SOURCES
  9. REFERENCES

BACKGROUND:

There is increasing evidence linking breast cancer and diabetes; however, few studies have explored the association between cancer treatments and risk of diabetes. Tamoxifen may increase diabetes incidence through its estrogen-inhibiting effects. This study assessed whether tamoxifen treatment in older breast cancer survivors is associated with an increased risk of diabetes.

METHODS:

This nested case-control study used population-based health databases in Ontario, Canada to identify women older than 65 years with early stage breast cancer between April 1, 1996 and March 31, 2006. Cases were defined as cohort members diagnosed with diabetes during follow-up (March 31, 2008), and each case was age-matched with up to 5 controls who did not develop diabetes. After adjusting for other risk factors, the authors compared the likelihood of diabetes between current tamoxifen users and tamoxifen nonusers, based on prescriptions at diabetes diagnosis. They also compared diabetes risk in current aromatase inhibitor users versus nonusers.

RESULTS:

Of 14,360 breast cancer survivors identified, mean age 74.9 years, 1445 (10%) developed diabetes over a mean follow-up of 5.2 years. Current tamoxifen therapy was associated with a significantly higher risk of diabetes compared with no tamoxifen therapy (adjusted odds ratio, 1.24; 95% confidence interval, 1.08-1.42; P = .002). There was no association between aromatase inhibitor therapy and diabetes.

CONCLUSIONS:

Current tamoxifen therapy is associated with an increased incidence of diabetes in older breast cancer survivors. These findings suggest that tamoxifen treatment may exacerbate an underlying risk of diabetes in susceptible women; further studies are needed to better explore this association. Cancer 2011. © 2011 American Cancer Society.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. FUNDING SOURCES
  9. REFERENCES

Diabetes is associated with a higher incidence of breast cancer1, 2 and worse breast cancer survival,3-5 and women with breast cancer may also have a higher risk of future diabetes.6, 7 Survival after breast cancer is increasing, such that 1% of women are now long-term breast cancer survivors.8 As survivorship continues to increase, long-term health consequences for this population are becoming more important. Well-recognized health issues for breast cancer survivors include the risk of cancer recurrence or second primary tumors, local complications related to surgery or radiation therapy, and side effects of adjuvant treatment.9 Despite an emerging link between diabetes and breast cancer, few studies have explored whether treatments used for breast cancer are associated with a higher risk of diabetes.

Tamoxifen, a selective estrogen receptor modulator, is the most commonly prescribed hormonal treatment in women with breast cancer.10 Women with estrogen receptor-positive breast cancer are generally prescribed tamoxifen as adjuvant therapy for a 5-year period.10 Side effects of tamoxifen include menopausal symptoms, thromboembolism, and endometrial cancer.9 Tamoxifen may also increase the risk of diabetes through its estrogen-inhibiting effects. A role for estrogen in glucose homeostasis is becoming increasingly recognized.11-15 Estrogen receptors are found on insulin-producing pancreatic beta cells, and estrogen inhibition is associated with increased insulin resistance16 and decreased insulin secretion.17 Estrogen replacement was also found to lower the risk of diabetes in clinical trials.18, 19 Furthermore, tamoxifen has been shown to directly suppress insulin secretion in mice.17 Finally, tamoxifen has been shown to increase weight gain, which may also promote diabetes.20 This effect may be of particular concern for women with breast cancer, who may already have a higher risk of diabetes.6, 7 To our knowledge, no studies have assessed whether tamoxifen may contribute to a higher risk of diabetes in breast cancer survivors.

The objective of this population-based study was to assess whether older breast cancer survivors prescribed tamoxifen therapy had a higher incidence of diabetes compared with those who were not prescribed tamoxifen therapy.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. FUNDING SOURCES
  9. REFERENCES

Data Sources

This study used linked population-based health databases from Ontario, Canada. We identified prescription records using the Ontario Drug Benefit database, which includes comprehensive data on publicly funded medications dispensed to Ontarians aged 65 years and older.21 Women with breast cancer were identified using the Ontario Cancer Registry, which contains data on all Ontario residents who have been newly diagnosed with cancer or who have died of cancer.22 The Canadian Institute for Health Information Discharge Abstract database was used to identify hospital admissions, and the Ontario Health Insurance Plan database provided information on physician service claims. We obtained information on demographics and deaths from the Registered Persons Database and diabetes status from the Ontario Diabetes Database, a previously validated registry of Ontarians with diabetes.23 These administrative health care databases were linked anonymously using encrypted health card numbers.

Study Population and Design

We conducted a population-based case-control study nested within a cohort of women aged 66 years and older who were newly diagnosed with invasive breast cancer as recorded in the Ontario Cancer Registry between April 1, 1996 and March 31, 2006. We limited our cohort to women with early stage breast cancer to avoid differential survival rates that may affect the opportunity to develop diabetes. Because data on cancer stage were not available from our databases, we defined early stage breast cancer as patients who had breast cancer surgery in the 4 months before or 12 months after the cancer diagnosis, who did not have evidence of metastases at diagnosis, and who were alive 1 year after the breast cancer diagnosis date. Because we wanted to assess the association between tamoxifen exposure and newly diagnosed diabetes, we excluded patients who had a history of diabetes before the breast cancer diagnosis based on the Ontario Diabetes Database. We also excluded patients who had ductal carcinoma in situ, and those with a previous nonskin cancer malignancy. Patients in the cohort were observed until the first occurrence of an outcome (see below), a breast cancer recurrence (defined as those receiving breast cancer surgery or chemotherapy >1 year after cohort entry), a new malignancy, death, or the end of the follow-up period (March 31, 2008).

Cases and Controls

Within the cohort of breast cancer survivors, we identified as cases all those who were newly diagnosed with diabetes during the follow-up period, based on entry into the Ontario Diabetes Database. The date of diabetes diagnosis served as the case index (event) date.

We then selected up to 5 control patients for each case from the cohort of women with breast cancer who were still at risk for an event at the case's index date (within ±30 days). Eligible controls could not become a case for at least 30 days after the index/event date of the case. Controls were matched to cases based on age (±1 year) and on fiscal year of cohort entry date (ie, date of breast cancer diagnosis). Controls were assigned the same index date as that of their respective case. Patients could serve as a control more than once and were eligible to become a case at a later time after selection as a control.

Exposure to Tamoxifen

The primary exposure of interest was tamoxifen treatment, defined as 2 consecutive prescriptions for tamoxifen before the index/event date. Prescriptions were defined as consecutive if the interval between 2 prescription refills did not exceed the days supplied by the first prescription by >20%.

We categorized exposure status into 1 of 4 mutually exclusive groups based on the date of the most recent tamoxifen prescription dispensed before the index date and its number of days supplied. Timing of use relative to the index date was classified as current, recent past, and remote use. Tamoxifen exposure was classified as current use if the last prescription's days supplied (+14 days) overlapped the index date. The exposure was classified as recent past exposure if the end of the last prescription was >14 days but ≤180 days before the index date. Tamoxifen use was classified as remote treatment if the last prescription ended >180 days but <5 years before the index date. No use of tamoxifen was the referent exposure, and it included patients who had no record of tamoxifen treatment as well as those who did not have 2 consecutive tamoxifen prescriptions or who had tamoxifen exposure >5 years before the index date.

Statistical Analysis

We used conditional logistic regression to estimate the unadjusted and adjusted odds ratios (ORs) and 95% confidence intervals (CIs) for the association between newly diagnosed diabetes and tamoxifen exposure. Because estimates were similar for the current and recent past (stopped drug within 180 days) tamoxifen users, we combined them into 1 category in our analyses. The primary analysis compared diabetes risk between current and recent past tamoxifen use and no tamoxifen use. We also compared diabetes risk between remote users and nonusers of tamoxifen. For all analyses, we adjusted for the following potential confounders: income quintiles, Charlson comorbidity index score in the year before the cohort entry date,24 rural residence, history of cardiovascular event, renal failure, stroke or venous thromboembolism in the 5 years before cohort entry, adjuvant chemotherapy, and other prescription drugs in the 1 year before the index date, including previous use of aromatase inhibitors (AIs; see Tables 1 and 2).

Table 1. Baseline Characteristics of Cases (Breast Cancer Patients With Diabetes) and Controls (Matched Breast Cancer Patients Without Diabetes)
CharacteristicsCases, n = 1445Controls, n = 7220
  • Abbreviation: SD, standard deviation.

  • a

    Income quintiles are derived by linking postal codes with Canadian census data, which provide the median household income level of neighborhood of residence.

  • b

    Cardiovascular diseases include hospitalization for acute myocardial infarction, congestive heart failure, arrhythmia, peripheral vascular disease, acute rheumatic heart disease, hypertension, myocarditis, and other.

  • c

    Diabetes-promoting drugs include thiazide diuretics, beta blockers, angiotensin-converting enzyme inhibitors, statins, angiotensin receptor blockers, and antipsychotic agents.

Demographics at cohort entry  
 Age at diagnosis date, mean ± SD73.80 ± 5.8073.78 ± 5.77
  >80 years196 (13.6%)951 (13.2%)
   65-80 years1249 (86.4%)6269 (86.8%)
 Income quintilea  
   1 [lowest]326 (22.6%)1337 (18.5%)
   2332 (23.0%)1524 (21.1%)
   3288 (19.9%)1495 (20.7%)
   4225 (15.6%)1348 (18.7%)
   5 [highest]270 (18.7%)1488 (20.6%)
 Rural residence  
  Yes214 (14.8%)1014 (14.0%)
  No1230 (85.1%)6200 (85.9%)
Comorbidity 5-years before cohort entry  
 Charlson comorbidity score  
   0 or missing1347 (93.2%)6823 (94.5%)
   182 (5.7%)318 (4.4%)
   2+16 (1.1%)79 (1.1%)
 Cardiovascular diseaseb292 (20.2%)1112 (15.4%)
 End-stage renal disease14 (1.0%)58 (0.8%)
 Stroke27 (1.9%)116 (1.6%)
 Venous thrombosis16 (1.1%)54 (0.7%)
Prescription drug use 1 year before event  
 Aromatase inhibitor therapy141 (9.8%)760 (10.5%)
 At least 1 diabetes-promoting drugc1170 (81.0%)4579 (63.4%)
 Oral glucocorticoid therapy111 (7.7%)376 (5.2%)
 Estrogen therapy31 (2.1%)168 (2.3%)
 Progesterone therapy19 (1.3%)95 (1.3%)
 Raloxifene therapy8 (0.6%)17 (0.2%)
 Adjuvant chemotherapy110 (7.6%)525 (7.3%)
Table 2. Baseline Characteristics by Tamoxifen Exposure for Cases and Their Matched Controls
CharacteristicsCurrent/Past UsersRemote UsersNon Users
Cases, n = 531Controls, n = 2399Cases, n = 202Controls, n = 1046Cases, n = 712Controls, n = 3775
  • Abbreviation: SD, standard deviation.

  • a

    Income quintiles are derived by linking postal codes with Canadian census data, which provide the median household income level of neighborhood of residence.

  • b

    Cardiovascular diseases include hospitalization for acute myocardial infarction, congestive heart failure, arrhythmia, peripheral vascular disease, acute rheumatic heart disease, hypertension, myocarditis, and other.

  • c

    Diabetes-promoting drugs include thiazide diuretics, beta blockers, angiotensin-converting enzyme inhibitors, statins, angiotensin receptor blockers, and antipsychotic agents.

Demographics at cohort entry      
 Age at diagnosis date, mean ±SD74.07 ± 5.7574.13 ± 5.9173.20 ± 5.6673.52 ± 5.4173.76 ± 5.8873.62 ± 5.77
  >80 years73 (13.7%)360 (15.0%)23 (11.4%)114 (10.9%)100 (14.0%)477 (12.6%)
  65-80 years458 (86.3%)2039 (85.0%)179 (88.6%)932 (89.1%)612 (86.0%)3298 (87.4%)
 Income quintilesa      
  1127 (23.9%)454 (18.9%)44 (21.8%)171 (16.3%)155 (21.8%)712 (18.9%)
  2114 (21.5%)512 (21.3%)48 (23.8%)239 (22.8%)170 (23.9%)773 (20.5%)
  3106 (20.0%)490 (20.4%)35 (17.3%)207 (19.8%)147 (20.6%)798 (21.1%)
  485 (16.0%)417 (17.4%)31 (15.3%)199 (19.0%)109 (15.3%)732 (19.4%)
  597 (18.3%)515 (21.5%)44 (21.8%)227 (21.7%)129 (18.1%)746 (19.8%)
 Rural residence      
  Yes76 (14.3%)331 (13.8%)31 (15.3%)122 (11.7%)107 (15.0%)561 (14.9%)
  No455 (85.7%)2068 (86.2%)171 (84.7%)924 (88.3%)605 (85.0%)3214 (85.1%)
Comorbidity [5-year precohort entry]      
 Charlson score      
   0 or missing493 (92.8%)2277 (94.9%)192 (95%)989 (94.6%)662 (93.0%)3557 (94.2%)
   133 (6.2%)93 (3.9%)9 (4.5%)41 (3.9%)40 (5.6%)184 (4.9%)
   2+≤5 (0.9%)29 (1.2%)≤5 (0.5%)16 (1.5%)10 (1.4%)34 (0.9%)
 Cardiovascular diseaseb107 (20.2%)365 (15.2%)35 (17.3%)134 (12.8%)150 (21.1%)613 (16.2%)
 Stroke7 (1.3%)30 (1.3%)≤5 (2.0%)14 (1.3%)16 (2.2%)72 (1.9%)
 Venous thrombosis8 (1.5%)14 (0.6%)0 (0.0%)≤5 (0.4%)8 (1.1%)36 (1.0%)
Prescription drug use [1 year pre-event]      
 Aromatase inhibitor therapy23 (4.3%)83 (3.5%)47 (23.3%)251 (24.0%)71 (10.0%)426 (11.3%)
 At least 1 diabetes-promoting drugc413 (77.8%)1477 (61.6%)174 (86.1%)686 (65.6%)583 (81.9%)2416 (64.0%)
 Oral glucocorticoid therapy33 (6.2%)105 (4.4%)13 (6.4%)42 (4.0%)65 (9.1%)229 (6.1%)
 Estrogen therapy≤5 (0.8%)40 (1.7%)0 (0.0%)≤5 (0.3%)27 (3.8%)125 (3.3%)
 Progesterone therapy≤5 (0.9%)28 (1.2%)≤5 (0.5%)10 (1.0%)13 (1.8%)57 (1.5%)
 Raloxifene therapy≤5 (0.2%)≤5 (0.0%)≤5 (0.5%)≤5 (0.2%)6 (0.8%)14 (0.4%)
 Adjuvant chemotherapy28 (5.3%)128 (5.3%)10 (5.0%)54 (5.2%)72 (10.1%)343 (9.1%)

We conducted a secondary analysis to test the influence of duration of tamoxifen therapy on risk of diabetes among current/recent past users. This analysis was conducted to assess for a dose-response effect, and to address the possibility of a surveillance bias, whereby there is an increase in diabetes diagnoses shortly after starting tamoxifen because of enhanced health care in that period. We further classified current/recent past use into treatment for ≤2 years and >2 years before index date (diagnosis of diabetes). Each category was then compared with tamoxifen nonusers.

As a sensitivity analysis to assess the robustness of our findings, we repeated the above analysis to examine for diabetes risk among patients prescribed AIs, who have similar indications for breast cancer treatment. Because only women with estrogen-receptor positive (ER+) cancers receive tamoxifen or AIs, this analysis was performed to help separate effects related to ER+ status and/or estrogen inhibition versus tamoxifen itself. The exposure window definitions of AI exposure were similar to tamoxifen exposure. The use of AIs was much lower than tamoxifen use during our study period, and our databases only captured AI prescriptions that received special permission for coverage by the insurance plan.

Ethics

This study was approved by the research ethics board of Sunnybrook Health Sciences Centre in Toronto.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. FUNDING SOURCES
  9. REFERENCES

The study cohort consisted of 14,360 breast cancer survivors (mean age at breast cancer diagnosis date, 74.9 ± 6.52 years) who met the inclusion and exclusion criteria and who were followed up for an average of 5.2 years (standard deviation, 2.9 years). During the follow-up period, we identified 1445 cases that were newly diagnosed with diabetes. Of the 1445 cases with new diabetes, almost all (1442) were matched to 5 controls. Of these cases, 531 (37%) were exposed to current or recent past tamoxifen use (ie, their most recent treatment with tamoxifen was terminated within 180 days before the index date), and 127 (9%) had current or past AI use. The characteristics of the cases and their matched controls are described in Tables 1 (cases and controls) and 2 (cases and controls within tamoxifen exposure groups). Diabetes cases were more likely to have a history of cardiovascular disease, and were more likely to be taking oral glucocorticoids or other drugs that are known to promote diabetes (Table 1). When baseline characteristics were compared between tamoxifen users and nonusers, there were no significant differences in age, income, comorbidity, or cardiovascular disease. Current tamoxifen users were less likely to have been treated with diabetes-promoting drugs, glucocorticoids, an AI, or adjuvant chemotherapy (Table 2).

Compared with no use of tamoxifen, current or recent past use of tamoxifen was associated with a statistically significant increase in the likelihood of newly diagnosed diabetes (adjusted OR, 1.24; 95% CI, 1.08-1.42; P = .002) (Table 3). Remote past use of tamoxifen (exposed to tamoxifen >180 days before the index date) was not associated with an increased likelihood of diabetes (adjusted OR, 1.05; 95% CI, 0.87-1.27; P = .634). The estimates were similar for current users with ≤2 years and >2 years of tamoxifen therapy, when they were compared with no use of tamoxifen (Table 3). Less than 1% of current/past tamoxifen users had been prescribed an AI before starting tamoxifen, and the proportion was similar between cases and controls.

Table 3. ORs of Incidence of Diabetes for Tamoxifen Use Versus No Tamoxifen Use and AI Use Versus No AI Use
Pattern of UseCase, No.Controls, No.Unadjusted ORAdjusted
ORa95% CIP 
  • Abbreviations: AI, aromatase inhibitor; CI, confidence interval; OR, odds ratio.

  • a

    Adjusted for age; income quintiles; Charlson score in the year prior to cohort entry; rural residence, cardiovascular disease, renal disease, stroke, or venous thromboembolism in the 5 years prior to cohort entry; use of glucocorticoids; previous use of AIs (for tamoxifen); previous use of tamoxifen (for AIs); use of estrogen, progestogens, or raloxifene in the year prior to index date; and adjuvant chemotherapy.

  • b

    Remote use: discontinued treatment from >180 days to ≤5 years before the index date.

  • c

    Current use: dispensed a prescription with the day supplied + 14 days overlapping the index date.

  • d

    Recent past use: discontinued treatment from >14 days to ≤180 days before the index date.

Tamoxifen exposure       
No use71237751.00
Remote useb20210461.011.050.871.27.634
Currentc or recent pastd tamoxifen use53123991.211.241.081.42.002
Duration on tamoxifen       
0-2 years22910611.221.241.031.49.023
2+ years30213381.211.231.031.48.025
AI exposure       
No use129564251.00
Remote useb231071.050.970.591.59.893
Currentc or recent pastd AI use1276880.910.810.611.07.143

We did not find any association between current or past AI use and newly diagnosed diabetes (adjusted OR, 0.81; 95% CI, 0.61-1.07, P = .143); however, the number of current and recent past AI users was small (127 cases and 688 controls; Table 3). The baseline characteristics between current/past tamoxifen and AI users were generally similar; however, AI users tended to be younger than tamoxifen users (mean age, 72.7 vs 74.1 years) and were twice as likely to have received adjuvant chemotherapy (11.7% vs 5.3%). Of the current/past AI users, 44.2% had been prescribed tamoxifen before AI use, and there was no difference in prior tamoxifen use between cases and controls. Interestingly, there was also no significant association between previous adjuvant chemotherapy and risk of diabetes in our study (data not shown).

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. FUNDING SOURCES
  9. REFERENCES

This population-based study found that tamoxifen therapy among older breast cancer survivors was associated with a significantly increased likelihood of diabetes, even after adjustment for other risk factors. The increase in diabetes incidence related to tamoxifen therapy persisted throughout the treatment period, arguing against a role for enhanced diabetes testing around the time of breast cancer diagnosis and treatment. In addition, this association was only noted for current or recently discontinued tamoxifen treatment, and an association was not seen when previous tamoxifen users were compared with nonusers. These findings suggest that current tamoxifen therapy may promote an increased risk of diabetes in susceptible individuals, but does not appear to have a persistent or delayed effect after discontinuation. Our study highlights a potential new side effect of tamoxifen therapy that warrants further investigation.

The potential mechanisms underlying our findings are uncertain, but may be related to effects of estrogen inhibition. There is increasing evidence that estrogen may play a role in glucose homeostasis.11-15 The overall prevalence of diabetes is lower in premenopausal women, a trend that is reversed after menopause.25 The decline in estrogen after menopause is also associated with weight gain, insulin resistance, and the metabolic syndrome.26 Moreover, postmenopausal estrogen replacement has been shown to reduce the incidence of diabetes in clinical trials.18, 19 Treatment with tamoxifen has also been associated with hypertriglyceridemia and steatohepatitis, both of which are features of insulin resistance and glucose intolerance.27 In addition, a small study found a significant increase in visceral fat deposition, a prominent feature of insulin resistance, in breast cancer patients treated with tamoxifen compared with controls.28 Preclinical studies indicate that estrogen receptors are present on insulin-producing pancreatic beta-cells, and estrogen appears to protect against beta-cell apoptosis.17 Estrogen inhibition has been associated with increased insulin resistance16 and decreased insulin secretion.17 Tamoxifen has been shown to directly increase beta-cell apoptosis and decrease insulin secretion in mice.17 Although this effect may not be sufficient to cause diabetes in persons with normal pancreatic beta-cell function, it may precipitate hyperglycemia in persons with mild abnormalities who are already at risk for diabetes. Our findings support the hypothesis that tamoxifen may exert direct effects on glucose homeostasis, and highlight the need for further studies to better explore this effect.

It is also possible that differences in hormone receptor status between tamoxifen users and nonusers may have contributed to our findings. Tamoxifen treatment is only used in women with ER+ cancer,10 and data on hormone receptor status were not available in our study. Women with ER+ breast cancer are more likely to have risk factors that increase estrogen exposure, such as obesity, which is a risk factor for diabetes. However, although some studies show a weak association between obesity and ER+ cancers, data are inconsistent.29, 30 Moreover, there is evidence that women with pre-existing diabetes31 and the metabolic syndrome,32 a strong risk factor for diabetes, have a higher prevalence of receptor-negative breast cancer. Our study also did not find an increased risk of diabetes in other women with ER+ cancers, such as previous tamoxifen users or those treated with AIs. Although we cannot exclude this explanation, these findings argue against a prominent role of risk factors such as obesity common to all women with ER+ cancers in our study.

Our study has implications for both patients with breast cancer and for clinicians who care for these women. Tamoxifen therapy may promote an increase in the development of diabetes, particularly in higher risk populations such as older women or those with underlying risk factors. Moreover, women with breast cancer may already have a higher risk of diabetes because of common risk factors such as obesity and insulin resistance.6, 7 Further attention to diabetes screening and counseling for diabetes prevention in breast cancer survivors being prescribed tamoxifen may be warranted, especially in those with a family history, obesity, or previous gestational diabetes. To our knowledge, this association between tamoxifen and diabetes has not been observed in previous studies or clinical trials. This may be because previous studies have not formally searched for an association, especially as diabetes is common, has many risk factors, and often presents with few symptoms. Further research is needed to explore this relationship in clinical trial populations and in cohorts with more detailed metabolic data.

Strengths of our study include the use of large population-based datasets, a comprehensive cancer registry and drug information, and a validated measure of diabetes diagnosis. However, there are several limitations that should be mentioned. As in all observational studies, the possibility of bias and misclassification cannot be excluded. In this case, we cannot rule out an effect of confounding, whereby the higher risk of diabetes associated with tamoxifen therapy was caused by baseline differences in tamoxifen users rather than the treatment itself. We attempted to address this possibility by matching and adjusting for several risk factors for diabetes such as age, income, cardiovascular disease, comorbidity, and other drug use that might promote diabetes. We did not have access to clinical data such as body mass index, reproductive factors, family history, education, the presence of the metabolic syndrome of prior gestational diabetes, breast cancer stage, or hormone receptor status. We also acknowledge that tamoxifen therapy is more likely to be avoided in women who are at risk for thromboembolic or cardiovascular disease. However, this bias would have attenuated any difference in diabetes risk based on contraindications to tamoxifen. Reassuringly, we did not find significant differences in baseline cardiovascular factors between tamoxifen users and nonusers. In addition, we cannot exclude the role of a survival bias, whereby women taking tamoxifen have a better cancer prognosis and are therefore more likely to survive long enough to develop diabetes than women not taking tamoxifen. To address this issue, we limited our cohort to early stage breast cancer patients and matched the cases and controls on time since breast cancer diagnosis. The lack of association between AI use and diabetes may have also been caused by the small number of AI users in our study, and further work is needed to confirm this finding. Other limitations include the lack of information on rates of undiagnosed diabetes and detailed metabolic measures such as glucose levels, insulin resistance, and beta cell function.

In conclusion, this population-based study found that current tamoxifen use was associated with a significantly increased risk of new diabetes among older breast cancer survivors. The reasons for this association are uncertain, and further research is needed. In the meantime, our findings suggest that closer monitoring for diabetes in tamoxifen-treated breast cancer survivors may be warranted.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. FUNDING SOURCES
  9. REFERENCES

We thank Wei Wu, Sunila Kalkar, and Vladana Urosevic for assistance with manuscript and table preparation.

FUNDING SOURCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. FUNDING SOURCES
  9. REFERENCES

This study was conducted with the support of funding provided by Cancer Care Ontario and the Ontario Institute for Cancer Research (through funding provided by the Ministry of Health and Long-Term Care and the Ministry of Research and Innovation of the Government of Ontario) and by a Canadian Diabetes Association/Canadian Institutes of Health Research-Institute of Nutrition, Metabolism, and Diabetes Clinician Scientist Award, which also provides salary support for L.L.L.; P.A. is supported by a Career Investigator Award from the Heart and Stroke Foundation of Ontario.

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. FUNDING SOURCES
  9. REFERENCES