Sodium–glucose cotransporter 2 (SGLT2) inhibitors are second- to third-line antidiabetes drugs that have been shown to have cardiovascular benefits. However, in premarketing trials of the SGLT2 inhibitor dapagliflozin, there were numerical imbalances in breast cancer events compared with placebo; all occurred within 1 year of randomization (rate ratio 2.47, 95% CI 0.64–14.10) (1). In contrast, no imbalances were observed in subsequent large cardiovascular outcome trials of dapagliflozin and other SGLT2 inhibitors (2–4).
While the discrepant findings between the pre- and postmarketing trials could be due to chance, one hypothesis is that the early breast cancer imbalances are the result of an accelerated tumor-promoting effect of SGLT2 inhibitors. However, this is unlikely, given that sodium–glucose cotransporter proteins are not expressed in mammary tissue, and animal studies have failed to demonstrate that SGLT2 inhibitors have any neoplastic activity (5). Another hypothesis relates to the weight-lowering effects of SGLT2 inhibitors, which could facilitate the detection of existing breast lumps, thus leading to a transient overdetection of breast cancer. To date, this possible association has not been investigated in the real-world setting.
We conducted a population-based cohort study using the U.K. Clinical Practice Research Datalink (CPRD) (protocol: 19_272). We identified all female patients newly treated with either an SGLT2 inhibitor (dapagliflozin, canagliflozin, empagliflozin) or a dipeptidyl peptidase 4 (DPP-4) inhibitor (sitagliptin, saxagliptin, linagliptin, alogliptin) between 1 January 2013 (the year the first SGLT2 inhibitor, dapagliflozin, entered the U.K. market) and 30 June 2018. Cohort entry was the date of the first prescription of either drug class during the study period. To be included in the cohort, patients were required to be at least 40 years of age and have at least 1 year of medical history in the CPRD before cohort entry. This 1-year period served as a washout window to identify new users and was used as a baseline covariate assessment period. We excluded patients previously diagnosed with end-stage renal disease or undergoing dialysis (contraindications to receiving SGLT2 inhibitors), as well as those previously diagnosed with breast cancer at any time before cohort entry. Patients were followed using an intention-to-treat approach until an incident diagnosis of breast cancer (identified using Read codes with a positive predictive value of 90%, compared with U.K. National Cancer Data Repository [6]), death from any cause, or end of study period (31 December 2018)—whichever occurred first.
We used propensity score fine stratification weighting to control for confounding. The propensity score model considered >40 covariates (Table 1), including age, BMI, smoking status, proxies of diabetes severity, prescription drugs, and breast cancer risk factors. Cox proportional hazards models were used to estimate weighted hazard ratios (HRs) of breast cancer, and corresponding 95% CIs, in comparison of new users of SGLT2 inhibitors with new users of DPP-4 inhibitors. We also assessed whether there was a duration-response relation in terms of cumulative duration of use and time since treatment initiation and investigated whether the association varied according to type of SGLT2 inhibitor. Finally, in sensitivity analyses, we imposed lag periods, accounted for death as a competing risk using inverse probability of censoring weighting, and repeated the analysis by matching on propensity score.
HRs for incident breast cancer in comparison of use of SGLT2 inhibitors with use of DPP-4 inhibitors
| . | Patientsa . | Events . | Person-years . | Incidence rate (95% CI)b . | Weighted HR (95% CI)c . | |
|---|---|---|---|---|---|---|
| Primary analysis | ||||||
| DPP-4 inhibitors | 36,631 | 382 | 103,228 | 3.7 (3.3–4.1) | 1.00 (Reference) | |
| SGLT2 inhibitors | 9,938 | 67 | 23,621 | 2.8 (2.2–3.6) | 1.00 (0.76–1.30) | |
| Type of SGLT2 inhibitor | ||||||
| DPP-4 inhibitors | 35,380 | 369 | 100,523 | 3.7 (3.3–4.1) | 1.00 (Reference) | |
| Canagliflozin | 1,726 | 7 | 4,083 | 1.7 (0.7–3.5) | 0.62 (0.29–1.31) | |
| DPP-4 inhibitors | 36,650 | 384 | 103,289 | 3.7 (3.4–4.1) | 1.00 (Reference) | |
| Dapagliflozin | 5,429 | 50 | 15,111 | 3.3 (2.5–4.4) | 1.16 (0.86–1.57) | |
| DPP-4 inhibitors | 36,044 | 377 | 101,978 | 3.7 (3.3–4.1) | 1.00 (Reference) | |
| Empagliflozin | 2,771 | 10 | 4,405 | 2.3 (1.1–4.2) | 0.83 (0.44–1.56) | |
| Sensitivity analyses | ||||||
| 1-year lagd | ||||||
| DPP-4 inhibitors | 30,620 | 262 | 67,937 | 3.9 (3.4–4.4) | 1.00 (Reference) | |
| SGLT-2 inhibitors | 8,143 | 48 | 14,250 | 3.4 (2.5–4.5) | 1.14 (0.83–1.56) | |
| 2-year lagd | ||||||
| DPP-4 inhibitors | 22,854 | 167 | 41,292 | 4.0 (3.5–4.7) | 1.00 (Reference) | |
| SGLT-2 inhibitors | 5,503 | 23 | 7,422 | 3.1 (2.0–4.6) | 1.00 (0.64–1.56) | |
| Competing riske | ||||||
| DPP-4 inhibitors | 36,631 | 382 | 103,228 | 3.7 (3.3–4.1) | 1.00 (Reference) | |
| SGLT-2 inhibitors | 9,938 | 67 | 23,621 | 2.8 (2.2–3.6) | 1.00 (0.76–1.30) | |
| Matching on propensity scoref | ||||||
| DPP-4 inhibitors | 8,642 | 78 | 25,577 | 3.0 (2.4–3.8) | 1.00 (Reference) | |
| SGLT-2 inhibitors | 8,642 | 54 | 20,056 | 2.7 (2.0–3.5) | 0.90 (0.63–1.28) |
| . | Patientsa . | Events . | Person-years . | Incidence rate (95% CI)b . | Weighted HR (95% CI)c . | |
|---|---|---|---|---|---|---|
| Primary analysis | ||||||
| DPP-4 inhibitors | 36,631 | 382 | 103,228 | 3.7 (3.3–4.1) | 1.00 (Reference) | |
| SGLT2 inhibitors | 9,938 | 67 | 23,621 | 2.8 (2.2–3.6) | 1.00 (0.76–1.30) | |
| Type of SGLT2 inhibitor | ||||||
| DPP-4 inhibitors | 35,380 | 369 | 100,523 | 3.7 (3.3–4.1) | 1.00 (Reference) | |
| Canagliflozin | 1,726 | 7 | 4,083 | 1.7 (0.7–3.5) | 0.62 (0.29–1.31) | |
| DPP-4 inhibitors | 36,650 | 384 | 103,289 | 3.7 (3.4–4.1) | 1.00 (Reference) | |
| Dapagliflozin | 5,429 | 50 | 15,111 | 3.3 (2.5–4.4) | 1.16 (0.86–1.57) | |
| DPP-4 inhibitors | 36,044 | 377 | 101,978 | 3.7 (3.3–4.1) | 1.00 (Reference) | |
| Empagliflozin | 2,771 | 10 | 4,405 | 2.3 (1.1–4.2) | 0.83 (0.44–1.56) | |
| Sensitivity analyses | ||||||
| 1-year lagd | ||||||
| DPP-4 inhibitors | 30,620 | 262 | 67,937 | 3.9 (3.4–4.4) | 1.00 (Reference) | |
| SGLT-2 inhibitors | 8,143 | 48 | 14,250 | 3.4 (2.5–4.5) | 1.14 (0.83–1.56) | |
| 2-year lagd | ||||||
| DPP-4 inhibitors | 22,854 | 167 | 41,292 | 4.0 (3.5–4.7) | 1.00 (Reference) | |
| SGLT-2 inhibitors | 5,503 | 23 | 7,422 | 3.1 (2.0–4.6) | 1.00 (0.64–1.56) | |
| Competing riske | ||||||
| DPP-4 inhibitors | 36,631 | 382 | 103,228 | 3.7 (3.3–4.1) | 1.00 (Reference) | |
| SGLT-2 inhibitors | 9,938 | 67 | 23,621 | 2.8 (2.2–3.6) | 1.00 (0.76–1.30) | |
| Matching on propensity scoref | ||||||
| DPP-4 inhibitors | 8,642 | 78 | 25,577 | 3.0 (2.4–3.8) | 1.00 (Reference) | |
| SGLT-2 inhibitors | 8,642 | 54 | 20,056 | 2.7 (2.0–3.5) | 0.90 (0.63–1.28) |
Data are n unless otherwise indicated.
The number of patients in the DPP-4 inhibitor and SGLT2 inhibitor groups differs slightly across analyses due to trimming of patients in nonoverlapping regions of the propensity score. In the primary analysis, most SGLT2 inhibitor users were prescribed dapagliflozin (n = 5,429 [54.6%]), followed by empagliflozin (n = 2,771 [27.9%]) and canagliflozin (n = 1,726 [17.4%]). Of the DPP-4 inhibitor users, most were prescribed sitagliptin (n = 18,360 [50.1%]), followed by linagliptin (n = 10,303 [28.1%]), alogliptin (n = 4,587 [12.5%]), saxagliptin (n = 3,087 [8.4%]), and vildagliptin (n = 294 [0.8%]).
Per 1,000 person-years.
Propensity score fine stratification weighting. The propensity score model included age; BMI; alcohol-related disorders; smoking status; duration of diabetes; hemoglobin A1c (last measure before cohort entry); microvascular complications (nephropathy, neuropathy, retinopathy); macrovascular complications (peripheral vascular disease, myocardial infarction, ischemic stroke); previous use of antidiabetes drugs (metformin, sulfonylureas, thiazolidinedione, meglitinides, α-glucosidase inhibitors, glucagon-like peptide 1 receptor agonists, and insulin, all entered as non–mutually exclusive variables and measured at any time before cohort entry); oophorectomy; cancer (other than nonmelanoma skin cancer); congestive heart failure; coronary artery disease; chronic kidney disease; ever use of antihypertension drugs (β-blockers, calcium channel blockers, ACE inhibitors, angiotensin receptor blockers, and diuretics); ever use of anti-inflammatory drugs, antiplatelets (aspirin, clopidogrel, ticagrelor, prasugrel), statins, hormone replacement therapy, and oral anticoagulants (warfarin, heparin); and mammographic screening history in the year before cohort entry.
The exposure groups were lagged by 1 and 2 years by censoring of breast cancer events that occurred in the 1 and 2 years after treatment initiation, respectively.
The analysis considered death from any cause as a competing risk using inverse probability of censoring weighting.
The analysis was repeated by matching of SGLT2 inhibitor users to DPP-4 inhibitor users on propensity score, in a 1:1 ratio, using nearest neighborhood matching within a caliper of 0.01.
The study included 9,938 new users of SGLT2 inhibitors and 36,631 new users of DPP-4 inhibitors. After a median follow-up of 2.6 years, these exposure groups generated 67 and 382 breast cancer events, yielding crude incidence rates of 2.8 (95% CI 2.2–3.6) and 3.7 (95% CI 3.3–4.1) per 1,000 person-years, respectively. Overall, compared with DPP-4 inhibitor use, SGLT2 inhibitor use was not associated with an overall increased risk of breast cancer (HR 1.00, 95% CI 0.76–1.30) (Table 1). Similarly, there was no clear duration-response relation with SGLT2 inhibitors in terms of cumulative duration of use and time since treatment initiation. The analysis by type of SGLT2 inhibitor generated HRs below the null for canagliflozin and empagliflozin, while the HR was above the null for dapagliflozin (HR 1.16, 95% CI 0.86–1.56). In a post hoc analysis, 1–2 years’ cumulative duration of dapagliflozin use generated a moderately elevated HR but with a wide CI (HR 1.58, 95% CI 0.95–2.64). Finally, the results remained consistent in the several sensitivity analyses.
In summary, the results of this large population-based cohort study indicate that use of SGLT2 inhibitors is not associated with an overall increased risk of breast cancer compared with DPP-4 inhibitors. While these findings provide some reassurance on the short-term effects of SGLT2 inhibitors on breast cancer incidence among female patients with type 2 diabetes, future studies will be needed to investigate their long-term effects on this malignancy.
Article Information
Acknowledgments. Dr. Yu holds a Chercheur-Clinicien Junior 1 award from Fonds de Recherche du Québec - Santé. Dr. Azoulay holds a Chercheur-Boursier Senior Award from Fonds de Recherche du Québec - Santé and is the recipient of a William Dawson Scholar award from McGill University.
Funding. This study was funded by a Foundation Scheme grant from the Canadian Institutes of Health Research (FDN-143328).
The sponsor had no influence on the design or conduct of the study; collection, management, analysis, or interpretation of the data; or preparation, review, or approval of the manuscript.
Duality of Interest. All authors have completed the ICMJE uniform disclosure form at www.icmje.org/coi_disclosure.pdf and declare that L.A. received consulting fees from Janssen and Pfizer for work unrelated to this study. No other potential conflicts of interest relevant to this article were reported.
Author Contributions. M.S., H.Y., O.H.Y.Y., S.M.W., and L.A. contributed to study concept and design. L.A. acquired data. M.S., H.Y., O.H.Y.Y., S.M.W., and L.A. contributed to analysis and interpretation of data. M.S. drafted the manuscript. M.S., H.Y., O.H.Y.Y., S.M.W., and L.A. contributed to critical revision of the manuscript for important intellectual content. M.S., H.Y., and L.A. contributed to statistical analysis. L.A. obtained funding. L.A. supervised the study. L.A. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
