The association of sodium‐glucose cotransporter 2 inhibitors with cancer: An overview of quantitative systematic reviews

Abstract Aims To summarize reported cancer events associated with SGLT‐2 inhibitors used in patients with type 2 diabetes mellitus, as well as assess the quality of included reviews. Materials and methods In May 2019, we searched PubMed, Embase and the Cochrane Library for quantitative systematic reviews assessing the safety of SGLT‐2 inhibitors. Data were abstracted using a standardized form, and methodological quality was assessed using the AMSTAR 2 tool. Main outcome measures included total cancer events and specific cancers such as breast cancer, bladder cancer, gastrointestinal cancer, prostate cancer, respiratory cancer, renal cancer and skin cancer. Pooled treatment effects from included reviews were summarized for SGLT‐2 inhibitors as a class and for individual SGLT‐2 inhibitors commonly used worldwide (canagliflozin, dapagliflozin and empagliflozin). Results We screened 1248 unique citations, of which eight quantitative systematic reviews meta‐analysed results from studies reporting the association between an SGLT‐2 inhibitor and any cancer. Only one review was rated as high quality according to AMSTAR 2 assessment. In total, data from 170 cancer‐related point estimates (PE) were reported. As a class, SGLT‐2 inhibitors were not associated with an increased risk of any cancer event versus placebo and active comparators. Most point estimates (7/143) were nonsignificant for individual cancers except for two associations. Empagliflozin was associated with an increased risk of bladder cancer versus placebo and active comparators in two reviews, while canagliflozin appeared protective for gastrointestinal cancer versus placebo and active comparators in one review. Conclusions It appears that SGLT‐2 inhibitors are not associated with an increased risk of total cancer or specific cancers in patients with type 2 diabetes. However, higher quality evidence is needed to derive confident conclusions.


| INTRODUC TI ON
Sodium-glucose cotransporter 2 (SGLT-2) inhibitors are a novel class of antihyperglycaemic agents used in the treatment of type 2 diabetes mellitus. These agents inhibit the SGLT-2 protein expressed in the proximal tubule within the kidney, which is responsible for the renal reabsorption of glucose. 1 Inhibition of these transporters facilitates blood glucose reduction via urinary excretion of glucose. 1 There are a wide variety of benefits associated with SGLT-2 inhibitor use in type 2 diabetes, including significant reduction in haemoglobin A1C, reduction in major cardiovascular adverse events (MACE) and significant reduction in the risk of end-stage kidney disease compared to placebo. [2][3][4] Due to these demonstrated benefits, the utilization of SGLT-2 inhibitors for the treatment of type 2 diabetes mellitus has rapidly increased since market approval. 5 However, these agents have undergone unprecedented postmarketing investigations given the FDA requirements to demonstrate cardiovascular safety of new antihyperglycaemic agents. Despite elusive mechanisms, cancer risk associated with SGLT-2 inhibitors has been reported in several quantitative systematic reviews. [6][7][8][9][10][11][12][13] There are signals in the literature that SGLT-2 inhibitors may affect cancer risk. It has been postulated that SGLT-2 inhibitors may activate medullary thyroid tumour growth in both rats and male mice; however, the relevance of this information in humans is not known. 14,15 Furthermore, in 2011, the US Food and Drug Administration (FDA) observed discrepancies in the risk of bladder and breast cancers with dapagliflozin versus comparators. 16 Regulatory concerns were also raised due to an imbalance of lung cancer and melanoma observed with empagliflozin use. 17 In contrast, canagliflozin has been associated with a decreased risk of stomach cancer. 10 SGLT-1 has been implicated in cancer cell survival via glucose uptake; therefore, canagliflozin's inhibition of both SGLT-1 and SGLT-2 receptors has been proposed for this agent's purported protective effect. 18 Interestingly, there have been several systematic reviews and meta-analyses reporting on cancer risk associated with SGLT-2 inhibitor use. A combination of low cancer event rates, poor diagnostic consistency and short follow-up times of studies included in quantitative reviews assessing cancer risk in SGLT-2 inhibitor users thus far have made it difficult for clinicians to draw confident conclusions on potentially relevant implications of this data in practice. Given these limitations as well as variance in the methodological rigour of published quantitative systematic reviews, there is a need to critically review, evaluate and summarize these studies. Therefore, we conducted an overview of reviews, adapted from Cochrane Overviews, which serves to effectively accomplish this task. 19 An overview of reviews provides clinicians, policymakers and clinical guideline developers with a summary of the available evidence for a topic of interest. We aimed to summarize evidence from and assess the quality of published quantitative systematic reviews evaluating the cancer risk associated with SGLT-2 inhibitor use in the treatment of type 2 diabetes.

| ME THODS
The protocol for this overview of reviews is registered with the PROSPERO international prospective register of systematic reviews (PROSPERO 2019:CRD42019135863). 20 This overview is part of a series of overviews of reviews exploring various adverse events associated with SGLT-2 inhibitor use in patients with type 2 diabetes mellitus.

| Eligibility criteria
Systematic reviews of randomized controlled studies, cohort or case-control studies with a meta-analysis (ie quantitative systematic reviews) that evaluated SGLT-2 inhibitor safety and collected data on adverse events (beyond hypoglycaemia) were included. Quantitative systematic reviews that did not use a systematic search strategy were excluded. Our outcomes of interest were any point estimates reporting on the association between SGLT-2 inhibitors and any type of cancer in quantitative systematic reviews. We did not restrict the inclusion of quantitative systematic reviews based on the timing of the outcome following drug exposure. We restricted the language of included reviews to English.

| Sources and searching
Potentially relevant quantitative systematic reviews were identi- Endocrinology. Third, we hand searched the references of included systematic reviews. The search strategy is available in Appendix S1.

| Quality assessment
Two independent reviewers assessed the quality of included systematic reviews using the 'A MeaSurement Tool to Assess systematic Reviews 2' (AMSTAR 2) checklist. 21 AMSTAR 2 is a validated tool consisting of 16 domains that assess the methodological quality of systematic reviews containing both randomized and nonrandomized studies of interventions. All discordant AMSTAR 2 quality ratings between reviewers were resolved by consensus.
Consistent with AMSTAR 2 published literature, systematic reviews having more than one critical flaw were rated as critically low quality, one critical flaw as low quality, more than one noncritical weakness as moderate quality and no or one noncritical weakness as high quality. Domains 2, 7, 4, 9, 11, 13 and 15 are considered critical in AMSTAR 2. 21

| Analysis
We conducted a descriptive analysis of our results by summariz-

| RE SULTS
We identified 1248 unique citations, of which eight quantitative systematic reviews met our inclusion criteria (

| Quality assessment
The complete AMSTAR 2 assessments and overall quality ratings for included systematic reviews are shown in Appendix S1: Figure S1.
Only one (12.5%) included review received an AMSTAR 2 quality rating of high. 8 Four (50%) reviews were considered critically low quality, one (12.5%) review was considered low quality and two (25%) reviews were considered moderate quality.

| D ISCUSS I ON
SGLT-2 inhibitors do not appear to be associated with an overall increased risk of cancer in patients with type 2 diabetes mellitus.

Point estimates reported for class effects of SGLT-2 inhibitors
on the risk of any cancer event, as well as specific cancer subtypes, showed no significant association with the use of these agents. This held true regardless of whether SGLT-2 inhibitor interventions were compared with placebo or active comparators.
However, practicing clinicians do not prescribe by class, but rather by individual SGLT-2 inhibitor agents for antihyperglycaemic management. Considering popular SGLT-2 inhibitor agents used globally (ie canagliflozin, dapagliflozin and empagliflozin), most cancer-related data collected for these individual agents also indicated there were no significant associations between their use and overall risk of any cancer event. Some individual point estimates from included reviews, specifically for canagliflozin and empagliflozin, reported a statistically significant decreased risk of gastric cancer and increased risk of bladder cancers for users of these agents, respectively. 10,13 There are several potential reasons that could account for the statistically significant associations observed between empagliflozin use and bladder cancer. First, detection bias is a plausible explanation for this increased risk. 22  Although several quantitative systematic reviews have been published assessing cancer-related events as primary and secondary outcomes in SGLT-2 inhibitor users, the methodological rigour of these studies appears to be inconsistent. Half of the included quantitative systematic reviews were considered to be of 'critically low quality' according to AMSTAR 2 assessment. Since systematic reviews are considered to be at the top of the scientific evidence pyramid, it is crucial that the methods undertaken to complete these reviews are transparent and replicable. 26 However, this does not necessarily mean that the point estimates reported within reviews deemed 'low quality' by AMSTAR 2 rating are sourced from low quality evidence. The quality of evidence contained within the included reviews was generally high (ie evidence from randomized controlled trials and government regulatory reports). It is important to remember that the AMSTAR 2 tool is used to measure the methodological quality of systematic reviews, not to assess the quality of evidence contained within the review.
Our review provides clinicians with a comprehensive summary that highlights important limitations of assessing SGLT-2 inhibitor-associated cancer risk using quantitative systematic reviews.
Despite using established methods (eg published protocol, comprehensive search strategy, screening and quality assessment performed by at least 2 independent reviewers), our overview also has some limitations. We did not meta-analyse the point estimates gathered from included reviews as this was beyond the scope of this study. Additionally, our unit of analysis was at the review level. Furthermore, with the large volume of reviews that have been published on SGLT-2 inhibitor safety, it is possible that additional studies and reviews assessing cancer risk in our population of interest are currently under consideration for publication. A potential resolution to prevent evidence from individual systematic reviews from becoming quickly outdated is to develop a 'living systematic review' that has been described by the Cochrane community. 27 With an updated literature search that is ideally conducted once monthly, living systematic reviews are continually updated with the most current evidence as it becomes available.

| CON CLUS ION
As current evidence stands, canagliflozin, dapagliflozin and empagliflozin do not appear to significantly impact cancer risk in patients with type 2 diabetes; however, long-term safety data are lacking.
Given the limitations of the included quantitative systematic reviews, as well as imprecise effect estimates reported in these reviews, more long-term data from high quality observational studies are needed to more precisely assess cancer risks associated with SGLT-2 inhibitor use. Future studies should focus on quantifying bladder and gastrointestinal cancers.

CO N FLI C T O F I NTE R E S T
None of us have any financial arrangements or any potential conflicts of interest to disclose with regard to the products in this manuscript.

AUTH O R CO NTR I B UTI O N
RP, KN and JMG conceptualized the review. RP wrote the first draft of the manuscript, made suggested changes from co-authors and formatted the paper for publication. All authors partook in the review selection and critical appraisal processes, as well as provided intellectual feedback on manuscript drafts. All authors approved the final draft of the manuscript prior to submission.

E TH I C S S TATEM ENT
Our study did not undergo review by a human research ethics board as it did not involve human subjects and consisted of a review of the literature using aggregated anonymous data.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data are available within the article and its Appendix S1 and Appendix S2.