SGLT-2i and Risk of Malignancy in Type 2 Diabetes: A Meta-Analysis of Randomized Controlled Trials

Background: Currently, the association between sodium-glucose cotransporter 2 inhibitor (SGLT-2i) and malignancy risk has yet to be fully elucidated. This meta-analysis aimed to determine the relationship between SGLT-2i and malignancy risk in type 2 diabetes (T2D) patients. Methods: We searched PubMed, ScienceDirect, EMBASE, Cochrane Central Register of Controlled Trials, and Web of Science to identify randomized controlled trials (RCTs) published up to August 2020 related to T2D patients treated with SGLT-2i vs. placebo or other hypoglycemic agents. The meta-analysis's primary outcome was malignancies' incidence, and the results were evaluated using risk ratio (RR) and 95% confidence interval (CI). Results: We reviewed 76 articles (77 RCTs), comprising 45,162 and 43,811 patients in SGLT-2i and control groups, respectively. Compared with the control group, SGLT-2i had no significant association with augmented overall malignancy risk in T2D patients (RR = 1.05, 95% CI = 0.97–1.14, P = 0.20), but ertugliflozin may upsurge the risk (RR = 1.80, 95% CI = 1.02–3.17, P = 0.04). Compared with active hypoglycemic agents, dapagliflozin may increase (RR = 2.71, 95% CI = 1.46–6.43, P = 0.02) and empagliflozin may decrease (RR = 0.67, 95% CI = 0.45–0.98, P = 0.04) the malignancy risk. Compared with placebo, empagliflozin may exhibit risk increase (RR = 1.25, 95% CI = 1.05–1.49, P = 0.01), primarily in digestive system (RR = 1.48, 95% CI = 0.99–2.21, P = 0.05). Conclusions: Our results proposed that in diverse comparisons, ertugliflozin and dapagliflozin seemed to increase the malignancy risk in T2D patients. Empagliflozin may cause malignancy risk reduction compared with active hypoglycemic agents but increase overall risk primarily in the digestive system compared with placebo. In short, the relationship between SGLT-2i and malignancy in T2D patients remains unclear.


INTRODUCTION
The incidence of diabetes rises annually, with about 463 million people living with the disease today and an estimated 578 million by 2030 (1). Poor blood sugar control in diabetics may cause blindness, kidney failure and lower limb amputations (2). In recent decades, type 2 diabetes (T2D) has become a global public health crisis with a severe impact on human health (3), accounting for about 90% of people with diabetes, and the second leading global death reason is cancer, representing one sixth (4). Diabetes is evidenced to associate with a potential malignancy, risk with diabetics having a 10-20% higher risk of malignancy than non-diabetics (5). Studies have demonstrated that T2D significantly increases specific cancers' risk, such as liver and pancreatic cancer (6). The tumor-causing mechanisms of T2D may include hyperinsulinemia, insulin resistance, hyperglycemia, oxidative stress, and chronic inflammation (7).
Sodium-glucose cotransporter 2 inhibitor (SGLT-2i) can selectively inhibit glucose renal reabsorption and increase urine glucose excretion, independent of insulin action to reduce the blood sugar level of drug (8). In addition to reducing blood sugar and weight and lowering blood pressure, studies have revealed that SGLT-2i is beneficial in slowing the progression of cardiovascular and kidney diseases (9,10). Based on the above advantages, SGLT-2i has a great application prospect. Evidence proposes that SGLT-2i is not significantly associated with increased overall cancer risk (11). However, some SGLT-2i can increase or decrease certain cancers' risk, such as dapagliflozin, which may increase the risk of bladder cancer and breast cancer in T2D patients (12), and canagliflozin may reduce the risk of gastrointestinal cancers (11). Given the low incidence of malignant tumors and the long incubation period, a longer follow-up time is mandatory. Our meta-analysis was conducted to investigate SGLT-2i impact on malignancy incidence in T2D patients.

Search Strategy
This meta-analysis was reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines (13). The included randomized controlled trials (RCTs) were SGLT-2i in T2D patients. SGLT-2i such as dapagliflozin, canagliflozin, empagliflozin, tofogliflozin, ertugliflozin, luseogliflozin, and bexagliflozin were compared with placebo or other active hypoglycemic drugs to explore malignancy incidence in patients during follow-up.
After conducting a comprehensive and systematic search in PubMed, ScienceDirect, EMBASE, Cochrane Central Register of Controlled Trials and Web of Science databases, only articles published in English by August 2020 and before Abbreviations: SGLT-2i, sodium-glucose cotransporter 2 inhibitor; RCT, randomized controlled trial; RR, risk ratio; CI, confidence interval; T2DM, type 2 diabetes mellitus; T2D, type 2 diabetes; GLP-1RA, glucagon-like peptide-1 receptor agonist; DPP-4i, dipeptidyl-peptidase-4 inhibitor; SGLT2, sodiumglucose cotransporter 2; UGT1A9, UDP glucuronosyltransferase family 1 member A9; AMPK, adenosine monophosphate activated protein kinase; kDa, kilodalton. are retrieved. The search formula was as follows: (type 2 diabetes OR type 2 diabetes mellitus OR T2DM OR T2D) AND (sodium-glucose cotransporter 2 inhibitor OR SGLT-2i OR sotagliflozin OR janagliflozin OR dapagliflozin OR canagliflozin OR empagliflozin OR ipragliflozin OR tofogliflozin OR ertugliflozin OR luseogliflozin OR sergliflozin OR licogliflozin OR remogliflozin OR bexagliflozin). Two researchers independently searched the articles, reviewed the title and abstract, viewed the full text, and selected the inclusion articles. To avoid missing negative results, the vocabulary related to malignant tumors was not limited. Instead, the full text (including Supplementary Materials) was scanned to extract relevant data.

Study Selection
Studies that fulfill the following criteria were encompassed in this meta-analysis: (1) participants were T2D patients; (2) RCTs compared the therapeutic efficacy of SGLT-2i with placebo or other hypoglycemic agents; (3) RCTs stated thorough information on malignancy occurrence; (4) the experimental group was provided SGLT-2i therapy (including single drug or combination drug), and the control group was supplied non-SGLT-2i therapy (placebo or other hypoglycemic drugs). Exclusion criteria: (1) non-RCTs, including review, observational research, cases; (2) patients with type 1 diabetes mellitus or healthy volunteers; (3) non-English language; (4) duplicate reports. When articles were repeatedly updated, the most recent or datacomplete one was involved herein. After a systematic search, the two authors evaluated all chosen works, and the questionable studies were further discussed to resolve various opinions.

Data Extraction and Quality Assessment
Data extraction for studies included was performed independently by two researchers and reviewed by a third one. The extracted data comprise (1) study characteristics, such as author, region, year of publication, and followup time; (2) participant characteristics, including age, gender and subject inclusion criteria; (3) total number of malignant neoplasms, including primary, recurrent and metastatic cancers and classification of different types of tumors; (4) drug dose utilized by the experimental and control groups.
RCTs were assessed utilizing the Cochrane Collaboration's tool. The evaluation criteria include review and judgment of "low risk, " "high risk, " or "unclear risk" in terms of sequence generation, allocation concealment, blinding, incomplete outcome data, as well as selective outcome reporting and free of other bias. Any differences between the two researchers were resolved by discussion or by a third person review.

Statistical Analysis
The Review Manager 5.3 statistical analysis software was employed for the above analysis. The risk ratio (RR) and 95% confidence interval (CI) were deployed to evaluate the results. Heterogeneity of included studies was assessed employing I 2 statistics, where I 2 < 50%, indicating low heterogeneity, and a fixed-effect model was used; otherwise, a random-effect model was utilized. P ≤ 0.05 was statistically significant, and P < 0.10 within the suspected influence scope.

Eligible Studies and Characteristics
A total of 14,260 articles were initially searched, leaving 3,331 articles after deletion of duplicates. By reviewing title and abstract information, 2,963 articles were excluded. By evaluating 368 fulltext articles, 292 articles were excluded, comprising 16 non-RCTs, 171 articles provided data without malignant tumors, and 105 repeated reports. Finally, 76 articles between 2012 and 2020 were chosen for this meta-analysis , with 77 RCTs. Typically, 59 articles existed on SGLT-2i vs. placebo [60 RCTs in total, one article containing 2 RCTs (64)], and 25 articles existed on SGLT-2i vs. other hypoglycemic agents (Figure 1).

Risk of Bias Assessment
In most trials, the sequence generation and allocation concealment were low risk of bias, while only one research was unclear for allocation concealment. Five trials had a high risk of bias in the blind method. All the included trials possess low risk of bias for incomplete outcome data. Regarding selective outcome reporting, 11 studies had low risk of bias, while the rest were each unclear risk. Finally, all studies were judged as unclear risk for free of other bias (Supplementary Table 1).

DISCUSSION
SGLT-2i possesses good benefits in lowering blood glucose, but some safety problems may result in urogenital infection, bone fractures, ketoacidosis, etc. (90), so its clinical use requires to be considered comprehensively. Epidemiological studies have manifested a link between T2D and cancer, and one of the reasons may be hyperglycemia itself (91). While SGLT-2i may affect malignant tumors occurrence by lowering blood glucose, its comprehensive impact is still uncertain. Tang et al. analyzed 46 RCTs from 24 to 160 weeks and stated that empagliflozin might correlate with increased risk of bladder cancer (11). Nevertheless, the data involved in this analysis were challenged, and the corrected data showcased that empagliflozin might not be linked to bladder cancer (92). Beyond that, Tang et al. also concluded that canagliflozin might have a protective effect on gastrointestinal cancer (11). A meta-analysis of 27 trials displayed that SGLT-2i were not statistically associated with any cancer type (93). The abovementioned studies may be due to the low incidence of malignant tumors, small statistical sample size and short follow-up time, making it challenging to get clear and unified results.
This meta-analysis showed neither a significant association between SGLT-2i and the overall risk of malignancy in T2D patients nor with medication duration, consistent with the results of previous analysis (11,93,94). In different types of SGLT-2i analyses, we found that ertugliflozin significantly increased overall malignancy incidence, had no statistically significant difference compared to other hypoglycemic drugs or placebo alone, and had no great risk of a specific malignant tumor. The above could be due to lack of test sample size. A pooled analysis of 7 RCTs concluded that ertugliflozin had no significant difference in malignancies incidence compared with placebo or other active hypoglycemic agents (95).
Moreover, we observed that dapagliflozin might reduce the risk of respiratory system malignancies compared with the control group, but without statistical significance. Sodiumglucose cotransporter 2 (SGLT2) expression increased at the lung premalignancy and early-stage lung adenocarcinoma (96), and dapagliflozin may be able to reduce cancer cells proliferation by inhibiting glucose transport. Villani et al. also found that canagliflozin could prevent lung cancer cells' proliferation by precluding respiration supported by mitochondrial complex-I (97). It is worth noting that our data also indicated that compared with other hypoglycemic drugs, dapagliflozin could increase the overall risk of malignant tumors, and the most likely one was the digestive system malignancy, but without statistically significant difference. According to preceding studies, some active antidiabetic drugs can impede tumors. For example, a meta-analysis by Dicembrini et al. indicates that DPP-4i may have a potential inhibitory effect on colorectal cancer (98). A meta-analysis of 21 studies showed that metformin might be beneficial for survival in patients with pancreatic cancer and diabetes (99). Wu et al. found that metformin had no effect on the overall esophageal cancer risk, but it may reduce esophageal The values in italics represent statistical differences in the results (i.e., P < 0.05).
Frontiers in Public Health | www.frontiersin.org trials displayed no significant association between SGLT-2i and pancreatic cancer incidence (106). Since our research records the digestive system malignant tumor sample size too small (includes only 4 RCTs, dapagliflozin group 6 cases, and other active hypoglycemic drugs group 0 cases), difference did not reach statistical significance. Accordingly, it is uncertain whether the dapagliflozin increase of digestive system malignancy arising from the overall risk is high. Simultaneously, our data revealed that compared with other hypoglycemic drugs, empagliflozin could reduce the overall risk of malignant tumors and may reduce the incidence of urinary system malignant tumors, but the latter without  (108). Data analysis from 20 empagliflozin and placebo-controlled trials revealed no significant association between empagliflozin and the incidence of bladder and renal malignancies (109). An observational, prospective follow-up study has reported that empagliflozin can enhance anti-inflammatory and antioxidant effects in T2D patients, including cardiovascular advantages (110). Antiinflammatory and antioxidant impacts may also be linked to reduced incidence of malignancies. However, due to the limited number of included RCTs and the difference was not statistically significant, it cannot be confirmed that the overall risk reduction of malignant tumors is ascribed to urinary malignancies reduction. Furthermore, compared with placebo, we stated that empagliflozin significantly increased malignant tumor risk, mainly those of the digestive system. Nevertheless, one summarizes 15 randomized phase I-III trials plus four extension studies of empagliflozin and a placebo-controlled study indicating that empagliflozin safety data had no linkage to T2D patients with malignant tumor (111). The conclusion is still out on whether empagliflozin increases the risk of malignancies.
Our study showcased that canagliflozin might potentially increase breast cancer risk compared with placebo, but no statistically significant difference existed. Earlier data submitted to the Food and Drug Administration (FDA) suggested that dapagliflozin might upsurge breast cancer risk, but subsequent studies have suggested that it may be increased risk due to early cancer diagnosis rather than actual increase in incidence (112). Studies have shown that breast cancer incidence in canagliflozin intervention groups was similar to that in non-canagliflozin groups, and both were lower (113). A large population cohort study with a median of 2.6 years of follow-up showed that SGLT-2i utilization was not associated with increased overall breast cancer risk than DPP-4i (114). Interestingly, a study has implied that canagliflozin holds anti-proliferation effect on breast cancer cells by increasing phosphorylation of adenosine monophosphate activated protein kinase (AMPK) and reducing the phosphorylation of 70 kilodalton (kDa) ribosomal protein S6 kinase 1, thereby blocking cell cycle and inducing apoptosis (115). Besides, a study has manifested that ipragliflozin can inhibit the proliferation of breast cancer cells (116).
The strength of our study lies in the large study scale and sample size. However, our limitations are also obvious. Few studies we referred primarily aimed at assessing the risk of malignancy, and the incidence of malignant tumors closely related to age, diabetes, gender and many other factors. However, due to limited data availability, we can not adjust these parameters. In summary, it is not sufficient to discuss or conclude the relationship between SGLT-2i and malignancy risk in this meta-analysis. More RCTs related to long-term use of SGLT-2i are required in the future to provide more evidence for the safety of such drugs in long-term use.

CONCLUSION
In summary, current data from RCTs had no significant association between SGLT-2i and overall malignant tumor risk. Our evidence proposes that ertugliflozin may increase the overall risk of malignancy. Compared with active hypoglycemic agents, dapagliflozin may increase the overall risk of malignant tumor, while empagliflozin may reduce its risk. But compared with placebo, empagliflozin may increase the overall malignancy risk, mainly in the digestive system. However, the follow-up time of the RCTs analyzed in our study were relatively short, and the data of various factors were incomplete, which could insufficient to account for the long-term effects of SGLT-2i on malignant tumors, and more data are required for comprehensive analysis.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

AUTHOR CONTRIBUTIONS
FX designed the research process. NS and YShi searched the database for corresponding articles. JX and YSi extracted useful information from the articles above. TY and MZ used statistical software for analysis. NS and XL drafted the meta-analysis. DN polished this article. All authors had read and approved the manuscript and ensured that this was the case.