METHODS: MEDLINE, EMBASE and CENTRAL were searched from inception to 13 July 2021 for randomised controlled trials comparing second-line glucose lowering therapies with placebo, standard care or one another. Primary outcomes included cardiovascular and renal outcomes. Secondary outcomes were non-cardiovascular adverse events. Risk ratios (RRs) and corresponding confidence intervals (CI) or credible intervals (CrI) were reported within pairwise and network meta-analysis. The quality of evidence was evaluated using the GRADE (Grading of Recommendations, Assessment, Development and Evaluation) criteria. Number needed to treat (NNT) and number needed (NNH) to harm were calculated at 5 years using incidence rates and RRs. PROSPERO (CRD42020168322).
RESULTS: We included 38 trials from seven classes of glucose-lowering therapies. Both sodium-glucose co-transporter-2 inhibitors (SGLT2i) and glucagon-like peptide 1 receptor agonists (GLP1RA) showed moderate to high certainty in reducing risk of 3-point major adverse cardiovascular events, 3P-MACE (network estimates: SGLT2i [RR 0.90; 95% CrI 0.84-0.96; NNT, 59], GLP1RA [RR 0.88; 95% CrI 0.83-0.93; NNT, 50]), cardiovascular death, all-cause mortality, renal composite outcome and macroalbuminuria. SGLT2i also showed high certainty in reducing risk of hospitalization for heart failure (hHF), ESRD, acute kidney injury, doubling in serum creatinine and decline in eGFR. GLP1RA were associated with lower risk of stroke (high certainty) while glitazone use was associated with an increased risk of hHF (very low certainty). The risk of developing ESRD was lower with the use of sulphonylureas (low certainty). For adverse events, sulphonylureas and insulin were associated with increased hypoglycaemic events (very low to low certainty), while GLP1RA increased the risk of gastrointestinal side effects leading to treatment discontinuation (low certainty). DPP-4i increased risk of acute pancreatitis (low certainty). SGLT2i were associated with increased risk of genital infection, volume depletion (high certainty), amputation and ketoacidosis (moderate certainty). Risk of fracture was increased with the use of glitazones (moderate certainty).
CONCLUSIONS: SGLT2i and GLP1RA were associated with lower risk for different cardiorenal end points, when used as an adjunct to metformin in people with type 2 diabetes. Additionally, SGLT2i demonstrated benefits in reducing risk for surrogate end points in kidney disease progression. Safety outcomes differ among the available pharmacotherapies.