SUMMARY  

• Several real-world analyses in type 2 diabetes mellitus (T2DM) patients found that risk of hospitalization for heart failure (HHF) was lower in patients treated with INVOKANA or other sodium glucose co-transporter 2 inhibitors (SGLT2i) compared to other antihyperglycemic agents (AHAs).^1-6
• A large database observational analysis found no difference in risk of HHF between new users of INVOKANA and new users of other SGLT2i in overall and established cardiovascular disease (CVD) populations.⁴
  • A statistically significant decrease in risk of HHF between new users of INVOKANA and all non-SGLT2i was reported in both the overall and established CVD populations.

REAL-world evidence

Observational Study

OBSERVE-4D, a large database observational analysis, utilized clinical characterization and population-level effect estimation to evaluate risks of HHF events associated with INVOKANA compared to other SGLT2i and non-SGLT2i in real clinical practice via a retrospective comparative cohort design. These risks were evaluated across all new users and in the subpopulation of patients with established CVD.⁴

Study Design

Four United States claims databases were included patients with a confirmed diagnosis of T2DM and first exposure to cohort-defining drug between 4/1/2013 and 5/15/2017.⁴

HHF was defined as all hospital admissions with a primary diagnosis of HF as identified by ICD-9-CM and ICD-10-CM codes.

The 6 exposure cohorts consisted of new users of:

1. INVOKANA
2. empagliflozin or dapagliflozin
3. empagliflozin
4. dapagliflozin
5. select non-SGLT2i (any dipeptidyl peptidase-4 inhibitor [DPP-4i], glucagon-like peptide-1 [GLP-1] receptor agonist, select other AHAs)
6. all non-SGLT2i (any DPP-4i, GLP-1 receptor agonist, thiazolidinediones [TZD], sulfonylureas [SU], insulin, select other AHAs)

Time at Risk Periods

• The on-treatment period evaluated risk of HHF during drug exposure, defined as 1 day after drug start date to the exposure end date (end of the first persistent period of exposure with 30-day gap between exposures).
• The intent-to-treat (ITT) period evaluated overall risk after starting drug therapy; defined as 1 day after start date to the end of the patient’s observation period.

Statistical Analysis

• Population-level effect estimation analyses were applied for seven pairwise comparisons. The baseline characteristics of populations varied greatly among the databases. A propensity score-adjusted new user cohort design was used to examined relative hazards of outcomes across all new users and the subpopulation with established CVD.
• A set of negative control outcomes was used to control for any potential systematic errors after propensity score adjustment.
• Propensity score adjustment was performed to reduce potential confounding due to imbalances between target and comparator cohorts with respect to baseline covariates.

Results

A total of 142,800 new users of INVOKANA (43,043 in the CVD subpopulation), 110,897 new users of other SGLT2i (31,011 in the CVD subpopulation), and 460,885 new users of all non-SGLT2i AHAs (141,579 in the CVD subpopulation) were evaluated.

The complete study results are available publicly through an interactive web-based application at: https://data.ohdsi.org/AhasHfBkleAmputation

• Of the HHF events that occurred, 75% were reported from the 30% of the overall population with established CVD.
• Meta-analytic results found no difference in risk of HHF in the overall and established CVD populations between new users of INVOKANA and new users of other SGLT2i, both during the on-treatment and ITT periods.
• Meta-analytic results found a statistically significant decrease in risk of HHF between new users of INVOKANA and all non-SGLT2i in the overall and established CVD populations. See Table: Meta-Analytic Estimate of the Risk of HHF in the On-Treatment Population.

• Sensitivity analyses did not find any differences in risk of HHF among SGLT2i. In the INVOKANA vs non-SGLT2i comparison, INVOKANA showed consistent decreased risk of HHF in both on-treatment and ITT populations; however, a larger decrease in risk was seen with the on-treatment population.
• The confirmatory findings of a reduction in risk of HHF with SGLT2i from this study are consistent with those from other real-world analyses of the SGLT2i class.^1,2,4,7,8

CVD-REAL

CVD-REAL, a real-world analysis conducted by Kosiborod et al (2017)¹ utilized health records from claims databases and registries across six countries to compare risk of HHF in T2DM patients (n=309,046) newly initiated on SGLT2i vs other glucose-lowering drugs (GLDs). Secondary endpoints compared the composite risk of HHF or all-cause death between the two treatment groups across 5 of the 6 countries (n=215,622).¹

Study Design

• A total of 309,056 new users receiving an SGLT2i or other GLD were identified after a 1:1 propensity match ratio was utilized to randomize patients.
• Baseline characteristics for propensity match cohorts were similar, including 87% who did not have known CVD and 13% who had established CVD.
• Among 154,528 patients who were initiated on SGLT2i in the HHF cohort, the proportion of exposure time was greatest with INVOKANA (52.7%) vs dapagliflozin (41.8%) vs empagliflozin (5.5%).
• Among 107,811 patients initiated on SGLT2i in the all-cause death or composite of HHF or all-cause death cohort across five countries, the proportion of exposure time with INVOKANA was 42.3%, 51% with dapagliflozin, and 6.7% with empagliflozin in the allcause death cohort and 45%, 49%, and 6%, respectively, in the composite cohort.
  • Among US patients initiated on SGLT2i in the all-cause death cohort and composite cohort, the proportion of exposure time was greatest with INVOKANA (75.4%) vs dapagliflozin (19.3%) vs empagliflozin (5.3%).

Results

• Compared to other GLDs, SGLT2i’s resulted in a significantly lower risk of HHF (hazard ratio [HR]: 0.61; 95% confidence interval [CI]: 0.51-0.73; P<0.001).
• Treatment with SGLT2i resulted in a lower risk for the composite endpoint of HHF or allcause death (HR: 0.54, 95% CI: 0.48-0.60; P<0.001) as compared with other GLDs.

CVD-REAL2

Kosiborod et al (2018)⁵ conducted another real-world analysis (CVD-REAL2) in geographical areas not studied in CVD-REAL, including South Korea, Japan, Singapore, Israel, Australia, and Canada.

• A total of 447,106 new users receiving an SGLT2i or other GLD were identified after a 1:1 propensity match ratio was utilized to randomize patients.
• Baseline characteristics for propensity match cohorts were similar, including 74% who did not have known CVD and 26% who had established CVD.
• As compared to treatment with other GLDs, SGLT2i’s resulted in a significantly lower risk of HHF (HR: 0.64; 95% CI: 0.50-0.82; P=0.001).
• Treatment with SGLT2i resulted in a lower risk for the composite endpoint of HHF or allcause death (HR: 0.60; 95% CI: 0.47-0.76; P<0.001) as compared with other GLDs.

EASEL Study

Udell et al (2017)² conducted a retrospective, population-based, event-driven, cohort study within the US Department of Defense (DoD) Military Health System to evaluate CV benefits in T2DM patients with established CVD (including HF) newly exposed to SGLT2i, compared to new users of non-SGLT2i AHAs from April 1, 2013 to December 31, 2016.

Study Design

• The primary outcome was the composite of all-cause mortality (ACM) and HHF. Other endpoints included a composite of MACE plus HHF, as well as HHF alone.
• A total of 25,258 patients were matched 1:1 (12,629 new users of SGLT2i with 12,629 new users of a non-SGLT2i AHA.
  • Of patients initiated on SGLT2i, 7,333 (58.1%) patients started INVOKANA, 3,341 (26.4%) empagliflozin, and 1,955 (15.5%) dapagliflozin. Baseline characteristics were well-balanced.

Results

• Incidence rate of ACM and HHF was 1.73 vs 3.01 per 100 patient-years (PY) among patients newly initiated on SGLT2i and non-SGLT2i, respectively (HR: 0.57; 95% CI: 0.50-0.65; P<0.0001).
• Initiation of an SGLT2i was associated with a lower rate of HHF (0.51 vs 0.90 events per 100 PY; HR: 0.57; 95% CI: 0.45-0.73; P<0.0001).
• For the composite endpoint of MACE and HHF, the rate was significantly lower among patients newly initiated on an SGLT2i compared with non-SGLT2i (2.72 vs 4.11 per 100 PY; HR: 0.66; 95% CI: 0.60-0.74; P<0.0001).

Other Real-World Analyses

Patorno et al (2017)³ conducted a retrospective real-world analysis of CV safety of INVOKANA compared to non-SGLT2i using data from a large US commercial healthcare dataset from April 2013-September 2015. The study used three pairwise 1:1 propensity score-matched cohorts of patients.

• Investigators examined HHF associated with initiation of INVOKANA vs non-SGLT2i (DDP4i, GLP-1 RA, and SU) in adult T2DM patients.
• Primary outcome included HHF.
• The investigators identified three pairwise 1:1 propensity score-matched cohorts of patients: (17,667 pairs initiated on INVOKANA or a DDP4i), (20,539 pairs initiated on INVOKANA or a GLP-1 RA), (17,354 pairs initiated on INVOKANA or a SU)
• Risk of HHF was lower in patients treated with INVOKANA vs other AHAs; INVOKANA vs DPP4i (8.9 vs 12.8 per 1000 PY; HR: 0.70; 95% CI: 0.54-0.92), INVOKANA vs GLP-1 RA (7.5 vs 12.4 per 1000 PY; HR: 0.61; 95% CI: 0.47-0.78), and INVOKANA vs SUs (7.3 vs 14.4 per 1000 PY; HR: 0.51; 95% CI: 0.38-0.67).

Pilon et al (2020)⁶ conducted a retrospective cohort study to compare HHF events in patients with T2DM and macroalbuminuria initiated on INVOKANA vs DPP-4i in a CREDENCE-like population using real-world claims data from Optum’s Clinformatics™ Data Mart database between March 2012 and March 2019.

• Inclusion criteria were as follows: adults with a T2DM diagnosis and a urine albumin-to-creatinine ratio >300 mg/g or albumin excretion rate >300 mg/day; initiated on INVOKANA or DPP-4i, with ≥6 months of pre-initiation insurance eligibility. Patients were excluded if they notably had a diagnosis of type 1 diabetes mellitus, acute kidney injury, or stage 5 chronic kidney disease at baseline.
• The observation period was censored at the earliest of the index drug discontinuation, a claim for a drug from the alternate index cohort (DPP-4i or INVOKANA), or the end of eligibility or data.
• Inverse probability of treatment weighting (IPTW) accounted for differences in baseline demographics, clinical characteristics, and healthcare costs.
• HHF events were defined as inpatient stays with a diagnosis of heart failure (ICD-9: 428.x; ICD-10: I50.x). Rates of HHF events per 100 PY were calculated and weighted Cox proportional hazards model accounting for repeated measurement was used to estimate the HR, 95% CI, and p-value.
• A total of 669 patients initiated on INVOKANA and 1870 patients initiated on a DPP-4i were identified.
• After IPTW balanced baseline characteristics, mean age was ~64 years old, ~40% of patients were female, and mean HbA1c was ~9.0%.
• Among INVOKANA and DPP-4i patients, respectively, the mean observation period was 6.6 and 8.6 months and the rate of HHF events per 100 PY was 1.16 and 4.40.
• At 12 months, the hazard for HHF was 70% lower among patients initiated on INVOKANA relative to patients initiated on DPP-4i (HR [95% CI]: 0.30 [0.09-0.94]; P=0.0396).

Literature Search

A literature search of MEDLINE^® (and/or other resources, including internal/external databases) pertaining to this topic was conducted through 31 January 2024.