# GLP-1 receptor agonists or SGLT2-inhibitors? Evaluation of a personalized treatment algorithm for individuals with type 2 diabetes: a registry-based cohort study

## TL;DR

A personalized treatment algorithm for type 2 diabetes patients did not clearly favor GLP-1 receptor agonists or SGLT2-inhibitors in preventing cardiovascular events.

## Contribution

A real-world algorithm was developed and evaluated to personalize GLP-1-RA or SGLT2i treatment for T2D patients based on cardiovascular risk.

## Key findings

- The algorithm predicted 48% of patients would benefit more from GLP-1-RA and 52% from SGLT2i.
- GLP-1-RA-optimal patients had higher BMI and lower eGFR compared to SGLT2i-optimal patients.
- The predicted optimal treatment did not significantly delay non-fatal ASCVD events compared to suboptimal treatment.

## Abstract

Guidelines recommend GLP-1 receptor agonists (GLP-1-RA) and SGLT2-inhibitors
(SGLT2i) for individuals with type 2 diabetes (T2D) at high risk of
atherosclerotic cardiovascular disease (ASCVD). In the context of precision
medicine, we evaluated a personalized treatment algorithm to guide the
initial decision between these therapies.

Using data from the observational Diabetes Prospective Follow-up registry
(Germany/Austria) we studied individuals with T2D who initiated GLP-1-RA
(n=1433) or SGLT2i (n=2547) in a multicenter, real-world setting. Baseline
characteristics included age, sex, body mass index (BMI), estimated
glomerular filtration rate (eGFR), HbA1c, diabetes duration, and history of
ASCVD. Non-fatal ASCVD events (myocardial infarction, angina,
revascularization, stroke, transient ischemic attack, and peripheral artery
disease) were analyzed using dynamic weighted survival modeling to predict
the optimal treatment for each individual.

The algorithm predicted 48% of individuals to have better ASCVD outcomes with
GLP-1-RA and 52% with SGLT2i. GLP-1-RA-optimal individuals had on average a
higher BMI (37 vs 31 kg/m
2
), lower eGFR (71 vs 93 ml/min per 1.73
m
2
) and less history of ASCVD (9 vs 18%) compared to
SGLT2i-optimal individuals. However, an internal model validation showed
that the predicted optimal treatment did not statistically significantly
prolong the average time to a non-fatal ASCVD event compared to the
suboptimal treatment (AFT parameter: 1.13; 95% CI: 0.83-1.56; HR: 0.88; 95%
CI: 0.64-1.21).

The personalized treatment algorithm for GLP-1-RA and SGLT2i did not result
in clear individual ASCVD benefits on either drug, a finding consistent with
the clinical equipoise reflected in current T2D treatment guidelines.