AUTHORS
Naseema Gangat, MBBS, Aysel Bayraktar2, Buraq Ahmed3, Merry Nakhleh4, Amina Akram, MBBS5, Sree Maneesha Sunkara, MBBS6, Lauren Fang, MD7, Animesh Pardanani, PhD, MBBS8, Ayalew Tefferi, MD9
AFFILIATIONS
1 Mayo Clinic, Rochester, United States
2 Mayo Clinic, Rochester, United States
3 Cambridge University, Cambridge, United Kingdom
4 Mayo Clinic, Rochester, United States
5 Mayo Clinic, Rochester, United States
6 Mayo Clinic, Rochester, United States
7 Mayo Clinic, Rochester, United States
8 Mayo Clinic, Rochester, United States
9 Mayo Clinic, Rochester, United States
ABSTRACT
Background: GLP-1 receptor agonists (GLP-1RA), approved for the treatment of type 2 diabetes mellitus and obesity have anti-inflammatory and antithrombotic properties. Emerging registry based studies suggest improved outcomes in patients with myeloproliferative neoplasms receiving GLP-1RA therapy (Chang et al., ISTH 2025; Cheema et al., Clin Lymphoma Myeloma Leuk., 2025). In the current study, we examined the association of GLP-1RA therapy with overall survival (OS) and incidence of thrombosis in patients with essential thrombocythemia (ET), with analyses accounting for immortal time bias.
Methods: A Mayo Clinic database of patients with ET was queried to identify GLP-1RA recipients. OS was analyzed using Kaplan–Meier and multivariable Cox regression, with GLP-1RA exposure modeled as a time-dependent covariate. Among GLP-1RA recipients, thrombotic events were divided into pre-exposure, exposure and post-discontinuation. Incidence rates were calculated as events per 100 person-years, and incidence rate ratios (IRRs) were used to compare thrombotic risk between exposure periods and with GLP-1RA non-recipients. Arterial and venous thrombosis were analyzed using competing-risk methods with death as the competing event.
Results:
Comparison of GLP-1 users and non-users
Among 1,340 patients with ET, 104 (8%) received GLP-1RA between October 2006 and April 2026. Treatment was initiated before ET diagnosis in 15 (14%) patients and during chronic-phase ET in 84 (80%), at a median of 82 months (range, 1–426) after diagnosis. Median treatment duration was 25 months (range, 1–204). Semaglutide (n=59, 57%), was the most frequently prescribed agent, followed by tirzepatide (n=32, 30%), dulaglutide (n=11, 10%), liraglutide (n=7, 7%), and exenatide (n=4, 4%).
Compared with non-users, GLP-1RA recipients were younger (median age, 52 vs 59 years; p<0.01), more likely to have diabetes (39% vs 8%; p<0.01) and hypertension (60% vs 47%; p=0.02), and displayed lower platelet counts (median, 671 vs 750 ×109/L; p<0.01). Prior arterial thrombosis was more frequent among GLP-1RA recipients (24% vs 13%; p<0.01), while prior venous thrombosis rates were similar (16% vs 11%; p=0.11). Sex distribution (female, 72% vs 66%; p=0.18) and driver mutation frequencies (p=0.44) were comparable between groups.
GLP-1RA recipients were more frequently classified as low/intermediate-1 risk by Triple AAA+ (71% vs 61%; p<0.01). By contrast, revised IPSET-thrombosis risk distribution was similar between groups (p=0.18). Aspirin use at diagnosis was comparable (85% vs 80%; p=0.20), whereas GLP-1RA recipients were less likely to receive cytoreductive therapy (49% vs 61%; p=0.02).
Overall Survival and incidence of thrombosis
After a median follow-up of 7.8 years (range; 0.01-52.6), 6 (6%) GLP-1RA recipients and 339 (27%) non-users had died. Median OS was not reached among GLP-1RA recipients compared with 20 years in non-users (10-year OS, 96% vs 80%; p<0.01). This association remained significant after adjustment for IPSET-survival (p=0.01) and Triple AAA+ model (p=0.02). However, after accounting for immortal time bias, GLP-1RA use was not associated with improved OS (p=0.52), including among patients with diabetes mellitus (p=0.67).
Arterial thrombosis occurred in 22 (21%) GLP-1RA recipients and 170 (14%) non-users. GLP-1RA use was associated with a higher cumulative incidence of arterial thrombosis (20 year cumulative incidence: 40.9%vs 20.6%; p=0.05) including after analysis adjusted for revised IPSET thrombosis (p=0.09), and among patients with diabetes (p=0.02).
Venous thrombosis occurred in 7 (7%) GLP-1RA recipients and 122 (10%) non-users, and cumulative incidence of venous thrombosis in GLP-1RA users and non-users was comparable (20 year cumulative incidence: 24.7%vs 16.6%; p=0.36), and among diabetics (p=0.98).
Among GLP-1RA recipients, incidence of thrombosis after ET diagnosis was higher during GLP-1RA treatment compared to before treatment (5.4 vs 1.9 per 100 person-years; IRR, 2.91; 95% CI, 1.39–6.12; p=0.005) and 3.9 per 100 person-years after treatment discontinuation, compared to non-users (5.4 vs 2.6 per 100 person-years; IRR, 2.10; 95% CI, 1.21–3.66; p=0.009), mainly due to arterial events (11 arterial vs 2 venous events).
Conclusion: GLP-1RA use was not associated with improved survival outcomes in patients with ET, after accounting for immortal time bias. The observed association with arterial thrombosis requires confirmation in prospective studies.
DISCLOSURES
Merry Nakhleh: No relevant conflicts of interest to declare.