500 - Risk of venous thromboembolism and thrombotic stroke in adolescents and premenopausal young women with chronic kidney disease on estrogen and progesterone therapy

Publication Number: 2486.500

Netanya Pollock, Childrens Hospital of Philadelphia, Philadelphia, PA, United States; Jarcy Zee, Childrens Hospital of Philadelphia, Philadelphia, PA, United States; Jordana Cohen, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA, United States; Lawrence Copelovitch, The Children's Hospital of Philadelphia, Philadelphia, PA, United States; Sogol Mostoufi-Moab, Children's Hospital of Philadelphia, Philadelphia, PA, United States; Michelle Denburg, Children's Hospital of Philadelphia, Philadelphia, PA, United States

Background: Chronic kidney disease (CKD) is associated with menstrual irregularities and hypoestrogenism. Combined estrogen and progesterone therapy, prescribed for menstrual disorders and contraception, increases the risk of venous thromboembolism (VTE) and thrombotic stroke in the general population. Women with CKD may already have an elevated baseline thrombotic risk. Data on safety of estrogen and progesterone in adolescents and young women with CKD is lacking.

Objective: To evaluate the association between estrogen/progesterone and the risk of VTE or thrombotic stroke in adolescents and young women with moderate to severe CKD using a real-world database.

Design/Methods: Using electronic health record data from the TriNetX Research Network, we identified females aged 12-40 years with stage 3-5D CKD and no prior estrogen/progesterone use. Eligible patients required >=180 days of follow up after cohort entry (CKD diagnosis). Patients with prior VTE, thrombotic stroke, inherited thrombophilia, or malignancy were excluded. Propensity score matching (1:m, m<=5) was used to balance baseline characteristics. A pooled logistic model was used to approximate a Cox model and estimate hazard ratios (HRs) and 95% confidence intervals (CIs). Robust standard errors accounted for clustering within matched sets and within individuals across matched sets, and inverse probability of censoring weights accounted for treatment switch. Analyses were performed in Stata 18.0.

Results: A total of 29,083 patients met study inclusion criteria. Among these, 1,339 estrogen/progesterone users were matched to non-users with replacement, yielding 2,637 unique patients for analysis. Baseline characteristics were well balanced post-matching (Table 1). During a median follow-up period of 8.2 years (IQR 0.75 - 25.5), 98 VTE events occurred, corresponding to crude incidence rates (IR) of 5.1 (95% CI 3.4 - 7.8) and 7.9 (95% CI 6.2 - 10.2) per 1,000 person-years (PY) among hormone users and non-users, respectively. The adjusted HR for VTE comparing users with non-users was 0.76 (95% CI 0.48 - 1.19) (Table 2). Sixty-four thrombotic stroke events occurred, with crude IR of 3.6 (2.2 - 6.3) and 5.5 (4.1 - 7.6) per 1,000 PY among users and non-users, respectively. The adjusted HR for thrombotic stroke was 0.70 (95% CI 0.4 - 1.2).

Conclusion(s): In this real-world study of adolescents and young women with CKD, estrogen/progesterone therapy was not associated with increased risk of VTE or thrombotic stroke. Future studies should account for indication and underlying CKD diagnosis to further define the thrombotic safety profile in this high-risk population.

Baseline Cohort Characteristics Before and After Propensity Score Matching
Abbreviations: E/P = estrogen/progesterone; CKD = chronic kidney disease; ESKD = end stage kidney disease; HD = hemodialysis; PD = peritoneal dialysis; ASMD = absolute standardized mean difference
Notes: Unmatched columns show unweighted characteristics. Matched columns show weighted summary statistics from 1:m (m<= 5) propensity score matching with replacement (1,139 matched sets, 2,637 unique patients). Covariate balance defined as ASMD < 0.1. ASMD are weighted by the matching weights.

Incidence Rates and Adjusted Hazard Ratios (95% CI) of Thrombotic Events
Abbreviations: Ref = reference group (non-users); CI = confidence interval; HR = hazard ratio
Notes: HR approximated by pooled logistic regression under the rare outcome assumption.