Frailty phenotypes and determinants of glucocorticoid-related adverse events in polymyalgia rheumatica at 12 months: a real-life single-center investigation.

BACKGROUND: Polymyalgia rheumatica (PMR) is a clinically heterogeneous disease with variable trajectories. Although glucocorticoids (GCs) are effective, prolonged exposure carries significant toxicity risks. This study was aimed at exploring whether baseline frailty phenotypes (stratified by age and comorbidity burden) and treatment exposure influenced relapse patterns and the occurrence of glucocorticoid-related adverse events (GC-related AEs).

METHODS: Fifty-eight patients with isolated PMR were retrospectively analyzed over a 12-month follow-up. Unsupervised hierarchical clustering incorporating baseline and follow-up data was performed to identify distinct longitudinal clinical trajectories. Longitudinal outcomes related to disease activity (remission/relapses) were assessed using generalized estimating equation models. Safety dynamics were evaluated, determining the cumulative prednisone dosage at the first adverse event and using multivariable Cox proportional hazards regression to identify independent predictors of GC-related AEs, accounting for time-varying exposure patterns.

RESULTS: Two clusters were identified: Cluster 1 ("frail": older, multimorbid, n = 22) and Cluster 2 ("robust": younger, fewer comorbidities, n = 36). Cluster membership was not associated with longitudinal disease activity (p = 0.129); relapse risk was instead associated with markers of treatment intensification, including DMARD requirement (OR 2.94, 95% CI 1.13-7.65, p = 0.028). GC-related AEs occurred in 40% of patients (95% CI 28-53%). In the multivariable Cox model, comorbidity burden was a significant independent predictor of GC-related AEs (hazard ratio [HR] 1.27, p = 0.044), while the "robust" Cluster 2 showed a trend towards reduced risk (HR 0.45, p = 0.09). Interestingly, the cumulative prednisone exposure at the first adverse event was similar between "frail" and "robust" patients (median 1112 mg vs 1287.5 mg; p = 0.86), suggesting a consistent dose-dependent pattern.

CONCLUSIONS: Relapse trajectories in PMR appear to be linked to intrinsic disease activity rather than baseline frailty alone. Safety analysis suggests a cumulative GC dose range (approximately 1.2 g of prednisone) across both phenotypes. However, multimorbidity may act as an additive risk factor, reducing the physiological safety margin. These preliminary findings suggest that early steroid-sparing strategies might be considered in multimorbid patients to prevent them from reaching this cumulative dose range. Key Points • Baseline clinical phenotypes defined by age and comorbidities identified distinct "frail" and "robust" clusters where, despite a similar longitudinal course of disease activity, the frail phenotype exhibited a significantly higher rate of glucocorticoid-related adverse events (59% vs. 28%). • A cumulative prednisone dose range (approximately 1.2 g) was consistently associated with the onset of adverse events across the cohort; however, multivariate time-to-event analysis identified multimorbidity as a significant independent predictor of GC-related AEs (HR 1.27, p = 0.044), suggesting a reduced physiological safety margin in frail patients. • As relapse risk appeared linked to markers of treatment intensification rather than baseline frailty, these preliminary findings suggest that early steroid-sparing strategies might be considered in multimorbid patients to prevent them from reaching hazardous cumulative glucocorticoid dosages.