Trajectory analysis of glucocorticoid treatment highlights issues in the current tapering strategy for polymyalgia rheumatica

Abstract

Objectives: To identify glucocorticoid (GC) treatment patterns in patients with polymyalgia rheumatica (PMR) and explore patient profiles that may benefit from GC-sparing interventions.

Methods: This descriptive study was conducted using an electronic medical record database in Japan. We identified patients with PMR aged ≥50 years who were initiated 5-<30 mg/day of GCs with increased inflammatory markers. Group-based trajectory modelling (GBTM) was used to characterise GC treatment patterns over 52 weeks. We analysed clinical characteristics, including changes in GC doses, longitudinal C-reactive protein levels, immunosuppressant use and GC-related toxicities.

Results: Among 452 eligible patients with PMR, four treatment trajectories were identified: rapidly-declining (19.0%), low-dose (36.9%), intermediate-dose (32.5%) and high-dose (11.5%). The rapidly declining and low-dose groups had more patients aged ≥80 years and with comorbidities. The median doses at week 52 in the low-dose, intermediate-dose and high-dose groups were 3.0, 4.0 and 7.5 mg/day, respectively. These groups had higher cumulative doses and greater GC-related toxicities compared with the rapidly declining group, which was reduced to 0 mg/day by week 8. The cumulative incidence of immunosuppressant use at week 52 was 6.1%-10.5%, even in the high-dose group.

Conclusions: GBTM analysis indicates that many patients who do not discontinue GC use within 1 year are exposed to high cumulative GC doses, which are associated with an elevated risk of GC-related toxicities. Our findings highlight the need to reconsider treatment strategies for patients with PMR, including the use of GC-sparing agents.

Keywords: Glucocorticoids; Polymyalgia Rheumatica; Treatment.

Figure 1. Glucocorticoid (GC) dose trajectories according to the group-based trajectory model. (A) Estimated values over time. (B) Observed median values over time. Detailed numerical data for this figure are provided in online supplemental table S8. (C) Cumulative GC dose over time.

Figure 2. Baseline characteristics of the study population. (A) Age. (B) C-reactive protein (CRP). Detailed numerical data for this figure are provided in online supplemental table S9. (C) Comorbidities. Detailed numerical data for this figure are provided in online supplemental table S7. The number of patients at baseline is as follows: (1) rapidly declining group (n=86); (2) low-dose group (n=167); (3) intermediate-dose group (n=147 patients); and (4) high-dose group (n=52).

Figure 3. Incidence rates of GC dose changes and the initiation of immunosuppressants over time. (A) Incidence rates (per person-year±95% confidence intervals) of GC dose increases or decreases. Detailed numerical data for this figure are provided in online supplemental table S10. (B) Cumulative incidence of the initiation of MTX. (C) Cumulative incidence of the initiation of other immunosuppressants, except for MTX. The inset shows the same data on an enlarged y axis. GC, glucocorticoid; MTX, methotrexate.

Figure 4. Cumulative incidence of glucocorticoid (GC)-related toxicities. (A) Heatmap of the 52-week cumulative incidence of GC-related toxicities. Values were normalised for each toxicity at week 52. (B–G) Cumulative incidence of the GC-related toxicities at weeks 4, 8, 12, 26 and 52. The original data of the survival time analysis are shown in online supplemental figure S5. (B) osteoporosis, (C) diabetes, (D) dyslipidaemia, (E) hypertension, (F) peptic ulcer, (G) serious infections. Data are expressed as percentages ±95% confidence intervals.