Abstract Rationale In routine care, gastrointestinal adverse events and liver test abnormalities with nintedanib often prompt dose reduction, interruption, or early discontinuation; in the randomized INBUILD program, adverse events led to discontinuation in ∼22% and diarrhea was the most frequent event. Whether starting at 100 mg twice daily (BID) rather than 150 mg BID improves persistence without sacrificing effectiveness is unknown. Methods We emulated a single-center, retrospective target trial comparing low-dose (100 mg BID) vs standard-dose (150 mg BID) initiation among adults with idiopathic pulmonary fibrosis (IPF) or progressive fibrosing interstitial lung disease (PF-ILD) who started nintedanib between September 2015 and June 2025. “Time zero” was drug initiation. The primary estimand was restricted mean time on treatment (RMST) up to 12 months, where “on-treatment” ended at discontinuation or death; loss to follow-up was censored. Confounding was addressed using stabilized inverse-probability-of-treatment weighting (IPTW) with 1st-99th percentile trimming; cumulative incidence functions (CIFs) treated death as a competing event. Among 12-month survivors, the average treatment effect on ΔFVC (FVC %pred at ∼12 months minus baseline) was estimated using multiple imputation with augmented IPTW; complete-case AIPW was a sensitivity analysis. Prespecified longer-horizon RMST and reason-specific competing-risk analyses supported the primary estimand. Results Of 193 screened, 172 initiators met eligibility (low-dose 94; standard-dose 78). Post-weighting covariate balance was good (most weighted SMDs ≤0.10; FVC and DLCO ≈0.12). Low-dose initiation increased 12-month RMST by 52.9 days (95% CI 11.8-97.1). The IPTW-adjusted CIF for discontinuation at 12 months was 0.281 (standard-dose) vs 0.135 (low-dose), risk difference (RD) 0.146 (95% CI − 0.003 to 0.299); RD rose to 0.206 (0.043-0.376) at 24 months and 0.282 (0.051-0.501) at 36 months. Within 12 months, discontinuations and deaths were 18 and 9 (standard) vs 14 and 7 (low-dose). Supportive hazard models showed higher discontinuation under standard-dose (cause-specific HR 2.48, 95% CI 1.11-5.58; subdistribution HR 2.96, 95% CI 1.33-6.58). ΔFVC (standard-minus-low) was −0.04 %pred (95% CI − 2.78 to 2.71); complete-case −0.03 %pred (95% CI − 5.01 to 4.82; n = 99). Reasons for discontinuation were most often decreased appetite and diarrhea; counts were small and reason-specific RDs were descriptive. Conclusions In this target-trial emulation, initiating nintedanib at 100 mg BID improved treatment persistence over 12 months and beyond without a detectable one-year decrement in lung-function trajectory. A “start low, escalate as tolerated” policy appears pragmatic for patients at risk of early intolerance, warranting multisite validation. This abstract is funded by: None
Kaburaki et al. (Fri,) studied this question.