Identifying thresholds for meaningful improvements in NTDT-PRO scores to support conclusions about treatment benefit in clinical studies of patients with non-transfusion-dependent beta-thalassaemia: analysis of pooled data from a phase 2, double-blind, placebo-controlled, randomised trial.

Objectives: To estimate thresholds for defining meaningful within-patient improvement from baseline to weeks 13-24 and interpreting meaningfulness of between-group difference for the non-transfusion-dependent beta-thalassaemia patient-reported outcome (NTDT-PRO) tiredness/weakness (T/W) and shortness of breath (SoB) scores. A secondary objective was to determine the symptom severity threshold for the NTDT-PRO T/W domain to identify patients with symptomatic T/W.

Design: Pooled blinded data from the phase 2, double-blind, placebo-controlled, randomised BEYOND trial in NTDT (NCT03342404) were used. Anchor-based analyses supplemented with distribution-based analyses and empirical cumulative distribution function (eCDF) curves were applied. Distribution-based analyses and receiver operating characteristic curves were used to estimate between-group difference and symptomatic thresholds, respectively.

Setting: Greece, Italy, Lebanon, Thailand, the UK and the USA.

Participants: Adults (N=145; mean age 39.9 years) with NTDT who were transfusion-free ≥8 weeks before randomisation.

Measures: Score changes from baseline to weeks 13-24 in PROs used as anchors (correlation coefficient ≥0.3): NTDT-PRO T/W and SoB scores, Patient Global Impression of Severity, Functional Assessment of Chronic Illness Therapy-Fatigue (Fatigue Subscale, item HI12 and item An2) and Short Form Health Survey version 2.

Results: The eCDF curves support the use of estimates from the improvement by one level group for all anchors to determine the threshold(s) for meaningful within-patient improvement. Mean (median) changes from these groups and estimates from distribution-based analyses suggest that a ≥1-point reduction in the NTDT-PRO T/W or SoB domains represents a clinically meaningful improvement. Meaningful between-group difference threshold ranges were 0.53-1.10 for the T/W domain and 0.65-1.15 for the SoB domain. The optimal symptomatic threshold for the T/W domain (by maximum Youden's index) was ≥3 points.

Conclusions: The thresholds proposed may support the use of NTDT-PRO in assessing and interpreting treatment effects in clinical studies and identifying patients with NTDT in need of symptom relief.