The current methods used to estimate disability progression in clinical trials may be overestimating true disability by up to 30%, according to an MSBase analysis (Kalincik et al. Brain 2015;138(Pt 11):3287-3298).

MSBase evaluated different criteria used to estimate disability progression with data from 16,636 patients (112,584 patient-years). Progression rates varied widely according to the criteria used; the length of time used to confirm progression was the most important determinant of variance. The overall concordance rate of the different methods was low (17.3%).

For the most commonly used interval – EDSS change confirmed at three months – there was persistent disability in only 70% of patients at five years, indicating that about 30% of patients demonstrate “pseudoprogression”, and that their EDSS scores will improve during long-term follow-up. A longer confirmation period produced somewhat better results. The proportion of confirmed events that persisted over five years was 74%, 80% and 89% for disability progression confirmed at 6 months, 12 months and 24 months, respectively.

Overall, 11-34% of progression events were associated with regression of disability (i.e. EDSS score reduction) at five years. EDSS improvement after progression was more common in younger patients, those with smaller changes in disability scores, and with more frequent follow-ups.

The authors concluded that current estimates of confirmed disability progression in clinical trials could lead to spurious results, most notably in cohorts of younger patients with RRMS.

Comment
Dr. François Grand’Maison: This MSBase real-world observational study confirms earlier reports from randomized clinical trials (Rio et al. Ann Neurol 2006;59:344-352). One can only speculate about the multiple causes of pseudoprogression: inter- and intra-rater variability in the EDSS at low scores, assessments at different times of day, intercurrent conditions, variability in fatigue or pain, deconditioning-reconditioning, medication side effects and unrecognized relapses. A more intriguing possibility, for which there is no immunological or pathological evidence at this time, are true fluctuations in secondary progression.