Utility of Progression Independent of Relapse Activity as a Trial Outcome in Relapsing-Remitting Multiple Sclerosis

Progression independent of relapse activity (PIRA) is increasingly used as a measure of disability worsening in multiple sclerosis (MS) and is believed to reflect the more chronic neurodegenerative aspect of MS. However, while conceptually appealing, PIRA and its counterpart relapse-associated worsening (RAW) have not been validated as outcome measures for clinical trials. Here, we study the co-occurrence of MRI activity in patients experiencing PIRA and RAW in a clinical trial setting. To illustrate the problem of random variation and measurement error of these new outcomes, we contrasted PIRA and RAW with similarly defined improvement.

We reanalyzed individual patient-level data of AFFIRM (NCT00027300) and SENTINEL (NCT00030966), 2 multicenter randomized controlled trials investigating natalizumab compared with interferon beta or placebo in RRMS, with trial visits occurring every 3 months for 2 years. We calculated 3-month-confirmed disability worsening (3M-CDW), RAW, and PIRA events based on worsening on the Expanded Disability Status Scale, 9-hole peg test, or timed 25-foot walk for every trial visit. We related worsening and improvement events to MRI activity throughout follow-up and contrasted worsening of disability with similarly defined improvement.

Our analysis included 2,113 participants, 42.4% of whom developed radiologic disease activity during follow-up. Only 8% of participants had a 3M-CDW event. Although the majority of those 3M-CDW events were PIRA (6.8%) and not RAW (0.9%), 42.2% of participants with PIRA had MRI activity in the first year of follow-up and 30.9% in the second. Improvement events exceeded PIRA events throughout follow-up and occurred in all trial arms. Finally, there was no difference in time-to-PIRA between participants with and without radiologic disease activity.

PIRA and RAW in their current definitions do not reliably distinguish between disability worsening due to inflammatory disease activity and neurodegeneration in RRMS. In addition, PIRA and RAW have similar and troubling issues of random variation and measurement error as currently used trial outcome measures. Our analysis requires confirmation in other clinical data sets; a meaningful next step would be to study the co-occurrence of PIRA with radiologic disease activity in a setting with more comprehensive MRI monitoring.