Patients with multiple sclerosis (MS) taking natalizumab are at risk for progressive multifocal leukoencephalopathy (PML). We sought to describe the outcomes of discontinuing natalizumab on the basis of PML risk and those of obtaining additional screening across a range of scenarios using decision tree models.
Health state probabilities and values, measured as the proportion of quality–adjusted life years (PQALY) relative to baseline health, were based on literature review. Probabilities of worsening MS while continuing and discontinuing natalizumab were set to 0.23 and 0.44. For discontinuing therapy, PML risk, worsening MS value, and PML value were varied. For additional screening, the probability of discontinuing natalizumab without screening, PML risk, worsening MS value, and PML value were set to 33%, 1.1%, 0.88, and 0.09, respectively, with test sensitivity and specificity varying from 0.50 to 1.
Discontinuing natalizumab provided no benefit until PML risk reached 2.9%, assuming an MS relapse value of 0.88 and a PML value of 0.09. Additional screening changed the PQALY by −0.3% to 1.5%, largely influenced by specificity. Assuming a sensitivity of 80% and a specificity of 99%, screening increases the PQALY by 1.2%.
The highest PML risk identified by stratification is below 2.9%, suggesting that continuing natalizumab outweighs PML risk for most patients on the basis of theoretical calculations. However, decisions based on additional screening with high-specificity tests, including polymerase chain reaction cerebrospinal fluid tests for John Cunningham virus, may provide benefit and should be clinically tested. Increased precision of probabilities and quality-of-life values are also needed to improve decision making.
Department of Public Health Sciences, University of Miami Miller School of Medicine, Miami, FL.
Conflicts of Interest and Source of Funding: The authors have no conflicts of interest to declare.
Supported by an investigator-initiated grant from Seedlings Life Sciences, LLC. The authors had full access to all data and independence with respect to any and all publication decisions.
Address correspondence and reprint requests to David Cooper Landy, PhD, MPH, Department of Public Health Sciences, University of Miami Miller School of Medicine, Miami, FL; E-mail: DCLandy@med.miami.edu