Statistics from Altmetric.com
Progressive multifocal leucoencephalopathy (PML), a devastating demyelinating disease caused by John Cunningham virus (JCV) reactivation, affects patients with severe immune deficiencies.1 Despite having tested numerous molecules over the last 25 years, their benefits have often been overstated based on isolated case reports or small series.1 Because those therapies offered patients with PML some, although limited, hope, the desire to try an inaccurately tested but potentially lifesaving treatment surpasses rigorous clinical trial evaluation before widespread use. Notably, cidofovir’s alleged promise, based on case reports, largely evaporated when prospective cohort studies demonstrated that it had no influence on AIDS-related PML (AR-PML) overall survival (OS) or residual disability.2 That story might be repeating itself with maraviroc (MVC) use, also based on only a few favourable case reports.3
Restoring JCV-specific immune responses is the most effective way to improve survival of patients with PML.1 Immune recovery is not always beneficial and a subset of patients with PML may worsen due to immune reconstitution inflammatory syndrome (PML-IRIS), mediated by CD8+ T cells strongly expressing CC chemokine receptor-5 (CCR5). MVC is a non-competitive CCR5 antagonist approved in the HIV armamentarium. Histological findings and some case reports inferred that MVC might work as a mediating double-edged sword by enhancing CD4+ T-cell reconstitution and its potential immunomodulatory properties to improve AR-PML and AR-PML-IRIS survival in various conditions (idiopathic CD4+ T-cell lymphocytopaenia, sarcoidosis and natalizumab-treated patients with multiple sclerosis).4 But, MVC had no beneficial effect on OS of patients with PML in one-third of reports (online supplementary table).
Supplementary file 1
Our objective was to draw lessons from repeatedly raised false hopes of unconfirmed PML treatments that might contribute to prescribing ineffective drugs for years based on claimed efficacy in case reports or small series and by failing to respect the need for clinical trial evaluation of new molecules before authorising their widespread use.1
Charts were reviewed retrospectively for all consecutive patients with definite AR-PML, newly diagnosed between January 2008 (when combined antiretroviral treatment (cART) including MVC started in France) and December 2015, followed in four University Hospitals’ Infectious Diseases Departments. Definite PML diagnosis was complied with the validated consensus criteria.5 The date of histological results or cerebrospinal fluid (CSF) JCV-PCR–positivity dated AR-PML diagnosis. Patients were classified according to MVC exposure (MVC+ vs MVC−) as part of cART. PML-IRIS was defined as neurological deterioration (≥1 point increase of the modified Rankin disability scale ‒MRDS), temporally associated with cART initiation or optimisation and subsequent plasma HIV-load decrease of ≥1 log10 copies/mL.
Study endpoint was OS, calculated from the AR-PML diagnosis date to that of death or censored on 31 December 2015 for survivors by using the Kaplan-Meier estimator. OS was analysed using a Cox model with MVC exposure (as a time-dependent variable) to overcome a potential immortal time bias and IRIS onset as a dichotomous one (yes vs no).
The Rothschild Foundation’s Institutional Review Board approved this study. All participants provided written informed consent.
We identified 34 patients with definite AR-PML among our cohort of ~14 000 HIV-infected individuals, among whom 17 (50%) were MVC+ as part of cART. MVC+ and MVC– patients, respectively, were similar for mean (IQR) time since HIV diagnosis (7.9 (0–14) vs 5.2 (0–10) years; P=0.38), median (IQR) age (46 (36–50) vs 45 (40–50) years; P=0.31), CD4 nadir (45 (15–69) vs 50 (20–87)/mm3; P=0.84), CD4 count (73 (20–160) vs 59 (19–97)/mm3, P=0.59), transmission (P=0.99), median MRDS at PML onset (3 (2–3) vs 3, (2–4) P=0.24) and radiological localisation (P=0.71). Six patients had PML-positive brain biopsies (3 in each group) and 28 CSF-JCV–positive PCRs (14 in each group). cART was highly effective, with HIV-load decreases ≥1 log10 copies/mL within 1 month for 25/34 (74%) patients with AR-PML.
PML-IRIS occurred similarly in MVC+ (n=11) and MVC– patients (n=10), with a median HIV-load decrease of –3.5 (range: –1.3 to –5.6) log10 copies/mL at IRIS diagnosis.
Sixteen (47%) patients with AR-PML died, 8 MVC+ and 8 MVC–, respectively: after a median (IQR) 4.3 (2.7–6.0) vs 3.1 (2.2–11.1) months, with all but one deaths directly attributed to PML. Survivors’ median follow-up lasted 39.3 (20.3–73.2) months.
Figure 1 reports Kaplan-Meier estimates of MVC+ and MVC– patients’ OS. For all and MVC+ and MVC– patients with AR-PML, respectively, median 1-year OS was 58.3% (43.8–77.7), 52.3% (33.8–82.8) and 64.7% (45.5–91.9); and 5 year OS 49.9% (34.8–71.6), 52.3% (33.8–82.8) and 45.3% (24.4–84.2). Univariate analyses showed no difference between OS of MVC-treated and MVC-untreated patients (HR for death for MVC+ patients 1.09, 95% CI 0.41 to 2.91; P=0.87). According to multivariable analyses adjusted for IRIS, MVC was not associated with improved AR-PML OS (adjusted HR 0.99, 95% CI 0.34 to 2.90; P=0.98).
Our results indicated no clinically relevant benefit of MVC as part of cART on OS of patients with AR-PML.
No clinical evidence supports the hypothesis that MVC has a real impact on AR-PML and AR-PML-IRIS outcomes. Indeed, one-third of the 17 published cases had negative outcomes (online supplementary table). Because some patients with PML and PML-IRIS could have evolved favourably, regardless of treatment, isolated case reports bias the literature by over-reporting and overciting positive and under-reporting negative results. That context provides a false sense of hope, which in turn drives a ‘quotation bias’ that becomes a major confounding effect for the molecule’s real efficacy.
Our study’s main strength is the accurate control of MVC exposure to avoid immortal time bias. The retrospective design and small sample size mean our findings cannot definitively exclude a potential MVC beneficial effect on AR-PML survival. Because all our patients are HIV-positive, we cannot extrapolate our results to individuals who are HIV-negative. However, histological features of natalizumab-induced PML-IRIS and AR-PML-IRIS were very similar.1
We would like to temper the MVC promise that has not yet been validated, to avoid repeating the previous cidofovir and other unconfirmed candidate ‘false-hope syndromes’. So as not to miss opportunities to show that MVC is a viable PML therapy, our observations should be considered preliminary findings to build a well-designed prospective, randomised, double-blind multicentre trial, like the one currently testing interleukin-7.1 2
The authors thank Janet Jacobson for editorial assistance and Gilbert Lesage for bibliographic assistance.
Contributors EJ: performed medical chart research and data acquisition, analysed the data and contributed to the writing of the manuscript. GM-B: contributed to study concept, acquisition and interpretation of the data, and writing of the manuscript. CL and F-XL: contributed to interpretation of the data and writing of the manuscript. HP: performed data analysis and interpretation, and critical revision of the manuscript. GP, CK and IC: contributed to interpretation of the data and critical revision of the manuscript. AM: was responsible for study concept, overall supervision of interpretation of the data and writing of the manuscript.
Funding GM-B has received lecture fees from Abbvie,Pfizer and ViiV Healthcare France, and travel and accommodations funding from Eumedica France. CL reports lecture fees and travel and accommodations funding from Allergan and Thea, and lecture fees from Alcon. GP reports grants from Gilead and BMS, lecture fees from Abbvie Viiv Healthcare, GSK, Janssen, MSD, and Nephrotek. CK reports grants and lecture fees from Merck-Sharp-Dome, Bristol-Myers-Squibb, Viiv Healthcare and Gilead Sciences, and grants from Janssen Pharmaceuticals. F-XL reports lecture fees from MSD, GILEAD and Biomerieux, travel and accommodations funding from Genmark and for meeting travel from Eumedica and Pfizer. AM reports lecture fees and travel and accommodations funding from Biogen Idec, Novartis and MSD, and travel and accommodations funding from Teva and Merck-Serono.
Competing interests None declared.
Patient consent Obtained.
Ethics approval The Rothschild Foundation’s Institutional Review Board.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.