The trials were conducted in accordance with the Good Clinical Practice guidelines of the International Council for Harmonisation and the ethical principles outlined in the Declaration of Helsinki. The protocols of VALOR and its open-label extension were approved by relevant ethics committees. An independent data monitoring committee reviewed safety data. Written informed consent for both the randomized phase and open-label extension was provided by participants or their legal representatives.
The sponsor, Biogen, and the authors designed these trials. Biogen provided tofersen and placebo, oversaw the trial, performed the statistical analyses, and paid for medical writing assistance. Biogen and the authors analyzed the data. The first draft of the manuscript was written by the first author and an author employed by Biogen. The sponsor reviewed the manuscript but could not delay or prevent publication of the results. The authors reviewed and approved revisions of the manuscript and vouch for the accuracy and completeness of the data, the fidelity of the trials to the protocols, and the accuracy of the reporting of adverse events. There were confidentiality agreements between the authors and Biogen.
The phase 3, double-blind, randomized, placebo-controlled VALOR component of the trial was conducted from March 2019 through July 2021.16 Participants were enrolled at 32 sites in 10 countries (see Section S1 in the Supplementary Appendix, available at NEJM.org). The trial included a 4-week screening period, a 24-week treatment period, and a follow-up period of 4 to 8 weeks followed by an ongoing extension phase.
Participants were randomly assigned in a 2:1 ratio to receive an intrathecal bolus injection through a lumbar puncture of a 15-ml solution of tofersen (100 mg) or an equivalent volume of placebo (artificial cerebrospinal fluid [CSF]) administered over a period of 24 weeks, as three doses once every 2 weeks, followed by five doses once every 4 weeks (Fig. S1 in the Supplementary Appendix). Randomization was stratified according to the use or nonuse of edaravone, riluzole, or both at baseline and according to whether participants met prognostic criteria for faster disease progression that were based on SOD1 mutation type and the estimated slope of the score on the Amyotrophic Lateral Sclerosis Functional Rating Scale–Revised (ALSFRS-R), calculated from the time of symptom onset until screening (“prerandomization ALSFRS-R slope”). Owing to the potential for nonlinear progression on the ALSFRS-R score and for intra-mutation variability confounding the prognostic value of these measures, as well as literature supporting the use of neurofilament light chains as a prognostic marker of disease progression,16-24 analyses in subgroups that were defined according to baseline concentrations of neurofilament light chains in plasma (above vs. below the median concentration for the trial population) were prespecified before VALOR results were available (see Section S2 in the Supplementary Appendix).
After completion of VALOR, participants were given the option to participate in an open-label extension for up to 236 weeks, while remaining unaware of their trial-group assignment in VALOR. The combined analysis at week 52 of VALOR and its open-label extension was prespecified and was intended to enable comparison of early-start and delayed-start tofersen in the full intention-to-treat population. The extension phase is ongoing, and analysis is planned when all participants have completed at least 3.5 years of follow-up, which has not been reached.
We enrolled adults with weakness attributable to ALS and a confirmed SOD1 mutation. The primary analysis population was the subgroup of participants who met the trial-defined prognostic criteria for faster-progressing disease (see Section S2 in the Supplementary Appendix) and is called the “faster-progression subgroup.” Also undergoing randomization was a subgroup of participants who did not meet these enrichment criteria and were predicted to have slower progression of disease, the slower-progression subgroup. These persons were not included in the primary end-point analysis but had the opportunity to enroll in the open-label extension to receive tofersen. In the evaluation of combined data from VALOR and its open-label extension, participants who initiated tofersen in VALOR are referred to as the “early-start cohort,” regardless of whether they were predicted to have faster-progressing or slower-progressing disease in the randomized part of the trial. Those who received placebo in VALOR and had the opportunity to cross over to tofersen in the open-label extension approximately 28 weeks later are referred to as the “delayed-start cohort.”
The primary efficacy end point in VALOR was the change from baseline to week 28 in the ALSFRS-R total score in the faster-progression subgroup. The ALSFRS-R consists of 12 items across four subdomains of function (bulbar, fine motor, gross motor, and breathing), with total scores ranging from 0 to 48 and higher scores indicating better function. Prespecified secondary end points included the change from baseline in the total concentration of SOD1 protein in CSF, the concentration of neurofilament light chains in plasma, the percentage of the predicted slow vital capacity (volumes were standardized to the percentage of the predicted normal value on the basis of age, sex, and height), the handheld dynamometry megascore (average of z-scores across 16 muscle groups in the arms and legs, with higher values indicating greater strength), the time to death or permanent ventilation (≥22 hours of mechanical ventilation per day for ≥21 consecutive days), the time to death, and safety. Prespecified exploratory end points included participant-reported outcome measures such as the five-item Amyotrophic Lateral Sclerosis Assessment Questionnaire, fatigue (Fatigue Severity Scale), and quality of life (EuroQol Group 5-Dimension questionnaire). The same end points were assessed as part of the combined analyses of VALOR and the open-label extension.
We calculated that a sample size of 60 participants (2:1 randomization ratio) in the faster-progression primary analysis subgroup would provide 84% power to detect a between-group difference on the basis of the joint rank test (described below), assuming a change in the ALSFRS-R score from baseline to week 28 of −4.8 in the tofersen group and −24.7 in the placebo group, with a standard deviation of 20.39 and survival of 90% in the tofersen group and 82% in the placebo group, at a two-sided alpha level of 0.05. All primary and secondary end points for the 28-week randomized part of the trial were formally tested in the faster-progression subgroup. In the slower-progression subgroup, only the total SOD1 concentration in CSF was powered to test for statistical significance and was the primary end point in this population (Table S3). The joint rank test was used for statistical inference in the analysis of the change in the ALSFRS-R score. This accounts for both functional decline and survival and allows for a statistical test of the treatment effect while accounting for truncation of data owing to deaths. The joint rank score was calculated by comparing the change in each participant’s ALSFRS-R score from baseline to week 28 with that of every other participant in the trial, resulting in a score of 1 if the outcome was better than that of the participant being compared, –1 if worse, and 0 if the same. Participants who died were ranked lowest on the basis of their time to death, with progressively lower ranks given to those who died in the shortest period of time after the first dose. The sum of individual scores for each participant (i.e., ranked score) was assessed with the use of analysis of covariance (ANCOVA).
The ANCOVA model for ranked scores on the ALSFRS-R included trial group as a fixed effect and was adjusted for covariates (baseline disease duration since symptom onset, baseline ALSFRS-R total score, and use of riluzole or edaravone). The estimated between-group difference was obtained from the ANCOVA model for change from baseline in the ALSFRS-R score. Formal statistical testing for the overall population of all randomly assigned participants (irrespective of faster or slower predicted progression) was not specified for VALOR, but estimates are provided from the ANCOVA for change from baseline. Joint rank analysis was performed in conjunction with multiple imputation to account for missing data due to withdrawals not accounted for by death. The multiple-imputation model included trial group, use of riluzole or edaravone, and the baseline ALSFRS-R score. Additional subgroup and exploratory end points and analyses are described in Sections S3 and S4 in the Supplementary Appendix.
If the results for the primary end point differed significantly between the two trial groups, secondary end points for the faster-progression subgroup were tested with the use of a sequential closed testing procedure in order of ranking: the change from baseline (ratio to baseline) to week 28 in the total concentration of SOD1 protein in CSF, the change from baseline (ratio to baseline) to week 28 in the concentration of neurofilament light chains in plasma, the change from baseline to week 28 in the percentage of the predicted slow vital capacity, the change from baseline to week 28 in handheld dynamometry megascore, ventilation assistance–free survival, and overall survival. ANCOVA for change from baseline was used for all continuous end points and in conjunction with multiple imputation for handling missing data for withdrawals. Primary statistical inference for slow vital capacity was by joint rank analysis with the use of multiple imputation. For survival analyses, data for participants who did not meet the end-point definition were censored at the end of the trial or on the date of withdrawal. Only events that were adjudicated by the independent end-point adjudication committee were included. Treatment effects were assessed at a two-sided significance level of 0.05.
The first data cutoff for the combined analysis of data from VALOR and its open-label extension to evaluate the effects of early as compared with delayed initiation of tofersen was performed in July 2021. A second data cutoff of the open-label extension was performed on January 16, 2022, when the last participant who underwent randomization in VALOR had the opportunity for at least 52 weeks of follow-up from the start of VALOR. The combined analyses of these data are presented here. At the time that data from the January 2022 data cutoff were analyzed, the final results from VALOR and the original analysis of VALOR and its open-label extension had been presented at a scientific congress; however, participants, investigators and site staff, and the trial team remained unaware during the extension phase of the original trial-group assignments in VALOR.
Prespecified analyses of the data from VALOR and the data as of the first data cutoff of the open-label extension were performed on the basis of enrichment criteria (fast-progression and slow-progression subgroups) and of categorical subgroups defined by the median concentration of neurofilament light chains in plasma at baseline. Recognizing that adjusting for a continuous variable as a covariate more precisely controls for individual heterogeneity than dichotomizing the population into categorical subgroups, we amended the statistical analysis plan before analysis of the January 2022 data cutoff to incorporate the baseline concentration of neurofilament light chains in plasma as a covariate across analyses (Sections S2 and S4 in the Supplementary Appendix).
The combined analyses of the data as the January 2022 data cutoff are based on the intention-to-treat principle, whereby all participants who underwent randomization in VALOR (108 participants) are included according to their original trial-group assignment, regardless of fast or slow progression, adherence to the trial agent, early termination of the trial, or crossover to the tofersen group. The ANCOVA analyses in conjunction with multiple imputation were conducted identically to the analyses in VALOR. Kaplan–Meier survival analyses included all data up to January 16, 2022, for time to death or permanent ventilation and time to death; between-group comparisons for these end points were based on a log-rank test stratified according to trial group and the median concentration of neurofilament light chains in plasma at baseline (Section S4 and Table S5 in the Supplementary Appendix). Because there was no plan for adjustment of the widths of confidence intervals for multiple comparisons in the combined analysis, no conclusions can be drawn from these results.