Fatigue Benefit With Janssen’s MS Drug Ponvory Not Clinically Relevant – US/EU Regulators
Executive Summary
In approving ponesimod for relapsing multiple sclerosis, neither the FDA nor EMA included in labeling data from a novel patient-reported outcomes fatigue instrument; FDA said the PRO was not shown to be capable of measuring within-patient change, while EMA said the difference relative to Sanofi’s Aubagio fell short of the threshold for a clinically meaningful benefit.
A statistically significant benefit on a novel, patient-reported outcomes fatigue instrument was not enough to convince US and EU regulators the data warranted inclusion in labeling for Janssen Pharmaceuticals Inc.’s multiple sclerosis drug Ponvory (ponesimod).
Ponesimod demonstrated a benefit on the symptom scale of the Fatigue Symptom and Impact Questionnaire – Relapsing Multiple Sclerosis (FSIQ-RMS), a PRO instrument developed by Janssen. However, reviewers from both the US Food and Drug Administration and the European Medicines Agency questioned the clinical meaningfulness of the small numerical change relative to Sanofi’s Aubagio (teriflunomide) in the Phase III trial.
While both agencies concluded the fatigue results did not demonstrate a clinically meaningful benefit, they focused, in part, on different issues in their respective reviews, showing how two regulators can get to the same place by taking somewhat different paths.
Both regulators concluded the PRO instrument was validated. However, FDA reviewers said the FSIQ-RMS was not sufficiently justified as an instrument capable of measuring a within-patient change in fatigue.
Furthermore, the agency’s interpretation of the fatigue results were complicated by missing data, the use of another PRO instrument that was not appropriate to “anchor” a determination on clinically meaningful change threshold for the FSIQ-RMS, and uncertainties as to teriflunomide’s effect on fatigue.
Similar to the FDA, EMA reviewers said their interpretation of the fatigue data was hampered because the FSIQ-RMS was not previously included in a clinical study and its sensitivity to change was not tested prior to use in the ponesimod Phase III trial.
However, the European regulator also said the mean difference between the ponesimod and teriflunomide groups in fatigue symptom scores fell short of the threshold determined to represent a clinically meaningful benefit. In addition, EMA review documents cite the lack of improvement in fatigue scores in the ponesimod group relative to baseline, and the fact that most subjects had mild fatigue to begin with, as reasons for finding no clinically relevant impact on fatigue.
Results from the fatigue instrument were not included in the product’s EU summary of product characteristics because such labeling “should be reserved to results that are statistically compelling and clinically relevant,” the EMA’s public assessment report states.
Superiority In Reducing Relapses
Ponesimod, a sphingosine 1-phosphate receptor modulator, was approved by the FDA on 18 March 2021 for treatment of relapsing forms of MS. The European Commission followed suit two months later, granting marketing authorization on 19 May 2021.
The drug recently won approval from England and Wales’ health technology assessment body NICE after facing an initial rejection. (Also see "English Reimbursement Win For Janssen’s Oral Multiple Sclerosis Drug" - Pink Sheet, 11 Jan, 2022.)
In both the US and the EU, the drug was approved on the basis of a single Phase III study (OPTIMUM, also known as AC-058B301), in which 1,133 patients were randomized to ponesimod or teriflunomide. Patients were treated for 108 weeks, and the primary endpoint was annualized relapse rate (ARR) over the study period.
The ponesimod group experienced a 30.5% relative reduction in ARR compared to teriflunomide, a statistically significant result.
Novel PRO Instrument
The first key secondary efficacy endpoint in OPTIMUM was change from baseline to week 108 in fatigue-related symptoms as measured by the FSIQ-RMS.
The sponsor’s decision to place fatigue higher than disability progression in the secondary endpoint hierarchy rankled the EMA. (See sidebar for story.)
The FSIQ-RMS is a 20-item PRO instrument developed by Actelion Pharmaceuticals Ltd. (acquired in June 2017 by Janssen parent Johnson & Johnson) to evaluate two domains of fatigue: symptoms (FSIQ-RMS-S) and symptom impact (FSIQ-RMS-I).
The FSIQ-RMS-S comprises seven items assessing fatigue-related symptoms over seven consecutive days, with a 24-hour recall period. It uses an 11-point numeric rating scale, meaning that the unscaled symptom domain score of the FSIQ-RMS ranges from 0-77, with a higher score indicating greater fatigue. Subjects entered their data into an electronic diary at baseline and at weeks 12, 24, 60, 84 and 108.
The single-item Patient Global Impression of Severity (PGI-S) scale also was administered in OPTIMUM as an anchor for interpreting the clinical meaningfulness of the results of the FSIQ-RMS-S.
Through development of the FSIQ-RMS, Janssen saw an opportunity to address the limitations of existing PRO instruments for measuring fatigue in multiple sclerosis patients.
Janssen told the Pink Sheet that although PRO instruments previously have been used to measure fatigue in MS patients, review of their measurement properties has suggested shortcomings relative to the current standards for PRO instrument development.
“As such, we worked to address limitations of existing MS-specific instruments by developing a new, content-valid, concise PRO instrument (FSIQ-RMS),” the company said. “FSIQ-RMS works to assess fatigue symptoms relevant to patients within the spectrum of relapsing MS and the relevant impact of these symptoms on patients’ lives, in accordance with FDA PRO guidance.”
FDA review documents reflect extensive discussion between the sponsor and agency about the FSIQ-RMS during the instrument’s development. (See sidebar for clinical development timeline.)
According to minutes from a September 2019 pre-NDA meeting, Janssen asked the agency whether a specific point change on the FSIQ symptom domain would be an acceptable threshold for interpreting within-subject change from baseline to week 108.
In response, the FDA said it did not agree there is sufficient evidence or justification to support that Janssen’s proposed change threshold is clinically meaningful to patients. Janssen needed to provide justification regarding the clinical meaningfulness for its proposed threshold change, the FDA said.
“It is important to understand what constitutes a meaningful improvement in the 11-point PGIS scale ratings based on the patient perspective; this would aid in determining an appropriate point change in the PGIS scale to be used as the anchor to define improvement in the FSIQ Symptoms domain score,” the agency said.
Although Janssen’s proposed threshold for a clinically meaningful change was redacted from FDA review documents, the proposed threshold at issue in the EMA review was a -6.3 point change, according to the EPAR.
“Clinically meaningful change in the FSIQ-RMS was not pre-specified in the study protocol,” the EPAR states. “According to the psychometric analysis based on the data from this pivotal study, a clinically meaningful change at the subject level in the FSIQ-RMS was defined. This was done by anchoring the PGI-S score to the FSIQ-RMS score. A -6.3 change on the FSIQ-RMS was concluded as the meaningful change threshold value at the subject level.”
Missing Data
From baseline to week 108, ponesimod demonstrated a statistically significant 3.57-point reduction in the FSIQ-RMS-S relative to teriflunomide. The mean change in the ponesimod group from baseline was -0.01.
Story continues after table….
|
|
Ponesimod (n=567) |
Teriflunomide (n=566) |
|
Number of patients with measures at baseline and week 108 |
449 |
458 |
|
Mean score change from baseline |
-0.01 |
+3.56 |
|
Treatment effect (score difference) |
-3.57 |
|
|
95% confidence interval |
(-5.83, -1.32) |
|
|
p-value |
0.0019 |
|
|
Source: FDA summary review |
||
FDA reviewers identified a host of statistical and clinical concerns with Janssen’s use of the FSIQ-RMS-S and their ability to interpret the data, including a large amount of missing data.
“FSIQ symptoms scores at Week 108 were available for only 70.0% and 66.6% of patients in the ponesimod and teriflunomide groups, respectively and, because of missing baseline assessments, a change from baseline to Week 108 was available for only 60.7% and 58.0%, respectively,” the FDA’s summary review states. The review was authored by: Paul Lee, deputy direct of the Division of Neurology 2; DN2 director Nick Kozauer; and Eric Bastings, deputy director of the Office of Neuroscience.
FDA clinical reviewer David Jones said the extent of missing data and resulting potential for bias are concerning because it is possible that more fatigued subjects would be less likely to complete the PRO assessment.
Although the FSIQ-RMS-S score difference was statistically significant, a tipping point analysis suggested this was a relatively fragile finding and that just a 1.5-point change provided by the missing data could have rendered the observed treatment effect not significant, the FDA’s summary review states.
Clinical Meaningfulness Questioned
FDA reviewers also questioned whether the observed change in the FSIQ-RMS symptom score was clinically meaningful.
FDA clinical outcome assessment reviewer Susan Pretko and COA associate director Elektra Papadopoulos concluded the FSIQ-RMS-S has content validity, but insufficient information was provided to support interpretation of clinically meaningful within-patient changes in the symptom instrument scores.
FDA clinical outcome assessment reviewers said the FSIQ-RMS-S has content validity, but insufficient information was provided to support interpretation of clinically meaningful within-patient changes in the symptom instrument scores.
Although anchor-based methods are the primary approaches used by the agency to interpret meaningful within-patient score changes in COA endpoints, the PGI-S administered in the OPTIMUM study was not an appropriate anchor scale for the FSIQ-RMS and should not have been used to establish the clinical meaningfulness of the Janssen-developed instrument, the COA and summary reviews state.
“Anchor scales should be easier to interpret than the COA endpoint and should have distinct and non-overlapping response categories,” the COA review states. “The PGI-S uses a 0-10 numeric rating scale (NRS) which has limitations as an anchor measure given its intermediate response categories do not have verbal descriptors, and it is unclear what difference on this scale is clinically meaningful.”
FDA reviewers also took issue with the small magnitude of change.
The change of 3.5 points in the FSIQ-RMS-S amounts to a mean change of approximately 4% in the raw score, “which is a very small absolute change and below the typical threshold of a 20% change often considered for a clinical outcome measure to be considered to be potentially meaningful,” the summary review states.
For EMA, Change Falls Below Threshold Level
EMA reviewers noted the observed mean difference in change between treatment groups of -3.57 points is less than -6.3 points, “which was determined to be meaningful change threshold at the subject level.”
The applicant cited a patient preference study in asserting that the -3.57 change in the FSIQ-RMS-S score is as important to patients as a change of -0.06 in yearly relapse rate, which matches the between-treatment difference in ARR observed in the pivotal study (-0.088,) the EPAR states.
“This is a somewhat difficult comparison as while the patient preference study measures absolute changes in the FSIQ-RMS and relapse rate, the comparison is to a between-treatment difference of ponesimod and teriflunomide. A fair comparison would be if the between-treatment comparison would be to placebo.”
“The reduction in ARR implies that in the ponesimod group, less new events occur that leads to fatigue. So, the implicit claim that ponesimod has an independent effect on fatigue is not warranted.” – EMA
EMA reviewers point out that compared to baseline, there was no improvement in the FSIQ-RMS-S in the ponesimod group; scores in this arm remained stable, whereas in the teriflunomide group they increased.
This finding of symptom stability in the ponesimod arm “appears logical,” EMA reviewers said. “The reduction in ARR implies that in the ponesimod group, less new events occur that leads to fatigue. So, the implicit claim that ponesimod has an independent effect on fatigue is not warranted.”
Furthermore, other endpoints based on the FSIQ-RMS instrument and the PGI-S did not show statistically significant difference between treatment groups. “It is noted that the baseline FSIQ-RMS, as well as PGI-S scores, indicate that majority of patients had mild fatigue thus little room for improvement.”
Uncertainty About Teriflunomide Effect
FDA reviewers also said consideration needs to be given to whether teriflunomide, the active comparator, has a positive or negative effect on fatigue in general and the FSIQ-RMS in particular.
The Sanofi drug did not have a statistically significant effect on the Fatigue Impact Scale (FIS) in one of its pivotal studies in RMS but did have a statistically significant effect on the FIS at the end of another pivotal study.
Pointing to contradictory information provided by Janssen, clinical reviewer Jones said it is unclear whether teriflunomide has a beneficial or detrimental effect on fatigue as measured by the FIS, “an instrument that is arguably less specific for MS fatigue than the FSIQ-RMS.”
In response to an FDA information request about teriflunomide’s effect on fatigue in individuals with RMS, “the applicant was unable to provide additional clinical trial information about the effect of teriflunomide on the FIS but offered ‘real world’ data suggesting stabilization of fatigue with teriflunomide,” the review states.
“Conversely, after the late-cycle meeting (LCM), the applicant submitted a meta-analysis suggesting that teriflunomide does not have an effect on MS fatigue, at least as measured by the FIS.”
The agency’s summary review concludes that teriflunomide’s effect on fatigue as measured by the FSIQ-RMS is not known, and any relative finding of ponesimod to teriflunomide “requires conjecture given the uncertainty regarding teriflunomide’s treatment effect.”