Biomarker Qualification for Neurofilament Light Chain in ALS: Theory and Practice

Abstract

Objective.

Explore whether the utility of neurofilament light chain (NfL), as a biomarker to aid amyotrophic lateral sclerosis (ALS) therapy development, would be enhanced by obtaining formal qualification from the FDA for a defined context-of-use.

Methods.

Consensus discussion among academic, industry, and patient advocacy group representatives.

Results.

A wealth of scientific evidence supports the use of NfL as a prognostic, response, and potential safety biomarker in the broad ALS population, and as a risk/susceptibility biomarker among the subset of SOD1 pathogenic variant carriers. Though NfL has not yet been formally qualified for any of these contexts-of-use, the FDA has provided accelerated approval for an SOD1-lowering antisense oligonucleotide, based partially on the recognition that a reduction in NfL is reasonably likely to predict a clinical benefit.

Interpretation.

The increasing incorporation of NfL into ALS therapy development plans provides evidence that its utility—as a prognostic, response, risk/susceptibility, and/or safety biomarker—is already widely accepted by the community. The willingness of the FDA to base regulatory decisions on rigorous peer-reviewed data absent formal qualification, led us to conclude that formal qualification, despite some benefits, is not essential for ongoing and future use of NfL as a tool to aid ALS therapy development. While the balance of considerations for and against seeking NfL biomarker qualification will undoubtedly vary across different diseases and contexts-of-use, the robustness of the published data and careful deliberations of the ALS community may offer valuable insights for other disease communities grappling with the same issues.

“In theory there is no difference between theory and practice, while in practice there is.”

Benjamin Brewster¹ (though often misattributed to Yogi Berra)

Introduction

Neurofilament light chain (NfL) has emerged as a leading biomarker candidate relevant to therapy development for a host of neurodegenerative disorders², including amyotrophic lateral sclerosis (ALS)³, frontotemporal dementia (FTD)⁴, multiple sclerosis (MS)⁵, Huntington’s disease (HD)⁶, and Alzheimer’s disease (AD)⁷. The applicability of NfL to a broad range of neurological disorders derives from its non-specific nature as a marker of axonal degeneration. Such broad relevance, however, should not mask the critical importance of recognizing, and indeed defining, the specific context-of-use in which NfL will be used as a biomarker⁸. For ALS, the body of scientific evidence supports its use as a prognostic^9,10, response^10–12, and potential safety biomarker^13–15 in the broad ALS population, as well as a risk/susceptibility biomarker at least among a subset of SOD1 pathogenic variant carriers^16,17. The potential for NfL to aid ALS therapy development, combined with the lack of any qualified biomarkers for ALS, has engendered consideration of whether to seek qualification of NfL, with a defined context of use (COU) through the Food and Drug Administration’s (FDA) Biomarker Qualification Program (BQP)¹⁸. Indeed, such efforts are already underway in both the MS¹⁹ and FTD²⁰ communities, and the Foundation for the National Institutes of Health (FNIH) Biomarker Consortium has explored whether to include ALS under the auspice of their ongoing work in the FTD space (personal communication).

The FDA’s BQP was born out of the need to bridge the gap between biomarker discovery research and efforts to validate biomarkers for use in clinical trials. As the regulatory authority overseeing drug development in the United States, the FDA aimed to codify a process for diverse stakeholders to submit information that would enable the FDA to formally recognize that a biomarker could be used to support a particular regulatory decision. In essence the FDA sought to establish rigorous scientific criteria to support development and subsequent adoption of biomarkers by the broader scientific community. In turn, this would enable every group planning a clinical trial to rely on publicly available information rather than needing to do their own evaluation of whether the biomarker is fit for purpose.

Through a Clinical Trial Readiness Grant (U01 NS107027) from the National Institutes of Health (NIH), the CReATe (Clinical Research in ALS and Related Disorders for Therapeutic Development) Consortium sought to validate NfL as a biomarker for ALS clinical trials³ and committed to exploring the value of seeking formal qualification of NfL with guidance from an External Advisory Committee (EAC) comprising experts in biomarkers, ALS, and industry drug development, with input from patient advocacy group partners. The complexity of the formal qualification process, however, combined with emerging data and the recent FDA decision to provide accelerated approval based on NfL²¹ despite it not being formally qualified for this purpose, are reasons to give pause. It is against this background that we convened a recent meeting of the EAC to decide whether to pursue formal qualification.

Biomarker Qualification - in Theory

The Evidentiary Framework, as defined by the FDA, for seeking qualification entails a needs assessment, definition of a context of use, and consideration of the benefits and risks of using the biomarker in a specified context²². In turn, these factors inform the evidence needed to support qualification, including articulation of a biological rationale, presentation of data to support the relationship between the biomarker and the clinical outcome of interest, and the analytical performance of relevant assays.

Needs Assessment.

The need for biomarkers to support ALS drug development has been well articulated²³. Briefly, the long history of ALS clinical trials has been characterized by broad eligibility criteria; and the use of clinical outcome measures such as change in function (measured using the ALS functional rating scale revised [ALSFRS-R]) or tracheostomy-free survival as endpoints in both phase 2 and phase 3 trials. Over-reliance on clinical outcomes, which are insensitive in the early-to-mid phases of clinical drug development, especially in broadly inclusive trial populations, has undermined effective decision-making. There are multiple examples of false positive trial results and an unknown number of false negatives²⁴. Biomarkers have been envisioned as important drug development tools to address these challenges.

Context-of-Use.

Based on the foregoing, four principal contexts of use have been considered for NfL. The first is as a prognostic biomarker, with substantial peer-reviewed evidence supporting the value, for example, of baseline plasma/serum NfL in predicting future functional decline and survival in the setting of both ALS observational studies¹⁰ and clinical trials¹². The second potential context-of-use for NfL is as a response biomarker. Although there is currently only a single published trial showing a meaningful reduction in NfL in response to an experimental therapeutic¹², supportive evidence may be found in unpublished data from an ongoing expanded access study of a FUS-targeting antisense oligonucleotide (ASO), the observed reduction in NfL following high lipid and calorie supplementation in the LIPCAL-ALS trial²⁵, as well as the pharmacodynamic utility of NfL in other therapeutic contexts, including the genetic motor neuron disease spinal muscular atrophy (SMA)^26–28. The limited evidence for NfL as a response biomarker in ALS may relate to the paucity of meaningfully effective therapies. Notably, however, there is no reduction in NfL following treatment with Riluzole²⁹ or in the related candidate biomarker, phosphorylated neurofilament heavy (pNfH) in the phase 2 trial of AMX0035³⁰. Whether these observations reflect the very modest effect of these agents or the limited sensitivity of an NfL response to therapeutic agents with diverse mechanisms of action, remains to be determined. Nevertheless, the promise of NfL as a response biomarker has inspired a platform trial in the UK to prioritize repurposed drugs for ALS based on their ability to lower NfL³¹. The third potential context-of-use is as a risk/susceptibility biomarker, with a rise in plasma NfL predicting imminent risk of phenoconversion among carriers of highly penetrant SOD1 mutations associated with rapidly progressive ALS¹⁶. NfL is currently being used in this context in the ongoing ATLAS trial³². Finally, NfL may have value as a safety biomarker, with an increase in NfL following exposure to an investigational agent, raising potential concerns about neurotoxicity.

Benefit and Risk Considerations.

From the FDA’s perspective, benefits are focused on the potential value the biomarker adds to drug development; and risks include the anticipated consequences if the biomarker is unsuitable for its intended use. We evaluate the benefits and risks separately for each of the potential contexts-of-use detailed above, although note that serum/plasma NfL increases with advancing age^33,34, a consideration that is potentially relevant to all proposed contexts-of-use. As a prognostic biomarker for ALS, serum/plasma NfL (which correlates well with cerebrospinal fluid NfL^3,35) has been found to add prognostic value above and beyond what can be learned from readily available clinical information – and indeed may outperform commonly used clinical tools^3,35. Whether NfL adds prognostic value to what can be determined from available multivariate prognostic models such as the tool develop by ENCALS³⁶, is not yet known. Inclusion of baseline NfL in multivariate models may control for the rate of disease progression, thereby facilitating easier detection of a therapeutic signal amidst the noise of variability in natural history. Indeed, such strategies are being used in many but not all recent and ongoing ALS clinical trials^12,37. The use of NfL concentration to stratify randomization, on the other hand, might be more speculative as this would require prospective delineation of threshold values or ranges, and such thresholds have not yet been defined, although this concern might be mitigated through use of dynamic randomization. Residual uncertainty about the use NfL poses some risk of inadequately controlling for the prognostic value of NfL.

The benefits of using a reduction in NfL as a response biomarker include the potential to show evidence of a biological response over a shorter time and in a smaller number of patients; and the face validity of lowering the concentration of a marker that is well established to reflect the rate of axonal injury/degeneration³. Notably, this is further supported by the fact that a rise in NfL in response to an investigational product is interpreted as a sign of potential toxicity^13–15,38. The potential risk of relying on NfL as a response biomarker is the residual uncertainty about both the positive and negative predictive values of a lowering of NfL in response to any experimental therapeutic, and for predicting future clinical benefit especially for a drug with a mechanism of action (e.g. muscle contractility) that is unrelated to axonal injury/degeneration. The potential risk in using NfL as a safety biomarker is the under-recognition that NfL may increase subtly over time in the untreated state^10,35, and that short-term toxicity (for example, to dorsal root ganglia) from viral delivery vectors may mask potential benefits of the gene therapy being delivered¹⁵.

The benefit of using NfL as a risk/susceptibility biomarker lies in its value in predicting imminent phenoconversion to clinically manifest ALS, with the attendant opportunity to use a rise in NfL (a marker of axonal injury) as a trigger for initiating an experimental therapeutic during the pre-symptomatic stage of disease. With gene therapies increasingly under development for ALS and the recent approval of tofersen, genetic testing among ALS patients and their family members is now pervasive. As a result, there is a rapidly growing population of family members who are known carriers of pathogenic ALS mutations who have not yet developed disease. Thus, the benefits of a biomarker of phenoconversion in this expanding population cannot be overstated. The risk lies in over-extrapolation from NfL data acquired specifically in carriers of highly penetrant SOD1 mutations associated with rapidly progressive ALS, to patients with pathogenic mutations in other genes, or even other SOD1 mutations, where the pre-symptomatic trajectory of NfL may differ. Simply put, the available evidence suggests that a rise in NfL, on its own, is likely not sufficient to function as a risk/susceptibility biomarker among the broader population at elevated genetic risk for ALS³⁹ or FTD⁴⁰. Moreover, it also remains to be determined whether the initiation of therapy prior to symptoms, but after the appearance of axonal degeneration, is sufficient or whether yet-to-be-developed biomarkers of cellular or molecular function, or even functional compensation might be preferable. The ongoing ATLAS trial will provide some insight into this important question³².

Scientific Evidence to Support the Proposed Context-of-Use.

Qualification requires a body of evidence to demonstrate analytic validation of assays for quantification of the biomarker, as well as clinical validation of the proposed context-of-use in a clinically well characterized cohort. For NfL, a variety of methods for its quantification in serum/plasma and CSF have undergone analytic validation, with the performance characteristics of a range of assays well established^41,42. Although qualification of NfL as a biomarker has potential to improve reproducibility and consistency in how NfL is measured, most of these issues have already been addressed by the field; and sources of variability such as operator dependencies, batching strategies, and standardizing automated analysis platforms would not be further resolved through qualification. Notably, the process of seeking qualification would also not address the current unmet need for an independent reference standard as a calibrator for generating the standard curve necessary for result interpretation. Moreover, the use of NfL as a prognostic biomarker is supported by robust body of peer-reviewed scientific evidence, with some (but thus far limited) evidence supporting its use as a response or risk/susceptibility biomarker (i.e. demonstrated value in the proposed context-of-use within well characterized clinical cohorts). In short, therefore, a significant body of literature supports both the analytic and clinical validity of NfL as a fit-for-purpose biomarker. Notably, formal qualification for one context-of-use would not preclude NfL from also being used for a different context-of-use in ALS or another neurodegenerative disease.

Biomarker Qualification - in Practice

Formal biomarker qualification is a long and complex process, requiring a Critical Path Innovation Meeting (CPIM) or pre-letter of intent (LOI) meeting with the FDA, followed by an LOI, submission of a qualification plan, and finally submission of a full qualification package. Despite the interest in biomarkers and their enormous potential to aid therapeutic development, only eight biomarkers have been qualified by the FDA since the initiation of BQP in 2007 – and none for a neurological disease⁴³. The potential benefits of pursuing formal qualification, therefore, must be weighed against the costs, which include significant time, effort, and resources given the complexity of the qualification process. This is true, not only for biomarkers such as NfL in ALS, but also for communities exploring the benefits of biomarker qualification for other diseases. Importantly, the FDA is very attuned to the weight of the published scientific literature and seems willing to make regulatory decisions based on these data, even absent formal qualification. For example, the FDA’s recent decision to provide accelerated approval for tofersen²¹, was based in part on a marked reduction in plasma NfL, which they concluded is reasonably likely to predict a clinically meaningful benefit. This conclusion was undoubtedly informed by (1) the volume and quality of published evidence supporting the use of NfL in this way, (2) data provided by the sponsor to demonstrate technical performance of the assay, and (3) the clinical trial data submitted by the sponsor.

Conclusions

We believe there is compelling evidence from the published literature to support the use of NfL as a prognostic biomarker in future ALS clinical trials. Moreover, there is emerging evidence to support NfL as a response, safety and risk/susceptibility biomarker. There is, however, work that still needs to be done. Optimal use of NfL as a prognostic marker will require definition of assay-specific thresholds of NfL that might be used to stratify randomization. Quantification of the sample size savings that could be yielded by using NfL as a prognostic marker would also be valuable, as would demonstration of the prognostic value that NfL might add to existing disease progression models. Additional natural history data for carriers of mutations other than SOD1 mutations associated with rapidly progressive disease, alongside information about the trajectory of NfL based on more frequent (in-home) monitoring, would add weight to the risk/susceptibility context of use. The same is true for the use of NfL as a risk/susceptibility biomarker for those at risk for FTD, where multi-modal biomarker data (including NfL) will likely be necessary⁴⁰. Moreover, further trials showing a reduction in NfL in response to an experimental therapeutic, and determination of the clinical correlate of lowering NfL by a defined amount, would help to advance development of NfL as a response biomarker and perhaps even as a surrogate outcome. Similarly, we need to better understand the temporal course of the subtle increase in NfL over time to aid interpretation of a rise in NfL as a potential sign of toxicity. None of these goals, however, require qualification or are necessarily achieved through the process of seeking qualification. As such, we remain unconvinced that an effort to seek formal qualification is the optimal path forward to enable the expanded utilization of this biomarker.

The increasing incorporation of NfL into ALS therapy development by academic groups and the pharmaceutical/biotech industry serves as evidence that the utility of NfL is already widely accepted by the ALS community for its potential as prognostic, response, safety, and risk/susceptibility biomarkers. Moreover, despite the limited extant literature on NfL as a response biomarker, the FDA has shown a willingness to provide accelerated approval for a novel therapeutic for SOD1 ALS, making it increasingly likely that others engaged in ALS therapy development will pursue a similar strategy. NfL has also displayed promise as a biomarker across neurodegenerative indications, and has already been used as an endpoint in over 100 trials of drugs in multiple sclerosis, Alzheimer’s disease, and Huntington’s disease, among others². It seems therefore that, in practice, there is indeed a difference between theory and practice.

What is the current knowledge on the topic?

Neurofilament light chain (NfL) has emerged as a promising biomarker with the potential to enhance drug development efforts across a range of neurodegenerative diseases, including ALS.

What question did this study address?

This study aimed to address the question of whether the utility of NfL as a biomarker to aid ALS therapy development, would be enhanced by obtaining formal qualification from the FDA for a defined context of use, and whether these benefits justify the costs.

What does this study add to our knowledge?

This study lays out the case for and against seeking formal qualification from the FDA, ultimately concluding that NfL is already being used, and can continue to be used, as a drug development tool to aid ALS therapy development, even without qualification.

How might this potentially impact on the practice of neurology?

While this study is more focused on the use of NfL as a drug development tool to aid ALS therapy development, the issues discussed are also relevant to clinical practice. NfL is currently available commercially as an orderable laboratory test; and a nuanced understanding of the pros and cons of using NfL for different purposes, will help to minimize the potential for its misuse.

Data Availability

Data sharing is not applicable to this article as no new data were created or analyzed for this manuscript.