Statistics from Altmetric.com
A fundamental rethink of objectives and methodology in clinical trials in the ALS syndrome is necessary
Twenty-five years have passed since the first successful trial of riluzole in amyotrophic lateral sclerosis (ALS), a study that concluded that the drug had a modest effect in extending life expectancy.1 Since then, there have been a disturbingly large number of failed trials of other compounds.2 This negative experience has led to much discussion. For example, was patient selection too narrow or too broad?3 Were these trials adequately powered, the primary and secondary end points adequate and sufficiently sensitive and relevant?2 Had the patients studied had ALS for too long?2 Most of these studies used survival as the critical endpoint, usually with a suitable definition of ventilatory assistance as a surrogate endpoint.2 However, ALS is a heterogeneous syndromic disorder, due to multiple causative and modifying factors. Many different genetic factors confer susceptibility and there are likely dietary, exercise-related and other complex environmental factors leading to so-called ‘sporadic ALS’. The ALS syndrome shows remarkable variations in age of onset, site of onset, rate of progression and even in its clinical semiology, including a frequent association with frontotemporal dementia (FTD), itself of varying severity. Inevitably, the question arises as to whether trials be inclusive, involving all clinical variants or designed to test only a narrow question within this clinical range?3 Since ALS presents with different patterns of anterior brain degeneration, involving FTD and combinations of upper and lower motor neuron features, it is logical to include all patients with a diagnosis of ALS, in a pivotal trial, whether possible, probable or definite, since this represents the conundrum of clinical practice. However, the variable prognosis and outcome in these patients requires very large numbers, indeed unacceptably large numbers, to reach a likelihood of a meaningful result.4 And failed trials are expensive in financial terms and, importantly, in terms of personal and social costs to patients and their families, who have invested so much hope and trust in the trial process, an act of faith in the investigators that is too easily ignored.
Clinical trials in ALS have been based on experience of trials in other disorders, notably in cancer, infections and other neurological disorders, but the challenges posed by the heterogeneity of the ALS syndrome and the absence of any easily measured biological variable, or marker, have proved difficult to address.2 Triallists are inevitably limited in their design considerations by the complex nature of the disease, not least the delay between biological onset, perhaps years before clinical recognition and diagnosis,5 the latter itself nearly a year after symptomatic presentation.6 7 Trials are therefore conducted in mixed cohorts of patients, many with advanced disease with rather slow rates of progression, as well as others with rapid progression. In this issue of the Journal, van Eijk et al 8 report a novel reanalysis of 13 completed and reported trials that involved nearly 5000 participants, focussing on sample size, duration, drug use and costs. Their analysis involved a rigorous methodology. They studied the historical assumptions underlying the trial designs, obtained real-life participant data and validated design and sample size calculations by simulation. They found that the mean hazard rate for the endpoint of death, the usual measure adopted in these trials, was overestimated by nearly 20%. In addition, the hazard rate for death was not linear but increased over time, largely as a consequence of increasing ventilatory hazard. They suggest that with attention to these findings sample size could be reduced by 33%, trial duration by 17%, drug use by 14% and trial costs by 21%. These results are clearly important for future triallists. Helpfully, the analyses have been posted in an open-source data base available to other triallists. These results have important implications. Have potentially useful therapies been rejected because previous pivotal trials were inadequately designed? That is a question that is difficult to address but which may, in some instances, be worth considering. For any clinical trial, its design is critical, particularly regarding sensitivity and specificity of measurements. This question also raises the more fundamental issue as to whether animal or human biological signals should be used for drug selection.
A major limitation of current preliminary research seeking plausible therapies for ALS is the reliance on SOD1 mouse models, for example, the G92A transgenic model. SOD1 ALS is a genetically driven disorder that is unique and discrete from the much more frequent ‘sporadic’ ALS syndromes that are characterised by ubiquitinated TDP43 protein storage, sometimes associated with a C9orf42 repeat mutation. In general, it is wise to be cautious in moving from an animal model of ALS to a major, expensive pivotal human trial. The search for a potential therapeutic signal in pretrial work should, therefore, be retuned to focus on Phase I clinical experiments in human subjects. The analysis of van Eijk et al does not address the methodology for drug selection for pivotal clinical trials, other than to note that small phase I/II studies have usually used mortality data as the test strategy, while acknowledging that such studies are inevitably grossly underpowered.8 A measure of disability and quality of life, the ALSFRS-R, is relatively insensitive, because it is smoothed by the multiplicity of data categories on which it is based.9 Since death is not an outcome with a unitary cause and time to death requires a long follow-up, a more sensitive test is required. As yet, there is no universally agreed surrogate marker of the disease that is sensitive to progression.
However, lower motor neuron degeneration is a cardinal and early objective feature of ALS, causing focal fasciculation, wasting and weakness, often especially evident in the distal upper limbs, that offers a potential, if neglected, measurement.10 11 Neurophysiological abnormalities in ALS are sensitive, objective and stable in repeat observations. Serial assessment over a short time, perhaps 3 months, using the Neurophysiological Index, calculated from (CMAP/DML × F-wave frequency%), in studies of the ulnar nerve-ADM system in one limb10 12 13 or MUNIX14 using a limited set of muscles,15 can both sensitively detect a therapeutic signal, for example, change in the rate of progression, in a relatively modest number of patients. Indeed, any meaningful slowing of progression of the disease, even in one limb, if consistent in a small number of individuals, would constitute such a signal. Thus, a potential therapy could be ascertained in human subjects with ‘sporadic’ ALS, the objective of any pivotal study. The issues raised by van Eijk et al 8 are fundamental to good trial design in Phase III trials, but there are other issues that deserve attention in planning clinical trials in the ALS syndrome,16 the most important of which are understanding the therapeutic signal leading to drug selection and decisions regarding disease stage, rate of progression and the clinical syndrome chosen for study, whether narrow or broad.
The simplest primary endpoint is still mortality. But do not rush into Phase III on weak evidence of efficacy!
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent for publication Not required.
Provenance and peer review Commissioned; internally 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.