From JAMA Neurology…
As we enter an era of secondary prevention trials to treat or prevent neurodegenerative diseases, the need to optimize clinical trial design and maximize efficiency is paramount. Biomarkers will be central to such efforts, wherein treatment is to be administered only to individuals affected by the targeted abnormalities and when symptoms are mild or absent. Cerebrospinal fluid (CSF) biomarkers will play a particularly vital role because they can provide a direct, real-time, changeable reflection of the physiology and biochemistry of the brain and spinal cord.
Foremost, useful biomarkers will guide enrollment into trials by ensuring accurate early symptomatic or presymptomatic diagnosis, uniform pathological staging, and selection of individuals at increased risk of a progressive course. Ideally, trials will also include biomarkers that can assess pharmacological “target engagement” by the therapeutic agent to verify delivery, efficacy, and appropriate dosage. Finally, biomarkers that are known to change over the natural course of the disease will be required to monitor enrollees for treatment effect by identifying any divergence from expected disease progression within treatment groups; such biomarkers can provide objective, short-term study end points and thus serve as surrogates for or foreshadowing adjuncts to symptomatic decline. In this manner, appropriate biomarkers will effectively strengthen, shorten, and economize clinical trials and hasten the discovery and evaluation of new treatments.
To date, there have been significant, invaluable advances in the discovery and validation of such CSF biomarkers, but most of these studies have focused on a specific disease—most often, Alzheimer disease (AD) (reviewed by Perrin et al1 and Fagan and Perrin2) but also Parkinson disease (PD) (reviewed by van Dijk et al3) and some others.
In comparison, although several valuable studies have been published (eg, Shi et al4 and Mollenhauer et al,5 among others), progress toward identifying and validating biomarkers with utility for differential diagnosis (ie, markers that can distinguish between neurodegenerative diseases) has been slower. In part, this slower pace might be attributed to the scarcity of some neurodegenerative diseases and of well-characterized representative cases from which research samples can be obtained. However, another contributing factor is the exclusive attention of many research efforts and funding sources on one disease to the exclusion of others; such focus limits the breadth of individual sample/data collections and, by requiring intercollection collaborations for differential diagnosis studies, can inadvertently introduce methodological incompatibilities between collections that can complicate multicenter biomarker studies.
Regardless of existing or potential obstacles, biomarkers useful for differential diagnosis are critical for trial design. This requirement is obvious for prevention trials that depend entirely on biomarker assessment for diagnosis but remains acute for trials that are designed to treat symptomatic stages of neurodegenerative diseases. Very different neuropathological processes can induce very similar clinical presentations, and a single neuropathological disease can present with protean manifestations. Moreover, not uncommonly, multiple neurodegenerative pathologies (eg, tauopathy, synucleinopathy, and AD pathology) can occur simultaneously in a single patient. Most important, however, misdiagnosed or unanticipated pathologies are unlikely to respond to disease-specific therapeutic interventions and could undermine the results of a clinical trial.
In this issue of the Archives, Hall and colleagues6 report that a modest panel of 5 previously reported CSF biomarkers (β-amyloid 42 [Aβ42], total tau, phosphorylated tau181, α-synuclein, and neurofilament light chain) has potential to distinguish common causes of dementia as well as common causes of parkinsonism. They reached this conclusion after examining an impressive 453 CSF samples, collected at 2 clinical sites in Sweden, from patients who met established criteria for the clinical diagnosis of dementia of the Alzheimer type (AD), PD with dementia (PDD), dementia with Lewy bodies (DLB), PD, or atypical parkinsonian disorders: multiple system atrophy (MSA), progressive supranuclear palsy (PSP), and corticobasal degeneration (CBD). Specifically, receiver operating characteristic curve analyses demonstrated that the panel could distinguish AD from PDD and DLB with an area under the curve of 0.90 and PD from the atypical parkinsonian disorders with an area under the curve of 0.93; such values might be considered optimal for clinically diagnosed cohorts.
Consistent with previous findings in a more limited study,7 the principal driver of the curve for differential diagnosis of parkinsonism was neurofilament light chain; its level in CSF is increased in atypical parkinsonian disorders and unchanged in PD. Neurofilament light chain contributed only mildly to the differential diagnosis of dementias. In contrast, generally consistent with previous findings in studies of comparable size and scope,4- 5 the main drivers of the curve for dementia were levels of total tau (elevated in AD relative to controls, mildly decreased in PDD, and unchanged in DLB), phosphorylated tau181 (elevated in AD relative to controls and unchanged in PDD/DLB), and α-synuclein (decreased in PDD and DLB relative to controls and increased in AD). Although Aβ1-42 levels were decreased in patients with AD, they were also decreased in those with DLB; this finding, as reported by others,8- 9 may be related to the Aβ plaques commonly observed in DLB.10 Consequently, the relative contribution of Aβ1-42 to the panel for the differential diagnosis of dementia was less pronounced.
Thus, the strengths of this validation study are in its synthesis of these candidate markers into a broadly applicable working panel and the size and quality of its CSF collection. The fact that 3 components of the panel—Aβ42, total tau, and phosphorylated tau181—are almost certainly to be implemented in several upcoming secondary prevention trials for AD11- 12 and that 4 of these markers (neurofilament light chain excluded) were “multiplexed” into a single high-throughput assay for this study make this panel particularly appealing.
Nevertheless, it remains to be seen whether this panel might have utility to classify vascular dementia or the frontotemporal dementias, which were not included in this study; additional markers might be required for that purpose. It would also be worthwhile to know how this panel might perform in clinically ambiguous or seemingly straightforward cases that involve more than one proteinopathy (eg, tauopathy and synucleinopathy); like most current studies, the vast majority of samples in this one lack corresponding neuropathological assessments. As this collection and other antemortem CSF collections around the world mature and accrue subsequent neuropathological data, uncertainties such as this latter one might be more readily resolved.
The cases included in this study—by necessity and by design—represent relatively ideal clinical examples, with patients who are symptomatic and presumably well into the course of pathology. It remains unclear whether the biomarker changes of the entire panel observed at these stages will also be detectable and diagnostic during the early symptomatic or presymptomatic stages that will be targeted in clinical trials. Additional studies, involving early-stage or preclinical samples and subsequent longitudinal data, will be required to determine whether these specific biomarkers—or others—will be relevant for differential diagnosis in the design of secondary prevention trials.
Clearly, more work needs to be done. Nevertheless, this study represents a significant step forward, demonstrating how a relatively modest panel of robust CSF protein biomarkers can categorize dementias and parkinsonian syndromes on the basis of pathology rather than clinical/behavioral changes. Implementation of CSF biomarker panels such as this one should improve the efficiency of clinical trials and accelerate the evaluation and discovery of new effective treatments for neurological diseases.