CNS Biomarkers. In the last two days, our team attended the Workshop for Multimodal Biomarkers in CNS Disorders held at the National Academies of Sciences, Engineering, and Medicine in Washington, DC. This conference provided a needed review of the current state of multimodal biomarker discovery and development. While most of the speakers focused on more common CNS disorders such as Alzheimer’s disease and neuropsychiatric disorders, there stands to be important lessons that can be translated into the rare disease field. Here is what we learned about the clinical utility of biomarkers and their potential as we move towards precision medicine in rare disease.
Biomarkers. Biomarkers are objective measurements of the underlying pathophysiology of disease. The rationale behind why we use biomarkers is rooted in the idea that clinical trial data is not enough. In addition to seizure outcomes and developmental endpoints, we need complementary measures that can be assessed systematically. To understand the role of biomarkers in CNS disorders, it is important to recognize that biomarkers serve different purposes. While biomarkers in one disease can be used for identification of risk or susceptibility, disease onset, etiology, and pathogenic drivers, biomarkers for another disease can include the ability to monitor disease progression, stratify patients into clinically meaningful subgroups, predict short-term or long-term trajectories, or to prognosticate drug response and guide treatment strategies. In the genetic epilepsies and neurodevelopmental disorders, it is more likely the case that biomarkers will take on the role of the latter, by measuring disease progression and allowing for prognostication of outcomes and treatment response. However, what are the current challenges and considerations of biomarker development in rare disease?
Disease-specific. Before we expand upon the potential application of biomarkers in rare disease, it is important to understand that biomarkers should be disease-specific. That is, reliable biomarkers have a mechanistic link between the measurement and the underlying disease pathophysiology. In the context of understanding Alzheimer’s disease progression and onset, for example, the aggregation of amyloid beta or phosphorylated tau can be measured via positron emission tomography (PET) or CSF levels. However, for genetic epilepsies and rare diseases, there are very few validated measurements. Nevertheless, in the context of biomarker discovery, the identification and advancement of a biomarker starts with first defining how the biomarker will be used.
Heterogeneity. The genetic epilepsies and neurodevelopmental disorders share overlapping challenges with other CNS disorders in biomarker development, primarily when confronting the heterogeneity of clinical presentations. Given the phenotypic variability with regards to clinical subtypes, longitudinal disease progression, age relative to disease onset, and wide spectrum of outcomes, it is not surprising that there will likely not be a single biomarker that can explain the whole variance of a disease. When considering biomarkers for genetic epilepsies, it is possible that measures such as quantitative EEG (qEEG) will need to be qualified and assessed alongside other measures such as neurofilament light chain (NfL) in the context of seizure-induced neuronal injury or functional MRI (fMRI) for related neurodevelopmental conditions.
Harmonization. However, this heterogeneity is not limited to the underlying biological variability of these disorders but also found in the differences in the collection and analysis of biomarkers across studies and institutions. On this note, many of the challenges in biomarker development parallels and overlaps with the progression of artificial intelligence (AI) and machine learning in rare disease – when thinking about clinical validity and utility, there is an increasing need for validation, replicability, and reproducibility. So, how can we overcome the variability and amplify the signal to noise? The fundamental and brief answer underlies much of what our lab aims to do in the domain of computational phenotyping: data harmonization. Despite biological and inherent sources of variability, we show that it is possible to sharpen the signal and identify genotype-phenotype signatures, even in rare genetic epilepsies with significant clinical variability such as STXBP1-related disorders. In the context of multimodal biomarker development, this will include standardized frameworks for the collection and analysis of multidimensional measurements where clinical features and seizure frequencies are replaced by biospecimen samples and EEG reports.
Trial-readiness. The goal of biomarker discovery in rare disease is to assist with drug development and help move the field into the space of “trial-readiness.” So, what could biomarker development in conditions such as STXBP1-related disorders or SYNGAP1-related disorders potentially look like? In an ideal scenario, we would identify a combination of disease-specific biomarkers, such as the relative amount of protein in CSF or specific proteomic signatures as well as more global biomarkers such as NfL concentrations in plasma. Some work has already been performed on qEEGs, and under ideal circumstances, measurements such as the excitation/inhibition imbalance will be disease-specific and function as “core” biomarkers, with related EEG signatures in other developmental and epileptic encephalopathies (DEE). Furthermore, the use of wearable sensors may be able to detect unique signatures across various genetic epilepsies, such as the STXBP1 tremor. This combination of core and non-core biomarkers will then allow us to monitor disease progression and treatment effect in parallel with standardized assessments ascertained by providers such as the Bayley Scales of Infant and Toddler Development or by families such as the Observer-Reported Communication Ability (ORCA) Measure. This synthesis of markers, jointly with more in-depth insight into the natural history of these conditions will then allow us to obtain a precise understanding of whether a disease-modifying therapy is effective, even when clinical outcome markers may initially not show a prominent signal.
What you need to know. While biomarker development in rare disease has been limited by low statistical power of rare subtypes, lessons learned with regards to challenges and opportunities in more common CNS disorders can be translated to rare disorders. The integration of multimodal biomarkers holds promise in not only advancement in the scientific understanding of the pathophysiology of a disease but also in the clinical utility as a modality for diagnosis, prognostication, and trial design and interpretation. Given the variability across disorders, translating the potential of biomarkers into the clinical sphere will hinge on significant efforts in developing the infrastructure for the qualification and systematic analysis of reliable measurements. Nonetheless, individualizing patient care will rely on validated and robust biomarkers with which will pave the way for future precision medicine approaches.