Challenges in Developing Alzheimer's Blood Tests
Challenges in Developing Blood Tests for Alzheimer's Disease
In recent years, liquid biopsy has revolutionized the field of oncology, enabling early cancer detection, monitoring of disease burden and treatment response, and personalized therapy selection through minimally invasive blood tests. Now, a similar transition is taking place in neurodegenerative diseases, particularly Alzheimer's disease (AD), with the development of blood-based assays for earlier disease detection in primary care settings rather than late-stage disease detection in specialty settings. Earlier detection of AD can lead to timely intervention, especially with the recent FDA approvals of new drugs, which, along with lifestyle changes, medications, or other therapies, might slow the progression of the disease or alleviate symptoms.
The recent advancements in blood-based assays mark a significant step forward in our ability to manage this complex condition. However, as we integrate these innovative technologies into clinical practice, several challenges arise—particularly concerning the selection and viability of biomarkers and the performance characteristics of these tests.
As a regulatory consultant collaborating with diagnostic companies on liquid biopsy assays in oncology and neurodegenerative diseases, I have observed several critical issues that need to be addressed to ensure these tests are reliable, accurate, and suitable for clinical use.
1. Selection of Biomarkers
One of the primary challenges in the development of blood-based assays for AD is selecting the most appropriate biomarkers. Ideal biomarkers should be highly specific to the pathological features of AD. Alzheimer's disease is characterized by the presence of amyloid-beta plaques and tau protein tangles in the brain, but the translation of these markers into blood-based assays has been difficult with complexity. While those biomarkers like p-tau217, the Aβ42/Aβ40 ratio and NFL show promise due to their strong association with amyloid pathology, the optimal combination of biomarkers for earlier AD detection remains a topic of ongoing research and the variability in their performance across different demographic and ethnic groups presents a substantial hurdle.
Most importantly, the pathophysiological complexity of AD means that no single biomarker may suffice. A combination of biomarkers might be necessary to achieve the required diagnostic specificity and sensitivity. This approach requires multiplexing immunoassay technologies capable of simultaneously measuring several biomarkers with high precision. Addiitonally, the sensitivity and specificity of these markers can vary significantly, influenced by genetic factors, underlying health conditions, and even lifestyle factors that are not yet fully understood. For instance, chronic conditions like kidney disease can skew biomarker levels, leading to potential misdiagnosis or false positives. This variability necessitates a cautious approach to biomarker selection, emphasizing the need for extensive validation studies that incorporate diverse population groups to ensure the reliability and accuracy of these tests. Choosing the wrong set of biomarkers can lead to suboptimal test performance and potentially misdiagnose individuals. Moreover, the selection of appropriate biomarkers for blood-based detection of AD is a technically demanding process. The technical challenge lies in ensuring these biomarkers' stability and detectability in blood, where they are often present only in trace amounts.
2. Viability of Biomarker Strategy for Blood
Blood presents as an excellent less invasive medium for biomarker detection, primarily due to its accessibility and cost-effectiveness. However, the biological complexity of blood, coupled with the dilute nature of neurological biomarkers in the plasma or serum, adds layers of complexity to assay development. Unlike CSF or PET imaging, which directly measure brain pathology, blood-based biomarkers are indirect measures of AD pathology. This indirectness can introduce variability and potential confounding factors, such as the influence of peripheral processes on biomarker levels.
Additionally, the dynamic range of biomarker concentrations in the blood is much lower compared to CSF, making it more challenging to develop sensitive and specific assays. This involves optimizing assay conditions to reduce background noise and enhance the signal from low-abundance biomarkers. Moreover, the pre-analytical and analytical variable, ranging from the method of blood collection and processing to the stability of biomarkers in blood, can significantly impact the test outcomes. Pre-analytical variables such as sample handling, processing, and storage need rigorous standardization to prevent degradation of biomarkers and ensure reproducibility of the test results. Assay developers must carefully validate their biomarker strategies and demonstrate a strong correlation between blood biomarker levels and brain pathology to ensure the reliability of their tests.
3. Performance Characteristics for Risk Prediction
Integrating AI/ML-driven algorithms to classify risk scores based on blood biomarkers introduces another layer of complexity. These algorithms must be trained on large, diverse datasets to accurately predict AD risk. However, the performance of these algorithms can vary widely depending on the characteristics of the training data and the chosen cut-offs for risk prediction. The selection of cut-off values is particularly critical, as it determines the sensitivity and specificity of the test.
When assessing the performance characteristics of blood-based AD tests, particularly those utilizing AI/ML-driven algorithms for risk prediction, two primary technical considerations emerge: the test's ability to rule in or rule out disease and its predictive value under different clinical scenarios. For both applications, the predictive values, PPV and NPV, are crucial. These values vary depending on the prevalence of AD in the tested population and dictate the clinical utility of the test. Technically, optimizing these values involves not only refining the test’s analytical accuracy but also customizing the algorithm’s threshold settings based on demographic and clinical data to enhance decision-making accuracy.
In assessing the performance characteristics necessary for blood-based assays in AD detection, specific benchmarks are recommended to ensure clinical utility. According to the consensus from the Global CEO Initiative, for a blood-based test to function effectively as a triaging tool, it should exhibit a sensitivity of at least 90% and a specificity that varies depending on the context of use ≥85% in primary care and potentially lower (≥75–85%) in specialized care settings where follow-up confirmatory tests are readily available. For confirmatory testing, where the blood test might directly influence treatment decisions without further invasive testing, the recommended sensitivity and specificity are both approximately 90%, underscoring the need for high accuracy to avoid false positives and negatives that could lead to inappropriate treatment choices (Nature Reviews Neurology, Consensus Statement).
Building on this understanding of necessary benchmarks, while the performance characteristics of FDA-authorized CSF tests for classifying amyloid PET status are publicly documented, currently, there is no FDA authorization for blood-based biomarker tests for AD, which leaves us with the lack of regulatory benchmarks on the agency’s expectations for sensitivity and specificity thresholds. Moreover, the dynamic nature of machine learning models means that they continuously evolve as new data becomes available. This may necessitate the adoption of Pre-determined Change Control Plans (PCCP) to secure approval from regulatory agencies, along with the necessary clearances for the test.
Importantly, to ensure the reliability and applicability of blood-based biomarker tests for AD, it is essential that their validation be conducted with clinical samples that accurately represent the intended use population. Often, samples collected at later stages of the disease or from secondary care settings do not adequately represent the broader population of individuals with early symptoms of the disease who might be tested in a primary care context. This discrepancy can lead to biased data that does not truly reflect the actual performance of the test. Biomarker levels can vary significantly based on the stage of the disease and the health status of the individual, which means that tests validated using samples from later stages or specialized care settings may not perform with the same accuracy when used earlier in the disease. Therefore, a rigorous validation process using appropriately diverse and representative clinical samples is critical to ensure that these tests provide accurate and unbiased results for intended users.
Conclusion
In conclusion, blood-based assays for early AD detection have the potential to transform the diagnosis and management of this devastating neurodegenerative disease, much like how liquid biopsies have been revolutionizing oncology. However, several key challenges must be addressed to ensure the reliability and clinical utility of these tests. By carefully considering biomarker selection, biomarker strategy viability, and performance characteristics, and by engaging in collaborative efforts with stakeholders and regulatory bodies, assay developers can work towards developing robust blood-based tests that can improve the lives of countless individuals affected by AD.
