You might also read
Articles linked to this work by shared authors, journal, and citation graph.
Updated: Apr 22, 2026

Selecting Multiple Biomarker Subsets with Similarly Effective Binary Classification Performances
Published on: October 11, 2018
Nazem Atassi1, James Berry2, Amy Shui2
1From the Neurological Clinical Research Institute (NCRI), Department of Neurology (N.A., J.B., A. Sherman, E.S., J.W., I.K., M.C.), and the Biostatistics Center (A. Shui, D.S.), Massachusetts General Hospital, Boston; and Prize4Life (N.Z., M.L.), Cambridge, MA. natassi@partners.org.
This study describes the creation of a large, open-access database that combines information from many past clinical trials for amyotrophic lateral sclerosis (ALS). By merging data from over 8,600 patients, researchers identified that higher baseline levels of uric acid, creatinine, and body mass index are linked to slower disease progression and longer survival.
Area of Science:
Background:
No comprehensive repository existed previously to synthesize fragmented clinical trial information for amyotrophic lateral sclerosis. Researchers lacked a unified platform to analyze longitudinal patient data across multiple completed studies. That uncertainty drove the development of a centralized resource to improve understanding of disease phenotypes. Prior research has shown that individual trials often remain isolated, limiting the statistical power needed for robust clinical insights. This gap motivated the consolidation of diverse datasets into a single, standardized framework. Investigators aimed to overcome these barriers by pooling information from sixteen phase II/III trials and one observational study. The resulting collection provides a unique opportunity to explore biological markers on a large scale. No prior work had resolved how disparate trial metrics could be harmonized to reveal consistent predictive patterns in patient outcomes.
Purpose Of The Study:
The aim of this work is to pool data from completed clinical trials to create an open-access resource for the research community. Investigators sought to enable a greater understanding of the phenotype and biology of this condition. This effort addresses the challenge of fragmented information across isolated studies. That uncertainty drove the team to consolidate longitudinal records into a single, standardized database. Researchers intended to provide a platform that supports large-scale analysis of patient outcomes. By merging diverse datasets, the authors aimed to identify consistent biological markers of disease progression. This initiative was motivated by the need for more robust statistical power in clinical research. The study establishes a foundation for future investigations by making these merged records publicly available for the first time.
Main Methods:
Review approach involved aggregating longitudinal records from sixteen phase II/III trials alongside one observational study. Investigators standardized diverse metrics to ensure consistency across the entire merged dataset. The team utilized mixed effects models to evaluate the monthly rate of decline for functional and respiratory indicators. Researchers applied Cox regression techniques to interpret survival outcomes across the patient population. Statistical rigor was maintained by implementing a Bonferroni correction to account for multiple testing. The design focused on creating an open-access resource to support broader scientific inquiry. This approach allowed for the synthesis of over 8 million de-identified data points. The final architecture provides a unified platform for examining clinical and laboratory variables from thousands of participants.
Main Results:
Key findings from the literature indicate that higher baseline uric acid levels predict a slower drop in functional scores and vital capacity. Statistical analysis showed these associations were significant, with p-values of 0.01 and less than 0.0001 respectively. Higher creatinine levels at baseline also correlated with a slower decline in functional scores and vital capacity. These results were supported by p-values of 0.01 and less than 0.0001. The study found that elevated body mass index at baseline is associated with longer survival duration. This specific survival benefit reached a significance level of p less than 0.0001. The average monthly decline for the functional rating scale was 1.02 points. Vital capacity decreased at an average rate of 2.24 percent of predicted values per month.
Conclusions:
The authors propose that their merged repository serves as a valuable asset for future investigations into disease progression. Synthesis and implications suggest that baseline metabolic markers provide significant prognostic information for patients. Researchers observed that higher uric acid levels correlate with improved functional outcomes and extended life expectancy. Similarly, the study highlights that elevated creatinine at the start of observation predicts a more favorable disease trajectory. The team notes that body mass index acts as a robust indicator for survival duration. These findings demonstrate the utility of large-scale data integration in identifying potential biomarkers. The investigators emphasize that their resource facilitates broader access to clinical information for the scientific community. They conclude that these identified predictors warrant further validation to refine clinical prognostic models for this condition.
The researchers propose that higher baseline levels of uric acid, creatinine, and body mass index independently predict slower functional decline and increased survival duration. This mechanism involves metabolic factors influencing the rate of disease progression measured by standardized clinical scales.
The database, known as the Pooled Resource Open-Access ALS Clinical Trials (PRO-ACT) resource, integrates longitudinal information from sixteen phase II/III trials and one observational study. It contains over 8 million de-identified data points from more than 8,600 individual participants.
The authors utilized mixed effects models to characterize the rate of decline in the Revised ALS Functional Rating Scale and vital capacity. Cox regression models were necessary to analyze survival data, ensuring statistical rigor across the large, merged dataset.
The researchers standardized longitudinal clinical and laboratory data, including demographics and family histories, across diverse trials. This standardization role allowed for the successful merging of disparate datasets into a single, cohesive, and accessible repository for the scientific community.
The team measured disease progression using the Revised ALS Functional Rating Scale and vital capacity. They observed a monthly decline rate of 1.02 points for the functional scale and 2.24 percent for vital capacity, providing a quantitative baseline for disease trajectory.
The researchers propose that their repository is the largest publicly available collection of merged clinical trial data for this condition. They imply that this resource enables greater understanding of the biology and phenotype of the disease by providing unprecedented access to longitudinal records.