Research Article: MortalityPredictors.org: a manually-curated database of published biomarkers of human all-cause mortality

Date Published: August 31, 2017

Publisher: Impact Journals LLC

Author(s): Maximus V. Peto, Carlos De la Guardia, Ksenia Winslow, Andrew Ho, Kristen Fortney, Eric Morgen.

http://doi.org/10.18632/aging.101280

Abstract

Biomarkers of all-cause mortality are of tremendous clinical and research interest. Because of the long potential duration of prospective human lifespan studies, such biomarkers can play a key role in quantifying human aging and quickly evaluating any potential therapies. Decades of research into mortality biomarkers have resulted in numerous associations documented across hundreds of publications. Here, we present MortalityPredictors.org, a manually-curated, publicly accessible database, housing published, statistically-significant relationships between biomarkers and all-cause mortality in population-based or generally healthy samples. To gather the information for this database, we searched PubMed for appropriate research papers and then manually curated relevant data from each paper. We manually curated 1,576 biomarker associations, involving 471 distinct biomarkers. Biomarkers ranged in type from hematologic (red blood cell distribution width) to molecular (DNA methylation changes) to physical (grip strength). Via the web interface, the resulting data can be easily browsed, searched, and downloaded for further analysis. MortalityPredictors.org provides comprehensive results on published biomarkers of human all-cause mortality that can be used to compare biomarkers, facilitate meta-analysis, assist with the experimental design of aging studies, and serve as a central resource for analysis. We hope that it will facilitate future research into human mortality and aging.

Partial Text

Mortality biomarkers are of great clinical and research interest. General clinical applications include identifying high-risk patient groups, prognosticating for individual patients, and helping healthcare providers decide among treatment options [1]. Examples of very well-studied mortality biomarkers include blood pressure, cholesterol, and waist circumference, which have well-established relationships with mortality in various populations documented in dozens of studies, some with thousands or millions of participants [2-4]. These traditional biomarkers have been joined in more recent years by many biomarkers utilizing modern assays, for example genome-wide methylation levels [5], cell-free DNA concentration [6], and leukocyte telomere length [7].

MortalityPredictors.org is a unique resource encompassing 1,587 entries from over 30 years of published research, and catalogs a diverse array of biomarkers, reports details of their relationships with all-cause mortality, and includes relevant meta-data. While certain biomarker databases already exist, for example the Infectious Disease Biomarker Database (IDBD; http://biomarker.cdc.go.kr; [14]) and the Early Detection Research Network Biomarker Database (EDRNBD; http://edrn.nci.nih.gov/biomarkers), these are limited to specialized applications or specific diseases and do not share the generalizability of our mandate, which emphasizes a) all-cause mortality and b) human studies in population-based or relatively disease-free cohorts. Some research groups have performed systematic reviews and/or meta-analyses on biomarkers of all-cause mortality, but these have been either within specific domains [15], or limited to very specific time-frames or purposes [1], and have not made their results available as a public database. Other groups have evaluated large panels of candidate biomarkers within specific cohorts, for example 106 metabolites in 10 thousand participants in the Estonian Biobank (Fischer, 2015), but these represent individual studies with little attempt to summarize or collate the work of others. Finally, a number of aging-related databases exist as well, including Human Aging Genomic Resources (HAGR; http://genomics.senescence.info/; [16]), GeroProtectors database (http://geroprotectors.org/; [17]), and the JenAge Aging Factor Database (AgeFactDB; http://agefactdb.jenage.de/; [18]), but none focus on mortality biomarkers and the curation of reported associations.

Mortality biomarkers have played a major role in facilitating healthcare and research for many decades, and promise to have an ever greater role in the near future, given an aging population in many countries, as well as the rise of frailty and age-related deterioration as health as major public health concerns and active targets for new therapies. Studies evaluating therapies that aim to address the roots of frailty within the aging process itself will benefit greatly from research to identify reliable predictors of mortality. These include the landmark Targeting Aging with Metformin (TAME) trial, now underway, as well as future research into promising therapies such as rapamycin and senescent cell removal. The creation of better predictive models for mortality begins with comprehensive, public data about prior biomarker research, toward which effort we contribute MortalityPredictors.org. Our future research will leverage high-throughput “-omics” technologies to screen large numbers of predictors and derive their combinations that most accurately predict mortality. These will be combined with the best predictors from prior research in order to build and validate powerful multi-predictor models that we hope will accelerate clinical research efforts to target and reduce mortality in older adults.

 

Source:

http://doi.org/10.18632/aging.101280

 

Leave a Reply

Your email address will not be published.