Publication highlights update - October 2024

Protein-based signatures to predict disease risk

Advancements in proteomic technologies have increased the precision
and scalability of protein analysis, enabling major population health studies and better disease risk predictions by identifying actionable biomarkers and providing new insights into disease biologically. Olink’s PEA technology has been utilized in several recent studies that have resulted in the identification of protein-based risk models that enhance or even surpass existing models based on clinical factors and polygenic risk scores (PRS). This blog post presents four recent examples of high-impact publications illustrating this increasingly common application of protein biomarker research.

If you are interested in reading more about this subject, please read our new white paper, “Integrating Polygenic Risk Scores and Proteomics: A New Paradigm for Predicting Health Risks” – DOWNLOAD NOW

Proteomic signatures improve risk prediction for common and rare diseases

A group from the MRC Epidemiology Unit at Cambridge University used data from the UK Biobank Pharma Proteomics Project (UKB-PPP) to to find “sparse protein-based predictors” of 5-20 proteins that could predict 10-year incident disease for rare and common diseases. They wanted to identify signatures based on limited numbers of proteins that could  eventually be translated into a clinical setting, and to evaluate their predictive power compared to individual and combined models incorporating clinical parameters, current clinical biomarkers, and polygenic risk scores (PRS). Remarkably, they could identify  67 sparse prediction models that provided clinically relevant improvements over current models.

Key highlights:

• As well as out-performing clinical models, the 67 multi-protein models substantially improved them in combined models (median improvement ~7%).
• Of 23 diseases with available PRSs in the UKB data, the genetic data only improved the basic clinical model in 7 cases, with a median improvement of just 3% (compared to 8% for the protein models for these 7 diseases).
• 6 of the models could be compared with Olink data from a much smaller, independent cohort and all were validated with high concordance (r=0.81 for predictive accuracy and r=0.97 for improvement of clinical models).
• While the UKB-PPP dataset was only trained for 10-year prediction, the models retained good performance for 20-year disease incidence in the validation cohort.

Proteomic aging clocks predict risks of aging-related diseases

Researchers from Oxford University developed an accurate, protein-based aging clock using data generated with the Olink® Explore 3072 high-throughput proteomics platform. After measuring ~3,000 proteins in over 51,000 individuals of UK European, Finnish, and Chinese genetic backgrounds from 3 independent cohorts, machine learning was applied to derive a 204-protein “ProtAge” model that could predict chronological age with almost 95% accuracy.

Key highlights:

• Further biostatistical analysis reduced the initial model to a more practicable aging clock comprising of just 20 proteins (“ProtAge20”) that retained ~95% of the original prediction performance.
• Cell adhesion, ECM interactions, immune response, hormone regulation, enzymatic activity, energy balance and neuronal structure & function were identified as key processes related to these proteins.
• The ProtAge score showed significant associations with 18 major chronic diseases, spanning CVD, metabolic diseases, organ-specific conditions, and cancer, as well as to all-cause mortality risk.
• Unlike recently described DNA methylation aging clocks, this protein-based score was strongly inversely associated with telomere length, a key cellular hallmark of aging.

Inflammatory proteins predict CVD risks in obese individuals

Scientists from Boehringer Ingelheim used the Olink Target 96 Inflammation panel to measure plasma proteomics in 6662 participants from the Gutenberg Health Study (GHS) to better understand the increased risk of cardiovascular disease (CVD) posed by obesity. They aimed to delineate an obesity-related inflammatory protein signature (OIPS) and evaluate its association with all-cause mortality, cardiac-specific cause of death, major adverse cardiovascular events (MACE), and incident coronary artery disease (CAD).

Key highlights:

• Analysis of baseline proteomics identified a 21-protein OIPS significantly associated with both obesity measures and subsequent CVD outcomes.
• Association of the signature with all-cause mortality, cardiac death, and incident coronary artery disease was independent of established risk factors.
• While many of the OIPS markers have been individually linked to obesity before, AXIN1 was a completely novel finding that may shed new light onto the underlying mechanisms.
• The OIPS model was successfully validated in an external cohort with very high correlation of inflammatory protein scores (r=0.93) and similar predictive values for CVD outcomes seen in both cohorts.

Protein signatures predict risk of myeloid neoplasms

Clonal hematopoiesis (CH) is an age-related condition that is believed to underlie myeloid neoplasms (MN), but this progression only occurs in a small proportion of cases. To address the lack of predictive tools, researchers from Washington University in St. Louis used the UKB-PPP dataset to identify proteins associated with MN and/or CH.

Key highlights:

• Plasma proteomics identified 115 MN-associated markers, 34 of which were also associated with CH, with enrichment for pathways related to both innate and adaptive immune regulators.
• A multi-protein model significantly improved MN risk prediction (AUC=0.85) compared to a composite model of clinical factors & CH status (AUC= 0.80).
• Extending the analysis to >380,000 UKB participants using genetically predicted protein levels from UKB-PPP proteogenomic data identified 10 markers with significant associations to MN risk.
• Mendelian Randomization analysis indicated that most of these proteins are likely downstream markers of pathways involved in MN pathogenesis, but that 2 (F7 & IL-17RA) are likely causally involved in progression to MN.