Proteomics Profiling Reveals Plasma Biomarkers of Hypertrophic Cardiomyopathy
Key findings
- In a case-control study that comprehensively profiled the plasma proteome, researchers discovered proteins that accurately distinguished patients with hypertrophic cardiomyopathy from controls
- Multiple proteins significantly correlated with objective and subjective markers of disease severity, notably troponin I and New York Heart Association functional class
- In an analysis using the 50 most discriminative proteins, the Ras-MAPK pathway and its upstream and downstream pathways were upregulated in patients with hypertrophic cardiomyopathy
Experts often find it challenging to diagnose hypertrophic cardiomyopathy (HCM). Distinguishing HCM from other conditions that cause morphological changes in the heart often requires genetic testing, cardiac magnetic resonance imaging or, if an athlete's heart is suspected, deconditioning. Each of these modalities has serious limitations.
Subscribe to the latest updates from Cardiovascular Advances in Motion
Michael A. Fifer, MD, co-director of the Hypertrophic Cardiomyopathy Program and medical director of the Knight Center for Interventional Cardiovascular Therapeutics at the Corrigan Minehan Heart Center at Massachusetts General Hospital, Yuichi J. Shimada, MD, MPH, of Columbia University and colleagues are working to find a set of plasma biomarkers for HCM. In the Journal of Cardiovascular Translational Research, they report on the first investigation to discover proteins that distinguish HCM from controls and correlate with known indicators of advanced disease.
Proteomics Profiling
The researchers analyzed blood samples from 15 patients at Mass General who had clinically overt HCM and 22 control patients without HCM who were being followed in the cardiology clinic.
Proteomics profiling was done with the SOMAscan, which can measure 1,129 proteins in a small amount of plasma (about 50 microliters). This assay is established to be highly sensitive and reproducible, and it has already been used to discover biomarkers in non-cardiac conditions including Duchenne muscular dystrophy, Alzheimer's disease and lung and prostate cancer.
Identification of Candidate Proteins
Using supervised machine learning, the researchers identified 50 candidate proteins that had the highest potential to discriminate cases from controls (These are listed in the paper). The overall proteomic profile was distinctly different between the cases and controls.
As a sensitivity analysis, a different list of the 50 most discriminant proteins was made using a similar form of machine learning. There was 94% overlap between the two lists.
Multivariable Analysis
The researchers generated receiver operating characteristic curves using Monte Carlo cross validation (MCCV) with balanced subsampling. In each MCCV, randomly selected two-thirds of the samples were used to examine the importance of each protein, and the remaining one-third was used to validate the model created in the first step. The five most discriminant proteins were used to build the biomarker discrimination models.
These steps were repeated multiple times to calculate the average of the area under the curve and to determine which proteins were most frequently selected in the five-protein discrimination model. Models with 10, 25, 50 and 100 proteins were also developed to compare discriminative accuracy (the number of accurately classified samples divided by the total number of samples).
The average area under the curve was 0.94. The discriminative accuracy of the five-protein model was 89%, higher than that of the other models.
As a sensitivity analysis, the researchers used data from the 27 male participants for derivation and data from the 10 female patients for validation. Nine of the female participants were classified correctly (discriminative accuracy of 90%).
Pathway and Network Analyses
The researchers then determined how the original 50 most discriminant proteins were associated with pathways in the Kyoto Encyclopedia of Genes and Genomes database. The Ras pathway and multiple upstream signaling pathways related to cell proliferation and angiogenesis were significantly upregulated in HCM cases compared with controls.
A network analysis was performed to reveal interactions between the discriminant proteins. The researchers expected 48 interactions, but the analysis demonstrated 98, and proteins that map to the Ras-MAPK pathway were at the hub of the network.
Correlation with Known Indicators of HCM Severity
The 50 most discriminant proteins were then tested for significant correlations with conventional biomarkers and clinical signs of HCM:
- Troponin I–13 proteins correlated, 12 of which were concordant (proteins increased in HCM had a positive correlation and those decreased in HCM had negative correlation)
- New York Heart Association (NYHA) functional class: 12 proteins correlated, all of which were concordant
- Left atrial diameter: 9 proteins correlated, 3 of which were concordant
- Interventricular septal thickness: 2 proteins correlated
- Brain natriuretic peptide: 0 proteins correlated
In a pathway analysis of the 23 proteins that correlated with either troponin I or NYHA class, the Ras pathway was again significantly upregulated, along with its upstream and downstream pathways. Network analysis of the proteins showed 51 interactions, compared with 16 expected, centered around members of the Ras-MAPK pathway.
Laying the Groundwork
These findings are an important step toward developing a panel of biomarkers that could be measured in blood. When combined with conventional clinical features, a diagnostic panel that is quick to perform, low-priced and readily obtainable would help physicians accurately and easily diagnose HCM.
view original journal article Subscription may be required
Learn more about the Hypertophic Cardiomyopathy Program at Mass General
Refer a patient to the Corrigan Minehan Heart Center