Clinical Proteomics: 9-Protein Panel Beats NT-proBNP
- Aug 17
- 4 min read
NT-proBNP anchors nearly every heart failure risk score, and it still leaves teams guessing about who dies within two years. A clinical proteomics study of 1,212 adults with reduced ejection fraction asked whether broader plasma protein profiling closes that gap. Patane and colleagues measured 734 circulating proteins and derived a nine-protein panel that beat NT-proBNP alone. The instructive part is where the gain vanished: chronic Chagas cardiomyopathy, the group they most wanted to help.
Key Takeaways
A nine-protein plasma panel classified two-year mortality in heart failure 20 percent better than NT-proBNP, and replicated in UK Biobank.
Clinical proteomics panels trained on mixed-etiology cohorts can fail inside subgroups; this one lost 16 percent in Chagas cardiomyopathy.
Chagas patients died at 26 percent over two years against 16 percent cohort-wide, arguing for etiology-specific risk models.
Pathway enrichment tied Chagas hearts to fibrosis and trafficking biology, which generates drug-target hypotheses rather than confirming them.
What Clinical Proteomics Added to a Standard Risk Model
Baseline plasma came from 1,212 adults with reduced ejection fraction, 191 with Chagas disease confirmed by dual Trypanosoma cruzi serology. Protein profiling ran on Olink Explore 384 panels, an affinity readout rather than a mass spectrometry workflow, which shapes how the numbers should be read. Feature selection landed on C1QA, CCL4, REN, EGLN1, COL9A1, GP1BA, ITM2A, CNPY2 and NT-proBNP. Proteome analysis at this scale builds on population work like Sun and colleagues (2023).
Key Findings
Nine proteins beat the standard marker: F1-macro reached 0.674 against 0.560 for NT-proBNP alone, with time-dependent discrimination up about 6 percent.
The gain concentrated in two etiologies: hypertensive heart failure improved 40 percent and ischemic 21 percent, the rest far less.
Chagas went the other way: in those 191 patients the panel scored 16 percent worse than NT-proBNP alone, despite carrying the highest mortality at 26 percent.
Replication outside the derivation cohort: in 431 UK Biobank participants F1-macro rose 18 percent, reaching 0.612 against 0.474 in the top-risk tertile.
Figure 1. Kaplan-Meier survival by heart failure etiology. Panel A plots two-year survival for the idiopathic, hypertensive, ischemic, Chagas, and alcoholic subgroups against the whole cohort. Panel B compares Chagas with non-Chagas etiologies, where the Chagas curve drops more steeply (p = 1.05e-05). Panels C through F repeat that comparison for each remaining etiology. Adapted from Patane et al. (2026), PLOS Neglected Tropical Diseases.
Why One Panel Cannot Cover Every Etiology
Pooling five etiologies into one derivation set optimizes for the average patient. Chagas patients are not average: they die faster, and the proteins tracking their risk differ from those tracking ischemic risk. That the panel lost ground in the sickest subgroup is the honest part of this paper, and the same caution belongs on cardiovascular signatures generally, including serum panels flagging arterial calcification.
Pathway Signals Behind Chagas-Specific Biology
Fourteen pathways were enriched only in the Chagas comparison, clustering around matrix remodeling, cytoskeletal signaling, secretory trafficking, and innate immunity. Chronic Trypanosoma cruzi infection leaves a fibrotic, inflamed myocardium, so that profile reads as coherent rather than surprising. The authors call the target-linked analyses exploratory, and the label is right: enrichment across 734 affinity-measured proteins nominates hypotheses, it does not establish mechanism.
Implications for Biomarker Programs
For teams building prognostic panels, the lesson is that the discovery cohort quietly sets the ceiling: mix five etiologies and you optimize for the average patient, not the hard one. We see this in client work at Dalton, where a signature that looks clean in a pooled cohort loses separation once samples are stratified by disease origin, which is why we pair affinity panels with mass spectrometry protein quantification before a panel is locked. That cross-check earns its cost in any multi-omics biomarker discovery program spanning mixed patient populations.
Frequently Asked Questions
What is clinical proteomics used for?
Clinical proteomics measures proteins in patient samples such as plasma or serum to find markers of diagnosis, prognosis, or drug response. It supports risk stratification, trial enrichment, and target work. The study here is prognostic: nine plasma proteins sorted two-year mortality better than one standard marker.
Can a plasma protein panel predict death in heart failure better than NT-proBNP?
In this cohort, yes for most etiologies: the nine-protein panel lifted F1-macro from 0.560 to 0.674 and held up in an independent UK Biobank sample. Chagas cardiomyopathy was the exception, where NT-proBNP alone performed better.
How many samples does a clinical proteomics biomarker study need?
This study used 1,212 plasma samples for discovery and 431 for external validation, a sensible order of magnitude for prognostic work with a hard clinical endpoint. Pilots of 100 to 200 samples rank candidates but rarely support subgroup analysis, which is where panels break.
Conclusion
A small plasma protein panel plausibly adds prognostic information beyond NT-proBNP in heart failure, and that addition survived an outside cohort. What isn't proven is that any single panel transfers into Chagas cardiomyopathy, where biology and performance push the other way. For programs in etiologically mixed disease, power the subgroups you care about, not the pooled cohort alone.
Related Reading
See how we run these analyses in one lab: Dalton's multi-omics CRO services.
Citation
Patane, J. S. L., Giugni, F. R., Rosa, R. S., Marcondes-Braga, F. G., Mansur, A. J., Pereira, A. C., & Krieger, J. E. (2026). Plasma proteomics improves risk prediction in heart failure and reveals unique biology in chronic chagas cardiomyopathy. PLOS Neglected Tropical Diseases, 20(6), e0014370. https://doi.org/10.1371/journal.pntd.0014370
Note
This blog post summarizes findings from the above-cited research. Figures are adapted from the original publication. For full details, please refer to the source article.
By Seungjun Yeo, CEO at Dalton Bioanalytics. Specializing in multi-omics mass spectrometry for drug discovery and biomarker research.