Saliva produces measurable biomarker signals, but acute myocardial infarction diagnosis is not there yet
A paired-matrix study in acute myocardial infarction tested whether saliva could carry the same biochemical information clinicians usually seek in venous blood. Investigators measured high-sensitivity cardiac troponin I (hs-cTnI), Apelin, Elabela, Endoglin, Irisin, and ischaemia-modified albumin in paired blood and saliva samples from 60 selected patients who presented within six hours of symptom onset and 30 healthy controls.
The design is clinically appealing because the emergency department rarely has time to spare when chest pain suggests evolving infarction. Saliva collection avoids needles, can be repeated easily, and may eventually support rapid triage in settings where venous access is delayed or difficult. The study, however, stays in exploratory territory. The findings show measurable signals, not a ready-to-use bedside test.
Biomarker shifts tracked across both matrices
In blood, Irisin, Elabela, and Apelin were lower in patients, while Endoglin was higher. All of those differences reached statistical significance, with p values at or below 0.020. In saliva, Endoglin was higher and Irisin lower, and both of those differences were highly significant with p values below 0.001. After Bonferroni correction and age restriction, only those salivary differences remained significant, with p values at or below 0.030.
Ischaemia-modified albumin did not differ in either matrix.
That negative result matters because it limits the temptation to overread the panel as a broad marker set for myocardial injury. The study does not show that every candidate biomarker behaves well in saliva. It shows a narrower pattern: some analytes moved in parallel with infarction, while others did not separate patients from controls in this setting.
Troponin separated groups, but the signal reflects selection rather than validation
Salivary hs-cTnI assay signals completely separated patients from healthy controls. The interpretation has to be disciplined, because elevated serum hs-cTnI was an inclusion criterion and no salivary assay was matrix-validated. In other words, the separation reflects the way participants were chosen, not proof that salivary hs-cTnI can independently diagnose infarction.
That distinction is central for emergency medicine. A biomarker can look excellent inside a selected cohort and still fail when tested in undifferentiated chest pain, where unstable angina, pericarditis, pulmonary embolism, reflux, musculoskeletal pain, and anxiety can all cloud the clinical picture. The study acknowledges that problem directly by limiting its claim to preliminary evidence.
Correlation was strong, agreement was not
Blood and saliva values correlated strongly across the analytes, with Spearman rs ranging from 0.773 to 0.979 and all p values below 0.001. Strong correlation suggests the matrices move in the same direction, but correlation is not the same as interchangeability. Bland–Altman analysis showed proportional differences and wide limits of agreement.
That pattern is exactly what makes translation to the bedside difficult. A saliva assay that tracks the same biological trend as blood may still be too imprecise to replace plasma or serum if the two matrices diverge by a predictable but clinically large amount. Wide limits of agreement leave uncertainty around any single patient result, especially when clinicians must decide quickly whether to activate reperfusion pathways or continue observation.
The adipokine panel did not beat hs-cTnI in exploratory cross-validation
The adipokine panel did not improve on hs-cTnI in the exploratory cross-validation analysis. This finding tempers enthusiasm for a multi-analyte saliva panel as an immediate substitute for established cardiac testing. The biology behind the panel still makes sense: myocardial ischemia and injury can influence inflammatory signaling, vascular remodeling, and metabolic mediators. Yet biological plausibility alone does not guarantee better diagnostic performance.
Endoglin and Irisin emerged as the most consistent salivary signals after adjustment, but the study does not establish a diagnostic threshold, a clinical cutoff, or a usable algorithm for triage. It also does not compare saliva against standard emergency-department serial troponin testing, electrocardiography, or imaging. Those are the benchmarks that matter when chest pain is on the line.
What this paired-matrix study can and cannot support
This work offers a careful proof-of-signal rather than proof-of-clinical utility. The six analytes generated measurable signals in saliva, and the blood-saliva relationship was strong enough to justify continued study. The data also suggest that Endoglin and Irisin may deserve further attention if future assays can show stable performance across broader patient groups.
At the same time, the study remains limited by its exploratory design, its selected infarction cohort, and the absence of matrix-specific validation for salivary hs-cTnI. Healthy controls are not the same as patients arriving with undifferentiated chest pain. Without that harder comparison, sensitivity, specificity, positive predictive value, and real-world false-positive rates remain unknown.
What needs to happen before saliva enters myocardial infarction triage
Future work will need prospective evaluation in unselected emergency patients, not just in confirmed infarction cases. Matrix-specific assay validation is essential, because a biomarker measured in saliva can behave differently from the same biomarker measured in serum. Reproducibility, sample handling, timing from symptom onset, and performance across age groups will also shape whether any saliva-based strategy can compete with current practice.
The study was funded by TÜBİTAK and conducted with ethics approval and informed consent, but the clinical question now is narrower: can salivary biomarkers add actionable information in the first hours of suspected infarction without creating diagnostic noise? This report does not answer that question. It shows where the signals are, and just as clearly where the clinical boundary still stands.