Predicting Intramyocardial Hemorrhage Volume from Serial Troponin Kinetics

Background: Intramyocardial hemorrhage (IMH) can complicate ST‑segment elevation myocardial infarction treated with primary percutaneous coronary intervention (PCI) [1]. Clinical studies have reported that post-PCI troponin levels are significantly higher in patients who develop IMH than in those without IMH [2]. Prior work showed a correlation between troponin peak value and IMH Volume, however, the focus was on IMH occurrence rather than its volume. To date no study has investigated which troponin kinetics features are optimal for predicting IMH volume.

Methods: We analyzed 183 patients with acute myocardial infarction who underwent troponin sampling at multiple time points (0-48hours) and T2* cardiac MRI to quantify IMH volume. Troponin kinetics features were derived from the time series (Figure 1). Illustrative examples of troponin kinetics features alongside late gadolinium enhancement and T2* images of two patients with and without IMH are shown (Figure 2). Relationships between troponin kinetics features and IMH volume are presented (Figure 3), highlighting which metrics increase monotonically with IMH burden and which show non‑linear trends. 5-fold cross‑validation is used to assess the ability of a random‑forest classifier [3] to predict IMH volume size categories of low(< 1 %), medium(1–9 %), high( >9 %) from troponin kinetics features.

Results: Early troponin exposure was the strongest correlate of IMH volume. The 12–24‑h AUC and 0–12‑h AUC displayed the highest positive correlations with IMH volume (R≈0.62 and R≈0.61), followed by peak troponin value (R≈0.57) and maximum rise slope (R≈0.50). Later AUC (24–48 h), ratio of the 1‑h value to peak, log decay rate and 10–90 % rise time showed weaker or U‑shaped associations. A tuned random‑forest classifier trained to predict IMH volume size categories attained a 5-fold cross‑validation mean accuracy of 74.3 %.

Conclusion: Previous work demonstrated that elevated troponin levels can predict the occurrence of IMH after reperfused myocardial infarction. In this study, we investigated the ability of troponin kinetics features to predict IMH volume percentage. We found that early troponin exposure and peak values correlate with IMH volume and hold predictive value of its size. Our findings suggest that detailed troponin kinetics may aid risk stratification beyond simple peak or single‑time‑point measurements and could complement imaging‑based assessments to identify patients at risk for significant intramyocardial hemorrhage.

Figure 1. Troponin kinetics feature maps. Example troponin time series (blue curve) with annotated features used in the study, including Peak value, time-to-peak, rise 10–90% duration and max rise slope, and AUC 0–12 h, AUC 12–24 h, and AUC 24–48, and the 1 h/peak ratio and post-peak log-linear decay rate.
Figure 2. Illustrative examples. This figure shows two illustrative examples of patients with negative (-) and positive (+) IMH post-PCI outcomes. Left Column: Troponin kinetics feature maps. Middle columns: short-axis CMR at 48–72 h—LGE and T2*. Top row (IMH −Ve): no hypointense myocardial region on T2*. Bottom row (IMH +Ve): hypointense region (arrowheads) consistent with intramyocardial hemorrhage (IMH Volume: 11%).
Figure 3. Relationship between troponin kinetics and intra myocardial hemorrhage (IMH) volume. Early troponin exposure is most informative: AUC of 12–24 h (R ≈ 0.62) and AUC of 0–12 h (R ≈ 0.61) show the strongest positive correlations with hemorrhage volume, followed by peak troponin (R≈ 0.57) and the maximum rise slope (R ≈ 0.50). Late exposure (AUC of 24–48 h) has a weaker association (R ≈ 0.19). By contrast, time to peak (R ≈ –0.10), the ratio of the 1 h value to peak (R ≈ –0.04), the log decay rate (R ≈ –0.18) and the 10–90 % rise time (R ≈ –0.10) exhibit weak negative correlations and show non linear trends. The panels collectively highlight that larger IMH volumes are accompanied by higher and more sustained troponin release, whereas timing related metrics change only modestly or in complex ways.