Development and Validation of Risk Stratification for Heart Failure After Acute Coronary Syndrome Based on Dynamic S100A8/A9 Levels

Figures

• 

  Figure 1. Study design. ACS indicates acute coronary syndrome; CABG, coronary artery bypass grafting; HF, heart failure; and PPCI, primary percutaneous coronary intervention.

• 

  Figure 2. Association of longitudinal S100A8/A9 levels with HF events in the longitudinal subset of discovery cohort. A, S100A8/A9 levels on admission and on days 1, 2, 3, and 4 after admission (following primary percutaneous coronary intervention). Data are presented as mean and mean±SD of patients who had HF event (red) and patients who did not have HF event (blue). Fold changes in S100A8/A9 between the 2 groups at each time are shown. B, Associations between the incidence of HF events and S100A8/A9 levels at different time points were calculated using univariate Cox proportional risk regression models. C, Predictive efficacy of the increase in S100A8/A9 levels at 1 to 4 days after admission and the area under the S100A8/A9 curve at 5 time points over the admission S100A8/A9 level was assessed using Harrell C statistic calculated from the Cox regression model and receiver operating characteristic. ***P<0.001. 0d S100A8/A9 indicates S100A8/A9 levels at admission; 1d S100A8/A9, S100A8/A9 levels at 1 day of admission; 2d S100A8/A9, S100A8/A9 levels at 2 days of admission; 3d S100A8/A9, S100A8/A9 levels at 3 days of admission; 4d S100A8/A9, S100A8/A9 levels at 4 days of admission; AUC…

• 

  Figure 3. Risk‐assessment tool for defining low‐ or high‐risk clinical events using day 1 S100A8/A9 cutoff concentrations. S100A8/A9 concentrations at 1 day after admission are shown for the entire discovery cohort (n=1062). The innermost circle shows the set low‐risk threshold for S100A8/A9 levels at day 1 postadmission (in nanograms per milliliter). The second circle shows the rate of total HF events. The third circle shows the rate of in‐hospital HF events. The fourth circle shows the proportion of patients designated as having low (A) or high (B) risk if either value is lower than or equal to the cutoff, and high risk if the value is greater than the cutoff. The outermost circle shows the NPV (A) or PPV (B) of S100A8/A9 on 1 day of admission. Based on the cutoff points of S100A8/A9, the corresponding NPV and PPV are indicated by bold black lines in (A) and (B), respectively. The darker the shade of blue, the lower the concentration of S100A8/A9, and the darker the shade of orange, the higher the concentration of S100A8/A9. HF indicates heart failure; NPV, negative predictive value; and PPV, positive predictive value.

• 

  Figure 4. Incidence of HF events compared with the levels of day 1 S100A8/A9. Kaplan‐Meier curves illustrating the timing of HF events in the 3 risk strata of S100A8/A9 levels at 1 day of admission in the discovery cohort (A), validation cohort 1 (B), and validation cohort 2 (C). HF indicates heart failure; and Interm, intermediate.

• 

  Figure 5. Prognostic impacts of β‐blockers on 1‐year HF events according to IPTW analysis in the low‐risk group and the intermediate‐high‐risk group of HF events. A, Patients with ACS without left ventricular dysfunction at discharge (n=2344). B, Patients with ACS with left ventricular dysfunction at discharge (n=864). ACS indicates acute coronary syndrome; HF, heart failure; and IPTW, inverse probability treatment weighting.