Does the paradoxical effect of smoking on troponin impact the safety of the ESC 0/1h algorithm?

Abstract Introduction Paradoxically, recent population-based studies reported lower levels of high-sensitivity cardiac troponin (hs-cTn) in current smokers than in never and former smokers. This counterintuitive finding raises questions about the yet not investigated association between smoking and hs-cTn levels and its potential impact on the safety and accuracy of the ESC 0/1h algorithm in patients with suspected acute myocardial infarction (MI). Purpose We sought to investigate the impact of smoking on i) quantitative levels of hs-cTnI and T in patients with symptoms indicative of MI, and ii) the performance of the ESC 0/1h algorithm using both assays. Methods We used data from a prospective cohort study investigating patients with symptoms suggestive of an acute MI who presented to the emergency department of a German tertiary care center. Final diagnoses were adjudicated according to the 4th Universal Definition of MI. Patients with ST-elevation MI were excluded. The population was categorized into three subgroups according to self-reported smoking status, i.e. current, never, and former smokers. Diagnostic accuracies of hs-cTnI and T levels at 0h and 1h, as assessed by the area under the receiver operating curve (ROC-AUC), and the performance, i.e. safety, accuracy and efficacy, of the ESC 0/1h algorithm using both assays were compared between smokers and never/former smokers. Results Among 2,527 patients with suspected MI, 562 (22.2%) were current smokers, 1,352 (53.5%) never and 613 (24.3%) former smokers. The prevalence of MI was comparable between the three groups, while age as well as several cardiovascular risk factors differed across the smoking categories. Hs-cTn levels at baseline were quantitatively lower in current smokers (hs-cTnI 4.8 [2.3, 14.8] ng/L; hs-cTnT 7.0 [4.0, 16.0] ng/L) than in never (hs-cTnI 5.6 [2.6, 16.2] ng/L; hs-cTnT 9.0 [5.0, 21.0] ng/L) and former smokers (hs-cTnI 6.2 [2.9, 15.9] ng/L; hs-cTnT 10.0 [5.0, 23.0] ng/L). Diagnostic accuracies did not differ between current and never/former smokers regarding the ROC-AUC for non-ST-elevation MI, neither based on 0h hs-cTnI (AUC 0.89 [0.85-0.93] vs. 0.88 [0.86-0.90], p=0.72) and 1h hs-cTnI (AUC 0.94 [0.91-0.97] vs. 0.92 [0.90-0.94], p=0.22), nor on 0h hs-cTnT (AUC 0.87 [0.82-0.91] vs. 0.87 [0.85-0.89], p=0.79) and 1h hs-cTnT (AUC 0.94 [0.92-0.96] vs. 0.92 [0.90-0.94], p=0.086). Applying the ESC 0/1h algorithm using hs-cTnI provided a very high and comparable safety for ruling-out MI in current and never/former smokers. Rule-in accuracy was also similar in current and never/former smokers, while the observe zone was slightly larger in never/former smokers (Figure 1A). Similar findings were confirmed for hs-cTnT (Figure 1B). Conclusion Despite paradoxically lower hs-cTnI and T concentrations in current than in never or former smokers with suspected MI, the ESC 0/1h algorithm achieves very high and comparable safety, accuracy and efficacy irrespective of smoking status.