There were a total of 382 consenting patients with acute coronary syndrome undergoing isolated first-time CABG with at least one available NT-proBNP as follows: preoperative (n=366), postoperative day 1 (POD1; n=320) and postoperative day 3 (POD3; n=325) and data from all three time points available in 267 patients. Preoperative, intraoperative and postoperative characteristics of the 382 patients are presented in Table 1 and Table 2.
Overall NT-proBNP increased from 420 [150-970] ng/L preoperatively to 2065 [1324-3650] ng/L (p<0.001) POD1 and to 3610 [2167-6010] ng/L (p<0.001) POD3.
Overall 88 patients (23%) were treated with inotropes at some stage intraoperatively or postoperatively (Additional table S1). Only 33 of these patients fulfilled criteria for PHF and these patients had significantly worse outcome and more pronounced increase of NT-proBNP postoperatively (table S1).
Overall 35 patients (9%) from the whole cohort fulfilled criteria for PHF. Seven of these patients were also classified to have severe PHF. In two patients who fulfilled criteria for PHF myocardial dysfunction was mild and resolved without inotropes.
Patients with PHF had a more pronounced risk profile preoperatively and extended cross-clamp and CPB times intraoperatively. Postoperatively they had more signs of myocardial injury, higher incidence of acute kidney injury, extended ventilation time, prolonged ICU stay and a higher hospital mortality compared to those without PHF. Clinical outcomes were further aggravated in patients with severe PHF (Table 2).
Postoperative NT-proBNP in relation to PHF
Patients with PHF had higher pre- and postoperative levels of NT-proBNP compared to those without PHF (Table 1, Figure 1).
After adjusting for glutamate treatment and known preoperative non-cardiac confounders age, eGFR, female gender and obesity, NT-proBNP POD1 was 1.46 times higher in patients with PHF than in patients without PHF (adjusted coefficient 0.165, 95%CI 0.062- 0.269, p=0.002; Additional table S2). Interaction of glutamate was not statistically significant and would have changed the adjusted coefficient for PHF by 3% if kept in the final model.
After similar adjustment for glutamate treatment and known preoperative non-cardiac confounders, NT-proBNP POD3 was 1.54 times higher in patients with PHF than in patients without PHF (adjusted coefficient 0.188, 95%CI 0.188- 0.289, p<0.0001; Additional table S3). Interaction of glutamate was not statistically significant and would not have changed the adjusted coefficient for PHF if kept in the final model.
NT-proBNP on POD1 demonstrated significant discrimination for PHF (AUC 0.70; 95% CI 0.61-0.79; p<0.0001). The best cut-off value of 1836 ng/L had a sensitivity of 90% and a specificity of 46% (Figure 2A). A similar discrimination was found for NT-proBNP on POD3 (AUC 0.70; 95% CI 0.60-0.81; p<0.0001). The best cut-off value 6065 ng/L had a sensitivity of 57% and a specificity of 79% (Figure 2B).
Postoperative changes of NT-proBNP
NT-proBNP increased postoperatively in almost all patients with the highest values recorded on POD3 (Table 1, Figure 1). The postoperative increase of NT-proBNP was significantly more pronounced in patients with PHF and the postoperative changes of NT-proBNP were associated with PHF.
Postoperative increase of NT-proBNP from preoperative level to POD3 demonstrated significant discrimination for PHF (AUC 0.68; 95% CI 0.56-0.79; p=0.002, best cut-off 7639 ng/L with a sensitivity of 40% and a specificity of 92%). Similar discrimination was found for postoperative change of NT-proBNP from preoperative level to POD1 (AUC 0.66, 95% CI 0.56-0.76, p=0.004, best cut-off 1372 ng/L with a sensitivity of 87% and a specificity of 46%) and for postoperative change of NT-proBNP from POD1 to POD3 (AUC 0.63; 95% CI 0.50-0.76, p=0.028, best cut-off 4299 ng/L with a sensitivity of 38% and a specificity of 90%).
In the multivariable logistic regression analysis, delta Troponin T POD3-Pre ng/L, delta NT-proBNP POD3-POD1 ≥ 4299 ng/L and severe LV dysfunction emerged as independent risk factors for PHF (Table 3). The univariable Odds ratios for variables tested in are shown in Additional table S4.
Postoperative NT-proBNP in relation to severe PHF
The highest pre- and postoperative NT-proBNP values were recorded in patients with severe PHF (Table 1, Figure 1).
Patients with severe PHF had significantly higher NT-proBNP preoperatively (1920 [1030-4202] v 750 [300-1265] ng/L, p=0.022) and on POD1 (5040 [3060-10200] v 2740 [1875-4600] ng/L, p=0.028) compared to patients with PHF that was not classified as severe.
Patients with severe PHF had significantly higher NT-proBNP preoperatively (1920 [1030-4202] v 380 [140-864] ng/L, p=0.001), on POD1(5040 [3060-10200] v 2040 [1260-3440] ng/L, p=0.001 and on POD3 (11680 [6070-17914] v 3450 [2003-5550] ng/L, p=0.012) compared to patients without PHF.
After adjusting for glutamate treatment and known preoperative non-cardiac confounders age, eGFR, female gender and obesity, NT-proBNP POD1 was 2.18 times higher in patients with severe PHF than in patients without PHF (adjusted coefficient 0.339, 95%CI 0.134-0.543, p=0.001; Additional table S5). Interaction of glutamate was not statistically significant would have changed the adjusted coefficient for severe PHF by 6% if kept in the final model.
After similar adjustment for glutamate treatment and known preoperative non-cardiac confounders, NT-proBNP POD3 was 1.81 times higher in patients with severe PHF than in patients without PHF (adjusted coefficient 0.258, 95%CI 0.042-0.474, p=0.019; Additional table S6). Interaction of glutamate was not statistically significant would have changed the adjusted coefficient for severe PHF by 0.3% if kept in the final model.
NT-proBNP on POD1 demonstrated significant discrimination for severe PHF (AUC=0.86; 95% CI 0.76-0.95; P=0.001). The best cut-off value of 4575 ng/L had a sensitivity of 71% and a specificity of 84% (Figure 3A). A similar discrimination was found for NT-proBNP on POD3 (AUC=0.79; 95% CI 0.55-1.00; P=0.015) (Figure 3B). The best cut-off value of 6065 ng/L had a sensitivity of 83% and a specificity of 77%.
The number of events was too few to permit multivariable analysis of risk factors for severe PHF.
Patients with NT-proBNP above the cut-offs had more pronounced Troponin T elevations, higher incidence of acute kidney injury, extended ventilation time, prolonged ICU stay and higher hospital mortality (Supplemental tables S7-S8).
Circulatory support and variables associated with PHF in incremental quartiles of postoperative NT-proBNP
The incidence of inotropic treatment, the use of more than one drug and the average duration of treatment was higher in patients with NT-proBNP POD1 and POD3 in the upper quartiles. Extended ICU stay and ventilator support, and renal dysfunction, were also more common in the upper quartiles of postoperative NT-proBNP levels.
Patients with PHF and severe PHF belonged to the upper quartiles of postoperative NT-proBNP. A notable exception was one patient who was in severe heart failure preoperatively. This patient improved markedly after surgery and despite fulfilling study criteria for severe PHF plasma levels of NT-proBNP decreased from 9250 ng/L preoperatively to 4640 ng/L POD1 and 1590 ng/L POD3 (lowest quartile).
Detailed results regarding circulatory support and postoperative outcome in incremental quartiles of postoperative NT-proBNP levels are presented in Table 4 and 5.