Patients and clinical characteristics
Overall, 47.4% patients (n = 295) were DM, 52.6% patients (n = 328) were nonDM according to diatebes status, and 48.5% patients (n = 302) were FCR (rSSQFR=0), 51.5% patients (n = 321) were FIR (rSSQFR>0) according to rSSQFR value (Fig. 1). The distribution of rSS and rSSQFR were shown in the violin plot (Supplemental Fig. 1). Functional CR in the rSSQFR distribution was not shown because it had the rSSQFR value of 0.
Clinical characteristics of patients are presented in Table 1. Results suggested that patients with nonDM + FIR had more hypertension (P = 0.011), aspirin use (P = 0.044) and P2Y12 inhibitors use (P = 0.044) than nonDM + FCR in nondiatebes cohort. In diatebes cohort, patients with DM + FIR had lower estimated glomerular filtration rate (eGFR) (P = 0.001) and more insulin use (P = 0.021) than DM + FCR. Besides, in the FCR layering, patients with DM + FCR had more previous MI (P = 0.032), higher low-density lipoprotein cholesterol (LDL-C) (P = 0.008), higher high-density lipoprotein cholesterol (HDL-C) (P = 0.003), higher hemoglobin A1c (HbA1c) (P < 0.001) and higher fasting blood glucose (FBG) (P < 0.001) than nonDM + FCR. In the FIR layering, patients with DM + FIR had more smoking history (P = 0.023), higher cardiac troponin I(cTNI) (P = 0.009), lower eGFR (P = 0.013), higher HbA1c (P < 0.001), higher FBG (P < 0.001) than nonDM + FIR.
There are some differences between diatebes cohort and nondiatebes cohort such as age (P = 0.017), current smoking (P = 0.001), TG (P = 0.003), LDL-C (P < 0.001), HDL-C (P = 0.002), cTNI (P = 0.003), HbA1c (P < 0.001), and FBG (P < 0.001). Other baseline clinical characteristics were not comparable (P > 0.05).
Procedural characteristics
Procedural characteristics of patients are presented in Table 2. Patients with nonDM + FIR had more 3-vessel disease (P = 0.004), higher SS (P < 0.001), higher rSS (P < 0.001), higher SSQFR (P < 0.001), higher rSSQFR (P < 0.001), more non-infarcted vessel number (P < 0.004), higher non-infarcted vessel with initial thrombolysis in myocardial infarction (TIMI) flow grade ≤ 1 (P < 0.001), higher non-infarcted vessel with diameter stenosis (DS) ≥ 90 (P < 0.001), higher non-infarcted vessel location at left circumflex artery (LCX) (P < 0.048), higher non-infarcted vessel with QFR ≤ 0.8 (P < 0.001), higher non-infarcted vessel with QF ≤ 0.8 location at left anterior descending artery (LAD) (P < 0.001), higher non-infarcted vessel with QFR ≤ 0.8 location at LCX (P < 0.001), and higher non-infarcted vessel with QFR ≤ 0.8 location at right coronary artery (RCA) (P < 0.001) than nonDM + FCR when compared nondiatebes cohort. Situation of diabetes cohorts is similar to that of nondiatebes cohorts, except for DM + FIR had smaller stent diameter (P = 0.020) and longer stent length (P = 0.040) than DM + FCR.
When compare FCR layering, results showed DM + FCR had more non-infarcted vessel with QFR ≤ 0.8 location at LCX (P = 0.008), and more number of stents (P = 0.024) than nonDM + FCR. Patients with DM + FIR had higher SSQFR (P < 0.001), higher rSSQFR (P < 0.001), higher non-infarcted vessel with QFR ≤ 0.8 (P < 0.001), and higher non-infarcted vessel with QFR ≤ 0.8 location at RCA (P = 0.008) than nonDM + FIR. Other baseline procedural characteristics were not comparable (P > 0.05). (Table 2).
In addition, the clinical and procedural characteristics were taken into univariate and multivariate logistic regression analyses to identify the independent predictors of FIR. Results showed hypertension (P = 0.003), SS (P < 0.001), rSS (P < 0.001), SSQFR (P < 0.001), non-infarcted vessels of DS ≥ 90 (P = 0.036), and non-infarcted vessels at LCX (P = 0.002) were associated with FIR in nondiatebes cohort, while eGFR (P = 0.041), SS (P < 0.001), rSS (P < 0.001), SSQFR (P < 0.001), non-infarcted vessels at LCX (P < 0.001) were associated with FIR in diatebes cohort (Supplemental Fig. 2). The reason for the correlation between LCX and FIR may be that the volume of LCX is smaller than the other two blood vessels, leading to a decrease in QFR [17].
Clinical outcomes
99.0% (617/623) patients were not lost during the clinical follow-up with a median follow-up period of 36 months. Overall, incidence of MACE in diatebes cohort was higher than that of nondiatebes cohort (22.9% vs 13.6%, HR = 1.79, 95%CI = 1.23–2.58, P = 0.002) (Fig. 2A). And incidence of MACE in the FIR layering was also higher than that of in the FCR layering (24.0% vs 12.6%, HR = 1.96, 95%CI = 1.35–2.83, P < 0.001) (Fig. 2B).
In the nondiatebes cohort, kaplan-Meier curves showed that nonDM + FIR patients had a higher 3-year MACE incidence (18.5% vs 9.8%, HR = 1.98, 95%CI = 1.06–3.72, P = 0.032) as compared to nonDM + FCR (Fig. 3A). In the diatebes cohort, DM + FIR patients also had a higher 3-year MACE incidence (27.9% vs 16.1%, HR = 1.76, 95%CI = 1.11–2.80, P = 0.017) as compared to DM + FCR (Fig. 3B). In the FCR layering, DM + FCR had a similar incidence of MACE with nonDM + FCR (P = 0.085) (Supplemental Fig. 3A), while in the FIR layering DM + FIR had a little higher incidence than nonDM + FIR (P = 0.044) (Supplemental Fig. 3B).
In addition, components of MACE included cardiac death, TVR, non-TVR, rehospitalization due to UAP, and non-fatal MI were also analysed. On the whole, incidence rate of rehospitalization due to UAP (19.1% vs 8.8%, HR = 2.22, 95%CI = 1.46–3.38, P < 0.001) and non-fatal MI (4.0% vs 90.7%, HR = 4.01, 95%CI = 1.46–11.05, P = 0.007) in the diatebes cohort were higher than nondiatebes cohort (Fig. 2A). The incidence rate of non-TVR (7.2% vs 2.3%, HR = 2.83, 95%CI = 1.38–5.80, P = 0.004), rehospitalization due to UAP (17.3% vs 10.9%, HR = 1.61, 95%CI = 1.06–2.45, P = 0.026), and non-fatal MI (3.8% vs 1.0%, HR = 3.18, 95%CI = 1.15–8.75, P = 0.026) in the FIR layering were higher than FCR layering (Fig. 2B).
Among them, nonDM + FIR and nonDM + FCR had difference in incidence rate of non-TVR (7.0% vs 6.9%, HR = 4.34, 95%CI = 1.51–12.51, P = 0.007), and non-fatal MI (5.4% vs 0%, HR = 5.75, 95%CI = 1.64 − 0.13, P = 0.006) (Fig. 3A), but DM + FIR and DM + FCR had no statistical difference about components of MACE (Fig. 3B). In the FCR layering, DM + FCR had no statistical difference about components of MACE with nonDM + FCR (Supplemental Fig. 3A), while DM + FIR had a higher incidence of rehospitalization due to UAP than nonDM + FIR (22.4% vs 9.9%, HR = 2.24, 95%CI = 1.31–3.82, P = 0.003) in the FIR layering (Supplemental Fig. 3B).
A multivariate Cox regression model was used to find independent predictors of MACE. Results showed that after adjustment for baseline clinical differences, non-infarcted vessel with DS ≥ 90 (HR = 1.56, 95%CI = 1.05–2.32, P = 0.027), DM (HR = 1.60, 95%CI = 1.03–2.49, P = 0.036), and FIR (HR = 1.71, 95%CI = 1.13–2.57, P = 0.011) were independent predictors for overall patients (Fig. 4). Non-infarcted vessel with DS ≥ 90 (HR = 1.70, 95%CI = 1.05–2.76, P = 0.030), and FIR (HR = 1.69, 95%CI = 1.01–2.83, P = 0.045) were independent predictors of 3-year MACE for DM cohort (Fig. 4). There was no independent predictor for nonDM cohort.
Receive operating characteristic curve (ROC) curves were plotted to find a better indicators for discrimination of 3-year MACE. Result showed that when adding clinical risk factors, rSSQFR in the nonDM + FIR group had a statistical difference with clinical risk factors alone ((area under curve) AUC = 0.715, 95%CI = 0.596–0.835 vs AUC = 0.613, 95%CI = 0.474–0.752, P = 0.040) and clinical risk factors + rSS (AUC = 0.715, 95%CI = 0.596–0.835 vs AUC = 0.621, 95%CI = 0.479–0.764, P = 0.045) (Fig. 5A).
Similarly, clinical risk factors + rSSQFR showed the biggest AUC when compare with clinical risk factors alone (AUC = 0.812, 95%CI = 0.750–0.874 vs AUC = 0.666, 95%CI = 0.581–0.751, P < 0.001) and clinical risk factors + rSS (AUC = 0.812, 95%CI = 0.750–0.874 vs AUC = 0.672, 95%CI = 0.587–0.757, P = 0.003) in the DM + FIR group (Fig. 5B). Due to the rSSQFR value of 0 in the FCR group, there is no analyticity in nonDM + FCR group and DM + FCR group.