Characteristics of the patient population

After the screening of 42743 surgical interventions, 1590 patients (42%) were finally included. Baseline clinical characteristics and outcomes of the study population, stratified according to the use of stress test modality, are shown in Table 1. Patients who underwent SPECT-MPI stress strategy showed higher ASA class and more often reported the assumption of cardioprotective therapies (i.e., anti-platelets and lipid-lowering drugs).

Table 1 Clinical characteristics of the study population divided in two groups (no stress test vs. MPI stress test). Values are indicated as number (n) and percentage (%), or median (interquartile range)

The median age of the patients was 71 years (65–77), and 38.8% (617) were female. Regarding cardiovascular risk factors, 81.1% (1290) of the patients had hypertension, 69.2% (1100) dyslipidemia, and 32.7% (520) diabetes mellitus. Moreover, 43.7% (695) of patients had a history of CAD; moderate-to-severe CKD had a prevalence of 16.3% (260).

Concerning symptoms, 42.8% (680) of patients suffered from dyspnea, 4.2% (67) from typical angina, and 26.8% (426) from atypical chest pain.

Electrocardiogram and echocardiogram were available in all patients: 94% (1495) were in sinus rhythm and 6% (95) in atrial fibrillation. Median left ventricular ejection fraction (LVEF) was 55% (51–55). There was a slightly greater incidence of grade II diastolic dysfunction in the stress-tested group (25.9% vs. 21.3%, p 0.035), whereas grade III dysfunction was more prevalent in the non-stress group (2.9% vs. 1.5%, p 0.013).

Type of surgery and surgical risk scores ( Fig. 2)Fig. 2

Type of surgeries, preoperative risk evaluation, and 30-day cardiac events. The most prevalent type of non-cardiac surgery was vascular (19.9%), followed by orthopedic (18.6%). About the half of surgeries were at high risk. Revised cardiac index was high (≥ 3) in 37.3% of patients, and ASA class III was present in 48.5% of cases. The study population was divided in two groups (no stress test vs. imaging stress test). Primary composite endpoint and secondary endpoint (ACS) were significantly lower in the imaging stress-tested group. ACS acute coronary syndrome, ASA American Society of Anesthesiologists, CMR cardiac magnetic resonance, GYNECOL gynecological surgery, ORT orthopedic surgery, SPECT-MPI single-photon emission computed tomography myocardial perfusion imaging, VASC vascular surgery

The most prevalent type of NCS was vascular (n = 316, 19.9%), followed by orthopedic (n = 296, 18.6%) and urological (n = 278, 17.5%). Almost half of surgeries were at high risk (50,1%). RCRI was calculated for each patient: 37.4% (595) had an index of 1, 39.6% (630) of 2, and 23% (366) had an index of more than 3. With regard to ASA class, 12.5% (199) of patients were in class I; 35.7% (567) in class II; 48.5% (772) in class III, and 3.2% (51) in ASA class IV.

Non-invasive imaging cardiac test

A total of 669 patients underwent a stress test before NCS. Results and outcomes of sCMR and SPECT-MPI are shown in Table 2; baseline clinical characteristics are available in supplemental Table 1. A sCMR was performed in 287 patients. Mean LVEF was 58.7% (± 10), and late gadolinium enhancement was present in 39.7% of cases. Inducible ischemia was detected in 52/287 patients (18.1%). The median value of ischemic segments was 3 (2–4), with a total percentage of stress myocardial ischemia of 8.6% (± 3.4).

Table 2 MPI stress test results and outcomes for sCMR and SPECT-MPI. Values are indicated as number (n) and percentage (%), mean ± SD or median and interquartile range

A SPECT-MPI was performed in 382 patients and resulted positive for inducible ischemia in 49/382 cases (12.8%). The median value of reversible stress ischemic segments was 4 (2–4) with a total percentage of stress myocardial ischemia of 8.2% (± 4.4). Rest ischemia was detected in 31.2% of patients. Physical stress was performed in 57% of cases, while pharmacological in 43% of cases.

Indication to ICA after stress imaging was upon clinical judgment of the referring cardiologist.

Invasive coronary angiography

In the stress-tested group, a total of 105 patients (15.7%) underwent ICA before NCS and 82 (78% of the patients who underwent ICA) were revascularized. The rate of ICAs was not significantly different between sCMR and SPECT-MPI groups (18.8% vs. 13.3%, respectively, 0.054), despite a tendency toward higher rate of ICA in sCMR group with a similar rate of revascularizations (84.5% vs. 77.6%, respectively, p 0.448).

Myocardial ischemia predictors in patients undergoing non-cardiac surgery

Univariate and multivariate logistic regression analysis was performed to assess the association between clinical and echocardiographic parameters and the presence of myocardial ischemia in patients undergoing MPI stress test (Table 3 and supplemental Table 2). Independent predictors of myocardial ischemia were hypertension (OR 5.72, 95%CI 1.70–19.15, p 0.005), male sex (OR 1.74, 95% CI 1.09–2.79, p 0.020), moderate–severe CKD (OR 1.74, 95%CI 1.06–2.84, p 0.026), known CAD (OR 4.01, 95%CI 2.30–6.99, p < 0.001) and high-grade diastolic dysfunction (OR 1.85, 95%CI 1.30–2.64, p 0.001).

Table 3 Independent variables associated with the composite primary endpoint (3.A) and acute coronary syndromes (3.B). Predictors of ischemia at MPI stress test (3.C)Myocardial ischemia and revascularization

AUC and the optimal cut-points were calculated both for sCMR and SPECT-MPI to identify the best ischemic thresholds for revascularization (Fig. 3). Stress CMR showed an AUC of 0.95 (SE: 0.98, SP: 0.92) when a threshold for myocardial ischemia of 5.5% was chosen. On the other side, the best cut-point for SPECT-MPI was a percentage of ischemia of 7.5% (AUC: 0.83, SE: 0.78, SP: 0.87).

Fig. 3

ROC, AUC, and the optimal cut-points of sCMR and SPECT-MPI to predict coronary artery revascularization. AUC area under the curve, ROC receiver operating characteristics, sCMR stress cardiac magnetic resonance, SE sensitivity, SP specificity, SPECT-MPI single-photon emission computed tomography myocardial perfusion imaging

Cardiac events in non-cardiac surgery

The composite primary outcome occurred in 39 cases (2.4%) with a significantly greater incidence in the non-stress-tested group compared to the stress-tested group (respectively, 3.4% vs. 1.,2%, p 0.006; Table 1 and Fig. 2). Myocardial infarction was the most frequent cardiac complication (n = 20, 1.3% of cases), followed by unstable angina (n = 11, 0.6%), cardiac death (n = 3, 0.3%), cardiogenic pulmonary edema (n = 3, 0.3%), and cardiogenic shock (n = 2; 0.2%). ACS occurred more frequently in the non-stress-tested group (2.6% vs. 1.0%; p 0.027; Table 1 and Fig. 2).

Both primary outcome and ACS did not differ between sCMR and SPECT (primary outcome: 1.1% vs. 1.3%, p 0.756; ACS 0.7% vs. 1.3% p 0.441; Table 2).

After adjusting with multivariable logistic regression analysis, CAD (OR 2.33, 95%CI 1.11–4.88, p 0.025) was associated with an increased risk of the primary study endpoint, while higher METs grade (OR 0.26, 95%CI 0.12–0.55, p 0.001) and a stress test strategy (OR 0.33, 95%CI 0.15–0.76, p 0.009) were protective (Table 3 and supplemental Table 3). The independent protective effect of the MPI stress test strategy was consistent among all sub-categories, as shown by test for interaction (Fig. 4). Similarly, we performed a multivariable logistic regression analysis for the secondary endpoint ACS (Table 3). The protective effect of stress test strategy was confirmed also for ACS alone (OR 0.41, 95%CI 0.17–0.98, p 0.046).

Fig. 4

Test for interaction of imaging stress test strategy among different subgroups. The independent protective effect of the MPI stress test strategy was consistent among all sub-categories. CAD Coronary artery disease, CKD chronic kidney disease, CI confidence intervals, LVEF left ventricular ejection fraction, METS metabolic equivalents, NYHA New York Heart Association, OR odds ratio

Propensity score analysis

Propensity score matching was utilized to create a balanced comparison, resulting in a cohort of 669 matched pairs. After matching, the baseline characteristics between the stress-tested and non-tested groups showed no significant differences, with the largest standardized mean difference being 0.08 for the presence of diabetes. The lower rate of postoperative cardiovascular events in the matched stress-tested group suggests a potential benefit of preoperative stress testing in reducing these events. The C-index values for different models were as follows: non-MPI (0.65), sCMR with revascularization (0.85), sCMR without revascularization (0.80), SPECT with revascularization (0.82), SPECT without revascularization (0.75). ROC analysis demonstrated AUC values ranging from 0.66 for non-MPI to 0.86 for sCMR with revascularization. The DeLong test indicated that sCMR with revascularization had significantly higher AUC compared to SPECT with revascularization (p = 0.03). Post-matching analysis indicated that the rate of 30-day postoperative cardiovascular events in the stress-tested group was significantly lower than the non-tested group (OR 0.34; 95% CI, 0.15–0.78; p = 0.009) and similarly resulted for the incidence of myocardial infarction (p = 0.046). When examining the outcomes of patients with positive stress test results, 18.1% in the sCMR group and 12.8% in the SPECT-MPI group had inducible ischemia, leading to ICA in 78% and a subsequent revascularization rate of 84.5% for the sCMR group and 77.6% for the SPECT-MPI group. The matched cohort analysis mitigates the effects of referral bias and provided a more accurate estimate of the impact of preoperative stress testing, overcoming the selection bias intrinsic to the retrospective design of the study. The lower rate of postoperative cardiovascular events in the matched stress-tested group also confirms a potential benefit of preoperative stress testing in reducing these events.