Prevalence, Clinical Characteristics and Prognosis of Vascular Disease in Valvular Heart Surgery: A Multi-Centre Study

Study population

From January 2010 to December 2021, 3879 patients who presented for valvular heart surgery at two tertiary referral centres in Hong Kong (Queen Mary Hospital, Queen Elizabeth Hospital) were recruited into the Chinese Valvular Heart Disease Study (CVATS) database and retrospectively analysed. Patients were excluded if clinical, surgical, or follow-up data were incomplete (n = 36). 3843 patients in total were included in the final analysis. This study was part of the Chinese Valvular Heart Disease Study evaluating the epidemiology, pathophysiology and outcomes in Chinese patients with VHD (13), and was approved by the Institutional Review Board of the Hong Kong Hospital Authority (Hong Kong West Cluster).

Clinical and surgical parameters

Baseline history of comorbidities (diabetes, hypertension, hyperlipidaemia, atrial fibrillation, heart failure and chronic renal disease) was obtained from the Clinical Data Analysis and Reporting System (CDARS), a centralised inter-hospital electronic database, at the time of surgery. Baseline prescription of antihypertensive, lipid-lowering and antiplatelet medications was reviewed, along with preoperative New York Heart Association (NYHA) functional status and laboratory data. Valvular lesions, left ventricular ejection fraction (LVEF) and systolic pulmonary artery pressure (SPAP) were documented during preoperative echocardiography and confirmed during surgery. Surgical risk was evaluated using the logistic EuroSCORE (14).

Vascular disease was defined as the presence of one or more of the following at baseline, based on clinical diagnosis, angiographic findings or intervention need: coronary artery disease (CAD), ischaemic cerebrovascular accidents (CVA; including ischaemic stroke and transient ischaemic attack) and peripheral vascular disease (PVD). Polyvascular disease was defined as vascular disease involving ≥2 vascular beds.

Outcomes

The primary outcome in this study was postoperative major adverse cardiac events (MACE), defined as the composite of all-cause mortality, heart failure (HF) readmission, myocardial infarction and stroke, diagnosed based on the ninth revision of the International Classification of Diseases (ICD-9). The secondary outcome was all-cause mortality.

Statistical analysis

Continuous data were presented as mean ± standard deviation, while categorical data were presented as frequencies and percentages. Comparisons between continuous variables were done with the Kruskal–Wallis rank sum test, while Fisher’s exact test or Pearson’s χ² test was used for categorical variables as appropriate. Multivariable Cox proportional hazard regression analyses were performed to assess the association between vascular disease and postoperative MACE and mortality, with four models incrementally adjusting for baseline demographics, comorbidities, medications and surgical risk factors. For postoperative MACE, Fine-Gray analysis was also performed to account for non-cardiovascular death as a competing risk.

Subgroup analyses were also performed according to vascular disease subtype and valve operated. Linear regression was performed to investigate potential collinearities between age and all comorbidities, medications and surgical risk factors.

Risk attenuation analysis was used to determine the extent of risk associated with vascular disease attributable to other risk factors. Percentage attenuation of associations between vascular disease and postoperative outcomes was measured as [β_{unadjusted model} – β_{adjusted model})/β_{unadjusted model}] × 100 (%), with β denoting regression coefficients of different adjustment models. 95% confidence intervals (95% CI) around percentage attenuation values were derived from 1000-time bootstrap resampling. Formal risk reclassification analyses were performed by calculating continuous net reclassification improvement (cNRI) for MACE and mortality. Nested Cox regression analysis was used to determine the incremental prognostic value of vascular disease compared to various risk prediction models. All statistical analyses were performed using R version 4.2.1. A two-tailed p-value < 0.05 denoted statistical significance.

Baseline characteristics

Of 3843 patients, 1266 (33%) had vascular disease in ≥1 vascular bed, including 207 (5.4%) with polyvascular disease (16.4% of those with vascular disease). The distribution of different vascular disease subtypes is shown in Figure 1. Baseline clinical characteristics are shown in Table 1. Patients with vascular disease were older and more often male, with higher BMI, smoking rates, cardiovascular comorbidities and medications, with a graded increase from no vascular disease to polyvascular disease.

Figure 1

Surgical characteristics

Table 2 summarises surgical risk factors and characteristics according to vascular disease status. Patients with vascular disease had poorer symptomatic status, higher surgical risk and longer operation times, especially those with polyvascular disease. In terms of valvular pathology, the prevalence of aortic stenosis and degenerative disease was highest in patients with polyvascular disease and lowest in those without vascular disease. Congenital lesions and mitral valve pathologies were more common in patients without vascular disease. Compared to those with monovascular disease, patients with polyvascular disease more frequently received concomitant CABG during valvular heart surgery.

Outcomes

During a median follow-up period of 6.37 years (IQR: 3.40–9.54 years), 1176 MACE (23%) [576 (45%) with vascular disease; 600 (23%) without vascular disease, p < 0.001] and 713 deaths (19%) [359 (28.4%) with vascular disease; 354 (14%) without vascular disease, p < 0.001] were recorded.

Patients with polyvascular disease experienced the highest rates of postoperative MACE and all-cause mortality, followed by patients with monovascular disease and no vascular disease (p < 0.001; Figure 2). Postoperative rates of HF readmission, myocardial infarction and stroke were also greatest in those with polyvascular disease (Supplementary Figure 1).

Figure 2

After adjustment for baseline demographics (age, sex) (Model 1), comorbidities (diabetes, hypertension, atrial fibrillation, hyperlipidaemia, heart failure, chronic renal disease and smoking; Model 2), medications (antihypertensives, antiplatelets, statins and warfarin; Model 3) and surgical risk factors (NYHA functional class, LVEF, logistic EuroSCORE, multi-valve surgery, concomitant aortic surgery and prior revascularisation; Model 4), both monovascular disease and polyvascular disease remained independently associated with both postoperative MACE and all-cause mortality (Table 3; Supplementary Table 1 and Supplementary Figure 2).

Polyvascular disease was independently associated with a greater risk of MACE than monovascular disease [adjusted HR: 1.25 (1.01–1.54, p = 0.036)].

The percentage risk attenuation of MACE associated with vascular disease was 31.3%, 38.8%, 39.0% and 49.0% by Models 1–4, respectively (Figure 3a), while the percentage risk attenuation of mortality was 43.2%, 51.1%, 43.2% and 53.6% by Models 1–4, respectively (Supplementary Figure 3a). Adding vascular disease yielded significant reclassification improvement and incremental prognostic value for MACE and all-cause mortality to models combining demographic, clinical and surgical risk factors (Figure 3b; Supplementary Figure 3b). There was no significant collinearity between adjusting risk factors in model development (R² = 0.241, adjusted R² = 0.236, p < 0.001; Supplementary Figure 6), and no significant overfitting was observed when vascular disease was added to Model 4 on calibration (Supplementary Figure 7).

Figure 3

Subgroup analysis

All three vascular disease subtypes (CAD, ischaemic CVA and PVD) were associated with increased rates of MACE and all-cause mortality (Supplementary Figures 4a and 4b). Subgroup analysis including only patients with vascular disease showed that ischaemic CVA was associated with independently higher rates of MACE (Table 4; Supplementary Figure 5a), while CAD was associated with lower MACE risk (Table 4; Supplementary Figure 5b). Ischaemic CVA and PVD remained independently associated with both MACE and all-cause mortality after adjustment (Supplementary Table 2) and extracardiac vascular disease was associated with greater MACE risk than CAD [adjusted HR: 1.56 (1.27–1.92)].

Subgroup analysis according to valve operated showed independent associations between monovascular disease and both mortality and MACE in aortic and mitral valve surgeries (Supplementary Table 3). Polyvascular disease was associated with MACE in both aortic and mitral valve surgeries.

Clinical implications

Surgical risk prediction scores such as EuroSCORE II and STS score are widely used to estimate short-term morbidity and mortality after valvular heart surgery (11). However, numerous studies observe notable underestimation of long-term mortality (33), especially in those undergoing valvular heart surgery (34). Moreover, while both scores include vascular disease as a predictor of operative risk, the additive risk posed by polyvascular disease is not well defined in either model, and their utility in predicting long-term postoperative outcomes remains suboptimal (34). With population ageing and continued expansion of surgical indications, more robust risk stratification is necessary to cater to growing patient demands and increasing prevalence of polyvascular disease.

Vascular disease, prevalently seen yet often overlooked in valvular heart surgery, confers risk largely uncaptured by current risk prediction algorithms and substantially improves prediction of long-term postoperative outcomes in this underrepresented population. In particular, our study demonstrates the compounding effect of polyvascular disease compared to monovascular disease on long-term postoperative outcomes, an aspect largely omitted in current risk scores. Our findings suggest that identifying occult vascular and polyvascular disease before valvular heart surgery may enable more accurate long-term risk stratification, which may inform more proactive and diverse anti-atherosclerotic strategies, including preoperative statin therapy (35), exercise and prehabilitation (36, 37).

Limitations

This was a retrospective cohort study with flaws and biases inherent to study design. The presence of vascular disease may have been influenced by demographic confounders or medical confounders as well as insufficient screening, despite extensive efforts to adjust for such factors in multivariable analysis. Limited patients with vascular disease were on antiplatelet or lipid-lowering medications at baseline likely from widespread intolerance of adverse effects in our locality (38, 39), which may introduce unmeasured confounding and overestimation of its impact on postoperative outcomes despite comprehensive adjustment. Owing to the study period, there was insufficient data documenting predicted mortality with newer surgical risk scores such as EuroSCORE II. Nevertheless, predictive accuracy was similar between logistic EuroSCORE and EuroSCORE II (11). Patients in this study were mostly Asian, and validation in cohorts of other ethnicities or with different burdens of VHD is warranted, although the prognostic significance of vascular disease persisted across various subgroups in this study.