Nonlinear association and threshold effect between preoperative estimated pulse wave velocity and postoperative atrial fibrillation after coronary artery bypass grafting

Abstract

Background

Postoperative atrial fibrillation (POAF) is a common complication after coronary artery bypass grafting (CABG), yet reliable preoperative risk markers remain limited. Estimated pulse wave velocity (ePWV), a simple noninvasive indicator of arterial stiffness, may enhance risk stratification; however, its association with POAF has not been fully elucidated. This study aimed to investigate the nonlinear association and potential threshold effect between preoperative ePWV and POAF in patients undergoing CABG.

Methods

This single-center retrospective cohort study included 8,570 patients who underwent CABG between 2021 and 2022 at Beijing Anzhen Hospital. ePWV was calculated using age and mean blood pressure. Logistic regression, restricted cubic spline (RCS) modeling, and two-piecewise linear regression were used to evaluate the association and potential threshold effect between ePWV and POAF. Predictive performance was evaluated using receiver operating characteristic (ROC) analysis.

Results

POAF occurred in 7.9% of patients. Higher ePWV was independently associated with POAF [adjusted odds ratio (OR): 1.15, 95% confidence interval (CI): 1.09–1.21; P<0.001]. A significant nonlinear relationship was observed, with an inflection point at 11.4 m/s. Below this threshold, each 1 m/s increase in ePWV was associated with a 27% higher risk of POAF, whereas no significant association was observed at ePWV values ≥11.4 m/s. The area under the curve (AUC) for ePWV was 0.583, and the optimal cutoff of 9.62 meters per second yielded a negative predictive value of 94.6%.

Conclusions

ePWV shows an independent and nonlinear association with POAF after CABG, with identifiable thresholds that may inform preoperative risk stratification. As a simple and noninvasive measure of arterial stiffness, ePWV may assist in identifying patients at elevated risk of POAF. Validation in multicenter prospective cohorts is warranted to establish its broader clinical utility.

Highlight box.

Key findings

• In a large cohort of 8,570 patients undergoing coronary artery bypass grafting (CABG), higher preoperative estimated pulse wave velocity (ePWV) was independently associated with postoperative atrial fibrillation (POAF).

• A significant nonlinear association was identified, with an inflection point at 11.4 m/s. Below this threshold, each 1 m/s increase in ePWV was linked to a 27% higher risk of POAF.

• The optimal predictive cutoff was 9.62 m/s, providing a high negative predictive value (94.6%).

What is known and what is new?

• Arterial stiffness is a recognized contributor to postoperative complications, and carotid-femoral pulse wave velocity has been associated with increased POAF risk in previous cardiac surgery studies.

• However, the role of ePWV—a simple, noninvasive surrogate of arterial stiffness—has not been evaluated in the context of CABG.

• This study is the first to demonstrate a nonlinear, threshold-dependent relationship between preoperative ePWV and POAF, and to propose clinically meaningful cutoff values for risk stratification.

What is the implication, and what should change now?

• ePWV may serve as an accessible preoperative marker to identify patients at low, intermediate, and high risk for POAF.

• An ePWV <9.62 m/s suggests low risk and may help avoid unnecessary intensive prophylaxis, whereas values between 9.62–11.4 m/s identify patients who may benefit from closer monitoring and preventive strategies.

• Incorporating ePWV into preoperative evaluation may enhance individualized risk assessment and guide perioperative management in CABG candidates.

Introduction

Coronary artery bypass grafting (CABG) is an essential surgical treatment for severe coronary atherosclerotic heart disease and remains one of the most frequently performed cardiac surgeries worldwide (1). Postoperative atrial fibrillation (POAF) represents the most common complication after CABG, occurring in approximately 30% of patients. The development of POAF not only increases both early and late mortality, but also prolongs intensive care unit and overall hospital stay, leading to increased healthcare costs and worse postoperative outcomes (2-4). Despite its clinical relevance, the pathogenesis of POAF is multifactorial and incompletely understood, and reliable preoperative tools for prediction and risk monitoring are still lacking.

Arterial stiffness reflects structural and functional alterations of the arterial wall and is recognized as an important risk factor for cardiovascular and cerebrovascular diseases. Accumulating evidence also suggests that increased arterial stiffness contributes to postoperative complications in cardiac surgery (5-7). Arterial stiffness is typically assessed using carotid-femoral pulse wave velocity (cf-PWV) (8). Prior studies have demonstrated that each 1 m/s increase in cf-PWV corresponds to a markedly higher risk of POAF following aortic or valvular surgery, supporting a potential mechanistic link between aortic stiffness and postoperative arrhythmogenesis (9). However, findings from the Atherosclerosis Risk in Communities (ARIC) study revealed a U-shaped association between cf-PWV and atrial fibrillation, indicating that moderate arterial stiffness may be protective, whereas very high or very low arterial stiffness increases the risk of atrial fibrillation (10). These findings highlight the complexity of arterial stiffness as a predictor and underscore the need to further clarify its relationship with POAF following CABG.

The assessment of cf-PWV requires specialized instrumentation and experienced operators. Estimated pulse wave velocity (ePWV), derived from age and mean blood pressure, has emerged as a simple and noninvasive surrogate for arterial stiffness with good validity and reproducibility (11,12). However, the association between ePWV and POAF after CABG has not been examined in previous research. Given the strong correlation between ePWV and cf-PWV, and the established relevance of arterial stiffness in postoperative arrhythmias, it is plausible that elevated ePWV may be linked to a higher risk of POAF. This hypothesis requires confirmation in appropriately designed clinical studies.

Therefore, this study aimed to evaluate the association between preoperative ePWV and POAF after CABG, with particular focus on potential nonlinear relationships and threshold effects. Clarifying these patterns may help refine preoperative risk stratification and inform early strategies for POAF prevention. We present this article in accordance with the STROBE reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-aw-2385/rc).

Methods

Study population

This study was a single-center retrospective cohort study that included patients who underwent CABG at Beijing Anzhen Hospital. The study period, encompassing patient recruitment, exposure (defined as the CABG procedure), and data collection, spanned from January 2021 to December 2022. Patients were eligible if they were 18 years of age or older and had complete clinical information required for calculating preoperative ePWV. Exclusion criteria were preoperative hemodynamic instability, concomitant valvular or other cardiac surgical procedures, and incomplete medical records. Patients with echocardiographically documented valvular regurgitation were included only if the condition was managed medically and did not necessitate concomitant surgical intervention during the CABG procedure. A total of 8,570 patients met these criteria and were included in the final analysis.

The study protocol was reviewed and approved by the Ethics Committee of Beijing Anzhen Hospital (approval No. 2025283x). Given the retrospective study design, the requirement for written informed consent was waived. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Data collection and definitions

The primary exposure of this study was the preoperative ePWV. This parameter was calculated using the established formula incorporating patient age and mean blood pressure values, which were obtained from preoperative electronic medical records. The study population was stratified into two groups based on the median ePWV value (high ePWV group vs. low ePWV group) for intergroup comparative analyses.

Preoperative clinical characteristics, laboratory measurements, and operative data were systematically extracted from the hospital’s electronic medical record system. Baseline variables included age, sex, body mass index (BMI), and major comorbidities such as hypertension, diabetes mellitus, prior myocardial infarction, history of percutaneous coronary intervention, transient ischemic attack or stroke, and respiratory disease. Preoperative laboratory parameters included white blood cell count, hemoglobin, platelet count, serum creatinine, troponin I, myoglobin, B-type natriuretic peptide, international normalized ratio (INR), plasma D-dimer, and estimated glomerular filtration rate (eGFR). Operative variables, including the use of the internal mammary artery, number of grafts, estimated blood loss, and volume of red blood cell transfusion, were considered as potential confounders or predictors in the analysis.

POAF was the primary outcome and was defined as a new-onset episode of atrial fibrillation lasting at least 30 seconds, detected by continuous electrocardiographic monitoring or standard electrocardiography during the index hospitalization after CABG.

Statistical analysis

All statistical analyses were performed using R software (version 4.2.2). The distribution of continuous variables was assessed using the Shapiro-Wilk test. Continuous variables were expressed as mean ± standard deviation or median (interquartile range), and comparisons between groups were made using the t-test or the Mann-Whitney U test, as appropriate. Categorical variables were presented as counts (percentages) and compared using the χ² test or Fisher’s exact test.

Univariable and multivariable logistic regression analyses were used to identify independent factors associated with POAF, thereby reducing confounding bias by adjusting for a comprehensive set of clinical, operative, and laboratory variables. Odds ratios (ORs) and corresponding 95% confidence intervals (CIs) were calculated to assess the association between preoperative ePWV and POAF. Restricted cubic spline (RCS) models were applied to examine potential nonlinear relationships between ePWV and the outcome. When a nonlinear association was observed, a two-piecewise linear regression model was used to identify the inflection point and evaluate threshold effects. The predictive performance of ePWV for POAF was assessed using receiver operating characteristic (ROC) curves and the area under the curve (AUC), and the optimal cutoff value was determined by the Youden index. All statistical tests were two-sided, and a P value <0.05 was considered statistically significant.

Results

Patient demographics and baseline characteristics

A total of 8,570 patients who underwent CABG were included in this study. Based on the median preoperative ePWV value of 10.04 m/s, patients were divided into a low ePWV group (<10.04; n=4,301) and a high ePWV group (≥10.04; n=4,269). The mean age of the cohort was 62±9 years, and patients in the high ePWV group were significantly older than those in the low ePWV group (68±6 vs. 56±7 years, P<0.001). Men comprised 75.8% of the cohort, with a higher proportion in the low ePWV group (82.4% vs. 69.2%, P<0.001). BMI was also higher in the low ePWV group compared with the high ePWV group (26.0±3.4 vs. 25.6±3.2, P<0.001). The prevalence of hypertension, previous transient ischemic attack or stroke, and prior myocardial infarction was higher in the high ePWV group (65.5% vs. 55.1%; 14.5% vs. 10.8%; 16.9% vs. 11.9%; all P<0.001). There were also differences between the groups in diabetes mellitus, history of stent implantation, respiratory disease, alcohol consumption, smoking, and the severity of valvular regurgitation (all P<0.05). In terms of laboratory data, the high ePWV group had lower eGFR levels (82±15 vs. 91±16, P<0.001), and hemoglobin and liver function tests also differed between the groups (P<0.05). Significant differences were also noted in hemoglobin and liver function tests (P<0.05) (Table 1). Regarding operative characteristics, the high ePWV group had a lower rate of left internal mammary artery utilization (50.4% vs. 69.0%, P<0.001). The choice of conduit and number of grafts were individualized based on intraoperative assessment and patient-specific factors (Table 2). The incidence of POAF was higher in the high ePWV group (9.9% vs. 6.0%, P<0.001). Postoperative intra-aortic balloon pump support was slightly more frequent in the high ePWV group (6.5% vs. 4.7%, P<0.001). Other postoperative complications, including postoperative myocardial infarction, extracorporeal membrane oxygenation support, pulmonary infection, postoperative hemorrhage, and in-hospital mortality, did not differ significantly between the two groups (all P>0.05) (Table 2).

Table 1. Patient demographics and baseline characteristics.

Characteristic	Overall	Low ePWV (<10.04 m/s)	High ePWV (≥10.04 m/s)	P
No. of patients	8,570	4,301	4,269
Age, years	62±9	56±7	68±6	<0.001
Sex, male	6,497 (75.8)	3,542 (82.4)	2,955 (69.2)	<0.001
BMI, kg/m2	25.8±3.3	26.0±3.4	25.6±3.2	<0.001
Symptom onset time, days	183 (30, 1,125)	180 (30, 1,095)	240 (30, 1,460)	<0.001
Previous TIA/stroke event	1,082 (12.6)	465 (10.8)	617 (14.5)	<0.001
Hypertension	5,166 (60.3)	2,368 (55.1)	2,798 (65.5)	<0.001
Diabetes mellitus	3,193 (37.3)	1,658 (38.5)	1,535 (36.0)	0.01
Hyperlipidemia	3,946 (46.0)	2,025 (47.1)	1,921 (45.0)	0.053
Renal insufficiency	130 (1.5)	63 (1.5)	67 (1.6)	0.69
History of atrial fibrillation	85 (1.0)	37 (0.9)	48 (1.1)	0.22
Respiratory disease	148 (1.7)	62 (1.4)	86 (2.0)	0.042
Remote myocardial infarction	1,233 (14.4)	725 (16.9)	508 (11.9)	<0.001
History of CABG	17 (0.2)	5 (0.1)	12 (0.3)	0.09
History of stent implantation	1,126 (13.1)	604 (14.0)	522 (12.2)	0.01
Alcohol consumption	2,216 (25.9)	1,275 (29.6)	941 (22.0)	<0.001
Smoking	3,603 (42.0)	2,030 (47.2)	1,573 (36.8)	<0.001
LVEF, %	60 (55, 65)	60 (54, 65)	60 (56, 65)	<0.001
LVEDD, mm	48.0 (45.0, 52.0)	48.5 (45.0, 53.0)	48.0 (44.0, 51.0)	<0.001
LVESD, mm	32 (29, 35)	32 (29, 36)	31 (28, 34)	<0.001
Aortic regurgitation				<0.001
None	6,253 (73.0)	3,448 (80.2)	2,805 (65.7)
Mild	2,139 (25.0)	776 (18.0)	1,363 (31.9)
Moderate	147 (1.7)	58 (1.3)	89 (2.1)
Severe	31 (0.4)	19 (0.4)	12 (0.3)
Mitral regurgitation				<0.001
None	3,643 (42.5)	1,935 (45.0)	1,708 (40.0)
Mild	4,337 (50.6)	2,077 (48.3)	2,260 (52.9)
Moderate	467 (5.4)	224 (5.2)	243 (5.7)
Severe	123 (1.4)	65 (1.5)	58 (1.4)
Tricuspid regurgitation				<0.001
None	5,689 (66.4)	2,996 (69.7)	2,693 (63.1)
Mild	2,699 (31.5)	1,234 (28.7)	1,465 (34.3)
Moderate	145 (1.7)	56 (1.3)	89 (2.1)
Severe	37 (0.4)	15 (0.3)	22 (0.5)
Pericardial effusion	104 (1.2)	55 (1.3)	49 (1.1)	0.58
WBC, ×109/L	4.44±0.43	4.48±0.44	4.39±0.43	<0.001
RBC, ×1012/L	7.08±1.67	7.14±1.70	7.02±1.63	<0.001
PLT, ×109/L	222±49	224±49	220±48	0.002
Hb, g/L	136±14	138±14	135±13	<0.001
ALT, U/L	22 (14, 28)	24 (16, 31)	20 (13, 26)	<0.001
AST, U/L	19 (15, 22)	19 (15, 22)	19 (15, 22)	0.09
Cr, μmol/L	78 (68, 86)	78 (68, 85)	79 (67, 86)	0.89
CK-MB mass, ng/mL	1.70 (1.10, 2.29)	1.60 (1.10, 2.29)	1.80 (1.20, 2.30)	<0.001
LDH, U/L	186±49	184±45	187±52	0.04
hs-TnI, pg/mL	11 (5, 52)	11 (5, 53)	12 (5, 51)	0.002
Myoglobin, ng/mL	41±67	40±78	42±55	0.28
eGFR, mL/min/1.73 m2	87±16	91±16	82±15	<0.001
BNP, pg/mL	160±195	155±185	165±205	0.02
PT, s	11.40±1.19	11.41±1.38	11.39±0.97	0.58
APTT, s	31.4±4.7	31.7±5.6	31.1±3.6	<0.001
INR	1.02±0.10	1.02±0.12	1.01±0.08	0.57
D-dimer, ng/mL	119 (72, 169)	98 (62, 169)	140 (88, 182)	<0.001
Lactic acid, mmol/L	1.64±0.78	1.63±0.75	1.66±0.81	0.21

Categorical variables are presented as n (%). Continuous variables are presented as mean ± standard deviation or median (interquartile range). ALT, alanine aminotransferase; APTT, activated partial thromboplastin time; AST, aspartate aminotransferase; BMI, body mass index; BNP, B-type natriuretic peptide; CABG, coronary artery bypass grafting; CK-MB, creatine kinase-MB; Cr, serum creatinine; eGFR, estimated glomerular filtration rate; ePWV, estimated pulse wave velocity; Hb, hemoglobin; hs-TnI, high-sensitivity troponin I; INR, international normalized ratio; LDH, lactate dehydrogenase; LVEDD, left ventricular end-diastolic diameter; LVEF, left ventricular ejection fraction; LVESD, left ventricular end-systolic diameter; PLT, platelet count; PT, prothrombin time; RBC, red blood cell count; TIA, transient ischemic attack; WBC, white blood cell count.

Table 2. Perioperative and postoperative complications.

Characteristic	Overall	Low ePWV (<10.04 m/s)	High ePWV (≥10.04 m/s)	P
No. of patients	8,570	4,301	4,269
Operation time, s	4.00 (4.00, 5.00)	4.00 (4.00, 5.00)	4.00 (4.00, 5.00)	<0.001
Number of bypass grafts	2.03±0.48	2.04±0.48	2.02±0.48	0.01
LIMA utilization	5,119 (59.7)	2,966 (69.0)	2,153 (50.4)	<0.001
RIMA utilization	33 (0.4)	21 (0.5)	12 (0.3)	0.12
RA utilization	63 (0.7)	48 (1.1)	15 (0.4)	<0.001
SVG utilization	8,299 (96.8)	4,136 (96.2)	4,163 (97.5)	<0.001
PRBC transfusion, U	0.12±0.59	0.09±0.50	0.15±0.66	<0.001
Plasma transfusion, mL	16.1878±78.2028	14.9372±76.2775	17.4484±80.0859	0.14
Platelet transfusion, U	0.0023±0.1002	0.0021±0.0977	0.0026±0.1027	0.82
Intraoperative blood loss, mL	700 (600, 800)	700 (600, 800)	600 (600, 800)	<0.001
Length of hospital stay, days	13.9±5.8	14.0±5.9	13.9±5.6	0.63
Postoperative awakening time, h	6.4±10.6	6.0±11.4	6.7±9.7	0.002
ICU stay duration, h	37±49	36±48	37±49	0.25
Organ hypoperfusion	292 (3.4)	122 (2.8)	170 (4.0)	0.003
Cerebral hemorrhage	9 (0.1)	4 (0.1)	5 (0.1)	0.75
Cerebral infarction	198 (2.3)	76 (1.8)	122 (2.9)	<0.001
Limbs motor dysfunction	30 (0.4)	14 (0.3)	16 (0.4)	0.70
Reintubation	56 (0.7)	22 (0.5)	34 (0.8)	0.10
Tracheotomy	46 (0.5)	23 (0.5)	23 (0.5)	0.98
AKI	32 (0.4)	14 (0.3)	18 (0.4)	0.47
Postoperative CRRT	91 (1.1)	43 (1.0)	48 (1.1)	0.57
Postoperative myocardial infarction	32 (0.4)	13 (0.3)	19 (0.4)	0.28
Postoperative atrial fibrillation	680 (7.9)	259 (6.0)	421 (9.9)	<0.001
Postoperative IABP	480 (5.6)	278 (6.5)	202 (4.7)	<0.001
Postoperative ECMO	63 (0.7)	34 (0.8)	29 (0.7)	0.55
Pulmonary infection	167 (1.9)	77 (1.8)	90 (2.1)	0.29
Pleural effusion	1,938 (22.6)	981 (22.8)	957 (22.4)	0.67
Respiratory failure	22 (0.3)	12 (0.3)	10 (0.2)	0.68
Postoperative hemorrhage	39 (0.5)	18 (0.4)	21 (0.5)	0.61
Surgical debridement	11 (0.1)	4 (0.1)	7 (0.2)	0.36
Sepsis (blood culture-proven infection)	17 (0.2)	9 (0.2)	8 (0.2)	0.82
In-hospital mortality	32 (0.4)	15 (0.3)	17 (0.4)	0.71

Categorical variables are presented as n (%). Continuous variables are presented as mean ± standard deviation or median (interquartile range). AKI, acute kidney injury; CRRT, continuous renal replacement therapy; ECMO, extracorporeal membrane oxygenation; ePWV, estimated pulse wave velocity; IABP, intra-aortic balloon pump; ICU, intensive care unit; LIMA, left internal mammary artery; PRBC, packed red blood cell; RA, radial artery; RIMA, right internal mammary artery; SVG, saphenous vein graft.

Analysis of factors associated with POAF (logistic regression)

Univariable and multivariable logistic regression analyses were performed to identify factors associated with the occurrence of POAF. In the univariable analysis, a higher preoperative ePWV demonstrated the strongest association with POAF (odds ratio (OR: 1.17; 95% CI: 1.12–1.22; P<0.001), underscoring the contribution of increased arterial stiffness to postoperative arrhythmogenesis. Several other clinical and laboratory variables—including renal insufficiency, respiratory disease, prior stent implantation, indicators of impaired cardiac function, greater intraoperative blood loss, elevated hs-troponin I, reduced eGFR, higher INR, and increased D-dimer—were also associated with POAF (all P<0.05), although their effects were comparatively modest (Table 3). In the multivariable analysis, after adjusting for age, sex, comorbidities, operative variables, and key preoperative laboratory parameters, ePWV remained an independent predictor of POAF (OR: 1.15, 95% CI: 1.09–1.21, P<0.001). History of stent implantation (OR: 1.31, 95% CI: 1.05–1.64, P=0.02) and greater intraoperative blood loss (OR: 1.23, 95% CI: 1.05–1.45, P=0.01) also remained significant. Lower eGFR (OR: 0.99, 95% CI: 0.98–1.00, P=0.001) and higher INR (OR: 2.07, 95% CI: 1.22–3.50, P=0.007) were independently associated with POAF as well. Variables that were significant in the univariable analysis, including renal insufficiency, respiratory disease, left ventricular ejection fraction (LVEF), left ventricular end-diastolic diameter (LVEDD), and left ventricular end-systolic diameter (LVESD), were no longer significant in the multivariable model (Table 4). Overall, higher preoperative ePWV was identified as an independent risk factor for POAF after CABG, and its association persisted even after adjusting for age, comorbidities, and operative factors. These findings suggest that ePWV may serve as a useful marker for preoperative risk stratification and may help inform targeted preventive strategies.

Table 3. Univariate analysis of influencing factors (logistic regression).

Characteristic	N	Event, N	OR	95% CI	P
Age, years	2,073	167	1.02	0.85, 1.23	0.81
BMI, kg/m2	8,570	680	0.99	0.97, 1.02	0.59
Symptom onset time ≥183 days	4,324	352	1.06	0.90, 1.24	0.48
Previous TIA/stroke event	1,082	90	1.06	0.84, 1.34	0.62
Hypertension	5,166	422	1.08	0.92, 1.27	0.32
Diabetes mellitus	3,193	235	0.88	0.75, 1.04	0.13
Hyperlipidemia	3,946	296	0.9	0.76, 1.05	0.17
Renal insufficiency	130	18	1.89	1.14, 3.13	0.01
Respiratory disease	148	19	1.73	1.06, 2.82	0.03
Remote myocardial infarction	1,233	115	1.23	1.00, 1.52	0.051
History of CABG	17	1	0.72	0.10, 5.47	0.76
History of stent implantation	1,126	113	1.35	1.09, 1.67	0.005
Alcohol consumption	2,216	165	0.91	0.76, 1.09	0.32
Smoking	3,603	274	0.92	0.79, 1.08	0.34
LVEF ≥60%	5,036	347	0.71	0.61, 0.83	<0.001
LVEDD ≥48 mm	4,768	429	1.4	1.19, 1.64	<0.001
LVESD ≥32 mm	4,355	407	1.49	1.27, 1.75	<0.001
Pericardial effusion	104	10	1.24	0.64, 2.39	0.52
Operation time ≥4 hours	7,236	583	1.12	0.89, 1.40	0.33
Number of bypass grafts	8,570	680	0.84	0.71, 0.99	0.04
PRBCs transfusion, U	8,561	679	1.19	1.07, 1.32	<0.001
Intraoperative blood loss ≥700 mL	4,387	381	1.24	1.06, 1.45	0.008
RBC, ×1012/L	8,570	680	0.9	0.75, 1.07	0.24
WBC, ×109/L	8,570	680	1.04	0.99, 1.09	0.08
PLT, ×109/L	8,570	680	0.9985	0.9976, 0.9995	0.007
Hb, g/L	8,570	680	1	0.99, 1.00	0.39
Cr ≥78.4 μmol/L	4,293	380	1.29	1.10, 1.51	0.002
CK-MB ≥1.7 ng/mL	4,458	364	1.07	0.91, 1.25	0.41
LDH, U/L	8,570	680	1	1.00, 1.00	0.83
Hs-TnI ≥11.1 pg/mL	4,299	374	1.23	1.05, 1.45	0.009
Myoglobin, ng/mL	8,570	680	1	1.00, 1.00	0.09
eGFR, mL/min/1.73 m2	8,570	680	0.98	0.98, 0.99	<0.001
BNP, pg/mL	8,570	680	1.0012	1.0006, 1.0018	<0.001
APTT, s	8,570	680	0.99	0.97, 1.01	0.26
INR	8,570	680	2.29	1.37, 3.82	0.002
D-dimer ≥119 ng/mL	4,290	382	1.31	1.12, 1.53	<0.001
Lactic acid, mmol/L	8,570	680	1.01	0.92, 1.12	0.84
ePWV, m/s	8,570	680	1.17	1.12, 1.22	<0.001

APTT, activated partial thromboplastin time; BMI, body mass index; BNP, B-type natriuretic peptide; CABG, coronary artery bypass grafting; CI, confidence interval; CK-MB, creatine kinase-MB; Cr, serum creatinine; eGFR, estimated glomerular filtration rate; ePWV, estimated pulse wave velocity; Hb, hemoglobin; hs-TnI, high-sensitivity troponin I; INR, international normalized ratio; LDH, lactate dehydrogenase; LVEDD, left ventricular end-diastolic diameter; LVEF, left ventricular ejection fraction; LVESD, left ventricular end-systolic diameter; OR, odds ratio; PLT, platelet count; PRBC, packed red blood cell; RBC, red blood cell count; TIA, transient ischemic attack; WBC, white blood cell count.

Table 4. Multivariate analysis of influencing factors (logistic regression).

Characteristic	N	Event, N	OR	95% CI	P
Renal insufficiency	130	18	1.07	0.61, 1.85	0.82
Respiratory disease	148	19	1.52	0.92, 2.50	0.10
Remote myocardial infarction	1,232	115	1.06	0.85, 1.33	0.58
History of stent implantation	1,125	113	1.31	1.05, 1.64	0.02
Ejection fraction ≥60%	5,031	346	0.9	0.75, 1.09	0.29
LVEDD ≥48 mm	4,763	428	1.17	0.94, 1.46	0.17
LVESD ≥32 mm	4,349	406	1.24	0.99, 1.57	0.07
Number of bypass grafts	8,561	679	0.88	0.74, 1.05	0.15
PRBC transfusion, U	8,561	679	1.1	0.99, 1.23	0.07
Intraoperative blood loss ≥700 mL	4,386	381	1.23	1.05, 1.45	0.01
WBC, ×109/L	8,561	679	1.04	0.99, 1.09	0.09
PLT, ×109/L	8,561	679	1	1.00, 1.00	0.07
Cr ≥78.4 μmol/L	4,290	380	0.96	0.79, 1.17	0.70
hs-TnI ≥11.1 pg/mL	4,296	374	1	0.84, 1.19	0.99
Myoglobin, ng/mL	8,561	679	1	1.00, 1.00	0.79
eGFR, mL/min/1.73 m2	8,561	679	0.99	0.98, 1.00	0.001
BNP, pg/mL	8,561	679	1.0012	1.0006, 1.0018	0.049
INR	8,561	679	2.07	1.22, 3.50	0.007
D-dimer ≥119 ng/mL	4,285	381	1.05	0.89, 1.24	0.56
ePWV, m/s	8,561	679	1.15	1.09, 1.21	<0.001

BNP, B-type natriuretic peptide; CABG, coronary artery bypass grafting; CI, confidence interval; Cr, serum creatinine; eGFR, estimated glomerular filtration rate; ePWV, estimated pulse wave velocity; Hb, hemoglobin; hs-TnI, high-sensitivity troponin I; INR, international normalized ratio; LVEDD, left ventricular end-diastolic diameter; LVESD, left ventricular end-systolic diameter; OR, odds ratio; PLT, platelet count; PRBC, packed red blood cell; WBC, white blood cell count.

Nonlinear association and threshold effect between preoperative ePWV and POAF

RCS analyses were performed to characterize the shape of the association between preoperative ePWV and the risk of POAF after CABG. In the unadjusted model, ePWV showed a significant inverse J-shaped nonlinear relationship with POAF (P for overall association <0.001; P for nonlinearity <0.001) (Figure 1A). After adjusting for age, sex, comorbidities, cardiac function, operative characteristics, and laboratory findings, the nonlinear association persisted (P for overall association <0.001; P for nonlinearity =0.008) (Figure 1B), and the curve showed a monotonic increase. To further quantify this nonlinear pattern, a segmented linear regression analysis was performed based on the breakpoint identified from the RCS curve (11.4 m/s) (Figure 1B, Figure 2). Threshold effect testing showed a significant difference across this breakpoint (likelihood ratio test, P<0.001). When ePWV was <11.4 m/s, each 1 m/s increase in ePWV was associated with a 27% higher risk of POAF (OR: 1.27, 95% CI: 1.18–1.36, P<0.001). When ePWV was ≥11.4 m/s, no statistically significant association was observed (OR: 0.87, 95% CI: 0.75–1.02, P=0.09) (Table 5).

Figure 1.

Nonlinear association between preoperative ePWV and risk of postoperative atrial fibrillation. The solid line represents the OR, and the shaded area represents the 95% CI. (A) Unadjusted model. (B) Model adjusted for age, gender, comorbidities, and operative variables. P for nonlinear <0.001 for both models. A break-point was identified at 11.4 m/s in the adjusted model. CI, confidence interval; ePWV, estimated pulse wave velocity; OR, odds ratio.

Figure 2.

Threshold effect of preoperative ePWV on postoperative atrial fibrillation risk: Piecewise regression model. The solid line shows the predicted probability of postoperative atrial fibrillation derived from the piecewise linear regression model. A significant break-point was identified at an ePWV of 11.4 m/s. Below this threshold, the risk of atrial fibrillation increased sharply with rising ePWV, whereas above this threshold, the risk plateaued. ePWV, estimated pulse wave velocity.

Table 5. Threshold effect analysis of ePWV on postoperative atrial fibrillation.

Model	OR (95% CI)†	P
Fitting by standard logistic regression model	1.15 (1.09, 1.21)	<0.001
Fitting by piecewise logistic regression model (break-point =11.4)
ePWV <11.4 m/s	1.27 (1.18, 1.36)	<0.001
ePWV ≥11.4 m/s	0.87 (0.75, 1.02)	0.09
Log likelihood ratio		<0.001

^†, adjusted for: renal insufficiency, respiratory disease, remote myocardial infarction, history of stent implantation, ejection fraction ≥60%, LVEDD ≥48 mm, LVESD ≥32 mm, number of bypass grafts, PRBCs transfusion (U), intraoperative blood loss ≥700 mL, WBC (×10⁹/L), PLT (×10⁹/L), Cr ≥78.4 μmol/L, hs-TnI ≥11.1 pg/mL, myoglobin (ng/mL), eGFR (mL/min/1.73 m²), BNP (pg/mL), INR, D-dimer ≥119 ng/mL. BNP, B-type natriuretic peptide; CI, confidence interval; Cr, serum creatinine; eGFR, estimated glomerular filtration rate; ePWV, estimated pulse wave velocity; hs-TnI, high-sensitivity troponin I; INR, international normalized ratio; LVEDD, left ventricular end-diastolic diameter; LVESD, left ventricular end-systolic diameter; OR, odds ratio; PLT, platelet count; PRBCs, packed red blood cell; WBC, white blood cell count.

Predictive performance and clinical threshold of ePWV for POAF

To evaluate the predictive value of preoperative ePWV for POAF and to identify the most suitable threshold for clinical application, we performed a combined analysis using ROC curves and threshold effect assessment. The ROC analysis showed that ePWV had an AUC of 0.583 for predicting POAF (Figure 3). The optimal cutoff value, determined by maximizing the Youden index, was 9.62 m/s, yielding a sensitivity of 71.9%, a specificity of 42.2%, and a negative predictive value (NPV) of 94.6% (Table 6). Notably, the optimal cutoff of 9.62 m/s and the 11.4 m/s breakpoint identified in the threshold analysis jointly defined the clinically meaningful risk range of ePWV. When ePWV was below 9.62 m/s, patients were classified as a low-risk group for POAF (NPV >94%). For ePWV values between 9.62 and 11.4 m/s, the risk of POAF increased substantially with higher ePWV (OR 1.27, 95% CI: 1.18–1.36). When ePWV exceeded 11.4 m/s, the risk reached a high plateau.

Figure 3.

Receiver operating characteristic curve of preoperative ePWV for predicting postoperative atrial fibrillation. The AUC was 0.583. The dashed diagonal line represents the reference line of no discriminatory ability. AUC, area under the curve; ePWV, estimated pulse wave velocity.

Table 6. Predictive performance of ePWV for postoperative atrial fibrillation.

Optimal cutpoint (m/s)	AUC	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)
9.62	0.583	71.9	42.2	9.7	94.6
9.91	0.583	65.4	48.7	9.9	94.2

AUC, area under the curve; ePWV, estimated pulse wave velocity; NPV, negative predictive value; PPV, positive predictive value.

Discussion

This large single-center retrospective study systematically evaluated the association between preoperative ePWV and the risk of POAF after CABG. Our analysis yielded several important observations. Higher preoperative ePWV remained independently associated with POAF after comprehensive adjustment for clinical, operative, and laboratory characteristics. In addition, ePWV demonstrated a significant nonlinear relationship with POAF, with a clear threshold effect and a steeper risk gradient at values below 11.4 m/s. The ROC-derived cutoff of 9.62 m/s further provided a high NPV, and together with the identified threshold, delineates a clinically useful framework for ePWV-based risk stratification.

Evidence on the association between ePWV and POAF remains limited. A large population-based cohort study reported that higher ePWV was associated with an increased risk of incident atrial fibrillation, supporting its utility as a surrogate measure of arterial stiffness in epidemiological research (13). Building on evidence from population-based cohorts, our study shows that among patients undergoing CABG, ePWV demonstrates a clear nonlinear association with POAF.

At moderate levels of arterial stiffness (ePWV <11.4 m/s), elevations in ePWV may promote atrial fibrillation initiation through several hemodynamic and electrophysiological mechanisms. We speculate that greater arterial stiffness increases systolic load and left atrial filling pressure, thereby promoting atrial wall stretch and creating a substrate susceptible to arrhythmogenesis (14,15). Concurrent atrial fibrosis reduces atrial compliance and disrupts conduction integrity, leading to conduction slowing and a substrate that favors reentrant activity (16-18). In parallel, ion-channel remodeling, including the shortening of the atrial effective refractory period, further augments atrial vulnerability (19). However, when arterial stiffness becomes markedly elevated (ePWV ≥11.4 m/s), the risk of POAF no longer increases proportionally (20). A plausible explanation is that progressive structural and electrical remodeling creates a substrate favoring more sustained forms of atrial fibrillation, diminishing the influence of acute perioperative triggers. This pattern suggests that atrial remodeling may have progressed to a stage where further pathological worsening is limited, such that additional increases in ePWV exert little or no incremental effect on POAF risk.

Although several predictive models for POAF after CABG have been proposed, ePWV offers a new perspective for preoperative risk assessment (21). As a noninvasive and convenient measure that has been validated in cardiovascular research, it may complement existing markers. In this study, we identified the optimal cutoff value of ePWV for predicting POAF after CABG and demonstrated its nonlinear association and corresponding inflection point using RCS analysis. These findings support a more refined approach to risk stratification. Patients with ePWV <9.62 m/s may be considered at low risk and may not require intensive preventive measures. Those with ePWV between 9.62 and 11.4 m/s warrant closer monitoring and early prophylaxis. Patients with ePWV ≥11.4 m/s may require individualized management strategies given their higher risk.

Taken together, these findings indicate that preoperative ePWV, as a practical and noninvasive measure of vascular function, may have meaningful value for POAF risk stratification in patients undergoing CABG. Future studies should aim to incorporate ePWV into multivariable prediction frameworks alongside established risk factors and further clarify the molecular pathways linking arterial stiffness to POAF, thereby supporting its broader use in perioperative management.

The observed incidence of POAF in our cohort was 7.9%, which is lower than the commonly reported rate of approximately 30% in the literature. Several factors may explain this discrepancy. First, we applied a stringent definition of POAF (new-onset episode lasting ≥30 seconds), which may have overlooked brief or asymptomatic episodes. Second, our study exclusively enrolled patients undergoing isolated CABG, excluding those undergoing concomitant valvular or other major cardiac surgeries, which are associated with a higher risk of POAF. Finally, as a high-volume referral center, our institution’s standardized perioperative management protocols may have effectively mitigated POAF risk. These considerations should be kept in mind when interpreting the generalizability of our absolute risk estimates.

There are several limitations in this study. First, as a retrospective observational analysis, it demonstrates an association between ePWV and POAF but cannot fully elucidate underlying mechanisms, which likely involve multiple hemodynamic and structural pathways. Second, the predictive ability of ePWV alone is modest, and future studies should integrate it with other variables to improve multivariable risk models. Finally, the single-center design may limit generalizability, and the identified ePWV threshold requires validation in larger, multicenter prospective cohorts.

Conclusions

Preoperative ePWV was independently associated with the risk of POAF after CABG and demonstrated a clear nonlinear relationship characterized by an optimal cutoff of 9.62 m/s and an inflection point of 11.4 m/s. These thresholds provide a practical framework for preoperative risk stratification. As a simple and noninvasive measure of vascular function, ePWV may aid in identifying patients at increased risk of POAF. Validation in multicenter prospective cohorts and incorporation into comprehensive prediction models will be essential to determine its broader clinical utility.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study protocol was reviewed and approved by the Ethics Committee of Beijing Anzhen Hospital (approval No. 2025283x). Given the retrospective study design, the requirement for written informed consent was waived. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Data Sharing Statement

Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-aw-2385/dss

jtd-18-02-147-dss.pdf ^{(73.3KB, pdf)}

DOI: 10.21037/jtd-2025-aw-2385