Preoperative echocardiography-based nomogram for postoperative cardiac complications in pancreatic cancer with vascular invasion

Abstract

BACKGROUND

Pancreatic cancer with vascular invasion requires complex surgery, which is associated with substantial perioperative cardiac risk. Existing risk stratification tools lack specificity for this population and do not incorporate quantitative echocardiographic parameters.

AIM

To develop a nomogram integrating preoperative cardiac biomarkers and echocardiography parameters for predicting major postoperative cardiac complications (MPCCs) in patients undergoing surgery for pancreatic cancer with vascular invasion.

METHODS

This retrospective cohort study included 138 consecutive patients who underwent surgery for pancreatic cancer with vascular invasion. Preoperative clinical data and transthoracic echocardiography parameters were collected. MPCCs (myocardial infarction, arrhythmia, acute heart failure, or cardiac death within 30 days) constituted the primary endpoint. Independent predictors were identified through multivariable logistic regression and incorporated into a nomogram. Model performance was assessed by discrimination (C-index), calibration (Hosmer-Lemeshow test), and clinical utility (decision curve analysis).

RESULTS

MPCCs occurred in 45 patients (32.6%). Multivariable analysis identified preoperative B-type natriuretic peptide [odds ratio (OR) = 1.037, 95% confidence interval (CI): 1.013-1.063], ratio of early mitral inflow velocity (E) to average early diastolic mitral annular velocity (e’) (average E/e′ ratio) (OR = 1.296, 95%CI: 1.044-1.610), and left atrial volume index (OR = 1.046, 95%CI: 1.017-1.076) as independent predictors. The nomogram demonstrated excellent discrimination (C-index: 0.874, 95%CI: 0.812-0.936) and good calibration (P = 0.392). Decision curve analysis confirmed positive net benefit across clinically relevant threshold probabilities.

CONCLUSION

A nomogram based on preoperative B-type natriuretic peptide, average E/e′ ratio, and left atrial volume index accurately predicts postoperative cardiac complications in pancreatic cancer patients with vascular invasion. This tool facilitates preoperative risk stratification and guides perioperative management decisions.

Key Words: Postoperative complications; Echocardiography; Left atrial volume index; B-type natriuretic peptide; Average E/e’ ratio

Core Tip: Pancreatic cancer with vascular invasion carries substantial perioperative cardiac risk. This study developed a novel nomogram integrating preoperative B-type natriuretic peptide, average E/e’ ratio (early mitral inflow velocity to average mitral annular early diastolic velocity ratio), and left atrial volume index to predict postoperative cardiac complications. The nomogram demonstrated excellent discrimination (C-index: 0.874) and calibration. This nomogram enables early identification of high-risk patients and facilitates tailored perioperative management strategies, advancing precision risk stratification in this vulnerable population.

INTRODUCTION

Pancreatic cancer is an aggressive malignancy with rising incidence and poor prognosis. Surgical complexity increases significantly when tumors invade major vessels (e.g., portal vein, mesenteric vessels, celiac axis), often requiring vascular resection and reconstruction[1,2]. These procedures prolong operative time, exacerbate hemodynamic instability, and elevate postoperative complication risks. Complications extend hospital stays, increase costs, impair quality of life, and may delay adjuvant therapy. Among these, cardiac complications, such as myocardial infarction, acute heart failure, arrhythmias, and cardiac death, are particularly dangerous due to their insidious onset and poor prognosis, acting as a “silent killer” during the perioperative period[3,4].

Cardiovascular risk is notably high in these patients for three reasons[5-7]: (1) Tumor invasion activates coagulation and causes endothelial injury, disrupting myocardial oxygen supply; (2) Patients are often elderly with comorbidities and may have chemotherapy-related cardiotoxicity; and (3) Prolonged surgery amplifies inflammatory stress, while fluid shifts strain cardiac function.

Integrating preoperative echocardiography with clinical factors is essential for identifying high-risk patients and improving outcomes. Transthoracic echocardiography (TTE) offers detailed structural and functional insights beyond left ventricular ejection fraction (LVEF)[8,9]. Diastolic parameters (E/A ratio, E/e’, left atrial volume index (LAVI), right heart indices [tricuspid annular plane systolic excursion (TAPSE), pulmonary artery systolic pressure], and morphological measures [left ventricular mass index (LVMI)] have shown predictive value for cardiac events after non-cardiac surgery[10,11], yet their role in pancreatic cancer remains underexplored.

Current risk models lack specificity for pancreatic cancer patients, ignoring factors such as cachexia, hypercoagulability, chemotherapy-related cardiotoxicity, and prolonged visceral surgery. This study aims to develop a tailored predictive model by analyzing preoperative TTE parameters and clinical data in patients with vascular-invasive pancreatic cancer. The goal is to create a nomogram for individualized risk assessment, enabling early intervention and personalized perioperative management.

MATERIALS AND METHODS

Patient selection

A retrospective analysis was conducted using data from 138 patients with pancreatic cancer involving vascular invasion who underwent surgical treatment in the Department of Hepatobiliary Surgery at our hospital between January 2020 and June 2025. Based on the following inclusion and exclusion criteria, a total of 138 eligible patients were selected for analysis.

Inclusion criteria: (1) Aged between 29 years and 81 years; (2) Postoperative pathological confirmation of pancreatic ductal adenocarcinoma; (3) Preoperative comprehensive assessment confirming no distant metastasis; (4) Complete perioperative clinical data; (5) No contraindications to surgery identified during preoperative evaluation; (6) Informed consent obtained from the patient and family members regarding the surgical procedure; (7) No history of clinically manifest cardiovascular disease, defined as: Prior myocardial infarction, prior coronary revascularization (percutaneous coronary intervention or coronary artery bypass grafting), documented history of heart failure (heart failure with reduced ejection fraction or heart failure with preserved ejection fraction), history of sustained arrhythmias requiring treatment, or moderate-to-severe valvular heart disease. Patients with controlled hypertension, diabetes-related cardiovascular risk factors, or subclinical echocardiographic abnormalities without prior cardiac events were eligible for inclusion; and (8) Intraoperative confirmation of vascular invasion necessitating resection and reconstruction.

Exclusion criteria: (1) Intraoperative discovery of distant metastasis; (2) Final pathological examination confirming a diagnosis other than pancreatic cancer; (3) Inability to complete the surgery successfully; and (4) Intraoperative confirmation of no vascular invasion, with vessels successfully dissected free without requiring resection and reconstruction.

The study protocol adhered to the Declaration of Helsinki and received formal approval from the Ethics Committee of Beijing Chaoyang Hospital (Approval No. 2024-D-512) in April 2024. Data collection was performed from June 2024 to June 2025, encompassing patients who underwent surgery between January 2020 and June 2025.

Data collection

Clinical and echocardiographic data were systematically collected for all patients. Clinical variables included age, sex, body mass index, American Society of Anesthesiologists classification, comorbidities (hypertension, diabetes, coronary artery disease, chronic kidney disease, chronic obstructive pulmonary disease), smoking history, medication use (beta-blockers, angiotensin-converting enzyme inhibitors/angiotensin receptor blockers, statins), and preoperative levels of cardiac tropin (cTn), B-type natriuretic peptide (BNP), N-terminal-proBNP, serum creatinine, and electrocardiogram findings.

Preoperative TTE was performed using the Philips EPIQ system. Two experienced sonographers (8 years and 12 years of experience) independently reviewed the images, blinded to clinical outcomes. Inter-observer agreement was assessed using the intraclass correlation coefficient (ICC) with absolute agreement and the two-way random-effects model. ICC values were interpreted as follows: < 0.50, poor; 0.50-0.75, moderate; 0.75-0.90, good; and > 0.90, excellent agreement. Measurements were performed in a random sequence, with a minimum 2-week washout period between assessments to minimize recall bias. Assessed parameters included: Systolic function: LVEF (%). Diastolic function: Mitral inflow E wave, A wave, E/A ratio; septal and lateral e’ velocities; average E/e’ratio; and LAVI (mL/m²). LAV was calculated as (π/6) × anteroposterior × mediolateral × superoinferior diameter; body surface area was calculated as 0.007184 × [height (cm)]^0.725 × [weight (kg)]^0.425. Right heart function: Right heart function: TAPSE (mm), tricuspid lateral annular peak systolic velocity (S’, cm/second) and pulmonary artery systolic pressure (mmHg, if measurable). Cardiac morphology: LVMI (g/m²), calculated using the standard formula.

Observation period and endpoint definitions

Patients were followed for 30 days post-surgery (± 3 days). The primary endpoint was the occurrence of major postoperative cardiac complications (MPCCs), defined as: Myocardial ischemia/infarction: CTn elevation above the 99^th percentile URL with a rising/falling pattern, plus ischemic symptoms, new electrocardiogram changes, imaging evidence of wall motion abnormality, or angiographic culprit lesion. Arrhythmia: New-onset atrial fibrillation/flutter (ventricular rate > 120 bpm), sustained ventricular tachycardia (> 30 seconds), ventricular fibrillation, or high-grade atrioventricular block. Acute heart failure: Symptoms/signs (e.g., dyspnea, orthopnea, pulmonary edema) with elevated BNP (> 400 pg/mL) or N-terminal-proBNP (> 1800 pg/mL), requiring intravenous diuretics, vasodilators, or inotropes. Cardiac death: Death from cardiac causes within 30 days, excluding non-cardiac origins. High-risk patients (age > 65 years, diabetes, LVEF < 50%) underwent routine cTn screening at 48/72 hours; a > 20% dynamic change was considered significant. Suspected cases required confirmation by echocardiography or chest X-ray. Endpoint adjudication was performed independently by two cardiologists blinded to predictor variables, with a third resolving disagreements. Primary cardiac events were strictly distinguished from secondary cardiac dysfunction. Source documents were archived, and lost-to-follow-up data were censored.

Statistical analysis

All statistical analyses were conducted using appropriate methods for data type and distribution. Continuous variables were expressed as mean ± SD for normally distributed data or median (interquartile range) for non-normally distributed data, while categorical variables were summarized as n (%). Group comparisons employed independent samples t-test, Mann-Whitney U test, or χ²/Fisher’s exact test as appropriate. We first performed univariable logistic regression with MPCCs as the dependent variable to identify potential predictors among clinical and echocardiographic parameters (P < 0.05 for inclusion in multivariable analysis). Backward stepwise selection (retention criterion: P < 0.05) was used to identify independent predictors while assessing multicollinearity (variance inflation factor < 5). The final model was translated into a nomogram for individualized risk calculation. The nomogram was constructed using regression coefficients (β) from the final multivariable logistic regression model. For each predictor, points were assigned by linear transformation proportional to coefficient weight. Model performance was comprehensively evaluated through discrimination (C-index), calibration (calibration curves; Hosmer-Lemeshow test), internal validation (bootstrap resampling with 200 repetitions), and clinical utility (decision curve analysis). Additionally, a clinical impact curve was generated per 1000 patients to illustrate the trade-offs between true-positive and false-positive classifications across different risk thresholds, depicting the number of high-risk patients identified, true positives, and false positives at varying probability cutoffs. For risk stratification, patients were categorized into four groups based on clinically meaningful predicted probability thresholds determined by the investigators: Low-risk (< 5%), intermediate-risk (5%-30%), high-risk (30%-70%), and very high-risk (> 70%). These cutoffs were selected to correspond to distinct clinical action levels for perioperative management. Analyses used SPSS (version 24.0) for basic computations and R (version 2025.05.0) with rms, pROC, and rmda packages for modeling. All tests were two-sided with significance at P < 0.05.

RESULTS

Baseline characteristics

A total of 138 patients (79 males, 59 females; ratio: 1.34:1) with a mean age of 61.5 ± 10.2 years (range: 32-78) were included. Presenting symptoms included abdominal pain (37.7%), jaundice (27.5%), gastrointestinal symptoms (10.1%), and incidental findings (24.6%). Preoperatively, 15.2% and 12.3% underwent percutaneous transhepatic biliary drainage and endoscopic retrograde cholangiopancreatography for biliary decompression, respectively. Comorbidities included diabetes (30.4%), and 5.1% had a history of upper abdominal surgery. All patients underwent successful resection. Pathological evaluation confirmed R0 resection in 89.9% and R1 in 10.1%. Tumor locations were the pancreatic head (34.1%), neck (28.3%), uncinate process (8.0%), and body/tail (29.7%). Differentiation varied from well to poorly differentiated, with 12.3% being adenosquamous carcinoma. Perineural invasion and lymph node positivity were observed in 22.5% and 63.8%, respectively. Surgical procedures included pancreaticoduodenectomy (38.4%), total pancreatectomy (46.4%), and distal pancreatectomy (15.2%). Venous and arterial resections were performed in 40.6% and 15.2%, respectively. Median blood loss was 800 mL (interquartile range: 400-1000 mL), with 56.5% receiving transfusion; mean operative time was 12.4 ± 2.1 hours.

Mean postoperative stay was 18.2 ± 9.3 days; all patients completed 30-day follow-up. Non-cardiac complications occurred in 25.4% of patients, with two perioperative deaths (1.4%) due to hemorrhage-related liver failure. Other complications included biochemical pancreatic leak (8.0%), delayed gastric emptying (10.1%), intra-abdominal infection (4.3%), pulmonary infection (1.4%), pleural effusion (1.4%), and diarrhea (12.3%). Within 30 days, 45 patients (32.6%) experienced MPCCs. Arrhythmias were the most common (30.4%), followed by acute heart failure (1.4%) and cardiac death (0.7%). Details are presented in Table 1.

Table 1 Univariate and multivariate analysis of cardiac complications within 30 days postoperatively.

Factors	Univariate analysis		Multivariate analysis
	χ2 value	P value	OR (95%CI)	Wald	P value
Age	0.530	0.467
Gender	0.755	0.385
BMI (kg/m2)	0.837	0.360
ASA grade (III/IV)	0.311	0.577
Smoking history	0.733	0.392
Diabetes history	1.090	0.296
Coronary heart disease history	0.444	0.505
Hypertension history	0.204	0.651
BNP (pg/mL)	19.409	0.000	1.038 (1.013-1.063)	9.397	0.002
cTn (ng/mL)	1.424	0.233
EDV (mL)	8.051	0.005	1.054 (0.969-1.146)	1.512	0.219
ESV (mL)	5.773	0.016	0.882 (0.709-1.096)	1.286	0.257
LVEF (%)	16.809	0.000	0.788 (0.609-1.019)	3.296	0.069
E/A ratio	0.520	0.471
Average E/e’ ratio	12.274	0.000	1.296 (1.044-1.610)	5.536	0.019
LAVI (mL/m2)	24.145	0.000	1.046 (1.017-1.076)	10.057	0.002
LVMI (g/m2)	1.016	0.439
TAPSE (mm)	1.460	0.227
S’ (cm/second)	0.117	0.732
Postoperative abdominal complications	0.270	0.603

All univariate-screened variables entered multivariate analysis; intermediate P values during stepwise elimination; only significant variables (P < 0.05) are displayed in the final model. Echocardiographic parameters were measured by two independent observers. Inter-observer agreement was excellent for all key parameters (intraclass correlation coefficient > 0.88). OR: Odds ratio; CI: Confidence interval; BMI: Body mass index; ASA: American Society of Anesthesiology; BNP: B-type natriuretic peptide; cTn: Cardiac tropin; EDV: End-diastolic volume; ESV: End-systolic volume; LVEF: Left ventricular ejection fraction; LAVI: Left atrial volume index; LVMI: Left ventricular mass index; TAPSE: Tricuspid annular plane systolic excursion.

Inter-observer agreement regarding echocardiographic measurements

Inter-observer agreement for key echocardiographic parameters was excellent. The ICCs were 0.94 [95% confidence interval (CI): 0.91-0.96] for LVEF, 0.91 (95%CI: 0.87-0.94) for average E/e’ ratio, 0.93 (95%CI: 0.90-0.95) for LAVI, 0.89 (95%CI: 0.84-0.93) for E/A ratio, 0.88 (95%CI: 0.83-0.92) for TAPSE, and 0.90 (95%CI: 0.86-0.93) for LVMI. These results confirm the reliability of echocardiographic measurements in this study.

Analysis of factors influencing postoperative cardiac complications

The results of the univariable analysis are also presented in Table 1. BNP, end-diastolic volume, end-systolic volume, LVEF, ratio of early mitral inflow velocity (E) to average early diastolic mitral annular velocity (e’) (average E/e’ ratio), and LAVI were identified as potential risk factors influencing postoperative cardiac complications. These indicators were subsequently included in a multivariable Cox proportional hazards regression model for further analysis. The multivariable analysis revealed that BNP (Wald = 9.397, 95%CI: 1.013-1.063), average E/e’ ratio (Wald = 5.536, 95%CI: 1.044-1.610), and LAVI (Wald = 10.057, 95%CI: 1.017-1.076) were independent risk factors for postoperative cardiac complications. Patients with normal BNP levels, an average E/e’ ratio < 14, and LAVI < 40 had a significantly lower likelihood of experiencing postoperative cardiac complications. As illustrated in the figures below, comparative echocardiographic images demonstrate the normal vs abnormal patterns for both the LAVI (Figure 1A and B) and the average E/e’ ratio (Figure 1C-H).

Figure 1 Echocardiographic assessment of left atrial size and left ventricular diastolic function. A and B: Apical four-chamber views showing left atrial size. Normal left atrial volume index (≤ 34 mL/m²) (A). Increased left atrial volume index (> 34 mL/m²) indicating left atrial enlargement (B); C-H: Parameters for diastolic function assessment. Mitral inflow pulse-wave Doppler spectra (E and A waves) in two patients (C, F). Septal (D, G) and lateral (E, H) early diastolic mitral annular velocities (e’) obtained via tissue Doppler imaging from an apical four-chamber view, displaying normal (septal e’ ≥ 7 cm/second, lateral e’ ≥ 10 cm/second; D, E) and reduced (septal e’ < 7 cm/second, lateral e’ < 10 cm/second; G, H) e’ velocities. Average E/e’ ratio ≤ 14 indicates normal filling pressure; > 14 suggests elevated filling pressure.

Establishment of a predictive model for postoperative cardiac complications

Multivariable logistic regression identified three independent predictors of postoperative cardiac complications: Preoperative BNP, average E/e’ ratio, and LAVI. These were integrated into a nomogram to predict the 30-day risk (Figure 2). For each predictor, points were assigned by linear transformation proportional to coefficient weight: Average E/e’ ratio (β = 0.259, highest weight) had the steepest point-per-unit gradient; LAVI (β = 0.045) and BNP (β = 0.036) had shallower gradients. Absolute point ranges for observed clinical values were approximately: BNP (0-100 points), average E/e’ ratio (0-56 points), and LAVI (0-98 points), summing to a total score (0-160 points). The nomogram assigns points for each variable: To use this nomogram: (1) Locate the patient’s value on each variable axis and draw a line upward to the points axis to obtain the individual score; (2) Sum the three scores; and (3) Plot the total score on the “Total Points” line, then project downward to the “Cardiac complications” axis to read the predicted probability. Total points convert to probability with: P = 1/{1 + exp[-(-4.2 + 0.05 × Total Points)]}. A total score below 60 corresponds to a < 5% risk, while a score above 148 corresponds to a > 95% risk. The steep slope of the probability curve between 92 points and 120 points (30%-70% risk) indicates a critical threshold for clinical decision-making. For example, a patient with BNP = 140 pg/mL, E/e’ = 13, and LAVI = 60 mL/m² would score approximately 60 points, 40 points, and 42 points, respectively, totaling 142 points. Using the logistic function P = 1/{1 + exp[-(-4.2 + 0.05 × 142)]} = 1/[1 + exp(-2.9)] ≈ 0.945, this corresponds to a predicted risk of 94.5%.

Figure 2 Nomogram for cardiac complications within 30 days postoperatively. BNP: B-type natriuretic peptide; LAVI: Left atrial volume index.

Model validation and clinical utility assessment

The nomogram demonstrated excellent calibration, with the bias-corrected curve closely aligning with the ideal line across the entire risk spectrum (Figure 3A). The mean absolute error was 0.027, and the Hosmer-Lemeshow test (χ² = 8.44, P = 0.392) confirmed good fit, indicating reliable agreement between predicted and observed outcomes. Discrimination was strong, with a C-index of 0.874 (95%CI: 0.812-0.936), equivalent to the area under the receiver operating characteristic curve (Figure 3B). Bootstrap validation with 200 replicates yielded a consistent mean C-index of 0.874 (Figure 3C), demonstrating minimal optimism and robust performance (Somers’ Dxy = 0.748). The clear separation in predicted probability distributions between patients with and without cardiac complications (Figure 3D) further confirmed the model’s discriminatory power.

Figure 3 Development and internal validation of the nomogram for predicting major postoperative cardiac complications. A: Calibration curve; B: Receiver operating characteristic curve; C: Distribution of the C-index from bootstrap internal validation; D: Distribution of predicted probabilities. AUC: Area under the receiver operating characteristic curve; MPCC: Major postoperative cardiac complication.

The nomogram’s clinical utility was evaluated through risk stratification and decision curve analysis. Risk stratification based on nomogram scores effectively categorized patients into four distinct risk groups with sharply graduated event rates: Low-risk (< 5% predicted) 3.1%, intermediate-risk (5%-30%) 26.7%, high-risk (30%-70%) 90.0%, and very high-risk (> 70%) 100% (Figure 4A). This stepwise increase confirms the model’s accuracy in discriminating risk levels and its usefulness in guiding perioperative management. Decision curve analysis demonstrated superior net benefit for the full nomogram across threshold probabilities of approximately 10%-100%, outperforming both single-parameter models and the “treat-all” or “treat-none” strategies (Figure 4B). The peak net benefit within the 10%-40% threshold range suggests optimal utility for clinical decision-making in this interval. Together, these results confirm that the nomogram provides statistically robust and clinically actionable risk stratification, supporting its integration into perioperative care pathways.

Figure 4 Clinical validation of the prediction nomogram. A: Nomogram-based risk stratification: B: Decision curve analysis; C: Clinical impact curve. BNP: B-type natriuretic peptide; LAVI: Left atrial volume index.

Analysis of the clinical impact curve reveals the trade-offs in clinical decision-making using the prediction model across different risk thresholds (Figure 4C): The total number of high-risk patients decreases as the threshold increases, while the number of true positives remains stable, and the number of false positives, initially high at low thresholds, decreases with rising thresholds. At low thresholds (0-0.2), the model exhibits high sensitivity but a high false positive rate, potentially leading to overtreatment. At moderate thresholds (0.2-0.6), an optimal balance is achieved, with high true positive detection and significantly reduced false positives. At high thresholds (> 0.6), specificity improves, false positives are minimized, making this range suitable for scenarios requiring avoidance of misdiagnosis. The curve demonstrates that the model robustly identifies true positives across a wide threshold range, while allowing clinicians to flexibly select thresholds based on their tolerance for false positives vs missed diagnoses, thereby optimizing individualized treatment strategies.

DISCUSSION

Pancreatic cancer with vascular invasion presents major challenges in surgical oncology, often requiring complex resections that elevate perioperative risk. MPCCs significantly contribute to morbidity and mortality in these patients[12], typically resulting from the interplay of aggressive tumor biology, comorbidities, and physiological stress due to prolonged surgery[13]. Conventional risk tools such as the Revised Cardiac Risk Index lack specificity for this population, failing to incorporate both the unique pathophysiology of pancreatic cancer and advanced echocardiographic parameters that reflect myocardial vulnerability[14]. To address this gap, we developed and validated a novel nomogram integrating preoperative BNP, average E/e’ ratio, and LAVI to predict postoperative cardiac complications risk in this high-risk cohort.

Our analysis of 138 patients who underwent radical surgery for pancreatic cancer with vascular invasion revealed a postoperative cardiac complications incidence of 32.6%, with arrhythmias being the most frequent event. Multivariable logistic regression identified BNP, average E/e’ ratio, and LAVI as independent predictors. The resulting nomogram, which integrates these three variables, demonstrated excellent discriminatory ability (C-index: 0.874, 95%CI: 0.812-0.936) and good calibration (mean absolute error = 0.027), indicating a high concordance between predicted and observed outcomes. Decision curve analysis further confirmed the model’s clinical utility, showing a positive net benefit across a wide range of probability thresholds, thereby supporting its potential for guiding preoperative decision-making and resource allocation.

The findings of our study are strongly supported by the existing literature in cardiothoracic surgery and surgical risk assessment[15]. In our cohort, elevated BNP reflects limited cardiac reserve. Prolonged surgery with vascular clamping causes volume shifts and inflammatory surge, unmasking occult heart failure[16,17]. This population’s hypercoagulable state and frequent neoadjuvant chemotherapy exposure further compromise myocardial tolerance to hemodynamic perturbations[18,19]. Average E/e’ ratio emerged as a critical predictor, reflecting diastolic dysfunction specifically relevant to this surgical population. These tumors require extensive dissection and vascular reconstruction, causing prolonged cross-clamping and fluid shifts[20]. Stiff ventricles (high E/e’) cannot accommodate these fluctuations: Hypovolemia causes hypotension, while fluid resuscitation precipitates pulmonary edema[21]. This surgical-specific hemodynamic vulnerability explains why E/e’ outperformed traditional systolic indices in our model.

LAVI captures chronic left atrial remodeling from sustained pressure overload, creating an arrhythmogenic substrate particularly vulnerable in this setting. The combination of surgical stress, electrolyte shifts, and sympathetic activation during major vascular manipulation precipitates atrial fibrillation - observed in 30.4% of our MPCC cases. Notably, left atrial enlargement may also reflect chronic venous congestion from portal hypertension in this population with peri-pancreatic vascular involvement, representing a distinct pathophysiological pathway not captured by conventional cardiac risk scores[22,23].

Together, BNP, E/e’, and LAVI assess cardiac function from biochemical, functional, and structural perspectives. This triad captures the multifaceted nature of perioperative cardiac risk more accurately than traditional tools. The high predictive accuracy demonstrated by our model suggests that this echocardiographic and biomarker triad offers a superior risk assessment compared to traditional evaluations relying primarily on functional status (e.g., New York Heart Association class or Metabolic Equivalents of Tasks).

It is crucial to emphasize the dual role of anesthetic management. Anesthetic agents themselves are contributing factors; their myocardial depressant and vasodilatory effects can easily disrupt the fragile hemodynamic balance in high-risk patients. More importantly, however, anesthesia can serve as a critical tool for risk modification[24]. The nomogram we developed holds significant clinical implications, providing surgeons, anesthesiologists, and cardiologists with a practical, evidence-based tool for individualized risk stratification. For patients identified as high-risk, a series of targeted interventions can be proactively implemented based on this assessment. This includes preoperative optimization (e.g., cardiology consultation, medical management of volume status and diastolic function), tailored intraoperative hemodynamic monitoring, and close postoperative surveillance in high-dependency settings[25,26]. The value of this model lies in its ability to shift the focus beyond the simplistic binary question of operability towards a more nuanced discussion about risk modification and evidence-based shared decision-making. It transforms anesthetic management from a reactive process to a proactive, defensive strategy, directly addressing the specific pathophysiological vulnerabilities revealed by these three parameters, thereby effectively mitigating risk.

The nomogram developed in this study offers a practical, bedside-ready tool for individualized perioperative cardiac risk assessment in patients with vascular-invasive pancreatic cancer. By translating three routinely available preoperative parameters into a quantitative probability, it enables clinicians to stratify patients into actionable risk categories and allocate perioperative resources accordingly. For low-risk patients, standard perioperative care suffices; for intermediate-risk patients, enhanced hemodynamic monitoring and anesthesia optimization are warranted; for high-risk patients, proactive cardiology consultation and tailored intraoperative management are indicated; and for very high-risk patients, intensive perioperative management in a high-dependency setting becomes imperative. This risk-stratified approach transforms perioperative care from a reactive process to a proactive, precision-based strategy, ultimately aiming to reduce cardiac morbidity and improve surgical outcomes in this vulnerable population.

This study has several limitations that must be acknowledged. First, the retrospective, single-center design introduces selection bias and limits generalizability to other populations. The homogeneity of our cohort, drawn from a single grade A tertiary hospital, may not reflect the full spectrum of disease severity, surgical expertise, or perioperative management practices encountered in broader clinical settings. Second, although the sample size of 138 patients was adequate for initial model development according to established statistical criteria, the relatively modest number of events (45 MPCCs) may restrict the precision of effect estimates and the stability of the predictive model. Consequently, these findings should be considered preliminary until they are externally validated in larger, multicenter, prospective cohorts. External validation is essential to confirm the model’s calibration, discrimination, and clinical utility across diverse healthcare systems and to establish its generalizability before routine clinical implementation. Future research should therefore prioritize prospective multicenter validation, cost-effectiveness analyses of nomogram-guided interventions, and the potential integration of novel biomarkers or genetic markers to further enhance predictive accuracy and refine risk stratification.

CONCLUSION

In summary, we developed and validated a robust nomogram leveraging routine preoperative parameters (BNP, average E/e’ ratio, LAVI) to effectively predict postoperative cardiac complications in patients with pancreatic cancer and vascular invasion. This tool enables early high-risk identification and facilitates targeted management, advancing cardiac risk assessment from a traditional functional evaluation to a comprehensive multidimensional profile for guiding precision perioperative care.

ACKNOWLEDGEMENTS

We thank the patients for their great help in this report.