Abstract
Background
Frailty is associated with adverse clinical outcomes and is recognized as an important predictor of prognosis. Yet, practical frailty-based tools to assess the risk of postoperative complications after radical esophagectomy are still inadequate. The aim of this study was to evaluate the predictive performance of the Memorial Sloan Kettering-Frailty Index (MSK-FI) and to establish a more comprehensive model for forecasting postoperative complications in this patient population.
Methods
We retrospectively analyzed 410 patients who underwent radical esophagectomy between 2022 and 2023. Preoperative frailty status of these patients was calculated using MSK-FI. Patients were accordingly classified as frail (MSK-FI ≥2, n=86) or non-frail (MSK-FI <2, n=324). A predictive nomogram was subsequently conducted by incorporating clinical and surgical characteristics.
Results
Compared with the non-frail group, patients in the frail group had significantly higher rates of severe postoperative complications (26.74% vs. 14.20%; P=0.006) and longer hospital stays (11 vs. 10 days, P=0.01). Univariate logistic regression analysis identified age ≥65 years [odds ratio (OR) =2.11, P=0.02], neoadjuvant therapy (OR =1.86, P=0.03), MSK-FI ≥2 (OR =2.21, P=0.008), cervical esophagogastric anastomosis (OR =2.02, P=0.01), and prolonged surgery duration (OR =1.79, P=0.04) as risk factors for severe postoperative complications. These variables were incorporated into multivariable stepwise logistic regression to develop a predictive nomogram model, which demonstrated an area under the curve (AUC) of 0.700 [95% confidence interval (CI): 0.637–0.767].
Conclusions
Preoperative frailty status is associated with an increased risk of adverse perioperative outcomes in patients after radical esophagectomy. The integration of MSK-FI and clinical characteristics in nomogram provides a practical tool for predicting severe postoperative complications.
Keywords: Frailty, esophagectomy, postoperative complications, Memorial Sloan Kettering-Frailty Index (MSK-FI)
Highlight box.
Key findings
• Patients with esophageal cancer who have a high Memorial Sloan Kettering-Frailty Index (MSK-FI) are more likely to develop severe complications after surgery. Combining MSK-FI with clinical characteristics can help predict severe complications after radical esophagectomy.
What is known and what is new?
• Several studies have indicated the association between frailty and postoperative complications of esophageal cancer. However, these studies mainly focus on using frailty scores alone and do not include patients’ clinical characteristics.
• We combined MSK-FI and clinical variables to develop a predictive model and validated it.
What is the implication, and what should change now?
• The predictive value of MSK-FI improves when combined with broader clinical factors. Taking proactive management for patients with a high predicted probability may bring benefits.
Introduction
The Global Burden of Disease (GBD) study reports that Asia carries a disproportionate burden of esophageal cancer, accounting for 75.0% of incident cases and 74.1% of related deaths worldwide (1). For patients with locally advanced esophageal cancer, multimodal therapy combining neoadjuvant treatment with esophagectomy has become the standard of care (2). However, despite substantial advancements in surgical techniques and perioperative care, esophagectomy remains one of the most invasive and high-risk procedures in thoracic surgery. Postoperative outcomes are often further compromised by the intrinsic nutritional deficiencies associated with the disease (3). In addition, the prevalence of elderly patients—a population often characterized by reduced physiological reserve—demands more comprehensive consideration when performing esophagectomy (4). These converging factors elevate the incidence of adverse outcomes and mortality rates after surgery, emphasizing the importance of improved, individualized risk stratification (5).
The concept of frailty has been introduced for decades, and refers to an age-related clinical condition and considered to be a decline in physiological function across multiple organ systems, leading to an increased susceptibility to stressors and adverse health outcomes (6). In recent years, frailty has been applied in multiple surgical outcome studies and has been shown to be associated with poor postoperative results (7-9). Although the Comprehensive Geriatric Assessment (CGA) remains the reference standard for frailty evaluation, its complexity limits its routine use (10). Simplified alternatives such as the Charlson Comorbidity Index (CCI) and Fried’s Frailty Phenotype (FFP) each capture partial aspects of frailty—disease burden and physical performance, respectively—but fail to encompass its multidimensional nature (11). The Memorial Sloan Kettering-Frailty Index (MSK-FI) is an effective tool for assessing frailty, incorporating 10 comorbidities and 1 functional status: chronic lung disease, diabetes, congestive heart failure, myocardial infarction, coronary artery disease, hypertension, peripheral vascular disease, impaired sensorium, cerebrovascular accident, and transient ischemic attack (Table S1) (12). Compared with CGA and FFP, the MSK-FI provides comparable predictive accuracy while being clinically feasible for routine use.
Although the MSK-FI is a valuable tool for assessing frailty and stratifying surgical risk, its use alone may lack sufficient discriminative capacity for robust prediction of postoperative complications. In this study, we conducted a retrospective study for patients who underwent radical esophagectomy at Jinling Hospital, Medical School of Nanjing University, between 2022 and 2023. Our primary goal was to evaluate the predictive performance of the MSK-FI and to develop a more comprehensive model for forecasting postoperative complications. Besides, we also collected clinical data of patients who underwent radical esophagectomy in the first half of 2024 for external validation. We present this article in accordance with the TRIPOD reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-aw-2357/rc).
Methods
Data collection
We retrospectively collected clinical records of patients who underwent radical esophagectomy at Jinling Hospital between 2022 and 2023 for developing the model, with data from the first half of 2024 used for validation. The main inclusion criteria were as follows: (I) pathologically confirmed esophageal cancer after surgery; (II) absence of other concurrent malignancies; and (III) completion of radical resection. The exclusion criteria were as follows: (I) missing clinical data; (II) inability to complete evaluation; and (III) pre-existing severe organ dysfunction. After excluding 4 patients with incomplete preoperative testing, 2 with concurrent pneumonectomy, and 1 with an incomplete case record, a total of 410 patients were included in the final analysis. Data were collected from the hospital’s electronic medical record system, including demographic characteristics, preoperative assessments, comorbidities, intraoperative variables, and postoperative complications. Prolonged operative duration was defined as an operative duration exceeding the 75th percentile within each category, stratified by the combination of surgical method and approach. The pathological characteristics of cancers were staged using the tumor-node-metastasis (TNM) classification provided by the 8th edition of the American Joint Committee on Cancer (AJCC) (13). The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Jinling Hospital (No. 2025DZKY-023-02) and individual consent for this retrospective analysis was waived. Postoperative complications were classified using the Clavien-Dindo (C-D) classification system, with major complications defined as those of grade III or higher (14), for example, pleural effusion requiring invasive drainage, unplanned reoperation, respiratory failure requires endotracheal intubation, etc. Anastomotic leakage was defined as an abnormal passage caused by the interruption or defect in the integrity of the esophagus, anastomosis, staple line, or conduit, typically diagnosed through diagnostic imaging, endoscopy, or clinical presentation (15).
Assessment of frailty status
Frailty status was assessed using the MSK-FI. The specific assessment variables were provided in the Table S1. Comorbidities were adjudicated by clinical severity (e.g., moderate or severe) through chart review and multidisciplinary consensus. Only those meeting predetermined severity criteria were assigned points in the MSK-FI calculation, in order to avoid overestimation of frailty due to minor or well-controlled chronic conditions. Based on the optimal cut-off value defined below, patients with an MSK-FI score of less than 2 were classified as non-frail. It was noteworthy that the diagnoses of cerebrovascular accident and transient ischemic attack were occasionally indeterminate during data collection, as the initial evaluations were conducted at external hospitals. However, neurologists’ consultation records were available, and these two items were combined into a single category in the analysis table.
Statistical analysis
All statistical analyses were conducted using R software (version 4.5.0) and Stata (version 18.0). Categorical variables were analyzed with the chi-square test or Fisher’s exact test, while continuous variables were described as median and interquartile intervals, and differences were compared using rank-sum test. Univariate analyses were first performed to identify potential risk factors associated with severe postoperative complications. Variables with a univariable P value <0.10 were subsequently entered into a multivariable logistic regression model, in which a backward stepwise elimination procedure was applied to identify independent predictors for inclusion in the nomogram. The final selected predictors were used to construct a point-based nomogram: the linear predictor (LP) from the logistic model is
| [1] |
and the predicted probability is
| [2] |
In the nomogram each predictor’s contribution (points) was proportional to ; total points were mapped back to p via the logistic transformation. The nomogram was a visual instrument to depict the impact of individual predictors within the multivariable regression model, and the model’s performance was then examined using the receiver operating characteristic (ROC) curve to evaluate discriminative accuracy, and decision curve analysis (DCA) to appraise its potential clinical benefit. The optimal cut-off value was determined post hoc using ROC curve analysis, based on the maximum Youden index. All P values were two-sided tests, and statistical significance was set at P<0.05.
Results
The cohort
The study included 410 consecutive patients who underwent radical esophagectomy for esophageal cancer. Based on frailty assessment, 324 (79.02%) were classified as non-frail and 86 (20.98%) as frail. Significant demographic and clinical differences were observed between the two groups. Frail patients had a higher median age {70 years [interquartile range (IQR), 66–75] vs. 67 (IQR, 61–72) years, P<0.001}. McKeown esophagectomy was the most common surgical procedure (224 cases, 54.63%), performed more frequently in the non-frail group (188 cases, 58.02%) than in the frail group (36 cases, 41.86%; P<0.05) (Table 1). The baseline characteristics of patients according to frailty status were presented in Table S2.
Table 1. Clinical characteristics of patients undergoing radical esophagectomy.
| Variables | Total (n=410) | MSK-FI <2 (n=324) | MSK-FI ≥2 (n=86) | P |
|---|---|---|---|---|
| Age, years | 68 [63, 73] | 67 [61, 72] | 70 [66, 75] | <0.001 |
| Gender | 0.37 | |||
| Female | 91 (22.20) | 75 (23.15) | 16 (18.60) | |
| Male | 319 (77.80) | 249 (76.85) | 70 (81.40) | |
| Surgery methods | 0.02 | |||
| Sweet | 82 (20.00) | 62 (19.14) | 20 (23.26) | |
| McKeown | 224 (54.63) | 188 (58.02) | 36 (41.86) | |
| Ivor Lewis | 104 (25.37) | 74 (22.84) | 30 (34.88) | |
| Operative approach | 0.33 | |||
| OE | 82 (20.00) | 62 (19.14) | 20 (23.26) | |
| VAMIE | 181 (44.15) | 139 (42.90) | 42 (48.84) | |
| RAMIE | 147 (35.85) | 122 (37.65) | 25 (29.07) | |
| Anastomosis location | 0.06 | |||
| Cervical | 250 (60.98) | 205 (63.27) | 45 (52.33) | |
| Intrathoracic | 160 (39.02) | 63 (36.73) | 41 (47.67) | |
| Histology | 0.12 | |||
| Squamous cell carcinoma | 269 (65.61) | 216 (66.67) | 53 (61.63) | |
| Adenocarcinoma | 93 (22.68) | 67 (20.68) | 26 (30.23) | |
| Others | 48 (11.71) | 41 (12.65) | 7 (8.14) | |
| T | 0.19 | |||
| 0 | 41 (10.00) | 36 (11.11) | 5 (5.81) | |
| 1 | 57 (13.90) | 49 (15.12) | 8 (9.30) | |
| 2 | 99 (24.15) | 75 (23.15) | 24 (27.91) | |
| 3 | 213 (51.95) | 164 (50.62) | 49 (56.98) | |
| N | 0.39 | |||
| 0 | 228 (55.61) | 177 (54.63) | 51 (59.30) | |
| 1 | 90 (21.95) | 73 (22.53) | 17 (19.77) | |
| 2 | 56 (13.66) | 48 (14.81) | 8 (9.30) | |
| 3 | 36 (8.78) | 26 (8.02) | 10 (11.63) | |
| Pathological grade | 0.08 | |||
| 0 | 37 (9.02) | 33 (10.19) | 4 (4.65) | |
| 1 | 58 (14.15) | 50 (15.43) | 8 (9.30) | |
| 2 | 177 (43.17) | 131 (40.43) | 46 (53.49) | |
| 3 | 102 (24.88) | 84 (25.93) | 18 (20.93) | |
| 4 | 36 (8.78) | 26 (8.02) | 10 (11.63) | |
| Number of dissected LNs | 21 [14, 27] | 21 [14, 27] | 18 [13, 27] | 0.41 |
| ASA | <0.001 | |||
| 1 | 1 (0.24) | 1 (0.31) | 0 (0.00) | |
| 2 | 290 (70.73) | 246 (75.93) | 44 (51.16) | |
| 3 | 117 (28.54) | 76 (23.46) | 41 (47.67) | |
| 4 | 2 (0.49) | 1 (0.31) | 1 (1.16) | |
| Neoadjuvant therapy | 0.09 | |||
| No | 310 (75.61) | 239 (73.77) | 71 (82.56) | |
| Yes | 100 (24.39) | 85 (26.23) | 45 (17.44) | |
| Infection | 0.19† | |||
| No | 376 (91.71) | 294 (90.74) | 82 (95.35) | |
| Yes | 34 (8.29) | 30 (9.26) | 4 (4.65) | |
| Arrhythmia | 0.31 | |||
| No | 390 (95.12) | 310 (95.68) | 80 (93.02) | |
| Yes | 20 (4.88) | 14 (4.32) | 6 (6.98) | |
| Urological diseases | 0.08 | |||
| No | 380 (92.68) | 304 (93.83) | 76 (88.37) | |
| Yes | 30 (7.32) | 20 (6.17) | 10 (11.63) | |
| Digestive diseases | 0.76† | |||
| No | 393 (95.85) | 311 (95.99) | 82 (95.35) | |
| Yes | 17 (4.15) | 13 (4.01) | 4 (4.65) | |
| Thyroid diseases | >0.99† | |||
| No | 404 (98.54) | 319 (98.46) | 85 (98.84) | |
| Yes | 6 (1.46) | 5 (1.54) | 1 (1.16) | |
| Malignant diseases | 0.29† | |||
| No | 398 (97.07) | 316 (97.53) | 82 (95.35) | |
| Yes | 12 (2.93) | 8 (2.47) | 4 (4.65) | |
| History of smoking | >0.99 | |||
| No | 267 (65.12) | 211 (65.12) | 56 (65.12) | |
| Yes | 143 (34.88) | 113 (34.88) | 30 (34.88) | |
| History of drinking | 0.55 | |||
| No | 270 (65.85) | 211 (62.73) | 59 (68.60) | |
| Yes | 140 (34.15) | 113 (37.27) | 27 (31.40) | |
| BMI, kg/m2 | 23.3 [21.5, 25.1] | 23.2 [21.3, 25.1] | 24.0 [22.0, 25.2] | 0.10 |
| Intraoperative blood loss, mL | 255 [200, 350] | 300 [200, 400] | 200 [200, 300] | 0.16 |
| Intraoperative blood transfusion performed | 0.50 | |||
| No | 396 (96.59) | 314 (96.91) | 82 (95.35) | |
| Yes | 14 (3.41) | 10 (3.09) | 4 (4.65) | |
| Surgery duration, min | 305 [210, 362] | 310 [220, 365] | 265 [210, 348] | 0.12 |
| Sweet | 200 [180, 230] | 190 [170, 230] | 210 [190, 250] | 0.12 |
| VA McKeown | 360 [320, 410] | 360 [310, 410] | 380 [320, 430] | 0.36 |
| VA Ivor Lewis | 340 [310, 380] | 340 [310, 380] | 340 [300, 380] | 0.65 |
| RA McKeown | 210 [175, 240] | 200 [160, 240] | 210 [180, 240] | 0.37 |
| RA Ivor Lewis | 325 [306, 368] | 330 [312, 367] | 320 [300, 370] | 0.79 |
| Surgery duration, min | 0.35 | |||
| Short-term group | 293 (71.46) | 235 (72.53) | 58 (67.44) | |
| Long-term group | 117 (28.54) | 89 (27.47) | 28 (32.56) | |
| Albumin, g/L | 40.6 [38.1, 43.4] | 40.8 [38.1, 43.6] | 40.2 [38.2, 41.8] | 0.09 |
Data are presented as median [IQR] or n (%). †, Fisher’s exact test, two-sided. ASA, American Society of Anesthesiologists; BMI, body mass index; IQR, interquartile range; LN, lymph node; MSK-FI, Memorial Sloan Kettering-Frailty Index; N, node; OE, open esophagectomy; RA, robot-assisted; RAMIE, robot-assisted minimally invasive esophagectomy; T, tumor; VA, video-assisted; VAMIE, video-assisted minimally invasive esophagectomy.
Postoperative outcomes
The frail group had a significantly higher incidence of postoperative complications compared to the non-frail group, including transfusion requirements, arrhythmia, need for invasive procedures, unplanned reoperations, and respiratory failure, with all P values less than 0.05. Totally, frailty group suffered severe complications (23 cases, 26.74% vs. 46 cases, 14.20%; P=0.006). Additionally, frail patients experienced prolonged hospitalization [median 11 (IQR, 8–18) vs. 10 (IQR, 8–14) days, P=0.01]. No other statistically significant differences were observed in the remaining perioperative parameters (Table 2).
Table 2. Postoperative outcomes following radical esophagectomy in patients stratified by MSK-FI.
| Variables | Total (n=410) | MSK-FI <2 (n=324) | MSK-FI ≥2 (n=86) | P |
|---|---|---|---|---|
| Pneumonia | 0.055 | |||
| No | 296 (72.20) | 241 (74.38) | 55 (63.95) | |
| Yes | 114 (27.80) | 83 (25.62) | 31 (36.05) | |
| Pleural effusion or pneumothorax | 0.17 | |||
| No | 377 (91.95) | 301 (92.90) | 76 (88.37) | |
| Yes | 33 (8.05) | 23 (7.10) | 10 (11.63) | |
| Transfuse | 0.004 | |||
| No | 350 (85.37) | 285 (87.96) | 65 (75.58) | |
| Yes | 60 (14.63) | 39 (12.04) | 21 (24.42) | |
| Incision complication | 0.72† | |||
| No | 398 (97.07) | 315 (97.22) | 83 (96.51) | |
| Yes | 12 (2.93) | 9 (2.78) | 3 (3.49) | |
| Chylothorax | 0.28† | |||
| No | 405 (98.78) | 321 (99.07) | 84 (97.67) | |
| Yes | 5 (1.22) | 3 (0.93) | 2 (2.33) | |
| Cerebral infarction | 0.51† | |||
| No | 407 (99.27) | 322 (99.38) | 85 (98.84) | |
| Yes | 3 (0.73) | 2 (0.62) | 1 (1.16) | |
| Delirium | 0.38† | |||
| No | 408 (99.51) | 323 (99.69) | 85 (98.84) | |
| Yes | 2 (0.49) | 1 (0.31) | 1 (1.16) | |
| Arrhythmia | 0.02 | |||
| No | 397 (96.83) | 317 (97.84) | 80 (93.02) | |
| Yes | 13 (3.17) | 7 (2.16) | 6 (6.98) | |
| Invasive procedures | 0.02 | |||
| No | 347 (84.63) | 281 (86.73) | 66 (76.74) | |
| Yes | 63 (15.37) | 43 (13.27) | 20 (23.26) | |
| Unplanned reoperation | 0.02† | |||
| No | 404 (98.54) | 322 (99.38) | 82 (95.35) | |
| Yes | 6 (1.46) | 2 (0.62) | 4 (4.65) | |
| Respiratory failure | 0.01 | |||
| No | 372 (90.73) | 300 (92.59) | 72 (83.72) | |
| Yes | 38 (9.27) | 24 (7.41) | 14 (16.28) | |
| Anastomotic leakage | 0.76 | |||
| No | 366 (89.27) | 290 (89.51) | 76 (88.37) | |
| Yes | 44 (10.73) | 34 (10.49) | 10 (11.63) | |
| C-D classification | 0.006 | |||
| <3 | 341 (83.17) | 278 (85.80) | 63 (73.26) | |
| ≥3 | 69 (16.83) | 46 (14.20) | 23 (26.74) | |
| Readmission | >0.99† | |||
| No | 396 (96.59) | 313 (96.60) | 83 (96.51) | |
| Yes | 14 (3.41) | 11 (3.40) | 3 (3.49) | |
| Hospital stay, days | 10 [8, 14] | 10 [8, 14] | 11 [8, 18] | 0.01 |
Data are presented as median [IQR] or n (%). †, Fisher’s exact test, two-sided. C-D, Clavien-Dindo; IQR, interquartile range; MSK-FI, Memorial Sloan Kettering-Frailty Index.
Factors associated with postoperative complications
Univariate logistic regression analysis identified that age ≥65 years [odds ratio (OR) =2.11, 95% confidence interval (CI): 1.12–3.94, P=0.02], robot-assisted minimally invasive esophagectomy (RAMIE) (OR =2.44, 95% CI: 1.11–5.39, P=0.03), neoadjuvant therapy (OR =1.86, 95% CI: 1.07–3.25, P=0.03), MSK-FI ≥2 (OR =2.21, 95% CI: 1.25–3.90, P=0.008), cervical esophagogastric anastomosis (OR =2.02, 95% CI: 1.13–3.61, P=0.01), and prolonged surgery duration (OR =1.79, 95% CI: 1.04–3.08, P=0.04) were significant predictors of severe complications, with all P values less than 0.05. Multivariable stepwise logistic regression analysis demonstrated that age ≥65 years (OR =2.12, 95% CI: 1.11–4.07, P=0.02), MSK-FI ≥2 (OR =2.28, 95% CI: 1.25–4.18, P=0.007), and cervical esophagogastric anastomosis (OR =2.23, 95% CI: 1.22–4.10, P=0.01) were independently associated with an increased risk of postoperative complications. Additionally, neoadjuvant therapy (OR =1.80, 95% CI: 1.00–3.24, P=0.051) and prolonged surgery duration (OR =1.75, 95% CI: 1.00–3.06, P=0.051) exhibited borderline significance. No multicollinearity was detected among the covariates (Table 3). Subgroup analyses were performed to evaluate the robustness of the association between MSK-FI and postoperative complications. The results demonstrated that the association between MSK-FI and postoperative complications was generally consistent across clinically relevant subgroups, with no significant interactions observed (Figure S1). Furthermore, interaction analysis confirmed that the effect of MSK-FI on postoperative complications was not significantly modified by age (Table S3).
Table 3. Univariate and multivariable stepwise logistic regression analyses of risk factors for short-term severe complications in patients after radical esophagectomy.
| Variables | Univariate analysis | Multivariate analysis | |||
|---|---|---|---|---|---|
| OR (95% CI) | P | OR (95% CI) | P | ||
| Gender | |||||
| Female | 1.00 (reference) | ||||
| Male | 0.93 (0.50–1.73) | 0.83 | |||
| Age | |||||
| <65 years | 1.00 (reference) | 1.00 (reference) | |||
| ≥65 years | 2.11 (1.12–3.94) | 0.02 | 2.12 (1.11–4.07) | 0.02 | |
| ASA | |||||
| 1–2 | 1.00 (reference) | ||||
| 3–4 | 0.84 (0.47–1.51) | 0.55 | |||
| Surgery methods | 0.07 | ||||
| Sweet | 1.00 (reference) | ||||
| McKeown | 2.10 (0.98–4.50) | 0.058 | |||
| Ivor Lewis | 1.26 (0.52–3.08) | 0.61 | |||
| Operative approach | 0.03 | ||||
| OE | 1.00 (reference) | ||||
| VAMIE | 1.36 (0.61–3.05) | 0.46 | |||
| RAMIE | 2.44 (1.11–5.39) | 0.03 | |||
| Anastomosis location | |||||
| Intrathoracic | 1.00 (reference) | 1.00 (reference) | |||
| Cervical | 2.02 (1.13–3.61) | 0.01 | 2.23 (1.22–4.10) | 0.01 | |
| Number of dissected LNs | |||||
| <15 | 1.00 (reference) | ||||
| ≥15 | 1.81 (0.93–3.52) | 0.07 | |||
| Intraoperative blood transfusion performed | |||||
| No | 1.00 (reference) | ||||
| Yes | 2.04 (0.62–6.69) | 0.27 | |||
| Histology | 0.40 | ||||
| Squamous cell carcinoma | 1.00 (reference) | ||||
| Adenocarcinoma | 0.65 (0.33–1.28) | 0.21 | |||
| Others | 0.51 (0.32–1.76) | 0.51 | |||
| T stage | |||||
| ≤2 | 1.00 (reference) | ||||
| 3–4 | 0.91 (0.61–1.35) | 0.64 | |||
| N stage | |||||
| 0–1 | 1.00 (reference) | ||||
| 2–3 | 0.69 (0.35–1.34) | 0.26 | |||
| Stage | |||||
| ≤II | 1.00 (reference) | ||||
| III–IV | 0.99 (0.65–1.51) | 0.98 | |||
| Neoadjuvant therapy | |||||
| No | 1.00 (reference) | 1.00 (reference) | |||
| Yes | 1.86 (1.07–3.25) | 0.03 | 1.80 (1.00–3.24) | 0.051 | |
| Infection | |||||
| No | 1.00 (reference) | ||||
| Yes | 1.31 (0.55–3.15) | 0.55 | |||
| Urological diseases | |||||
| No | 1.00 (reference) | ||||
| Yes | 1.26 (0.49–3.20) | 0.64 | |||
| Digestive diseases | |||||
| No | 1.00 (reference) | ||||
| Yes | 0.30 (0.04–2.29) | 0.17 | |||
| History of malignant diseases | |||||
| No | 1.00 (reference) | ||||
| Yes | 1.68 (0.44–6.36) | 0.47 | |||
| Thyroid diseases | |||||
| No | 1.00 (reference) | ||||
| Yes | 0.00 (0.00–4.22) | 0.39 | |||
| MSK-FI | |||||
| <2 | 1.00 (reference) | 1.00 (reference) | |||
| ≥2 | 2.21 (1.25–3.90) | 0.008 | 2.28 (1.25–4.18) | 0.007 | |
| History of smoking | |||||
| No | 1.00 (reference) | ||||
| Yes | 0.79 (0.45–1.37) | 0.39 | |||
| History of drinking | |||||
| No | 1.00 (reference) | ||||
| Yes | 0.96 (0.55–1.66) | 0.88 | |||
| BMI, kg/m2 | 0.50 | ||||
| <18.5 | 1.00 (reference) | ||||
| 18.5–24 | 1.03 (0.22–4.86) | 0.97 | |||
| >24 | 1.37 (0.81–2.32) | 0.24 | |||
| Albumin, g/L | |||||
| ≥35 | 1.00 (reference) | ||||
| <35 | 1.32 (0.52–3.37) | 0.57 | |||
| Surgery duration | |||||
| Short-term group | 1.00 (reference) | 1.00 (reference) | |||
| Long-term group | 1.79 (1.04–3.08) | 0.04 | 1.75 (1.00–3.06) | 0.051 | |
ASA, American Society of Anesthesiologists; BMI, body mass index; CI, confidence interval; LN, lymph node; MSK-FI, Memorial Sloan Kettering-Frailty Index; N, node; OE, open esophagectomy; OR, odds ratio; RAMIE, robot-assisted minimally invasive esophagectomy; T, tumor; VAMIE, video-assisted minimally invasive esophagectomy.
Development and evaluation of the nomogram
A nomogram based on the multivariable stepwise regression results was constructed to stratify the risk of severe postoperative complications. The nomogram showed moderate discriminative performance, with an area under the curve (AUC) of 0.700 (95% CI: 0.637–0.767), outperforming MSK-FI alone (AUC =0.599, 95% CI: 0.528–0.671). The calibration curve showed good concordance between predicted and observed outcomes, supporting its reliable accuracy. Furthermore, DCA further confirmed the model’s clinical utility, indicating its potential value in postoperative risk assessment and individualized patient management (Figure 1).
Figure 1.
Nomogram constructed based on the results of multivariable stepwise regression. (A) Nomogram for predicting postoperative complications. (B) ROC curve of the nomogram. (C) DCA evaluating the clinical utility of the nomogram. (D) Calibration curve assessing the agreement between predicted and observed outcomes. AUC, area under the curve; DCA, decision curve analysis; MSK-FI, Memorial Sloan Kettering-Frailty Index; ROC, receiver operating characteristic.
To validate our newly developed nomogram model, we collected clinical data from patients who underwent radical esophagectomy at our clinical center in the first half of 2024. The data were processed using the same methodology as previously described (n=68). The model’s performance was assessed using DCA, calibration curve, and ROC curve (Figure 2). The AUC was 0.785 (95% CI: 0.603–0.968). Based on the model-derived cut-off value of 0.158, we stratified the patients in the validation cohort into distinct groups, the results were presented in Table 4.
Figure 2.
External validation of the nomogram. (A) DCA evaluating the clinical utility of the nomogram in the external validation cohort. (B) Calibration curve assessing the agreement between predicted and observed outcomes. (C) ROC curve of the nomogram in the external validation cohort. AUC, area under the curve; DCA, decision curve analysis; ROC, receiver operating characteristic.
Table 4. Postoperative complications stratified by C-D grade and risk group based on the predictive model.
| Variables | Low-risk group, n (%) | High-risk group, n (%) | P |
|---|---|---|---|
| C-D <3 | 33 (94.29) | 25 (75.76) | 0.04† |
| C-D ≥3 | 2 (5.71) | 8 (24.24) |
†, Fisher’s exact test, two-sided. C-D, Clavien-Dindo.
Discussion
The prevention of short-term postoperative complications is widely acknowledged as a critical goal across all surgical specialties, as it directly influences patient recovery, length of hospital stay, and even long-term prognosis (16). Frailty, defined as a state of diminished physiological reserve and resilience, has been consistently associated with increased perioperative morbidity and mortality in various surgical populations (17,18). These studies support the incorporation of frailty assessment into surgical decision-making, particularly in oncologic surgery. The MSK-FI has been widely used as a practical frailty assessment tool, although most studies have focused on older populations due to its original development in elderly cohorts (12,19). Increasing evidence now suggests that chronological age and frailty are not necessarily synonymous; frailty represents a decline in physiological reserve and homeostatic capacity that can also manifest in younger individuals (20). In our study, both subgroup and interaction analyses demonstrated that age and MSK-FI contributed independently to postoperative risk, indicating that frailty, as captured by the MSK-FI, reflects vulnerability beyond the effects of aging itself. These findings support the broader applicability of the MSK-FI in patients with esophageal cancer across different age groups, although further validation in diverse and younger populations remains warranted.
It is noteworthy that although recent research has validated the prognostic value of frailty indices for postoperative complications, most studies lack a visual and integrative approach to facilitate clinical application (7,19). Specifically, in our study, the MSK-FI alone yielded an AUC of 0.599, indicating it is insufficient to use MSK-FI as a standalone indicator. Moreover, as described, the standard treatment for esophageal cancer involves neoadjuvant therapy followed by radical esophagectomy, which has been shown in the literature to significantly improve patient survival (21). However, with the advancement of techniques and therapeutics, the impact of neoadjuvant therapy on postoperative cardiopulmonary complications and surgical complications remains controversial (22,23). Some studies have shown that with increasing age, patients undergoing neoadjuvant therapy may be more susceptible to adverse reactions induced by surgical or anesthetic stimuli, leading to an elevated risk of severe postoperative complications (24,25). Nevertheless, the associated tissue edema, fibrosis, and potential immune dysregulation increase the technical complexity of the surgical procedure, and consensus is lacking regarding the optimal timing for surgery following such treatment (26). Given that radical esophagectomy is a highly invasive procedure with complex surgical techniques, such patients are consequently more susceptible to serious complications after systemic therapy. Furthermore, intrathoracic anastomosis has been demonstrated to be superior to cervical anastomosis with respect to complication rates (27); additionally, extended operative duration correlates with inferior short-term surgical outcomes (28).
To this end, we constructed a composite model that integrates the MSK-FI with clinical characteristics to predict short-term postoperative complications in patients with esophageal cancer, aiming to inform decision-making. As shown in the results, we employed a stepwise logistic regression approach to develop this prediction model. We performed sensitivity analyses by constructing models with and without borderline predictors (neoadjuvant therapy and operative time). Although these variables did not substantially enhance discrimination or calibration, a modest improvement was observed (AUC: 0.660 vs. 0.700). Although the model’s overall performance was comparable to prior studies (29), its key advantage lies in integrating both preoperative and intraoperative determinants, emphasizing that complication prevention depends not only on patient frailty but also on surgical and anesthetic factors.
In the external validation cohort, the DCA showed a higher net benefit within the threshold probability range of 10% to 40%. This indicates that the model is clinically useful for guiding individualized decision-making in this risk interval. The calibration curve also demonstrated good agreement between predicted and observed probabilities, with a mean absolute error of 0.032, confirming the reliable estimation of individual risks. Although the validation sample was relatively small (n=68), the bootstrap-corrected results remained stable, further supporting the robustness of the model. Collectively, these results underscore the potential clinical applicability of our prediction model, which may assist clinicians in refining risk stratification and optimizing perioperative management strategies. Importantly, stratification by predicted risk further highlighted its utility: patients classified as high risk showed a statistically significant increase in the incidence of severe postoperative complications. Nevertheless, the limited cohort size raises the possibility of estimation error, underscoring the need for further validation in larger populations.
The introduction of the enhanced recovery after surgery (ERAS) concept has markedly improved postoperative outcomes in thoracic surgery, emphasizing the importance of a multidisciplinary and integrated perioperative care pathway (30). At our center, surgical management for esophageal carcinoma typically involves two simultaneous procedures: esophageal resection and alimentary tract reconstruction, most commonly using a gastric conduit. Despite the preservation of organ integrity, patients nonetheless face a profound metabolic stress response in the early postoperative period. Early and adequate postoperative nutritional support has been shown to enhance nutritional status and confer significant benefit (31). Research also shows that promptly advancing to a combined enteral and oral feeding regimen prevents postoperative malnutrition and does not appear to increase perioperative morbidity (32). Additionally, prior literature had demonstrated that immunonutrition could bolster host immunity, showed optimal potential to reduce the incidence of postoperative complications (33). This underscores that nutritional support has become an integral component of the ERAS protocol, and a more proactive management strategy for esophageal patients was described that comprehensive nutrition supported from neoadjuvant therapy before surgery to adjuvant therapy after surgery (34,35), which provided a perspective for proactive management of patients with high predicted risks.
Some other limitations of this study should be acknowledged. First, the retrospective and single-center design of our analysis may restrict the external validity of our findings, and validation in larger, multicenter or prospective cohorts is needed to confirm the generalizability of the model. Moreover, how to effectively translate these predictive outcomes into clinical benefits remains an area of further exploration. Second, although the MSK-FI is a convenient and validated tool, it assigns equal weight to all included variables, which may limit its ability to capture disease-specific physiological deficits (36). Future frailty indices should explore the use of weighted or disease-tailored scoring systems to enhance predictive precision. Third, as this study focused solely on short-term postoperative complications, the absence of long-term follow-up data prevented evaluation of whether preoperative optimization strategies translate into sustained clinical benefits (37).
Conclusions
In summary, while the MSK-FI remains a valuable tool for assessing frailty among esophageal cancer patients, its predictive value improves when combined with broader clinical and surgical variables. Our findings support a multidimensional approach to perioperative risk assessment and highlight the importance of individualized treatment planning. Further research is warranted to refine frailty assessment tools, validate our predictive model in larger cohorts, and develop strategies to reduce postoperative complication rates in this high-risk population.
Supplementary
The article’s supplementary files as
Acknowledgments
None.
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 was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Jinling Hospital (No. 2025DZKY-023-02) and individual consent for this retrospective analysis was waived.
Footnotes
Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-aw-2357/rc
Funding: This study was supported by Natural Science Foundation of Jiangsu Province Basic Research Program General Project, China (No. BK20181239).
Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-aw-2357/coif). The authors have no conflicts of interest to declare.
Data Sharing Statement
Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-aw-2357/dss
References
- 1.Filho AM, Laversanne M, Ferlay J, et al. The GLOBOCAN 2022 cancer estimates: Data sources, methods, and a snapshot of the cancer burden worldwide. Int J Cancer 2025;156:1336-46. 10.1002/ijc.35278 [DOI] [PubMed] [Google Scholar]
- 2.Ajani JA, D'Amico TA, Bentrem DJ, et al. Esophageal and Esophagogastric Junction Cancers, Version 2.2023, NCCN Clinical Practice Guidelines in Oncology. J Natl Compr Canc Netw 2023;21:393-422. 10.6004/jnccn.2023.0019 [DOI] [PubMed] [Google Scholar]
- 3.Cao J, Xu H, Li W, et al. Nutritional assessment and risk factors associated to malnutrition in patients with esophageal cancer. Curr Probl Cancer 2021;45:100638. 10.1016/j.currproblcancer.2020.100638 [DOI] [PubMed] [Google Scholar]
- 4.Mazidimoradi A, Banakar N, Khani Y, et al. Current status and temporal trend in incidence, death, and burden of esophageal cancer from 1990-2019. Thorac Cancer 2023;14:2408-58. 10.1111/1759-7714.15028 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Chen S, Zhang S, Cai S, et al. Impact of frailty on outcomes following coronary artery bypass grafting: a systematic review and meta-analysis. BMC Surg 2024;24:419. 10.1186/s12893-024-02728-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Hoogendijk EO, Afilalo J, Ensrud KE, et al. Frailty: implications for clinical practice and public health. Lancet 2019;394:1365-75. 10.1016/S0140-6736(19)31786-6 [DOI] [PubMed] [Google Scholar]
- 7.Shinall MC, Jr, Arya S, Youk A, et al. Association of Preoperative Patient Frailty and Operative Stress With Postoperative Mortality. JAMA Surg 2020;155:e194620. 10.1001/jamasurg.2019.4620 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Tjeertes EKM, van Fessem JMK, Mattace-Raso FUS, et al. Influence of Frailty on Outcome in Older Patients Undergoing Non-Cardiac Surgery - A Systematic Review and Meta-Analysis. Aging Dis 2020;11:1276-90. 10.14336/AD.2019.1024 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Bai YX, Wang ZH, Lv Y, et al. Association between frailty and acute kidney injury after cardiac surgery: unraveling the moderation effect of body fat through an international, retrospective, multicohort study. Int J Surg 2025;111:761-70. 10.1097/JS9.0000000000001861 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Pilotto A, Custodero C, Maggi S, et al. A multidimensional approach to frailty in older people. Ageing Res Rev 2020;60:101047. 10.1016/j.arr.2020.101047 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Howlett SE, Rutenberg AD, Rockwood K. The degree of frailty as a translational measure of health in aging. Nat Aging 2021;1:651-65. 10.1038/s43587-021-00099-3 [DOI] [PubMed] [Google Scholar]
- 12.Shahrokni A, Tin A, Alexander K, et al. Development and Evaluation of a New Frailty Index for Older Surgical Patients With Cancer. JAMA Netw Open 2019;2:e193545. 10.1001/jamanetworkopen.2019.3545 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Rice TW, Patil DT, Blackstone EH. 8th edition AJCC/UICC staging of cancers of the esophagus and esophagogastric junction: application to clinical practice. Ann Cardiothorac Surg 2017;6:119-30. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Clavien PA, Barkun J, de Oliveira ML, et al. The Clavien-Dindo classification of surgical complications: five-year experience. Ann Surg 2009;250:187-96. 10.1097/SLA.0b013e3181b13ca2 [DOI] [PubMed] [Google Scholar]
- 15.Fabbi M, Hagens ERC, van Berge Henegouwen MI, et al. Anastomotic leakage after esophagectomy for esophageal cancer: definitions, diagnostics, and treatment. Dis Esophagus 2021;34:doaa039. 10.1093/dote/doaa039 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.van Hootegem SJM, van der Linde M, Schneider MA, et al. Impact of postoperative complications on clinical outcomes after gastrectomy for cancer: multicentre study. Br J Surg 2025;112:znaf043. 10.1093/bjs/znaf043 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Thommen R, Kazim SF, Rumalla K, et al. Preoperative frailty measured by risk analysis index predicts complications and poor discharge outcomes after Brain Tumor Resection in a large multi-center analysis. J Neurooncol 2022;160:285-97. 10.1007/s11060-022-04135-z [DOI] [PubMed] [Google Scholar]
- 18.Han B, Li Q, Chen X. Effects of the frailty phenotype on post-operative complications in older surgical patients: a systematic review and meta-analysis. BMC Geriatr 2019;19:141. 10.1186/s12877-019-1153-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Boerner T, Sewell M, Tin AL, et al. A Novel Frailty Index Can Predict the Short-Term Outcomes of Esophagectomy in Older Patients with Esophageal Cancer. Curr Oncol 2024;31:4685-94. 10.3390/curroncol31080349 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Loecker C, Schmaderer M, Zimmerman L. Frailty in Young and Middle-Aged Adults: An Integrative Review. J Frailty Aging 2021;10:327-33. 10.14283/jfa.2021.14 [DOI] [PubMed] [Google Scholar]
- 21.Yang H, Wang F, Hallemeier CL, et al. Oesophageal cancer. Lancet 2024;404:1991-2005. 10.1016/S0140-6736(24)02226-8 [DOI] [PubMed] [Google Scholar]
- 22.Sirviö VEJ, Räsänen JV, Helminen O, et al. Neoadjuvant Treatment and Postoperative Complications After Surgery for Esophageal Cancer: A Population-Based, Nationwide Study in Finland. Ann Surg Oncol 2025;32:7914-28. 10.1245/s10434-025-17945-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Du W, Qiao X, Yao J, et al. Effect of Neoadjuvant Immunochemotherapy on Postoperative Pulmonary Complications for Locally Advanced Esophageal Cancer: A Propensity Score Matching Cohort Study. Drug Des Devel Ther 2025;19:7637-51. 10.2147/DDDT.S537794 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Horne ZD, Wegner RE, Colonias A, et al. Drivers of 30- and 90-day Postoperative Death After Neoadjuvant Chemoradiation for Esophageal Cancer. Ann Thorac Surg 2020;109:921-6. 10.1016/j.athoracsur.2019.10.057 [DOI] [PubMed] [Google Scholar]
- 25.Yang Y, Zhang H, Yu B, et al. Risk factors and prediction of intensive care unit readmission after oesophagectomy for cancer‡. Eur J Cardiothorac Surg 2025;67:ezaf124. 10.1093/ejcts/ezaf124 [DOI] [PubMed] [Google Scholar]
- 26.Ke J, Xie Y, Liang J, et al. Surgical intervention after neoadjuvant therapy in esophageal cancer: a narrative review. J Thorac Dis 2023;15:2261-76. 10.21037/jtd-23-420 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.van Workum F, Verstegen MHP, Klarenbeek BR, et al. Intrathoracic vs Cervical Anastomosis After Totally or Hybrid Minimally Invasive Esophagectomy for Esophageal Cancer: A Randomized Clinical Trial. JAMA Surg 2021;156:601-10. 10.1001/jamasurg.2021.1555 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Unruh KR, Bastawrous AL, Kanneganti S, et al. The Impact of Prolonged Operative Time Associated With Minimally Invasive Colorectal Surgery: A Report From the Surgical Care Outcomes Assessment Program. Dis Colon Rectum 2024;67:302-12. 10.1097/DCR.0000000000002925 [DOI] [PubMed] [Google Scholar]
- 29.van Nieuw Amerongen MP, de Grooth HJ, Veerman GL, et al. Prediction of Morbidity and Mortality After Esophagectomy: A Systematic Review. Ann Surg Oncol 2024;31:3459-70. 10.1245/s10434-024-14997-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Powers BK, Ponder HL, Findley R, et al. Enhanced recovery after surgery (ERAS(®)) Society abdominal and thoracic surgery recommendations: A systematic review and comparison of guidelines for perioperative and pharmacotherapy core items. World J Surg 2024;48:509-23. 10.1002/wjs.12101 [DOI] [PubMed] [Google Scholar]
- 31.Carr RA, Harrington C, Stella C, et al. Early implementation of a perioperative nutrition support pathway for patients undergoing esophagectomy for esophageal cancer. Cancer Med 2022;11:592-601. 10.1002/cam4.4360 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Na KJ, Kang CH, Kim YR, et al. Comparison of Clinical Outcomes and Postoperative Nutritional Status Between Early and Late Oral Feeding After Esophagectomy: An Open Labeled Randomized Controlled Trial. Ann Surg 2025;281:388-94. 10.1097/SLA.0000000000006441 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Tian X, Jin YF, Liu XL, et al. Network meta-analysis of the optimal time of applying enteral immunonutrition in esophageal cancer patients receiving esophagectomy. Support Care Cancer 2022;30:7133-46. 10.1007/s00520-022-07058-7 [DOI] [PubMed] [Google Scholar]
- 34.Weimann A, Bezmarevic M, Braga M, et al. ESPEN guideline on clinical nutrition in surgery - Update 2025. Clin Nutr 2025;53:222-61. 10.1016/j.clnu.2025.08.029 [DOI] [PubMed] [Google Scholar]
- 35.Lovey J, Molnar A, Banky B. Long-term nutrition in patients candidate to neoadjuvant and adjuvant treatments. Eur J Surg Oncol 2024;50:106850. 10.1016/j.ejso.2023.02.007 [DOI] [PubMed] [Google Scholar]
- 36.Charlson ME, Carrozzino D, Guidi J, et al. Charlson Comorbidity Index: A Critical Review of Clinimetric Properties. Psychother Psychosom 2022;91:8-35. 10.1159/000521288 [DOI] [PubMed] [Google Scholar]
- 37.Navarrete-Reyes AP, Mateos-Soria AS, Sánchez-Hernández JJ, et al. Frailty and Cancer Prognosis. Curr Oncol Rep 2024;26:991-1020. 10.1007/s11912-024-01558-x [DOI] [PubMed] [Google Scholar]