A short-term postoperative prediction model for colorectal cancer using comprehensive geriatric assessment Authors: Di Yang, Mengyu Cao, Shaokang Yang, Yujia Chen, Qinglong Jin, Jiyan Leng Published: 2026 ABSTRACT BackgroundOlder colorectal cancer (CRC) patients face significant postoperative risks. This study aimed to develop a prediction model to quantify the short-term prognosis of these patients following laparoscopic surgery.MethodsThis prospective study enrolled patients aged ≥ 60 years undergoing elective laparoscopic radical resection (May 2021–April 2024). The primary outcome was 90-day Major Adverse Postoperative Events (MAPE). A comprehensive geriatric assessment (CGA) was performed to identify independent risk factors.ResultsMAPE occurred in 23.7% of the derivation cohort and 22.5% of the validation cohort. Multivariable logistic regression identified six independent preoperative predictors: weight loss (OR = 2.20, 95% CI 1.20–4.22), ADL scores (OR = 1.11, 95% CI 1.03–1.20), preoperative BUN (OR = 1.14, 95% CI 1.01–1.29), ASMI (OR = 0.68, 95% CI 0.47–0.98), hypoalbuminemia (OR = 1.33, 95% CI 1.02–1.89), and TNM stage (OR = 1.75, 95% CI 1.18–2.60). The prediction model demonstrated robust discrimination (AUC: derivation 0.802; validation 0.779) and excellent calibration. Decision curve analysis confirmed its clinical utility across a wide range of threshold probabilities.ConclusionsThis CGA-based tool effectively predicts short-term prognosis, enabling personalized risk stratification and identifying high-risk candidates to guide future perioperative management strategies.Graphical abstract CGA-based prediction model for short-term prognosis in older colorectal cancer patients Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12877-026-07880-2. Introduction Colorectal cancer (CRC) is one of the most common malignant tumors [1]. Laparoscopic radical resection effectively removes tumors and locoregional lymph nodes, reducing recurrence risk and improving survival. In advanced CRC, neoadjuvant chemotherapy and radiotherapy decrease tumor burden and recurrence rates [2–4]. Advances in surgical and anesthetic techniques have diminished age as a surgical contraindication, increasing older adult for intervention [5]. However, the “young-old” (60–70 years) and “old-old” (> 80 years) display significant heterogeneity in surgical resilience. Organ dysfunction, cognitive impairment, sarcopenia, frailty, and comorbidities —rather than age itself—predispose older patients to postoperative complications, readmission, and mortality [6, 7]. Hence, moving beyond mere age to a more nuanced preoperative risk stratification is imperative for personalized geriatric oncology. Comprehensive Geriatric Assessment (CGA) has emerged as the multidimensional gold standard for evaluating physiological, psychological, and social domains in older adults [8, 9]. Unlike single-marker assessments, CGA provides a holistic view of a patient’s “biological age.” While international guidelines strongly endorse CGA for geriatric cancer care [10–12], its clinical application is often hindered by its complexity. There is a pressing need for a simplified, validated prediction model that translates complex CGA data into a quantifiable risk score, specifically for the short-term recovery period following CRC surgery. Most existing models focus on isolated outcomes, such as intra-hospital complications, which may underestimate the true burden of surgical recovery in the elderly. To address this, we defined a more rigorous and clinically relevant primary outcome: Major Adverse Postoperative Events (MAPE). This composite endpoint integrates 90-day mortality, unplanned readmission, and major complications. By utilizing a 90-day window, MAPE captures the complete trajectory of postoperative “failed recovery,” including out-of-hospital deaths and readmissions—a period where elderly patients are most vulnerable. Methods Study design and participants This prospective study enrolled patients aged ≥ 60 with primary colorectal adenocarcinoma at the First Hospital of Jilin University (May 2021–April 2024). All participants underwent elective radical laparoscopic resection. Inclusion required histological confirmation and preoperative elective status. We excluded emergency surgeries, palliative procedures, conversion to open surgery, and patients with prior malignancies or severe organ failure. The age cutoff of ≥ 60 years was selected as it captures the population where age-related decline in physiological reserve becomes clinically significant in surgical oncology [13]. Participant flow is summarized in Fig. 1. Fig. 1The experimental process The experimental process To ensure model stability, we utilized temporal validation: the derivation cohort (May 2021–July 2023, n = 417) and the validation cohort (August 2023–April 2024, n = 142). This design evaluates the model’s predictive utility in a subsequent, independent patient population [14]. Surgical and pathological standards All surgical procedures were performed laparoscopically by a single specialized team following standardized Total Mesorectal Excision or Complete Mesocolic Excision. Preoperative TNM staging was determined by contrast-enhanced CT, then confirmed by postoperative pathology. Standardized neoadjuvant chemotherapy protocols were documented. For patients with temporary diverting ileostomy, stoma management followed standard nursing protocols. Primary outcome: major adverse postoperative events (MAPE) The primary endpoint was 90-day MAPE, a composite of: Major complications: Clavien-Dindo classification Grade ≥ III during the index hospitalization [15].Unplanned readmission: Excluding routine admissions for scheduled chemotherapy or planned stoma reversal.90-day mortality: All-cause death including out-of-hospital events. Major complications: Clavien-Dindo classification Grade ≥ III during the index hospitalization [15]. Major complications: Clavien-Dindo classification Grade ≥ III during the index hospitalization [15]. Unplanned readmission: Excluding routine admissions for scheduled chemotherapy or planned stoma reversal. Unplanned readmission: Excluding routine admissions for scheduled chemotherapy or planned stoma reversal. 90-day mortality: All-cause death including out-of-hospital events. 90-day mortality: All-cause death including out-of-hospital events. For the purpose of identifying risk factors in the derivation cohort, patients were categorized into two groups based on the occurrence of the primary endpoint: the MAPE group (patients who experienced at least one event) and the Non-MAPE group (patients free from any event). Comprehensive geriatric assessment A structured, preoperative CGA was performed for all eligible patients by a dedicated team of experienced geriatricians. To ensure the baseline nature of the assessment, all evaluations were conducted following hospital admission but prior to the elective surgical intervention. The CGA encompassed the following five dimensions:Demographic and General Characteristics: Patient demographics, including age, sex, and lifestyle factors, were collected through standardized, structured interviews.Anthropometric Measurements: Height and weight were measured according to standard protocols using calibrated instruments. Body mass index (BMI) was calculated as weight divided by height squared (kg/m²). Weight loss was defined as an unintentional reduction of ≥ 3 kg within the three months preceding admission.Comorbidity Burden: The cumulative burden of chronic diseases was quantified using the Age-adjusted Charlson Comorbidity Index (aCCI) [16]. This validated tool incorporates both the severity of co-existing conditions and the physiological impact of age to provide a comprehensive score of patient complexity (Supplementary Material 1).Sarcopenia Assessment: Appendicular skeletal muscle mass (ASM) was measured using bioelectrical impedance analysis (Inbody S10; Biospace, Seoul, Korea) [17]. The appendicular skeletal muscle mass index (ASMI) was calculated as ASM/height² (kg/m²). Handgrip strength (HS) was measured using a dynamometer (Jamar®Plus). Gait speed (GS) was assessed over a 6-meter walk. Sarcopenia was diagnosed when participants presented with both low muscle mass (ASMI < 7.0 kg/m² for men, < 5.7 kg/m² for women) and either low muscle strength (handgrip strength < 28 kg for men, < 18 kg for women) or low physical performance (gait speed < 1.0 m/s) [18].Functional Status: Functional independence and activities of daily living were evaluated using the Activities of Daily Living (ADL) scale, encompassing both basic and instrumental domains. Higher scores indicated greater degrees of functional impairment and dependency (Supplementary Material 2). Demographic and General Characteristics: Patient demographics, including age, sex, and lifestyle factors, were collected through standardized, structured interviews. Demographic and General Characteristics: Patient demographics, including age, sex, and lifestyle factors, were collected through standardized, structured interviews. Anthropometric Measurements: Height and weight were measured according to standard protocols using calibrated instruments. Body mass index (BMI) was calculated as weight divided by height squared (kg/m²). Weight loss was defined as an unintentional reduction of ≥ 3 kg within the three months preceding admission. Anthropometric Measurements: Height and weight were measured according to standard protocols using calibrated instruments. Body mass index (BMI) was calculated as weight divided by height squared (kg/m²). Weight loss was defined as an unintentional reduction of ≥ 3 kg within the three months preceding admission. Comorbidity Burden: The cumulative burden of chronic diseases was quantified using the Age-adjusted Charlson Comorbidity Index (aCCI) [16]. This validated tool incorporates both the severity of co-existing conditions and the physiological impact of age to provide a comprehensive score of patient complexity (Supplementary Material 1). Comorbidity Burden: The cumulative burden of chronic diseases was quantified using the Age-adjusted Charlson Comorbidity Index (aCCI) [16]. This validated tool incorporates both the severity of co-existing conditions and the physiological impact of age to provide a comprehensive score of patient complexity (Supplementary Material 1). Sarcopenia Assessment: Appendicular skeletal muscle mass (ASM) was measured using bioelectrical impedance analysis (Inbody S10; Biospace, Seoul, Korea) [17]. The appendicular skeletal muscle mass index (ASMI) was calculated as ASM/height² (kg/m²). Handgrip strength (HS) was measured using a dynamometer (Jamar®Plus). Gait speed (GS) was assessed over a 6-meter walk. Sarcopenia was diagnosed when participants presented with both low muscle mass (ASMI < 7.0 kg/m² for men, < 5.7 kg/m² for women) and either low muscle strength (handgrip strength < 28 kg for men, < 18 kg for women) or low physical performance (gait speed < 1.0 m/s) [18]. Sarcopenia Assessment: Appendicular skeletal muscle mass (ASM) was measured using bioelectrical impedance analysis (Inbody S10; Biospace, Seoul, Korea) [17]. The appendicular skeletal muscle mass index (ASMI) was calculated as ASM/height² (kg/m²). Handgrip strength (HS) was measured using a dynamometer (Jamar®Plus). Gait speed (GS) was assessed over a 6-meter walk. Sarcopenia was diagnosed when participants presented with both low muscle mass (ASMI < 7.0 kg/m² for men, < 5.7 kg/m² for women) and either low muscle strength (handgrip strength < 28 kg for men, < 18 kg for women) or low physical performance (gait speed < 1.0 m/s) [18]. Functional Status: Functional independence and activities of daily living were evaluated using the Activities of Daily Living (ADL) scale, encompassing both basic and instrumental domains. Higher scores indicated greater degrees of functional impairment and dependency (Supplementary Material 2). Functional Status: Functional independence and activities of daily living were evaluated using the Activities of Daily Living (ADL) scale, encompassing both basic and instrumental domains. Higher scores indicated greater degrees of functional impairment and dependency (Supplementary Material 2). The study was approved by the Institutional Review Board of the First Hospital of Jilin University (Approval No: 21K103-002) and conducted in accordance with the Declaration of Helsinki. Statistical analysis Univariate analysis identified candidate predictors (P < 0.05), which were subsequently entered into multivariable logistic regression to develop the model. Discrimination was evaluated by area under the curve (AUC); calibration by Hosmer-Lemeshow tests and calibration plots. Clinical utility was assessed via Decision Curve Analysis (DCA). The stability of the multivariable model was ensured by an events per variable (EPV) ratio of 16.5 (99 events and 6 predictors), exceeding the recommended minimum of 10. Analyses were conducted using SPSS 26.0 and R 4.5.0. Results Participant characteristics and cohort distribution A total of 859 patients were screened, and 559 met the eligibility criteria. Based on the date of surgery, patients were assigned to either the derivation cohort (n = 417) or the validation cohort (n = 142). As detailed in Supplementary Material 3, the two cohorts were well-balanced with no significant differences in baseline characteristics. The median age was 67.0 years in the derivation cohort and 67.5 years in the validation cohort (P = 0.592). Both groups showed similar distributions in gender (P = 0.580), BMI (P = 0.856), and preoperative CGA domains, including the prevalence of sarcopenia (P = 0.765) and functional ADL scores (P = 0.209). Surgical parameters, such as tumor location (P = 0.300) and TNM staging (P = 0.759), were also comparable. The incidence of 90-day MAPE was 23.7% (n = 99) in the derivation cohort and 22.5% (n = 32) in the validation cohort (P = 0.421). Composition and intersection of 90-day MAPE In the derivation cohort, 99 unique patients (23.7%) experienced at least one event defining MAPE. The composite endpoint consisted of major complications (n = 60), unplanned readmissions (n = 43), and 90-day mortality (n = 9). Detailed intersection analysis (Fig. 2) revealed that 12 patients experienced overlapping adverse events: five patients had both major complications and readmissions. Five patients died following major complications. One patient died following an unplanned readmission. One patient experienced the full spectrum of all three adverse events. Fig. 2Composition and intersection of 90-day major adverse postoperative events (MAPE) Composition and intersection of 90-day major adverse postoperative events (MAPE) Note: The diagram illustrates the number of unique patients experiencing major complications, unplanned readmissions, and 90-day mortality, as well as the overlaps among these adverse events in the derivation cohort (n = 99). Univariate analysis and predictor identification Comparisons between the MAPE group (n = 99) and the non-MAPE group (n = 318) in the derivation cohort were presented in Table 1. The MAPE group was significantly older (P = 0.017) and had a higher proportion of patients ≥ 80 years (P = 0.028). They exhibited lower BMI (P = 0.014), longer hospital stays (P = 0.007), and higher hospitalization costs (P = 0.006). Preoperative laboratory results showed significantly lower hemoglobin (P = 0.044) and albumin (P = 0.001), alongside elevated BUN (P = 0.003) and CA19-9 (P = 0.028). Regarding CGA domains, the MAPE group presented with higher aCCI scores (P = 0.011), more frequent weight loss (P = 0.001), higher sarcopenia prevalence (P = 0.001), slower gait speed (P = 0.033), lower ASMI (P = 0.001), and higher ADL scores (P = 0.001). Furthermore, these patients were characterized by larger tumor diameters (P = 0.001) and more advanced TNM stages (P = 0.001). Postoperatively, significant differences persisted in albumin (P = 0.028) and BUN (P = 0.008). Table 1Comparison of characteristics between the MAPE and non-MAPE group in the derivation cohortCharacteristicsNon-MAPE Group (N = 318)MAPE Group(N = 99)P-valuePreoperative General dataGender (Female), n (%)129 (40.6)42 (42.4)0.743Age, Median (IQR)66 (63–72)69 (66–74)0.017*Age Group0.028* 60–69, n (%)208 (65.4)56 (56.6) 70–79, n (%)91 (28.6)30 (20.3) ≥ 80, n (%)19 (6.0)13 (13.1)Smoke, n (%)32 (10.1)5 (5.1)0.126Drink, n (%)29 (9.1)6 (6.1)0.338Length of Stay, day10.64 ± 4.9312.31 ± 6.590.007**Cost of Hospitalization, ¥53,310 ± 10,06758,013 ± 23,9900.006**Surgical dataTumor Diameter (cm)4.57 ± 1.685.38 ± 2.110.001Bleeding, n (%)132 (41.5)36 (36.4)0.362Transfusion, n (%)19 (6.0)9 (9.1)0.279TNM Stage0.001** 1, n (%)27 (8.5)16 (16.2) 2, n (%)161 (50.6)29 (26.3) 3, n (%)116 (36.5)46 (46.5) 4, n (%)14 (4.4)8 (8.1)Pathological Classification0.077 Infiltrative type, n (%)12(3.8)5(5.1) Ulcerative type, n (%)191 (60.1)62(62.6) Elevated type, n (%)115 (36.1)32(32.3)Location0.117 Colon, n (%)185 (58.2)66 (66.7) Rectum, n (%)115 (36.2)25 (26.3) Colorectal junction, n (%)18 (5.7)8 (8.1)Postoperative dataHB (g/L)112.07 ± 20.87110.33 ± 21.240.472AST (U/L)18.91 ± 11.2718.62 ± 13.650.833ALT (U/L)14.36 ± 11.2513.73 ± 12.330.636Albumin (g/L)31.11 ± 3.5032.21 ± 4.350.028*BUN (mmol/L)4.83 ± 2.055.47 ± 2.120.008**Creatinine (µmol/L)58.70 ± 20.2259.32 ± 22.220.795Laboratory ParametersHB (g/L)120.33 ± 25.21116.51 ± 24.290.044*AST (U/L)20.77 ± 9.8120.03 ± 12.930.540ALT (U/L)16.11 ± 10.716.87 ± 6.890.051ALP (U/L)81.24 ± 28.0180.39 ± 18.870.780ChE (U/L)6414.9 ± 1558.206055.05 ± 1715.470.051Albumin (g/L)37.35 ± 4.5634.02 ± 4.640.001**Hypoalbuminemia0.001** Normal (≥ 35 g/L), n (%)248 (78)62 (62.6) MHA (30–35 g/L), n (%)63 (19.8)22 (22.2) MSHA (< 30 g/L), n (%)7 (2.2)15(15.1)BUN (mmol/L)5.20 ± 1.615.79 ± 2.030.003**Creatinine (µmol/L)66.86 ± 19.9069.58 ± 24.860.265TG (mmol/L)1.43 ± 0.911.27 ± 0.730.128HDL-c (mmol/L)0.98 ± 0.330.94 ± 0.290.267LDL-c (mmol/L)2.81 ± 1.112.60 ± 0.900.076CEA6.29 ± 13.1014.99 ± 62.750.174AFP10.64 ± 4.9212.31 ± 6.590.061CA12524.88 ± 64.1125.54 ± 52.800.926CA7247.95 ± 30.337.47 ± 11.010.426CA199 (U/ml)15.21 ± 26.8115.11 ± 17.350.028*Comprehensive Geriatric AssessmentComorbidity BurdenaCCI, Median (IQR)6 (4–8)8 (5–8)0.011*Anthropometric & Nutritional StatusBMI (kg/m2)23.15 ± 3.4722.17 ± 3.330.014*Weight Loss, n (%)90 (28.3)55 (55.6)0.001**Sarcopenia AssessmentSarcopenia, n (%)75 (23.5)45 (45.5)0.001**HS (kg)26.75 ± 6.4625.65 ± 6.250.116GS (m/s)1.20 ± 0.231.14 ± 0.250.033*ASMI (kg/m2)7.18 ± 0.946.56 ± 1.030.001**Functional StatusADL score19.19 ± 3.2422.19 ± 5.390.001**BMI body mass index, aCCI age-adjusted Charlson comorbidity index, ADL activities of daily living, AFP alpha-fetoprotein, ASMI appendicular skeletal muscle mass index, ALT alanine aminotransferase, AST aspartate aminotransferase, BMI body mass index, BUN blood urea nitrogen, CA carbohydrate antigen, CEA carcinoembryonic antigen, ChE cholinesterase, CRC colorectal cancer, GS gait speed, HB hemoglobin, HDL-c high-density lipoprotein cholesterol, HS handgrip strength, LDL-c low-density lipoprotein cholesterol, MAPE major adverse postoperative events, TG triglyceride, TNM tumor-node-metastasis*: P < 0.05, **: P < 0.01 Table 1Comparison of characteristics between the MAPE and non-MAPE group in the derivation cohortCharacteristicsNon-MAPE Group (N = 318)MAPE Group(N = 99)P-valuePreoperative General dataGender (Female), n (%)129 (40.6)42 (42.4)0.743Age, Median (IQR)66 (63–72)69 (66–74)0.017*Age Group0.028* 60–69, n (%)208 (65.4)56 (56.6) 70–79, n (%)91 (28.6)30 (20.3) ≥ 80, n (%)19 (6.0)13 (13.1)Smoke, n (%)32 (10.1)5 (5.1)0.126Drink, n (%)29 (9.1)6 (6.1)0.338Length of Stay, day10.64 ± 4.9312.31 ± 6.590.007**Cost of Hospitalization, ¥53,310 ± 10,06758,013 ± 23,9900.006**Surgical dataTumor Diameter (cm)4.57 ± 1.685.38 ± 2.110.001Bleeding, n (%)132 (41.5)36 (36.4)0.362Transfusion, n (%)19 (6.0)9 (9.1)0.279TNM Stage0.001** 1, n (%)27 (8.5)16 (16.2) 2, n (%)161 (50.6)29 (26.3) 3, n (%)116 (36.5)46 (46.5) 4, n (%)14 (4.4)8 (8.1)Pathological Classification0.077 Infiltrative type, n (%)12(3.8)5(5.1) Ulcerative type, n (%)191 (60.1)62(62.6) Elevated type, n (%)115 (36.1)32(32.3)Location0.117 Colon, n (%)185 (58.2)66 (66.7) Rectum, n (%)115 (36.2)25 (26.3) Colorectal junction, n (%)18 (5.7)8 (8.1)Postoperative dataHB (g/L)112.07 ± 20.87110.33 ± 21.240.472AST (U/L)18.91 ± 11.2718.62 ± 13.650.833ALT (U/L)14.36 ± 11.2513.73 ± 12.330.636Albumin (g/L)31.11 ± 3.5032.21 ± 4.350.028*BUN (mmol/L)4.83 ± 2.055.47 ± 2.120.008**Creatinine (µmol/L)58.70 ± 20.2259.32 ± 22.220.795Laboratory ParametersHB (g/L)120.33 ± 25.21116.51 ± 24.290.044*AST (U/L)20.77 ± 9.8120.03 ± 12.930.540ALT (U/L)16.11 ± 10.716.87 ± 6.890.051ALP (U/L)81.24 ± 28.0180.39 ± 18.870.780ChE (U/L)6414.9 ± 1558.206055.05 ± 1715.470.051Albumin (g/L)37.35 ± 4.5634.02 ± 4.640.001**Hypoalbuminemia0.001** Normal (≥ 35 g/L), n (%)248 (78)62 (62.6) MHA (30–35 g/L), n (%)63 (19.8)22 (22.2) MSHA (< 30 g/L), n (%)7 (2.2)15(15.1)BUN (mmol/L)5.20 ± 1.615.79 ± 2.030.003**Creatinine (µmol/L)66.86 ± 19.9069.58 ± 24.860.265TG (mmol/L)1.43 ± 0.911.27 ± 0.730.128HDL-c (mmol/L)0.98 ± 0.330.94 ± 0.290.267LDL-c (mmol/L)2.81 ± 1.112.60 ± 0.900.076CEA6.29 ± 13.1014.99 ± 62.750.174AFP10.64 ± 4.9212.31 ± 6.590.061CA12524.88 ± 64.1125.54 ± 52.800.926CA7247.95 ± 30.337.47 ± 11.010.426CA199 (U/ml)15.21 ± 26.8115.11 ± 17.350.028*Comprehensive Geriatric AssessmentComorbidity BurdenaCCI, Median (IQR)6 (4–8)8 (5–8)0.011*Anthropometric & Nutritional StatusBMI (kg/m2)23.15 ± 3.4722.17 ± 3.330.014*Weight Loss, n (%)90 (28.3)55 (55.6)0.001**Sarcopenia AssessmentSarcopenia, n (%)75 (23.5)45 (45.5)0.001**HS (kg)26.75 ± 6.4625.65 ± 6.250.116GS (m/s)1.20 ± 0.231.14 ± 0.250.033*ASMI (kg/m2)7.18 ± 0.946.56 ± 1.030.001**Functional StatusADL score19.19 ± 3.2422.19 ± 5.390.001**BMI body mass index, aCCI age-adjusted Charlson comorbidity index, ADL activities of daily living, AFP alpha-fetoprotein, ASMI appendicular skeletal muscle mass index, ALT alanine aminotransferase, AST aspartate aminotransferase, BMI body mass index, BUN blood urea nitrogen, CA carbohydrate antigen, CEA carcinoembryonic antigen, ChE cholinesterase, CRC colorectal cancer, GS gait speed, HB hemoglobin, HDL-c high-density lipoprotein cholesterol, HS handgrip strength, LDL-c low-density lipoprotein cholesterol, MAPE major adverse postoperative events, TG triglyceride, TNM tumor-node-metastasis*: P < 0.05, **: P < 0.01 Comparison of characteristics between the MAPE and non-MAPE group in the derivation cohort BMI body mass index, aCCI age-adjusted Charlson comorbidity index, ADL activities of daily living, AFP alpha-fetoprotein, ASMI appendicular skeletal muscle mass index, ALT alanine aminotransferase, AST aspartate aminotransferase, BMI body mass index, BUN blood urea nitrogen, CA carbohydrate antigen, CEA carcinoembryonic antigen, ChE cholinesterase, CRC colorectal cancer, GS gait speed, HB hemoglobin, HDL-c high-density lipoprotein cholesterol, HS handgrip strength, LDL-c low-density lipoprotein cholesterol, MAPE major adverse postoperative events, TG triglyceride, TNM tumor-node-metastasis *: P < 0.05, **: P < 0.01 Multivariable Model and Validation Multivariable logistic regression confirmed six independent preoperative predictors of 90-day MAPE (Table 2): Weight loss (OR = 2.20, 95% CI 1.20–4.22, P = 0.012), ADL scores (OR = 1.11, 95% CI 1.03–1.20, P = 0.005), preoperative BUN (OR = 1.14, 95% CI 1.01–1.29, P = 0.041), ASMI (OR = 0.68, 95% CI 0.47–0.98, P = 0.037), hypoalbuminemia (OR = 1.33, 95% CI 1.02–1.89, P = 0.001) and TNM stage (OR = 1.75, 95% CI 1.18–2.60, P = 0.005). Table 2Logistic regression analysis in the derivation cohortNon-MAPE Group (N = 318)MAPE Group(N = 99)OROR 95%CIP-valueAge Group--0.900.59–1.460.673Weight Loss, n (%)95 (29.9)49 (49.5)2.201.20–4.220.012*aCCI, Median (IQR)6 (4–8)8 (5–8)1.070.94–1.210.308GS (m/s)1.20 ± 0.231.15 ± 0.250.980.78–1.120.146ASMI (kg/m2)7.18 ± 0.946.56 ± 1.030.680.47–0.980.037*ADL score19.19 ± 3.2422.19 ± 5.391.111.03–1.200.005**HB (g/L)120.33 ± 25.21116.51 ± 24.291.050.92–1.190.474Hypoalbuminemia--1.331.02–1.890.001**BUN (mmol/L)5.20 ± 1.615.79 ± 2.031.141.01–1.290.041*CA199 (U/ml)15.21 ± 26.8115.11 ± 17.350.9770.96-1.020.241Tumor Diameter (cm)4.57 ± 1.685.38 ± 2.111.070.91–1.250.405TNM Stage--1.751.18–2.600.005**OR odds ratio, CI confidence interval*: P<0.05, **: P<0.01 Table 2Logistic regression analysis in the derivation cohortNon-MAPE Group (N = 318)MAPE Group(N = 99)OROR 95%CIP-valueAge Group--0.900.59–1.460.673Weight Loss, n (%)95 (29.9)49 (49.5)2.201.20–4.220.012*aCCI, Median (IQR)6 (4–8)8 (5–8)1.070.94–1.210.308GS (m/s)1.20 ± 0.231.15 ± 0.250.980.78–1.120.146ASMI (kg/m2)7.18 ± 0.946.56 ± 1.030.680.47–0.980.037*ADL score19.19 ± 3.2422.19 ± 5.391.111.03–1.200.005**HB (g/L)120.33 ± 25.21116.51 ± 24.291.050.92–1.190.474Hypoalbuminemia--1.331.02–1.890.001**BUN (mmol/L)5.20 ± 1.615.79 ± 2.031.141.01–1.290.041*CA199 (U/ml)15.21 ± 26.8115.11 ± 17.350.9770.96-1.020.241Tumor Diameter (cm)4.57 ± 1.685.38 ± 2.111.070.91–1.250.405TNM Stage--1.751.18–2.600.005**OR odds ratio, CI confidence interval*: P<0.05, **: P<0.01 Logistic regression analysis in the derivation cohort OR odds ratio, CI confidence interval *: P<0.05, **: P<0.01 Six predictors—weight loss, ADL scores, BUN, ASMI, hypoalbuminemia and TNM stage—were integrated into a nomogram predicting short-term postoperative outcomes in older CRC patients (Fig. 3). Fig. 3Postoperative short-term prediction model for CRC Postoperative short-term prediction model for CRC The model demonstrated robust discrimination, with an AUC of 0.802 (95% CI 0.766–0.843) in the derivation cohort and 0.779 (95% CI 0.748–0.826) in the validation cohort (Fig. 4a and b). Calibration plots confirmed high predictive accuracy in both cohorts (Fig. 4c and d). DCA further supported the model’s clinical utility, demonstrating superior net benefit across a wide range of threshold probabilities (Fig. 4e and f). Fig. 4The validation of prediction model The validation of prediction model Discussion This study developed and validated a CGA-based nomogram to predict 90-day Major Adverse Postoperative Events (MAPE) in older adults undergoing laparoscopic colorectal cancer (CRC) surgery. The final model, incorporating weight loss, ADL scores, ASMI, preoperative albumin, BUN, and TNM stage, demonstrated robust discrimination and clinical utility. The observed MAPE rate of 23.7% in our cohort underscores the necessity of such a tool. This rate reflects the composite nature of our endpoint, which captures the full spectrum of recovery failure—including in-hospital complications, post-discharge readmissions, and mortality—a period where older adults are most vulnerable. Our findings emphasize that in this high-risk population, short-term surgical recovery is determined by a synergy between tumor biology and the patient’s underlying physiological reserve. Nutritional and body composition metrics: weight loss and ASMI Preoperative weight loss and low ASMI were identified as independent predictors of MAPE. Weight loss in older patients with CRC often signals a state of cancer-associated cachexia, characterized by systemic inflammation and metabolic derangement, which significantly erodes surgical resilience [19, 20]. This aligns with evidence linking preoperative BMI decline to reduced survival in CRC [21] and associating treatment-related weight loss with diminished overall survival in cancer patients [22]. While BMI is a conventional metric, it often lacks the sensitivity to detect “sarcopenic obesity,” where muscle depletion is masked by adipose tissue [23]. In contrast, ASMI provides a direct quantification of skeletal muscle mass. Our findings confirm that muscle wasting is a potent driver of postoperative “failed recovery,” supporting the inclusion of body composition analysis in preoperative risk assessments to identify vulnerable individuals who might otherwise appear fit [24]. Functional independence: the impact of ADL scores Impaired functional status, as reflected by higher ADL scores, was another critical driver of 90-day adverse events. ADL scores capture the cumulative impact of physiological decline on a patient’s daily independence [25]. For “young-old” patients (60–70 years) who represented the majority of our cohort, functional assessment was particularly valuable in uncovering “hidden frailty.” As advocated in recent geriatric oncology guidelines, integrating functional dependency into prognostic models significantly enhances the precision of risk quantification compared to using chronological age alone [12]. In our multivariable model, chronological age was not an independent predictor (P = 0.673), whereas CGA-based markers such as ASMI and ADL scores remained significant. This suggests that for older CRC patients, ‘biological age’ and physiological reserve are more critical determinants of short-term prognosis than chronological age alone, supporting the use of CGA for nuanced risk stratification even in ‘young-old’ populations. Preoperative biomarkers and clinical utility: BUN and albumin A notable finding of this study is the independent predictive value of preoperative BUN and albumin in identifying high-risk patients. Specifically, BUN, but not creatinine, was independently associated with MAPE. This likely reflects metabolic disturbances related to a hypercatabolic state, occult dehydration, or diminished renal functional reserve. While elevated BUN is a known risk factor for mortality in contexts such as acute COPD exacerbations, our findings extend its utility to surgical oncology as a sensitive surrogate for physiological reserve. Unlike creatinine, which is often confounded by sarcopenia in older adults, BUN more accurately reflects a patient’s hydration status and metabolic stress. Consequently, elevated preoperative BUN—often in synergy with hypoalbuminemia—signals a precarious homeostatic balance and diminished resilience to surgical trauma, effectively identifying individuals who require intensified perioperative optimization. Preoperative hypoalbuminemia demonstrated significant prognostic value in our model. This aligns with a broad spectrum of oncology literature where nutritional and inflammatory markers are critical; for instance, grade ≥ 2 hypoproteinemia has been identified as a significant independent prognostic factor for worse disease-free survival in esophageal cancer [26]. Furthermore, reduced survival has been observed in cervical cancer patients with concurrent hypoproteinemia and anemia [27]. Preoperative hypoalbuminemia reflects both nutritional deficiency and systemic inflammation, further compromising the patient’s recovery potential. The synergy of CGA and TNM staging While TNM stage remains the gold standard for long-term oncologic prognosis, its role in predicting acute postoperative recovery is often secondary to physiological factors [21, 28]. In our multivariable model, advanced TNM stage remained an independent predictor, likely reflecting the increased technical complexity and physiological toll of advanced resections. However, the addition of CGA domains significantly improved the model’s predictive power [29]. This suggests that while tumor stage dictates the surgical challenge, the patient’s physiological reserve determines their ability to withstand that challenge. A holistic approach accounting for both “host” and “disease” factors is essential for personalized surgical care in older populations. Limitations This study has several limitations. First, as a single-center study focusing on laparoscopic surgery, external validation in diverse populations is needed. Second, while surgical techniques and the need for ostomies differed, tumor location was not a significant independent predictor of MAPE. Third, unmeasured social factors might influence 90-day readmission rates. Finally, the 90-day window may capture some events unrelated to the index procedure, though its relevance in assessing recovery remains high. Conclusion We developed and validated a practical prediction model combining nutritional, functional, and oncological markers. This CGA-based tool effectively identifies older CRC patients at high risk for 90-day adverse events, enabling personalized risk stratification and identifying high-risk candidates to guide future perioperative management strategies. Supplementary Information Supplementary Material 1. Supplementary Material 1. Supplementary Material 2. Supplementary Material 2. Supplementary Material 3. Supplementary Material 3.