Clinical and pathological characteristics of the patients
The characteristics of the 171 participants in this study are detailed in Table 1. Among the participants, 113 were male (66.08%), 58 were female (33.92%), and the average age was 63.27 years ± 13.49 years. In this study, the AUC for the ability of PA to predict severe postoperative complications of MBO was 0.888 (95% CI: 0.832 ~ 0.944). The optimal cut-off point determined by the Youden index was 89.00, with a corresponding Youden index of approximately 0.722. This cut-off point exhibited a sensitivity of approximately 94.3% and a specificity of approximately 77.9% (Supplementary Fig. 1). Using this cut-off point, all patients were divided into a superlow PA group (PA < 89.0 µg/L, n = 63) and a low PA group (PA ≥ 89.0 µg/L, n = 108). Diabetes was present in 154 patients (90.06%), and 122 patients had hypertension (71.35%). There were 21 patients with preoperative infection (12.19%), and there was no statistically significant difference between the two groups in terms of prealbumin levels (p = 0.274). The tumour was primarily located in the left colon (n = 76, 44.44%), presenting mainly as incomplete obstruction (n = 94, 59.87%), ulcerative type (n = 55, 72.37%), moderately differentiated (n = 74, 83.15%), or tubular adenocarcinoma (n = 76, 64.41%). A few patients experienced intestinal perforation (n = 15, 8.77%), but there was no statistically significant difference between the two groups. Fibrinogen (p = 0.012), albumin (ALB) (< 0.001), and total protein (p = 0.009) were significantly different between the two PA groups. In the PA groups, there were 26 emergency surgery patients (26.90%), with no significant difference between the two groups. The main types of surgery included radical surgery (n = 93, 54.39%) and palliative surgery (n = 78, 45.61%), with palliative surgery comprising palliative tumour resection (n = 26, 15.20%) and simple ostomy (n = 52, 30.41%). Compared with the superlow PA group, the low PA group underwent postoperative chemotherapy in more cases (43 vs. 36, p = 0.028) had longer postoperative hospital stays (16.19 ± 8.38 vs. 12.24 ± 8.7, P = 0.004), had longer total hospital stays (22.67 ± 9.28 vs. 18.94 ± 10.01, P = 0.017), and incurred greater hospitalization costs [72604.88 (46392.01, 98785.77) vs. 58462.5 (42733.65, 77809.62), p = 0.02].
Table 1 Correlations between the adjusted preoperative PA and clinicopathological features in MBO patients
PA is an independent influencing factor for early severe complications (CD grade ≥ 3) and infectious complications.
In this study, all 171 patients experienced postoperative complications of varying severity, with 95 patients (55.56%) presenting with infectious complications. Specifically, 64 patients (37.43%) had incisional infections, 50 patients (29.24%) had pulmonary infections, 20 patients (11.70%) had abdominal infections, and 10 patients (5.85%) developed sepsis (Supplementary Fig. 1). Complications in the low PA group were predominantly classified as CD1 (n = 96, 88.89%), while in the superlow PA group, CD3 complications were more prevalent (n = 26, 41.27%). Overall, there were significant differences between the two groups in terms of different CD classifications of the complications (p < 0.001) (Fig. 2 and Supplementary Table 1). Notably, the superlow PA group exhibited a greater rate of CD grade ≥ 3 complications (n = 32, p < 0.001), indicating that as the PA decreased, the severity of overall postoperative complications gradually increased (p < 0.001) (Fig. 3). The superlow PA group also had a greater incidence of infectious complications (n = 55,87.30% vs. n = 40,37.04%, p < 0.001), with statistically significant differences in various CD classifications between the two groups (p < 0.001), notably higher rates of CD grade ≥ 3 complications (n = 31,56.36% vs. n = 1,2.50%, p < 0.001). This suggests that as the PA decreases, the severity of postoperative infectious complications gradually increases (p < 0.001) (Fig. 3). In this study, both univariate and multivariate logistic regression analyses were conducted on early postoperative severe complications and infectious complications. According to the multivariate analysis of all patients, preoperative PA (OR = 0.95, p < 0.001) and preoperative colonic metal stent placement (OR = 10.03, p = 0.015) were significantly associated with early severe complications (CD grade ≥ 3). Preoperative PA (OR = 0.96, p < 0.001) was significantly associated with infectious complications. However, no significant correlations were found between other indicators and overall complications or infectious complications (Table 2).
Table 2 Logistic regression analysis of early postoperative complications
PA is an independent influencing factor for OS.
In this study, the median follow-up period was 1.58 years, with 19.08% of patients surviving beyond 3 years and a mortality rate of 61.84%. Cox regression analysis stratified by TNM stage revealed sex, PA level, surgical approach, and postoperative chemotherapy as independent influencing factors for OS. Specifically, sex had a hazard ratio (HR) of 0.48 [0.29, 0.8], P = 0.005; the PA HR was 0.99 [0.99, 1], P = 0.045; the surgical approach HR was 2.57 [1.38, 4.78], P = 0.003; and the postoperative chemotherapy HR was 0.28 [0.17, 0.46], P < 0.001 (Table 3). A multivariate Cox analysis-based line graph for 171 patients with MBO was established for survival prediction. Sex, PA, TNM stage, tumour history, and postoperative chemotherapy indicators were included in this nomogram (Fig. 4). According to Kaplan‒Meier survival curves, patients in the ultralow PA group had shorter survival than did those in the low PA group (P = 0.0016). Factors such as lymph node metastasis, TNM stage, emergency surgery, preoperative complications, infectious complications, severe early postoperative complications, surgical approach, neoadjuvant chemotherapy, and postoperative chemotherapy were significantly associated with survival time (p < 0.05) (Fig. 5).
Table 3 Univariate and multivariate stratified Cox analyses by TNM staging to evaluate the predictors of OS in patients with MBO
Compared to traditional nutritional indicators, PA exhibits the highest testing efficiency for early severe postoperative complications in patients with MBO.
In this study, ROC curves were generated to compare the ability of PA and other indicators to predict early postoperative complications (Fig. 6 and Table 4). The AUC values for BMI, ALB, FPR, FAR, and AFR were 0.519, 0.643, 0.871, 0.674, and 0.781, respectively. Notably, the AFR exhibited a greater AUC (0.888) for predicting early postoperative complications (CD grade ≥ 3) after MBO surgery, surpassing BMI, ALB, FPR, FAR, and the AFR.
Table 4 Diagnostic performance of BMI, ALB, PA, FPR, FAR and the AFR for early postoperative complications of MBO