The time and mode of early and late recurrence of gastric cancer

Among the 521 postoperative patients included in this study, there were 365 males and 156 females. The overall median recurrence time was 16 months (IQR, 9–27 months). The mean recurrence time for patients was 22.23 months (95% CI: 20.51–23.96 months). Postoperative recurrence mainly occurred within 2 years. The recurrence rates at 1-, 2-, 3-, 4-, and 5- years after surgery were 32.1% (167/521), 72.0% (375/521), 83.3% (434/521), 91.5% (477/521), and 94.2% (491/521), respectively (Fig. 1A). Among them, there were 375 patients with early recurrence; the median recurrence time was 12 months (IQR, 8–18 months). Additionally, there were 146 patients with late recurrence, and the median recurrence time was 40 months (IQR, 32–55 months).

Fig. 1

Recurrence regularity of gastric cancer patients. Recurrence cases (A) and recurrence modes (B) within 5 years in all patients with gastric cancer

In this study, abdominal metastasis (n = 168, 32.2%) was the most common mode of recurrence, followed by liver metastasis (n = 127, 24.4%). Patients with ovarian metastases (n = 31, 6.0%) had the least recurrence. Additionally, the remaining metastatic sites included retroperitoneal lymph node metastasis (n = 38, 7.3%), in situ recurrence (n = 51, 9.8%), and others (n = 106, 20.3%) (Fig. 1B).

Risk factors for early recurrence of gastric cancer

A total of 365 patients in the training cohort were classified into early and late recurrence groups, with the division occurring at the 2-year mark. Univariate and multivariate analyses were conducted to explore the risk factors that could influence early gastric cancer recurrence. Age, serosa infiltration, lymph node metastasis, mode of recurrence, and tumour markers CA19-9 and CA72-4 were found to be closely associated with early gastric cancer recurrence (Table 1). All covariables were included in a multivariate logistic regression analysis to account for the influence of these variables. Consequently, the multivariate logistic regression model revealed that age (OR = 1.789, P < 0.05), serosa infiltration (OR = 2.293, P < 0.05), lymph node metastasis (OR = 2.162, P < 0.05), liver metastasis (OR = 7.031, P < 0.05), abdominal metastasis (OR = 2.731, P < 0.05), retroperitoneal lymph node metastasis (OR = 5.696, P < 0.05), and the tumour marker CA19-9 (OR = 1.970, P < 0.05) were independent risk factors for early recurrence (Table 1).

Risk factors for survival of gastric cancer recurrence

The overall median postoperative survival for all 521 patients was 28 months (IQR, 17–50 months), and the average survival time was 37.46 months (95% CI: 34.92–40.00 months). The survival rates at 1-, 3-, and 5- years after surgery were 87.5% (456/521), 38.6% (201/521), and 17.7% (92/521), respectively.

Furthermore, the survival analysis of all patients included in this study revealed that age > 60 (HR = 1.357, P < 0.05), early recurrence (HR = 3.265, P < 0.05), vascular tumour thrombus (HR = 1.572, P < 0.05), non-postoperative chemotherapy (HR = 2.069, P < 0.05), lymph node staging (HR = 1.409, P < 0.05), and abdominal metastasis (HR = 1.513, P < 0.05) were independent risk factors affecting the survival of patients with postoperative recurrence (Table 2). Among these factors, early recurrence was identified as the most significant risk factor influencing postoperative survival. This observation is supported by the survival curve depicted in Fig. 2A, indicating that the survival of patients with late recurrence was significantly better than that of patients with early recurrence (HR = 3.447, P < 0.05).

Table 2 Univariate and multivariate analysis of OS in postoperative patients with gastric cancerFig. 2

Cumulative survival curves after surgery for different risk factors among all recurrent patients. Survival curves comparing the early recurrence group with the late recurrence group (A). Survival curves comparing groups that received preoperative neoadjuvant chemotherapy with those that did not (B). Survival curves comparing groups that received postoperative adjuvant chemotherapy with those that did not (C)

Additionally, an analysis was conducted to examine the relationship between adjuvant chemotherapy and postoperative survival. The results revealed that there was no significant association between preoperative neoadjuvant chemotherapy and postoperative survival (Fig. 2B). On the other hand, patients who underwent postoperative adjuvant chemotherapy exhibited significantly better survival rates compared to those who did not receive chemotherapy (HR = 1.839, P < 0.05) (Fig. 2C). These findings indicate a clear positive effect of postoperative adjuvant chemotherapy on postoperative survival.

Subgroup survival analysis of postoperative adjuvant chemotherapy

Further subgroup analysis was conducted regarding the impact of postoperative adjuvant chemotherapy on early and late recurrence patients. The analysis demonstrated a substantial survival benefit in early recurrence patients (HR = 2.223, P < 0.05) (Fig. 3A), whereas there was no significant difference observed in late recurrence patients (Fig. 3B).

Fig. 3

Subgroup survival analysis of postoperative adjuvant chemotherapy. Prognostic differences between early relapse (A) and late relapse (B) patients with adjuvant chemotherapy after surgery. Postoperative adjuvant chemotherapy in patients with TNM stage I/II (C) and TNM stage III (D)

Moreover, the study classified patients into two groups based on TNM staging (I/II, III). Figures 3C and 3D illustrate the differential impact of postoperative adjuvant chemotherapy on patients with different TNM stages. Remarkably, only stage III gastric cancer patients experienced a significant survival benefit with postoperative adjuvant chemotherapy (HR = 2.321, P < 0.05). These results suggest that the positive survival effects of postoperative adjuvant chemotherapy are more significant in early recurrent cases and in patients with stage III gastric cancer.

Development and validation of model for predicting early recurrence

A clinical nomogram was developed based on independent predictors of early recurrence to estimate the risk of early postoperative recurrence in gastric cancer patients (Fig. 4A). The nomogram included age, serous infiltration, lymph node metastasis, recurrence mode, and the tumour marker CA19-9 as predictors. In the training cohort, Fig. 4B displays the calibration curve used for predicting the probability of early recurrence. The results indicated that the nomogram’s predictions aligned well with the actual observations, with a mean absolute error (MAE) of 0.027. The area under the curve (AUC) for this prediction model was 0.739 (95% CI: 0.682–0.798) (Fig. 4C), and the C-index for predicting early recurrence was 0.739.

Fig. 4

Development and validation of prediction model. Nomogram for predicting early postoperative recurrence in patients with gastric cancer (A). Calibration curves of nomogram in training cohort (B) and verification cohort (C). The diagonal line represents the performance of ideal nomogram, and the solid line represents the consistency between the built nomogram and the actual nomogram. ROC curves of nomogram in the training cohort (D) and in the verification cohort (E). The area below the red line (AUC) represents the performance of nomogram

Validation the predictive accuracy of nomogram for early recurrence

To assess the suitability of the model, a validation cohort consisting of 165 patients from the same center was used. Independent risk factors included in the nomogram were evaluated in the validation cohort. Figure 4D indicates good consistency of the nomogram calibration curves for predicting the risk of early recurrence, with a MAE of 0.027. The AUC for this prediction model in the validation cohort was 0.743 (95% CI: 0.652–0.833) (Fig. 4E), and the C-index for predicting early recurrence in the validation cohort was 0.743.

Decision curve analysis

Figure 5 displays a decision curve analysis comparing the developed nomogram with the TNM staging system. The analysis shows that both the nomogram and TNM staging are beneficial for predicting early recurrence when the patient’s probability exceeds 30%. The nomogram outperforms the 8th AJCC-TNM system in both cohorts, indicating its superior predictive ability for early recurrence in gastric cancer patients.

Fig. 5

Decision curve analysis for prediction of early recurrence. The black line represents the net benefit of none of the patients receiving general treatment interventions; the gray line represents the net benefit of patients receiving general treatment interventions; the green line represents the net benefit for patients receiving pTNM staging interventions; the red and blue lines represent the net benefit for patients receiving nomogram interventions in the training cohort and validation cohort, respectively. The red and blue lines is above the rest of lines, indicating that the nomogram provides clinical benefit