Effects of Clavien-Dindo Classification on Long-Term Survival of Patients With Advanced Gastric Cancer After Radical Resection: A Propensity Score-matched Study

Background The impact of postoperative complications (POCs) classified by the Clavien-Dindo (C-D) system on long-term survival after radical resection in patients with advanced gastric cancer (AGC) is not yet clear. Methods This study analyzed 531 patients with AGC who underwent radical resection in an institution between January 2015 and December 2017. Patients were divided into 2 groups according to the occurrence of POCs and recorded according to C-D classifications. The long-term survival outcomes of the entire cohort after propensity score matching (PSM) were compared. Results After PSM, there was no significant difference in baseline data between the complications (C) group (n = 92) and the non-complications (NC) group (n = 92). Survival analysis showed that the 5-year overall survival (OS) and relapse-free survival (RFS) were lower in the C group (48.9% vs 62.0%, P = .040; 38.5% vs 54.9%, P = .005; respectively). Subgroup analysis showed that severe complications (C-D grade > II) were associated with a decrease in 5-year OS and RFS compared with the matched NC group (40.0% vs 62.0%, P = .008; 29.4% vs 54.9%, P = .001; respectively). Multivariate analysis confirmed adjuvant chemotherapy, tumor size, and complications were independent risk factors for poor survival outcomes. Further multivariate analysis showed that older age, combined excision, and comorbidities were independent risk factors for POCs. Conclusions Severe complications reduced the survival outcome of patients. More attention should be paid to perioperative management of patients with high risk factors for complications.


Introduction
Gastric cancer is a global disease, the fourth leading cause of cancer-related deaths worldwide, 1 and the second leading cause of cancer-related deaths in China. 2 Radical resection of gastric cancer lesions combined with adequate lymph node dissection is key to treatment of gastric cancer.4][5] The most common POCs are anastomotic problems, obstruction, and infectious complications such as pneumonia, abdominal abscess, urinary tract infection, and incision infection.
Postoperative complications (POCs) were defined as any deviation from normal postoperative course, 6 which were always associated with longer hospital stays, increased economic costs, and increased mortality. 7,80][11] Some researchers have suggested that the immune response of POCs may provide a suitable microenvironment for tumor cells remaining after surgery, thus promoting tumor recurrence and deterioration of survival outcomes. 12,13n this study, propensity score matching (PSM) was used to assess the impact of POCs on patient long-term survival outcomes to reduce potential confounders.

Patients
Patients were selected from our prospectively maintained gastric cancer database from January 2015 to December 2017.Inclusion criteria for the study were as follows: (1) adenocarcinoma confirmed by gastroscopy and pathological biopsy; (2) invasion depth greater than pT1; (3) no distant metastasis; (4) radical resection (R0) was performed; (5) other malignant tumors were not present; (6) emergency surgery was not performed; and (7) preoperative chemotherapy or radiotherapy was not performed.Pathological staging was graded according to the 8th Union for International Cancer Control (UICC)/ American Joint Committee on Cancer (AJCC) staging system of gastric cancer. 14Propensity score matching with a .02caliper width was used to match the 2 groups on a 1:1 basis.The study was approved by our institute's ethics review committee.

Postoperative Evaluation and Follow-up
Postoperative Complications (POCs) were classified according to the Clavien-Dindo (C-D) classification system. 6,15Clavien-Dindo grade above level I was considered to have complications, otherwise it was considered to have no complications.Complications above grade II were considered serious.Multiple complications occurred in 1 patient, with only the most severe recorded.Patients were followed up every 3 months for the first 2 years, then every 6 months for 2 to 5 years, and then once a year.Overall survival was defined as the time from surgery to death or the last follow-up.Relapse-free survival was defined as the time of surgery to recurrence, death, or the last follow-up.The final follow-up date was October 2020.

Statistical Analysis
To reduce the impact of clinicopathological characteristics on survival outcomes, 1:1 PSM was performed using a logistic regression model and the following covariates: sex, age, body mass index (BMI), preoperative albumin, surgical method, resection range, history of upper abdominal surgery, combined resection, tumor size, comorbidities, pT stage, pN stage, pTNM stage, and adjuvant chemotherapy.The chisquare or Fisher's exact test was used to compare categorical variables between the 2 groups.The survival rate was compared by the Kaplan-Meier method and long-rank test.Multivariate analysis was performed using the Cox proportional hazard model to determine independent risk factors for OS and RFS reduction.Multivariate analysis was performed using a binary logistic regression model to identify independent risk factors for POCs.P-values less than .05were considered statistically significant.SPSS version 25.0 (SPSS Inc., Chicago, IL, USA) was used for statistical analysis and PSM.

Clinicopathological Characteristics
Table 1 shows the patient characteristics of the entire cohort (n = 531) and the PSM cohort (n = 184).There were differences in age, preoperative albumin, combined resection, comorbidities, and pT staging between the C and NC groups (P < .05).After PSM, patient characteristics were balanced in terms of age, sex, BMI, preoperative albumin, comorbidities, surgical method, resection range, combined resection, tumor size, pT stage, pN stage, pTNM stage, and postoperative chemotherapy.

Postoperative Complications
Table 2 shows the details of POCs classified by C-D classification.The overall incidence of POCs was 17.7% (94/531) among the 531 patients in the entire cohort.The incidence of serious complications was 6.6% (35/531).

Survival Outcomes
The median follow-up time was 47 months.Postoperative complications were associated with a decreased 5-year OS (48.9% vs 62.0%, P = .040)and RFS (38.5% vs 54.9%, P = .005)(Figure 1), indicating that the survival status of the C group was better than that of the NC group, and this difference was statistically significant.
To further assess the impact of different C-D levels, we analyzed the survival outcomes of grade II POCs and matched NC groups.Grade II POCs had a tendency to decrease 5-year OS (54.4% vs 62.0%, P = .262)and RFS (43.9% vs 54.9%, P = .074)(Figure 2), although the difference was not statistically significant.Another subgroup analysis showed that severe complications (C-D grade > II) reduced OS (40.0%vs 62.0%, P = .008)and RFS (29.4% vs 54.9%, P = .001)(Figure 3) compared with the matched NC group; this difference was statistically significant.
We also assessed the differences between grade C-D II and severe complications.The results showed that severe complications reduced OS (37.2% vs 59.2%, P = .023)in a statistically significant manner and tended to reduce RFS (28.6% vs 46.9%, P = .054),but this difference was not statistically significant (Figure 4).

Risk Factors for Postoperative Complications
Multivariate analysis showed that age over 50, combined resection, and comorbidities were independent predictors of POCs (Table 3).

Risk Factors Associated With Decreased Overall Survival and Relapse-Free Survival
Multivariate analysis showed that POCs, stage pTNM III, tumor diameter > 5 cm, and absence of adjuvant chemotherapy were independent predictors of OS and RFS reduction (Table 4).

Discussion
This study investigated the relationship between POCs and OS and RFS after radical gastrectomy for advanced gastric cancer.Postoperative complications can prolong hospital stay and increase cost and postoperative mortality.Studies have shown that POCs can reduce the longterm survival outcome of colorectal cancer; [16][17][18] the same trend has also been observed in gastric cancer. 9,19,20n this study, we use the C-D classification to classify POCs; this system has been regarded as a classical classification method and has been used to classify a variety of POCs. 6,15It has also been used to classify POCs of  gastric cancer. 21,22The cohort was divided into 2 groups according to the presence of complications.Because the baseline characteristics of patients between the 2 groups were not comparable, we used PSM to reduce potential confounding factors and selection bias.The present study showed that POCs were related to poor long-term survival outcomes.To evaluate the relationship between complication grade and survival outcome, we further analyzed the impact of C-D grade complications and severe complications on survival outcome.We found that grade complications had a negative effect on 5-year OS and RFS, but the results were not statistically significant, in contrast with a previous study. 23nother subgroup analysis showed that severe complications reduced OS and RFS in patients with severe complications; this difference was statistically significant.Further subgroup analysis showed that severe complications reduced RFS in patients with severe complications compared with grade II complications.At the same time, there was a trend of negative impact on OS, but the results did not reach statistical significance (P > .05).
Because of the negative impact of POCs on survival outcome, it is necessary to understand the risk factors and take intervening measures.We carried out single-and multi-factor analyses in the whole cohort in this study.Multivariate analysis showed that age over 50, combined resection, and comorbid diseases were independent risk factors for POCs.Many studies have shown that there is a relationship between advanced age and the occurrence of POCs, 24,25 which is consistent with our research.Old age will lead to a decrease in the physiological function of various organs, deterioration of nutritional status, and a decrease of anesthetic tolerance, which may be related to the increase of POCs.Therefore, in the perioperative management of patients, more attention should be paid to elderly patients.A number of studies have shown that combined resection is associated with an increase in POCs; 24,26 our study came to the same conclusion.Combined organ resection can prolong the operation time and increase the scope of surgical injury, which may increase POCs compared with gastrectomy alone.8][29] In addition, studies have shown that combined resection may negatively affect the recurrence-free survival of patients. 30herefore, more caution should be taken in deciding whether to perform multiple organ resection.
In terms of comorbidities, many studies have shown that comorbidities increase POCs, [31][32][33] which is consistent with our results.This may be related to the fact that codisease can reduce the patient's respiratory and cardiac reserves.In addition, studies have shown that diabetes is associated with infectious complications. 34,35Therefore, appropriate perioperative management of patients with comorbid diseases is necessary.
7][38] In our study, the chemotherapy data of the 2 groups were similar after PSM.As a result, the potential effect of adjuvant chemotherapy on survival outcome was eliminated to the greatest extent.Chinese experts agree to recommend preoperative neoadjuvant therapy for advanced   46 Due to the small number of patients receiving preoperative neoadjuvant therapy and different chemotherapy regimens in our institution during the time span of this study, patients who did not receive preoperative neoadjuvant therapy were excluded.Thus far, the mechanism by which complications lead to poor survival outcomes is unclear.The generally accepted explanation is that the inflammatory microenvironment can promote tumor development, while complications can lead to a severe inflammatory response, which suppresses the patient's immune system and promotes tumor recurrence and metastasis. 12,13here were some limitations of this study.First, this study was a single-center retrospective study.Second, the operation was performed by multiple groups of physicians, and the influence of surgical experience on the outcome of the operation could not be ruled out. 31Finally, PSM could not rule out all deviations; there may be other variables that affect surgical outcomes and survival outcomes that were not included in the study.
In conclusion, POCs had a negative impact on the longterm survival of patients with advanced gastric cancer, and severe complications (C-D grade > II) significantly reduced survival.Therefore, more attention should be paid to the perioperative management of patients older than 50 years, with combined resection and comorbid diseases.

Figure 1 .
Figure 1.Comparison of Overall survival (A) and Relapse-free survival (B) between the C group and NC group of all patients in the propensity score-matched cohort.

Figure 2 .
Figure 2. Comparison of Overall survival (A) and Relapse-free survival (B) between matched NC group and grade II complications.

Figure 3 .
Figure 3.Comparison of Overall survival (A) and Relapse-free survival (B) between matched NC groups and severe complications.

Figure 4 .
Figure 4. Comparison of Overall survival (A) and Relapse-free survival (B) between grade II complications and severe complications.

Table 1 .
Characteristics of Patients.

Table 3 .
Multivariate Analysis of Risk Factors for POCs.

Table 4 .
Analysis of Prognostic Factors Associated With Decreased Survival Outcomes.