Laparoscopic versus open gastrectomy for locally advanced gastric cancer after neoadjuvant chemotherapy: a comprehensive contrastive analysis with propensity score matching

This study included patients with LAGC who received surgery after NC between January 2012 and September 2020 in the Department of Gastrointestinal Surgery, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology. The first OG and LG for LAGC patients after NC were conducted at January 2012 and August 2012, respectively. NC was conducted on LAGC patients with a clinical tumor stage of cT2-4aN1-3M0. The treatment regimens included FOLFOX (oxaliplatin, leucovorin, and 5-fluorouracil), SOX (oxaliplatin and TS-1), XELOX (oxaliplatin and capecitabine), and FLOT (docetaxel, oxaliplatin, leucovorin and fluorouracil), which were selected based on the Chinese Society of Clinical Oncology (CSCO) clinical guidelines of gastric cancer [13] and the tolerance of patients. The following were the criteria for inclusion: (1) clinical tumor stage of cT2-4aN1-3M0, (2) underwent surgery after NC. And the following were the criteria for exclusion: (1) palliative gastrectomy, (2) diagnosis of any other malignant tumor in the past 5 years and (3) incomplete clinicopathological or follow-up data. All patients were separated into two groups: those who received LG (LG group) and those who underwent OG (OG group) after NC. The flow chart for the selection of cases is shown in Fig. 1. The Tongji Medical College Ethics Committee of Huazhong University of Science and Technology ratified the protocol of this study and the approved number is UHCT-IEC-SOP-016–03-01. The informed consent was signed by all patients to have their data used for the study.

All surgeries were performed by experienced surgeons of the Department of Gastrointestinal Surgery, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology. Based on previous literatures [4, 7, 12], and physician experience, LG is only considered for patients who meet the following conditions: (1) without abdominal operation before, (2) patients could tolerate laparoscopic surgery, and (3) without obvious metastasis and R0 resection is possible. The criteria for OG are as follows: (1) patients could tolerate surgery, and (2) without obvious metastasis and R0 resection is possible. For patients according to the above criteria, the benefits and risks of the two surgical methods would be fully explained by the surgical physician, and the final decision was made by the patients. The performance of proximal gastrectomy, distal gastrectomy, or total gastrectomy was depended on the tumor location. All patients underwent standard gastrectomy with D2 lymphadenectomy and standard reconstruction based on the Japanese gastric cancer treatment guidelines (5th edition) [14]. Reconstruction after proximal gastrectomy was performed via esophagogastrostomy, while reconstruction after total gastrectomy was performed via Roux-en-Y esophagojejunostomy. Distal gastrectomy reconstruction was performed via standard Roux-en-Y gastrojejunostomy or Billroth II gastrojejunal anastomosis depending on the surgeon’s preference. All reconstruction processes were performed in an open manner. Postoperative adjuvant chemotherapy was routinely applied to all patients unless they could not put up with it due to serious adverse effects. The postoperative adjuvant chemotherapy regimens included XELOX, SOX, FOLFOX, and S-1. For patients who achieved R0 resection after NC, postoperative adjuvant chemotherapy was carried out consistent with the original regimens if NC was effective. If the disease progressed after NC, subsequent chemotherapy regimens would be discussed through multidisciplinary team. Postoperative adjuvant chemotherapy and NC were 8 cycles in total and were adjusted according to the patient’s disease condition and tolerance.

Demographic data, involving age, sex, body mass index, Charlson Comorbidity Index (CCI) [15], and American Society of Anesthesiologists (ASA) score, were collected from all patients. Enhanced abdominal computed tomography (CT) was performed before NC to evaluate the clinical tumor stage and after NC to assess the chemotherapy response. The American Joint Committee on Cancer (AJCC) Cancer Staging Manual (8th) [16] was used to define the clinical tumor stage, and the Response Evaluation Criteria in Solid Tumors (RECIST version 1.1) was applied to evaluate the chemotherapy response [17].

A standard clinical pathway was used for management of all patients. Liquid diet was started after first flatus, and patients were discharged after the absence of complications. Data related to surgery, including the length of incision, operative time, estimated blood loss, time to first flatus, time to first liquid diet, postoperative complications, and length of postoperative hospital stay, as well as pathological characters, were collected. The Clavien–Dindo classification system was applied to grade postoperative complications, which were defined as incidents happening within 30 days after surgery [18, 19]. Tumor regression grade was evaluated according to AJCC standard established by College of American Pathologists (CAP) [20], and pathological stage was defined according to the AJCC Cancer Staging Manual (8th) [16]. A 30-day mortality was defined as death due to any cause that occurred within 30 days after surgery.

Laboratory data such as hemocytes, hemoglobin (Hb), total protein (TP), albumin (Alb), and prealbumin (PAB) at 1 day before surgery and 1 day, 3 days, and 7 days after surgery were collected. To avoid the potential influence of inconsistent preoperative baseline levels, the reduced value was calculated as preoperative value subtracted postoperative value to evaluate the outcomes of different surgical methods. The systemic immune-inflammation index (SII) and systemic inflammatory response index (SIRI) were adopted to assess the immune and inflammation conditions at 1 day before surgery and 1 day, 3 days, and 7 days after surgery. The SII was calculated as neutrophil count × platelet count/lymphocyte count. The SIRI was calculated as neutrophil count × monocyte count/lymphocyte count. Tumor markers including CEA, CA199, and CA724 at 1 week before surgery and 1 month after surgery were collected to evaluate the short-term oncological outcomes of two surgical methods.

Postoperative follow-up was performed every 3–6 months for the first 2 years and every 6–12 months thereafter via outpatient clinics or telephonic interviews. During follow-up, patients’ complete blood count, liver and kidney function, and tumor markers were examined. Chest-, abdomen-, and pelvis-enhanced CT were examined once every 6 months. Gastroscopy was conducted once every year. The last follow-up date was December 30, 2022. After excluding the cases with incomplete clinicopathological or follow-up data, all 148 cases included in this study achieved complete follow-up, and the median follow-up period was 48.0 (3.0–128.8) months. Median follow-up period in LG group was 44.0 (3.2–119.0) months and was 50.0 (3.0–128.8) months in OG group. Relapse-free survival (RFS) was defined as the duration from surgery to recurrence, and OS was defined as the duration from surgery to death. Local recurrence refers to the recurrence at the original surgical site, while distant metastasis refers to the metastasis of organs or parts other than the original surgical site, such as the liver, peritoneum, lung, and bone.

PSM was conducted by matching patients who underwent LG after NC with those who underwent OG after NC to eliminate differences in baseline statistics. Seven covariates (age, sex, CCI, ASA score, clinical stage, tumor size, and resection type) were chosen to calculate the propensity score. PSM was conducted by one-to-one nearest neighbor method, and the caliper was set on 0.1.

Quantitative variables were presented as mean ± standard deviation, and the differences were compared with independent t-test. Quantitative variables with high deviation were expressed as medians (interquartile range), and the differences were compared with Mann–Whitney U-test. Categorical variables were expressed as frequencies with the differences compared by chi-square test or Fisher’s exact test. Standardized difference was calculated to compare the balance of baseline data before and after PSM. Kaplan–Meier analysis was used to generate the survival curves, and the differences were evaluated by log-rank test. Univariate and multivariate analyses of risk factors for RFS and OS were conducted by the Cox proportional hazards regression model. The cutoff point of the continuous variables associated with survival was determined using the median. The multivariate Cox proportional hazards regression model included variables with P < 0.1 in the univariate analysis. Statistical analysis was conducted by SPSS software (SPSS 20.0, Chicago, IL, USA). R (version 4.0.4) was used for PSM. GraphPad Prism software (version 8.0, USA) was used for image processing. P < 0.05 was considered statistically significant.

As shown in Fig. 1, 148 patients who underwent gastrectomy after NC were enrolled in the study based on inclusion and exclusion criteria. The LG and OG groups consisted of 80 and 68 participants, respectively. The resection type and anastomotic methods were significantly different before matching. After PSM, 55 patients remained in each group, and the baseline data were comparable between the two groups (Table 1). The LG group included 45 men (81.8%) and 10 women (18.2%) with a mean age of 56.9 ± 9.2 years. The OG group included 46 men (83.6%) and 9 women (16.4%) with a mean age of 57.4 ± 10.8 years.

As shown in Table 2, the most common NC regimens were FOLFOX and SOX in both of the groups. According to the RECIST criteria, 3 patients (2.7%) achieved complete response (CR), 48 patients (43.6%) achieved partial response (PR), 55 patients (50.0%) had stable disease (SD), and 4 patients (3.6) had progressed disease (PD) after NC. Grade 3/4 adverse events of chemotherapy happened in 9.1% (10/110) of patients during the NC period based on the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE, version 4.0) [21]. The most common adverse events were neutropenia and myelosuppression. No NC-related death was observed. The postoperative adjuvant chemotherapy portion of the LG group and OG group was 81.8% (45/55) and 83.6% (46/55), respectively, and the difference was not statistically significant (P = 0.801).

The variables related to the short-term surgical outcomes are shown in Table 3. Compared with the OG group, the LG group had a shorter length of incision (6.4 ± 3.2 cm vs. 19.1 ± 4.5 cm, P < 0.001) and less blood loss (121.7 ± 71.5 ml vs. 180.0 ± 89.4 ml, P < 0.001), but the operative time was longer (282.0 ± 64.5 min vs. 254.6 ± 74.2 min, P = 0.041). The proportion of R0 resection (90.9% vs. 90.9%, P = 1.000) and number of lymph node dissection (LND) (22.5 ± 6.0 vs. 24.5 ± 9.2, P = 0.180) were not significantly different between the two groups. As for postoperative recovery, it was shown that the LG group had a shorter time to first flatus (3.4 ± 1.2 days vs. 4.5 ± 1.8 days, P < 0.001) and shorter time to first liquid diet (4.3 ± 1.8 days vs. 5.4 ± 2.0 days, P = 0.004) than those of the OG group, but there was no significant difference in the median postoperative hospital stay (10.0 days vs. 11.0 days, P = 0.291).

The overall incidence of postoperative complications was 20.9% (23/110), with 10 patients (18.2%) in the LG group and 13 patients (23.6%) in the OG group; the difference was not statistically significant (P = 0.482) (Table 4). Grade III or higher postoperative complications occurred in 7 patients (6.4%), including intraperitoneal hemorrhage in 1 patient (OG), delayed gastric emptying in 1 patient (LG), duodenal stump fistula in 1 patient (LG), anastomotic stenosis in 1 patient (OG), respiratory failure in 2 patients (both OG), and cerebral embolism in 1 patient (OG). All these complications were managed by conservative treatment or reoperation as appropriate. No 30-day mortality was recorded.

In order to comprehensively figure out the influence of different surgical methods on LAGC patients following NC, we collected perioperative laboratory data to analyze the nutritional and immune-inflammation status of patients during the perioperative period. By calculating the reduction of red blood cells (RBCs) and Hb, we found that LG group had less reduced RBCs (P = 0.039) and Hb (P = 0.018) in 3 days after surgery (Fig. 2A–B), which confirmed that LG group had less blood loss than OG group during surgery. This may be attributed to the smaller incision size, better visualization of anatomical structures, and more precise hemostasis achieved by laparoscopic surgery. The reduced Alb was also calculated and found to be less in LG group both in 1 day (P = 0.029), 3 days (P = 0.015), and 7 days (P = 0.035) after surgery (Fig. 2D). PAB is a good indicator for short-term nutritional status due to its short half-life period [22, 23]. Our results showed that LG group had less reduced PAB in 3 days after surgery (P = 0.010) (Fig. 2E). All above results indicated that LG is less harmful to the nutritional status of LAGC patients after NC. The SII and SIRI are generally acknowledged indicators for the immune-inflammation status of patients [24, 25]. By calculating the SII and SIRI before surgery and 1 day, 3 days, and 7 days after surgery, we found no significant difference between the two groups (Fig. 3), which further proved the safety of LG for LAGC patients after NC.

To evaluate the short-term oncological outcomes of different surgical methods, the data of tumor markers at a week before surgery and 1 month after surgery were collected. The results showed no significant difference between the two groups (Fig. 4). Regarding the long-term oncological outcomes, we recorded and analyzed the recurrence and survival status of patients after a median follow-up of 48.0 (3.0–128.8) months. The overall recurrence within 3 years occurred in 24 (43.6%) and 28 (50.9%) patients in the LG and OG groups, respectively (P = 0.445). And there were no significant differences in local recurrence (P = 0.768) and metastasis (P = 0.550) between the two groups. The 5-year RFS and 5-year OS rates in the LG group were 46.9% and 46.7%, respectively, and the corresponding rates in the OG group were 43.3% and 52.1%, respectively. No significant difference was found in RFS (P = 0.446) or OS (P = 0.742) (Table 5). The Kaplan–Meier analysis showed no significant difference in RFS and OS both before and after PSM (Figs. 5 and 6).

Univariate Cox regression analysis showed that surgical procedures were not relevant to RFS (P = 0.448) or OS (P = 0.742). The risk factors related to RFS were tumor size ≥ 4 cm, lymph-vascular invasion, nerve invasion, R1 resection, clinical response of RECIST being SD or PD, ypTNM stages III or IV, and the absence of postoperative adjuvant chemotherapy. The risk factors related to OS were tumor size ≥ 4 cm, lymph-vascular invasion, nerve invasion, R1 resection, clinical response of RECIST being SD or PD, ypTNM stages III or IV, and the absence of postoperative adjuvant chemotherapy (Table 6). Multivariate Cox regression analysis showed that tumor size ≥ 4 cm (P = 0.021) and the absence of postoperative adjuvant chemotherapy (P = 0.012) were independent risk factors of OS (Table 7).