Potential survival benefits of open over laparoscopic radical gastrectomy for gastric cancer patients beyond three years after surgery: result from multicenter in-depth analysis based on propensity matching

We selected two hospitals (Fujian Medical University Union Hospital and Qinghai University Affiliated Hospital) that have facilities for electronic storage of clinical data, including medical records, images, or laboratory data, for all consecutive patients with GC who underwent gastrectomy during the study period. All patients at these institutions who met the inclusion criteria were enrolled. This study was approved by the appropriate Institutional Review Board (Fujian Medical University Union Hospital and Qinghai University Affiliated Hospital). All the patients were well informed and gave their full consent after receiving a verbal explanation of this study and an information document.

The patients enrolled in this study had histologically confirmed gastric adenocarcinoma, diagnosed as clinical stage cT2-4aN0-3M0, and had undergone radical distal gastrectomy with D2 lymphadenectomy between April 2007 and December 2014. Eligible patients with the lesion located in the lower or middle third of the stomach on preoperative evaluation, including gastroscopy, full-abdominal CT enhancement and upper gastrointestinal angiography, were included. Conversely, patients with cT1/T4b or unknown stage, positive residual tumor margin, other malignant tumors or a history of malignant tumors, D1 + or D3 lymphadenectomy, and conversion to open surgery were excluded (Fig. 1). Finally, 1256 patients were enrolled, including 505 patients who underwent open surgery and 751 patients who underwent laparoscopic surgery. All the patients selected in this study were performed by surgeons with rich experience in open and laparoscopic radical distal gastrectomy. All patients received preoperative conversation with the surgeon before operation and chose the open or laparoscopic surgical methods and signed the informed consent. In this study, preoperative cT and cN staging of patients was conducted according to the Eighth Union for International Cancer Control tumor-node-metastasis (TNM) classification [15], which was evaluated by two physicians, respectively, in combination with Computerized tomography(CT), Magnetic Resonance Imaging(MRI) and ultrasonic gastroscopy of the patients. If there were differences in the results, PETCT would be used to further confirm the clinical staging if necessary. cT staging was mainly evaluated by ultrasound gastroscopy and CT, and the criteria were: The presence of a mass in the stomach usually is diagnosed as a hypoechoic or dark thickening in one or more layers. A dark thickening extending to but not completely through the fourth layer with a preserved smooth outer echogenic border is associated with a T2 tumor. Complete loss of all the layers, associated with a smooth white outer layer representing the serosal surface, suggests penetration through the muscularis propria into the subserosal fat is consistent with a T3 tumor. If there is extension of the dark wall thickening with loss of the serosal echogenic stripe, the tumor is staged T4a. the criteria of cN + was: tumor involvement if round and/or > 8 mm in short axis diameter in preoperative imaging [16‐18]. In this study, the D2 lymph node dissections of LDG and ODG were consistent, and standard D2 lymph node dissections including N0.1–7, 8A, 9, 11p, and 12a were performed in accordance with the Japanese Guidelines for the Treatment of Gastric Cancer.

Postoperative follow-up was performed in outpatient and hospital settings, every 3 months within 2 years, every 6 months within 3–5 years, and once a year after 5 years. The vast majority of patients routinely underwent physical examination, laboratory tests, chest radiography, full-abdominal color Doppler ultrasound or full-abdominal CT, and annual gastroscopy. Recurrence was identified by medical history and physical examination in combination with imaging evaluation, cytology, or tissue biopsy (preferred when feasible). The endpoints were overall survival (OS) and disease-free survival (DFS).

The incidence of postoperative complications with a severity of grade III or higher according to the Clavien–Dindo classification was assessed [19].

We calculated the life years gained using differences in restricted mean survival time [20] (RMST). The RMST is the mean survival time of all subjects in the study population followed up to t, defined as RMST(t) = E(min(T, t)), which is the area under the survival curve of T up to time t. the RMST(t) can be estimated well with the area under the corresponding Kaplan–Meier curve up to time t. Note that the area above the survival curve up to time t is the restricted mean time lost(RMTL(t)), which is t-RMST(t). With the RMST(t) or RMTL(t) summary measure, one can then construct the corresponding mean survival time curve over time, which provides a temporal, stochastic profile of T, expressed in units of time, for evaluating, for example, the benefit or safety of a new therapy. The advantage of using such a quantification over the survival rate is the setting of a fixed-time analysis.

We performed landmark analyses to assess outcomes before and beyond 3 years. Landmark analysis was introduced in 1983 by Anderson et al.[21] The goal is to estimate, in an unbiased way, the survival probabilities in ODG and LDG conditional on the group membership of patients at the landmark time (beyond 3 years of postoperatively).

Kaplan–Meier curves were used to analyze the OS and DFS of different surgical methods, and log-rank test was used to test whether there were significant differences between survival curves. Cox regression was used to conduct univariate analysis on the factors affecting OS, and the meaningful results of univariate analysis were incorporated into multivariate analysis to determine the independent prognostic factors that affect OS.

PSM [22] was conducted by a biostatistician who was blinded to the outcomes. The propensity score was estimated using a logistic regression model, and a 1:1 (ODG to LDG) matching (matched without replacement) with a caliper of 0.01 standard deviation of the estimated logit was performed. In addition to the propensity score, four preoperative factors (age, body mass index [BMI], clinical T and N factors) were exactly matched.

Continuous variables are expressed as mean ± standard deviation, and categorical variables are expressed as percentages. The chi-square test, Fisher’ s exact probability method, or unpaired t test were used to compare the differences in clinicopathologic data between the ODG group and the LDG group. Analyses were performed using SPSS 23.0 and RStudio version 1.1.419 (RStudio Inc.). P < 0.05 on both sides was considered to indicate a significant difference.

A total of 1256 patients with ADGC were included in this study. Of them, 505 were categorized into the ODG group and 751 in the LDG group before PSM. Before PSM, the patients in the ODG group had younger age, lower BMI, more advanced stage (including cT stage, cN stage, pT stage), more lymphatic vessel invasion, and higher proportion of those with adjuvant chemotherapy than patients in the LDG group (all P < 0.05). After matching, there were 461 patients in each group and the general data between the groups were comparable (P > 0.05) (Table 1).

Supplemental Table 1 showed that the ODG group had a longer operation time (199.0 vs. 169.8 min, P < 0.001) and more bleeding volume (121.3 vs. 78.7 mL, P < 0.001) than the LDG group. Moreover, the fluid intake time (5.3 vs. 4.7 days, P < 0.001), drainage tube removal time (8.7 vs. 8.3 days, P = 0.001), hospital stay (14.1 vs. 13.0 days, P = 0.027), and postoperative recovery were longer in the ODG group. The number of lymph node dissections and the postoperative exhaust time were comparable in the two groups. A total of 80 cases of postoperative complications occurred in the ODG group, including 12 cases of Clavien–Dindo grade III–IV. Meanwhile, 68 cases of postoperative complications occurred in the LDG group, 9 of which were grade III–IV, the rate of complications were comparable in the two groups (P = 0.535).

The median follow-up time for the entire group of patients was 88 months (0–137 months).The Kaplan–Meier survival curve showed that the 3- and 5-year OS and DFS in the ODG group were comparable to those in the LDG group (3-year OS: 74.2% vs. 76.1%, P = 0.259; 3-year DFS: 63.2% vs. 65.1%, P = 0.331; 5-year OS: 64.8% vs. 67.6%, P = 0.369; 5-year DFS: 62.8% vs. 63.6%, P = 0.469) (Supplemental Fig. 1).

Figure 2A shows the results of RMST analysis on the effect of tumor size on the survival of patients with different surgical methods. At 1 cm intervals, when the tumor size was ≤ 5 cm, the 3 -and 5-year RMST of the ODG group were shorter than those of the LDG group (3-year RMST: − 1.11, − 0.96, − 2.05, − 1.54, and − 3.57 years; 5-year RMST: − 0.38, − 1.87, − 4.45, − 3.45, and − 16.75 years). When the tumor size was > 5 cm, the 3 and 5-year RMST of the ODG group were longer than those of the LDG group (3-year RMST: 0.37, 2.66, 3.88, and 2.22 years; 5-year RMST: 0.83, 5.10, 5.74, and 3.56 years). Therefore, we analyzed the survival with a tumor size of 5 cm as the cutoff point.

Figure 2B shows the results of stratified analysis according to cT stage. The 3-year RMST of cT2-4a patients in the ODG and LDG groups were comparable (RMST: cT2, 0.21 year, P = 0.835; cT3, − 0.850 year, P = 0.233; cT4a, − 0.138 year, P = 0.16). In the 5-year RMST, although the two groups were comparable in cT2 and cT3 (RMST: cT2, 0.03 year, P = 0.987; cT3, − 1.72 years, P = 0.368), the RMST in cT4a in the ODG group within 5 years was statistically different from that in the LDG group (RMST: cT4a, − 8.46 years, P = 0.005).

Supplemental Table 2 showed that the operation time of the ODG group was significantly longer than that of the LDG group (200.1 vs. 180.9 min, P = 0.048) in cT4a patients. In terms of postoperative complications, 33 cases occurred in both the ODG and LDG groups, of which three cases were Clavien–Dindo grade III–IV, with no statistically significant difference (P = 0.982).

In patients with cT4a, when the tumor was > 5 cm, there was no difference in the 3-year OS and DFS between the two groups (3-year OS: 53.4% vs. 37.9%, P = 0.216; 3-year DFS: 55.2% vs. 45.1%, P = 0.431); however, the 5-year OS and DFS of the ODG group were significantly better than those of the LDG group (5-year OS: 49.8% vs. 37.1%, P = 0.033; 5-year DFS: 52.2% vs. 38.2%, P = 0.045) (Supplemental Fig. 3). Using RMST to verify this result. In patients with cT2-3 patients, the RMST of ODG was all comparable to LDG (P > 0.05) (Supplemental Fig. 4A). In patients with cT4a and tumors ≤ 5 cm, the 3-year RMST and 5-year RMST in ODG group were all shorter than LDG group(3-year RMST:− 2.11, P = 0.265, 5-year RMST: − 6.42, P = 0.012). In patients with cT4a and tumors > 5 cm, the 3-year RMST in the ODG group was comparable to that in the LDG group (3-year RMST: 0.98 year, P = 0.480), but the 5-year RMST in the ODG group was significantly longer than that in the LDG group (5-year RMST: 4.67 years, P = 0.042) (Supplemental Fig. 4B). Figure 3 shows the results of further landmark analysis. In patients with cT4a and tumors > 5 cm, the 3-year OS of patients in the ODG and LDG groups was comparable (P = 0.895), and the OS beyond 3 years of postoperatively was significantly better in the ODG group than in the LDG group (P = 0.033).

Supplemental Table 3 shows the results of multifactor risk analysis performed to study the impact of ODG and LDG on the OS of cT4a patients, including the factors (age, BMI, tissue type, Tumor size, cN, operation time, intraoperative blood loss, pT, pN, surgery, prealbumin, preoperative hemoglobin, CEA, and postoperative complications), that affect survival in these patients. The results showed that ODG was a risk factor for the 3 and 5 year OS in cT4a patients (3-year OS, hazard ratio [HR] 1.67 [0.65, 4.26]; 5-year OS, HR 1.13 [0.58, 2.50]).

Analysis of recurrence after matching showed that patients in the ODG and LDG group had similar in overall recurrence (137 vs. 140, P = 0.286), local recurrence (44 vs. 33, P = 0.234), peritoneal recurrence (32 vs. 46, P = 0.123), multiple-site recurrence (13 vs. 11, P = 0.837), and other or unknown-site recurrence (48 vs. 50, P = 0.915) (Supplemental Table 4). The overall recurrence time distribution was similar in the two groups (Fig. 4A). The same results showed that in cT4a patients, the two groups showed the similar number of recurrences and similar time of recurrence at each site (Supplemental Table 5, Fig. 4B).

cT4a patients were further stratified. Among patients with cT4a and tumors > 5 cm, the ODG and LDG groups showed non-significant differences in total recurrence (35 vs. 60, P = 0.178, Fig. 4C), local recurrence (13 vs. 15, P = 0.682), and other or unknown-site recurrence (14 vs. 22, P = 0.711). In terms of peritoneal recurrence, the number of recurrences in the ODG group was significantly fewer than that in the LDG group (4 vs. 17, P = 0.033) (Supplemental Table 6), and the recurrence time was later than in the LDG group (Fig. 4D).With respect to multiple-site recurrences, the multiple-site recurrence time was later in the ODG group (Fig. 4E). Supplemental Fig. 5 compares the recurrence risk of patients in the ODG and LDG groups. In cT4a patients, the recurrence risk was comparable between the ODG and LDG groups; however, in cT4a patients with tumors > 5 cm, the recurrence risk of the ODG group was significantly lower than that of the LDG group. A further landmark analysis was performed to compare the OS of patients with cT4a and tumors > 5 cm, excluding the patients who suffered peritoneal recurrence and multiple-site recurrence, there was not significantly different within 3 years and beyond 3 years of postoperatively in OS between two groups (P = 0.715 and P = 0.500, respectively) (Supplemental Fig. 6).