From the perspective of the routine clinical application, objective assessments of clinical data about surgical resection in GCLM are essential to investigate the surgery strategies in GCLM. Although surgery is recommended to alleviate severe gastrointestinal symptoms in consensus, the utility of surgery in GCLM still remains highly controversial. More recently, inconsistent and contradictory findings of surgical resection in GCLM have emerged in published literature (22, 23).

A public clinical trial (REGATTA) (22) failed to improve the overall survival (OS) rate in advanced gastric cancer patients assigned to gastrectomy plus post-operative chemotherapy than in those assigned to chemotherapy alone (14.3 vs. 16.6 months). However, evidence from a clinical trial (AIO-FLOT3) (23) showed different outcomes. Compared with patients who experienced chemotherapy alone, patients who experienced neoadjuvant chemotherapy followed by surgical resection had superior OS (22.9 vs. 10.7 months). Notably, the design of the REGATTA trial differed from the design of the AIO-FLOT3 trial in some respects, which possibly had influenced outcomes of the trial. First, most GCLM patients enrolled in the REGATTA trial were accompanied by peritoneal metastases, who were recognized as the worst kind in advanced GC patients in prognosis. Second, the surgical management in the REGATTA trial was restricted to D1 lymphadenectomy only, whereas in the AIO-FLOT3 trial, the surgical management adopted gastrectomy with D2 lymphadenectomy, which was recommended for total or subtotal distal gastrectomy (24). Third, compared with gastrectomy plus chemotherapy adopted in the REGATTA trial, the AIO-FLOT3 trial utilized neoadjuvant chemotherapy followed by surgical resection in the treatment plan. Collectively, the above evidence reveals the crucial factors including patient selection, surgical procedures, and treatment options in a multimodality approach to GCLM.

Available evidence of surgery for patients with GCLM mostly relies on retrospective studies, systematic reviews, and prospective trials. Data published after 2000 mostly showed significant and prognostic benefits of surgical resection for GCLM (Table 1) (5, 6, 20, 21, 25, 73), and the benefits were in continuous increase owing to advancements in accurate diagnosis, patient selection, perioperative nutritional support, anesthetic techniques, surgery approaches, management of post-operative complications, and enhanced recovery after surgery.

Recently, principally from East Asia and Europe, large retrospective studies on the surgical resection of GCLM have shown continuously acceptable survival outcomes for selected patients. Nishi et al. (51) demonstrated that the overall 1- and 3-year survival rates after hepatic resection for GCLM were 88.9 and 17.8% in 10 selected patients, respectively, with an MST of 21.5 months and no post-operative mortality. Similarly, in a retrospective single-center study involving 34 patients with GCLM, Ryu et al. (59) investigated the significance of surgical procedures including hepatic resection for more massive metastases and surgical microwave ablation for patients who had a high operative risk and identified prognostic factors. The results showed acceptable morbidity and favorable long-term outcomes, as the 1-, 3-, and 5-year OS rates after surgery were 86.5, 51.4, and 42.1%, respectively, and the 1-, 3-, and 5-year recurrence-free survival (RFS) rates were 38.5, 28.0, and 28.0%, respectively, with no significant survival differences for varied surgical treatments (P = 0.213).

Meanwhile, a nationwide retrospective study from England also showed that gastrectomy combined with hepatectomy for synchronous GCLM might carry survival benefits in selected patients (48). Kaplan–Meier curve analyses showed that patients who were selected to have gastrectomy with additional hepatectomy for liver metastases (GGH group) had survival similar to that of patients who had gastrectomy in the absence of liver metastases (GG group) (P = 0.196) and improved survival than did patients who had gastrectomy without liver resection for liver metastases (GGNH group) (P < 0.001) and patients with GCLM who had no surgery (GNS group) (P < 0.001). As for mortality, the GGH group and GGNH group had similar 30-day mortality (P = 0.246), whereas the former had significantly improved 90-day mortality (P = 0.009), 1-year mortality (P < 0.001), and 5-year mortality (P < 0.001); and the GNS group had the worst OS and highest mortality at 30, 90 days, 1, and 5 years (P < 0.001) in the four groups. The results of this study revealed that gastrectomy combined with additional surgical resection of liver metastases was better than palliative treatment or gastrectomy without resection of liver metastases for patients with GCLM in survival benefits.

To reassess this bias problem in full measure, many systematic reviews and pooled analyses were conducted. A systematic review launched by Liao et al. (74) included eight non-randomized studies, representing a total of 677 patients with GCLM. The median OS time in patients who underwent gastrectomy combined with hepatectomy was significantly prolonged, as compared with the median OS time of those who underwent palliative therapy (23.7 vs. 7.6 months), with survival rates of the two arms of 69, 40, 33%, and 27, 8, 4% at 1, 2, and 3 years, respectively. Compared with palliative therapy, hepatectomy was associated with significantly lower mortality at 1-year (OR 0.17, P < 0.001) and 2-year (OR 0.15, P < 0.001). Owing to the disparity in the stage of disease, differences of the regimen of chemotherapy, and preference of surgery of surgeons (75), patients who underwent hepatectomy in Western countries showed lower median rates of OS at 1 year (60 vs. 76%), 2 years (30 vs. 47%), and 3 years (23 vs. 39%) than did those in Asian countries.

As previously stated, most of the published papers on surgical resection in patients with GCLM came from retrospective data, whereas only four randomized controlled trials (RCTs) investigated the role of surgery for patients with GCLM so far. The REGATTA trial was the first RCT to compare gastrectomy followed by chemotherapy with chemotherapy alone concerning OS in patients with GCLM (22). Findings from this trial denied the survival efficacy of palliative gastrectomy followed by chemotherapy from an interim analysis, which had caused the interruption of this trial in 2016. However, to some extent, results from the AIO-FLOT3 trial countered those of the REGATTA trial by strict inclusion criteria, surgical approaches, and treatment regimens (23). The AIO-FLOT3 trial exhibited favorable survival in patients with GCLM who received neoadjuvant chemotherapy and later underwent surgical resection, which had provided a rationale for the ongoing AIO-FLOT5 trial (NCT02578368) (76). Compared with the REGATTA trial, the AIO-FLOT5 trial excludes the enrollment of patients with clinically visible tumors of the peritoneum and >P1 peritoneal tumors, adopts a complete resection of a primary tumor including standardized lymphadenectomy (R0 and at least D2), and adjusts the place of chemotherapy and surgery. Hopefully, if this trial was proved to be effective, it could potentially lead to a new standard of therapy. Another ongoing trial named SURGIGAST (NCT03042169), which has not recruited patients, aims to compare the OS of palliative surgical resection plus chemotherapy with that of chemotherapy alone for stage IV gastric cancer including GCLM (77).

Despite the significant survival benefits from gastrectomy combined with hepatectomy over non-resectional management in patients with GCLM, as well as favorable published outcomes from chemotherapy followed by surgery over chemotherapy alone, it must be stressed that most of data came from retrospective studies and systematic reviews. Thus, outcome data from the AIO-FLOT5 trial and the SURGIGAST trial are awaited to verify the survival benefit of surgical resection suggested by retrospective studies and systematic reviews.

A considerable amount of published literature about surgery in GCLM illustrates the ascendency of surgery. However, it is conspicuous that not every patient will benefit from surgery. Hence, prognostic evaluation is crucial to identify the suitable candidates for radical surgery, which are of gastric cancer, liver metastases (synchronous disease), and liver metastases alone (metachronous disease), from those who will not benefit from surgery.

Lately, in a multicenter retrospective study, Tiberio et al. (78) compared the application of radical surgery vs. palliative gastrectomy or palliative surgery without resection in GCLM, in which radical surgery had achieved better long-term results than others in the 5-year survival rate (9.3, 2.1, and 0%, respectively). In light of this, they further recognized the best candidates for radical surgery through systematically investigating the patient-related, gastric cancer-related, metastasis-related, and treatment-related prognostic factors. Results confirmed that the invasive depth of primary tumor (P < 0.001), curative surgical procedure (R0 resection; P = 0.001), timing of hepatic involvement (P < 0.001), and adjuvant chemotherapy (P < 0.001) were associated with long-term survival, independently. Especially in R0 resection, results implied that it can significantly reduce the possibility of recurrence in GC patients with liver oligometastasis, even in patients with multiply scattered metastases in both lobes of the liver.

Accordingly, in the metachronous disease, Tiberio et al. (79) also revealed that T4 gastric cancer (P = 0.019), the presence of lymph node metastases (P = 0.05), and grade 3 GC (P = 0.018) displayed negative prognostic factors. Moreover, a multivariate analysis demonstrated that a therapeutic strategy of liver metastases was highly associated with survival as well, in particular when R0 resection was performed (P < 0.001).

Likewise, based on real-world data, the AGAMENON registry involving 1,792 patients with advanced GC (80), distal esophagus, or gastroesophageal junction revealed higher 3-year survival rate after metastasectomy than non-metastasectomy (30.6 vs. 8.4%; P < 0.001) and median OS since metastasectomy of 16.7 months. With the use of a state-arrival extended Markov proportional hazard (PH) model, a multivariate analysis indicated the presence of a HER2-positive tumor treated with trastuzumab (P = 0.001) and chemotherapy followed by surgical procedure (P < 0.001) as favorable predictors of survival. Moreover, they also found that the unreasonable interval time between the initiation of chemotherapy and surgery appeared to worsen outcomes. Their results also recommended that 5 months as interval time benefits most patients, which is consistent with the AIO-FLOT3 trial.

Also, Takemura et al. (66) reported the overall 5-year survival rate of 37% and the MST of 34 months in 64 patients achieved macroscopically complete (R0 or R1) resections. Among 64 patients, 50 patients had the largest hepatic metastasis of more than 5 cm in diameter, and 14 patients had <5 cm in diameter (P = 0.07). Results demonstrated that patients with a maximum diameter of hepatic metastasis >5 cm had poorer long-term survival (P = 0.018).

Above all, most of identified prognostic factors were similar with those in various literature through multivariate analyses, which could be roughly divided into five major categories that consisted of primary tumor-associated, liver metastasis-associated, extrahepatic metastasis-associated, and treatment-associated prognostic factors and others, as shown in Table 2. However, these factors were mainly identified from retrospective studies in single center or multicenter, which need to be validated in prospective clinical studies to further confirm their prognostic role.

Although the Lauren classification and the WHO classification are popular in pathological grading of GCs, they are insufficient to guide personalized treatments, especially in GCLM. New classified evaluation for GCLM is thus required. Encouragingly, recent advancements in retrospective studies and prospective studies have greatly facilitated the identification of potential candidates.

Referring to the clinical study on GCLM and classification of stage IV GC (83), we divided GCLM patients into three categories, as shown in Figure 1. First, GCLM could be divided into the potentially resectable tumor (category I), marginally resectable tumor (category II), and unresectable tumor (category III) according to the analysis of clinical decision making in multidisciplinary treatment. For example, macroscopic peritoneal dissemination was considered an essential factor during the classification process, because patients with peritoneal dissemination or positive peritoneal cytology had significantly poor prognosis (84). Second, patients of category I were recommended to undergo surgery followed by post-operative chemotherapy or to receive neoadjuvant chemotherapy combined with surgery. Patients in category II were suggested to adopt conversion therapy aimed to an R0 resection after combined chemotherapy. Patients in category III, who also had obstruction and bleeding of the gastrointestinal tract in some cases, were advised to receive palliative chemotherapy.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.