GC is one of the most common malignant tumors in the digestive tract in China, with a low early diagnosis rate, a low surgical resection rate, and a high mortality rate [11]. Due to the lack of typical clinical symptoms in patients with GC, most patients are already in the advanced stage at diagnosis. Most advanced patients present distant metastases and the most common metastatic sites of GC include the liver, peritoneum, and distant lymph nodes [8]. The incidence of liver metastasis in advanced GC patients is 44.5% [12], but due to the lack of effective treatment, the prognosis of these patients is usually poor, which places a heavy burden on patients and their families. How to improve the prognosis of these patients is also an important challenge for clinicians. Compared to patients with liver metastases from colorectal cancer, the prognosis of GCLM is much worse. The molecular mechanisms of metastasis between the two diseases may be different, although the malignancy of GC is higher, and the mechanisms underlying GCLM may be more complex. Conversely, a standard set of diagnostic and treatment guidelines has been established for liver metastasis of colorectal cancer, while a similar systematic treatment plan for GCLM has not yet been developed, and various treatments are also being explored.
Currently, the TNM staging system is commonly used in clinical practice to assess the prognosis of cancer patients, and is considered the standard for cancer diagnosis, treatment, and prognosis [13]. Given that factors other than TNM staging may also affect the prognosis of patients with cancer, we often see certain differences in prognosis between patients at the same stage in clinical practice, because the TNM staging system ignores the biological heterogeneity of patients. Therefore, using the TNM staging system alone to predict the prognosis of patients is not accurate. A more accurate prediction model should be developed to evaluate the prognosis of patients with GCLM, to provide more convenient and reliable survival prediction and treatment strategies for individual patients with GCLM. Nomograms are currently a commonly used prediction model that can predict the probability of clinical outcomes or certain types of events. Nomograms can also be used to straightforwardly visualize the results of data analysis. Currently, nomograms have been widely applied to predict the prognosis of patients with different types of cancer [14].
Firstly, this study comprehensively explored the clinical and pathological characteristics of patients with GCLM in the SEER database. Through univariate and COX multivariate analysis, we found that age, pathological grade, surgery, and chemotherapy are independent risk factors for OS in patients with GCLM. Second, we established a nomogram model to predict the 1- and 3-year OS probability of patients with GCLM and conducted internal and external verification analysis using the cohort from the SEER database and the database of the First Affiliated Hospital of Sun Yat-sen University, respectively. Finally, we found that our model had good predictive accuracy.
When conducting prognostic factor analysis, we found that older patients often present a poorer prognosis consistent with clinical observations, as well as with other types of cancer. This may be related to the decline in immunity observed in the older patients, multiorgan aging, and the comorbidity due to multiple diseases. Furthermore, for older patients, the choice of treatment options is more limited, leading to treatment that is not very aggressive. Compared to young patients, older patients are more likely to experience postoperative complications and slow postoperative recovery, all of which contribute to poor surgical tolerance in older patients [15, 16]. At the same time, we found that the higher the pathological grade, the poorer the prognosis. This is consistent with the findings of Hu et al. [17], who also believe that the degree of tumor differentiation is significantly correlated with OS in patients with GCLM, and poorly differentiated adenocarcinoma is a high-risk factor for GCLM. According to the nomogram scores, patients who did not receive chemotherapy or surgical treatment will have higher scores and the corresponding 1- and 3-year survival probabilities are low, which is consistent with current clinical results. According to the NCCN [18] and the Japanese Guidelines for the Treatment of Gastric Cancer [19], it is recommended to use systemic chemotherapy or a combination of targeted and immunotherapy for patients with advanced metastatic unresectable GC. Systemic chemotherapy includes preoperative neoadjuvant chemotherapy, palliative chemotherapy, conversion chemotherapy, and postoperative adjuvant chemotherapy. Currently, commonly used therapeutic drugs in clinical practice include mainly fluorouracil (5-FU), Calcium levofolinate, capecitabine, Tegafur (S-1), cisplatin, oxaliplatin, irinotecan, docetaxel, paclitaxel, trastuzumab and nivolumab. Based on previously reported clinical trial evidence, these drugs can be used as monotherapy or in combination. At present, more commonly used first-line chemotherapy schemes include the combination of two drugs (fluorouracil + platinum) or three-drug combinations (fluorouracil + paclitaxel/anthracycline + platinum). For patients with HER-2(+++) tumors, trastuzumab can be added [20]. For patients with HER-2(-) tumors, further options can be selected based on PD-L1 expression status; whereas, patients with mismatch repair protein deficiency/highly unstable microsatellites (dMMR/MSI-H) status may consider the addition of immunotherapy [21]. For some patients with multiple metastases who can tolerate chemotherapy side effects, the combination of targeted drugs and immune drugs is also the preferred option [22]. A clinical trial in Japan [23] found that the objective response rate (ORR), the median progression-free survival period (mPFS), and the median total survival period (mOS) of patients with GCLM in the treatment queue of nivolumab combined with LEN/PEM were 46%, 7.8 months, and 15.6 months, respectively, while in the treatment queue for the anti PD-1 single drug, ORR mPFS and mOS were 9, 1.4, and 6.4 months, respectively. Therefore, we can see that the combination of targeted immunotherapy has marked benefits for patients with GCLM. In addition, the combination of neoadjuvant chemotherapy and surgery also has significant benefits for patients. A multicenter experiment in Germany found that patients who received neoadjuvant chemotherapy and surgery had higher mOS than those who received only chemotherapy [24]. For patients with multiple intrahepatic metastases and poor systemic chemotherapy efficacy, transcatheter arterial chemoembolization (TACE) or hepatic arterial infusion chemotherapy (HAIC) may also be considered. Interventional therapy can increase the local concentration of chemotherapy drugs in the tumor microenvironment, improve the local tumor control rate, and result in fewer adverse reactions than systemic chemotherapy.
There are still controversies regarding the surgical treatment of GCLM. GCLM usually manifests as multifocal metastasis, often accompanied by extrahepatic metastasis (such as to the peritoneum and lymph nodes), and the opportunity for curative resection is often lost. When symptoms such as obstruction and bleeding caused by tumors occur, palliative surgery can be performed to improve the patient's quality of life [25]. However, both the fourth edition of the Japanese Guidelines for the Treatment of Gastric Cancer [26] and the latest Chinese Consensus on the Diagnosis and Treatment of Liver Metastasis of Gastric Cancer [9] recommend that for patients with primary GC and those with GCLM that can be R0 resected without extrahepatic metastasis, curative resection should be chosen. The Chinese Expert Consensus on the Diagnosis and Comprehensive Treatment of Gastric Cancer Liver Metastasis divides GCLM into three subtypes based on the resectability of primary and metastatic lesions: resectable type (Type I), potentially resectable type (Type II), and non-resectable type (Type III). Their indications for a feasible and curative resection are that the primary gastric lesion does not invade adjacent organs, the metastatic lymph nodes can be cleared, the liver metastasis is < 3 and the maximum diameter is ≤ 4 cm (or the lesion is limited to one liver lobe), and important blood vessels and bile ducts are not involved. For potential resectable patients who initially do not meet the surgical requirements, targeted and systemic chemotherapy can be performed before tumor evaluation to determine whether they are eligible for resection surgery. The guidelines also recommend administering neoadjuvant chemotherapy prior to surgery, followed by resection of the primary GC lesion, D2 lymph node dissection, and R0 resection of liver metastases. At the same time, adjuvant chemotherapy must be administered for at least 4–8 cycles after surgery [9]. A multicenter study in the UK found that compared to patients with GCLM who did not receive surgical treatment, patients with GCLM who received surgical treatment had significantly higher OS, and surgical treatment can benefit patients with GCLM [27]. A meta-analysis by Danny et al. also suggests that surgical treatment for GCLM can achieve good survival results [28]. Furthermore, the study also showed that surgery has a significant impact on the prognosis of patients with GCLM. The score displayed by surgery on the nomogram is only lower than that of chemotherapy, indicating that surgery is only second in importance for the prognosis of GCLM after chemotherapy. In recent years, radiofrequency ablation has achieved excellent results in the treatment of liver metastases of colorectal cancer and is gradually being used for the local treatment of GCLM [29]. For patients with fewer liver metastases and a lesion diameter less than 3 cm, local radiofrequency ablation is recommended [30]. Previous studies have shown that compared to palliative therapy, the survival rate of patients with liver metastases from GC treated with radiofrequency ablation is significantly improved [31]. In addition, further combination chemotherapy is required after ablation therapy, which can significantly prolong the patient's survival period [32].
Our research has the following advantages. First, this study established a prognostic nomogram to predict OS in patients with GCLM, and internal and external cohort validation showed that the nomogram had high accuracy. Second, this study identified other risk factors that could affect the prognosis of GCLM, except for TNM staging, and stratified patient risk through the median risk score. For higher risk patients with GCLM, doctors can intervene in a timely manner to improve prognosis. Finally, the model we established can supplement the AJCC TNM staging profile, and calculate a personalized survival rate for patients with GC, and more importantly, it can be easily used without the aid of a computer (such as in community clinics). However, our study also has some limitations. First, this was a retrospective study in which there may have been a bias in patient selection during the predictive model construction process. Second, the information in the SEER database is not comprehensive, as it lacks information on serum tumor markers levels, detailed treatment information (such as specific surgical and chemotherapy methods), the number and size of liver metastases, and other prognostic risk factors related to survival, such as history of smoking and alcohol consumption, family history of tumors, Helicobacter pylori infection, and whether the primary lesion has neural and vascular infiltration. The lack of these factors may reduce the ability to evaluate the prognosis of patients with GCLM. Furthermore, only data on patients with concomitant GC and liver metastases are available in the SEER database, and it is not possible to establish a predictive model for GC patients with metachronous liver metastases. Finally, the data used to develop and validate the nomogram derived only from our hospital and the SEER database. Our hospital has a relatively small number of patients with GC accompanied by simultaneous liver metastasis and our hospital is composed mainly of Asian patients, while most patients in the SEER database are of black and white race. Therefore, the predictive capacity of this prognostic model for Asian patients is limited. Although our findings indicate that our nomogram has good predictive accuracy, future large-scale and multicenter clinical studies are still needed in different countries to evaluate its performance.