HAC is a rare special type of extrahepatic adenocarcinoma, the stomach is the most common site of HAC. The mechanism of HAS is not fully understood. Ishikura et al. believed that HAS originated from endoderm stem cells, which can differentiate into hepatocyte cell lines and/or intestinal cell lines; HAS can also be caused by the transdifferentiation of cells from adenoid to hepatoid[12].Research by Akiyama et al. also showed that hepatoid adenocarcinoma has the same origin as coexisting tubular adenocarcinoma[11].Gastric cancer is closely related to liver cancer. During the embryonic development, the stomach and liver are both developed from the foregut. Therefore, some primary gastric cancers have a differentiation disorder, which eventually leads to their differentiation into hepatocytes and the formation of hepatoid adenocarcinoma [5]. Histopathological features are the gold standard for the diagnosis of HAS. As long as hepatocellular differentiation areas appear in the primary focus of gastric cancer, regardless of whether serum AFP is elevated or immunohistochemical staining is positive for AFP, it can be diagnosed as HAS [3, 5]. The typical histology of HAS is usually the coexistence of adenocarcinoma area and hepatocellular differentiation area, the two can migrate between each other, and a few are all hepatocellular differentiation areas. The adenocarcinoma area is often located in the gastric mucosa, showing tubular and/or papillary structures with different degrees of differentiation. The tumor cell nuclei are deeply stained, and some tumor cells have transparent cytoplasm and nuclear shift. The hepatocellular differentiation area is often located in other layers of the stomach wall, usually showing morphological similarity to HCC histologically, and polyhedral tumor cells with large central nuclei, prominent nucleoli, and abundant eosinophilic cytoplasm were found in hematoxylin-eosin (HE) staining. Blood sinuses are quite abundant in hepatocellular differentiation area, the tumor cells are arranged in a string, and bile-like substances are visible around some cancer cells (Fig. 2) [1, 2, 4].
HAS has a lower incidence rate, previous studies have shown that it accounts for only 0.17–1% of all gastric cancers. The difference in proportion is not only related to race and region, but also closely related to the level of diagnosis and treatment. In our study, HAS accounted for 0.51% of all gastric cancers during the same period, which was consistent with previous studies. Compared with CGC, the clinical manifestations of HAS are not specific[13]. Our analysis shows that the clinical manifestations of HAS include upper abdominal pain, upper abdominal discomfort, meconium, nausea, and anorexia. Wang et al. found that HAS is more common in the elderly, with a higher proportion of males and a higher frequency of intestinal type and liver metastases [14]. Qu et al. studied 95 HAS patients from China, and the results showed that HAS mainly occurs in the elderly, males are more common than females, mainly in the gastric antrum, usually ulcerative, and the degree of differentiation is mostly poorly differentiated[15]. Lin et al. reported that the most common type in HAS is Borrmann type III, and the presence of recent bleeding ulcers can often be seen. In the present study, most of the patients were elderly men with a median age of 62 years. 54.8% (17/31) were Borrmann III type, 64.5% (20/31) were poorly differentiated type, consistent with previous studies, but the tumor sites were more common in the upper third of the stomach (20/31). And 22.6% (7/31) of patients presented with black stool as the initial symptom, which is consistent with the recent bleeding ulcers often seen in tumor lesions in previous studies [10].
Elevated serum AFP levels are an important feature of HAS, but not all HAS patients have elevated serum AFP levels. Su et al. reported an elevated serum AFP level in 84.8% of patients with HAS, ranging from 1.0 to 475,000 ng/mL [1]. Another study reported that about 80.6% of patients with HAS had elevated serum AFP levels [16]. Of the 31 patients in this study, 27 had serum AFP levels detected at the time of diagnosis, and 81.5% (22/27) patients had elevated serum AFP levels, of which 26.0% (7/27) patients had higher serum AFP levels than the instrument measurable upper limit. AFP is a glycoprotein produced and secreted by the fetal liver, yolk sac, and gastrointestinal tract, and serum AFP levels drop rapidly after birth. The increase in AFP levels is mainly related to the occurrence of hepatocellular carcinoma (HCC). In addition, many other tumors, such as nonseminoblastic germ cell tumors and endodermal sinus tumors, can also cause AFP levels to increase. Serum AFP level has the significance of disease monitoring, the level generally begin to decrease 1 to 2 weeks after tumor resection, return to normal 2 months after surgery, and increase again when distant metastasis or recurrence occur [17]. Ishikura et al. believe that AFP has an immunosuppressive effect, and its production will affect the prognosis of HAS [12].Wang et al. believe that preoperative serum AFP levels ≥ 500 ng/ml are related to OS (P = 0.007) and DFS (P = 0.05), and preoperative serum AFP level is a sensitive prognostic biomarker for DFS and OS[14]. For patients with HAS, elevated serum AFP level is an important feature, which is great helpful to confirm the diagnosis. It is recommended as a routine test item for patients with gastric cancer to reduce the missed diagnosis rate of HAS.
Immunohistochemical detection is also of great significance for the diagnosis of HAS, but there is no very clear marker. AFP is often diffusely positive in the hepatocellular differentiation area, while it is weakly positive or focally positive in the adenocarcinoma area [2]. The adenocarcinoma area is composed of well-differentiated intestinal epithelial cells and usually contains CEA [5]. The positive rate of Glypican-3 in HAS is higher than that of CGC, and it can be used for the diagnosis of HAS [1]. SALL4 is not expressed in adult normal gastric tissues and HCC, but is partially expressed in normal gastric cancer. In addition, it is diffusely expressed in HAS [18]. Akiyama et al. found that the hepatocellular differentiation area contains AFP positive cells [11]. Kumashiro et al. found that 83% of patients showed AFP expression [19]. Osada et al. reported that the positive rate of AFP staining in HAS patients was 80%, and the positive rate of Glypican 3 staining was 56% [20]. Another study reported that AFP staining was found to be positive in 91.6% of patients, and CEA staining was found to be positive in 78.7% of patients [1]. In our study, 77.4% (24/31) patients were positive for AFP staining, 64.7% (11/17) patients were positive for CEA staining, and 77.8% (14/18) patients were positive for Glypican-3 staining. We also found that Ki67 positive rate was 100%, of which 71.0% (22/31) positive area was > 50%. Previous studies also reported that the expression of Ki-67 in the tumor cell nucleus was moderately increased [21]. These results indicate that HAS has a high malignant potential, such as high proliferative activity, weak apoptosis and abundant neovascularization [22], which may be the reason for its poor prognosis.
Compared with CGC, HAS is more common in lymph node metastasis, liver metastasis, and extensive vascular invasion, with a worse prognosis. Lin et al. found that 90% of patients had lymph node metastasis and/or distant metastasis at the time of diagnosis [10]. Another study found that the total incidence of liver metastases in the HAS group was 75.6%, while the total incidence of liver metastases in the CGC group was 11.5% (P < 0.01) [13]. Osada et al. found that the incidence of vascular infiltration (P = 0.005) and distant metastases (P = 0.0458) was higher in hepatoid adenocarcinoma compared with non-hepatoid adenocarcinoma[20]. In the present study, 80.6% (25/31) of patients had regional lymph node metastasis at first visit, and 29.0% (9/31) of patients had distant metastasis at first visit. 25.8% (8/31) of the patients had liver metastasis at the time of first diagnosis, and 4 patients had liver metastasis during follow-up. That is, the incidence of liver metastasis was 38.7% (12/31), including 25.8% (8/31) simultaneous liver metastasis and 12.9% (4/31) metachronous liver metastasis. 66.7% (14/21) patients had vascular tumor thrombus, and 68.8% (11/16) patients had nerve invasion. This aggressive biological behavior of HAS leads to a poor prognosis. According to Liu et al., the one-year, three-year and five-year survival rates of HAS are 30%, 13% and 9%, respectively, while the CGC group is 95%, 57% and 38% (P < 0.01) [13]. Qu et al. reported that the 3-year survival rate of HAS was 7.4%, and the median OS was 10 months. Compared with patients with metastasis, the survival time of patients without metastasis was longer (P = 0.001), and the survival of patients receiving surgical treatment longer than patients treated with non-surgical methods (P = 0.046). Survival analysis showed that survival time was associated with metastasis (P = 0.002) and liver metastasis (P = 0.036)[15]. Another study found that the interval between gastrectomy and liver metastases in patients with HAS is shorter than CGC, resulting in a poorer prognosis[8]. Baek et al. reported that the median OS of patients with stage I-III and stage IV HAS was 28.0 and 8.2 months, respectively[6]. Zeng et al. found that vascular infiltration, pTNM stage and adjuvant therapy are independent risk factors for prognosis[16]. The median OS of the patients in this study was 28 months. The 1-year survival rate was 56%, the 3-year survival rate was 47%, and the 5-year survival rate was 40%. The survival rate in this study is higher than previous studies, probably because the data was collected retrospectively in a single center, more patients underwent radical surgery and adjuvant chemotherapy, and fewer patients in stage IV, resulting in selection bias. Univariate Cox regression analysis revealed that an elevated CA125 level, distant metastasis, and radical surgery were associated with prognosis (P < 0.05). A multivariate analysis further revealed that distant metastasis remained an independent prognostic risk factor for survival (P = 0.038; HR = 27.016, 95%CI: 1.205 ~ 605.664).
The prognosis of HAS is poor, but due to its low incidence, there is currently no standard treatment plan, and clinical treatment of HAS is similar to the treatment of CGC. Radical surgery, including the resection of liver metastases and adjuvant therapy, is considered the best treatment option. However, compared with CGC, the radical resection of HAS is more challenging because of its higher incidence of vascular invasion, lymph node metastasis and distant metastasis before surgery. Radical surgery, early pTNM and adjuvant therapy can significantly improve the prognosis of HAS [16]. One study described several different treatment options that can delay the progression of HAS, including cisplatin plus fluorouracil and epirubicin or irinotecan combined with mitomycin C[23]. Fang et al. reported a case of AFPGC patients with liver metastasis who underwent multimodal treatment including surgery, chemotherapy, interventional therapy and sorafenib targeted therapy, with an OS of 30 months[24]. Petrelli et al. reported that a case of hepatoid adenocarcinoma of pancreatic with liver, lung, and lymph node metastases treated with sorafenib had a progression-free survival time of 7 months[25]. In our study, 71.0% (22/31) of patients underwent radical surgery, and 72.7% (16/22) of patients undergoing radical surgery received adjuvant chemotherapy. The survival rate of patients was higher than previous studies, which may be due to the fact that more patients undergo radical surgery and adjuvant chemotherapy in our group.Univariate analysis showed that radical surgery is associated with prognosis(P = 0.0001;HR = 0.087,95%CI: 0.025 ~ 0.299), but between receiving adjuvant chemotherapy or not, the difference was not statistically significant (P = 0.936༛HR = 1.068༌95%CI༚0.214 ~ 5.319).Multivariate analysis revealed that only distant metastasis is an independent prognostic risk factor for survival (P = 0.038; HR = 27.016, 95%CI: 1.205 ~ 605.664). Due to the sample size is small, risk stratification was not done based on the TNM staging of this group of patients and other risk factors. In addition, the single-center retrospective study may have a selection bias. The benefit of radical surgery and adjuvant chemotherapy need to be explored by further large-scale clinical research. And whether targeted drugs such as sorafenib can be used to treat HAS also look forward to additional studies.