Multimodality Treatment Including Triplet Regimen as First-Line Chemotherapy May Improve Prognosis of Serum AFP-Elevated Gastric Cancer with Liver Metastasis

Serum α-fetoprotein- (AFP-) elevated gastric cancer is a rare tumor that has a poor prognosis due to high incidence of liver metastasis. This study sought to investigate the optimal treatment modality. A total of 319 gastric cancer patients with liver metastasis (GCLM) whose serum AFP levels were tested before treatment were enrolled in this study. They were classified as the serum AFP ≥ 20 ng/ml group (n = 74) and the AFP < 20 ng/ml group (n = 245). Median OS of the AFP < 20 ng/ml group was significantly longer than that of the AFP ≥ 20 ng/ml group (15.7 m versus 10.9 m, P = 0.004). ORR of first-line chemotherapy was 43.3% and 56.1% of the two groups, respectively (P = 0.024). Of patients who received doublet regimen, ORR of the AFP ≥ 20 ng/ml group was significantly lower (38.2 versus 56.9%, P = 0.013), while in those received triplet regimens, ORR between two groups was similar (66.7% versus 66.7%, P = 0.676). Moreover, for patients of the AFP ≥ 20 ng/ml group, those who reached PR had a longer survival period (15.4 m versus 9.4 m, P = 0.017), and combined with local treatment for liver metastasis also seemed to improve prognosis (19.2 m versus 8.4 m, P = 0.003). In conclusion, serum AFP-elevated GCLM had a poorer prognosis. Multimodality treatment including aggressive first-line chemotherapy with triplet regimen may be needed when treating them.


Introduction
Serum α-fetoprotein (AFP) has been proposed as a tumor marker for screening liver tumor and germ cell tumors in the clinic [1][2][3]. 70-95% of hepatocellular cancers are associated with increased AFP level. Serum AFP-elevated gastric cancer was first reported by Alpert et al. in 1970 [4]. Many other reports followed. The incidence of AFP-producing gastric cancer was merely 1.3-15.0% worldwide [5]. In most literatures, the gastric cancer patients with serum AFP elevation were found to have doughty invasiveness and poor prognosis [6][7][8][9]. Also, Liu et al. reported that the dismal prognosis of serum AFP-elevated gastric cancer was mainly due to high incidence of synchronous and metachronous liver metastasis, even when radical operation was done [5]. Therefore, systemic chemotherapy became the predominant treatment method for serum AFP-elevated gastric cancer with liver metastasis (GCLM).
Due to the rarity of this special cancer, there is limited data in the literature about optimal treatment modality. Although previous studies reported that conventional chemotherapy was predominantly ineffective for these patients [10,11], it remains controversial whether to perform systemic chemotherapy for this subtype of GCLM, and there had been so far no suggestion for choosing the optimal regimen.
The potential underlying molecular mechanism of AFPproducing gastric cancer may be the common embryonic origin of the stomach and liver from the foregut [12]. Koide et al. found that AFP-related gastric cancers had higher proliferative activity, weaker apoptosis, and richer neovascularization, compared with that of AFP-negative gastric cancers [13]. As the precise underlying mechanism of serum AFP-elevated GC remains to be elucidated, the optimal treatment approach requires further consideration. We aim to discover the optimal treatment modality for this special subtype GCLM.

Patients and Methods
2.1. Patient Selection. Between 2005 and 2016, 2047 patients were diagnosed with advanced gastric adenocarcinoma in our institute. 516 of them were diagnosed with liver metastasis (LM), including postoperative LM and LM at the initial diagnosis. We included subjects who had serum AFP test result before treatment, leaving 319 patients for analysis. Pretreatment serum AFP was assessed by radioimmunoassay (normal value: <7.0 ng/ml).

Data Collection.
We collected age, gender, ECOG, primary lesion site, histological type, Lauren classification, human epidermal growth factor receptor-2 (HER2) status, serum AFP level before treatment, first-line chemotherapy regimens, response, local treatment for LM, and survival information.
2.3. Follow-Up Care. All patients were regularly followed up from the date of the first hospitalization at our center. Objective response rate (ORR) were evaluated by RECIST version 1.0 (before 2009) and RECIST version 1.1, and severe adverse events (≥grade 3) were recorded. Overall survival (OS) was defined as the time from inspection of liver metastasis to death from any cause or last follow-up.

Statistics.
The Pearson chi-square test was used to measure the differences between variables. The Fisher exact test was used when the numbers were less than five. To identify prognostic factors of overall GCLM patients and the AFP ≥ 20 ng/ml subgroup, survival durations were calculated using the Kaplan-Meier method and Cox regression. For all tests, a P value < 0.05 was defined as significant. The SPSS software program (version 21.0; SPSS, Chicago, Illinois) was used for the analyses. The GraphPad Prism 6 (GraphPad Software Inc., La Jolla, CA) was used for chart making.

Results
3.1. Characteristics of GCLM of the AFP ≥ 20 ng/ml Group and AFP < 20 ng/ml Group. Of the 319 eligible patients, 74 (23.2%) were found to have serum AFP ≥ 20 ng/ml. Table 1 compared the clinicopathologic features of patients between the AFP ≥ 20 ng/ml group (n = 74) and AFP < 20 ng/ml group (n = 245). Results of age, gender, ECOG, disease status, primary lesion site, Lauren classification, HER2 status, peritoneal metastasis, and number of LM were similar between two groups.
Notably, compared with the serum AFP < 20 ng/ml group, 10 (13.5%) patients were diagnosed with hepatoid adenocarcinoma in the AFP ≥ 20 ng/ml group. Gastric hepatoid adenocarcinoma (GHA) was defined as a special subtype of primary gastric adenocarcinoma characterized by the histologic structures of "hepatocellular carcinoma-(HCC-) like differentiation" with or without excessive production of AFP [14,15]. In addition, portal vein tumor thrombus (PVTT) occurred frequently in the AFP ≥ 20 ng/ml group, while it is rarely observed in the AFP < 20 ng/ml group (14.9% versus 2.0%, P < 0 001). The clinicopathologic features of the two groups were detailed in Table 1.
With regard to second-line chemotherapy, there were fewer patients of the AFP ≥ 20 ng/ml group who received second-line chemotherapy than the AFP < 20 ng/ml group (40.0% versus 53.0%, P = 0 055). Regimens mainly involved taxanes and fluorouracil. Moreover, there were 1 (1.4%) and 8 (3.3%) patients who received surgery treatment after first-line chemotherapy in the two groups. In addition, 23 (31.1%) and 60 (24.5%) patients received local treatment for LM in the two groups, respectively, and there were no significant differences between them, either. Comparison of treatments and response between two groups were shown in Table 2.

Objective Response Rate (ORR) of Doublet and Triplet
Regimens between the AFP ≥ 20 ng/ml Group and AFP < 20 ng/ ml Group. Table 3 summarized the response to doublet and triplet regimen between two groups. Result showed that compared with the AFP < 20 ng/ml group, the AFP ≥ 20 ng/ml group had a significantly poor response to platinum/taxanebased doublet regimen (38.2% versus 56.9%, P = 0 013). However, with chemotherapy of triplet regimen, ORR was similar between two groups (66.7% versus 66.7%, P = 0 676).
We further compared ORR and occurrence of severe adverse events of different regimens in the AFP ≥ 20 ng/ml group in Table 4. Result showed that ORR of triplet regimen   In the AFP ≥ 20 ng/ml group, nine patients received triplet regimen as first-line chemotherapy. Among them, six were male, and only one patient was more than 60 years old. Serum AFP levels ranged from 22 ng/ml to 208,072 ng/ml. One patient's primary tumor was located at GEJ, others' primary tumor was located at gastric body or antrum. Three patients were diagnosed with hepatoid adenocarcinoma, the six left were common adenocarcinoma. Of five cases that Lauren's classification and HER2 status were known, one case was with intestinal type, two with diffuse type, and two with mixed type, only one case was examined as HER2 positive.
Two-thirds of patients (6/9) achieved PR after triplet regimen as first-line chemotherapy, and two achieved SD with tumor shrinkage of 18%. Only one patient's disease progressed quickly after only one cycle of chemotherapy. Two-thirds of patients (6/9) suffered from ≥grade 3 adverse events, and four of them had to change treatment regimens because of intolerable toxicity. Disease-free survival (DFS) and overall survival (OS) data were also included in Table 5.

Prognostic Factors of Overall GCLM and GCLM with
Serum AFP ≥ 20 ng/ml. Median serum AFP level was 480.9 ng/ml and 3.1 ng/ml in the AFP ≥ 20 ng/ml group and AFP < 20 ng/ml group, respectively. Median overall survival period was 10.9 m and 15.7 m in the two groups (P = 0 004, Figure 1). Furthermore, multivariate analysis revealed that besides typical prognostic factors of histologic type, extrahepatic unresectable advanced/metastatic sites, response to chemotherapy, and so on, elevation of serum AFP was also an independent prognostic factor for overall GCLM (details were shown in Table 6).
We further investigated the prognostic factors of GCLM patients with serum AFP ≥ 20 ng/ml. While the ORR of triplet regimen was excellent in AFP-elevated GCLM, analysis showed no significant difference in survival between doublet and triplet regimens (37.6 m versus 9.9 m, P = 0 157) due to the rather small number (9/74, 12.2%) of patients receiving triplet regimens. In addition, for GCLM patients with serum AFP ≥ 20 ng/ml, univariate and multivariate analysis revealed that response to first-line chemotherapy was an independent prognostic factor. Patients reached PR had a better prognosis, similar to overall population, while patients evaluated as SD/PD had the worst survival prognosis (P < 0 001, Figure 2). Also, survival analysis showed that combined with local treatment for LM may result in better prognosis and significant difference exist (19.2 m versus 8.4 m, P = 0 003) (Figure 3).

Discussion
The main findings of this study are as follows: (1) serum AFP ≥ 20 ng/ml GCLM showed a poorer prognosis than the AFP < 20 ng/ml group. (2) Doublet regimen was significantly less effective for the AFP ≥ 20 ng/ml group than in the AFP < 20 ng/ml group. (3) Triplet regimen increased ORR compared to doublet regimen when treating serum AFP ≥ 20 ng/ml GCLM, but result showed no significance on survival.
AFP is a fetal serum protein by fetal and yolk sac cells and by some fetal gastrointestinal cells [16]. After birth, the level of AFP rapidly decreased. The elevation of AFP in serum of people older than one year is indicative of either HCC or yolk sac tumor. In addition, some reports showed that AFP could also be produced by other tumors, including gastric cancer, rectal cancer, pancreas cancer, gallbladder cancer, lung cancer, and bladder cancer [17].
In 1970, Alpert et al. first reported a case of AFPproducing GC, which refers to a type of gastric cancer that AFP is positive in the immunohistochemical staining of pathological specimen [4]. In 1985, Ishikura et al. proposed a new entity, hepatoid adenocarcinoma of the stomach, which showed a histologic appearance typical of HCC, including solid, trabecular, and pseudogranular structure, tumor cells were round or polygonal in shape [18]. In addition, Nagai et al. clarified that hepatoid adenocarcinoma of the stomach had characteristic histologic features and a poor prognosis irrespective of AFP production and should be distinguished from AFP-positive GC without hepatoid features [14]. However, due to focal distribution and high heterogeneity of gastric hapatoid adenocarcinoma (GHA) [9,19], almost all GHA cases reported in previous literatures were diagnosed from postoperative specimens. On the other hand, due to aggressive behavior and high frequency of liver metastasis [8], most patients had lost operation opportunity at diagnosis. Thus, most of our patients' feature was just with serum AFP elevation, with only ten patients diagnosed as GHA. The definition of AFP-producing GC varies between studies owing to difficulty in setting the cut-off value; considering liver metastasis can be a factor for mild increase in AFP level, we chose AFP ≥ 20 ng/ml as a cut-off value in this study.
Serum AFP-elevated gastric cancer is rare, only accounts for 2.3-7.1% of all gastric cancers [6,20], but in GCLM population, our result showed that 23.2% (73/319) patients' serum AFP exceeded 20 ng/ml. To clarify the variables associated with the poor prognosis of the AFP ≥ 20 ng/ml group, we next reviewed the data of patients and analyzed the differences between two groups. Result showed the serum AFP ≥ 20 ng/ml group had a significantly poorer response to first-line chemotherapy in comparison to the AFP < 20 ng/ml group. Also, survival analysis revealed that response to firstline chemotherapy was significantly associated with survival prognosis, and for the AFP ≥ 20 ng/ml group, those who reached PR after first-line chemotherapy had a similar survival period as those AFP < 20 ng/ml. These results indicate that choosing effective chemotherapy regimen may improve prognosis of serum AFP-elevated GCLM.
In general treatment of inoperable locally advanced and/or metastatic (stage IV) GC, doublet combinations of platinum and fluropyrimidines are generally used, with an overall ORR of 52.2-58.7% [21,22]. There remains controversy regarding the utility of triplet regimes, especially in China and Japan [23]. Although there is considerable improvement in medicine science, serum AFP-elevated GC is found to have a poor response to chemotherapy and thus associated with a poor prognosis [24]; basic research indicated that AFP-producing cell lines were not sensitive to many drugs [25]. This clinical study further suggest that this special subtype of gastric cancer may be less sensitive to doublet regimen including platinum and fluoropyrimidines, which was in accordance with a previous study reported that for GHA, ORR and disease control rate (DCR) to palliative chemotherapy was only 7.7% (1/13) and 45.1% (6/13), respectively [24]. However, by comparing ORR and AEs of different regimens used in the AFP ≥ 20 ng/ml group, we found that triplet regimens combining platinum, taxanes, and fluoropyrimidines achieved a satisfactory ORR in this special subtype of GCLM. Although ≥grade 3 adverse events were reported in 66.7% patients, they were all reversible and there was no treatment-related death occurred.
Despite the excellent ORR of triplet regimen, median OS was similar in patients who received triplet regimen and doublet regimen. This phenomenon that could mainly attribute to the rate of patients who received triplet regimen was relatively low in our study (9/73), and four of nine patients were still alive until the last follow-up. On the other hand, in clinic practice, triplet regimen was always used in patients with heavy tumor burden, which may be associated with poor prognosis as well. Case by case analysis revealed there were two patients who lived longer than two years. Both of them reached PR after triplet regimen chemotherapy. One of them was examined as HER2 positive and received Herceptin treatment at second-line chemotherapy. In addition, the patient also received TACE for LM, which may also improve survival prognosis of gastric cancer with liver metastasis [11,26]. The other case was diagnosed as GHA, who also received multiline systemic treatment, including apatinib. We noted that there were two patients complicated with PVTT, which is a special characteristic of AFP-related gastric cancer [27]; a high rate of PVTT in AFP-producing gastric cancer (14.9% in our study) may indicate high intendancy of vascular invasion and angiogenesis [28,29]. It was reported that silencing AFP inhibits VEGF production in human HCC cells [30]. The function of apatinib, as a small molecular tyrosine kinase targeting VEGF-R2 (vascular endothelial growth factor receptor-2), is antiangiogenesis. There is another case report of targeted therapy with apatinib in a patient with advanced gastric cancer and high serum level of AFP and PFS achieved five months [31]. Thus, the inhibitors of VEGF or VEGFR might become potential drugs to treat this special subtype of gastric cancer. This long-time survivor in our study also received TACE for LM during treatment process.
We also showed that the only patient resistant to triplet regimen was a young female, with diffuse type Lauren classification and peritoneal metastasis and ascites, which were associated with poor prognosis and bad response to conventional systemic chemotherapy [32]. Furthermore, in our study, almost half of patients (cases 1, 4, 7, and 8) had to stop triplet regimen during treatment process because of intolerable adverse events; thus, severe toxicity of triplet regimen may also attribute to poor survival prognosis, and optimizing triplet regimens deserves further study. In a word, survival analysis and the two successfully treated cases indicate that although AFP-producing gastric cancer is often advanced and complicated with liver metastasis, long-term survival can be achieved by multimodality treatment including triplet regimen chemotherapy; those had a PR response of first-line chemotherapy could get more chance to be treated.

Conclusions
Serum AFP-elevated gastric cancer is a small subgroup of gastric carcinoma with high metastatic potential to the liver and poor prognosis. Multimodality treatment including aggressive chemotherapy of triplet regimen may be worthwhile to improve prognosis of serum AFP-elevated GCLM; better tolerated regimens should be investigated further in the future.

Shortcomings and Perspectives
Although the retrospective nature of this study and the number of cases treated with triplet regimen were relatively small, the results could still provide some clinical value. With such rare tumors, for which large clinical trials are not feasible, it became very important to summarize clinical experience retrospectively. Not limited within GCLM, maybe triplet regimen can be tried to be used in all AFP-elevated gastric cancer in future clinical practice. Although this type of gastric cancer is rare, it deserves further studies.

Ethical Approval
This study was approved by the Ethics Committee of Beijing University Cancer Hospital.

Consent
Patients were not required to give informed consent for this study because the analysis used anonymous clinical data that were obtained after each patient agreed to treatment by written consent.