A total of 127 patients with pre-treatment elevated serum AFP at National Clinical Research Center for Cancer (Tianjin, China) between August 2017 and May 2023 were included in this study. Of these, thirty-five patients presented with elevated serum CEA levels exceeding 5 ng/mL.

The inclusion criteria were as follows:

The exclusion criteria included:

Upon admission, all patients underwent blood tests within one week prior to initiating treatment. The threshold values for serum AFP, CEA, carbohydrate antigen 125 (CA125), carbohydrate antigen 19-9 (CA19-9), carbohydrate antigen 72-4 (CA72-4), carbohydrate antigen 242 (CA242), and lactate dehydrogenase (LDH) were set at 7 ng/mL, 5 ng/mL, 35 U/mL, 6.9 U/mL, 20 IU/mL, and 250 U/L, respectively, as per the manufacturer’s recommendations. The eighth edition of the Tumor-Node-Metastasis (TNM) staging system developed by American Joint Committee on Cancer was used to stage tumors in all patients (14).

A comprehensive follow-up was conducted on a cohort of 127 cases, albeit with the unfortunate loss of some to follow-up. Out of the 109 patients successfully tracked, 29 underwent radical surgical procedures, 34 were treated with chemotherapy, 29 received a combination of immunotherapy and chemotherapy, 14 initiated their treatment with preoperative neoadjuvant chemotherapy, and 3 patients opted for self-discharge. Kaplan- Meier Survival Curves was plotted by https://www.bioinformatics.com.cn, an online platform for data analysis and visualization.

Chi-squared tests, and Student’s t-test tests were applied to the clinical data, with a two-tailed p-value of < 0.05 indicating statistical significance (*p < 0.05, **p < 0.01, ***p < 0.001). All statistical computations were performed with SPSS software, version 27.0 (SPSS Inc., Illinois, USA).

In our cohort of 127 AFPGC patients, the majority were male (97 individuals) with a median age of 60.4 years, ranging from 27 to 90. The tumors were predominantly located in the distal third of the stomach (69 patients), followed by the upper third (34 patients), and the middle third (19 patients). Quite notably, 5 patients presented with extensive disease affecting more than two-thirds of the stomach. Concerning the surgical interventions applied, 37 patients underwent curative surgery, while only 3 received palliative operations, and 6 patients opted for self-discharge. Chemotherapy was administered to 69 patients, which included 14 who received neoadjuvant chemotherapy prior to undergoing curative surgery and 32 who were treated with a combination of immunotherapy and chemotherapy ( Table 1 ).

In our study, nearly half of the patients had mildly elevated pretreatment serum AFP (7-20ng/mL) ( Figure 1A ). We found elevated serum AFP levels (AFP≥20 ng/mL, AFP-H) did not affect patient survival (P=0.37, Figure 1B ). Nevertheless, serum CEA demonstrated superior predictive ability for survival compared to other serum gastrointestinal tumor markers, including AFP, CA19-9, as well as CA72-4, with AUC values of 0.758, 0.533, 0.619 and 0.604, respectively ( Figure 1C ).

Consequently, we postulate that serum CEA promotes tumor progression in AFPGC and patients with elevated serum CEA may have a distinct prognosis. We have categorized this cohort as ‘dual positive gastric cancer (DPGC)’, while those without elevated CEA are termed ‘single positive gastric cancer (SPGC)’. Notably, the survival rate for DPGC patients was substantially lower than for those with SPGC, suggesting a grimmer outlook for individuals with this particular subtype of gastric cancer (P=0.011, Figure 1D ).

Our further investigative work into the malignancy of DPGC tumors found that the majority (98.25%) were classified as TNM stage III-IV, compared to the lesser proportion (75.71%) seen in SPGC cases (P<0.001, Figure 2A ). This trend was also evident in the rates of lymph node metastasis and distant metastasis: 98.25% and 61.40% in DPGC patients, respectively, substantially exceeding the 77.14% and 32.86% observed in the SPGC group (P<0.001 for both, Figure 2A ). Furthermore, a possible organ preference for metastasis in DPGC was hint, as indicated by 82.86% DPGC patients with metastasis presenting with liver involvement, in sharp contrast to the 43.48% rate among SPGC counterparts (P=0.004, Figure 2B ). Additionally, a greater number of DPGC patients exhibited HER2 expression, which is associated with more aggressive tumor behavior (60.42% vs. 34.69%, P=0.011, Figures 2C, D ) ( 15).

Classical tumor markers such as CA19-9, CA72-4, CA242, and CA125 play significant roles in the diagnosis and monitoring of gastric cancer (16, 17). When comparing DPGC patients to those with SPGC, as shown in Figure 2E , there was a marked increase in the proportion of individuals with elevated levels of these markers: CA19-9 (50.94% vs. 22.22%, P=0.002), CA125 (31.03% vs. 8.33%, P=0.026), CA72-4 (56.86% vs. 28.85%, P=0.005), and CA242 (45.65% vs. 13.33%, P=0.001). Moreover, lactate dehydrogenase (LDH), a tumor biomarker linked to cancer metabolism, invasion, and immune system evasion, showed a higher prevalence of increased levels in the DPGC group (43.9%) compared to the SPGC group (12%, P<0.001).

Cancer-related inflammation is acknowledged as a critical intermediary in tumorigenesis, frequently preceding the onset of cancer and promoting its advancement (18). Notably, various hematological parameters are considerably altered in DPGC patients, including reduced lymphocyte count and increased neutrophil count (mean 1.27×10⁹/L vs. 1.58×10⁹/L, P=0.003; mean 4.97×10⁹/L vs. 3.94×10⁹/L, P=0.013, Figure 3A ). Neutrophil-to-lymphocyte ratio (NLR), lymphocyte-to-monocyte ratio (LMR) and platelet-to-lymphocyte ratio (PLR) are cancer-related inflammatory markers that could predict prognoses after treatment in cancer patients (19, 20). We observed that the NLR, PLR, and LMR underwent considerable changes in DPGC patients (mean 4.80 vs. 2.85, P<0.001; mean 269.43 vs. 187.63, P=0.001; mean 2.71 vs. 3.77, P<0.001, Figure 3B ). The systemic immune-inflammation index (SII), derived from lymphocyte, neutrophil, and platelet counts, was notably higher in DPGC (mean 1525.32 vs. 794.32, P=0.005, Figure 3B ), suggesting an association with poorer survival outcomes in gastric cancer (21, 22).

In DPGC patients, positive acute phase reactant proteins (APPs), including α1-globulin and β1-globulin, are significantly elevated (mean values of 6.11% vs. 5.07%, P=0.005; and 6.98% vs. 6.58%, P=0.030, respectively), as shown in Figure 3C . The notable decreases of negative APPs, such as albumin and prealbumin in DPGC (mean 0.17g/L vs. 0.20g/L, P=0.013; 39.64g/L vs. 41.62g/L, P=0.030, Figure 3D ) suggest more pronounced APR activation.

Studies have shown that there is a strong relationship between immune dysfunction and malnutrition, and that the two are mutually reinforcing (23). In our cohort, other than serum albumin and prealbumin, DPGC patients presented reduced hemoglobin levels compared to SPGC patients (mean 113.74g/L vs. 123.37g/L, P=0.042, Figure 3D ). The prognostic nutritional index (PNI), recognized as a predictive marker for survival in gastric cancer, was also significantly lower in the DPGC group (mean 143.67 vs. 176.59, P<0.001, Figure 3E ) (24). Collectively, these nutritional indicators highlight the compromised nutritional status of DPGC patients, which may contribute to their unfavorable prognosis.

Cancer-associated thrombosis (CAT) ranks as a leading cause of mortality in cancer patients, surpassed only by cancer progression itself (25). Both fibrinogen and D-dimer levels in DPGC surpassed the upper normal limit and were significantly elevated compared to SPGC (mean 4.13g/L vs. 3.61g/L, P=0.016; and 2232.02 ng/mL vs. 1323.53 ng/mL, P=0.016, respectively, Figures 3F, G ).The Khorana score, a reliable measure for ascertaining venous thromboembolism risk in cancer patients, indicated that a greater proportion of DPGC patients were classified as high-risk (Khorana score ≥3) in comparison to SPGC patients (47.37% vs. 30.00%, P=0.045, Figure 3H ) (26).

Thereafter we explored the factors contributing to the elevated CEA level (>5 ng/mL) in AFPGC using a binary regression model, as depicted in Figure 4 . The results showed that liver metastasis and increased neutrophil were positively correlated with elevated CEA level (OR 4.64, 95%CI (1.47-14.64), P=0.009; OR 4.49, 95%CI (1.02-19.36), P=0.046). Additionally, AFPGC patients with lymphopenia may also exhibit DPGC (OR 2.60, 95%CI (0.97-6.96), P=0.057).

We continue to explore whether there are differences in the response to treatment regimens between two subtypes of gastric cancer. Initially, radical surgery remains the optimal treatment option for all AFPGC patients. Among non-surgical treatments, patients who received combined immunotherapy and chemotherapy have a longer overall survival (OS) compared to those who only received chemotherapy (median OS 22.0 months vs. 11.0 months, P=0.024, Figure 5A ). Specifically, for the DPGC and SPGC subtypes of gastric cancer, while radical surgery continues to be the best treatment option for both, the response to the combined immunotherapy and chemotherapy regimen is significantly different. Patients with DPGC benefit more substantially from combined immunotherapy and chemotherapy than chemotherapy, while those with SPGC do not (median OS 18.0 months vs. 6.0 months, P=0.005, median OS 22.0 months vs. 22.0 months, P=0.91, Figures 5B, C ). There is no significant difference in overall survival between DPGC and SPGC patients who received combined immunotherapy and chemotherapy (median OS 18.0 months vs. 22.0 months, P=0.64, Figure 5D ).