Recurrence risk prediction of small-molecule drugs plus chemo in postoperative gastric cancer: Retrospective analysis via molecular markers

Abstract

BACKGROUND

Gastric cancer (GC) exhibits high postoperative recurrence rates even after adjuvant small-molecule and chemotherapy therapy. Reliable molecular markers are urgently needed to accurately predict recurrence risk and guide precise clinical treatment.

AIM

To investigate the clinical value of molecular markers in predicting recurrence risk in patients with postoperative GC receiving small-molecule drugs combined with chemotherapy.

METHODS

The clinical data of 136 patients with postoperative GC who received adjuvant therapy with small-molecule drugs combined with chemotherapy at our hospital between September 2023 and January 2025 were retrospectively analyzed. Based on the follow-up outcomes, the patients were divided into a recurrence group (38 cases) and a non-recurrence group (98 cases). Baseline data and molecular marker test results were compared between the groups. Factors influencing recurrence after treatment and the predictive value of molecular markers for recurrence risk were analyzed.

RESULTS

There were no significant differences in baseline data, including sex, body mass index, primary tumor location, The Cancer Genome Atlas subtype, or serum levels of carcinoembryonic antigen, alpha-fetoprotein, and CA125, between the two groups (P > 0.05). The recurrence group showed significantly higher proportions of age ≥ 60 years, poorly differentiated tumors, TNM stage III-IV, programmed death-ligand 1 (PD-L1) positivity, and higher tumor mutation burden (TMB) expression than the non-recurrence group (P < 0.05). Binary Logistic regression analysis showed that age (≥ 60 years), poor differentiation, TNM stages III-IV, PD-L1 positivity, and high TMB expression were independent risk factors for recurrence (OR = 3.126, 1.852, 4.109, 6.174, and 1.627, respectively; all P < 0.05). The area under the curve of the combined detection of PD-L1 positivity and TMB for predicting recurrence was 0.983, which was higher than that of single-molecule marker detection (P < 0.05).

CONCLUSION

Detection of molecular markers such as PD-L1 and TMB is highly valuable for predicting recurrence risk in GC patients. A combined analysis of baseline characteristics can further clarify the risk of recurrence after adjuvant therapy with small-molecule drugs combined with chemotherapy.

Key Words: Smallmolecule drugs; Chemotherapy; Gastric cancer; Recurrence risk; Molecular markers

Core Tip: This study retrospectively analyzed 136 patients with postoperative gastric cancer receiving small-molecule drugs and chemotherapy. Programmed death-ligand 1 and tumor mutation burden are independent risk factors for recurrence, and their combined detection has a high predictive value.

INTRODUCTION

Gastric cancer (GC) is a malignant gastrointestinal tumor with high morbidity and mortality rates worldwide. In recent years, the incidence of GC in China has been increasing annually owing to multiple factors, such as changes in dietary habits and lifestyle[1]. Radical gastrectomy (RG) is the core approach for achieving a clinical cure; however, patients with intermediate and advanced GCs still face a high risk of postoperative recurrence. Despite the implementation of postoperative adjuvant therapy, a combination of small-molecule drugs and chemotherapy, the recurrence rate remains at approximately 27%, which has a significant impact on patient prognosis and quality of life[2]. Therefore, identifying the risk factors for the postoperative recurrence of GC and establishing a precise recurrence prediction system are of great clinical significance.

The combination of small-molecule drugs and chemotherapy is the standard regimen for the adjuvant treatment of GC after surgery. Small-molecule drugs such as apatinib and regorafenib act synergistically with chemotherapeutic agents to significantly reduce the risk of postoperative recurrence[3]. However, clinical practice has revealed differences in the treatment response and recurrence risk among patients. Although traditional clinical baseline characteristics enable a preliminary evaluation of recurrence risk, they fail to accurately reflect tumor biology and therapeutic sensitivity[4].

In recent years, the use of molecular markers to predict tumor recurrence has attracted a great deal of attention. Programmed death-ligand 1 (PD-L1), an immune checkpoint molecule, is closely associated with tumor immune escape and recurrence[5]. Tumor mutation burden (TMB) reflects the degree of genomic mutations in tumors and is strongly associated with tumor invasiveness and drug resistance[6].

Therefore, this retrospective analysis was conducted on the clinical data of patients who received small-molecule drugs in combination with chemotherapy following RG. This study aimed to explore the correlation between baseline clinical characteristics, molecular markers, and postoperative recurrence risk; identify independent risk factors for recurrence after treatment; and evaluate the clinical value of molecular markers, including PD-L1 and TMB, in predicting recurrence risk. The results are summarized as follows.

MATERIALS AND METHODS

General data

We retrospectively analyzed the clinical data of 136 patients with GC who received small-molecule drugs combined with chemotherapy after surgery at the Affiliated Hospital of North Sichuan Medical College. The clinical data of patients with postoperative GC treated at our hospital between September 2023 and January 2024 were retrospectively analyzed. Follow-up was conducted until January 2025, with a minimum follow-up duration of 12 months for all enrolled patients. According to the recurrence status, patients were divided into a recurrence group (38 cases) and a non-recurrence group (98 cases).

Inclusion criteria: (1) All patients underwent radical resection; (2) All received adjuvant therapy with smallmolecule drugs (apatinib, regorafenib, etc.) combined with chemotherapy postoperatively; (3) All completed at least 4 cycles of standard treatment; (4) Complete and traceable clinical data were available; and (5) Karnofsky Performance Status score ≥ 60.

Exclusion criteria: (1) Patients who had their adjuvant therapy regimen adjusted during treatment; (2) Pathological diagnosis of gastric pulmonary adenocarcinoma; (3) Complications with primary malignant tumors at other sites; and (4) Follow-up duration less than 12 months or loss to follow-up.

This retrospective study was approved by the Medical Ethics Committee of Affiliated Hospital of North Sichuan Medical College (Approval No. 2025ER715-1). The requirement for informed consent was waived due to the retrospective nature of the study.

Methods

Clinical data were obtained by reviewing medical records and accessing the follow-up system. The entire process complied with clinical research data specifications. Raw medical records and test reports were collated via double independent checking to ensure the accuracy of data collection.

Basic clinical data, including sex, age, body mass index (BMI), and demographic information were collected. Pathological reports of postoperative resected specimens were retrieved from the picture archiving and communication system of the Department of Pathology, covering primary tumor location (antrum/cardia, body/fundus), degree of differentiation (poor differentiation, moderatewell differentiation), TNM stage, and The Cancer Genome Atlas (TCGA) subtype. Clinical treatment-related information, including specific types of small-molecule drugs (apatinib, regorafenib, etc.), dosage, administration route, and treatment cycles, was collected synchronously from the physician-order system. The postoperative adjuvant treatment regimen mainly included apatinib (250-500 mg orally once daily) or regorafenib (120-160 mg orally once daily for 3 weeks, followed by 1 week off) combined with platinum-based doublet chemotherapy regimens, such as S-1 and oxaliplatin or capecitabine and oxaliplatin. Chemotherapy was administered every 3 weeks for at least 4 cycles. The dosage and treatment duration were adjusted according to patient tolerance and adverse reactions.

The detection results for the molecular markers were obtained using the laboratory information systems of the Department of Laboratory Medicine. Original detection values and standardized interpretation results of carcinoembryonic antigen (CEA), alpha-fetoprotein (AFP), PDL1, CA125, TMB, and other indicators were fully documented. Recurrence was defined as the appearance of local recurrence, regional lymph node metastasis, peritoneal dissemination, or distant metastasis confirmed by imaging [computed tomography (CT), magnetic resonance imaging, or positron emission tomography/CT], gastroscopy, or pathological biopsy during follow-up after RG.

Observation indicators

Baseline data, including sex, age, BMI, degree of differentiation, primary tumor location, TNM stage, and TCGA subtype, were compared between the two groups. The detection results of molecular markers, such as CEA, AFP, PD-L1, CA125, and TMB, were also collected and compared. Binary Logistic regression analysis was used to identify factors influencing recurrence after adjuvant therapy, and the predictive value of single and combined molecular markers for recurrence risk was evaluated.

Statistical analysis

Statistical analyses were performed using SPSS 25.0 software. Enumeration data were presented as n (%) and analyzed by the χ² test. Measurement data were expressed as mean ± SD and examined using the t-test. A multivariate analysis was performed using binary Logistic regression. Receiver operating characteristic (ROC) curves were constructed to evaluate the value of molecular markers in predicting recurrence risk. Statistical significance was set at P < 0.05.

RESULTS

Baseline characteristics of the two groups

There were no statistically significant differences between the two groups in terms of sex, BMI, primary tumor location, TCGA subtype, or other baseline characteristics (P > 0.05). However, significant differences were observed between the two groups with respect to age, degree of differentiation, TNM stage, and other baseline data (P < 0.05) as shown in Table 1.

Table 1 Basic characteristics of patients in both groups, n (%) or mean ± SD.

Baseline characteristics	Recurrence group (n = 38)	Non-recurrence group (n = 98)	χ2/t	P value
Gender			0.039	0.843
Male	21 (55.26)	56 (57.14)
Female	17 (44.74)	42 (42.86)
Age (year)			7.742	0.005
≥ 60	26 (68.42)	41 (41.84)
< 60	12 (31.58)	57 (58.16)
BMI (kg/m2)	24.16 ± 1.75	23.94 ± 1.92	0.614	0.540
Differentiation			51.193	< 0.001
Poorly differentiated	27 (71.05)	10 (10.20)
Moderately-well differentiated	11 (28.95)	88 (89.80)
Primary tumor location			0.067	0.795
Antrum/cardia	20 (52.63)	54 (55.10)
Body/fundus	18 (47.37)	44 (44.90)
TNM stage			26.668	< 0.001
III-IV	27 (71.05)	23 (23.47)
I-II	11 (28.95)	75 (76.53)
TCGA somatotype			0.434	0.664
EBV-positive type	17 (44.74)	46 (46.94)
Microsatellite unstable type	15 (39.47)	41 (41.84)
Genomic stable type	6 (15.79)	11 (11.22)

BMI: Body mass index; TCGA: The Cancer Genome Atlas; EBV: Epstein-Barr virus.

Detection results of molecular biomarkers in the two groups

There were no statistically significant differences between the two groups in the levels of molecular biomarkers such as CEA, AFP, and CA125 (P > 0.05). However, significant differences were observed in the proportion of PD-L1 positivity and TMB expression levels between the two groups (P < 0.05), as shown in Table 2.

Table 2 Comparison of molecular biomarker detection results between the two groups, n (%) or mean ± SD.

Baseline characteristics	Recurrence group (n = 38)	Non-recurrence group (n = 98)	χ2/t	P value
CEA (ng/mL)	39.65 ± 4.17	38.79 ± 4.52	1.017	0.311
AFP (ng/mL)	23.17 ± 3.92	22.83 ± 4.05	0.443	0.658
PD-L1 positivity			13.322	< 0.001
Positive	29 (76.32)	36 (40.91)
Negative	9 (23.68)	52 (59.09)
CA125 (U/mL)	30.17 ± 2.94	29.56 ± 3.18	1.025	0.307
TMB (mut/Mb)	14.51 ± 2.82	10.13 ± 1.56	16.782	< 0.001

CEA: Carcinoembryonic antigen; AFP: Alpha-fetoprotein; PD-L1: Programmed death-ligand 1; TMB: Tumor mutation burden.

Analysis of factors influencing post-treatment recurrence

Post-treatment recurrence status was used as the dependent variable (recurrence = 1; no recurrence = 0). Variables with P < 0.05 in the comparisons of baseline characteristics and molecular biomarker results between the two groups were included as independent variables in a binary Logistic regression analysis. The variable assignments are presented in Table 3. The results demonstrated that age (≥ 60 years), poor differentiation, TNM stage III-IV, PD-L1 positivity, and high TMB expression were independent risk factors for postoperative recurrence (P < 0.05; Table 4).

Table 3 Assignment status.

Variable	Assignment
Age	≥ 60 years = 1; < 60 years = 0
Differentiation	Poorly differentiated = 1; moderately-well differentiated = 0
TNM stage	Stage III-IV = 1; stage I-II = 0
PD-L1 positivity	Positive = 1; negative = 0

PD-L1: Programmed death-ligand 1.

Table 4 Analysis of factors influencing disease recurrence after treatment.

Factor	B	SE	Wald χ2	P value	OR	95%CI
Age	1.140	0.319	12.794	< 0.001	3.126	1.674-5.837
Differentiation	0.616	0.235	6.840	0.009	1.852	1.167-2.939
TNM stage	1.391	0.493	7.948	0.005	4.109	1.528-10.571
PD-L1 positivity	1.820	0.451	16.295	< 0.001	6.174	2.551-14.942
TMB	0.487	0.163	8.962	0.003	1.627	1.183-2.238

PD-L1: Programmed death-ligand 1; TMB: Tumor mutation burden.

Predictive value of molecular biomarkers for recurrence

The area under the curve (AUC) of PD-L1 positivity for predicting recurrence was 0.698, while the AUC of TMB for predicting recurrence was 0.934. The combined detection of these two molecular biomarkers achieved an AUC of 0.983 for predicting recurrence after adjuvant treatment with small-molecule targeted drugs combined with chemotherapy in GC. In addition, the sensitivity and specificity of combined detection were higher than those of either biomarker alone (Table 5 and Figure 1).

Figure 1 Value of molecular biomarkers in relapse prediction. TMB: Tumor mutation burden; PD-L1: Programmed death-ligand 1.

Table 5 Values of molecular markers for recurrence prediction.

Variable	AUC	95%CI	Sensitivity	Specificity	Cut-off value
PD-L1 positivity	0.698	0.613-0.774	76.3	63.3	-
TMB	0.934	0.878-0.969	78.95	95.92	≥ 12.37
Combined detection	0.983	0.945-0.998	94.97	98.98	-

PD-L1: Programmed death-ligand 1; TMB: Tumor mutation burden.

DISCUSSION

GC is among the most common malignant tumors of the digestive tract. Global cancer statistics indicate that there are more than one million new cases of GC annually, with approximately 650000 deaths each year[7]. As a country with a high incidence of GC, China accounts for more than 40% of new cases globally each year, and GC-related mortality ranks second among all cancer-related deaths in China[8]. In the present study, postoperative GC patients with recurrence exhibited a significantly higher proportion of elderly age, poor differentiation, advanced TNM stage, PD-L1 positivity, and elevated TMB levels. Multivariate analysis further confirmed that these variables were independent predictors of recurrence risk.

Because the initial symptoms of GC are often subtle, most patients present in the early stages with nonspecific symptoms such as upper abdominal discomfort, acid reflux, and belching. This results in a low rate of early diagnoses. Consequently, most patients are diagnosed in the intermediate or advanced stages. Furthermore, data have shown that the postoperative recurrence rate of GC remains high, reaching 30%-50% in patients with advanced-stage disease after five years[9].

Currently, surgery is the primary treatment for GC, often combined with multidisciplinary therapy. RG is a core therapeutic approach that enables complete resection of the primary tumor, also standardized dissection of regional lymph nodes and removal of invading adjacent tissues. This eliminates most of the tumor burden in the body[10].

Numerous clinical studies have confirmed that patients who undergo standardized radical surgery achieve significant long-term survival benefits, whereas those who do not undergo radical surgery have a five-year survival rate of less than 20%[11,12]. Simultaneously, radical surgery can effectively treat tumor-related symptoms such as bleeding, obstruction, and pain. This creates a favorable physical condition for subsequent adjuvant therapy.

Although radical surgery can remove most tumor tissue, patients with advanced GC may still harbor residual microscopic lesions and potential micrometastases after surgery. These residual microscopic lesions and potential micrometastases are the main causes of postoperative recurrence[13]. It is vital to use small-molecule targeted drugs in combination with chemotherapy as adjuvant therapy after surgery. This is a key strategy to reduce recurrence rates and improve patient prognosis.

The 2022 NCCN guidelines for treating GC state that patients with stage III or above disease that has spread to the lymph nodes or veins have a 50%-70% chance of the cancer recurrence after surgery. It has been established that surgery alone cannot eradicate all cancer cells, and that such patients are at a high risk of recurrence; therefore, routine adjuvant therapy is given[14]. The risk of recurrence can be reduced to a certain extent by single-agent chemotherapy in its traditional form; however, it has some limitations. These include an inability to specifically target tumor cells and a high rate of drug resistance. On the other hand, small-molecule agents like apatinib and regorafenib can selectively stop tumor growth by blocking new blood vessel formation, while also inhibiting cancer cells from spreading. The synergistic effects of chemotherapeutic drugs may be exerted by these agents, thereby improving the eradication rate of residual cancer cells with a relatively low risk of resistance[15]. Clinical studies have demonstrated that patients with postoperative GC who receive adjuvant therapy combining small-molecule drugs and chemotherapy can achieve recurrence rates below 30%, while also extending disease-free survival[16]. Additionally, most small-molecule drugs are administered orally, which is associated with better treatment compliance and good applicability for long-term therapy in patients with physical weakness.

In the present analysis of 136 postoperative patients with GC who received adjuvant treatment with small-molecule drugs combined with chemotherapy, recurrence occurred in 38 patients after treatment, suggesting that a certain risk of recurrence still exists despite postoperative adjuvant therapy. Further analysis of baseline characteristics and molecular biomarkers related to recurrence revealed that age, degree of differentiation, TNM stage, PD-L1 positivity, and TMB expression levels were all important factors influencing recurrence after treatment.

From the perspective of age, elderly patients aged 60 years or above generally experience a decline in immune function and organ performance, resulting in reduced T cell proliferation and differentiation capacity, as well as decreased natural killer cell activity. Furthermore, elderly patients frequently have multiple health issues, such as diabetes and high blood pressure, in addition to their primary disease. The body's resistance to tumor cells is further weakened by these underlying diseases, which continuously disrupt the internal microenvironment and facilitate tumor proliferation and dissemination. Regarding clinical treatment, elderly patients have a lower tolerance for combined therapy involving small-molecule drugs and chemotherapy than younger patients. Reduced drug metabolism and excretion are also consequences of age-related hepatic and renal dysfunction. Therefore, it is often necessary to reduce the dose or shorten the chemotherapy cycle to manage these adverse reactions. This may result in insufficient treatment intensity and an increased risk of recurrence[17].

The study also established that inadequate differentiation was an independent risk factor for disease recurrence post-treatment. Poorly differentiated GC is characterized by cells that are very different from normal cells and are not tightly packed together. This makes it easier for tumor cells to break through the basement membrane of the first tumor and spread to nearby tissues. These cells are more likely to enter the blood and lymphatic systems, thereby increasing the risk of recurrence after treatment. Clinical reports have shown that even after R0 radical resection, surgical removal of microscopic metastatic lesions is difficult, and the effectiveness of small-molecule drugs combined with chemotherapy against these highly invasive cells is limited[18]. Studies have revealed that poorly differentiated GC cells have reduced genomic stability and are more susceptible to genetic mutations. This means that tumor cells are less likely to bind to small-molecule targeted agents and is less sensitive to DNA damage caused by chemotherapeutic drugs. This creates a significant risk of recurrence[19]. Other clinical data have shown that patients with poorly differentiated GC have a higher tumor proliferation index, with recurrence peaking within 1-2 years after surgery. Cancer cells proliferate much faster than moderately or well-differentiated tumors. Although small-molecule drugs combined with chemotherapy may suppress tumor growth to some extent, they cannot completely block the rapid proliferation and spread of these cells, resulting in a relatively high risk of recurrence after surgery[20].

The presence of TNM stages III-IV disease has been shown to increase the risk of recurrence after surgery. This is because advanced TNM stage is a key pathological factor contributing to recurrence. Clinical practice has shown that stage I-II gastric tumors can often be completely removed via radical surgery. However, stage III GC is characterized by extensive regional lymph node metastasis, and stage IV GC has already developed into distant micrometastases. In such cases, radical surgery can only remove the main lesion and regional lymph nodes. It cannot eliminate microscopic metastatic lesions in the blood or lymphatic system. This means that patients with stage III-IV disease are more likely to experience disease[21]. With regard to treatment limitations, small-molecule drugs combined with chemotherapy can inhibit tumor angiogenesis, and the tumor microenvironment in stage III-IV patients has already undergone remodelling. The increased presence of immunosuppressive cells further weakens the immune-mediated cytotoxic effects of the combined regimen, thereby increasing the risk of postoperative recurrence in patients with TNM stages III-IV disease.

Comparative analyses of CEA, AFP, PD-L1, CA125, and TMB levels with respect to molecular biomarkers were conducted. The results showed that PD-L1 positivity and high TMB expression are independent risk factors for postoperative recurrence. PD-L1 is an important immune checkpoint molecule that is expressed on the surface of tumor cells. Under typical circumstances, T cells are capable of identifying tumor antigens and eliminating cancerous cells. However, when PD-L1 positive tumor cells bind to the PD-1 receptor on T cells, they inhibit T-cell activation and proliferation, which makes the tumor difficult to spot. Consequently, residual tumor cells remaining after surgery can survive and proliferate despite immune surveillance[22]. Clinical studies have confirmed that the effectiveness of treatment is reduced by PD-L1 positivity. Although they can inhibit tumor angiogenesis, small-molecule drugs cannot improve the immunosuppressive microenvironment mediated by PD-L1[23]. Furthermore, although chemotherapeutic agents can kill tumor cells, they cannot reverse the PD-L1-induced immune escape. Other clinical studies have shown that patients with PD-L1-positive expression develop resistance to combined small-molecule therapy and chemotherapy more quickly and that residual cancer cells proliferate more rapidly. These mechanisms are the main reasons for the higher recurrence rates observed after treatment in PD-L1-positive patients[24].

The TMB is used to measure the number of mutations in the tumor genome. High TMB expression indicates numerous genetic mutations in tumor cells and can result in significantly increased tumor heterogeneity[25]. In terms of treatment responsiveness, GC cells with high TMB expression are more likely to exhibit abnormal activation of signalling pathways, which further accelerates tumor cell proliferation. Therefore, treatment with a combination of small-molecule drugs and chemotherapy may fail to completely inhibit tumor cell growth and spread. Furthermore, high TMB expression is strongly associated with tumor immune evasion. Although more neoantigens can be generated and immune recognition can be facilitated, in theory, by high TMB, the synergistic immune effects of combined small-molecule therapy and chemotherapy can be offset by PD-L1 overexpression in some high-TMB tumors, thereby resulting in the evasion of immune destruction.

The predictive value of detecting molecular biomarkers for recurrence risk after treatment with small-molecule drugs combined with chemotherapy in GC was clarified using ROC curve analysis. The results revealed variations in the predictive performance of different biomarkers, with the combined detection of PD-L1 positivity and TMB demonstrating the highest predictive efficacy for recurrence risk. In terms of the predictive performance of individual biomarkers, the AUC for PD-L1 positivity was 0.698, indicating a moderate predictive ability. One possible explanation for this is that PD-L1 promotes recurrence by mediating tumor immune escape. However, PD-L1 expression in GC is highly heterogeneous, with some minimally invasive tumors showing false-positive results for low PD-L1 expression. Thus, PD-L1 expression is an imprecise standalone predictor of recurrence risk.

TMB demonstrated excellent predictive performance for recurrence, with an AUC of 0.934. This may be due to the fact that elevated TMB expression frequently signifies a substantial tumor mutational burden. Consequently, tumor cells are more likely to develop drug resistance. Genetic mutations also increase the likelihood of cell proliferation and metastasis. The best way to predict the response to therapy is to test PD-L1 and TMB levels together. It had an AUC of 0.983. It also exhibited high sensitivity and specificity. From a mechanical point of view, PD-L1 shows how a tumor mediates immune system, whereas TMB shows how much the tumor has changed and how it might spread. A combined evaluation provides a more exhaustive appraisal of the biological behavior of a tumor and a more dependable foundation for predicting the risk of postoperative recurrence in patients with GC.

CONCLUSION

PD-L1 positivity and elevated TMB are independent predictors of postoperative recurrence in patients with GC receiving adjuvant small-molecule targeted therapy combined with chemotherapy. The combined detection of these molecular biomarkers demonstrated excellent predictive value and may contribute to the development of individualized postoperative surveillance and treatment strategies.