Construction of clinical evaluation index system for first-line ICIs in HER-2-negative advanced gastric cancer using Delphi-AHP

Abstract

To develop a scientific, objective, and quantitative clinical comprehensive evaluation index system for the first-line treatment of HER-2-negative advanced gastric cancer with immune checkpoint inhibitors, to provide a reference index for the clinical comprehensive evaluation of similar drugs in clinical settings, and to inform decision-making regarding the rational clinical use of medication and directory access in hospitals. A preliminary indicator pool was established based on the literature analysis method, and the final indicator system and weights were clarified by combining the use of two rounds of the Delphi expert consultation method and analysis hierarchical process (AHP). After expert consultation and validation, a comprehensive evaluation system of four primary indicators, 10 secondary indicators, and 18 tertiary indicators for the first-line treatment of HER-2-negative advanced gastric cancer with immune checkpoint inhibitors was finalized. The primary indicators were effectiveness (47%), safety (34%), economy (10%), and pharmacological properties (9%). The top three weights of the combination of secondary indicators were adverse reactions (26.8%), clinical efficacy (20.8%), and indications (15.6%), respectively. The top-ranked secondary indicators among the economic and pharmacological characteristics indicators were whether they were included in the national essential drug list (3.9%) and whether their pharmacological effects were clear (6.3%). The weights of the combination of tertiary indicators ranged from 0.4% to 15.3%, with the highest weights being the incidence of CTCAE grade ≥ 3 adverse events under the safety adverse reaction indicator (15.3%) and the incidence of adverse events leading to treatment interruption (11.6%). The comprehensive evaluation system based on the Delphi method and AHP for constructing the first-line treatment of HER-2-negative advanced gastric cancer with immune checkpoint inhibitors has a high degree of authority, scientificity and reliability. In view of the differences in drug attributes, disease characteristics, and medication populations, it is necessary to carry out a comprehensive clinical evaluation based on these three to meet the needs of the current medication.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-50727-4.

Introduction

Gastric cancer (GC) is a malignant epithelial neoplasm that originates in the stomach. It is the fifth most common cancer worldwide and the fourth leading cause of cancer-related mortality. Approximately 1.2 million new cases of gastric cancer are diagnosed globally each year, with China accounting for around 40%¹. Among patients with gastric cancer, 80–85% are HER-2 negative, and those with advanced disease have a particularly poor prognosis, with a five-year survival rate below 10%^2–4. According to The Cancer Genome Atlas Program (TCGA ) classification, molecular subtype analysis within the HER-2 negative population reveals substantial heterogeneity: the Epstein-Barr virus (EBV) -positive subtype comprises 8–10%, the microsatellite instability-high (MSI-H) subtype accounts for 15–20%, the chromosomal instability (CIN) subtype represents 50%, and the genomically stable (GS) subtype constitutes 20%. This classification underscores the heterogeneity within this disease category^5,6. Current treatments for gastric cancer include chemotherapy, molecular targeted therapies, and immune checkpoint inhibitors. Among these, chemotherapy regimens are supported by robust evidence from evidence-based medicine and extensive clinical practice. Pivotal trials such as CheckMate-649 (the KEYNOTE-859 China subgroup) have demonstrated marked survival improvements with the addition of immunotherapy to chemotherapy, extending survival from 8 to 10 months in the chemotherapy era to over 20 months in patients with high PD-L1 expression^7,8. Accordingly, immune checkpoint inhibitors combined with chemotherapy have become the standard first-line treatment for patients with HER-2 negative, unresectable, locally advanced, or metastatic gastroesophageal junction or gastric cancer^9–11. Moreover, in the nivolumab cohort, the three-year overall survival (OS) rate exceeds 20%, and a subset of patients may achieve what can be described as a “clinical cure.” Improved response and disease control rates have also made surgical intervention feasible for selected patients with advanced disease, offering the possibility of long-term survival. Beyond single-target immune checkpoint inhibitors, the dual-target immunotherapeutic drug cardunculizumab has been approved as first-line treatment for patients with locally advanced unresectable or metastatic gastric or gastroesophageal (G/GEJ) adenocarcinoma. In addition, phase III studies of PD-L1/TGF-β bifunctional fusion protein antibodies (e.g., SHR-1701) in combination with chemotherapy have shown substantial progress, further accelerating the development of new targets and mechanisms of action^12–14.However, a broad range of immune checkpoint inhibitor (ICI) monoclonal antibodies has been approved, and notable differences in baseline characteristics and disease features across trial populations contribute to considerable variability in efficacy and in the stratification of survival benefit. Among Chinese cancer patients receiving ICI treatment, the overall incidence of immune-related adverse events (irAEs) is higher than that reported in Europe and the United States, with particularly increased risks of hepatotoxicity and pneumonitis^8,15,16. Additionally, the high cost of immunotherapy, asymmetric domestic medical insurance coverage, and limited testing capacity at primary healthcare facilities have further intensified patients’ financial burden and treatment-related risks^17,18. Addressing the urgent need to identify the optimal ICI for complex clinical scenarios to maximize patient benefit remains critical. This study employed the Delphi expert consultation method in conjunction with the analytic hierarchy process (AHP) to construct a scientifically rigorous, objective, and quantitative comprehensive evaluation system for first-line immune checkpoint inhibitors in treating HER-2 negative advanced gastric cancer. This system defines evaluation indicators for the comprehensive clinical assessment of comparable drugs in routine practice and provides references for rational drug use and formulary inclusion decisions in hospitals, thereby supporting precise, stratified treatment of HER-2 negative gastric cancer with ICIs.

Objects and methods

Establish preliminary evaluation indicators

The “Management Guidelines for Comprehensive Clinical Evaluation of Drugs (2024 Trial)” and the “Technical Guidelines for Comprehensive Clinical Evaluation of Antineoplastic Drugs (2022 Edition)” provide a structured framework for the clinical comprehensive evaluation of immune checkpoint inhibitors in the first-line treatment of HER-2-negative advanced gastric cancer. This framework is organized into six key dimensions: safety, efficacy, economy, innovation, appropriateness, and accessibility. By organizing authoritative guidelines, searching journal databases, and reviewing drug information and policies and regulations issued by the U.S. Food and Drug Administration (FDA), National Medical Products Administration (NMPA), and European Medicines Agency (EMA), we have developed comprehensive evaluation evidence. The initial pool of indicators for preselection was developed through the above evidence-based research and collaborative discussions among project team members.

Delphi method

To accurately assess the effectiveness, the selection of evaluation indicators should adhere to the principles of operability, independence, and applicability. This study employed the Delphi method to screen evaluation indicators¹⁹, as illustrated in Fig. 1. The initial evaluation criteria ultimately determine retention based on the outcomes of expert assessments. To enhance data reliability and protect confidentiality, experts were selected based on their experience, expertise, and familiarity with the study domain; responses were collected anonymously to facilitate independent judgment; and only aggregated summaries of panel feedback were provided between rounds, without disclosure of any individual expert’s identity.

Fig. 1.

The process of Delphi (A) and AHP (B).

Selection of experts

The selection of experts adhered to principles of authority and representativeness. The inclusion criteria, as depicted in Fig. 2, were as follows: ① The expert group comprised clinicians, clinical pharmacists, health technology assessors, or health administration experts; ② They possessed 5 to 40 years of professional experience, held intermediate or higher titles, and were recognized as experts in their respective fields; and ③ The total number of experts ranged from 30 to 40.

Fig. 2.

The basic information of the Delphi experts regarding region (A), work experience (B), professional title (C), and specialty (D).

Design expert letter inquiry form and indicators assignment

The questionnaire was self-developed by the project design team for the purposes of this study. This study employed a generalized Likert scale to assess the significance of primary, secondary, and tertiary indicators, as evaluated by experts on a 5-point scale. The experts’ ratings categorized the indicators as very important (5 points), relatively important (4 points), generally important (3 points), unimportant (2 points), or very unimportant (1 point). The degree of expert authority is determined by the foundation and familiarity of their judgments regarding various indicators, which encompass practical experience, theoretical knowledge, references to domestic and international materials, and intuitive insights from supervisors. These factors are classified into three levels—large, medium, and small—based on their influence, facilitating a quantitative assessment. Furthermore, experts’ familiarity with all indicators across five dimensions—safety, efficacy, economy, pharmacological properties, and other attributes—was evaluated and categorized as familiar, more familiar, general, not too familiar, and unknown, with corresponding values of 0.9, 0.7, 0.5, 0.3, and 0.1, respectively. No separate pilot survey was conducted. Instead, the questionnaire underwent iterative revision and optimization through two rounds of Delphi expert consultation. Multidisciplinary experts assessed the content of the items/indicators during the Delphi process to support content validity and ensure the scientific relevance and appropriateness of the questionnaire items. The questionnaires were distributed and collected via an online survey platform (https://www.wjx.cn).

Data analysis

Data for scoring were obtained using a questionnaire, and the results from the pre-selected indicator pool were quantified using SPSS and Excel software. The quantitative indicators comprised the coefficient of expert positivity, concentration of expert opinions, degree of expert coordination, and coefficient of expert authority (Cr). The variable Cr represents the mean value of the expert’s familiarity with the indicator (Cs), and the coefficient of the expert’s judgmental basis for the indicator (Ca). The criterion for reliability of the consulting results was defined as Cr ≥ 0.7. The positive coefficient of experts, represented by the return rate of the questionnaire (return rate = total number of returns / total number of distributions), indicates the level of interest of experts in the consultation process. The arithmetic mean () served as an indicator of the concentration of expert opinions. A higher value signifies greater significance of the j_th indicator within the indicator system. The degree of coordination among expert opinions was evaluated using coefficient of variation (CV) and Kendall’s coefficient of coordination (W). A smaller CV indicates a higher degree of coordination among experts, with a generally accepted threshold of CV < 0.25. The W value ranges from 0 to 1, where a larger value signifies better coordination among experts, and a smaller value indicates poorer coordination. The χ2 test was applied to W, and if P < 0.05, it was concluded that the experts’ opinions exhibited consistency, thereby enhancing the credibility of the assessment and rendering the evaluation results favorable. In this study, indices are selected based on the criteria that they meet both an importance mean of ≥ 3.5 and CV < 0.25, while also incorporating expert opinions in the screening process.

Analytic hierarchy process (AHP)

The Analytic Hierarchy Process (AHP) is a multi-objective decision analysis method that integrates both qualitative and quantitative approaches²⁰. It employs pairwise comparisons to construct a judgment matrix that is used to derive single-level indicator weights. Subsequently, the total ranking of indicators across levels was calculated to determine the relative weight of each indicator relative to the overall indicator. The consistency test was considered successful when the random consistency ratio (CR) was < 0.10. The specific process and steps are illustrated in Supplementary (Method, The AHP statistical workflow). As a robustness check, we conducted a brief one-way sensitivity analysis on the aggregated (group) AHP pairwise-comparison matrices (geometric-mean aggregation). Each element was perturbed by ± 10% while preserving reciprocity (), and priorities were recomputed using the geometric-mean (root) method to assess the stability of weights and rank ordering.

Ethics approval and informed consent

This study involved expert consultation using Delphi questionnaires. The study protocol was reviewed and approved by the Ethics Committee of the Second Affiliated Hospital, Zhejiang University School of Medicine (approved on 14 April 2025; Approval No.: 20250655). All participating experts were informed about the study objectives and procedures prior to participation, and informed consent was obtained before completion of the questionnaires. The study procedures were conducted in accordance with applicable ethical guidelines and regulations (including the Declaration of Helsinki and relevant national requirements).

Results

Indicator preselection

In the context of the rational use of medicines in clinical practice, drawing upon the “Management Guidelines for Comprehensive Clinical Evaluation of Pharmaceuticals (2024 Edition for Trial Implementation)” and the “Technical Guidelines for Comprehensive Clinical Evaluation of Antitumor Drugs, 2022 Edition,” as well as a systematic analysis of the literature, a framework for an evaluation index system has been developed. This framework comprises five primary indices, 13 secondary indices, and twenty-nine tertiary indexes, encompassing aspects of safety, efficacy, economy, pharmacological properties, and other relevant attributes (see Table 1).

Table 1.

The framework for an evaluation index system has been developed.

Primary indicator	Secondary indicator	Tier-3 indicator
1 Safety	1.1 Adverse reactions	1.1.1 Incidence of adverse events (Common Terminology Criteria for Adverse Events (CTCAE) grade ≥ 3)
		1.1.2 Incidence of treatment interruption due to adverse events
	1.2 Special Populations	1.2.1 Usable in Elderly Population
		1.2.2 People with impaired liver function
		1.2.3 People with renal insufficiency
2 Effectiveness	2.1 Indications	2.1.1Approved by NMPA
		2.1.2Approved by FDA/EMA
		2.1.3 Limited to PD-1 Expression Population
	2.2 Clinical Efficacy	2.2.1Overall Survival (OS)
		2.2.2Progression-Free Survival (PFS)
		2.2.3Objective Response Rate (ORR) or Disease Control Rate (DCR)
	2.3 Clinical Guideline Recommendations	2.3.1Guideline Recommendation Level
3 Economy	3.1 Minimum Cost Analysis (Daily Treatment Costa)	3.1.1% ≤ 30%
		3.1.2% > 30% to 70%
		3.1.3% > 70% to 100%
	3.2 Inclusion in National Essential Drug List	3.2.1None
	3.3 Inclusion in National Reimbursement Drug List	3.3.1National Reimbursement Drug List - Category A
		3.3.2National Reimbursement Drug List - Category B
		3.3.3Reimbursable via Other Channels (Specific Disease Categories, Other Commercial Supplemental Insurance, etc.)
4 Pharmaceutical Properties	4.1 Clarity of Pharmacological Action	4.1.1 None
	4.2 Physicochemical Stability of Final Solutionb	4.2.1 Stable at 2–8 °C for ≤ 24 h
		4.2.2 Stable at 2–8 °C for > 24 h
5 Other Attributes	5.1 Innovativeness	5.1.1 Mechanism Innovation
		5.1.2 Original Research Drug / Improved New Drug
	5.2 Global Marketing Status	5.2.1 Marketed Domestically and Internationally
		5.2.2 Marketed Domestically, Not Marketed Internationally
	5.3 Drug Shelf Life	5.3.1 ≥ 36 months
		5.3.2 24 to < 36 months
		5.3.3 < 24 months

(a) The percentile calculation of daily treatment cost was performed by anchoring the price of the drug with the highest daily cost at 100%, sequentially deriving the corresponding values for the 70th and 30th percentiles, and then categorizing each drug into the respective percentile interval based on its price. (b) If the diluted solution cannot be used immediately, it remains physiochemically stable under refrigerated storage conditions (2–8 °C).

Basic information of the Delphi experts

The study involved 41 antitumor experts from seven provinces, 16 regions, and 21 tertiary hospitals to conduct consultations (Table S1). The ratio of medical experts was approximately 1:1, with 83% possessing more than 10 years of professional experience and 83% holding senior or higher titles. The overall academic level of the experts was high, and they had substantial experience in antitumor treatment, conferring a certain degree of authority and representativeness. Basic information regarding the experts is presented in Fig. 2.

The coefficient of expert authority (Cr)

In the initial phase, 41 questionnaires were distributed, followed by 27 in the subsequent phase. The response rate for both phases was 100%, with effective response rates of 95% and 100%. The experts’ positive coefficient was 100%. The Cs and Ca values for the questionnaire experts were 0.96 and 0.95, respectively, and the calculated Cr was 0.955 (rounded to 0.96), indicating that the experts’ opinions were both authoritative and credible.

Compile & analyze the degree of coordination among expert opinions

The results indicated a progressive reduction in the CV range for the indicators across all levels for each successive round (Fig. 3). W surpassed 0.3, indicating a moderate level of agreement among experts’ ratings. Furthermore, the P-values associated with W in each round were below 0.05, suggesting that the observed agreement was statistically significant. Indicator validity was supported by the Delphi-based content development and screening process (importance mean and CV thresholds) and by the experts’ authority assessment (Cr).

Fig. 3.

Delphi expert consensus analysis dashboard.

Delphi method expert consultation results

The selection of indicators was predicated on screening criteria of > 3.5 and CV < 0.25, with subsequent revisions informed by expert opinions (Table S2). Following the initial round of results, the tertiary indicator “limited PD-1-expressing population” within the “efficacy-indications” dimension was eliminated. The tertiary indicator under “efficacy-clinical guidelines” was refined to “guideline recommendation level,” further specified as “national diagnostic and therapeutic standard recommendation/clinical guideline recommendation level.” This refinement included distinctions such as “national diagnosis and treatment standard recommendation/clinical diagnosis and treatment pathway recommendation,” “CSCO/NCCN guideline level I recommendation,” and “CSCO/NCCN guideline level II and below,” as well as “expert consensus recommendation.” In addition, the ABC level of the recommended evidence was differentiated within the scoring rules. The primary indicator “other attributes” and all associated secondary and tertiary indicators were removed. The tertiary indicator under “Least Cost Analysis” was replaced with a differentiation of the average daily treatment cost of the drug within the scoring rules. After the first round of indicator optimization, there were four primary indicators, 10 secondary indicators, and 18 tertiary indicators. The revised questionnaire on the indicators was redistributed to experts, and an analysis of the statistical results indicated that the indicators met the retention criteria, with no further contentious opinions from the experts. Consequently, these indicators were retained in their entirety (Table 3).

Table 3.

Index system weight vector. The combined weights reflect multiplicative calculations (primary × Secondary or Primary×Secondary×Tier-3.

Primary indicator	Weight	Secondary indicator	Weight	Combined weight	Tier-3 indicator	Weight	Combined weight
1 Safety	0.34	1.1 Adverse reactions	0.8	0.268	1.1.1 Incidence of adverse events (CTCAE grade ≥ 3)	0.5695	0.153
					1.1.2 Incidence of treatment interruption due to adverse events	0.4305	0.116
		1.2 Special Populations	0.20	0.067	1.2.1 Usable in Elderly Population	0.3148	0.021
					1.2.2 People with impaired liver function	0.345	0.023
					1.2.3 People with renal insufficiency	0.3401	0.023
2 Effectiveness	0.47	2.1 Indications	0.33	0.156	2.1.1 Approved by NMPA	0.557	0.087
					2.1.2 Approved by FDA/EMA	0.2629	0.041
					2.1.3 Approved in Japan and Some Regions	0.1801	0.028
		2.2 Clinical Efficacy	0.44	0.208	2.2.1 Overall Survival (OS)	0.5452	0.114
					2.2.2 Progression-Free Survival (PFS)	0.3187	0.066
					2.2.3 Objective Response Rate (ORR) or Disease Control Rate (DCR)	0.1361	0.028
		2.3 Clinical Guideline Recommendations	0.23	0.110	2.3.1 Recommendation in Nationally Recognized Clinical Diagnosis/Treatment Pathways	0.3914	0.043
					2.3.2 Recommendation in CSCO/NCCN Guidelines	0.3819	0.042
					2.3.3 Conditional Recommendation in CSCO/NCCN Guidelines	0.1563	0.017
					2.3.4 Expert Consensus Recommendation Only	0.0704	0.008
3 Economy	0.10	3.1 Minimum Cost Analysis (Daily Treatment Cost)	0.25	0.025	3.1.1 None
		3.2 Inclusion in National Essential Drug List	0.39	0.039	3.2.1 None
		3.3 Inclusion in National Reimbursement Drug List	0.36	0.036	3.3.1 National Reimbursement Drug List - Category A	0.5515	0.020
					3.3.2 National Reimbursement Drug List - Category B	0.3326	0.012
					3.3.3 Reimbursable via Other Channels (Specific Disease Categories, Other Commercial Supplemental Insurance, etc.)	0.1159	0.004
4 Pharmaceutical Properties	0.09	4.1 Clarity of Pharmacological Action	0.70	0.063	4.1.1 None
		4.2 Physicochemical Stability of Final Solution	0.30	0.027	4.2.1 None

Assessing the weight of the index system utilizing the analytic hierarchy process

In this study, we developed a questionnaire titled “Comprehensive Evaluation Index System for First-Line Treatment of HER-2 Negative Advanced Gastric Cancer with Immune Checkpoint Inhibitors” to assess the relative importance of the two indices within each index dimension (Table 2). The questionnaire was distributed to authoritative experts who had previously participated in the expert consultation questionnaire. A total of 13 questionnaires were distributed, 11 of which were returned, resulting in a response rate of 84.6%. After confirming that the judgment matrices of the 11 experts satisfied the consistency requirements (CR < 0.1), the weights of the indicators at each level were calculated and normalized. The scores were assigned using the percentage method, and the results were listed in Table 3. The primary indicators were effectiveness (47%), safety (34%), economy (10%), and pharmacological properties (9%). The top three weights of the combination of secondary indicators were adverse reactions (26.8%), clinical efficacy (20.8%), and indications (15.6%), respectively. The top-ranked secondary indicators among the economic and pharmacological characteristics indicators were whether they were included in the national essential drug list (3.9%) and whether their pharmacological effects were clear (6.3%). The weights of the combination of tertiary indicators ranged from 0.4% to 15.3%, with the highest weights being the incidence of CTCAE grade ≥ 3 adverse events under the safety adverse reaction indicator (15.3%) and the incidence of adverse events leading to treatment interruption (11.6%). Besides, the results of the one-way sensitivity analysis were listed in Table S3. The primary-level and all secondary-level rankings remained unchanged across all perturbation scenarios; limited rank reversals were observed only in two tertiary matrices and where the top-two weights were nearly tied (very small margins), consistent with expected AHP behavior under near-tie conditions. To facilitate real-world implementation, we provide a detailed, study-specific scoring rubric for the comprehensive evaluation of the drugs assessed in this work (Table S4). This rubric operationalizes the proposed indicator framework by specifying clear scoring rules for each indicator, enabling transparent, reproducible calculation of overall scores to support hospital formulary (directory access) and related decision-making.

Table 2.

Fundamental scale for pairwise comparisons in the judgment matrix.

Scale	Evaluation rule
1	Equal importance of both elements
3	Moderate importance of one element over another
5	Strong importance of one element over another
7	Very strong importance of one element over another
9	Extreme importance of one element over another
2, 4, 6, 8	Intermediate values between adjacent scales
Reciprocal	If element i has weight ⍺ vs. j, then j has weight 1/⍺ vs. i

Discussion

The Analytic Hierarchy Process (AHP) and Delphi methods are widely used for the clinical evaluation of antitumor drugs, and their integration enables a structured workflow from “qualitative consensus” to “quantitative weighting,” which is well suited to decisions that must balance scientific evidence with clinical practicality. To avoid diluting the influence of key criteria by introducing an excessive number of indicators, we constructed the final framework using four overarching dimensions—safety, efficacy, economy, and pharmacological characteristics—which together capture the core considerations that routinely drive clinical and pharmacy decision-making. The resulting weights indicate that efficacy and safety dominate the evaluation, consistent with fundamental requirements for clinical use, whereas economy and pharmacological characteristics currently contribute less; however, their relative importance may change as precision medicine and health economics continue to develop and receive greater clinical attention. For operationalization in practice, the framework should be implemented with explicit, evidence-informed scoring rules tailored to real-world decision contexts—for example, incorporating PD-L1 expression requirements within approved indications and specifying reimbursement or payment restrictions (e.g., Class A vs. Class B medical insurance) under the economic dimension—so that each drug can be assigned a transparent and reproducible comprehensive score based on expert consensus and survey evidence.

There are several limitations to this study. First, the evidence supporting each indicator was primarily derived from the published literature, with many key sources being phase III randomized controlled trials (e.g., CheckMate-649). This may introduce population-selection bias (e.g., enrollment of patients with better performance status), potentially leading to an overestimation of real-world effectiveness. Second, the network meta-analysis lacked direct head-to-head comparisons; although we prioritized studies with larger sample sizes and higher levels of evidence during evidence retrieval, residual bias cannot be fully excluded. Third, the expert panel composition may have influenced indicator selection and weighting: experts were mainly recruited from tertiary hospitals and were geographically concentrated in East China (and largely in economically active, relatively healthcare-developed areas in China’s eastern/southern regions). Moreover, experts with backgrounds most directly relevant to health economics, pharmacoeconomics, and health management/policy accounted for a relatively small proportion of the panel, which may have reduced the emphasis placed on economic and policy-related indicators and could partially explain the comparatively lower weight assigned to the economic dimension. Despite these limitations, the study provides a structured and operational framework that can be further validated and refined in broader, more geographically diverse panels with greater representation from pharmacoeconomics and health-management disciplines.

Conclusions

The study integrated qualitative and quantitative methods to develop a four-dimensional comprehensive evaluation system for immune checkpoint inhibitors in the first-line treatment of HER-2-negative advanced gastric cancer. This system reflects the clinical value of drugs from multiple perspectives, thereby strengthening the foundation for guiding clinical applications, hospital drug admission, and facilitating the scientific and rational selection of drugs. Given variations in drug attributes, disease characteristics, and patient populations, comprehensive clinical evaluation methods for drugs require continuous innovation and improvement. Decision-makers can select appropriate evaluation dimensions, indicators, and methods according to their decision-making objectives. Ultimately, through a scientific evaluation system, the maximization of efficacy, minimization of toxicity, and rationalization of resources can be achieved, promoting the transition of immunotherapy for gastric cancer from “broad-spectrum application” to “precise stratification.”

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Made substantial contributions to conception: Donghang Xu; Design of the study and performed data analysis and paper writing: Tiantian Wang, Difei Yao; Lingyan Yu.

Funding

This study was supported by Zhejiang pharmaceutical association (2022ZYYL05, 2020ZYY15); Zhejiang Provincial Natural Science Foundation of China (No. ZCLQ24H0901); the research project on high quality development of Hospital pharmacy, National Institute of Hospital Administration, NHC, China (No. NIHAYSZX2533); Project of Development Center for medical science and technology National Health Commission (WKZX2024CX501205).

Data availability

The datasets used and/or analyzed during the current study available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.