Predictive value of PD-L1 expression and dMMR/pMMR status for immune checkpoint inhibitor combined with chemotherapy in advanced/recurrent endometrial cancer: a meta-analysis

Abstract

Background

Advanced/recurrent endometrial cancer (EC) poses a significant therapeutic challenge. Immune checkpoint inhibitor (ICI) combined with chemotherapy (CT) represents a promising first-line approach, yet the predictive value of deficient/proficient mismatch repair (dMMR/pMMR) status and PD-L1 expression in therapeutic efficacy remains unclear. This study aimed to systematically evaluate the predictive value of dMMR/pMMR status and PD-L1 expression for progression-free survival (PFS) and overall survival (OS) in patients with advanced/recurrent EC treated with ICI + CT.

Methods

A systematic review and meta-analysis of randomized controlled trials (RCTs) from four databases (PubMed, Embase, Cochrane Library, Web of Science) was performed. PFS and OS were synthesized as hazard ratios (HRs) with 95% confidence intervals (CIs) using Review Manager 5.4.

Results

Five RCTs (2,707 patients) were included in this meta-analysis. Compared with CT alone, ICI + CT significantly improved PFS in all subgroups: dMMR (HR = 0.36, 95% CI:0.28–0.45, P < 0.00001), pMMR (HR = 0.78, 95% CI:0.70–0.87, P = 0.009), PD-L1-positive (HR = 0.52, 95% CI:0.38–0.70, P < 0.0001), and PD-L1-negative (HR = 0.66, 95% CI:0.44–0.98, P = 0.04). For OS, only the dMMR subgroup showed a significant benefit (HR = 0.41, 95% CI:0.28–0.61, P < 0.00001), with no improvement observed in pMMR patients (HR = 0.88, 95% CI:0.73–1.07, P = 0.20).

Conclusion

For advanced/recurrent EC, ICI + CT enhanced PFS across dMMR/pMMR and PD-L1 expression subgroups, with OS benefit apparently confined to dMMR patients and not observed in pMMR patients. These results present the clinically valuable predictive value of MMR status for ICI + CT efficacy and suggest that PD-L1expression did not influence the benefit of ICI on patient PFS, which deserve high clinical attention.

Trial registration

https://inplasy.com/inplasy-2026-03-0015/, identifier INPLASY202630015.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12957-026-04330-6.

Introduction

Endometrial cancer is the sixth most common cancer among women worldwide, and its incidence continues to rise with population aging and the increasing prevalence of obesity [1–3].The 5-year survival rate for patients with stage I disease is approximately 90%–95%, and the 10-year survival rate is about 77%; in contrast, the 5-year survival rate for stage IV disease is only 15%–20%, and the median survival time among patients with recurrence is 12–18 months [4]. Although the overall prognosis of early-stage endometrial cancer is favorable, patients with advanced, recurrent, and high-risk disease have poor outcomes, and effective treatment strategies remain very limited, representing a major challenge in current clinical management of endometrial cancer [1, 4]. For advanced/recurrent endometrial cancer, conventional chemotherapy and radiotherapy are suboptimal, with low 5-year survival rates, and there is an urgent need for novel therapeutic approaches to overcome this bottleneck [5, 6].

At present, immune checkpoint inhibitors (ICIs), including programmed cell death protein-1(PD-1)/programmed cell death ligand-1 (PD-L1) inhibitors, have shown remarkable activity in the treatment of advanced endometrial cancer owing to their unique antitumor immune mechanisms, and have become an important therapeutic option [7, 8]. Evidence has demonstrated that, compared with chemotherapy alone, ICI combined with chemotherapy significantly improves the prognosis of patients with advanced endometrial cancer [8]. Therefore, ICI plus chemotherapy has very promising prospects in endometrial cancer and has been increasingly applied in clinical practice [9]. However, multiple studies have reported differences in the efficacy of ICI-based regimens between deficient mismatch repair (dMMR) and proficient mismatch repair (pMMR) patients, with dMMR patients appearing more likely to benefit from ICI therapy than pMMR patients [10–13]. In addition, PD-L1 expression has been shown to have potential value for predicting the efficacy of ICIs in patients with various cancers [13–16]. Nevertheless, in endometrial cancer, whether PD-L1 expression is associated with the efficacy of ICI therapy remains elusive.

Therefore, in order to explore the prognostic predictive value of dMMR/pMMR and PD-L1 in the treatment of patients with advanced/recurrent endometrial cancer, we performed this meta-analysis, aiming to elucidate the differences in ICI efficacy among populations with different MMR and PD-L1 statuses and to thoroughly evaluate the predictive value of dMMR/pMMR and PD-L1, thereby providing a theoretical basis for the clinical management of endometrial cancer [17–19].

Methods

Literature search and study selection

We conducted a systematic search across four electronic databases—PubMed, Embase, the Cochrane Library, and Web of Science—from their inception until October 31, 2025. The search strategy employed a combination of controlled vocabulary (e.g., MeSH terms) and keywords related to endometrial/uterine cancers and ICIs, with database-specific search strings provided in Supplementary Materials Table S1. The search strategy employed a combination of controlled vocabulary (e.g., MeSH terms) and keywords related to endometrial/uterine cancers and immune checkpoint inhibitors. Key terms included, but were not limited to: “Endometrial Neoplasms”, “Uterine Cancer”, “checkpoint inhibitor”, “PD-1”, “PD-L1”, “CTLA-4”, and both generic and specific drug names(e.g.,pembrolizumab, nivolumab, cemiplimab, dostarlimab, retifanlimab, tislelizumab, camrelizumab, toripalimab, sintilimab, penpulimab, zimberelimab, sugemalimab, serplulimab, atezolizumab, durvalumab, avelumab, Ipilimumab, tremelimumab, cadonilimab, relatlimab, Ivonescimab). To ensure literature saturation, the reference lists of all retrieved review articles and eligible trials were manually examined for additional relevant studies. This systematic review and meta-analysis was registered in the INPLASY database (registration number: INPLASY202630015) and was conducted following PRISMA guidelines.

Eligibility criteria were restricted to randomized controlled trials (RCTs) investigating the combination of immune checkpoint inhibitors (ICIs) with chemotherapy for recurrent or metastatic endometrial cancer. We excluded non-randomized studies, reviews, meta-analyzes, case reports, and publications in languages other than English. Studies were also omitted if they presented ambiguous data or failed to report survival outcomes stratified by mismatch repair (MMR) status or PD-L1 expression levels. In cases of duplicate publications, only the most comprehensive and recent report was included. Two independent reviewers (L.L.F. and L.H.M.) performed the study selection, with any discrepancies adjudicated through discussion and consensus with a senior reviewer (Y.Y.H.).

Clinical outcomes and data extraction

The primary endpoints of this meta-analysis were progression-free survival (PFS) and overall survival (OS). Two independent reviewers (L.L.F. and L.H.M.) extracted the following data from each eligible trial: study identifiers (e.g., first author, publication year), design characteristics (phase, sample size), patient baseline demographics (e.g., median age), biomarker status (MMR and PD-L1), specific ICI-based treatment regimens, and the reported efficacy outcomes for PFS and OS. The comparative treatment effects on PFS and OS were expressed as hazard ratios (HRs) with their corresponding 95% confidence intervals (CIs). In the NRG GY018 trial, two completely independent subgroups were pre-specified in the original study design, namely the dMMR subgroup and the pMMR subgroup. These two subgroups differ significantly in their molecular characteristics and potential treatment responses, which may affect the consistency of the study results. To fully explore the potential differences in survival stratified by PD-L1 expression, we separated the data of the NRG GY018 trial into two entries (labeled as NRG GY018 (2025a) for the dMMR subgroup and NRG GY018 (2025b) for the pMMR subgroup) during the data extraction and analysis process. All subgroup analyses were conducted in accordance with the pre-specified protocol of this meta-analysis.

Quality assessment and risk of bias

Two independent reviewers (T.Z.Y. and Y.X.W.) evaluated the methodological quality of the included randomized controlled trials (RCTs) using the the Cochrane Collaboration’s tool, for assessing risk of bias [20]. This tool appraises bias across seven specific domains: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessors, management of incomplete outcome data, selective outcome reporting, and other potential sources of bias. For each domain and each study, the risk of bias was judged as “low,” “high,” or “some concerns” (indicating unclear risk). Publication bias across the included studies was visually inspected using funnel plots. Any discrepancies in assessment between the initial reviewers were resolved through discussion and consensus with a third senior reviewer (Y.Y.H.).

Statistical analysis

All meta-analyzes were conducted utilizing Review Manager (RevMan) software, version 5.4. Pooled treatment effects for progression-free survival (PFS) and overall survival (OS) are reported as hazard ratios (HRs) with 95% confidence intervals (CIs). The inverse variance (IV) method was used for effect size synthesis. Between-trial heterogeneity was quantified using the I² statistic. A random-effects model was employed in cases of considerable heterogeneity (I² > 50%); otherwise, a fixed-effects model was applied. A two-sided P value of less than 0.05 was considered statistically significant for all analyzes Funnel plots were used to visually assess publication bias for PFS and OS, with no significant publication bias indicated by the symmetrical distribution of study points.

Results

Study selection

Using the initial search strategy, we identified 4,056 potentially relevant records. After screening and removal of duplicates, four phase III trials (DUO-E [21], RUBY [22], NRG GY018 [23], and AtTEnd [24]) and one phase II RCT (MITO END-3) [25] published between 2023 and 2025. The primary results of these trials were published from 2023 to 2025, with updated follow-up data for the NRG GY018 trial released in 2025, comprising a total of 2,707 patients, were finally included (Table 1; Fig. 1). Among these five RCTs, DUO-E was a global study conducted at multiple centers across 22 countries; RUBY was a global trial involving 113 centers in 19 countries; NRG GY018 was a global trial conducted at multiple centers in 4 countries; AtTEnd covered Asia, Oceania, and Europe, involving 89 clinical research centers in 11 countries; and MITO END-3 was conducted at 31 clinical research centers in Italy, Europe. All patients were diagnosed with recurrent or metastatic endometrial cancer and received PD-1/PD-L1 inhibitors plus chemotherapy versus chemotherapy alone. Across these five RCTs, the ICI agents included two anti–PD-1 antibodies (pembrolizumab, dostarlimab) and three anti–PD-L1 antibodies (avelumab, atezolizumab, durvalumab). The characteristics and detailed information of each study are summarized in Table 1.

Table 1.

Baseline characteristics of incl uded studies

Study	Phase	Sample size		Age Median (Range)	dMMR	pMMR	PD-L1 (+)	PD-L1 (-)	Regimen
		Arms	N
MITO END-3 (2023) [25]	2	Study	63	66(61–72)	26	35	23	39	Avelumab plus carboplatin and paclitaxel
		Control	62	65(56–70)	31	29	24	35	Carboplatin and paclitaxel
DUO-E (2023) [21]	3	Study	477	64(22–84)	46	192	170	61	Durvalumab plus carboplatin and paclitaxel
		Control	238	64(31–85)	49	192	163	75	Carboplatin and paclitaxel
NRG GY018 (2025) [23]	3	Study	404	67.2 (39.0–82.0)a	110	294	294	102	Pembrolizumab plus carboplatin and paclitaxel
				66.2 (31–94.0)b
		Control	406	66.0 (37.0–86.0)a	112	294	302	97	Carboplatin and paclitaxel
				66.1 (29.0–91.0)b
AtTEnd (2024) [24]	3	Study	360	67 (61–73)	81	269	86	247	Atezolizumab plus carboplatin and paclitaxel
		Control	189	65 (60–73)	44	140	44	129	Carboplatin and paclitaxel
RUBY (2023) [22]	3	Study	245	64(41–81)	53	192	NR	NR	Dostarlimab plus carboplatin and paclitaxel
		Control	249	65(28–85)	65	184	NR	NR	Carboplatin and paclitaxel

PD-L1 programmed cell death ligand 1, dMMR deficient mismatch repair, pMMR proficient mismatch repair, RCT randomized controlled trial, NR not reported

^a dMMR subgroup; ^b pMMR subgroup

Fig. 1.

PRISMA flow diagram depicting the process of literature search and study selection

PFS outcome

All five included studies reported PFS outcomes for the dMMR and pMMR subgroups, and three studies reported PFS outcomes according to PD-L1 status. The meta-analysis indicated that, in the dMMR population, ICI + CT significantly improved PFS compared with CT alone (HR = 0.36, 95% CI [0.28, 0.45], I² = 0%, P < 0.0001; Fig. 2A). In the pMMR population, ICI + CT also yielded better PFS (HR = 0.78, 95% CI [0.64, 0.94], I² = 57%, P = 0.009; Fig. 2B). In addition, in both PD-L1–positive (HR = 0.52, 95% CI [0.38, 0.70], I² = 60%, P < 0.0001; Fig. 3A) and PD-L1–negative (HR = 0.66, 95% CI [0.44, 0.98], I² = 66%, P = 0.04; Fig. 3B) populations, the ICI group showed superior PFS benefits. Specifically, the PD-L1 expression data for the dMMR (2025a) and pMMR (2025b) cohorts within the NRG GY018 trial are fully characterized and displayed in Fig. 3.

Fig. 2.

Forest plots of progression-free survival (PFS) stratified by mismatch repair (MMR) status. A PFS in the dMMR subgroup; B PFS in the pMMR subgroup. Abbreviations: IV, Inverse Variance; Fixed/Random, Fixed-effect or Random-effects model; 95% CI, 95% Confidence Interval; dMMR, deficient mismatch repair; pMMR, proficient mismatch repair

Fig. 3.

Forest plots of progression-free survival (PFS) stratified by PD-L1 expression status. A PFS in the PD-L1-positive subgroup; B PFS in the PD-L1-negative subgroup. Abbreviations: IV, Inverse Variance; Fixed/Random, Fixed-effect or Random-effects model; 95% CI, 95% Confidence Interval; PD-L1, programmed death-ligand 1; a=dMMR subgroup; b= pMMR subgroup

OS outcome

Of the five included studies, four reported OS data for the dMMR and pMMR populations. None of the five studies reported OS data according to PD-L1 status. The meta-analysis results are shown in Fig. 4. In the dMMR population, compared with CT alone, ICI + CT significantly prolonged OS (HR = 0.41, 95% CI [0.28, 0.61], I² = 0%, P < 0.0001,Fig. 4A). In the pMMR population, there was no difference in OS (HR = 0.88, 95% CI [0.73, 1.07], I² = 49%, P = 0.20, Fig. 4B) between the two treatment groups.

Fig. 4.

Forest plots of overall survival (OS) stratified by mismatch repair (MMR) status. A OS in the dMMR subgroup; B OS in the pMMR subgroup. Abbreviations: IV, Inverse Variance; Fixed/Random, Fixed-effect or Random-effects model; 95% CI, 95% Confidence Interval; dMMR, deficient mismatch repair; pMMR, proficient mismatch repair

Quality of the included RCTs

The five included RCTs were thoroughly evaluated using the Cochrane Collaboration’s tool. Four of the five trials (DUO-E, RUBY, NRG GY018, AtTEnd) were judged to be of high methodological quality with no high risk of bias in any core domain. The MITO END-3 (2023) trial was assessed to have high risk of bias for allocation concealment (selection bias) and blinding of outcome assessment (detection bias), with “some concerns” for no other domains. The detailed risk-of-bias assessment was presented in Fig. 5. Funnel plots (Fig. 6) indicated that for both PFS and OS endpoints in the dMMR subgroup, individual study points were roughly symmetrically distributed around the vertical line of the pooled effect size (PFS: HR = 0.36; OS: HR = 0.41), with no obvious asymmetry observed.This suggested the absence of significant publication bias in the dMMR subgroup analysis, which was consistent with the high methodological quality of most included RCTs (four out of five trials with no high risk of bias in any domain). However, the small sample size of included studies (n = 5 for PFS, n = 4 for OS) limited the statistical power and interpretability of this visual publication bias assessment, and the results should thus be interpreted with caution.

Fig. 5.

Risk of bias assessment for the included randomized controlled trials (RCTs)

Fig. 6.

Funnel plots for publication bias assessment of PFS and OS in the dMMR subgroup of advanced/recurrent endometrial cancer

Discussion

To our knowledge, the current study is the first comprehensive meta-analysis focused on evaluating the roles of MMR status and PD-L1 expression in recurrent and metastatic endometrial cancer treated with ICIs. This meta-analysis demonstrates that both dMMR and pMMR subgroups derive a PFS benefit from ICI plus chemotherapy, although the magnitude of benefit differs between the two groups. Of note, in the first-line setting, a statistically significant OS benefit was observed only in the dMMR population. Furthermore, the current evidence does not support a predictive role of PD-L1 expression for ICI efficacy in endometrial cancer. These findings highlight the distinct predictive roles of MMR status and PD-L1 expression, providing novel evidence for biomarker-guided ICI combination therapy in recurrent or metastatic endometrial cancer. This suggests that MMR is of great predictive value in assessing outcomes in recurrent and metastatic endometrial cancer treated with ICIs, and that the dMMR subgroup may derive the greatest OS benefit, further confirming the central value of dMMR status as a predictive biomarker for the efficacy of immunotherapy in endometrial cancer and supporting routine MMR testing before receiving ICI treatment in clinical practice.

From a mechanistic perspective, relevant studies have shown that MLH1 deficiency (one of the main causes of dMMR) leads to aberrant DNA excision by exonuclease 1, resulting in cytosolic DNA accumulation and subsequent activation of the cGAS–STING pathway. This process provides a key molecular basis for the efficacy of tumor immunotherapy and directly links mismatch repair deficiency to activation of the cGAS–STING pathway; thus, tumors with deficient mismatch repair (dMMR) respond to immunotherapy because dMMR induces neoantigen generation and activation of the cGAS–STING pathway [26]. Marco Gerlinger and colleagues proposed that the immunotherapy sensitivity of dMMR tumors cannot be attributed solely to high TMB and neoantigen load, but also requires cGAS–STING pathway activation, which induces type I interferon secretion and promotes T-cell activation; they further demonstrated that loss of the MLH1 gene drives activation of the cGAS–STING pathway, underscoring its critical role in enhancing tumor immunogenicity and increasing sensitivity to ICIs [27]. Xu et al. reported that in dMMR/MSI-H tumors, replication protein A depletion leads to cytosolic DNA accumulation, which can activate the cGAS–STING pathway, and noted that this pathway enhances tumor immunogenicity, whereas dysregulation of its negative regulatory mechanisms can impair T-cell–mediated cytotoxicity, providing evidence from both positive and negative aspects for the association between this pathway and the sensitivity of dMMR tumors to ICIs [28]. The association of dMMR and pMMR status with outcomes of immunotherapy has also been observed in other tumor types.

In the KEYNOTE-177 and CheckMate 142 studies, dMMR/MSI-H metastatic colorectal cancers exhibited superior responses to immunotherapy [13, 29]. Among patients with locally advanced dMMR colon cancer, neoadjuvant immunotherapy also achieved higher rates of pathological response [30]. By contrast, pMMR colorectal cancers respond poorly to immunotherapy [31–33]. In gastric or gastroesophageal adenocarcinoma, multiple studies have shown that dMMR/MSI-H patients derive higher clinical response rates from immunotherapy [34–36], whereas pMMR patients have low response rates; combined chemotherapy can improve efficacy in this group but remains inferior to that in dMMR patients [36]. Guidelines recommend routine dMMR/MSI testing in patients with advanced gastroesophageal cancer, with immunotherapy prioritized for positive patients regardless of prior lines of therapy [37, 38]. In gastric cancer, neoadjuvant PD-1 blockade has enabled organ preservation in a substantial proportion of patients with early-stage dMMR solid tumors amenable to curative surgery, further validating dMMR as a marker of benefit from immunotherapy in other solid tumors [39].

PD-L1 is currently one of the most widely used predictive biomarkers for ICI immunotherapy benefit and is recommended by guidelines for patient selection across multiple tumor types. In non–small cell lung cancer, guidelines explicitly recommend assessment of PD-L1 status [40, 41]; in gastric and esophageal cancers, PD-L1 testing is advised for patients scheduled to receive PD-1/PD-L1 inhibitors [42, 43]; and in the NCCN guidelines for persistent/recurrent/progressive cervical cancer, PD-L1 expression testing is explicitly required [44]. In our study, ICI plus chemotherapy improved PFS in recurrent/metastatic endometrial cancer in both PD-L1–positive and PD-L1–negative populations, indicating that the expression level of PD-L1 did not affect the PFS benefits of patients and the predictive role of PD-L1 in PFS may be relatively limited, thus future research needs to focus on its OS prediction results. Because OS data were incomplete, we were unable to evaluate the predictive value of PD-L1 for OS, underscoring the need for further studies. Although the predictive value of PD-L1 varies across tumor types and immunotherapy regimens, it has consistently attracted attention as one of the most important predictive factors for ICI therapy.

Despite the encouraging findings, this study has several limitations. First, the number of included studies was relatively small, and substantial heterogeneity was observed in the PFS analysis of the pMMR population (I² = 57% or 68%) as well as in the PD-L1-positive (I² = 60%) and PD-L1-negative (I² = 66%) subgroups, which may affect the stability of the results. As noted, the moderate to high heterogeneity in PFS analysis of these subgroups was mainly attributed to different ICI agents (two anti-PD-1 antibodies and three anti-PD-L1 antibodies, a major potential source of heterogeneity), non-uniform PD-L1 testing methods, minor differences in chemotherapy regimens and patient baseline characteristics, as well as high bias risk of the MITO END-3 trial (though it had little impact on the overall PFS benefit direction). Due to the limited number of included RCTs (n = 5), we did not perform leave-one-out sensitivity analysis, as excluding any single study would lead to unstable estimates and insufficient statistical power; instead, we adopted a random-effects model to account for potential heterogeneity and ensure the robustness of the pooled results.Second, two types of ICI agents were included, but the limited number of studies precluded further subgroup analyzes according to ICI type, potentially introducing bias; therefore, the results should be interpreted with caution. Third, a potential limitation is the nested subgroup analysis of the NRG GY018 trial (MMR + PD-L1 stratification), which resulted in small sample sizes for individual nested subgroups (e.g., dMMR/PD-L1-negative, pMMR/PD-L1-positive). Thus, the observed trends in benefit magnitude should be interpreted with caution, and future large-scale prospective studies are needed to validate the combined predictive value of MMR status and PD-L1 expression. Fourth, we did not further analyze the combined predictive value of molecular subtypes (such as POLE-mutated and TP53-mutated subtypes) with MMR and PD-L1 status. Future studies should explore joint predictive models integrating MMR, PD-L1, and molecular subtypes, and investigate novel combination regimens for the pMMR population (e.g., ICI plus antiangiogenic agents plus antibody–drug conjugates) to further improve survival outcomes in these patients. Finally, owing to limited data resources, MSI status of patients could not be adequately analyzed in parallel, which to some extent affects the evaluation of our results. Additionally, two potential confounding factors should be noted: extensive immunotherapy crossover in control arms post-progression may have diluted the OS benefit of ICI + CT (impacting pMMR subgroup OS analysis), and non-uniform PD-L1 testing methods may have reduced cross-study result comparability.

Conclusion

This meta-analysis demonstrates that ICI plus chemotherapy confers a significant PFS benefit in patients with recurrent/metastatic endometrial cancer, regardless of their MMR or PD-L1 biomarker status. Notably, however, only dMMR status may emerge as a potential independent predictor of OS benefit with this combination regimen, whereas pMMR status did not appear to be associated with improved OS outcomes. Furthermore, PD-L1 expression did not influence the benefit of ICI on patient PFS, suggesting its relatively limited predictive value for this endpoint, though further evaluation of its role in OS remains warranted. These observations hold important clinical implications for guiding ICI-based treatment decision-making in this patient cohort. Given the inherent limitations of this meta-analysis, additional well-designed prospective studies are urgently needed to validate these findings and refine biomarker-driven therapeutic approaches.

Abbreviations

ICI

Immune checkpoint inhibitor

RCT

Randomized controlled trial

OS

Overall survival

PFS

Progression-free survival

PD-1

Programmed cell death protein-1

PD-L1

Programmed cell death ligand-1

MMR

Mismatch repair

dMMR

Deficient mismatch repair

pMMR

Proficient mismatch repair

MLH1

MutL homolog 1

DNA

Deoxyribonucleic acid

cGAS-STING

Cyclic GMP–AMP synthase–stimulator of interferon genes

TMB

Tumor mutational burden

MSI-H

Microsatellite instability-high

POLE

DNA polymerase epsilon, catalytic subunit

NCCN

National Comprehensive Cancer Network

TP53

Tumor protein 53

Authors’ contributions

Yiua Yang coordinated the data collection and conceived the original idea, Lifang Lan and Hongmei Liao provided statistical analysis, Lifang Lan wrote the manuscript, all other Authors facilitated data collection and critically reviewed the manuscript for important intellectual contents. Yihua Yang and Lifang Lan accept direct responsibility for the manuscript.

Funding

The National Natural Science Foundation of China (Nos.82571880 and 82360308 ), the Guangxi Medical University Training Program for Distinguished Young Scholars provided funding for this work.

Data availability

The original data used during the current study can be obtained by contacting the corresponding author.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.