Introduction

Immune checkpoint inhibitors (ICIs) have revolutionized the field of cancer treatment [1,2,3,4]. This groundbreaking approach has shifted the focus from indiscriminate targeting of cancer cells by traditional chemotherapy to enhancing the immune system's ability to selectively attack tumor cells. However, the efficacy of ICIs is still far from satisfactory due to resistance [5, 6], and the distinction of biomarkers of response is an intense area of research [7]. This could potentially be linked to the variability observed in the tumor microenvironment (TME) across different types of tumors [8, 9]. Thus, it is crucial to identify the populations that benefit from ICIs treatment.

Based on the infiltration of T cells, Chen and Mellman have classified the tumor immune microenvironment into three different phenotypes: immune-desert, immune-excluded, and immune-inflamed. Among them, the immune-desert and immune-excluded phenotypes, also known as “cold tumors”, were non-inflamed tumors that were typically insensitive to ICIs. In contrast, the immune-inflamed phenotype, referred to as "hot tumors", exhibited a significantly stronger response to ICIs [8, 10]. Studies demonstrated that tumors associated viral infection often exhibit an “hot tumor” [11, 12]. However, the impact of viral infection in tumors on the efficacy of ICIs remains a topic of debate in clinical practice, with no established consensus. While some researchers supported a positive impact [13, 14], others advocated for non-inferior survival outcome [15, 16]. Additionally, the types of tumors, viruses and ICIs were various. Therefore, it remains to be fully illuminated that the impact of viral status on the efficacy of ICIs in cancer patients.

Previous studies had preliminarily explored the effect of human papilloma virus (HPV) on ICIs through meta-analysis [17], but they focused on single type of tumor and lack of subgroup analyses. Here, we conducted a comprehensive survey based on a large sample size (29 cohorts incorporating 3,255 individuals), multiple types of viruses and tumors to evaluate the impact of viral status on ICIs efficacy for malignancies.

Materials and methods

Literature searches

PubMed, Cochrane Library, and EMBASE were systematically searched to identify relevant studies up until December 15th, 2023 by entering the following keywords: “immune checkpoint inhibitors”, “ICI”, “immunotherapy”, “PD-1”, “PD-L1”, “CTLA-4”, “programmed cell death protein 1”, “programmed cell death protein ligand 1”, “cytotoxic T lymphocyte-associated protein 4”, “pembrolizumab”, “nivolumab”, “atezolizumab”, “ipilimumab”, “tremelimumab”, “avelumab”, “durvalumab”, “carelizumab”, “tislelizumab”, “cemiplimab”, “toripalimab”, “penpulimab”, “cemiplimab”, “adebrelimab”, “sugemalimab”, “Epstein Barr virus”, “EBV”, “Hepatitis B virus”, “HBV”, “Human Papilloma virus”, “HPV”, “hepatitis C virus”, “HCV”, “gastric cancer”, “GC”, “stomach Cancer”, “hepatic cancer”, “liver cancer”, “HCC”, “nasopharyngeal carcinoma”, “NPC”, “head and neck cancer”, “HNSCC”, “lymphoma”, “cancer”, “neoplasm”, “tumor”, “carcinoma”, and “malignancy”. Moreover, the reference lists of related articles were scrutinized for additional studies.

Selection of studies

Two investigators respectively performed an initial screening of titles and abstracts, and then scrutinized the full texts to identify eligible studies.

Inclusion criteria

Inclusion criteria were included: (1) individuals were pathologically confirmed as malignancies; (2) therapeutic outcomes were analyzed on the efficacy of ICIs according to viral status (including Epstein Barr virus (EBV), hepatitis B virus (HBV), hepatitis C virus (HCV), and HPV; (3) A hazard ratio (HR) accompanied by a 95% confidence interval (CI) for progression-free survival (PFS) and/or overall survival (OS) and/or odds ratios (OR) with 95% CI for objective response rate (ORR) could be obtained or calculated from the original literature.

Data extraction

Data from all enrolled studies were independently collected by two investigators. The data was collected from each publication as follow: publication year, first author, number of patients, primary tumor, immunotherapy agents, viral types, HR for OS and/or PFS, and OR for ORR between the viral infection group and viral uninfection group.

Quality assessment

The quality of studies was assessed using the Newcastle–Ottawa quality assessment scale (NOS), with scores of more than six indicating medium to high quality [18]. Discrepancies were settled through a consensus reached among all investigators.

Statistical methods

The primary endpoints of the study were OS and PFS. The association between viral status (infection vs uninfection) and the efficacy ICIs was measured applying HR with the corresponding 95% CI. Subgroup analyses were accomplished based on the viral types, treatment regimen, patient locations, and ICI agents. Statistical analysis was performed by R 4.2.2 statistical software. Heterogeneity was evaluated through the I-square tests and Cochran’s Q test. if P < 0.05 or I2 > 50%, it indicated remarkable heterogeneity, and a random effect model was employed. Otherwise, a fixed effect model was adopted. Publication bias was assessed using funnel plot, Egger’s test, and trim-and-fill method [19].

Results

Study selection and characteristics of trials

A total of 18,347 potentially relevant articles were intensively scrutinized. Among them, 1,658 were removed for duplication, while 16,689 were filtered out for digressing from the subject after screening the titles and abstracts. Subsequently, the full texts of 163 articles were thoroughly reviewed, of which 134 were excluded for the following reasons: repeated study cohort (n = 23), unavailable data to evaluate the efficacy of ICIs (n = 47), non-human research (n = 22), reviews or meta-analysis (n = 42). Finally, a total of 29 studies incorporating 3,255 participants were identified (The links of original article and details were shown in Supplementary 1). The elaborate procedure is displayed in Fig. 1.

Fig. 1
figure 1

Flowchart on selection including trials in the meta-analysis

Full size image

A total of 3255 individuals in 13 retrospective studies and 16 prospective studies were recruited. All 29 adopted studies were rated as moderate or high quality. Furthermore, the sample size ranged from 12 to 421. Of these studies, 3 focused on gastric cancer (GC), 1 on nasopharyngeal carcinoma (NPC), 14 on hepatocellular carcinoma (HCC), 11 on headneck squamous cell carcinoma (HNSCC). Principal traits and details were presented in Table 1.

Table 1 The principal characteristics and further details of eligible articles
Full size table

Main results

The impact of viral status on malignancy patients treated with ICIs

Pooled results showed that tumor patients with viral infection who received ICI agents had a significantly favorable OS [HR = 0.67, 95%CI (0.57–0.79), P < 0.0001] by a random-effect model (I2 = 42%, P = 0.02) (Fig. 2a), and a trend towards improved PFS [HR = 0.75, 95%CI (0.56–1.00), P = 0.05] based on a random-effect model (I2 = 58%, P < 0.01)(Fig. 2b). Furthermore, There was an increased ORR [OR = 1.43, 95%CI (1.14–1.80), P = 0.0018] in viral positive group according to a fixed-effect model (I2 = 12%, P = 0.31) (Fig. 2c).

Fig. 2
figure 2

Forest plots for a overall survival (OS), b progression-free survival (PFS), and c objective response rate (ORR)

Full size image

Subgroup analysis for the impact of viral status

We performed subgroup based on the viral types, treatment regimen, and patient locations (Fig. 3a–i). The result showed that patients treated with ICIs who were exposed to HBV/HCV or HPV infection exhibited an evidently superior OS without heterogeneity, compared to those without HBV/HCV [HR = 0.79, 95%CI (0.65–0.96)] and HPV [HR = 0.64, 95%CI (0.53–0.76)]. While the OS was similar between the EBV-positive group and the EBV-negative group [HR = 0.58, 95%CI (0.21–1.63)] (Fig. 3a). Furthermore, we found that patients with viral infection who received ICIs had a significantly better OS, in contrast to those without infection, regardless of the treatment types (monotherapy or combined therapy) [HR = 0.64, 95%CI (0.54–0.75) and HR = 0.78, 95%CI (0.64–0.94), respectively] (Fig. 3b) and patient locations (eastern countries or western countries) [HR = 0.68, 95%CI (0.55–0.84) and HR = 0.57, 95%CI (0.45–0.71), respectively] (Fig. 3c). Additionally, the groups with HBV/HCV or HPV infection achieved a higher ORR compared to the groups without HBV/HCV or HPV infection)[OR = 1.58, 95%CI (1.19–2.10) and OR = 1.57, 95%CI (1.0–2.47), respectively] (Fig. 3g).

Fig. 3
figure 3

a The pooled HRs for overall survival (OS) stratified on viral types (EBV, HBV/HCV, and HPV); b treatment regions (monotherapy or combined therapy); c patients locations (western countries and eastern countries); d the pooled HRs for progression-free survival (PFS) stratified on viral types; e treatment regions; f patients locations; g the pooled ORs for objective response rate (ORR) stratified on viral types; h treatment regions; i patients locations

Full size image

Publication bias

The shape of the funnel plot suggested no publication bias for recruited studies on PFS (Egger: P = 0.11) (Fig. 4a) and ORR (Egger: P = 0.98) (Fig. 4b). However, there was a publication bias for OS (Egger: P = 0.02) (Fig. 4c). Nevertheless, in the results of the trim and fill method, the publication bias corrected overall effect size was 0.76 (95% CI: 0.62–0.93) (Fig. 4d), even though the effect size increased compared to the original ones. This implied that the results obtained for this study were reliable and consistent.

Fig. 4
figure 4

Funnel plot of publication bias on a progression-free survival (PFS), b objective response rate (ORR), and c overall survival (OS) in the meta-analysis; d the corrected HRs for OS based on the trim and fill method

Full size image

The sensitivity analysis

Sensitivity analyses were performed by excluding one single study from the primary analyses. The results showed that no single study significantly influenced the pooled HRs or ORs, suggesting that the data of this meta-analysis were relatively credible and stable (Fig. 5).

Fig. 5
figure 5

Sensitivity analysis of a overall survival (OS), b progression-free survival (PFS), and c objective response rate (ORR)

Full size image

Discussion

Recently, increasing evidences highlighted the importance of viral infection in influencing the efficacy of immunotherapy. Pre-clinical and clinical evidences have recognized that the viral infection may play a crucial role in boosting the immune response and improve prognosis of cancer patients undergoing ICIs treatment [11,12,13,14]. To the best of our knowledge, this study was firstly investigated the impact of viral infection on outcomes of cancer patients treated with ICIs based on a comprehensive survey (29 cohorts incorporating 3,255 individuals), multiple viruses types (EBV, HBV, HCV, and HPV) and multiple tumor types (including GC, NPC, HCC and HNSCC). The result demonstrated a significant association between viral infection and improved outcomes for cancer patients receiving ICIs treatment.

Mechanically, PD-1 and its ligands played a crucial role in enabling tumor cells to evade the anti-tumor response of immune system [20]. Less widely recognized was that the PD-1/PD-L1 axis also played a role in regulating immune responses against viral infection and can be influenced by various viruses [21, 22]. Upregulation of PD-1 and its ligands PD-L1 were observed during acute viral infection and after infection with persistent viruses including important human pathogens such as HBV, HCV, and EBV [23,24,25,26]. Moreover, viral infection associated carcinomas were typically characterized by abundant immune cell infiltration [11, 12], which might further positively affect the efficacy of ICIs.

Notably, our study exhibited that EBV infection did not impact the efficacy of ICI treatment, despite the frequent association of EBV infection with high PD-L1 expression in tumors was discovered [25, 26]. The reason remains to be elucidated. However, it should be noted this study comprising EBV associated tumors was only included 4 cohorts with a total of 408 participants. Therefore, caution should be advised when interpreting the result.

However, this study encountered two flaws: firstly, some of recruited studies were retrospective, although we had comprehensively analyzed the articles; secondly, due to the limited availability of comprehensive data, subgroup analysis based on specific ICI agents could not be conducted.

In conclusion, this study demonstrated that the presence of viral infection was positively associated with better outcomes, with improved OS, increased ORR, and potential benefits in PFS in cancer patients undergoing ICIs therapy. And subgroup analyses on therapy regimen and patient locations exhibited similar results, indicating the positive impact of viral infection on ICIs therapy in clinical practice.