Prognostic Impact of Memory CD8(+) T Cells on Immunotherapy in Human Cancers: A Systematic Review and Meta-Analysis

Objective The objective of this systematic review and meta-analysis was to determine the prognostic value of memory CD8(+) T cells in cancer patients with immunotherapy. Methods EMBASE, MEDLINE (PubMed), and Web of Science databases were searched to identify suitabile articles published before March 2021. Risk of bias on the study level was assessed using the Cochrane Bias Risk Assessment Tool. The hazard ratios (HRs) and 95% confidence intervals (CIs) of pooled progression-free survival (PFS) and overall survival (OS) were calculated using RevMan 5.4 to evaluate the prognostic impact of memory CD8(+) T cells. Results In total, nine studies were included in the final analysis. High levels of memory CD8(+) T cells were significantly closely correlated with better progression-free survival (PFS) and overall survival (OS) of cancer patients with immunotherapy (PFS, HR 0.64, 95% CI 0.53–0.78; OS, HR 0.37, 95% CI 0.21–0.65). Memory CD8(+) T cells still have significant prognostic value in cancer patients given immunotherapy alone after excluding of other interfering factors such as chemotherapy, radiotherapy, and targeted therapy (PFS, HR 0.65, 95% CI 0.48–0.89; OS, HR 0.23, 95% CI 0.13–0.42). However, high memory CD8(+) T cells levels did not correspond to a longer PFS or OS in cancer patients with non-immunotherapy (PFS, HR 1.05, 95% CI 0.63–1.73; OS, HR 1.29, 95% CI 0.48–3.48). Thus, memory CD8(+) T cells might be a promising predictor in cancer patients with immunotherapy. Conclusions The host’s overall immune status, and not only the tumor itself, should be considered to predict the efficacy of immunotherapy in cancer patients. This study is the first to show the significant prognostic value of memory CD8(+) T cells in immunotherapy of cancer patients through systematic review and meta-analysis. Thus, the detection of memory CD8(+) T cells has a considerable value in clinical practice in cancer patients with immunotherapy. Memory CD8(+) T cells may be promising immunotherapy targets.


INTRODUCTION
Tumor-infiltrating CD8(+) T cells can predict patient survival and response to immunotherapy in many cancers (1)(2)(3)(4)(5)(6)(7)(8)(9). However, it is not clear why some patients with high CD8(+) T cell levels respond significantly to immunotherapy, while others do not. One reasonable explanation is that there are multiple subtypes of tumor-infiltrating CD8(+) T cells with different phenotypes and functions (10)(11)(12)(13)(14)(15). Han et al. (14). found three main subtypes of tumor-infiltrating CD8(+) T cells: memory CD8(+) T cells, exhausted CD8(+) T cells, and effector CD8(+) T cells. Exhausted CD8(+) T cells are characterized by low proliferation in response to neoantigen stimulation, progressive loss of effector function, expression of multiple inhibitory receptors such as PD-1, Tim3, and LAG3, and ultimately loss of their tumor-killing effect. Although effector CD8(+) T cells and memory CD8(+) T cells share some similarities at the epigenetic, metabolic, and functional levels, memory CD8(+) T cells can persist for a long time when effector CD8(+) T cells undergo severe contractions. Unlike effector CD8(+) T cells, they are capable of dramatic proliferation after contact with neoantigens and they can maintain the host's continuous anti-tumor immune state. This suggests that memory CD8(+) T cells play a key role in the sustained tumor killing (16)(17)(18). Therefore, it is not comprehensive to predict the efficacy of immunotherapy in patients solely by the number of tumor infiltrating CD8(+) T cells or the level of expression of PD-L1 (19)(20)(21)(22). At present, some studies have found that the levels of memory CD8(+) T cells measured by immunohistochemical (IHC) staining can predict the efficacy of immunotherapy in cancer patients, but some studies contradict this conclusion (23). These studies all had insufficient sample size and low universality. Therefore, the objective of the present systematic review and meta-analysis was to determine the prognostic roles of memory CD8(+) T cells in cancer patients who had undergone immunotherapy.

METHODS
The present systematic review and meta-analysis were conducted following the Preferred Reporting Items for Systematic Reviews and Meta Analyses (PRISMA) statement (24).

Eligibility Criteria
The inclusion criteria were (1) The subject was pathologically diagnosed as a malignant tumor (2), Study subjects must be treated with immune checkpoint inhibitors (3), the relationship between memory CD8(+) T cell levels and PFS or OS was investigated (4), studies containing enough data to estimate the effects HR and 95% CI for PFS or OS. The exclusion criteria included (1): studies without enough data to estimate a HR with a 95%CI (2), studies from review articles, case reports, commentaries and letter.

Outcome Definitions
Progression-free survival (PFS) is defined as the time from the beginning of immunotherapy to the first disease progression.
Overall survival (OS) was defined as the period from the date of tumor diagnosis to the time of death with any cause (25).

Search Strategy and Study Selection
A systematic literature search was conducted using a combination of key words and Medical Subject Headings (MeSH) terms in the following databases: EMBASE, MEDLINE (PubMed), and Web of Science, to identify eligible articles published before March 2021. The search terms mainly include: stem-like T cell, exhausted T cell, memory T cell, Tumor infiltrating lymphocytes, Cancer and immune checkpoint. Details the retrieval strategy in the Supplementary material.

Data Extraction
Indeed data were extracted from the included studies using a pilot-tested data extraction form. In addition to the information in Table 1, the extracted data included HRs with 95% CIs for PFS or OS.

Statistical Analysis
We performed meta-analysis to obtain a pooled estimate of the prognostic role of memory CD8(+) T cells using RevMan 5.4. A Pvalue less than 0.05 was set as indicative of statistical significance. Between-study heterogeneity was measured using the Higgins I 2 statistic and Cochrane's Q test (P < 0.10 or I 2 > 50% was considered indicative of statistically significant heterogeneity) (26). A random effects model was applied if heterogeneity was present. However, the fixed effect model was used in the absence of between-study heterogeneity (P > 0.10 or I 2 < 50%) (27).

Search Results and Study Characteristics
Eligible studies were identified and selected as shown in Figure 1. A total of 5797 articles were for initial evaluation, 4833 studies were eligible after exclusion of duplicates. Abstracts and titles of these studies were reviewed and 4799 studies were excluded. After abstract review, we identified 34 articles for full manuscript review and 28 of these articles were excluded for the reasons delineated in Figure 1. Finally, nine studies from six articles (23,(28)(29)(30)(31)(32) were included in the final meta-analysis ( Table 1).

Memory CD8(+) T Cells and OS
The association between the levels of memory CD8(+) T cells and OS of cancer patients with immunotherapy was extracted from five studies. The results showed that higher levels of memory CD8(+) T cells correspond to better OS, with pooled HR of 0.37 (95% CI 0.21-0.65) for memory CD8(+) T cell level (high vs. low) ( Figure 3A).
However, there was relatively significant heterogeneity between the main group and the subgroup. Through sensitivity and subgroup analysis, it was found that the heterogeneity was mainly derived from study Wong 2019, and the heterogeneity was significantly decreased in the main group (I 2 = 0%) and subgroup (I 2 = 0%) after its elimination. This may be due to the T cells phenotype is GZMB in Wong 2019, which differed significantly from the definition of the T cells phenotype in other studies. In addition, the final conclusions remain the same in the main group (HR 0.30, 95% CI 0.19-0.47) and in the subgroup (HR 0.26, 95% CI 0.14-0.51) after exclusion of Wong 2019. Thus, this analysis confirmed the stability of our results.

Memory CD8(+) T Cells in the Control Arms
There are a total of 3 studies with immunotherapy as the experimental group and non-immunotherapy as the control group focused on the predictive value of memory CD8(+) T cells. Our study found that high levels of memory CD8(+) T cells have no effect on PFS and OS in cancer patients with non-immunothrapy (PFS, HR 1.05, 95% CI 0.63-1.73; OS, HR 1.29, 95% CI 0.48-3.48) (Figures 4A, B). However, there is significant heterogeneity between the IMmotion 150_control and POPLAR_control groups, which may be due to the difference in baseline treatment regimens between the two

Risk of Bias in Included Studies
Risk of bias were assessed for all included studies (n=9) as shown in Figure 5. Since the articles are all prospective clinical trials, the overall risk of bias is relatively low.

Publication Bias
Funnel plot analysis did not indicate apparent publication bias affecting the HRs for PFS and OS ( Figure 6).

DISCUSSION
In the past, tumor immunotherapy focused on T cell immunity, which enabled 20% of patients to enjoy greater clinical benefits (33,34). In order to further improve the efficacy of immunotherapy and predict the target population, it is necessary to study the roles of T cells of different phenotypes and functions in anti-tumor immunity in depth. In chronic infections, memory CD8(+) T cells have been shown to be a key factor in persistent immune activity. However, its role in anti-tumor immunity is still unclear (35)(36)(37)(38). This is the first systematic review and metaanalysis to synthesize nine studies to evaluate the association between memory CD8(+) T cell levels and prognosis of immunotherapy in cancer patients. Our results indicate that high levels of memory CD8(+) T cells in cancer patients with immunotherapy may significantly prolong PFS and OS. In our study, the indicators of outcome prognosis were PFS and OS. Previous studies reported that higher memory CD8(+) T cell levels predict a worse response to immunotherapy in cancer patients (23). According to our pooled results, cancer patients with high memory CD8 T cell levels showed better PFS (HR 0.64, 95% CI 0.53-0.78) and OS (HR 0.38, 95% CI 0.19-0.73) than those with low memory CD8(+) T cell levels. Furthermore, our study also found a significant predictive effect of memory CD8(+) T cell levels in cancer patients with immunotherapy alone after controlling for interference factors such as chemotherapy, radiation and targeted therapy (PFS; 0.65, 95% CI 0.48-0.89, OS; OS; 0.23, 95% CI 0.13-0.42). One potential explanation for these findings is the influence of memory CD8(+) T cells levels on tumor immunorecognition and tumor immunosuppression (39)(40)(41) Tumors are a chronic disease, and patients who develop curative immunity to tumor cells are thought to need longlived T cell immunity, which is mediated by memory CD8(+) T cells (10,36). A correlation between the numbers of memory CD8(+) T cells and the level of tumor immunity has been firmly established, functional ability of memory CD8(+) T cells also determines the degree of memory CD8(+) T cell-mediated tumor immunity (10,42,43). Unlike effector T cells and exhausted T cells, memory CD8 T cells can persist and function throughout host and tumor tissues, making them a potential target for anti-tumor immunity (12,44). Therefore, patients with high memory CD8(+) T cell levels tend to have better tumor immunosuppressive effects.
Our study also has some specific limitations. First, the number of included studies was limited and the sample size of the included study was insufficient. Second, most of the subjects were still treated with combined immunotherapy, which may cause interference from other factors, such as chemotherapy, radiotherapy, and targeted therapy. In addition, there are also differences in immune checkpoint inhibitors among different studies, and there may be bias errors. Next, there were subtle differences in the definition of memory CD8(+) T cells in different studies, which may be the source of the heterogeneity of the paper. This will also be the direction of further attention and research. Furthermore, our results support prognostic value of memory CD8(+) T cells, however the number of studies in the control arms was relatively limited. Thus, there need more research data to further support the predictive value of memory CD8(+) T cells. Finally. the prognostic value of memory CD8(+) T cells has only been confirmed in lung cancer and urothelial carcinoma, and its value in other types of cancer merits further exploration.

CONCLUSION
In recent years, immunotherapy has brought considerable survival benefits for cancer patients, but it has also encountered therapeutic bottlenecks. Currently, biomarkers of the efficacy of immunotherapy, such as TMB, PDL-1 and MSI-H, are all based on tumor cells themselves (19,45,46). However, comprehensive prognosis assessment should pay attention to both tumor and host immune status. Therefore, it is urgent to study the immune microenvironment in depth in order to further improve the efficacy of immunotherapy (47).
Based on the above analysis, we can definitively conclude that a higher level of memory CD8(+) T cells corresponds to a better prognosis for immunotherapy in cancer patients, and significant differences were also observed in patients given immunotherapy alone after exclusion of other interfering factors such as chemotherapy, radiotherapy and targeted therapy.
This study is the first to demonstrate that memory CD8(+) T cells play a key role in tumor immunotherapy. More noteworthy is that memory CD8(+) T cells can not only predict the efficacy of patients before immunotherapy, but also may become a promising target for anti-tumor therapy, which may enhance the efficacy of immunotherapy by increasing the content of memory CD8(+) T cells in the immune microenvironment.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding authors.

AUTHOR CONTRIBUTIONS
YJ, AT, MP and YW were involved in the conception and design of the study. PW retrieves the database, and ZX extracts the data. YJ and AT were used for statistical analysis. YJ and AT wrote the first draft of the manuscript. JP, PR, RL, and MP wrote parts of the manuscript. All authors contributed to the article and approved the submitted version.