Association of Peripheral Blood Biomarkers with Response to anti-PD-1 Immunotherapy for Patients with dMMR Metastatic Colorectal Cancer: A Multicenter Cohort Study

Background: Mismatch repair decient (dMMR) is an established biomarker for response to anti-PD-1 immunotherapy in metastatic colorectal carcinoma (mCRC). Although patients with dMMR mCRC could achieve a high incidence of disease control and favorable progression-free survival (PFS), reported response rates to PD-1 inhibitors are variable with 28%–52%, indicating that additional predictive biomarkers are warranted. Methods: This multicenter cohort study enrolled patients with dMMR mCRC receiving anti-PD-1 immunotherapy at Sun Yat-sen University, the Sixth Aliated Hospital and Sun Yat-sen University Cancer Center between December, 2016 to December, 2019. A total of 20 peripheral blood biomarkers, including T cells (frequency of CD4+ T cell, frequency of CD8+ T cell, ratio of CD4+/CD8+), carcinoembryonic antigen (CEA), inammatory markers and lipid metabolism markers. The association between response or survival and peripheral blood parameters was analyzed. Results: Among tested parameters, ratio of CD4+/CD8+, frequency of CD4+ T cell was signicantly associated with PFS (P=0.023, P=0.012) and OS (P=0.027, P=0.019) in univariate analysis. Lower level of CD4+/CD8+ ratio or frequency of CD4+ T cell showed signicant association with better overall response rates (ORR; P = 0.03, P=0.01). Ratio of CD4+/CD8+, and frequency of CD4+ T cell maintained signicance in multivariate Cox model for PFS (HR=9.23, P=0.004; HR=4.83, P=0.02) and OS (HR=15.22, P=0.009; HR=16.21, P=0.025). Conclusions: Ratio of CD4+/CD8+ and frequency of CD4+ T cell might be crucial independent biomarkers within dMMR mCRC to better identify patients for response to PD-1 inhibitors. If validated in

Tumor in ltrating lymphocytes (frequency of CD8 + T cells) mainly contributes to the antitumor immune response and is a reliable prognostic indicator for CRC [8,9], but it is not an optimal predictor for anti-PD-1 immunotherapy. Furthermore, several peripheral blood indexes-including T cells (CD4 + , CD8 + Tlymphocytes), systemic in ammation (neutrophil-to lymphocyte ratio (NLR), absolute neutrophil count (ANC), C-reactive protein (CRP), and lactate dehydrogenase (LDH)) have been associated with response or survival outcomes in patients with melanoma and non-small cell lung cancer (NSCLC) receiving immune checkpoint inhibitors (ICIs) [10][11][12][13][14][15]. In addition, lipid metabolism has been demonstrated to have a crucial role in the promotion of migration [16], invasion [17] and be related with tumor immune milieu [18]. However, it remains unclear whether peripheral blood pro ling could detect responses to anti-PD-1 immunotherapy in mCRC with MSI-H tumors. Thus, this multicenter study analyzed 41 mCRC patients with dMMR tumors to investigate the potential association between peripheral biomarkers with response to anti-PD-1 immunotherapy.

Patients
A total of 41 patients with mCRC with dMMR tumors treated with anti-PD-1 inhibitor (nivolumab, pembrolizumab, triprizumab, toripalimab and camrelizumab) were identi ed at Sun Yat-sen University, the Sixth A liated Hospital and Sun Yat-sen University Cancer Center between December, 2016 to December, 2019 ( Figure 1). The end of follow-up was June 30, 2020. The study was approved by the institutional review board of Sun Yat-sen University, the Sixth A liated Hospital. Written informed consent from patients was waived due to the retrospective nature of our study.
Pretreatment values were de ned as the values obtained before the initiation of anti-PD-1 immunotherapy.

Statistical analysis
Tumor response was evaluated according to the Response Evaluation Criteria in Solid Tumors, version 1.1. Progression-free survival (PFS) was de ned as the duration from the date of immunotherapy initiation to clinical or radiographic progression or death. Overall survival (OS) was de ned as the duration from the date of immunotherapy initiation to death. Fisher's exact test and Mann-Whitney U test were performed to compare distribution between groups based on response for categorical variables and continuous variables, respectively. Univariate Cox regression model was performed to estimate the hazard ratios (HRs) and 95% con dence intervals (CIs) of survival based on clinicopathologic parameters and peripheral blood indexes. The receiver operating characteristics (ROC) curve analysis was used to determine the cut-off point for the continuous variables including peripheral blood parameters. Kaplan-Meier method was used to perform survival analysis, with P values compared by the log-rank test. Only parameters with statistical signi cance in univariate analysis were included in multivariable analysis.
HRs and 95% CIs of survival was estimated multivariate Cox regression models. A two-tailed P value < 0.05 was considered statistical signi cance. All statistical analyses were performed in R software (version 3.5.1; http://www.Rproject.org).

Patients characteristics
A total of 41 mCRC patients with dMMR tumors were identi ed to be treated with PD-1 inhibitors. Clinical outcomes are depicted in supplementary Table S1. Overall, 4 patients achieved a complete response (CR), 19 patients achieved a partial response (PR), 10 patients achieved stable disease (SD), and 8 patients achieved a progressive disease (PD), which led to an overall response rate (ORR) of 56% (23/41). Patients clinicopathological characteristics are detailed in Table 1 Association between biomarkers and objective response Characteristics and overall response (OR) were compared between responders (CR/PR) and nonresponders (SD/PD). frequency of CD4+ T cell and CEA as continuous variables were signi cantly associated with ORR (all P values <0.05, Table 1), while other investigated parameters were similar in spite of the signi cant association between lower level for the ratio of CD4/CD8 T cells with ORR (P=0.03). For mutation data, KRAS or BRAF mutations did not show any signi cant difference (Table 1). Frequency of CD4+ T cell, ratio of CD4+/CD8+, high dense lipoprotein (HDL), and ApoA1 were associated with PFS in a univariate Cox regression model ( Table 2). Frequency of CD4+ T cell, ratio of CD4+/CD8+, NLR, high dense lipoprotein (HDL), and ApoA1 were associated with OS in a univariate Cox regression model (Table S2). Using ROC curves, the cut-off values of the above variables for PFS were identi ed (Table 2, Figure S2). The potentially survival-related factors (HDL, ApoA1, NLR) were not signi cantly associated with frequency of CD4+ T cell or ratio of CD4+/CD8+ (Table S3). Frequency of CD4+ T cell and ratio of CD4+/CD8+ remained signi cance in multivariate analysis for both PFS and OS (Table 3, Table S4). The optimal predictive cut-points of CD4+/CD8+ratio and frequency of CD4+  Figure   2). Log rank analysis revealed that lower level of CD4+/CD8+ ratio was associated with a better PFS (P=0.002) and OS (P=0.007) (Figure 2 Figure 3). Log rank analysis revealed that lower level of frequency of CD4+ T cell was associated with a better PFS (P=0.017) and OS (P=0.0495) (Figure 3). In a multivariate analysis, frequency of CD4+ T cell remained the signi cance in predicting PFS (P=0.02, HR=4.83, 95% CI=1.28-18.27) and OS (P=0.025, HR=16.21, 95% CI=1.43-184.2) ( Table 3, S4). Furthermore, NLR was signi cantly associated with OS in univariate and multivariate analysis. HR= hazard ratio, CI= con dential interval, CEA=carcinoembryonic antigen, CRP=C-reactive protein, LDH=lactate dehydrogenase, ALB= albumin, NLR=neutrophil-to-lymphocyte ratio, PLR=platelet-tolymphocyte ratio, LMR=lymphocyte-to-monocyte ratio, CHO=cholesterol, TG= triglyceride, HDL=highdensity lipoprotein, LDL=low-density lipoprotein, ApoA1=apolipoprotein A1, and ApoB= apolipoprotein B HR= hazard ratio, CI= con dential interval, HDL=high-density lipoprotein, ApoA1=apolipoprotein A1

Discussion
Anti-PD-1 immunotherapy is approved by FDA for refractory dMMR CRC. In the present multicenter cohort study, response data for 41 patients with dMMR mCRC treated with anti-PD-1 inhibitors were analyzed and potential blood parameters were identi ed as predictive biomarkers for response. Although tumor mutational burden has potential predictive value for anti-PD-1 therapy in MSI-H mCRC patients, the identi cation of additional peripheral blood biomarkers is crucial because the access and assay of biomarkers from blood is more easily than those from tumor tissue. Considering the limited experience with anti-PD-1 therapy in patients with dMMR mCRC, the evidence of potential blood biomarkers for these patients was scarce. To the best of our knowledge, the present study with 41 patients with dMMR mCRC is the rst to show that baseline level of frequency of CD4 + T cell and the ratio of CD4+/CD8 + are independent potential biomarkers for ORR and survival. Moreover, the present study indicated the potential prognostic value for NLR with regard to OS.
Previously, pretreatment counts of peripheral blood cells or LDH have been investigated as potential biomarkers for clinical outcomes in patients with melanoma [10,11,19] and non-small cell lung cancer (NSCLC) [13,14,20] treated with ICIs. Even though the present study showed that NLR could potentially predict OS, it failed to indicate the predictive value of NLR for ORR and PFS in anti-PD-1 therapy, which indicated dMMR mCRC might be distinct with melanoma or NSCLC. As far as we known, this study with 41 patients with dMMR mCRC is the rst to show that pretreatment frequency of CD4 + T cell and the ratio of CD4+/CD8 + are independent potential biomarkers for both ORR and survival. Our analysis thus showed that a low ratio of CD4 T cell (≤ 39.5) was signi cantly associated with a better ORR and PFS/OS in dMMR patients with mCRC. The potential mechanism may be that CD4 + lymphocytes are anergised rather than being stimulated and this might therefore correlate with a poor prognosis [21].
Moreover, the domain type of frequency of CD4 + T cell in the peripheral blood may be regulatory T cells, which have been recently reported to inhibit the anti-tumor activity of cytotoxic frequency of CD4 + T cell and then negatively affect the response and survival of patients undergoing anti-PD-1 immunotherapy [22]. Moreover, the ratio of CD4+/CD8 + is a predictor for ORR and survival. This may be explained not only by the potential pro-tumor activity of regulatory frequency of CD4 + T cell but also by the anti-tumor of frequency of CD8 + T cell. More frequency of CD8 + T cell in blood represents systematic anti-tumor immune features and they could migrate to the tumor site, lymph nodes and distal sites to enhance anti-cancer ability [17,23,24], which was consistent with the ndings from a recent study [12] to investigate the peripheral cell to predict response to anti-PD-1 immunotherapy in melanoma. They found the reduction of frequency of CD8 + T cell in the peripheral blood responders as compared to that in the blood of non-responders, which also indicated the crucial role in response to anti-PD-1 therapy. Moreover, NICHE trial indicated that increase of frequency of CD8 + T cell counts in CRC might re ect an underlying immune activation [4].
Although our data revealed that HDL and ApoA1 were signi cantly associated with PFS and OS in univariate analysis, the signi cance of HDL and ApoA1 did not maintain in multivariate analysis. ApoA1, as a prominent protein component in HDL, not just has antiapoptotic, anti-in ammatory, and antioxidant functions [25], but also alters tumor associated macrophages (TAMs) from a protumor M2 to an antitumor M1 phenotype[26] and modulates regulatory T cells [27]. A very recent study[18] also inferred that high ApoA1correlated with higher TIL, which might be the reason for its potential positive impact on PFS.
Limitations of the present study include relatively small sample size of patients and its retrospective nature. Another limitation lies in the absence of external validation of the associations detected in the present study, which need further large-scale study to validate our ndings. Since the present study has not performed the associations for other variables especially TMB that has been indicated predictive value in dMMR cancers, future integrative analysis of circulating immune-based biomarkers with genomic and epigenetic biomarkers for clinical response or survival and prospective trials of MSI-H cancers are warranted to validate their predictive potential. In addition, the speci c subtypes of peripheral leukocyte excluding CD4 + and CD8 + immune cells have not been analyzed, although these immune cells had different roles and prognoses in response to anti-PD-1 therapy. Thus, these ndings require high content data-generating technologies to provide potential mechanism for the circulating immune system and its correlation with the tumor-immune microenvironment.
This is the rst multicenter study to reveal that frequency of CD4 + T cell and ratio of CD4+/CD8 + are biomarkers to predict response to anti-PD-1 therapy and survival within an dMMR population. The nding indicates that patients with very low frequency of CD4 + T cell or low ratio of CD4+/CD8 + might respond well, and this subset of patients might be further selected to receive rst-line treatment with anti-PD-1 immunotherapy, which was consistent with the recent concept that anti-PD-1 immunotherapy is moved to rst-line treatment for mCRC [28]. These ndings might provide a potential explanation for the variability in response to anti-PD-1 inhibitors in numerous prospective clinical trials in dMMR mCRC and support the potential predictive role of frequency of CD4 + T cell and ratio of CD4+/CD8 + in anti-PD-1 immunotherapy.

Conclusions
Ratio of CD4+/CD8 + and frequency of CD4 + T cell might be crucial independent biomarkers within dMMR mCRC to better identify patients for response to PD-1 inhibitors. If validated in prospective clinical trials, ratio of CD4+/CD8 + and frequency of CD4 + T cell might provide aid in guiding the treatment of PD-1 inhibitors in dMMR mCRC.

Declarations
Ethics approval and consent to participate The study was approved by the institutional review board of Sun Yat-sen University, the Sixth A liated Hospital. Written informed consent from patients was waived due to the retrospective nature of our study.

Consent for publication
Not applicable.

Availability of data and materials
The data used and/or analyzed during the current study are available from the corresponding author on reasonable request.  Frequency of CD4+ T cell is predictive of response and survival outcome. Optimal cut-off point was calculated by receiver operating characteristics (ROC) curve analysis to dichotomize patients into high and low groups. (A) Frequency of CD4+ T cell distribution is visualized by a histogram between treatment response groups (CR, complete response; PR, partial response; SD, stable disease; PD, progressive