Immune checkpoint inhibitors (ICIs) targeting programmed cell death1 (PD­1) and its ligand (PD-L1) have greatly improved therapeutic outcomes in multiple carcinomas including advanced non-small cell lung cancer (NSCLC). However, the limited benefit population of ICI monotherapy makes it necessay to screen predictive biomarkers for stratifying the patients. Currently, PD-L1 expression on tumor cells has been used in clinic as an indicative biomarker (1). In addition, a high tumor mutation burden (2, 3), intratumor immunological signatures (4, 5), as well as increased diversity of gut microbiota (6, 7) are also reported to be associated with better responses to ICI treatment. Nevertheless, due to the difficulties in the collection of tumor biopsies from cancer patients and insufficient biomarkers to identify benefit patients, it is still necessary to investigate novel indicators associated with the responses to ICI treatment.

Peripheral blood is the most widely used specimen in clinical diagnosis. ICI treatment is demonstrated to affect the peripheral immune profiles. For instance, PD-1 blockade induces Ki67 expression in a subset of peripheral PD-1⁺CD8⁺ T cells and the expansion of neoantigen-specific T cell clones after anti-PD-1 treatment, which suggests the restoration of systematic anti-tumor immunity of CD8⁺ T cells after receiving ICI immunotherapy (8, 9). Therefore, several peripheral CD8⁺ T cell signatures have been reported to be associated with clinical outcomes of anti-PD-1/PD-L1 immunotherapy (8, 10). Our previous study also illustrated that high percentages of IFN-γ-producing naïve CD4⁺ T cells and PD-1⁺CD4⁺ memory T cells were associated with better response to anti-PD-1 treatment in NSCLC patients (11). Recently, B cells and tertiary lymphoid structures (TLSs) in the tumor microenvironment (TME) are identified to promote the responses to ICI treatment, illustrating the significance of B cells in anti-PD-1/PD-L1 immunotherapy (12–14). Whether systematic B cell signatures are associated with the responses to ICI therapy needs to be further explored.

In this study, peripheral B cell signatures from advanced NSCLC patients receiving anti-PD-1/PD-L1 monotherapy were systematically assessed by multiplex flow cytometry. Peripheral IgM⁺ memory B cells were identified to be associated with the responses to anti-PD-1 monotherapy in advanced NSCLC patients, providing the evidence on peripheral B cell signatures as a potential biomarker for stratifying the patients before the treatment.

T cells including CD8⁺ and CD4⁺ T cells play significant roles in mediating durable anti-tumor immunity (18–21), making them feasible in predicting the responses to ICI immunotherapy in clinic (8, 11, 22–24). Although B cells are mostly dedicated to humoral immunity against microbial infections, recent findings also illustrate response prediction significance of tumor-infiltrating B cells and TLSs in the patients with melanoma and sarcoma upon ICI treatment (12–14). In the present study, we have investigated the potential of peripheral B cells in predicting the responses to anti-PD-1/PD-L1 treatment in Chinese advanced NSCLC patients. In line with the findings in the TME, our results show that the percentage of CD19⁺ B cells at the baseline is higher in R than that in NonR group. What is more, the percentage of IgM⁺ memory B cells at the baseline is the most potential signature in predicting the responses to anti-PD-1 monotherapy.

B cells exert anti-tumor immunity mainly through antigen presentation and cytokine secretion for T cell activation (17, 25, 26). Although the exact mechanisms of B cells in facilitating the responses to ICI therapy need to be further investigated, it has been proposed that B cells as well as TLSs in tumor regions played critical roles in the activation of regional tumor-specific T cells through presenting tumor antigens (12). We also compared the infiltrating B cells in tumor biopsies between R (n = 3) and NonR (n = 12) patients before the treatment. No significant difference was observed between these two groups ( Figure S4 ), which is similar to the results by Helmink BA et al. (13). Other studies also revealed the properties of infiltrating B cells associated with the responses to ICI treatment. For instance, clonal counts for both immunoglobulin heavy and light chains as well as BCR diversity were increased in the responders than those in the non-responders of melanoma patients upon neoadjuvant ICI therapy (13). Additionally, tumor associated B cells were found to sustain the inflammation in melanoma, which also facilitated the responses to ICI therapy (27). These results therefore provide the possible mechanism of infiltrating B cells on anti-tumor immunity in the TME.

In this study, we have demonstrated that high percentages of peripheral CD19⁺ B cells before the treatment were associated with a longer PFS in advanced NSCLC patients receiving nivolumab monotherapy. Compared to CD8⁺ T cells or NK cells, B cells account for a large proportion in PD-1 expressing lymphocytes ( Figure 1A ). PD-1 is previously reported to be engaged in regulating B cell-mediated humoral immunity (28, 29). We have also reported that blockade of PD-1 in mice augmented humoral immunity with the accumulation of germinal center B cells (GCBs) and memory B cells in the spleens together with elevated percentages of plasma cells (30). We therefore determined IgG and IgM levels in the patients receiving anti-PD-1 treatment in some patients of cohort 1, where we deduced that IgG and/or IgM might be also related to the efficacy of ICIs treatment. However, both IgG and IgM before anti-PD-1 monotherapy were comparable between R and NonR groups ( Figure S5 ). Interestingly, IgM⁺ B cells highly express HLA-DR when compared to other B cell subsets ( Figure S6B ). These data suggests that the effects of PD-1 blockade on human B cells might be more related to their antigen presentation capacity with the upregulation of MHC molecules (12) rather than their differentiation into antibody-producing plasma cells.

What is more, logistical regression analysis screened out IgM⁺ memory B cells as a noval indicator to predict the efficacy of anti-PD-1 monotherapy. Memory B cells are a B cell subset which is responsible for the maintenance of memory response upon microbial infection. Based on IgG and IgM expressions, they can be subdivided into long-lived IgM⁺ and class-switched IgG⁺ memory B cells. IgM⁺ memory B cells are demonstrated to play important roles in lasting long term immunity with less turnover (31). It has been reported that when encountering the same or similar antigens IgM⁺ memory B cells may reenter GCs to undergo further affinity maturation and isotype switch (32, 33). This subpopulation in the circulation has been associated with the outcome of the infections such as COVID-19 (34, 35). We have found that they expressed high level of PD-1 when compared to IgM⁺ mature B cells ( Figure 5B ). In addition, PD-L1 expression was much higher on IgM⁺ mature B cells than on both IgM^- (mainly IgG⁺) memory B cells and IgM⁺ mature B cells ( Figure 5C ). More interestingly, upon anti-PD-1 treatment, there exhibited the decrease in peripheral PD-1⁺IgM⁺ memory B cells in R group than in NonR group whereas no similar trend was observed in PD-1⁺IgM^- memory B cells (data not shown). This might be explained that PD-1⁺IgM⁺ memory B cells might be one of the targets of anti-PD-1 antibody, which is similar to PD-1⁺CD8⁺ T cells reported previously (8, 10). Herein, we did not define how IgM⁺ memory B cells modulate anti-PD-1 regimen-mediated anti-tumor immunity in NSCLC patients. Given that IgM⁺ memory B cells are the main subpopulation in the secondary responses with high expressions of HLA-DR ( Figure S6C ), they might be engaged in promoting anti-tumor immunity through remodeling the capacity of antigen presentation upon nivolumab immunotherapy. In the future, whether IgM⁺ memory B cells migrate into the TME and how they interplay with T cells to enhance anti-tumor immunity merit further investigation. This might shed light on the mechanisms of B cells contributing to enhance ICI therapy.

In this study, since all the patients recruited have received at least one or two-rounds of chemotherapy before anti-PD-1 monotherapy, we also analyzed the effects of prior chemotherapy on peripheral B cells. It was showed that the prior chemotherapy has little effects on the proportions of peripheral CD19⁺ B cells at the beginning of anti-PD-1 monotherapy as well as the prognosis ( Figure S7 ). Moreover, there was no significant difference in peripheral CD19⁺ B cell percentages between R and NonR patients at 2 and 8 weeks after the treatment ( Figures S3A, C ). No obvious changes were observed in the percentages of peripheral CD19⁺ B cells from the baseline to 8 weeks after immunotherapy either ( Figures S3E, F ). However, both peripheral CD19⁺ B cells and IgM⁺ memory B cell at the baseline were not associated with a longer OS ( Figure S8 ). This might be due to the fact that B cell subsets are not the populations undergoing direct cytotoxicity against tumor cells. They are more likely to perform indirect function such as presenting tumor antigen for T cell activation, which might be more related to the PFS rather than OS.

Another interesting observation we got from this study is that the prediction roles of peripheral IgM⁺ memory B cells for anti-PD-1 monotherapy is not suitable for anti-PD-L1 monotherapy ( Figure 4E ). One of the reasons might be the sample size in anti-PD-L1 monotherapy cohort (n = 30). Another reason might be due to the different mechanisms of anti-PD-1 and anti-PD-L1 treatment against tumor. This is also demonstrated by our previous studies both in CD4⁺ T cell subsets (11) or metabolic biomarker signatures (36) for response prediction of ICI treatment in the same cohort. Our findings warrant further exploration in the prospective study with a larger advanced NSCLC population, as well as other types of tumors (37).

In summary, our study identified peripheral IgM⁺ memory B cells with high PD-1 and PD-L1 expressions as an alternative potential indicator to predict the responses to anti-PD-1 monotherapy in Chinese advanced NSCLC. Since plenty of the signatures in the periphery have been identified independently including CD8⁺ T cells (8, 10), CD4⁺ T cells (24, 38), neutrophil-to-lymphocyte ratio (NLR) (39), monocytes (40), IL-8 (41, 42), LDH (43), as well as ctDNA (43, 44). How to integrate different signatures to establish multiparametric approaches to improve prediction efficacy will be significant in the future.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

All samples were collected in accordance with the Ethics Committee of Shanghai Chest Hospital-approved protocol (Number KS1732). All patients have provided written consent prior to blood collection.

YW, JY, and SL conceived and designed the study. LX, LG, WX, RS, and SZ performed the flow cytometric experiments. JK, YY, YXY, WL, and YG were working on the clinical sample collection. LX, LG, JK, and YW performed data analysis. HC and ZL offered assistance and helpful discussions for data analysis. YW, SL, JY, and LX interpreted the results and drafted the manuscript. All authors contributed to the article and approved the submitted version.

This research was funded by the National Key R&D Program of China (2016YFC1303300), National Natural Science Foundation of China (82073152, 81802264, 82030045), Technology Innovation Program of Shanghai (19411950500), Talent Training Program of Shanghai Chest Hospital in 2019, Incubation Project Plan for Research in Shanghai Chest Hospital (2019YNJCM07), and Shanghai Chest Hospital Project of Collaborative Innovative Grant (YJT20191015).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.