Immune checkpoint inhibitors (CPIs) that block PD-1/PD-L1 or CTLA-4 have transformed the treatment of many cancers and their use continues to expand. Anti-CTLA-4 was approved for treatment of melanoma followed by inhibitors of PD-1 or PD-L1, which have been approved to treat, in addition to melanoma, multiple tumor types [1]. In addition to use as single agents, dual CTLA-4/PD-1 inhibiting regimens have been approved for melanoma and renal cell carcinoma [1].

The anti-tumor mechanisms of action of CPIs involve relief of negative T cell costimulatory signals [2]. However, this effect is not specific to anti-tumor T cells, and the role of B cells, which also express PD-1, in anti-tumor immune response during CPI therapy has not been well established. Activity of CPIs in patients with underlying B cell deficiencies, such as X-linked agammaglobulinemia, immunoglobulin deficiencies, and common variable immunodeficiency has not been studied, as these conditions are rare, and these patients have typically been excluded from clinical trials for oncologic drugs due to their underlying immune disorders.

PD-1 inhibitors have been studied in patients who have been treated with B cell depleting antibodies, such as patients with B cell lymphomas previously treated with rituximab [3]. A number of ongoing clinical trials are assessing the activity of PD-1 inhibitors in combination with rituximab in B cell lymphomas, such as NCT03401853 and NCT02446457, and others. Activity has been seen in clinical trials of PD-1 inhibitors following rituximab therapy [3], yet it remains unknown whether the B cell depletion impedes the anti-tumor activity in this setting. B cell depleting drugs are being used for autoimmune disorders such as rheumatoid arthritis and multiple sclerosis, and the safety of their continued use in patients with malignancies requiring treatment with PD-1 inhibitors has not been explored.

T cell stimulation in many cases is not specific to tumor-infiltrating T cells, and new clinical problems, immune related adverse events (irAEs), have emerged. The precise mechanisms of these irAEs are not well understood, and likely vary by toxicity [4]. PD-1 is expressed on B cells, including immunosuppressive B regulatory cells and has a well-established role in B cell tolerance [5]. Interestingly, changes in B cells induced by checkpoint inhibitors have been shown to specifically correlate with irAEs [6].

High grade (grade 3–4) irAEs are seen in ~ 27% of patients treated with anti-CTLA-4, the most common being hepatotoxicity, dermatitis, and colitis [4]. Grade 3–4 irAEs occur in 10–15% of patients treated with PD-1/PD-L1 inhibitors and 55% of patients treated with the combination [4]. Autoimmune endocrine disorders such as insulin-dependent diabetes, and neurologic disorders such as multiple sclerosis, Guillain-Barre Syndrome and myasthenia gravis can also occur [7–9]. While most irAEs are reversible, some can result in long term morbidities, such as brittle diabetes [9]. Moreover, irAEs can be lethal in the case of pneumonitis and myocarditis; most trials report a treatment related death rate of < 1% for patients treated with anti-PD-1 monotherapy [4]. Although severe irAEs are fairly rare, this highlights the importance of developing appropriate methods to modulate irAEs without impeding anti-tumor effects. Although many autoimmune toxicities are thought to be purely T cell-mediated, some irAEs such as select endocrinopathies, autoimmune hemolytic anemia (AIHA), bullous skin eruptions and some neurologic toxicities appear to be more specifically mediated by B cells through autoantibody secretion. This observation suggests that B cell depletion might be useful a strategy to treat certain irAEs. Limited case reports and/or small series have suggested that anti-B cell therapy can be effective in treating such irAEs (Additional file 1: Table S1).

Given the importance of determining the role of B cells in response to PD-1 inhibitors we first studied B cell density in tumors from melanoma patients treated with PD-1 inhibitors. Subsequently, we employed tumor bearing animal models treated with PD-1 inhibitors alone or with B cell depleting antibodies to assess the role of B cells in anti tumor immune responses. Similar experiments were also performed in muMT mice which lack mature B cells. Although we did not study the efficacy of B cell depletion in treating irAEs herein, these data have clear implications for future study of B cell depleting drugs for treating select irAE in patients undergoing CPI therapy.

We studied B cell content in melanoma tumors from 40 patients treated with PD-1 inhibitors. Tumor infiltrating B cells were fairly sparse in most samples, 0–131 cells per 0.6 mm histospot (median - 2.5, mean - 17.4), 3 histospots per patient. Figure 1 A-B shows examples of heavily and poorly infiltrated tumors. Seeing that B cell content can vary within a tumor, we used the highest B cell density in the three histospots to evaluate associations with outcome. Thirty-four patients were evaluable for response using RECIST1.1. The objective response rate was 32%. There was no association between B cell content and radiographic response (Fig. 1C). We defined “high” B cell content as above the median value in at least one histospot and found no difference in overall survival (P = 0.64) among tumors with high versus low B cell content by log rank statistics (Fig. 1D).Fig1Fig. 1

Tumor infiltrating B cell density is not associated with response to anti-PD-1. a, b Tumor is masked by antibodies to S100 and HMB45 conjugated to Cy2 (green), and the area of B cell infiltrate is determined by the relative area of CD20 positive cells within the tumor mask, identified by anti-CD20 antibodies conjugated to Cy5 (red). Nuclei are identified by DAPI (blue). Examples of high B cell density (a) and low B cell density (b) are shown. c B cell quantification in responders (CR: complete response, PR: partial response) (n = 11) vs non-responders (SD: stable disease, PD: progressive disease) (n = 23), p = 0.22, by unpaired t-test. d Presence or absence of tumor infiltrating B cells does not correlate with overall survival, log-rank p = 0.64

Our purpose was to determine whether B cells play an essential role in anti-tumor activity of PD-1 inhibitors, and whether use of B cell inhibitors hampers anti-tumor activity of anti-PD-1. While the importance of tumor-infiltrating T cells in anti-tumor immunity is indisputable, the role of B cells is less clear. B cell depletion could affect production of auto-antibodies to melanoma antigens; which could impair anti-tumor immune responses given the observations that in melanoma auto-antibodies have been shown to be associated with improved outcomes, higher tumor-specific CD8 counts and response to ipilimumab [18]. Endogenous immunoglobulins can also regulate antibody dependent cellular cytotoxicity (ADCC) and could be impaired after rixutimab administration. B cells are also known to modulate T cell responses directly by presenting antigens and activating T cells and indirectly by modulating myeloid and/or antigen presenting cell function. Despite all of these considerations, it is unclear if the immune-stimulating role of B cells in advanced melanoma is clinically meaningful, particularly in the setting of metastatic disease and response to anti-PD-1. In humans, CD20 targeting drugs such as rituximab cause apoptosis of mature B cells and lead to an absence of circulating B cells for about 6 months after treatment. However, rituximab spares stem cells and plasma cells and typically does not affect overall immunoglobulin levels [19].

Conversely, some studies have suggested that B cell depletion might actually have anti-cancer effects in certain contexts [20]. For example, anti-CD20 treatment in mice bearing squamous cell carcinoma potentiated the efficacy of chemotherapy and enhanced tumor infiltration by CD8+ T cells [21]. Tumor-associated B cells have even been proposed to promote resistance to BRAF and MEK inhibitors in melanoma through secretion of IGF-1 and clinical trials evaluating therapeutic B cell depletion in this context are underway (NCT01376713) [22]. Immunoregulatory tumor infiltrating B cells (Bregs) have been shown to inhibit anti-tumor T cell responses in some models through IL-10 production [23]. These considerations suggest that B cell depletion might not only be neutral, it might even enrich anti-tumor immunity in certain contexts.

Seeing that core biopsies often do not reflect cellular content of entire tumor, we employed three cores per patient in this study, using the highest value obtained from the cores for our analysis. We found similar B cell content in tumors from responsive and non-responsive patients treated with anti-PD-1, suggesting that elimination of B cells in humans might not significantly affect anti-tumor immunity induced by PD-1 inhibitors. We corroborated this in several murine models. The anti-PD-1-sensitive MC38 model and the UV-mutagenized, neoantigen rich, YUMMER1.7 melanoma model were used. We showed partial B cell depletion with anti-CD20 did not impair anti-tumor immunity in either model. Spontaneous rejection of YUMMER1.7 in muMT mice, which lack B cells, both in tumors and in the circulation and lymphoid organs, was not impaired. Administering anti-PD-1 in muMT mice bearing MC38 or YUMMER1.7 tumors revealed no difference in complete response rate or survival compared to WT C57BL/6 J mice. These data suggest that B cell function may not be strictly required for either effective endogenous or anti-PD-1 elicited anti-tumor immune responses. Further, these results suggest that study of B cell-targeting therapies to treat select B cell-dependent irAE in patients receiving CPI therapy is warranted. It should be noted that despite the absence of mature B cells, muMT mice have been shown in some studies to still produce some immunoglobulins [24].

As immune checkpoint inhibitors are becoming widely used in multiple tumor types, they are also being used in B cell malignancies. Many patients with B cell malignancies have received B cell eliminating drugs, raising the question of whether this is determental for anti-tumor activity. Moreover, use of PD-1 inhibitors in patients with underlying autoimmune disorders treated with B cell eliminating drugs who develop cancer has not been studied, although case-series suggest that in some it is tolerated [25]. Once CPIs are initiated in patients with underlying autoimmunity, their disease might flare. Our data from the MC38 and YUMMER 1.7 models suggest that use of rituximab in this setting might not be detrimental and requires further study in humans. Finally, our results from the muMT mice bearing MC38 or YUMMER1.7 tumors suggests that patients with underlying B cell deficiencies or prior therapy with B cell depleting drugs might similarly benefit from PD-1 inhibitors shold they develop malignancies.

With widespread use of CPIs for malignancies, the incidence of toxicities will likely rise, and previously undescribed irAEs, such as cardiac toxicities, have only recently been reported [26]. Although it is unclear whether mechanisms of classic autoimmune disorders are exactly recapitulated in patients treated by CPIs who develop autoimmunity, diseases such as AIHA and bullous pemphigoid which are bonafide antibody-mediated disorders, have been shown to to respond to rituximab when triggered by CPI [27, 28]. Others, such as autoimmune diabetes, rheumatoid arthritis, type I diabetes mellitus, and multiple sclerosis are thought to be predominantly T cell mediated. However, anti-B cell therapy in classic rheumatoid arthritis and multiple sclerosis have shown efficacy, further supporting the study of anti-CD20 antibodies in treating these irAE. Several groups have already reported successful use of rituximab for treatment of B-cell mediated irAEs as summarized in Additional file 1: Table S1.

In summary, this is the first study look for a potential association between B cell infiltrates and benefit from PD-1 inhibitors in melanoma. Although B cell content appeared to be slightly higher in patients who responded to anti-PD-1 than patients who did not, this did not reach statistical significance in this cohort. Furthermore, using two, well-established murine cancer models we found that B cell inhibition did not affect T cell dependent anti-tumor responses induced by PD-1 inhibitors in these models. Taken together, these studies support further research into the use and impact of anti-B cell therapy in patients receiving CPIs who have underlying B cell deficiencies, have received B cell depleting antibodies for underlying autoimmune disorders or B cell malignancies, or patients with select irAEs induced by PD-1 inhibitors.