Examining Peripheral and Tumor Cellular Immunome in Patients With Cancer

Immunotherapies are rapidly being integrated into standard of care (SOC) therapy in conjunction with surgery, chemotherapy, and radiotherapy for many cancers and a large number of clinical studies continue to explore immunotherapy alone and as part of combination therapies in patients with cancer. It is evident that clinical effectiveness of immunotherapy is limited to a subset of patients and improving immunotherapy related outcomes remains a major scientific and clinical effort. Understanding the immune cell subset phenotype and activation/functional status (cellular immunome) prior to and post therapy is therefore critical to develop biomarkers that (1) will predict if a patient will respond to immunotherapy and (2) are a result of immunotherapy. In this study, we investigated local (tumor) and peripheral (blood) cellular immunome of patients with melanoma, breast cancer, and brain cancer using a rapid and reliable standardized, multiparameter flow cytometry assay. We used this approach to monitor changes in the peripheral cellular immunome in women with breast cancer undergoing SOC therapy. Our analysis is unique because it is conducted using matched fresh tumor tissue and blood from patients in real-time, within 2–3 h of sample acquisition, and provides insight into the innate and adaptive immune cell profile in blood and tumor. Specific to blood, this approach involves no manipulation and evaluates all immune subsets such as T cells, B cells, natural killer (NK) cells, monocytes, dendritic cells (DCs), neutrophils, eosinophils, and basophils using 0.5 ml of blood. Analysis of the corresponding tumor provides much needed insight into the phenotype and activation status of immune cells, especially T and B cells, in the tumor microenvironment vs. the periphery. This analysis will be used to assess baseline and therapy-mediated changes in local and peripheral cellular immunome in patients with glioblastoma, breast cancer, and melanoma in planned immunotherapy clinical studies.


INTRODUCTION
Over the past 7 years, immunotherapy has dramatically changed the treatment landscape for many solid tumor and hematologic malignancies (1)(2)(3)(4)(5). Cancer immunotherapy strategies include cytokines, immune checkpoint blockade, adoptive T cell therapies, vaccines (including dendritic cell (DC) based cellular vaccines) and oncolytic viruses (6)(7)(8)(9)(10). In particular, immune checkpoint blockade with antibodies that block CTLA-4, PD-1 or PD-L1 have resulted in durable clinical responses in patients with advanced melanoma, lung, kidney, bladder, and colorectal cancer, but both primary and acquired resistance occur in the majority of patients (1)(2)(3)(4)(5)11). Several components of pre-existing anti-tumor immunity in patients have emerged as key regulators in determining sensitivity to checkpoint blockade (3, [12][13][14]. This suggests that the therapeutic efficacy of immunotherapies can be enhanced with novel combination therapies that promote immunological responses. Novel combination strategies can help overcome the resistance currently seen and may also provide rationale for expansion of these same combination therapies to other solid tumors (15).
Intratumoral injection of an oncolytic virus to promote local recruitment of immune cells to the tumor is being developed as a strategy to overcome resistance to immune blockade (15,16). We are investigating oncolytic poliovirus (oncPV) therapy, which has demonstrated unprecedented responses in the treatment of recurrent glioblastoma multiforme (GBM), with 21% of patients experiencing durable, long-term (>3 years) survival (17). Although this was remarkably successful given recurrent GBM is a uniformly fatal disease, the majority of patients did not respond to therapy. Therefore, identification of key immune biomarkers is essential to appropriately select patients that can benefit from therapy. These same patterns of cellular immunome may also elucidate the mechanism of tumor responses to the combination of oncPV plus immune checkpoint blockade. The ability to reliably examine the immune cell subset phenotype and activation/functional status (cellular immunome) of cancer patients before and after therapy is critical to the development of predictive markers.
Toward the goal of utilizing the cellular immunome to predict response to immune-based therapies, we analyzed matched blood and tumor tissue from melanoma, breast cancer and brain cancer patients. Previous studies have shown that peripheral immune subtypes can be predictive of intratumoral immune responses and serve as a prognostic factor in cancer patients (18). Analysis of blood and tumor was conducted in real-time, shortly after sample collection with minimal processing to prevent exclusion or death of innate immune cells (e.g., neutrophils, basophils, and eosinophils). This also allows for a true representation of all immune cells in blood and therapy-mediated changes in immune cells. The immune cell subsets that infiltrate the tumors and the percentage of total tumor tissue infiltrated by these cells can also be determined more precisely by immediate processing and analysis of tumor tissue. Additionally, we use pre-cocktailed, dried reagents from the same lot to maintain consistency between samples for every patient throughout therapy. In doing so, we reduce the cost of the process and eliminate operator error. Here we describe a method to query the patient's cellular immunome rapidly during the course of immunotherapy with the goal to advance the development of predictive immune biomarkers that can then be applied as a companion diagnostic in the clinic.

Patient Recruitment
Eligible patients, 18 years or older, who underwent SOC surgical resection of melanoma, breast cancer, and brain cancer were consented to research tumor tissue and blood collection though the Duke BioRepository and Precision Pathology Center (BRPC) or the Brain Tumor BioRepository (BTBR) at Duke University School of Medicine. All patients were selected based on ability to get both tumor tissue and blood at the same time. Breast cancer patients (n = 15) were diagnosed with localized disease and received SOC neoadjuvant chemotherapy followed by surgery. Melanoma patients (n = 9) included in this study had stage I-IV disease. Primary tumors or metastatic lymph nodes were resected during SOC therapy. Brain tumor patients (n = 3) were newlydiagnosed with grade III glioma or grade IV glioblastoma (GBM) and received SOC surgery. De-identified matched blood and tumor tissue from consented patients was obtained by the BRPC and BTBR and used for examination of the cellular immunome.

Blood and Tumor Processing
Blood was obtained by venipuncture and collected in Vacutainer collection tubes containing EDTA anticoagulant (BD Biosciences). Blood was rotated on shaker until flow cytometry analysis (10 min-4 h post collection).
Surgically resected tumors were collected and stored in MACS tissue storage solution at 4 • C (Miltenyi). Storage time was 2-16 h post-tumor collection. Tumor cells and tumor infiltrating immune cells were analyzed after tumors were processed using the Tumor Dissociation Kit and Gentle MACS mechanical dissociator (Miltenyi) following the manufacturer's recommendations. Single cell suspension was immediately analyzed by flow cytometry.

RESULTS
We analyzed matched blood and tumor tissue from patients with melanoma, breast cancer, and brain cancer. Figure 1 demonstrates basic panel analysis of whole blood (WB) obtained from melanoma (1A), breast cancer (1B), and brain cancer (1C) patients. We examined lymphocytes including B cells (CD19+), CD3+/CD4+ T cells, CD3+/CD8+ T cells, and NK T cells (CD3+CD56+). Total monocytes (CD14+CD16+) (Table S1), were then separated into classical monocytes (CD14+CD16-), intermediate monocytes (CD14+CD16+) and non-classical monocytes (CD14-CD16+). Basic phenotyping analysis revealed that brain tumor patients had 3-8-fold lower percentages of circulating lymphocytes compared to melanoma or breast cancer patients. This was observed in all of the brain cancer patients analyzed, and is in agreement with a recent study demonstrating that patients with newly-diagnosed GBM have very low numbers of T cells in peripheral blood (19).
Duraclone IM T cell panel was used to examine the phenotype and activation status of circulating T cells in melanoma, breast cancer and brain cancer patients (Figure 2 and Table S2). Immune cells were identified using CD45 and side scatter and CD3 was used to separate T cells within the lymphocyte gate followed by analysis of CD4 and CD8 expression within the CD3 subset. The analysis described below was conducted on both CD4 and CD8 T cells. Figure 2 depicts analysis conducted on CD4 T cells and Table S2 provides percentage of individual CD4 and CD8 T cell subsets. Human T cell subsets and differentiation status was assessed using the following cell surface markers: CD45RA, CCR7, CD62L, CD27, and CD28. Loss of CD45RA on naive cells is accompanied by a gain of CD45RO on differentiated cells, thus CD45RA is sufficient to identity naïve vs. differentiated cells (20,21). In Figure 2, CD4+ T cells were characterized for expression of CD45RA and CCR7 to define naïve (CCR7+CD45RA+), central memory (CCR7+CD45RA-), effector memory (CCR7-CD45RA-), and effector T cells (CCR7-CD45RA+, also referred to as T effector memory cells re-expressing CD45RA or T EMRA cells). CD27 and CD28 (receptors involved in T cell activation) are used to study cellular activation history and are present on naïve and central memory T cells but absent on effector memory and effector T cells (22,23). Expression of PD1 is normally associated with T cell differentiation and activation, but sustained expression may indicate exhausted T cells in cancer and chronic infections (24). CD57 expression allows the identification of terminally differentiated T cells that have limited proliferative capacity (25). We observed that patients with melanoma had higher percentages of effector T cells (these are CCR7-CD45RA+) as compared to breast and brain cancer patients (Table S2). We therefore examined the effector T cell population in melanoma patients for expression of PD1 and CD57 (markers used to distinguish T effector memory cells re-expressing CD45RA, T EMRA, Figure S1) (24). All cells had high expression of both PD1 and CD57 markers which is consistent with T EMRA phenotype suggesting a potential role for this cell type in melanoma.
Cell surface T cell receptor expression was analyzed in melanoma, breast cancer, and brain cancer patient blood using the DuraClone IM TCR panel (Figure 3 and Table S3). Lymphocytes were identified based on CD3, CD4, and CD8 expression. CD3+ cells were assessed for surface expression of HLA-DR, a marker used to identify activated T cells. CD3+ T cells were then assessed for expression of TCRαβ and TCRγδ. TCRαβ cells account for 95% of all T cells that are found in circulation. TCRγδ cells (which comprise 5% of total T cells) exhibit an innate immune cell-like phenotype and are thought to respond rapidly and broadly to foreign antigens compared to the TCRαβ subtype cells (26). TCRγδ+ T cells were further assessed for expression of TCRVδ1 and TCRVδ2 (27). Interestingly, TCRVδ2 are capable of professional phagocytosis, a characteristic that is not shared by the TCRαβ T cell subset (28). Despite reduced percentages of circulating lymphocytes, brain tumor patients exhibited the highest percentage of TCRVδ2 compared to breast cancer and melanoma patients.
Regulatory T cells (Tregs) and the ratio of regulatory T cells to conventional T cells in blood are used to measure therapy outcomes in patients treated with immunotherapy. To examine Tregs, lymphocytes were identified based on their CD3, CD4, and CD8 expression (Figure 4 and Table S4). CD3+CD4+ T cells were evaluated for Treg markers, including Foxp3, CD25, Helios1, and CD39 (29)(30)(31)(32)(33). Treg percentages in blood ranged from 0.83 to 2.83% in melanoma, 3.24 to 9% in breast cancer, and 0.26 to 8.31% in brain tumor.
Because we analyzed whole blood, we were able to examine populations that are excluded when analysis is conducted on peripheral blood mononuclear cells (PBMCs), namely granulocytes ( Figure 5 and Table S5). Granulocytes, which represent 45-75% of total leukocytes, control inflammatory responses and are the first responders to sites of infection. Granulocytes include neutrophils (45-75%), eosinophils (0-7%), and basophils (0-2%). The resolution of inflammation is directly tied to how long granulocytes, specifically neutrophils, persist in tissues (34). In the context of cancer, neutrophils are known to modulate the tumor microenvironment by exerting both anti-tumor and pro-tumor effects (35). The analysis presented in Figure 5 not only provides an overall assessment of granulocyte composition in blood, but also examines the  Table S1.

into T cells and B cells based on expression of CD3 (T cells) and CD19 (B cells). NK cells were defined as CD56+ and CD3-and NK T cells were defined as CD56+ and CD3+. CD3+ cells were separated into CD4+ T cells and CD8+ T cells. Range in frequency of cell subsets is shown in
activation state of neutrophils. Cells that were positive CD294 and negative for lineage markers CD3, CD19, CD56, and CD14 were gated and analyzed for expression of CD15. Eosinophils were characterized as CD294+CD15+ and basophils were characterized CD294+CD15-. Neutrophils were characterized as CD294-CD15+CD3-CD19-CD56-CD14-. Neutrophils were further analyzed for expression of CD62L (adhesion marker that is downregulated upon activation), CD11b (upregulated upon activation), and PDL1 (upregulated upon activation). Our data reveal that breast cancer and melanoma patients had similar percentages of neutrophils, granulocytes, and basophils. The activation status of neutrophils (%PDL1+ cells) was also similar in breast cancer and melanoma patients. Unlike melanoma and breast cancer patients, all brain tumor patients lacked circulating basophils and eosinophils. The absence of eosinophils and basophils was observed in all brain cancer patients. Proportionally, these patients have increased percentages of circulating neutrophils (Table S5).
Recent studies have highlighted the importance of B cells in cancer, especially in breast cancer (36)(37)(38)(39)(40)(41). B cells are known to contribute to antitumor immunity in two ways; produce antitumor antibodies and present antigens to T cells. However, an in-depth analysis of B cells is often not conducted and few studies examine B cell subsets and B cell activation in cancers. We used the DuraClone IM B cell panel to analyze the B cells in the blood of melanoma, breast cancer, and brain cancer patients (Figure 6 and Table S6). B lymphocytes were identified as CD45 and CD19 expressing cells (Figure 6). Their activation status was examined using CD27 and CD28 marker expression. Plasmablasts, which are recently activated antibody producing cells, lack expression of IgM and IgD and express high levels of CD27 and CD38. This analysis identifies naive B cells (IgD+CD27-), marginal zone-like/natural effector B cells (IgD+CD27+), class-switched memory B cells (IgD-CD27+), CD21 low CD38 low B cells, transitional B cells (IgM+CD38+), and plasmablasts (IgM-CD38+). Increased numbers of memory B cells and plasmablasts in blood and tumor would indicate B cell responses in patients with cancer (42). This would also highlight the need to examine changes in B cell receptor (BCR) clonality and diversity in patients with cancer undergoing therapy.
Dendritic cells (DCs) are potent antigen presenting cells (43) and have been used in multiple clinical trials across tumor types as cellular vaccines. Although DCs represent 1% or less of circulating lymphocytes, the importance of DC analysis in blood cannot be understated given their innate ability to respond to pathogen-and danger-associated molecular patterns and consequently their ability to stimulate naïve T and B lymphocytes. It is important to examine DC subsets and changes in DC subset frequency because DCs are critical in immunostimulation and immunoregulation. Using the FIGURE 2 | T cell subsets in peripheral blood of patients with cancer. Non-single events were excluded as described in Figure 1 and lymphocytes were gated based on CD45 expression vs. side scatter. (A) melanoma; (B) breast cancer; (C) brain cancer. Gating of T cells was done based on expression of CD3 and side scatter, followed by CD4 and CD8. CD45RA and CCR7 expression was used to identify naïve (CD45RA+CCR7), central memory (CD45RA-CCR7+), effector memory (CD45RA-CCR7-), and effector (CD45RA+CCR7-) T cells. PD1, CD57, CD27, and CD28 expression was examined in CD4 and CD8 T cells. Range in frequency of cell subsets is shown in Table S2. CM, Central Memory; E, effector; EM, Effector Memory; N, naïve.
DuraClone IM DC panel, we examined DC populations in the peripheral blood of cancer patients (Figure 7 and Table S7). We gated on HLA-DR+ (class II+) cells that were negative for all lineage (Lin) markers (CD3, CD14, CD19, CD20, CD56). These cells were then analyzed based on expression of the CD11c surface marker to identify HLA-DR+CD11c+ myeloid DCs (mDCs) and HLA-DR+CD11c-plasmacytoid DCs (pDCs). All mDCs were further divided into DC subsets as follows: CD11c+Clec9A+CD16-(normally 0.02-0.06% of PBMCs), CD11c+CD1c+CD16-(normally 0.3-0.8% of PBMCs), and CD11c+CD16+Clec9A-inflammatory mDCs (normally 0.75-2% of PBMCs). Expression of CD123 was used to confirm pDC phenotype (normally 0.3-0.8% of PBMCs). Although circulating DCs were found in both melanoma and breast cancer patients, brain tumor patients lacked all DCs in circulation and flow plots are therefore not presented in Figure 7. Interestingly, melanoma patients had an increased average of pDCs in circulation. pDCs are known to play an important role in melanoma and are actively being studied for their potential to induce type I IFN production and antitumor immunity (44).
Next we examined the local (tumor) cellular immunome in patients with melanoma, breast cancer and brain cancer. Surgically resected fresh tumor tissue (matched to blood from same patient) was processed into a single cell suspension within 16 h of tumor harvest followed immediately by flow cytometry. Figure 8 shows analysis of melanoma tumors, basic analysis of all tumor-infiltrating lymphocytes, T cell subset, Treg, and B cell analysis. Dead cells were excluded and CD45 was used to distinguish immune cells from tumor cells followed by examination of CD4+ and CD8+ T cells, CD19+ B cells and CD56+ NK T cells. We collected data from a total of nine patients presenting with localized stage I disease, stage IIB, stage IIIC, and metastatic stage IV disease. In Figure 8A we show overall composition of immune cell infiltrates found in a representative melanoma tumor. Overall lymphocyte infiltration in the analyzed tumors varied between 2.02 and 39.12% of total live cells. CD3+ T cells accounted for the majority of infiltrating lymphocytes 54.77-89.95%. In the shown example, B cells accounted for 36% of all lymphocytes. This large B cell infiltration was only observed in two of the patients. Three out of nine tumors showed low percentages of infiltrating B cells (∼3-6% of infiltrating lymphocytes) and the range in the other four tumors tested was 9-17% of all lymphocytes. Overall monocyte percentages were low in all of the tumors analyzed and the largest percentage of these cells were of the classical or intermediate phenotype (Table S8). NK Cells  Table S3. TCRab refers to TCRαβ and TCRgd refers to TCRγδ. and NK T cells observed in these tumors are also presented in Table S8. These percentages varied greatly between the tumors analyzed.
We analyzed T cell subsets to examine the phenotype and differentiation status of these T cells. In Figure 8B and Table S9, we show analysis of both CD4+ and CD8+ T cell subsets. Both subtypes had similar overall expression of CD57. PD1 expression was elevated in both T cell subsets, which is in agreement with previously published studies in melanoma patients (45). This discovery is not surprising given the success that immunotherapy treatments have had in melanoma patients with pre-existing anti-tumor immunity. Further analysis revealed that both T cell subsets express CCR7 and lack CD45RA, a hallmark of central memory T cells. These cells were also positive for both CD27 and CD28 cell-surface markers. Taken together these data show that although T cells are abundant in melanoma tumors, these cells display a central memory phenotype with significant expression of PD1.
Regulatory T cells are an important CD4+ T cell subtype that can impact anti-tumor immune responses (46). In the shown analysis, regulatory T cells accounted for ∼50% of all CD4+ T cells (Figure 8C), a phenotype that potentially contributes to anti-cancer therapy resistance.
In the two melanoma patients with high percentage of B cell infiltrates in the tumor, we further phenotyped the infiltrating B cell subpopulations. Data presented in Figure 8D demonstrates that these cells primarily display a naïve B cell phenotype. Plasmablasts and transitional B cells were scarce. It is unclear if these cells display a regulatory B cell phenotype. These studies are ongoing as part of a larger initiative and will be reported in a separate manuscript.
Breast cancer analysis was conducted with tumor tissue from 15 women presenting with localized disease who underwent surgery as part of SOC therapy. We processed all samples fresh in order to avoid cell death and cell surface marker alteration as a result of the freeze-thawing cycle (47). In Figure 9 we show overall flow cytometry analysis of infiltrating immune cells in a representative example. T cells account for the majority of infiltrating lymphocytes in all tumors (average and percentage ranges are shown in Table S8). Of all T cells, CD4 T cells were the predominant subpopulation observed in 11 out of 15 tumors. B cells (>10% of all lymphocytes) were present in five out of the 15 tumors tested. The remaining infiltrating immune cells were monocytes, NK cells, and NK T cells. It is important to note that immune cell infiltration in the majority of the breast tumors FIGURE 4 | Regulatory T cell analysis in peripheral blood of patients with cancer. Non-single events were excluded as described in Figure 1. (A) melanoma; (B) breast cancer; (C) brain cancer. CD4 T cell populations were identified based on expression of CD45, CD3, and CD4 expression. CD25, Foxp3, Helios1, and CD39 expression was assessed against CD4 expression. Range in frequency of cell subsets is shown in Table S4. examined was relatively low (1.59-10% of all live cells found in the tumor).
Next we analyzed CD4+ and CD8+ T cells for expression of CD57 and PD1. Data presented in Figure 9B and Table S9 show CD57 expression percentages in the analyzed breast tumors. PD1 expression was high in all of the tumor infiltrating T cells that were analyzed (>60.22 of all CD4+ or CD8+ T cells) suggesting exhaustion of these infiltrating cells and a potential role for checkpoint therapy in breast tumors with pre-existing infiltrating T cells. These cells displayed a central memory phenotype, similar to T cells observed in melanoma patients. Unlike CD4+ and CD8+ T cells observed in melanoma patients, T cells observed in breast tumors showed differential expression of the CD27 cell surface marker, where only half of the CD28+ cells were also positive for cell-surface CD27. Although a role has not been assigned to this cell population in breast cancer, previous studies in other tumor types suggest that CD27 co-expression on the surface of CD8+ T cells directly correlates with the ability of these CD8+ T cells to mediate tumor regression (48).
We were able to conduct a comprehensive analysis of the T cell subset in breast tumors obtained from patients who underwent SOC therapy, however, our analysis was limited with regards to B cell analysis. Although, a fraction of the breast tumors were infiltrated by significant numbers of B cells, we lacked sufficient tumor sample to complete a detailed B cell analysis. In future studies, we plan to focus specifically on the B cell subsets that are found in breast tumors.  Table S5.
Brain tumor immune analysis was conducted in three patients whose tumors were resected as part of SOC therapy. In Figure 10 we show a representative example of the immune infiltrates in brain tumors. Brain tumors were characterized by low percentages of infiltrating immune cells (Table S8). Lymphocytes accounted for <2% of all analyzed cells. This finding is not surprising given the low numbers of circulating lymphocytes in these patients and recent reports showing lymphocyte sequestration of T cells in the bone marrow of brain tumor patients (19). Unlike other tumor types analyzed, brain tumors showed high percentages of infiltrating NK cells (14.23-50.45% of all infiltrating lymphocytes), which suggests a potential role for this cell type in brain cancer.
Analysis of T cell subsets in brain tumor revealed that infiltrating CD4+ and CD8+ T cells expressed CD57 (Table S9). PD1 was also highly expressed on both CD4+ and CD8+ T cell subtypes suggesting T cell exhaustion in these tumors. Unlike the other tumor types analyzed, T cells infiltrating brain tumors display an effector memory phenotype based on their lack of CCR7 and CD45RA expression. Further analysis in larger cohort of brain tumor patients is currently ongoing in the context of immunotherapy clinical trials to examine how and which immune infiltrating cells impact therapeutic outcomes.
We analyzed melanoma, breast, and brain tumors for the presence of granulocytes and dendritic cells and observed that these cells were not present in the tumors.
The analysis that we have presented was developed to examine baseline and therapy-mediated changes in the peripheral (blood) cellular immunome in the context of therapy along with companion analysis in tumors before and after therapy when feasible. Therefore, to validate our peripheral cellular immunome analysis, we longitudinally examined blood from women with breast cancer undergoing SOC therapy. Analysis of blood was performed as described above using the DuraClone IM basic panel (Figure 11). Blood was obtained prior to chemotherapy (Figure 11A), post-chemotherapy at the time of surgery ( Figure 11B) and 2 months after surgery ( Figure 11C). Table S10, we observed changes in the peripheral cellular immunome during the course of treatment. The percentage of natural killer cells doubled post chemotherapy (at the time of surgery) but returned to pre-therapy levels 2 months after surgery. Non-classical monocytes (CD16+CD14-) increased >4-fold and the percentage of natural killer cells doubled post-chemotherapy (at the time of surgery) but both returned to pre-therapy levels 2 months after surgery. This preliminary analysis suggests that we can capture changes in the peripheral cellular immunome. Such an analysis will accurately allow us to monitor the baseline cellular immunome and longitudinal changes in all cellular subsets (including granulocytes) during the course of SOC chemotherapy, surgery, and immunotherapy. The reproducibility of this analysis is guaranteed because we use pre-standardized dried reagents from the same lot to allow precise measurement of changes in the cellular immunome.

DISCUSSION
The ability to precisely and reproducibly understand immune cell phenotype and activation status (cellular immunome) in tumor and blood is critical to assess if a patient has responded to therapy and to develop biomarkers of response to therapy. The analysis presented in this manuscript was done using paired tumor and blood samples from patients with cancer with the following objectives: (1) examine blood and tumor cellular immunome in patients with cancer; (2) determine feasibility of this approach across different tumor types, and (3) assess if the experimental protocol is reproducible and precise to allow longitudinal monitoring of the peripheral cellular immunome to measure changes in response to therapies. Although cost of real time analysis can be a concern during immune monitoring, a side by side comparison between real times analysis of samples and analysis of frozen PBMCs revealed comparable overall costs.
In this study we present analysis conducted in patients with melanoma, breast cancer, and brain cancer. Our data demonstrate that we can indeed reliably and accurately monitor patient cellular immunome in tumor and blood in a matter of 2-5 h (blood) and 2-16 h (tumor) after sample acquisition and get a complete profile that encompasses 77 immune subsets. The value of developing an assay that has a quick turnaround and can be conducted in real-time is important in biomarker development; specifically if we want to use blood  and tumor cellular immunome to drive decisions related to therapy. Conducting analysis of blood in real-time is important for many reasons. The use of fresh blood eliminates the need to purify and freeze peripheral blood mononuclear cells (PBMCs), which introduces variability during isolation, freezing, storage, and thawing procedures. The use of whole blood FIGURE 11 | Basic immune cell analysis in blood of women with breast cancer undergoing standard of care therapy. Blood was obtained from women undergoing standard of care therapy (chemotherapy and surgery) for breast cancer; (A) prior to therapy; (B) post-chemotherapy and pre-surgery; (C) 2-months post-surgery follow-up. The same patient was followed at each time point as described in Figure 1. Range in frequency of cell subsets is described in Table S10. Data are representative of analysis conducted in three women with breast cancer.
permits analysis of all cell populations in blood (including neutrophils, eosinophils, and basophils). For example, our analysis of granulocytes in Figure 5 provides a broad view of the complete profile of the peripheral cellular immunome profile, given that neutrophils represent ∼45-75% of all leukocytes. The analysis provides information on neutrophil activation status and PDL1 expression. PDL1 expression on neutrophils in patients ranged from 5.04 to 31.24% in breast cancer and 11.77-49.65% in melanoma (  (49). A recent study demonstrated a negative correlation between neutrophil and CD8 T cell levels in patients with non-small cell lung cancer (50).
Our data in Figures 1-11 and Tables S1-S10 provides a snapshot of the local (tumor) and peripheral (blood) cellular immunome in patients with melanoma, breast tumor, brain tumor. We also saw qualitative differences in our samples, for example lack of eosinophils and basophils and fewer lymphocytes in patients with brain tumor. These qualitative differences are highlighted in the results section in the context of the cell subset examined. These changes could be in response to ongoing therapies and studies to analyze these changes are currently underway. For example, we are currently investigating therapy-induced changes in women with breast cancer (n = 100). Specifically, we are monitoring the peripheral cellular immunome at baseline, post-chemotherapy, and pre-surgery and 2 months post-surgery to comprehensively address how the peripheral cellular immunome is impacted during SOC therapy. Also part of this planned study is the investigation of the local/tumor cellular immunome, in patients undergoing SOC surgery. An example of longitudinal analysis of the peripheral cellular immunome in women with breast cancer undergoing SOC therapy is presented in Figure 11. Here we present the basic phenotype panel analysis at three time-points: baseline, post-chemotherapy, and pre-surgery and 2 months post-surgery. Our analysis revealed an increase in NK cells and NKT cells, post-chemotherapy and these numbers returned to pre-therapy levels 2 months post-surgery. The monocyte compartment also changed with therapy. The monocyte compartment in whole blood has three subsets: CD14+CD16-classical monocytes (phagocytosis), CD14+CD16+ intermediate monocytes (proinflammatory, phagocytosis) and CD14-CD16+ non-classical monocytes (patrolling, antiviral, pro-inflammatory) (51,52). As shown in Figure 11 and Table S10, we observed an increase in non-classical monocytes post-treatment (presumably indicating inflammation following chemotherapy or surgery) and return to pre-therapy levels 2 months after surgery. An increase in the number of CD4+CD8+ T cells was also observed posttreatment (Figures 11A-C; panel 3, top right), with a return to pre-therapy numbers post-therapy. As such, double positive CD4 and CD8 T cells have been described in both healthy donors, patients with inflammatory disorders and patients with cancer, including breast cancer (53). What is not completely clear is the function of these T cells as they have been reported to be suppressive or cytotoxic. Clearly our next task is to elucidate the function of these cells and such studies are planned.
In ongoing studies we are also developing immune function analysis using whole blood. This ensures immune function analysis is done in the context of all cell types in blood, for example platelets, or soluble serum factors that may be relevant to immune function. Studies suggest that whole blood function more accurately represents in vivo immune competence than PBMC function (54)(55)(56). We did not examine the function of monocytes, NK cells, and T cells in this study. We have developed these panels and will present our findings in a follow-up study. Lastly, this study does not describe the macrophage compartment in the tumors. As such, the presence and relevance of macrophages in human tumors has been a subject of considerable research (57)(58)(59). We did observe the presence of CD14+ and CD16+ cells that displayed high side scatter within the tumors (Figure S2), which may be either macrophages or Tie-2-expressing monocytes (60,61). To elucidate this further, a macrophage-specific panel is being developed for melanoma, breast, and brain tumors (this panel includes antibodies to CD45, CD11b, CD16, CD14, CD68, CD163, HLA-DR, and CD66b). We plan to conduct a focused analysis of macrophages in 10-15 tumor tissues for each of these cancers and perform parallel immunohistochemistry (IHC) analysis to examine macrophages in the context of tumor microenvironment.
These studies were initiated to develop a robust method for cellular immunome monitoring in the context of our current immunotherapy clinical trials. In a recent clinical study with a novel recombinant oncolytic poliovirus, PVSRIPO, we reported unprecedented responses in the treatment of recurrent glioblastoma, with 21% of patients experiencing durable, longterm (>3 years) survival (17). Our team also demonstrated that the oncolytic poliovirus, PVSRIPO, infects, and kills tumor cells while simultaneously activating innate immune cells, which leads to efficient priming of adaptive antitumor immune response (62). In the clinical study in patients with recurrent glioblastoma (17), 79% of patients did not respond to PVSRIPO, making the identification of biomarkers that predict response to therapy and mechanisms of resistance critical areas of research. The development of rapid, reliable, and reproducible monitoring of the cellular immunome is required for immune biomarker development. We will use the analysis presented in this study in the planned clinical studies in patients with recurrent glioblastoma (ClinicalTrials.gov Identifier: NCT02986178), breast cancer (ClinicalTrials.gov Identifier: NCT03564782), and melanoma (ClinicalTrials.gov Identifier: NCT03712358).    (C) brain cancer. Tumors immune infiltrates from melanoma, breast, and brain tumor patients were analyzed as shown in Figures 8-10. All cells that were positive for CD14 with high SSC (side scatter) and distinct from blood monocytes were assessed for expression of CD14 and CD16.

ETHICS STATEMENT
Table S1 | The frequency of leukocytes (as percentage of singlets), lymphocytes (as percentage of singlets), CD19+ B cells, CD3+ T cells, NK T cells and NK Cells (as percentage of all lymphocytes), CD4+ and CD8+ T cells (as percentage of CD3+ T cells), monocytes (as percentage of all leukocytes) and monocyte subsets (as percentage of all monocytes). Data shows average for all patients and low-high range. Table S2 | The frequency of CD4+ or CD8+ T cell subsets (as percentage of CD4+ or CD8+ T cells). Data shows average for all patients and low-high range.      Table S7 | The frequency of Lin-HLADR+ cells (as percentage of singlets), pDCs (as percentage of Lin-HLADR+ cells), mDCs (as percentage of Lin-HLADR+ cells), CD16+, CD11c+Clec9A+ CD16-, and CD11c+CD1c+CD16-subsets (as percentage of mDCs). Data shows average for all patients and low-high range. Table S8 | The frequency of leukocytes (as percentage of singlets), lymphocytes (as percentage of singlets), CD19+ B cells, CD3+ T cells, NK T cells and NK Cells (as percentage of all lymphocytes), CD4+ and CD8+ T cells (as percentage of CD3+ T cells), monocytes (as percentage of all leukocytes), and monocyte subsets (as percentage of all monocytes). Data shows average for all patients and low-high range. Table S9 | The frequency of CD4+ or CD8+ T cell subsets (as percentage of CD4+ or CD8+ T cells). Data shows average for all patients and low-high range.
Table S10 | The frequency of leukocytes (as percentage of singlets), lymphocytes (as percentage of singlets), CD19+ B cells, CD3+ T cells, NK T cells and NK Cells (as percentage of all lymphocytes), CD4+ and CD8+ T cells (as percentage of CD3+ T cells), monocytes (as percentage of all leukocytes) and monocyte subsets (as percentage of all monocytes). Data shows average for all patients and low-high range. The patients presented in this study had varying degrees of response to chemotherapy. Patient 1 exhibited a complete pathologic response, patient 2 had a partial response to chemotherapy with sub-millimeter foci of residual tumor in the tumor bed and patient 3 had a minimal response to chemotherapy.