Immune Desert Research Articles

Abstract P2-08-35: A significance of SUVmax levels on FDG-PET as a prognostic factor may be mediated by local immune environment of breast cancer

Abstract (Background) SUVmax levels (maximum radioactivity concentration per a pixel) on FDG-PET reflect glucose uptake and it is clinically useful as a prognostic factor. It is reported that breast cancer with high levels of SUVmax causes insufficient glucose concentration in stromal tissue, which results in suppressed cytotoxic T-lymphocytes function. These data may indicate that the prognostic significance of SUVmax levels is influenced by local immune environment of breast cancer. The aim of this study is to investigate whether local immune responses of breast cancer affect correlation of SUVmax levels and prognosis. (Method) The 278 invasive breast cancer patients were recruited who underwent surgery at Hyogo College of Medicine Hospital and whose SUVmax levels in the breast were examined before surgery or neo-adjuvant therapy. The cutoff value of SUVmax levels was set at 3.585. Tumor infiltrate lymphocytes (TILs) were evaluated as a local immune response and the distributions of TILs were divided into three groups, inflamed (intra-tumoral lymphocytes, Inf), immune excluded (peri-tumoral lymphocytes, IE) and immune desert (very few lymphocytes, ID). During follow-up period (median 39 months), 21 patients relapsed. (Results) Relapse free survival (RFS) in the SUVmax-high group was significantly worse than in the SUVmax–low group (p=0.0026). There was no correlation between TILs distribution patterns and RFS. In the IE+ID group (175 patients) SUVmax levels were not correlated with prognosis, but in the Inf group (103 patients) RFS of SUVmax-high was significantly worse than of SUVmax-low (p=0.0051). In the multivariate analysis including nodal status and nuclear grade, SUVmax levels of the Inf group was an independent prognostic factor. (Discussion and conclusion) SUVmax levels in primary lesions were correlated with prognosis only in the Inf group and were not in the IE+ID group. A significance of SUVmax levels as a prognostic factor may be diverse depending on the local immune environment of breast cancer. A novel therapeutic strategy such as inducing suppression of glucose uptake in cancer cells is suggested for breast cancer with immune inflamed. Citation Format: Fujimoto Y, Higuchi T, Watanabe T, Hida AI, Imamura M, Kitajima K, Miyoshi Y. A significance of SUVmax levels on FDG-PET as a prognostic factor may be mediated by local immune environment of breast cancer [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P2-08-35.

Abstract P2-09-02: Modulation of the immune microenvironment in high risk DCIS by intralesional injection of anti-PD-1 (pembrolizumab)

Abstract Background. Ductal carcinoma in situ (DCIS) is a risk factor for the subsequent development of invasive breast cancer. Features of DCIS that are associated with a high risk of a subsequent event include large size (> 5 cm), high grade, comedo necrosis, palpable mass, hormone receptor negativity, and HER2 positivity. We have previously shown that immune infiltrates are positively associated with these high-risk features, suggesting that manipulating the immune microenvironment in high-risk DCIS, for example via checkpoint blockade, could potentially alter disease progression. Methods. In this phase 1 pilot study we investigated changes in the immune microenvironment of high risk DCIS after intralesional injection of anti-PD-1 (pembrolizumab). Study participants received 2 intralesional injections of pembrolizumab, 3 weeks apart, with surgery approximately 3 weeks after the 2nd dose. The study started with a dose of 2 mg/injection (1/100 of the standard 200 mg iv dose), then escalated to 4 mg and 8 mg, with 3 patients at each dose. Tissue samples from pre-treatment biopsies and post-treatment surgical resections were stained with two 6-plex immune panels using Opal immunofluorescence reagents (Perkin Elmer) on a fully automated Ventana Discovery platform, imaged with a Vectra 3 system and analyzed with inForm software (Perkin Elmer). An algorithm for tumor/stroma segmentation developed in inForm was used to randomly select 10 high power fields (hpfs) for imaging. Cell phenotype maps were generated for each of these hpfs for each sample. Cell densities were determined per area of stroma, DCIS, or total tissue and averaged across all hpfs for a given case. Spatial analyses were performed to quantitate co-localization of immune cells with DCIS cells. Results. The intralesional injections were easily administered and well tolerated. No systemic toxicities were observed at any dose. MRI imaging demonstrated no change in the size of lesions after treatment. Multiplex immunofluorescence analyses demonstrated heterogeneous responses ranging from dramatic increases in T cells, in particular CD8+ T cells, in cases which had a T cell infiltrate prior to therapy, to no post-therapy T cell infiltrate in cases with a pre-therapy immune desert. We also observed increases in B cells and macrophages and a decrease in the ratio of FoxP3+ T cells to CD8+ T cells, the latter mainly due to a significant increase in CD8+ cells, as opposed to a decrease in FoxP3+ cells. Spatial analyses indicated that in some cases, despite a marked increase in T cells post therapy, these cells did not co-localize with DCIS cells, indicating a state of immune exclusion. Conclusions. We have demonstrated the safety and feasibility of intralesional injection of an immune checkpoint inhibitor (pembrolizumab) in high risk DCIS. In some patients we observed a dramatic change in the immune microenvironment, with an increase in T cells, B cells, and macrophages, and a decrease in the FoxP3:CD8 ratio, even at a dose that is 1/100 of the standard intravenous clinical dose. An expansion study is underway in which patients will receive 4 injections of pembrolizumab at 3 week intervals prior to going to surgery to determine if more injections/time will increase response rate. Citation Format: Campbell MJ, McCune E, Rothschild H, Bolen J, VandenBerg S, Chien J, Wong J, Esserman L. Modulation of the immune microenvironment in high risk DCIS by intralesional injection of anti-PD-1 (pembrolizumab) [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P2-09-02.

Abstract 3689: Identification of candidate therapeutic targets in BCG unresponsive bladder cancer- inflammatory subtypes of BCG unresponsive bladder cancer

Abstract Intravesical immunotherapy with Bacillus Calmette-Guérin (BCG) is used for the first line treatment of high risk non-muscle invasive bladder cancer. Despite high initial response rates (70%), recurrence is a major problem and many patients develop BCG unresponsive disease, for which the primary treatment option is definitive surgery (cystectomy). In order to define the biological properties of BCG unresponsive disease, we performed whole transcriptome RNAseq on 29 matched tumors obtained from patients before and after the development of BCG resistance. Unsupervised cluster analysis revealed the presence of two clusters - BCG cluster 1, containing 8 pre- and 19-post BCG treatment tumors, and BCG cluster 2, containing 21 pre- and 10- post BCG tumors (Fisher's exact test, p<0.01). To characterize the biological properties of the two clusters, we extracted the significantly differentially expressed genes and analyzed them by Ingenuity Pathway Analysis (Sigma). Interestingly, multiple immune response pathways (T cell receptor signaling, IL8 signaling, IL12-mediated signaling, IFN-gamma pathways, immune cell trafficking, etc) were significantly enriched (p<0.01) in the BCG cluster 1 that was enriched for BCG unresponsive tumors. To determine the potential relationship between BCG clusters 1 and 2 and the previously defined basal and luminal molecular subtypes, we generated a batch-corrected bladder cancer meta-dataset combining two publically available datasets (GSE48075 and GSE32894) consisting of mixtures of non-muscle invasive and muscle-invasive tumors. We performed consensus cluster analysis using the meta-data and identified 3 distinct molecular clusters. The results indicated that the gene expression signature that characterized BCG cluster 1 was also present in tumors assigned to the basal/SCC-like and p53-like/infiltrated tumors defined previously. Together, the results indicate that the tumors in BCG cluster 1 may be enriched with T cells and therefore may have increased sensitivity to immune checkpoint blockade. However, it is likely that alternative therapeutic targets will need to be identified for BCG unresponsive tumors that display the ‘immune desert' phenotype characteristic of BCG cluster 2. Citation Format: Woonyoung Choi, Roger Li, Chinedu Mmeje, I-ling Lee, Shanna Pretzsch, Jolanta Bondaruk, Max Kates, Trinity Bivalacqua, Bogdan Czerniak, Ashish M. Kamat, Colin Dinney, Peter Black, David J. McConkey. Identification of candidate therapeutic targets in BCG unresponsive bladder cancer- inflammatory subtypes of BCG unresponsive bladder cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 3689.

Abstract 1023: Release of cancer cells from the prostate tumor microenvironment to promote neoantigen exposure and enhance anti-tumor immune response

Abstract Prostate tumors are widely considered to be immunologically silent and resistant to immunotherapy. PCa is widely considered to be an immune desert, with little infiltration of cytotoxic T lymphocytes (CTLs). However, anti-tumor immune responses have been seen in PCa in the right context. We believe we can boost immune response and T cell infiltration into PCa tumors by inhibiting the stromal derived factor 1-CXC chemokine receptor 4 (SDF1-CXCR4) signaling axis and releasing tumor cells from the immune-suppressive bone marrow niche. We have shown that CXCR4 antagonism causes PCa disseminated tumor cells (DTCs) to leave the bone marrow niche and enter the circulation as released circulating tumor cells (CTCs). Our hypothesis is that the release of PCa cells from the tumor microenvironment will enhance chemotherapy-induced cell death and the subsequent presentation of tumor-associated antigens, allowing increased anti-tumor response to immunotherapies. I generated a novel selection-free immunofluorescence-based protocol to confirm this in dissemination models of cancer in mice. After CXCR4 inhibition, I observed an approximate five-fold increase in the number of released CTCs in the blood of mice 24 hours after treatment. Very little research has been performed on the proliferation of CTCs or DTCs, particularly when released from the bone marrow niche. I have been able to mark proliferating cells in blood and bone marrow with EdU and quantify proliferating CTCs. We determined that while released cancer cells are proliferating more than DTCs, the ratio of released proliferating cells is not sufficient to cause a robust response to chemotherapy alone. This led to the hypothesis that released cells could provide an excellent substrate for the immune system. Evidence to support this idea is that we consistently observe the engulfment of CTCs by CD45-positive immune cells from the blood of mice, indicating that immune cells may respond to CTCs but not DTCs. Our approach is unique in that we are using CXCR4 inhibition to release cancer cells into the bloodstream to expose them to immune cells and initiate an immune response via activated CTL infiltration into PCa tumors. We believe this represents an innovative polyvalent autologous vaccine approach as it does not rely on a single antigen to stimulate CTLs, and also relies on each individual patient's cancer cells and immune cells. We have also generated some immunofluorescent protocols to determine the immune composition of circulating cells in relation to CTCs. While we believe that CTLs are responsible for most of the anti-tumor immune response, we are using this opportunity to discover novel immune cell types that may be present before and after therapy. This approach, when combined with classic cancer therapy and immunotherapy, may represent a novel way to boost response to cancer therapies in metastatic PCa. Citation Format: Kenneth C. Valkenburg, Kenneth J. Pienta. Release of cancer cells from the prostate tumor microenvironment to promote neoantigen exposure and enhance anti-tumor immune response [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 1023.

Abstract 3638: Quantitative assessment of tumor-infiltrating lymphocytes and immunotherapy outcome in metastatic melanoma

Abstract Background: Durable responses to immune checkpoint blockade against programmed cell death 1 (PD-1) are limited to a subset of metastatic melanoma patients, so there is a need for predictive biomarkers. Tumor-infiltrating lymphocytes (TILs) are the major cellular target of anti-PD-1 therapy, so we hypothesized that pretreatment TIL profiles would be associated with response. Methods: Pretreatment whole-tissue sections from 94 melanoma patients treated with anti-PD-1 therapy (pembrolizumab, nivolumab, or ipilimumab plus nivolumab) from 2011-17 were collected from Yale Pathology archives, and stained to visualize nuclei (DAPI) and melanoma cells (S100 & HMB45) in combination with two multiplex immunofluorescence panels to: (1) perform TIL quantitation of helper T cells by CD4 (SP35, Spring), cytotoxic T cells by CD8 (C8/144B, Dako), and B cells by CD20 (L26, Dako); and (2) assess TIL activation by identifying T cells by CD3 (SP7, Novus), cytolytic activity by GZMB (4E6, Abcam), and proliferation by Ki67 (MIB-1, Dako). Cell phenotyping and counting were performed using inForm software (PerkinElmer) and protein expression was measured by the AQUA method of quantitative immunofluorescence (QIF). Response Evaluation Criteria In Solid Tumors (RECIST) 1.1 were used to classify best overall response as complete response (CR), partial response (PR), stable disease (SD), or progressive disease (PD). Objective response rate (ORR; CR/PR), disease control rate (DCR; CR/PR/SD), and progression-free survival (PFS) were correlated with TIL parameters measured by both methodologies. Results: Pretreatment lymphocytic infiltration, by cell counts or QIF, was significantly higher in CR/PR than in SD/PD, particularly for CD3 and CD8 (P < 0.0001). Neither TIL activation (CD3 high, Ki67 and/or GZMB high) nor TIL dormancy (CD3 high, Ki67 and GZMB low) was significantly associated with outcome. Multivariable analyses revealed significant CD8 associations (HR > 3) with PFS independent of age, sex, mutation, stage, treatment, and prior immune checkpoint blockade, which accounted for similar CD3 PFS associations as expected. The favorable predictive performance of CD8 cell count (and QIF) had an area under the receiver operating characteristic (ROC) curve of 0.75 for ORR and 0.78 for DCR, which reached 0.83 (ORR or DCR) for dual therapy (ipilimumab plus nivolumab). Interestingly, in contrast to previous hypothetical classifications, there were a number of responders (CR/PR) in the CD3 or CD8 low (immune desert) category. Conclusions: Pretreatment lymphocytic infiltration, by cell counts or QIF, is associated with anti-PD-1 response in metastatic melanoma. Multiplex analysis of the tumor microenvironment has the potential to be used as a companion diagnostic test for precision immunotherapy. Citation Format: Pok Fai Wong, James W. Smithy, Kim R. Blenman, Harriet M. Kluger, David L. Rimm. Quantitative assessment of tumor-infiltrating lymphocytes and immunotherapy outcome in metastatic melanoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 3638.

Abstract 5070: Chemokine expression signatures in infiltrated vs non-infiltrated tumors

Abstract Background: Previous studies in animal models have shown that expression of specific chemokines determines immune cell infiltration in the tumor microenvironment. More specifically, low expression of CXCL9/10/11, CXCR3, and CCL5, coupled with high expression of CCL2, CCR2, CCR4, CCR5, CCL22, CXCL12, and CXCR4, leads to an exclusion of effector T-cells in the tumor microenvironment while allowing the entry of Treg and MDSC(s). Methods: 300 formalin-fixed, paraffin-embedded (FFPE) metastatic cutaneous melanoma samples were evaluated by the RNA-seq component of a comprehensive immune profile panel to measure transcript levels of chemokine genes. Resultant data was QC filtered, normalized and ranked based on an assorted reference population of various tumor types. CD8 expression was used to categorize tumors as inflamed (CD8 Rank ≥ 75), borderline ≥ 25 CD8 Rank < 75) and immune deserts (CD8 Rank < 25). T-cell infiltration is defined by CD8 immunohistochemistry with following definitions: Noninfiltrated - Sparse number of CD8+ T-cells that infiltrate nests of neoplastic cells and represent less than 5% of the tumor. Infiltrated - Frequent CD8+ Tcells that infiltrate nests of neoplastic cells in an overlapping fashion at least focally and represent more than 50% of the tumor cells. Results: CD8 infiltration by immunohistochemistry showed high correlation with CD8 gene expression by RNAseq with infiltrated tumors showing significantly higher expression of CD8 than non-infiltrated tumors (v.test:9.47, p= 2.79E-21; Wilcoxon rank sum test p<0.05). Moreover, 73% of inflamed tumors were categorized as “infiltrated” by IHC while 93% of Immune desert tumors were categorized as “non-infiltrated” by IHC. Chemokine expression significantly correlated with infiltration status by IHC, wherein, CCL5, CCR5, CCL4, CXCR3, CXCL9, CCL2, CCL22, CCL3, and CXCL10 were significantly under expressed in tumors lacking infiltrating effector T-cells (Wilcoxon rank sum test p<0.05, Tukey HSD adjusted p <0.001). Overall ANOVA results showed significant relationship between gene expression of these chemokines and Infiltration status by CD8 IHC (p <0.05). Conclusion: In 300 metastatic cutaneous melanoma cases, we demonstrated that tumor inflammation status by CD8 expression by RNAseq correlated with CD8 infiltration pattern by IHC. Moreover, expression of Infiltrated and non-infiltrated tumors shows distinct chemokine signatures where higher CD8 T-cell infiltration correlates with higher expression of studied chemokines in the tumor microenvironment. It requires further investigation to better understand the interplay of chemokines and cytokines in the tumor microenvironments that drive the immune cycle in melanoma. Citation Format: Sarabjot Pabla, Jeffrey Conroy, Mary Nesline, Sean Glenn, Blake Burgher, Maochun Qin, Jonathan Andreas, Vincent Giamo, Felicia L. Lenzo, Angela Omilian, Wiam Bshara, Antonios Papanicolau-Sengos, Yirong Wang, Marc Ernstoff, Mark Gardner, Carl Morrison. Chemokine expression signatures in infiltrated vs non-infiltrated tumors [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 5070.

Immune deserts: Correlation of low CD8 gene expression with non-response to checkpoint inhibition.

23 Background: Associations between presence and density of effector T-cells (CD8+) and response to checkpoint inhibitors (CPIs) has been established for protein biomarkers using immunohistochemistry but less work has been done to characterize this relationship through gene expression. Methods: We collected clinical response data (RECIST v1.1) for 184 patients who had been treated with CPIs across tumor types with FDA indications. We tested formalin-fixed paraffin embedded (FFPE) samples by NGS with a comprehensive immune response panel which interrogates the gene expression profile of 54 validated immune-related genes. Ranked gene expression was compared to a reference population. 85/184 (46%) cases were considered “responders” after six months, and 66/184 cases (36%) were considered responders after twelve months on treatment. We hypothesized that patients with in the bottom quartile of CD8 expression, as measured by the rank of the sum of normalized CD8A and CD8B transcripts compared to a reference population, would be unlikely to respond to checkpoint therapy (immune deserts), and that “hot” tumor environments (TME) with the highest quartile of expression would be most likely to respond. Results: After six months of therapy, 13/40 (32.5%) patients with immune desert tumors responded to therapy, whereas 39/82 patients (47.6%) in the middle 50% of expression responded, and 33/62 (53.2%) of patients with “hot” TME responded to treatment. The two-tailed p-value of the null hypothesis, that TME status does not correlate with response, was 0.018 and could be rejected. After twelve months of observation, 9/40 (22.5%) immune desert patients were considered “responders”. 28/82 (34.2%) of the middle 50% of expression were classified as responders, and 29/62 (46.8%) of top quartile samples came from responders. The two-tailed p-value of the null hypothesis was 0.005 and could be rejected. Conclusions: Gene expression of CD8 as measured by RNA-seq from FFPE samples can be used to compare these samples versus a reference population to predict an increase or decrease in likelihood of response to CPIs. Patients in the lowest quartile of CD8 expression were the least likely to respond to these promising therapies.

With the Tsunami of Immunotherapy, Will Soft Tissue Sarcoma Remain an Immune Desert?

Soft tissue sarcomas (STSs) are rare, aggressive and heterogeneous malignancies with poor outcomes. They include more than 50 different histologic subtypes with variable molecular and cytogenetic profiles responsible for the heterogeneity of these tumors. Radical surgery, chemotherapy and radiation therapy are still the mainstay of treatment with limited therapeutic options in case of metastatic and locally advanced unresectable disease. The current available cytotoxic and targeted therapies are only offering short living disease control. In the era of immunotherapy, STSs were affected by this ‘tsunami’ with very promising results, but larger randomized trials are still needed to prove their benefit in terms of survival and efficacy. They are not included yet in the standard of care. In this review of the literature, we are developing the active immunotherapeutic strategies (vaccines), as well as the passive strategies (adoptive cell therapy) and the checkpoint inhibitors emphasizing on the most recent results and on the limitation of the immunotherapy in this heterogenic media.

Checkpoint Blockade Plus Oncolytic Virus: A Hot Therapeutic Cancer Strategy

How can we transform an immune desert into a 'hot tumor' that is prone to respond to anti-programmed death (PD)-1 immunotherapy? This might be possible by injecting an oncolytic virus, engineered to induce local immune stimulation, prior to anti-PD-1 therapy. A recent study demonstrated that this combination - evaluated in a Phase Ib metastatic melanoma clinical study - yields promising results.

981P - Reactive stroma mediates CD8+ T cell spatial distribution and function in ovarian cancer

Background: Close proximity between cytotoxic T cells and tumor cells is key to effective immunotherapy. Ovarian cancer exhibits diverse immune phenotypes with distinct prevalence and spatial localization of CD8+ T cells. This study is aimed to characterize the molecular mechanisms orchestrating the localization and function of CD8+ T Cells in ovarian cancer. Methods: CD8 IHC and RNAseq were performed on 277 ovarian tumor tissues from ICON7 phase 3 trial. CD8 T-cells in tumor vs. stromal area was assessed by digital pathology. A Random Forest regression model was constructed to identify molecular features associated with enumeration or spatial localization of CD8+ T cells. In situ validation was performed by MHCI IHC and FAP RNAish. Functional role of ovarian fibroblasts was characterized by ex vivo T cell function assays. Results: We identified three main immune phenotypes, including T-cell infiltrated, T-cell excluded, and immune desert. The immune phenotypes are highly associated with prognosis and the molecular subtypes of ovarian cancer. The T-cell infiltrated phenotype is denoted by high expression of T-effector signatures and antigen presentation machinery. The T-cell excluded phenotype showed similar expression of T-effector signatures as the T-cell infiltrated phenotype, however, most of the CD8+ T-cells were excluded from the tumor bed. The T-cell excluded phenotype showed high expression of the reactive stroma signatures (i.e. FAP), and low expression of class I antigen presentation genes. Lastly, the immune desert phenotype featured low prevalence of CD8+ T-cells, and high expression of neuroendocrine and metabolic pathways. In situ analysis confirmed the two key molecular features associated with the T-cell excluded phenotype: 1) loss of the MHC I expression in the tumor compartment, and 2) high FAP expression in CAFs. Co-culturing of ovarian fibroblast cells with T-cells resulted in reduced T-cell activation and proliferation. Conclusions: Our study uncovered key molecular mechanisms mediating the interplay between CD8+ T cell localization and function in ovarian cancer. Our findings underscore the potential of targeting reactive stroma as a novel therapeutic strategy to optimize immunotherapy for ovarian cancer patients. Legal entity responsible for the study: Genentech Funding: Genentech Disclosure: All authors have declared no conflicts of interest.