Bases legales

Identification and profiling of neoantigen-specific T cells in NSCL cancer patients treated with atezolizumab

Michael Fehlings1, Suchit Jhunjhunwala, BS, PhD2, Marcin Kowanetz, PhD2, Bill O'Gorman2, Priti Hegde, PhD2, Jessica Li, Master of Science2, Hermi Sumatoh1_{, Boon Heng Lee}1_{, Alessandra Nardin, DVM}1_{, Mahesh}

Yadav, PhD2_{, Leesun Kim}2_{, Susan Flynn}2_{, Marcus Ballinger, PhD}2_,

Evan Newell, PhD3

1

immunoSCAPE, Singapore, Singapore;2Genentech, South San Francisco, CA, USA;3_{SiGN, Singapore, Singapore}

Correspondence:Mahesh Yadav ([email protected])

Journal for ImmunoTherapy of Cancer2018,6(Suppl 1):O1

Fig. 1 (abstract P705).CD4 and CD8 T cells redistribute in the spleen. Animals were implanted for 14 days with B16 melanoma cells, after receiving 3 doses of anti-PD-L1 the CD8/CD4 T cell ratio increases

Fig. 2 (abstract P705).Splenic CD4 T cells show an activated phenotype. After tumor implant and antibody treatment, CD4 T cells show increased expression of CD25, CD69, CD278, and CD279. Similar results were observed in CD8 T cells (not shown)

Fig. 3 (abstract P705).CD8+_{cells infiltrate the tumor and co-}

localize with CD11c*cells. After tumor implant and antibody treatment, the tumors were excised and analyzed by fluorescent microscopy. Pictures were taken with a 40X objective

Fig. 4 (abstract P705).Cytokine and chemokine profile in serum. After tumor implant and antibody treatment, the analysis of the samples with LEGENDplexTM_{show A) Increased Th Cytokines. B) Decreased}

pro-inflammatory cytokines. C) Increased chemokine production

Fig. 5 (abstract P705).A) Tumor-specific infiltrating CD8+_{T cells}

increase 24 days after treatment. Animals were implanted with SIYRYYGL-expressing B16 cells and 24 days after treatment the tumor was collected and analyzed. B) Anti-PD-L1 treatment reduces tumor growth

Background

There is strong evidence that immunotherapy-mediated tumor rejec- tion is associated with the reinvigoration of tumor-specific CD8+ T cells most likely recognizing neoantigens derived from tumor somatic muta- tions. However, despite a substantial number of mutations present in some tumors, only a small fraction of neoantigens have been shown to be immunogenic, partly due to the challenge in identifying rare neoantigen-specific CD8+ T cells in tumor-bearing individuals.

Methods

We employed mass cytometry and highly-multiplexed combinatorial tetramer staining together with cellular barcoding and high- dimensional phenotypic characterization to longitudinally monitor neoantigen-specific CD8+ T cells in PBMC from 14 NSCL cancer patients treated with atezolizumab. Close to 800 candidate tumor neoantigens and 73 viral-derived control peptides were screened across all patient samples; T cells were simultaneously profiled using 30 or more markers.

Results

Virus-specific T cells were detected in most patient samples at frequen- cies as low as 0.004% of total CD8+ T cells. T cells reactive for 13 differ- ent neoantigens were also identified with a medium to high confidence across all patients and time points. Interestingly, the major- ity of medium-to-high confidence hits (9/13) were detected among the 8 patients who presented an objective response to treatment, with only 4 of the 13 hits detected in the 6 patients with progressive disease. The neoantigen-specific cells differed phenotypically from bulk CD8+ T cells in the peripheral blood and displayed a diverse phenotype.

Conclusions

This study demonstrates the utility of the use of mass cytometry to- gether with a combinatorial tetramer staining for the ex vivo identifi- cation, characterization, and longitudinal follow-up of rare tumor- specific T cells. Importantly, it suggests that the detection of neoantigen-specific T cells may be used as a predictor of response to checkpoint blockade and supports further research into this.

Trial Registration

NCT01903993

References

1. Atezolizumab versus docetaxel for patients with previously treated non-small- cell lung cancer (POPLAR): a multicentre, open-label, phase 2 randomised controlled trial The Lancet. Volume 387, No. 10030, p1837–1846, 30 April 2016

Ethics Approval

This study was performed on de-identified samples from POPLAR trial, a multicentre, randomized, open-label, allcomer, multicenter phase 2 trial. The POPLAR trial was performed in full accordance with the guidelines for GCP and the Declaration of Helsinki. Protocol; approval was obtained from an independent ethics committee for each trial site.

Consent

All patients gave appropriate ethical approval for this analysis. Statements confirming compliance with ethical regulations, the committees that approved the POPLAR study protocol, and confirmation of informed consent from all study participants are included in the previous publications describing the POPLAR trial [1].

O2

Using artificial intelligence to distinguish subjects with prostate cancer (PCa) from benign prostate hyperplasia (BPH) through immunophenotyping of MDSCs and lymphocyte cell populations

John Roop1, Alex Polo1, Anthony Campisi, BS1, Dmitry Gabrilovich, MD, PhD2_{, Amit Kumar, PhD}1_{, George Dominguez, PhD}1

1

Anixa Biosciences, San Jose, CA, USA;2The Wistar Institute, Philadelphia, PA, USA

Correspondence:George Dominguez ([email protected])

Journal for ImmunoTherapy of Cancer2018,6(Suppl 1):O2

Background

Myeloid-derived suppressor cells (MDSCs) are key contributors in sup- porting tumor progression and tumor escape through their ability to suppress anti-tumor responses mediated through T cell and natural killer (NK) cell activity [1,2]. Several studies have quantified MDSCs to detect tumor development, monitor progression, and/or predict thera- peutic responses [3, 4]. The objective of this study was to determine if

flow cytometry data analysis of MDSCs and other leukocytes could be incorporated with a supervised machine learning classifier to identify individuals presenting either with a prostate malignancy (PCa) or be- nign condition (such as benign prostate hyperplasia or BPH).

Methods

We used standard multiparametric flow cytometry techniques to immu- nophenotype MDSCs and other leukocytes found in the peripheral blood of 73 PCa, 48 BPH, and 73 control subjects; all prostate patholo- gies were confirmed with a transrectal ultrasound guided prostate (TRUSP) biopsy. Subjects were excluded if they had a previous history of cancer (not including subjects under active surveillance), had a med- ical intervention for prostate cancer, or were receiving a dihydrotestos- terone (DHT) or alpha-1 blocker for active treatment of BPH. Next, a series of neural networks were created with inputs consisting of the nu- merical event counts from the flow cytometry FCS file (fluorescent or scatter intensity values). Three datasets were constructed: the training dataset– to ‘teach’ two output categories through backpropagation and parameter fitting; the validation dataset– to evaluate the fit to minimize overfitting; and the test dataset_–to rank the trained networks against each other and estimate the classification performance. Finally, a naïve testing set (i.e. never seen by the network) was used to deter- mine the overall performance of the top-ranking networks after voting.

Results

With this approach, we were able to distinguish PCa subjects (both high and low grade) from control subjects with 91.7% accuracy (AUROC = 0.929; 95% CI: 0.8418 to 1.016). Using the same approach, we can further identify PCa from BPH subjects with 87.5% accuracy (AUROC = 0.869; 95% CI: 0.7938 to 0.9442).

Conclusions

By pairing supervised machine learning with the immunophenotyp- ing of MDSCs and other leukocytes using flow cytometry, we have developed a novel method for distinguishing PCa from control or BPH subjects with high levels of accuracy. We believe this method- ology could be used to predict patient responses to immunother- apies and/or to monitor tumor recurrence.

References

1. Kumar V, Patel S, Tcyganov E, Gabrilovich D. The nature of myeloid- derived suppressor cells in the tumor microenvironment. Trends Immu- nol. 2016; 37:208-220.

2. Marvel D, Gabrilovich D. Myeloid-derived suppressor cells in the tumor micro- environment: expect the unexpected. J Clin Invest. 2015; 125:3356-3364. 3. Elliott L, Doherty G, Sheahan K, Ryan E. Human tumor-infiltrating myeloid

cells: phenotypic and functional diversity. Front Immunol. 2017; 8:86. 4. Okla K, Wertel I, Wawruszak A, Bobinski M, Kotarski J. Blood-based ana-

lyses of cancer: circulating myeloid-derived suppressor cells_–is a new era coming? Crit Rev Clin Lab Sci. 2018.

Ethics Approval

The study was approved by The Wistar Institute's Institutional Review Board, protocol number 21802305.

O3

Anti-tumor immune responses in metastatic breast cancer exceptional responder patients

Alusha Mamchak, PhD1, Ngan Nguyen, PhD1, Danhui Zhang, MD PhD1, Felix Chu, MS1_{, Mike Harbell, BS}1_{, Beatriz Millare, BS}1_{, Kevin Williamson}1_,

Xiaomu Chen1, Xiaobin Tang1, Shuwei Jiang1, Dongkyoon Kim, BS PhD1, Nicole Hasser1_{, Sarah Hippely}2_{, Maren Levin}2_{, Amy Manning-Bog, PhD}1_,

Jeff DeFalco1, william Robinson, MD1, Daniel Emerling, PhD1, Norman Greenberg, PhD1_{, Guy Cavet, PhD}1_{, Joyce O}

’Shaughnessy, MD2 1

Atreca Inc, Redwood City, CA, USA;2Baylor University Medical Center, Dallas, TX, USA

Correspondence:Alusha Mamchak ([email protected])

Journal for ImmunoTherapy of Cancer2018,6(Suppl 1):O3

Background

Analyzing anti-cancer immune responses can offer insights into the mechanisms that underlie successful cancer therapies. While the role of T cell immunity in anti-cancer responses has been well characterized, the role of the humoral immune response to cancer remains less clear. Thus,

we set out to identify anti-tumor antibodies from 11 metastatic breast cancer (MBC) patients who had exceptional responses to systemic ther- apy with multi-year benefit, all of whom were disease-free or long-term non-progressors. The patients’breast cancers were diverse with respect to hormone receptor status, HER2 status, and treatment history.

Methods

We used flow cytometry to isolate plasmablasts, which are antibody se- creting cells produced in lymphoid tissues through activation, affinity maturation, and differentiation of antigen-specific naive and memory B cells. Natively paired IgG sequences were generated from individual cells using Immune Repertoire Capture® (IRC_™) technology. Similar im- munoglobulin sequences were grouped into putative antibody lineages based on germline gene usage and CDR3 sequence features.

Results

A total of 9160 native pairs of expression-ready, heavy and light chain im- munoglobulin sequences were generated. The antibody sequences were grouped into putative clonal lineages, with 931 of these lineages being expressed by two or more plasmablasts, providing evidence of selection and expansion. Comparison of the immunoglobulin repertoires across patients re- vealed several cases in which similar families of antibodies were identified in more than one patient, suggesting convergent selection. Antibodies in the putative convergent families were predominantly IgG2 (86%) which was sig- nificantly higher than the frequency of IgG2 in non-convergent lineages.By computationally selecting immunoglobulin sequences and expressing them as recombinant proteins, patient-derived antibodies were identified that bind to human and mouse cancer cell lines, including human breast and lung can- cer, and tissues derived from xenogeneic murine models of human breast and prostate cancer. In addition, several antibodies were shown by immuno- histochemistry to bind specifically to non-autologous human breast cancer tis- sue but not to adjacent breast tissue.

Conclusions

Through the use of IRC™technology and antibody binding assays, the humoral immune response in MCB exceptional responder patients has been quantified. Of particular interest, this study has identified patient derived-antibodies that bind specifically to non-autologous breast cancer tissue and have potential to form the basis of new cancer therapeutics.

O4

Immune monitoring after NKTR-214 plus nivolumab (PIVOT-02) in previously untreated patients with metastatic Stage IV melanoma

Adi Diab, MD1, Scott Tykodi2, Brendan Curti, MD3, Daniel Cho, MD4, Michael Wong, MD PhD FRCPC1_{, Igor Puzanov, MD, MSCI, FACP}5_{, Karl}

Lewis, MD6, Michele Maio, MD, PhD7, Gregory Daniels, MD, PhD8, Alexander Spira, MD, PhD, FACP9_{, Mary Tagliaferri, MD}10_{, Alison Hannah,}

MD10, Wendy Clemens, PhD10, Michael Imperiale10, Chantale

Bernatchez1_{, Cara Haymaker, PhD}1_{, Salah Eddine Bentebibel}11_{, Jonathan}

Zalevsky, PhD10, Ute Hoch, PhD10, Christie Fanton, PhD10, Ahsan Rizwan, MPharm, PhD10_{, Sandra Aung, PhD}10_{, Fiore Cattaruzza}10_{, Ernesto}

Iaccucci10, Dariusz Sawka12, Mehmet Bilen, MD13, Paul Lorigan14, Giovanni Grignani15_{, James Larkin, MD}16_{, Sekwon Jang, MD}17_{, Ewa Kalinka-Warzocha,}

PhD, MD18, Mario Sznol, MD19, Michael Hurwitz, MD, PhD19

1_{The University of Texas MD Anderson Cancer Center, Houston, TX, USA;} 2

University of Washington and Fred Hutchinson Cancer Research Center, Seattle, WA, USA;3_{Providence Cancer Institute and Earle A. Chiles}

Research Institute, Portland, OR, USA;4NYU Medical Oncology Associates, New York, NY, USA;5_{Roswell Park Cancer Institute, Buffalo,}

NY, USA;6University of Colorado Denver, Aurora, CO;7Azienda Ospedaliera Universitaria Senese, Siena, Italy;8_{Moores Cancer Center,}

University of California San Diego, La Jolla, CA, USA;9Virginia Cancer Specialists, PC, Fairfax, VA, USA;10_{Nektar Therapeutics, San Anselmo, CA,}

USA;11The University of Texas MD Anderson Canc, Houston, TX, USA;

12_{Szpital Specjalistyczny w Brzozowie Podkarpacki Osrodek}

Onkologiczny, Brzozów, Poland;13Emory University Hospital (Winship Cancer Institute), Atlanta, GA, USA;14_{The Christie NHS Foundation Trust,}

Manchester, UK;15Institute for Cancer Research and Treatment (IRCC), Candiolo, Italy;16_{The Royal Marsden, London, UK;}17_{Inova Schar Cancer}

Institute, Fairfax, VA, USA;18Instytut Medyczny Santa Familia, Lodz, Poland;19_{Yale School of Medicine, New Haven, CT, USA}

Correspondence:Adi Diab ([email protected])

Journal for ImmunoTherapy of Cancer2018,6(Suppl 1):O4

Background

In patients with melanoma, low levels of tumor-infiltrating lymphocytes and low/absent PD-L1 expression are associated with limited response to anti-PD-1/anti-PD-L1 therapies. NKTR-214 (IL-2Rβγ-biased cytokine) monotherapy stimulates proliferation and activation of lymphocytes in blood and tumor and increases PD-1/PD-L1 expression. The impact of NKTR-214 and nivolumab on the systemic immune system and local tumor microenvironment is presented.

Methods

The melanoma cohort is closed; 41 patients were enrolled with 38 evaluable for efficacy (_≥1 follow-up scan). Tumor biopsies were ana- lyzed using multispectral IHC, gene expression, and TCR sequencing. Flow cytometry and hematology were used to evaluate blood cells. PD-L1 expression was evaluated using DAKO, 28-8 PharmDx Assay.

Results

Immune monitoring of blood revealed clear activation of the IL-2 path- way following administration of NKTR-214 plus nivolumab. Lymphocyte numbers increased 9x (N=41) from nadir reaching their peak 7 days post dose and maintained that magnitude of increase after each cycle. The proportion of proliferating (Ki67+, n=12) CD4+, CD8+, and NK cells in- creased 13x, 20x, and 6x over baseline, respectively. Similar immune acti- vation was reported with NKTR-214 monotherapy (8x, 8x, and 7x over baseline, respectively). Immune cells demonstrated an antigen- experienced phenotype with an increased proportion of HLA-DR expres- sion on CD4+, CD8+, and NK cells 3x, 2x, and 6x over baseline, respect- ively. ICOS levels increased 2x on CD8+ T cells. Baseline and week 3 biopsies (n=12, evaluable) showed local effects on the tumor microenvir- onment including elevated expression of PD-L1 on the tumor (patients converted from PD-L1 negative to positive), increased total numbers of CD8 infiltrate, and increased proportion of proliferating cells all ranging from 6-17x over baseline. Following treatment, intratumoral gene expres- sion analyses showed elevations in networks associated with the NKTR- 214 mechanism of action, including induction of an interferon-gamma gene signature. The investigator-assessed objective response rate as of 12 July 2018 was 50% (N=38), and no responder has relapsed. Deepening of response was observed over time and was associated with immune activation, consistent with the MOA of NKTR-214 plus nivolumab. The median duration of response has not been reached.

Conclusions

NKTR-214 is a robust agonist of the IL-2 pathway and together with nivolumab promotes immune activation in the periphery and tumor microenvironment for significant clinical activity. A global phase 3 trial in treatment-naïve advanced melanoma patients of NKTR-214 plus nivolumab versus nivolumab (1:1) will be open for enrollment in 2018.

Trial Registration

Clinicaltrials.gov NCT02983045 (PIVOT-02)

O5

B-cells and tertiary lymphoid structures (TLS) predict response to immune checkpoint blockade (ICB)

Sangeetha Reddy, MD, MSci1_{, MD, MSci}1_{, Beth Helmink, MD PhD}1_,

Jianjun Gao, MD PhD1, Shaojun Zhang, PhD1, Keren Yizhak2, Moshe Sade-Feldman3_{, Jorge Blando, DVM}1_{, Guangchun Han}1_{, Vancheswaran}

Gopalakrishnan, MPH, PhD1, Hao Zhao1, Wenbin Liu1, Hussein Tawbi, MD, PhD1_{, Rodabe Amaria, MD}1_{, Michael Davies, MD, PhD}1_{, Patrick Hwu,}

MD1, Jeffrey Lee, MD1, Jeffrey Gershenwald, MD1, Scott Woodman1, Elizabeth Burton1_{, Lauren Haydu, MS, BChe, MIPH}1_{, Alexander Lazar, MD,}

PhD1, Courtney Hudgens, MS1, Alexandria Cogdill, MEng1, Oscar Krijgsman, PhD4_{, Elisa Rozeman, MD}4_{, Daniel Peeper, PhD}4_{, Christian Blank, MD}4_{, Ton}

Schumacher, PhD4, Emily Keung1, Pierre-Olivier Gaudreau1, Alexandre Reuben1_{, Christine Spencer, PhD}1_{, Lisa Butterfield, PhD}5_{, James Allison, PhD}1_,

Michael Tetzlaff, MD PhD1, Florent Petitprez, MS6, Wolf Herman Fridman, MD, PhD6_{, Catherine Sautes-Fridman, PhD}6_{, Nir Hacohen, PhD}2_{, Padmanee}

Sharma, MD, PhD1, Linghua Wang, PhD1, Jennifer Wargo, MD, MMSc1

1_{MD Anderson Cancer Center, Houston, TX, USA;}2_{Broad Institute,}

Cambridge, MA, USA;3Massachusetts General Hospital, Boston, MA, USA;

4_{The Netherlands Cancer Institute, Netherlands, Netherlands;}5_University

of Pittsburgh Medical Center, Pittsburgh, PA, USA;6Centre de Recherche des Cordeliers, Paris, France

Correspondence:Jennifer Wargo ([email protected])

Background

Mutational load, cytotoxic T-cell markers, and PD-L1 have been iden- tified as biomarkers of response to ICB. However, there is a growing understanding of the contribution of B-cells in shaping response to ICB. We conducted a neoadjuvant ICB trial in patients with high-risk resectable melanoma ((NCT02519322), and identified B-cell signa- tures in responders by protein expression profiling.

Methods

To further investigate this, we performed transcriptomic profiling of longitudinal specimens in this melanoma cohort. Differentially expressed genes (DEG) were assessed in baseline samples with ad- equate tumor purity. Targeted immune profiling was further performed using immune deconvolution tool MCP-counter in all baseline and on- treatment samples, with additional validation from a metastatic renal cell carcinoma (RCC) trial of ICB (NCT02210117) and the melanoma TCGA dataset. Cases were dichotomized by CD8 T-cell scores to study interaction between B- and T-cells. Singlet and multiplex immunohisto- chemistry assessed spatial organization of the tumor infiltrating B-cells. Single cell RNA sequencing was performed in an independent cohort of metastatic melanoma patients treated with ICB.

Results

The most DEG at baseline in melanoma responders to ICB were B-cell related genes such as MZB1, BTLA, and IGLL5 (NR) (p<0.0001 for all). B- cell signatures were confirmed by a targeted immune gene assessment using MCP-counter, showing higher B lineage signatures among re- sponders at baseline (p=0.036) and on-treatment (p=0.038). Among baseline cases, B-cells were more strongly predictive among CD8 T-cell low subsets (p=0.085) compared to T cell high (p=0.833). The applic- ability of these findings to other tumors was demonstrated in an RCC cohort, in which B lineage scores was predictive of response (p=2.6e- 03), and differential effects were again seen between CD8 T-cell low (p=0.008) and high cases (p=0.564). In the melanoma TCGA, B-cell lineage score was correlated with improved survival (p<0.0001 for over- all and disease-specific survival), in particular in CD8 T-cell low cases in- cluding after multivariable adjustment (p=0.001 for overall survival and 0.006 for disease-specific survival). Single cell sequencing in an inde- pendent melanoma cohort identified DEG within B-cells by response, providing insights into B-cell phenotypes associated with outcomes. As- sessment of tissue sections from tumor samples in the neoadjuvant melanoma ICB cohort demonstrated co-localization of the B cells in TLS with CD8 and CD4 T-cells and CD21 follicular dendritic cells. The ratio of tumor area occupied by TLS was higher in responders (p=0.037 at baseline and 0.002 on-treatment).

Conclusions

Together, these results highlight the potential significance for B-cell signatures as prognostic and predictive factors for response to ICB.

O6

Comparison of biomarker assay modalities in anti-PD-(L)1 monotherapy: a meta-analysis

Steve Lu, BS1_{, Steve Lu, BS}1_{, Ludmila Danilova, PhD}2_{, David Rimm, MD,}

PhD3, Clifford Hoyt, MS4, Matthew Hellmann, MD5, Janis Taube, MD1

1_{Johns Hopkins University School of Medicine, Baltimore, MD, USA;} 2

Johns Hopkins Medical Institutions, Baltimore, MD, USA;3Yale University