Tissue-resident memory and circulating T cells are early responders to pre-surgical cancer immunotherapy.

Neoadjuvant immune checkpoint blockade has shown promising clinical activity. Here, we characterized early kinetics in tumor-infiltrating and circulating immune cells in oral cancer patients treated with neoadjuvant anti-PD-1 or anti-PD-1/CTLA-4 in a clinical trial (NCT02919683). Tumor-infiltrating CD8 T cells that clonally expanded during immunotherapy expressed elevated tissue-resident memory and cytotoxicity programs, which were already active prior to therapy, supporting the capacity for rapid response. Systematic target discovery revealed that treatment-expanded tumor T cell clones in responding patients recognized several self-antigens, including the cancer-specific antigen MAGEA1. Treatment also induced a systemic immune response characterized by expansion of activated T cells enriched for tumor-infiltrating T cell clonotypes, including both pre-existing and emergent clonotypes undetectable prior to therapy. The frequency of activated blood CD8 T cells, notably pre-treatment PD-1-positive KLRG1-negative T cells, was strongly associated with intra-tumoral pathological response. These results demonstrate how neoadjuvant checkpoint blockade induces local and systemic tumor immunity.

AI and immunology.

AI is rapidly becoming part of many aspects of daily life, with an impact that reaches all fields of research. We asked investigators to share their thoughts on how AI is changing immunology research, what is necessary to move forward, the potential and the pitfalls, and what will remain unchanged as the field journeys into a new era.

LSD1 Ablation Stimulates Anti-tumor Immunity and Enables Checkpoint Blockade.

Chromatin regulators play a broad role in regulating gene expression and, when gone awry, can lead to cancer. Here, we demonstrate that ablation of the histone demethylase LSD1 in cancer cells increases repetitive element expression, including endogenous retroviral elements (ERVs), and decreases expression of RNA-induced silencing complex (RISC) components. Significantly, this leads to double-stranded RNA (dsRNA) stress and activation of type 1 interferon, which stimulates anti-tumor T cell immunity and restrains tumor growth. Furthermore, LSD1 depletion enhances tumor immunogenicity and T cell infiltration in poorly immunogenic tumors and elicits significant responses of checkpoint blockade-refractory mouse melanoma to anti-PD-1 therapy. Consistently, TCGA data analysis shows an inverse correlation between LSD1 expression and CD8+ T cell infiltration in various human cancers. Our study identifies LSD1 as a potent inhibitor of anti-tumor immunity and responsiveness to immunotherapy and suggests LSD1 inhibition combined with PD-(L)1 blockade as a novel cancer treatment strategy.

Cancer-Germline Antigen Expression Discriminates Clinical Outcome to CTLA-4 Blockade.

CTLA-4 immune checkpoint blockade is clinically effective in a subset of patients with metastatic melanoma. We identify a subcluster of MAGE-A cancer-germline antigens, located within a narrow 75 kb region of chromosome Xq28, that predicts resistance uniquely to blockade of CTLA-4, but not PD-1. We validate this gene expression signature in an independent anti-CTLA-4-treated cohort and show its specificity to the CTLA-4 pathway with two independent anti-PD-1-treated cohorts. Autophagy, a process critical for optimal anti-cancer immunity, has previously been shown to be suppressed by the MAGE-TRIM28 ubiquitin ligase in vitro. We now show that the expression of the key autophagosome component LC3B and other activators of autophagy are negatively associated with MAGE-A protein levels in human melanomas, including samples from patients with resistance to CTLA-4 blockade. Our findings implicate autophagy suppression in resistance to CTLA-4 blockade in melanoma, suggesting exploitation of autophagy induction for potential therapeutic synergy with CTLA-4 inhibitors.

Structural Alterations Driving Castration-Resistant Prostate Cancer Revealed by Linked-Read Genome Sequencing.

Nearly all prostate cancer deaths are from metastatic castration-resistant prostate cancer (mCRPC), but there have been few whole-genome sequencing (WGS) studies of this disease state. We performed linked-read WGS on 23 mCRPC biopsy specimens and analyzed cell-free DNA sequencing data from 86 patients with mCRPC. In addition to frequent rearrangements affecting known prostate cancer genes, we observed complex rearrangements of the AR locus in most cases. Unexpectedly, these rearrangements include highly recurrent tandem duplications involving an upstream enhancer of AR in 70%-87% of cases compared with <2% of primary prostate cancers. A subset of cases displayed AR or MYC enhancer duplication in the context of a genome-wide tandem duplicator phenotype associated with CDK12 inactivation. Our findings highlight the complex genomic structure of mCRPC, nominate alterations that may inform prostate cancer treatment, and suggest that additional recurrent events in the non-coding mCRPC genome remain to be discovered.

A Cancer Cell Program Promotes T Cell Exclusion and Resistance to Checkpoint Blockade.

Immune checkpoint inhibitors (ICIs) produce durable responses in some melanoma patients, but many patients derive no clinical benefit, and the molecular underpinnings of such resistance remain elusive. Here, we leveraged single-cell RNA sequencing (scRNA-seq) from 33 melanoma tumors and computational analyses to interrogate malignant cell states that promote immune evasion. We identified a resistance program expressed by malignant cells that is associated with T cell exclusion and immune evasion. The program is expressed prior to immunotherapy, characterizes cold niches in situ, and predicts clinical responses to anti-PD-1 therapy in an independent cohort of 112 melanoma patients. CDK4/6-inhibition represses this program in individual malignant cells, induces senescence, and reduces melanoma tumor outgrowth in mouse models in vivo when given in combination with immunotherapy. Our study provides a high-resolution landscape of ICI-resistant cell states, identifies clinically predictive signatures, and suggests new therapeutic strategies to overcome immunotherapy resistance.

Oncogenic Signaling Pathways in The Cancer Genome Atlas.

Genetic alterations in signaling pathways that control cell-cycle progression, apoptosis, and cell growth are common hallmarks of cancer, but the extent, mechanisms, and co-occurrence of alterations in these pathways differ between individual tumors and tumor types. Using mutations, copy-number changes, mRNA expression, gene fusions and DNA methylation in 9,125 tumors profiled by The Cancer Genome Atlas (TCGA), we analyzed the mechanisms and patterns of somatic alterations in ten canonical pathways: cell cycle, Hippo, Myc, Notch, Nrf2, PI-3-Kinase/Akt, RTK-RAS, TGFß signaling, p53 and ß-catenin/Wnt. We charted the detailed landscape of pathway alterations in 33 cancer types, stratified into 64 subtypes, and identified patterns of co-occurrence and mutual exclusivity. Eighty-nine percent of tumors had at least one driver alteration in these pathways, and 57% percent of tumors had at least one alteration potentially targetable by currently available drugs. Thirty percent of tumors had multiple targetable alterations, indicating opportunities for combination therapy.

Comprehensive Characterization of Cancer Driver Genes and Mutations.

Identifying molecular cancer drivers is critical for precision oncology. Multiple advanced algorithms to identify drivers now exist, but systematic attempts to combine and optimize them on large datasets are few. We report a PanCancer and PanSoftware analysis spanning 9,423 tumor exomes (comprising all 33 of The Cancer Genome Atlas projects) and using 26 computational tools to catalog driver genes and mutations. We identify 299 driver genes with implications regarding their anatomical sites and cancer/cell types. Sequence- and structure-based analyses identified >3,400 putative missense driver mutations supported by multiple lines of evidence. Experimental validation confirmed 60%-85% of predicted mutations as likely drivers. We found that >300 MSI tumors are associated with high PD-1/PD-L1, and 57% of tumors analyzed harbor putative clinically actionable events. Our study represents the most comprehensive discovery of cancer genes and mutations to date and will serve as a blueprint for future biological and clinical endeavors.

Tumor Evolution: A Problem of Histocompatibility.

In this issue of Cell, two articles show that tumor-specific changes in HLA-mediated antigen presentation affect tumor immunogenicity and may play a role in shaping cancer cell survival.

Integrative clinical genomics of advanced prostate cancer.

Toward development of a precision medicine framework for metastatic, castration-resistant prostate cancer (mCRPC), we established a multi-institutional clinical sequencing infrastructure to conduct prospective whole-exome and transcriptome sequencing of bone or soft tissue tumor biopsies from a cohort of 150 mCRPC affected individuals. Aberrations of AR, ETS genes, TP53, and PTEN were frequent (40%-60% of cases), with TP53 and AR alterations enriched in mCRPC compared to primary prostate cancer. We identified new genomic alterations in PIK3CA/B, R-spondin, BRAF/RAF1, APC, ß-catenin, and ZBTB16/PLZF. Moreover, aberrations of BRCA2, BRCA1, and ATM were observed at substantially higher frequencies (19.3% overall) compared to those in primary prostate cancers. 89% of affected individuals harbored a clinically actionable aberration, including 62.7% with aberrations in AR, 65% in other cancer-related genes, and 8% with actionable pathogenic germline alterations. This cohort study provides clinically actionable information that could impact treatment decisions for these affected individuals.

Germline polygenic risk scores are associated with immune gene expression signature and immune cell infiltration in breast cancer.

The tumor immune microenvironment (TIME) plays key roles in tumor progression and response to immunotherapy. Previous studies have identified individual germline variants associated with differences in TIME. Here, we hypothesize that common variants associated with breast cancer risk or cancer-related traits, represented by polygenic risk scores (PRSs), may jointly influence immune features in TIME. We derived 154 immune traits from bulk gene expression profiles of 764 breast tumors and 598 adjacent normal tissue samples from 825 individuals with breast cancer in the Nurses' Health Study (NHS) and NHSII. Immunohistochemical staining of four immune cell markers were available for a subset of 205 individuals. Germline PRSs were calculated for 16 different traits including breast cancer, autoimmune diseases, type 2 diabetes, ages at menarche and menopause, body mass index (BMI), BMI-adjusted waist-to-hip ratio, alcohol intake, and tobacco smoking. Overall, we identified 44 associations between germline PRSs and immune traits at false discovery rate q < 0.25, including 3 associations with q < 0.05. We observed consistent inverse associations of inflammatory bowel disease (IBD) and Crohn disease (CD) PRSs with interferon signaling and STAT1 scores in breast tumor and adjacent normal tissue; these associations were replicated in a Norwegian cohort. Inverse associations were also consistently observed for IBD PRS and B cell abundance in normal tissue. We also observed positive associations between CD PRS and endothelial cell abundance in tumor. Our findings suggest that the genetic mechanisms that influence immune-related diseases are also associated with TIME in breast cancer.

Biologically informed deep neural network for prostate cancer discovery.

The determination of molecular features that mediate clinically aggressive phenotypes in prostate cancer remains a major biological and clinical challenge1,2. Recent advances in interpretability of machine learning models as applied to biomedical problems may enable discovery and prediction in clinical cancer genomics3-5. Here we developed P-NET-a biologically informed deep learning model-to stratify patients with prostate cancer by treatment-resistance state and evaluate molecular drivers of treatment resistance for therapeutic targeting through complete model interpretability. We demonstrate that P-NET can predict cancer state using molecular data with a performance that is superior to other modelling approaches. Moreover, the biological interpretability within P-NET revealed established and novel molecularly altered candidates, such as MDM4 and FGFR1, which were implicated in predicting advanced disease and validated in vitro. Broadly, biologically informed fully interpretable neural networks enable preclinical discovery and clinical prediction in prostate cancer and may have general applicability across cancer types.

Toward a CRISPR-based mouse model of Vhl-deficient clear cell kidney cancer: Initial experience and lessons learned.

CRISPR is revolutionizing the ability to do somatic gene editing in mice for the purpose of creating new cancer models. Inactivation of the VHL tumor suppressor gene is the signature initiating event in the most common form of kidney cancer, clear cell renal cell carcinoma (ccRCC). Such tumors are usually driven by the excessive HIF2 activity that arises when the VHL gene product, pVHL, is defective. Given the pressing need for a robust immunocompetent mouse model of human ccRCC, we directly injected adenovirus-associated viruses (AAVs) encoding sgRNAs against VHL and other known/suspected ccRCC tumor suppressor genes into the kidneys of C57BL/6 mice under conditions where Cas9 was under the control of one of two different kidney-specific promoters (Cdh16 or Pax8) to induce kidney tumors. An AAV targeting Vhl, Pbrm1, Keap1, and Tsc1 reproducibly caused macroscopic ccRCCs that partially resembled human ccRCC tumors with respect to transcriptome and cell of origin and responded to a ccRCC standard-of-care agent, axitinib. Unfortunately, these tumors, like those produced by earlier genetically engineered mouse ccRCCs, are HIF2 independent.

Loss of PTEN Is Associated with Resistance to Anti-PD-1 Checkpoint Blockade Therapy in Metastatic Uterine Leiomyosarcoma.

Response to immune checkpoint blockade in mesenchymal tumors is poorly characterized, but immunogenomic dissection of these cancers could inform immunotherapy mediators. We identified a treatment-naive patient who has metastatic uterine leiomyosarcoma and has experienced complete tumor remission for >2 years on anti-PD-1 (pembrolizumab) monotherapy. We analyzed the primary tumor, the sole treatment-resistant metastasis, and germline tissue to explore mechanisms of immunotherapy sensitivity and resistance. Both tumors stained diffusely for PD-L2 and showed sparse PD-L1 staining. PD-1+ cell infiltration significantly decreased in the resistant tumor (p = 0.039). Genomically, the treatment-resistant tumor uniquely harbored biallelic PTEN loss and had reduced expression of two neoantigens that demonstrated strong immunoreactivity with patient T cells in vitro, suggesting long-lasting immunological memory. In this near-complete response to PD-1 blockade in a mesenchymal tumor, we identified PTEN mutations and reduced expression of genes encoding neoantigens as potential mediators of resistance to immune checkpoint therapy.

Germline predisposition to pediatric Ewing sarcoma is characterized by inherited pathogenic variants in DNA damage repair genes.

More knowledge is needed regarding germline predisposition to Ewing sarcoma to inform biological investigation and clinical practice. Here, we evaluated the enrichment of pathogenic germline variants in Ewing sarcoma relative to other pediatric sarcoma subtypes, as well as patterns of inheritance of these variants. We carried out European-focused and pan-ancestry case-control analyses to screen for enrichment of pathogenic germline variants in 141 established cancer predisposition genes in 1,147 individuals with pediatric sarcoma diagnoses (226 Ewing sarcoma, 438 osteosarcoma, 180 rhabdomyosarcoma, and 303 other sarcoma) relative to identically processed cancer-free control individuals. Findings in Ewing sarcoma were validated with an additional cohort of 430 individuals, and a subset of 301 Ewing sarcoma parent-proband trios was analyzed for inheritance patterns of identified pathogenic variants. A distinct pattern of pathogenic germline variants was seen in Ewing sarcoma relative to other sarcoma subtypes. FANCC was the only gene with an enrichment signal for heterozygous pathogenic variants in the European Ewing sarcoma discovery cohort (three individuals, OR 12.6, 95% CI 3.0-43.2, p = 0.003, FDR = 0.40). This enrichment in FANCC heterozygous pathogenic variants was again observed in the European Ewing sarcoma validation cohort (three individuals, OR 7.0, 95% CI 1.7-23.6, p = 0.014), representing a broader importance of genes involved in DNA damage repair, which were also nominally enriched in individuals with Ewing sarcoma. Pathogenic variants in DNA damage repair genes were acquired through autosomal inheritance. Our study provides new insight into germline risk factors contributing to Ewing sarcoma pathogenesis.

Empirical evaluation of language modeling to ascertain cancer outcomes from clinical text reports.

BACKGROUND: Longitudinal data on key cancer outcomes for clinical research, such as response to treatment and disease progression, are not captured in standard cancer registry reporting. Manual extraction of such outcomes from unstructured electronic health records is a slow, resource-intensive process. Natural language processing (NLP) methods can accelerate outcome annotation, but they require substantial labeled data. Transfer learning based on language modeling, particularly using the Transformer architecture, has achieved improvements in NLP performance. However, there has been no systematic evaluation of NLP model training strategies on the extraction of cancer outcomes from unstructured text. RESULTS: We evaluated the performance of nine NLP models at the two tasks of identifying cancer response and cancer progression within imaging reports at a single academic center among patients with non-small cell lung cancer. We trained the classification models under different conditions, including training sample size, classification architecture, and language model pre-training. The training involved a labeled dataset of 14,218 imaging reports for 1112 patients with lung cancer. A subset of models was based on a pre-trained language model, DFCI-ImagingBERT, created by further pre-training a BERT-based model using an unlabeled dataset of 662,579 reports from 27,483 patients with cancer from our center. A classifier based on our DFCI-ImagingBERT, trained on more than 200 patients, achieved the best results in most experiments; however, these results were marginally better than simpler "bag of words" or convolutional neural network models. CONCLUSION: When developing AI models to extract outcomes from imaging reports for clinical cancer research, if computational resources are plentiful but labeled training data are limited, large language models can be used for zero- or few-shot learning to achieve reasonable performance. When computational resources are more limited but labeled training data are readily available, even simple machine learning architectures can achieve good performance for such tasks.

Impact of an Expanded Definition of Family History on Outcomes of Active Surveillance for Prostate Cancer.

PURPOSE: Despite family history being an established risk factor for prostate cancer, the role of a broader definition of family history inclusive of not just prostate cancer but other genetically related malignancies has not been investigated in the active surveillance population. Here, we evaluate the impact of an expanded definition of family history on active surveillance outcomes. MATERIALS AND METHODS: Patients undergoing active surveillance for prostate cancer at Massachusetts General Hospital from 1997-2019 with detailed data available on family cancer history were identified. Primary outcome was biopsy progression-free survival, and secondary outcomes were treatment-free survival, adverse pathological features at prostatectomy, and biochemical recurrence after treatment. Statistical analyses were conducted using the Kaplan-Meier method and Cox regression. RESULTS: Among 855 evaluable patients, 300 (35.1%) patients had any family history of prostate cancer, and 95 (11.1%) had a family history of related malignancies suggestive of a hereditary cancer syndrome. Family history of prostate cancer alone was not associated with biopsy progression, whereas family history suggestive of a hereditary cancer syndrome was associated with a significantly increased risk of biopsy progression (HR 1.43, 95%CI 1.01-2.02), independent of other known clinicopathological risk factors in multivariable analysis. Similarly, family history suggestive of a hereditary cancer syndrome was associated with significantly lower treatment-free survival (HR 1.58, 95%CI 1.14-2.18) in multivariable analysis. No significant association was found between family history and adverse features on surgical pathology or biochemical recurrence. CONCLUSIONS: An expanded family history suggestive of a hereditary cancer syndrome is an independent predictor of biopsy progression during active surveillance. Men with such a family history may still be offered active surveillance but should be counseled regarding the higher risk of disease progression.

Implementation of a prostate cancer-specific targeted sequencing panel for credentialing of patient-derived cell lines and genomic characterization of patient samples.

BACKGROUND: Primary and metastatic prostate cancers have low mutation rates and recurrent alterations in a small set of genes, enabling targeted sequencing of prostate cancer-associated genes as an efficient approach to characterizing patient samples (compared to whole-exome and whole-genome sequencing). For example, targeted sequencing provides a flexible, rapid, and cost-effective method for genomic assessment of patient-derived cell lines to evaluate fidelity to initial patient tumor samples. METHODS: We developed a prostate cancer-specific targeted next-generation sequencing (NGS) panel to detect alterations in 62 prostate cancer-associated genes as well as recurring gene fusions with ETS family members, representing the majority of common alterations in prostate cancer. We tested this panel on primary prostate cancer tissues and blood biopsies from patients with metastatic prostate cancer. We generated patient-derived cell lines from primary prostate cancers using conditional reprogramming methods and applied targeted sequencing to evaluate the fidelity of these cell lines to the original patient tumors. RESULTS: The prostate cancer-specific panel identified biologically and clinically relevant alterations, including point mutations in driver oncogenes and ETS family fusion genes, in tumor tissues from 29 radical prostatectomy samples. The targeted panel also identified genomic alterations in cell-free DNA and circulating tumor cells (CTCs) from patients with metastatic prostate cancer, and in standard prostate cancer cell lines. We used the targeted panel to sequence our set of patient-derived cell lines; however, no prostate cancer-specific mutations were identified in the tumor-derived cell lines, suggesting preferential outgrowth of normal prostate epithelial cells. CONCLUSIONS: We evaluated a prostate cancer-specific targeted NGS panel to detect common and clinically relevant alterations (including ETS family gene fusions) in prostate cancer. The panel detected driver mutations in a diverse set of clinical samples of prostate cancer, including fresh-frozen tumors, cell-free DNA, CTCs, and cell lines. Targeted sequencing of patient-derived cell lines highlights the challenge of deriving cell lines from primary prostate cancers and the importance of genomic characterization to credential candidate cell lines. Our study supports that a prostate cancer-specific targeted sequencing panel provides an efficient, clinically feasible approach to identify genetic alterations across a spectrum of prostate cancer samples and cell lines.