Genetic changes in repetitive sequences are a hallmark of cancer and other diseases, but characterizing these has been challenging using standard sequencing approaches. We developed a de novo kmer ...finding approach, called ARTEMIS (Analysis of RepeaT EleMents in dISease), to identify repeat elements from whole-genome sequencing. Using this method, we analyzed 1.2 billion kmers in 2837 tissue and plasma samples from 1975 patients, including those with lung, breast, colorectal, ovarian, liver, gastric, head and neck, bladder, cervical, thyroid, or prostate cancer. We identified tumor-specific changes in these patients in 1280 repeat element types from the LINE, SINE, LTR, transposable element, and human satellite families. These included changes to known repeats and 820 elements that were not previously known to be altered in human cancer. Repeat elements were enriched in regions of driver genes, and their representation was altered by structural changes and epigenetic states. Machine learning analyses of genome-wide repeat landscapes and fragmentation profiles in cfDNA detected patients with early-stage lung or liver cancer in cross-validated and externally validated cohorts. In addition, these repeat landscapes could be used to noninvasively identify the tissue of origin of tumors. These analyses reveal widespread changes in repeat landscapes of human cancers and provide an approach for their detection and characterization that could benefit early detection and disease monitoring of patients with cancer.
Despite progress in immunotherapy, identifying patients that respond has remained a challenge. Through analysis of whole-exome and targeted sequence data from 5,449 tumors, we found a significant ...correlation between tumor mutation burden (TMB) and tumor purity, suggesting that low tumor purity tumors are likely to have inaccurate TMB estimates. We developed a new method to estimate a corrected TMB (cTMB) that was adjusted for tumor purity and more accurately predicted outcome to immune checkpoint blockade (ICB). To identify improved predictive markers together with cTMB, we performed whole-exome sequencing for 104 lung tumors treated with ICB. Through comprehensive analyses of sequence and structural alterations, we discovered a significant enrichment in activating mutations in receptor tyrosine kinase (RTK) genes in nonresponding tumors in three immunotherapy treated cohorts. An integrated multivariable model incorporating cTMB, RTK mutations, smoking-related mutational signature and human leukocyte antigen status provided an improved predictor of response to immunotherapy that was independently validated.
In this study, we incorporate analyses of genome-wide sequence and structural alterations with pre- and on-therapy transcriptomic and T cell repertoire features in immunotherapy-naive melanoma ...patients treated with immune checkpoint blockade. Although tumor mutation burden is associated with improved treatment response, the mutation frequency in expressed genes is superior in predicting outcome. Increased T cell density in baseline tumors and dynamic changes in regression or expansion of the T cell repertoire during therapy distinguish responders from non-responders. Transcriptome analyses reveal an increased abundance of B cell subsets in tumors from responders and patterns of molecular response related to expressed mutation elimination or retention that reflect clinical outcome. High-dimensional genomic, transcriptomic, and immune repertoire data were integrated into a multi-modal predictor of response. These findings identify genomic and transcriptomic characteristics of tumors and immune cells that predict response to immune checkpoint blockade and highlight the importance of pre-existing T and B cell immunity in therapeutic outcomes.
Display omitted
Unmet need for integrated molecular models that interpret immunotherapy responseGenomic, transcriptomic, and T and B cell sequence data integration by machine learningT cell dynamism is a hallmark of response to immune checkpoint blockadeThe combined contributions of B, T, and tumor cell features predict clinical outcome
Anagnostou et al. integrate genomic, expression, and immune cell repertoire analyses to gain insights into the crosstalk between cancer and immune cells during immunotherapy for melanoma. These findings suggest that the complex phenotype of tumor immune infiltrates combined with genomic features of tumor cells are relevant for determining clinical outcomes.
Abstract Noninvasive approaches for detection of tumor-specific mutations in cell-free DNA (cfDNA) have the potential to track a patient’s response to treatment, enabling effective and timely ...decisions on therapy. However, mutations in cfDNA arising from clonal hematopoeisis (CH) are common and tumor biopsies for definitive identification of the origin of these mutations are not always available. Sequencing of matched cells from buffy coat and the absence of mutations in these cells has been used to rule-out white blood cell (WBC) mutations, but uneven sequencing depths between matched cfDNA and buffy coat and the fraction of mutant alleles are generally ignored by rule-based tests. A probabilistic approach that quantifies the evidence of tumor-derived mutations in cfDNA is needed. We developed a Bayesian framework to estimate the probability that a mutation identified in cfDNA is tumor specific. Our approach requires the number of reads with a mutant allele in plasma (yp) and WBCs (yw) and the corresponding number of total distinct reads at these locations. The posterior odds that a mutation is tumor-derived (S) versus CH or germline (H) is given by the ratio of the probabilities of observing the distinct reads given each model times the prior odds. Estimation of the Bayes factor is obtained by integrating over the unobserved mutant allele fractions in plasma and WBCs using Monte Carlo importance sampling. We applied this approach to 52 patients with initially unresectable colorectal cancer (CRC) liver metastases in the CAIRO5 clinical trial (NCT02162563), using ultra-deep targeted sequencing of cfDNA from plasma and matched WBCs. Among the CAIRO5 patients analyzed, we identified 95 mutations with moderate evidence of tumor-derived cfDNA mutations (Bayes factor > 10) and 19 mutations that were CH-derived (Bayes factor < 0.1). For a subset of 47 cfDNA mutations with no corresponding mutation identified by WBC sequencing, the evidence of tumor origin was highly variable (Bayes factor range: 0.03 to 5.6). While the standard rule-based approach identifies all of these mutations as tumor-derived, none of these mutations reach a moderate evidence cutpoint (all Bayes factors < 6). As a false positive would lead to identification of cfDNA mutations that do not track tumor burden, requiring even higher levels of evidence (Bayes factor > 99) for the selection of cfDNA mutations could be warranted, and still identifies one or more cfDNA mutations in 43 of the patients. This approach is implemented in the R package PLASMUT available from Bioconductor (doi:10.18129/B9.bioc.plasmut). We developed an approach that quantifies the evidence between two competing models for the origin of mutations in cfDNA. A cutpoint for determination of the probability of tumor-derived cfDNA mutations can be tailored to the disease application, balancing the potential benefits of noninvasive testing with the harms of false positives and negatives. Citation Format: Adith S. Arun, Jamie E. Medina, Stephen Cristiano, Daniel C. Bruhm, Remond J. Fijneman, Gerrit A. Meijer, Alessandro Leal, Victor E. Velculescu, Robert B. Scharpf. PLASMUT: An R Package for estimating the probability of tumor-specific mutations in cell-free DNA abstract. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 6101.
Abstract only
10009
Background: The complex crosstalk between tumor and immune cells during immune checkpoint blockade mandates the development of integrated models to interpret the antitumor immune ...response and predict clinical outcome. Methods: We performed comprehensive genomic, transcriptomic and T cell repertoire analyses on tumor biopsies from 64 patients with advanced melanoma receiving nivolumab +/- ipilimumab on CheckMate-038 (NCT01621490). Tumor biopsies were obtained at baseline and 2-4 weeks on therapy. Machine learning and Cox proportional hazards regression analyses were employed to integrate multi-omics features in predictive models of response, defined by RECISTv1.1 as complete and partial response, and survival (PFS and OS). Results: Responding patients had a higher tumor mutation burden (TMB) than non-responders. Expressed TMB more accurately predicted overall survival than genomic TMB (log rank p = 0.028 vs 0.078). High tumor aneuploidy was associated with worse prognosis especially for the patients in the nivolumab + ipilimumab group (log rank p = 0.01). TCR sequencing of paired tumors before and on-treatment revealed that responders had a significantly higher number of unique TCR clones at baseline and more clonotypic shifts on-treatment (p = 0.0018). Gene rearrangement analyses using transcriptome data identified a higher number of rearrangements involving immunoglobulin (Ig) genes in baseline tumors from responders. Deconvolution of transcriptomic data confirmed an enrichment in tumor associated B cells in baseline tumors of responders, suggesting that pre-existing B cell infiltration is a predictor of clinical outcome. Random forests were utilized to integrate Ig rearrangements, expressed TMB and tumor aneuploidy, into a predictive model of response that was superior to TMB (AUC = 0.89 and 0.65 respectively). Multivariate Cox proportional hazards analysis incorporating the same features was utilized to generate a risk score for each patient; those with high risk scores had a significantly shorter PFS compared to low risk patients (median PFS 1.45 months vs 29.01 months, log rank p = 3.4e-06, HR = 9.18, 95% CI: 3.14-26.85). Conclusions: Our findings highlight the multi-faceted interactions between the tumor and the immune system and the importance of pre-existing T and B cell immunity in driving clinical response and PFS after immune checkpoint blockade, laying the groundwork for integration of genomic and immune features into predictive models that may ultimately optimize therapeutic decisions.
Abstract Introduction: Pancreatic cancer has a poor prognosis especially when identified at advanced stages. Globally in 2020, >450,000 people died from the disease. For those with locally advanced ...or metastatic cancer, the standard of care treatment is chemotherapy. Preliminary studies show that some patients with advanced-stage disease respond to immune checkpoint blockade treatment. Determining response to therapy using imaging techniques can be challenging. There is a clinical unmet need for noninvasive approaches that can provide a real-time assessment of response to therapy in these patients. We evaluated two methods for assessing response to therapy using circulating cell-free DNA (cfDNA) in patients with metastatic pancreatic cancer treated with immune checkpoint inhibition and radiation as part of the CheckPAC study (NCT02866383). Methods: For all patients with available samples (n=40), we evaluated one pre-treatment blood draw and one on-treatment blood draw (26-65 days after treatment start). We extracted cfDNA from each sample and performed low-coverage 1-2x whole genome sequencing (WGS). We evaluated tumor and matched buffy coat for the 34 patients who had > 10% tumor purity using WGS with at least 60x and 30x coverage, respectively. We compiled the mutated bases in the tumor genome and examined these positions in the cfDNA. We counted the number of mutant observations from cfDNA WGS and divided them by the total number of observations at these bases to calculate whole-genome mutant allele frequency (MAF). We categorized responders based on >30% drop in MAF from baseline. In a tissue independent approach, we detected changes in genome-wide cfDNA fragmentation patterns using the DNA evaluation of fragments for early interception (DELFI) approach. We evaluated survival using DELFI scores of on-treatment samples. Samples were designated DELFI high (above median) or DELFI low (below median.) Log-Rank tests were used to compare survival curves. Results: For the mutation based approach, molecular responders (n=7) had a median progression free survival (PFS) of 485 days compared to 57 days for non-responders (n=9) (p=0.02). Molecular responders had a median overall survival (OS) of 694 days compared to 200 days for non-responders (p=0.062). Patients with DELFI low scores after therapy (n=14) had a longer median PFS than those with DELFI high scores (n=14) (307 days versus 70 days, respectively) (p=0.013). DELFI low patients also had a longer median OS than DELFI high patients (627 vs 158 days, respectively) (p<0.0001). Conclusions: These analyses suggest that WGS mutation-based and fragmentation cfDNA approaches can identify individuals with metastatic pancreatic cancer who respond to immune checkpoint inhibition. Incorporation of these molecular methods for evaluation of tumor burden may provide information for patient-physician decision making to improve patient quality of life. Citation Format: Carolyn Anna Hruban, Inna Chen, Daniel C. Bruhm, Shashikant Koul, Jennie Yao, Susann Theile, Kavya Boyapati, Nicholas A. Vulpescu, Akshaya Annapragada, Leonardo Ferreira, Stephen Cristiano, Zachariah H. Foda, Vilmos Adleff, Julia Johansen, Jillian Phallen, Robert B. Scharpf, Victor E. Velculescu. Liquid biopsy approaches for monitoring metastatic pancreatic cancer in immunotherapy treated patients abstract. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 2422.
Abstract Introduction: Repeat sequences comprise >50% of the human genome and structural and epigenetic changes in these regions are implicated in cancer. However, no systematic analysis of the ...compendium of repeat sequences has ever been performed in human cancer or cell-free DNA (cfDNA), largely due to the inability to identify and quantify repeat sequences genome-wide. We describe here the first comprehensive analysis of genome-wide repeat landscapes in cancer and demonstrate their utility in cfDNA liquid biopsies. Methods: We developed ARTEMIS (Analysis of RepeaT EleMents in dISease) an alignment-free, genome-wide approach for analyzing repeat landscapes in short read sequencing. This approach uses a de novo search of short sequences (kmers) in the telomere to telomere (chm13) reference genome to identify 1.2 billion 24-mers uniquely defining 1280 individual repeat types occurring genome-wide across 57 subfamilies and 6 families. We analyzed ARTEMIS kmers in whole genome sequences of 525 matched tumor/normal pairs from breast, colorectal, liver, lung, ovarian, cervical, prostate, thyroid, head and neck, gastric, and bladder cancers in the Pan Cancer Analysis of Whole Genomes (PCAWG), and in low coverage (1-2x) whole genome sequences of 1450 cfDNA samples from individuals with and without 8 types of cancer. Results: Analysis of ARTEMIS kmer repeat landscapes in 525 PCAWG tumors identified changes in all 1280 repeat element types, including 820 novel elements not previously known to be altered in cancer. A median of 807 repeat elements (range 246-1280) were altered in each tumor compared to its matched normal. The majority of changes were in repeat elements not previously described as altered in tumorigenesis and were most frequently found within Satellites, LINEs and SINEs, though changes were also observed in LTRs, Transposable Elements, and RNA Elements. A cross-validated cfDNA model using repeat landscapes (ARTEMIS) and fragmentation features (DELFI) detected individuals in a diagnostic cohort (n=287) across all stages of lung cancer with high performance (AUC 0.91, 95% CI 0.88-0.95) and was externally validated in a separate population (n=513). The locked model generated scores that correlated with circulating tumor mutant allele fractions for patients (n=19) undergoing targeted lung cancer therapy (r=0.80, p<2.2e-16), and stratified progression-free survival (p<0.001). ARTEMIS repeat landscape analyses of cfDNA also detected liver cancer in a high-risk cohort (n=208) of individuals with cirrhosis or viral hepatitis (AUC 0.91, 95% CI 0.87-0.95), and identified tissue of origin among seven tumor types (n=423). Conclusions: ARTEMIS reveals genome-wide repeat landscapes in human cancer, including in 820 novel elements not previously known to be altered in tumorigenesis. These repeat landscapes that can now be described are evaluable in the circulation and provide an avenue for noninvasive detection and characterization of cancer. Citation Format: Akshaya Annapragada, Noushin Niknafs, James R. White, Daniel C. Bruhm, Christopher Cherry, Jamie E. Medina, Vilmos Adleff, Carolyn Hruban, Dimitrios Mathios, Zachariah H. Foda, Jillian Phallen, Robert B. Scharpf, Victor E. Velculescu. Genome-wide repeat landscapes in cancer and cell-free DNA abstract. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 988.
Abstract Cell-free DNA (cfDNA) in the bloodstream is increasingly gaining attention as a diagnostic tool for the early detection of cancer. Nevertheless, the characteristics and sources of cfDNA ...fragmentation in the blood remain poorly understood. In this study, we sought to unravel the impact of DNA methylation and gene expression on the naturally occurring genome-wide fragmentation of cfDNA. We performed a comprehensive analysis of plasma samples from 969 individuals, including 182 individuals diagnosed with cancer (pancreatic, colorectal, ovarian, lung and breast cancer). In healthy individual cfDNA fragmentation patterns like DNA motifs at cfDNA fraqgments ends, recurrent cfDNA fragment start sites, cfDNA coverage and cfDNA fragment size were compared to publicly available data on cfDNA methylation and gene expression from myeloid cells. We discovered that the ends of cfDNA fragments, particularly those containing CGs or CCGs, displayed a distinct pattern of enrichment or depletion, respectively, at methylated CpG positions throughout the genome. This phenomenon is consistent with structural models suggesting an increased interaction of CG fragment ends with nucleosomes. cfDNA fragments were more prevalent (up to 3.7 fold) and of larger sizes (up to 4-5 base larger) when derived from regions of methylated CpGs or transcriptional start sites of genes that were not actively expressed. Differences in cfDNA fragmentation reflected specific biological pathways associated with their tissue of origin (like E2f transcription factors and blood cell metabolism genes in cfDNA from healthy individuals). Through analyses of cfDNA from patients with cancer, we identified a connection between tumor-related hypomethylation, increased gene expression, and a global reduction in cfDNA fragment size. This connection may offer an explanation for the generally smaller cfDNA fragments found in individuals with cancer. We found that cancer-specific methylation at CpG sites from pancreatic cancers were associated with widespread changes in cfDNA fragment ends, particularly for patients with pancreatic (test statistic = Welch two sample t-test; p<2.2e-16) and other cancers. These findings establish a direct link between epigenetic modifications and cfDNA fragmentation that may have implications for non-invasive detection of cancer. Citation Format: Michaël Noë, Akshaya V. Annapragada, Zachariah H. Foda, Jamie E. Medina, Dimitrios Mathios, Stephen Cristiano, Christopher Cherry, Daniel C. Bruhm, Noushin Niknafs, Vilmos Adleff, Leonardo Ferreira, Hari Easwaran, Stephen Baylin, Jillian Phallen, Robert B. Scharpf, Victor E. Velculescu. DNA methylation and gene expression as determinants of genome-wide cell-free DNA fragmentation abstract. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 974.
Abstract only
3018
Background: Analyses of cell-free DNA (cfDNA) in the blood provide a noninvasive diagnostic avenue for patients with cancer. However, cfDNA analyses have largely focused on ...targeted sequencing of specific genes, and the characteristics of the origins and molecular features of cfDNA are poorly understood. We developed an ultrasensitive approach that allows simultaneous examination of a large number of abnormalities in cfDNA through genome-wide analysis of fragmentation patterns. Methods: We used a machine learning model to examined cfDNA fragmentation profiles of 236 patients with largely localized breast, colorectal, lung, ovarian, pancreatic, gastric, or bile duct cancer and 245 healthy individuals. Estimation of performance was determined by ten-fold cross validation repeated ten times. Results: cfDNA profiles of healthy individuals reflected nucleosomal patterns of white blood cells, while patients with cancer had altered fragmentation patterns. The degree of abnormality in fragmentation profiles during therapy closely matched levels of mutant allele fractions in cfDNA as determined using ultra-deep targeted sequencing. The sensitivity of detection ranged from 57% to > 99% among the seven cancer types at 98% specificity, with an overall AUC of 0.94. Fragmentation profiles could be used to identify the tissue of origin of the cancers to a limited number of sites in 75% of cases. Combining our approach with mutation-based cfDNA analyses detected 91% of cancer patients. Conclusions: This effort is the first study to demonstrate genome-wide cell-free DNA fragmentation abnormalities in patients with cancer. Results of these analyses highlight important properties of cfDNA and provide a facile approach for screening, early detection, and monitoring of human cancer.
Abstract
Introduction: Ovarian cancer is the leading cause of death from gynecological related cancers worldwide. Most patients are diagnosed at late stages due to asymptomatic disease and lack of ...effective screening modalities. Additionally, in women with an ovarian mass the diagnosis of ovarian cancer can be challenging with many women having surgery only to discover a benign pathology. Liquid biopsies, including analyses of cell-free DNA (cfDNA) fragmentomes in the circulation, have shown promise for the early detection of cancer and may provide a useful avenue for detection of ovarian cancer.
Methods: To evaluate cfDNA fragmentomes for detecting ovarian cancer, we assessed plasma from 507 women, including 128 with ovarian cancers comprising high grade serous, endometrioid, mucinous and clear cell subtypes, 48 with benign masses, and 223 without cancer, as well as a validation cohort of 108 women with (n=14) and without (n=94) ovarian cancers from a separate institution. We obtained cfDNA from each individual and performed low-coverage (1-2x) whole genome sequencing. The cfDNA fragmentome data were analyzed with our DELFI (DNA evaluation of fragments for early interception) approach optimized for high specificity. We used a cross-validated machine learning model, and the fixed DELFI model was evaluated in the external validation cohort.
Results: Individuals with ovarian cancer had significantly higher DELFI scores than those without cancer (mean 0.59 vs 0.18, respectively, p < 0.0001) resulting in an AUC of 0.85 (95% CI = 0.80-0.90). DELFI was successful in identifying high-grade serous ovarian cancer across all stages, with sensitivities of 56%, 60%, 58% and 100% for stages I - IV, respectively, at 99% specificity. We further applied DELFI to evaluate women with ovarian masses in a prospective observational cohort (NL58253.031.16). Women with benign masses had DELFI scores lower than those with ovarian cancer (mean 0.23 vs 0.59, p< .0001) and were distinguished with an overall AUC of 0.80 (95% CI = 0.73-0.86). In the external validation cohort, women with cancer were distinguished from women without cancer with an AUC of 0.88 (95% CI = 0.72 - 1.0). At the DELFI score threshold with 99% specificity in the cross-validated cohort, the external validation cohort had specificity of 97% with an overall sensitivity of 79%. The cfDNA fragmentome profiles reflected chromosomal, chromatin, transcription factor binding site, and disease-specific pathway changes known to be altered in ovarian cancer. We are extending these efforts to over 1000 individuals with and without ovarian cancer and the integrated results will be presented.
Conclusion: Overall, we demonstrate the utility of cfDNA fragmentomes for noninvasive detection of ovarian cancer. These results may provide a feasible approach for ovarian cancer screening and management of patients with ovarian masses.
Citation Format: Akshaya V. Annapragada, Jamie E. Medina, Pien Lof, Dimitrios Mathios, Zachariah H. Foda, Michaël Noë, Sarah Short, Adrianna Bartolomucci, Daniel C. Bruhm, Euihye Jung, Jenna Canzoniero, Noushin Niknafs, Stephen Cristiano, Vilmos Adleff, Heather Symecko, Daan van de Broek, Stephen B. Baylin, Michael F. Press, Dennis Slamon, Gottfried Konecny, Susan Domchek, Ronny Drapkin, Jillian Phallen, Robert B. Scharpf, Christianne Lok, Victor E. Velculescu. Early detection of ovarian cancer using cell-free DNA fragmentomes abstract. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 773.