Genomic evolution and acquired resistance to preoperative chemoradiation therapy (CRT) in rectal cancer

613

Background: Following pre-operative CRT for locally advanced rectal cancer, ~20% of patients achieve a pathologic complete response and 20-40% have little to no response (NR) at time of surgery. Pathologic downstaging after CRT is associated with improved outcomes but molecular predictors of response are poorly understood. The objective of this study was to characterize somatic mutations and derive genomic biomarkers distinguishing complete and excellent partial responders (R) vs NR through whole exome sequencing (WES) in pre- and post-CRT tumor samples. Methods: Rectal cancer patients treated with pre-operative CRT and having either a R or NR were identified. DNA extraction from clinical FFPE samples and WES were performed on pre/post-CRT tumor and matched germline DNA followed by genomic analysis to identify somatic mutations, insertion/deletions, and clonal/subclonal mutational evolution. RNA extraction/sequencing and GSEA was performed in parallel. Results: We identified 34 pre- and post-CRT matched tumor samples from 17 patients treated with pre-operative fluoropyrimidine-based CRT (50.4 Gy) followed by surgery within 8-10 weeks. Nine and 8 patients were classified as NR and R, respectively. All FFPE passed quality control metrics and were successfully sequenced. Mean somatic mutational burden pre-CRT was 3.78 mutations/Mb (rg, 2.25-8.43), which, unexpectedly, was not significantly increased in the post-CRT samples (4.23 mutations/Mb; rg, 1.74-8.96). NR cases had more concurrent KRAS/TP53 (KP) mutations than R cases (67% vs 13%, p < 0.05). Related to this finding, NR was associated with epithelial-to-mesenchymal transition (p < 0.001) and potentially linked to high intra-tumoral heterogeneity pre-CRT (p = 0.08). Conclusions: We have established feasibility and a pipeline for performing WES on pre- and post-CRT FFPE tumor samples. Our data suggest that evolution of CRT resistance is not through hypermutation. In conjunction with recent observations from our group (Hong, JNCI 2017; Wang, Cancer Res 2017), our findings strongly support the clinical utility of the KP genotype as a biomarker of CRT resistance and begin to offer insight into the underlying mechanisms on a genomic level.

Metabolomic correlates of response in nivolumab-treated renal cell carcinoma and melanoma patients

3036

Background: Immune-checkpoint inhibition has been shown to be effective in a variety of cancers, including renal cell carcinoma (RCC) and melanoma. However, only a subset of patients with RCC and melanoma respond to anti-PD1 therapy. Given the importance of metabolism in the tumor immune microenvironment, we performed serum metabolomics in nivolumab-treated patients towards identifying novel non-invasive correlates of response and progression-free survival in immunotherapy-treated patients. Methods: We performed liquid chromatography-mass spectrometry on pre- and on-treatment serum samples from 79 patients with advanced melanoma (CA209-038 study) and 82 patients with metastatic RCC (CA209-009 study) receiving nivolumab. We precisely measured more than one-hundred named metabolites at baseline (prior to starting nivolumab), at 4 weeks and at 6 (melanoma) or 9 weeks (RCC) after initiation of treatment and correlated these with best overall response as well as progression-free survival (PFS). Results: In melanoma patients treated with nivolumab, the difference in mean levels of kynurenine (the product of IDO / TDO activity in tryptophan catabolism) between weeks 4 and 6 compared to baseline was significantly different between responders and non-responders (t-test with unequal variance p-value = 0.043 and p-value = 0.044 respectively). In RCC patients, we observed that patients with no response to nivolumab had significantly higher adenosine levels, than those who responded, at baseline and at 4 weeks after initiation of treatment (158% and 138% higher, t-test p-value = 0.0019 and p-value = 0.0011 respectively). RCC nivolumab-treated patients with higher (top quartile) baseline adenosine levels also had a significantly worse PFS (log rank test p-value = 0.004). Conclusions: In this first-of-its kind metabolomic analysis of peripheral blood from nivolumab-treated patients, we find that the change in kynurenine levels in melanoma patients correlates to response. In addition, higher baseline levels of adenosine in RCC patients are associated with worse PFS and lack of response to nivolumab. These results suggest a possible role for serum metabolites as biomarkers of benefit to PD1 inhibition.

Clinical actionability of germline testing in patients with limited colorectal polyps

e13027

Background: As classic adenoma-carcinoma sequence is the main process underlying most colorectal cancer (CRC), early detection and removal of colorectal adenomas is crucial in preventing CRC. Although adenomatous polyps are usually sporadic, several inherited CRC syndromes such as Familial adenomatous polyposis, MUTYH-associated polyposis and Lynch syndrome can present initially with colon polyps. The identification of germline defects in patients with colon polyps is thus critical for proper cancer risk counseling and CRC prevention. Current guidelines recommend germline testing for patients with more than 20 polyps or those with more than 10 polyps and a family history of CRC. However, the diagnostic yield of germline testing on otherwise healthy individuals with 10 or fewer colon polyps has not been well studied. Here, we performed a pilot study to evaluate the clinical actionability of germline genetic testing on these patients. Methods: A total of 13 cancer-free adults, who presented with colon polyps (n < 10) and who otherwise were not selected based on age of onset or family history underwent germline Exome Sequencing. Variants in 13 well-established CRC risk genes ( APC, CHEK2, MYH, MLH1, MSH2, MSH6, PMS2, NTHL1, BMPR1A, SMAD4, PTEN, STK11, TP53) were evaluated for pathogenicity. Results: A total of 13 patients (12 male, 1 female) were evaluated. The median age of presentation was 69 (range 49 to 88). The median number of adenomatous colon polyps was 1 (range 1 to 8). Two (15.4%, 95% CI = 1.9 - 45.4, Binomial Exact) patients had at least one disruptive mutation in the examined genes. One of these patients had a truncating mutation in APC (p.Arg216*) and presented with two tubulovillous adenomas at age 49. The second patient had 8 adenomas in distal colon and rectum at age 64, and harbored a known pathogenic mutation in MSH6(p.Arg1035*). Conclusions: This pilot study provides evidence that a relatively high percentage of patients presenting with a few colon polyps may have inherited defects in highly actionable genes. If validated in larger cohorts with appropriate population controls, these findings may influence the clinical care of such patients and their families and suggest germline molecular testing in those patients.

Loss-of-function of PBRM1 to predict response to anti-PD-1/PD-L1 therapy in metastatic renal cell carcinoma

3016

Background: Immune checkpoint inhibitors targeting programmed cell death-1 (PD-1) substantially improve patient survival in clear-cell renal cell carcinoma (ccRCC), but predictive biomarkers for efficacy have not yet been identified. Methods: We analyzed whole exome sequencing (WES) from a clinical trial of anti-PD-1 monotherapy (nivolumab) for ccRCC (N = 34) to discover genomic predictors of response to immune checkpoint therapy, and validated our findings in 28 ccRCC patients from 2 institutions treated with anti-PD-1 or anti-PD-L1 therapies. We defined 3 response groups: clinical benefit (CB) – complete or partial response by RECIST or stable disease with objective decrease in tumor burden and progression free survival (PFS) > 6 months - and no clinical benefit (NCB) – progressive disease with PFS < 3 months, with all other patients in intermediate benefit (IB). We further validated our findings in WES from 212 melanoma patients treated with immune checkpoint therapies in 3 published cohorts. Results: Biallelic loss of the chromatin remodeling subunit PBRM1, mutated in 34/62 (55%) patients across both cohorts and up to 41% of ccRCC overall, was the only gene mutation associated with CB in both the training (p = 0.0064; Pearson’s chi-squared) and validation cohorts (p = 0.043), and predicted both PFS and overall survival (OS) (p = 0.042 and 0.014, respectively; Kaplan-Meier). In 212 melanomas, truncating alterations in ARID2 – a closely related chromatin remodeler - were also enriched in responders after correcting for tumor mutational burden (p = 0.036), and having a truncating alteration in either PBRM1 or ARID2 significantly predicted overall survival (p = 0.022). In this ccRCC cohort, tumor mutational burden and loss of antigen presentation machinery were not associated with CB or NCB. Conclusions: Loss of chromatin remodeling subunits may impact response to immune checkpoint therapy in both ccRCC and melanoma. Further study in larger cohorts of immunotherapy-treated patients and functional characterization of ARID2 and PBRM1 in the context of the tumor-immune microenvironment will help to determine potential for further biomarker development.

A phase III randomized study comparing perioperative nivolumab vs. observation in patients with localized renal cell carcinoma undergoing nephrectomy (PROSPER RCC)

TPS4596

Background: The anti-PD-1 antibody nivolumab (nivo) improves overall survival (OS) in metastatic treatment refractory RCC and is generally tolerable. In 2017, there is no standard adjuvant therapy proven to increase OS over surgery alone in non-metastatic (M0) disease. Mouse solid tumor models have revealed an OS benefit with a short course of neoadjuvant PD-1 blockade compared to adjuvant therapy. Two ongoing phase 2 studies of perioperative nivo in RCC patients (pts) have shown preliminary feasibility and safety with no surgical delays or complications. The PROSPER RCC trial will examine if the addition of perioperative nivo to radical or partial nephrectomy can improve clinical outcomes in pts with locally advanced RCC. With the goal of increasing cure and recurrence-free survival (RFS) rates in M0 RCC, we propose a three-pronged, multidisciplinary approach of presurgical priming with nivo followed by resection and adjuvant PD-1 blockade. Methods: Tumorbiopsy prior to randomization is mandatory to ensure the correct diagnosis and will permit unparalleled correlative science in this global, randomized, unblinded, phase 3 National Clinical Trials Network study. 766 pts with clinical stage ≥T2 or any node positive M0 RCC of any histology will be enrolled. The study arm will receive nivo 240mg IV for 2 doses prior to surgery followed by nivo adjuvantly for 9 months (q2 wks x 3 mo followed by q4 wks x 6 mo). The control arm will undergo the current standard of care: surgical resection followed by observation. Pts are stratified by clinical T stage, node positivity, and histology. There is 84.2% power to detect a 14.4% absolute increase in the primary endpoint of RFS from the ASSURE historical control of 55.8% to 70.2% at 5 yrs (HR 0.70). The study is also powered to detect a significant OS benefit (HR 0.67). Key safety, feasibility, and quality of life endpoints are incorporated. PROSPER RCC exemplifies team science with a host of planned correlative work to investigate the significance of the baseline immune milieu and changes after neoadjuvant priming and to identify predictive gene expression patterns. Additional collaborations are welcomed.

Subclonal mutational heterogeneity and survival in cisplatin-resistant muscle-invasive bladder cancer

4512

Background: Biomarkers of survival and resistance in chemotherapy-resistant muscle-invasive bladder cancer (MIBC) are not well-characterized, but may inform management in this setting. Methods: Matched pre- and post-neoadjuvant cisplatin-based chemotherapy (NAC) tumor samples were obtained from 30 MIBC patients with gross residual disease (≥ pT2) at cystectomy, followed by whole exome sequencing of these “trios” (pre- and post-NAC tumor with matched germline samples). Phylogenetic analysis of matched tumor samples was performed to identify subclones, their associated mutations, and the corresponding enrichment in post-treatment tumors. Intratumoral heterogeneity was assessed by the proportion of mutations that were subclonal; the number of inferred subclones; and associated with overall survival using a Cox Proportional Hazards model. Results: Increased proportion of subclonal mutations in post-treatment tumors was associated with worse overall survival (HRR 1.86 [95% CI 1.12-3.06], p = 0.02), whereas pre-treatment proportion of subclonal mutations was only borderline statistically significant (HRR 1.48 [95% CI 0.99-2.20], p = 0.052). The total number of inferred tumor subclones in pre- or post-treatment tumor (or both) was associated with overall survival (HRR 1.60 [95% CI 1.05-2.43], p = 0.03), interpreted as a 60% increase in death rate per additional inferred subclone. While no single gene was statistically significantly enriched for new alterations in the post-chemotherapy resistant samples, we observed new post-treatment amplifications in cell-cycle genes ( E2F3, c-JUN), biallelic events in cell-cycle regulators ( FBXW7), and amplification of immune checkpoint genes ( PDL1/2). Conclusions: These results suggest that intratumoral heterogeneity (particularly post-therapy) predicts survival in a chemotherapy-resistant cohort. Further, alterations in cell cycle regulation may contribute to the mechanism of chemotherapy resistance. Finally, we observe evidence of immune checkpoint gene amplification post-treatment, suggesting that testing immune checkpoint blockade during NAC or, in high risk patients, following NAC may be warranted.

Whole exome analysis of patients (pts) with metastatic GIST (mGIST) demonstrating exceptional survival with imatinib (IM) therapy compared to those with short term benefit

11513

Background: Most patients with mGIST initially benefit from IM therapy with durable disease control (DC), i.e. objective responses and stable disease, with median duration of approximately two years. We reported exceptional long-term benefit (LTB) with DC and overall survival (OS) >14 years in a subset of mGIST pts treated with IM. We aimed to characterize tumor and normal genomes of exceptional LTB pts treated with IM and compare with short-term benefit (STB) pts. Methods: Among 87 mGIST pts enrolled between July 2000 and June 2001 in the B2222 trial of IM and followed prospectively at the Dana Farber Cancer Institute, we identified 10 LTB (>14 years of DC) pts, and 6 STB (<2 years of DC) pts on IM. Targeted genotyping ( KIT/PDGFRA)was performed in all tumors (n=16). Whole exome sequencing (WES) was performed on archival FFPE tumor samples from LTB and STB pts prior to any IM treatment. We compared WES results from LTB with STB pts to identify unique features of long-term DC and OS with IM. Results: KIT mutation in LTB pts were as follows: exon 11(6 pts), exon 9 (3 pts), and SDH-deficient with KIT/PDGFRA wild type (1 pt). In STB pts, mutated KITwas found 4 pts (exon 11) and 2 pts (exon 9). WES was successful in six LTB (five exon 11, one exon 9) and three STB (two exon 11, one exon 9) pts. A total of 1211 somatic mutations were observed (546 missense, 37 nonsense, 256 silent, 285 indels, 36 splice mutations). The mean somatic mutational burden was 3.42 mutations/Mb (range 1.18-4.93) and 3.34 mutations/Mb (range 1.06-6.68) among LTB and STB, respectively. Genes mutated in LTB but not in STB were MUC7 (4 pts), H1F0 (3 pts), ZKSCAN1 (3 pts), SLC24A1 (3 pts) and USP4 (2 pts). Conclusions: KRAB domain containing zinc finger (KRAB-ZNF) gene expression signatures have been associated with prediction of response to IM, and a possible role in response modulation to tyrosine kinase inhibitors in GIST. We found variants in ZKSCAN1, a gene encoding a transcriptional regulator of the KRAB subfamily of zinc finger, to be present in LTB but not in STB. KRAB-ZNF family of genes may be linked to LTB and exceptional survival with IM in mGIST; functional analyses will be important to test such hypotheses.

Enrichment of germline DNA-repair gene mutations in patients with colorectal cancer

1500

Background: Twin studies showed that 30% of all colorectal cancer (CRC) patients have an inherited genetic susceptibility. Several CRC predisposition genes have been described to date. However, mutations in these genes explain the risk in only 5-10% of CRC cases. In this study, we hypothesized that some of the CRC heritability may be explained by excess disruptive germline mutations in DNA repair genes (DRGs). Methods: Exome sequencing data of 716 in the discovery cohort (CanSeq and NHS/HPFS studies) and 1609 CRC patients in the validation cohort (TCGA and NSCCG studies) were used to evaluate germline variants in a pre-selected group of 42 DRGs and 12 known CRC risk genes. Frequencies of disruptive mutations in our cohorts were examined relative to 27173 non-Finnish European cancer-free adults from the ExAC cohort to evaluate for enrichment. Results: Of 716 patients in the discovery cohort, 27 (3.8%) patients harbored germline mutations in APC (n = 11), MSH6 (n = 2), MUTYH (n = 11), CHEK2 (n = 1) and TP53 (n = 2). Interestingly, germline mutations in ATM and PALB2 were significantly enriched in our CRC discovery cohort (OR = 2.7; P = 0.044; and OR = 4.8; P = 0.026, respectively). Evaluation of germline data from another 1609 CRC patients (validation cohort) also showed significantly higher rates of ATM mutations (5; 0.7%; OR = 2.1; P = 0.044), and a trend for enrichment of PALB2 mutations (3; 0.4%; OR = 2.8; P = 0.056). Secondary analysis of actionable germline mutations in a highly penetrant cancer risk gene set ( ATM, BRCA1, BRCA2, BRIP1 and PALB2) suggest a broader enrichment trend in CRC patients for these genes (Discovery: OR = 1.7; P = 0.06; Validation: OR = 2; P = 1.96e-04). Conclusions: Our analysis of germline variants in 2325 CRC patients showed the first robust evidence for germline ATM mutations to confer a higher risk of developing CRC. We also presented evidence to support PALB2 as a potential novel CRC risk gene. Overall, our study shows that mutations in some DRGs may explain some of the missing CRC heritability. It also indicates that a significant percentage of CRC patients, who carry mutations in highly actionable genes where cancer screening recommendations for patients and families do exist, are not captured with current testing recommendations.

Optimizing somatic genomic reporting and physician interpretation with web-based, interactive technologies

1517

Background: The increased availability of tumor genomic profiling is revolutionizing oncology. However, the promise of precision care will not be realized if providers misinterpret complex genomic data. Methods: We created web-based, interactive reports with enhanced data visualization elements and embedded decision support for > 300 gene panels. We conducted a randomized vignette-based survey study to determine whether exposure to the interactive reports, as compared to static reports, improves physicians’ genomic understanding and report-based satisfaction. Overall comprehension and satisfaction scores were calculated across three vignettes (possible range 0-18 and 1-4 respectively, higher score correspond to improved endpoints). Results: 105 physicians at a major cancer center participated (29% participation rate); 67% medical, 20% pediatric, 7% radiation and 7% surgical oncology; 37% female. Prior to viewing the case-based vignette reports, 34% of physicians reported that they found it difficult to make treatment recommendations based on the standard report in their routine practice. After viewing the case-based vignettes, physicians’ overall comprehension scores did not differ significantly by report type (mean score interactive 11.6 vs. static 10.5, difference = 1.1, 95% CI -0.3, 2.5, p = 0.13). However, physicians who viewed the interactive report were more likely to correctly assess sequencing quality (p < 0.001) and understand when reports needed to be interpreted with caution (e.g., if low tumor purity, p = 0.02). Overall satisfaction scores were significantly higher in the interactive group than the static group (mean score 2.5 vs. 2.1, difference = 0.4, 95% CI 0.2, 0.7, p = 0.001). Of the 92 physicians who endorsed the need for additional genomic support for providers, 66% reported that interactive genomic reports would be helpful. Conclusions: Interactive, genomic reports may improve physicians’ ability to accurately assess genomic data and increase physician satisfaction. To advance the field, further research in representative provider populations is warranted and efforts to integrate interactive genomic reports into electronic health records are needed.

The long tail of significantly mutated genes in prostate cancer

131

Background: The mutational landscapes of primary and metastatic PCa have been robustly analyzed in multiple whole exome sequencing (WES) studies. We hypothesized that an aggregate, uniform analysis of all data generated to date would enable discovery of new significantly mutated genes and pathways not previously associated with PCa, and shed more light onto the genetic differences between primary and metastatic PCa. Methods: We assembled and uniformly analyzed a cohort of 1,021 tumor and matched germline primary and metastatic PCa whole exomes (686 primary, 335 metastatic), and performed mutational significance analysis using statistical and biological approaches to determine which genes and pathways are recurrently altered. Results: We identified 117 significantly mutated genes (Mutsig q<0.1) in PCa, which included many novel genes and pathways. These include epigenetic modifiers [KMT2C (6%), KMT2D (6%), and KDM6A (2.7%)], regulators of the SWI/SNF complex [SMARCA1 (1.1%), ARID1A (1.5%), ARID1B (1.3%), ARID2 (1.3%), and PBRM1 (0.7%)], and the splicing pathway [SF3B1 (1.1%) and U2AF1 (0.5%)]. In addition, we identified mutations in FUBP1 (0.4%), a splicing regulator involved in the regulation of MDM2 splicing. We also found truncating mutations in SPEN, a hormone inducible transcriptional repressor, in 2.8% of samples, similar to the frequency observed in breast tumors. Finally, a comparison of primary and metastatic samples enabled discovery of a genetic profile associated with metastatic disease, including AR amplifications and mutations, and loss of TP53, PTEN, and RB1. At lower frequency, metastatic tumors showed enrichment in mutations in MLLs (KMT2C/D), APC, CDK12, BRCA2, CTNNB1, and amplifications of MYC and CCND1. Conclusions: Through aggregation and uniform genomic analysis, we refined the map of somatic mutations in PCa and identified cancer genes and pathways not previously associated with this disease. Our findings may inform patient stratification and translational investigation.

The Prostate Cancer Project (PC Project): Translational genomics through direct patient engagement

199

Background: While there has been substantial advancement in the genomic understanding of prostate cancer (PCa), there is still much to be discovered. Additional progress is dependent upon obtaining a large amount of clinically annotated genomic data. As PCa is often treated in a community setting, where research samples are not collected, we are starting a direct-to-patient nationwide research initiative where patients can donate their medical records and biospecimens to accelerate research. Previously, we launched the metastatic breast cancer project (MBCproject; mbcproject.org) that leverages social media to engage the MBC community. Based on the initial success with this approach, we now aim to build out the PCproject. Methods: In collaboration with patients, we are developing a website to enable participation in the PCproject. Enrolled patients will be sent a saliva kit, used for germline DNA. We will also obtain medical records. Metastatic patients will also be sent a blood draw kit for circulating tumor DNA (ctDNA). Whole exome sequencing of the ctDNA will be performed. We will use the recruitment infrastructure, clinical record abstraction, and biospecimen processing developed for the MBC project. The data will be shared widely with the research community. Aggregate study results will be reported to patients. Results: In the first year of the MBCproject, 2912 MBC patients from all 50 states enrolled. 2766 (95.0%) completed the 16-question survey about their cancer, treatments, and demographic information. 1716 (58.9%) completed the online consent form permitting acquisition and analysis of medical records, tumor tissue, and saliva samples. 936 (68.8% success rate) saliva samples have been received. To date, we have obtained medical records from 155 patients (72.1% success rate) and tumor samples from 60 patients (72.3% success rate). Based on initial recruitment and surveys among PCa patients, we estimate that 500 patients will enroll in 2017. Conclusions: Based on experience from the MBC project, we will partner directly with patients to recruit and drive the PCproject forward. Remote acquisition of medical records, saliva samples, and tumor tissue for patients located throughout the US is feasible.

Impact of immune checkpoint protein expression in tumor cells and tumor infiltrating CD8+ T cells on clinical benefit from PD-1 blockade in metastatic clear cell renal cell carcinoma (mccRCC)

477

Background: PD-1 blockade with nivolumab has demonstrated efficacy in mccRCC patients who have failed prior therapy; however, durable benefit is observed only in a subset of patients. Correlative studies have demonstrated that tumor PD-L1 expression alone fails to reliably identify patients likely to benefit. Therefore, the development of more robust predictive markers is warranted. Methods: We studied 20 mccRCC patients treated with nivolumab (n=17) or atezolizumab (n=3) with distinct clinical outcomes: 8 pts who experienced durable clinical benefit (DCB) for ≥ 12 months and 12 pts with limited clinical benefit (LCB) for ≤ 6 months from start of therapy. Expression of PD-L1 and PD-L2 on tumor cells and density of tumor infiltrating CD8+ T cells and Foxp3+ cells were evaluated by immunohistochemistry. Percentages of tumor infiltrating CD8+T cells expressing the immune-inhibitory molecules PD-1, TIM-3, or LAG-3 either alone or in combination were determined by multiplex immunofluorescence, using the Inform algorithm (Perkin Elmer). The association between clinical benefit (CB) and biomarker expression was assessed using Fisher exact test. Results: Baseline patient characteristics were similar in the DCB and LCB groups. When analyzed separately, tumor cell expression of PD-L1 or PD-L2 showed no association with CB. However, when positivity was defined as expression of either or both PD-L1(≥ 1% cutoff) and PD-L2 (≥ 5% cutoff), a significant association with CB was found (7/8 DCB vs 4/11 LCB, p = 0.04). A low percentage of tumor infiltrating CD8+PD-1+ T cells co-expressing either or both TIM-3 and LAG-3 (75thpercentile cutoff) was also significantly associated with CB (8/8 DCB vs 7/12 LCB, p = 0.05). Conclusions: Our results suggest that expression of a PD-1 ligand on tumor cells (PD-L1 and/or PD-L2), and a low percentage of severely exhausted T cells, i.e. CD8+PD-1+ T-cells co-expressing the immune-inhibitory receptors TIM-3 and/or LAG-3, might represent valuable predictive biomarkers for response to PD-1 blockade in mccRCC. Independent validation in a larger patient cohort is ongoing.

Genomic features of primary and recurrent anal squamous cell carcinoma

556

Background: Approximately 7000 cases of anal squamous cell carcinoma (ASCC) are diagnosed annually in the US, and the incidence is increasing. Combination chemoradiotherapy (CRT) is the standard of care for locally-advanced ASCC, and the 5-year survival rate is approximately 70%. Unlike many other tumor types, no large-scale genomic studies have been reported for anal cancer. In order to characterize the mutational landscape of ASCC and identify potential therapeutic targets, we perform comprehensive genomic analysis of a pilot cohort of ASCC cases. Methods: We performed whole exome sequencing of tumor and matched germline DNA from a pilot cohort of twelve patients with locally advanced (Stage II-IVA) ASCC cases treated at Dana-Farber Cancer Institute. All patients received concurrent chemoradiotherapy (CRT): seven ‘responders’ had complete response and no evidence of recurrence with a minimum of one year follow-up, while five ‘non-responders’ had residual or recurrent disease following CRT. For non-responders, both primary and residual/recurrent tumors were analyzed. Results: The overall mutation rate was 3.7 mutations per megabase (Mb). Recurrent mutations in several known cancer genes were observed. Five of twelve cases had a hotspot mutation in FBXW7, including 3 of the 5 non-responders. Known oncogenic mutations were also observed in PIK3CA (3 tumors) and NFE2L2/KEAP1 (3 tumors). Analysis of paired primary and recurrent samples revealed surprising examples of distinct driver mutations in clonally-related tumors. Copy number analysis revealed focal amplifications of chromosome 3q. Conclusions: This study represents one of the first genome-scale analyses in ASCC. The overall mutation rate was similar to other HPV-associated squamous cell carcinomas, and recurrent mutations in several known cancer genes were observed. Analysis of paired primary and residual/recurrent tumors revealed surprising heterogeneity. To extend our findings, we are currently performing a similar analysis in a larger extension cohort of ASCC tumors.

Integrated genomic correlates of response to PD-1 inhibitor nivolumab in metastatic renal cell carcinoma (mRCC)

545

Background: Nivolumab (nivo) is a programmed death-1 (PD-1) inhibitor that has recently been shown to increase overall survival in patients with mRCC. The association between benefit from nivo with neoantigen load, mutation load and signatures of immune infiltration has yet to be explored in mRCC. Methods: Publicly available whole exomes (WES) and whole transcriptomes (RNA-seq) from over 300 clear-cell RCC patients in the Cancer Genome Atlas (TCGA) were analyzed for mutational load, neoantigen load and expression of immune-related genes. HLA types were inferred from WES with Polysolver and neoantigens were predicted from somatic nonsynonymous mutations with NetMHCpan3.2. The same analysis pipeline was applied to paired pre-treatment germline and metastatic tumor specimens from 9 patients who received nivo for mRCC: 3 responders (CR or PR by RECIST; R) and 6 non-responders (SD or PD by RECIST; NR), all with available pre-nivo fresh biopsies of metastatic lesions. Matched RNA-seq was available on a subset of these patient samples (3 R, 3 NR). Results: Across the TCGA and nivo-treated patients cohorts, RCC had relatively few nonsynonymous mutations and neoantigens. Among the nivo-treated patients, neoantigen load was significantly higher in NR compared to R (p = 0.048), but nonsynonymous mutation load was not. An exceptional R who experienced CR (PFS > 30 months) had outlying higher expression of selected immune-related genes compared to the 8 other patient samples (p < 0.05 for PD-L1, PD-L2; p < 0.01 for CTLA4, PD-1, PRF1; p < 0.001 for GZMA, BTLA, CD8A), and was in the top 1-5% of expression of these genes among all TCGA data. Bootstrapping analysis comparing mean expression in R vs NR compared to an empirical distribution of TCGA RNA-seq data showed significantly higher expression of PD-1, PD-L1, GZMA, LAG3, IDO1, ICOS, and BTLA in R vs. NR (p < 0.05). No genes were enriched for nonsynonymous mutations in R or NR. Conclusions: This analysis suggests that both DNA- and RNA-level data may be relevant for explaining clinical benefit from nivo in mRCC. In contrast with results from other tumor types, responders to nivo did not have more mutated tumors, though immune-related gene expression may be related to response.

Genomic landscape of high-grade T1 micropapillary bladder tumors

299

Background: The genomic landscape of high-grade T1 micropapillary bladder tumors (HGT1micropap) is unknown. Clinically, micropapillary bladder cancer is an aggressive and possibly lethal disease. Our main objective was to assess the genomic landscape of HGT1micropap through identifying mutations, insertions/deletions (indels), translocations, and copy number variations (CNVs). Methods: We prospectively identified nine HGT1micropap with 45.4 months of median follow up. Patients were treated in a uniform manner using TUR, BCG, and appropriate follow up. We performed whole exome sequencing using Ilumina Exome _v5 plus translocation. Mutations and indels were called using the Firehose pipeline. CNVs were called using ExomeCNV. We examined the mutational landscape and compared the genomic alterations to TCGA (>T2, n=131)2 and publicly available data on non-muscle invasive bladder tumors (Ta/T1, n=37)1. Results: Within the HGT1micropap, mutations on TP53, KMT2D, TSC1, and ATM were suggested to occur more frequently compared to the NMIBC control group1. FGFR3 was seen at the expected frequency for NMIBC. The mutations of interest are presented in Table 1 with the percentage seen in the other cohorts. Of interest, TSC1 was seen in higher frequency in micropapillary than in the NMIBC or the TCGA cohort2. We did not see any patterns between CNVs and mutations. We also saw two patients with severe chromothripsis. 3 patients had loss of chromosome 9 or 9q without any other severe chromosome alterations. CNV alterations will be presented and compared to MIBC. Conclusions: In this preliminary analysis, our HGT1micropap, showed a mutational landscape more similar to MIBC compared to NMIBC bladder landscape. We did not find any clear driver of the micropapillary histology at the exome level in this limited sample of patient, which may indicate that tumor heterogeneity or epigentic changes may be driving this aggressive phenotype. [Table: see text]

Comprehensive genomic profiling of 443 cases of renal cell carcinoma to reveal frequent clinically relevant genomic alterations

433

Background: Despite the availability of targeted therapy, effective management of advanced RCC is an unmet medical need as treatment is not personalized and is not guided by patient-specific genomic alterations (GAs). To assess the spectrum of clinically relevant GAs (CRGAs) in advanced RCC, comprehensive genomic profiling (CGP) of clinical RCC samples was performed with the goal of informing use of existing and novel targeted therapies. Methods: DNA was extracted from 40 microns of FFPE sections from 443 consecutive patients with relapsed/metastatic RCC. CGP was performed on hybridization-captured, adaptor ligation based libraries to a mean coverage depth of 646X for 3,230 exons of 182 cancer-related genes plus 37 introns from 14 genes frequently rearranged in cancer. The CGP assay included base substitutions (SUB), INDELs, copy number alterations (CNA) and fusions/rearrangements. CRGA were defined as GA linked to drugs on the market or under evaluation in mechanism driven clinical trials. Results: There were 73% male and 27% female patients with a mean age of 56 years, and the majority of cases were advanced stage with 198/443 specimens (44.6%) from metastatic sites. 400/443 patients (89%) had at least 1 GA on CGP with a mean 3.1 GA/case. 396/400 RCC harboring GA had at least 1 CRGA involving 111 individual genes with a mean of 1.32 CRGA/UC. The most common novel CRGA in order of frequency were: CDKN2A (21%), BAP1 (12%), ATM (11%), PTEN(8.5%), TSC1(8.3%), mTOR (7%), MET (6.5%), AR (5.3%), DNMT3A (5%) and TSC2 (5%). Moreover, VHL harbored a diversity of GA’s in in 49% of cases. Collecting duct carcinomas harbored an enrichment of NF2 truncating alterations (>40%), distinct from renal medullary carcinoma which did not harbor such GA. Multiple clinical antitumor responses to targeted therapies will be presented. Conclusions: Using a CGP assay capable of detecting all classes of GA simultaneously, a high frequency of CRGA was identified in a large series of patients with advanced RCC. The diversity of CRGA suggests opportunities for the rational application of existing and investigational targeted therapies, and for possible deeper characterization of histological types of RCC.

Whole-exome sequencing (WES) predicting two extreme phenotypes of response to VEGF-targeted therapies (VEGF-TT) in patients with metastatic clear cell renal cell carcinoma (mRCC)

422

Background: There is significant variability in the response to VEGF-TT in mRCC with no validated predictive biomarkers. We hypothesized that whole exome analysis might identify markers of response and resistance to VEGF-TT in mRCC. Methods: mRCC patients who received first-line sunitinib or pazopanib and were in two extreme phenotypes of response were identified. Extreme responders (ER) were defined as PR or CR for ≥3 years (n=10) and primary refractory patients (PRP) were defined as PD within the first 3 months of therapy (n=10). WES was performed in pre-treatment specimens and established Broad Institute analytical pipelines were utilized to identify point mutations and copy number alterations across the exome. ER (n=4) and PRP (n=4) who were part of TCGA project for clear-cell RCC were included in this analysis. Nonsense, missense and indel mutations in established or novel RCC genes were investigated. Results: IMDC prognostic scores were not significantly different between the two groups. VHL mutations were observed at similar frequency in ER and PRP, overall 57%. Mutations in PBRM1 were identified in 7 ER (50%) vs. 1 PRP (7%) (p=0.03). In addition, mutations in TP53 were only found in PRP (p=0.09). No other gene had mutations that were associated with either response or primary refractory disease (Table). Conclusions: In this pilot study, PBRM1 mutations were associated with extreme response to VEGF-TT. However, this was exploratory, and multivariable analysis was not performed. Analysis of additional patient samples is ongoing to confirm or refute this association. If true it suggests that epigenetic effects of PBRM1 mutation may contribute to response to VEGF-TT in mRCC. [Table: see text]

A phase II study of everolimus in patients with aggressive RAI refractory (RAIR) thyroid cancer (TC)

6023

Background: We present results of an open label phase II study of the mTOR inhibitor Everolimus in patients (pts) with RAIR TC. Methods: Pts with metastatic, incurable RAIR TC who had shown radiographic progression within 6 months prior to enrollment received Everolimus 10mg orally once daily. Responses were monitored by CT's every two months. The primary endpoint was progression free survival. Sequential biopsies were obtained in selected pts. Results: Enrollment to the differentiated TC (DTC) cohort finished in Jan 2013 and included 33 pts, among them 11 with Hurthle cell TC. Exploratory cohorts enrolled 10 pts with medullary [MTC] and 5 with anaplastic [ATC] with 2 added openings remaining for ATC. For the DTC cohort, median time on study to date is 10 months (mo) (<1-23+). 31 pts are evaluable at this time. PFS in the DTC cohort by Kaplan-Meier (K-M) analysis is 16.0 mo (95%CI 10-NR). Currently, disease stability for 6 and 12 mo or more was achieved in 18 and 10/31 pts, respectively, 11 pts remain on study. Median OS was not reached but 1 year survival by K-M analysis was 76%. One pt achieved a PR. 3 pts with DTC underwent sequential biopsies which revealed activation of autophagy while markers for apoptosis were not detected. Among 10 MTC pts, one achieved a PR and 9 pts had stable disease for 6 mo or more (6-33+). Among 5 ATC pts, 3 progressed, one has ongoing disease stability for 5 mo. One patient achieved a complete response that lasted for 18 mo and whole exome sequencing revealed somatic loss of function mutation affecting the Tuberous Sclerosis 2 (TSC2) protein, a negative regulator of mTOR activity [TSC2 (Q1178*) and FLCN (R17fs)]. Most common treatment-related adverse events were as anticipated and included fatigue, stomatitis and infections. Grade (gr) 3 events included infection 5, weight loss 3, leukopenia 3, thrombocytopenia 3, fatigue 3, hypophosphatemia 2, stomatitis 2, pneumonitis 1 and thrombosis 1pts. One pt had gr 4 hypercholesterinemia and one pt had gr 4 leukopenia. Conclusions: Everolimus has significant anti-tumor activity in pts with advanced TC. Activation of autophagy could account for high rate of disease stability. Sequencing may identify mechanistic basis and predictive markers for treatment response. Clinical trial information: NCT00936858.

The genetic landscape of clinical resistance to RAF inhibition in melanoma

11009

Background: Although single-agent RAF inhibition has proved effective in metastatic BRAFV600-mutant melanoma, most patients relapse and some are intrinsically resistant. While several genetic resistance effectors have been identified, a comprehensive assessment of the genetic resistance spectrum in a large patient cohort may further inform resistance patterns and treatment strategies. Methods: Pre-treatment and post-relapse biopsies were obtained from BRAFV600melanoma patients treated with vemurafenib or dabrafenib. Whole exome sequencing of tumor and normal samples was performed to identify exome-wide mutations, insertion/deletions, and chromosomal copy number alterations. Since somatic allelic fraction comparisons across treatment time points must account for variable stromal admixture between biopsies, novel algorithms were applied to impute purity and tumor cell ploidy, thus determining the “cancer cell fraction” (CCF) of each alteration. Those with enriched CCF in post-relapse samples were nominated as potential resistance effectors. Results: 30 patients (towards a goal of 77) were sequenced to a mean depth of 145-fold coverage. In acquired resistance patients (on drug ≥ 12 weeks, n=18), NRAS Q61R (28%) and BRAF amplifications (6%) were observed. The remaining patients had multiple clonally enriched alterations in the post-relapse tumors that may drive resistance, including MEK1 mutations at residues previously noted in an in vitro mutagenesis screen for vemurafenib resistance. In the intrinsic resistance cohort (on drug < 12 weeks, n=12), multiple putative resistance mutations not previously observed in the setting of acquired resistance were identified. Conclusions: In addition to reported genetic resistance mechanisms, we identified new MEK1 mutations in resistant samples. For the majority of patients (67%) without events in these genes, computational algorithms identified new candidate drivers of RAF inhibitor resistance. Alterations that predict intrinsic or acquired resistance may also inform new approaches to melanoma therapy. Future studies incorporating longitudinal biopsies and genetic profiling will underpin a robust assessment of clinical resistance in melanoma and other cancers.

Whole exome and whole transcriptome sequencing in melanoma patients to identify mechanisms of resistance to combined RAF/MEK inhibition

9015

Background: The RAF inhibitors vemurafenib and dabrafenib (D) and the MEK inhibitor trametinib (T) improve survival as monotherapies in BRAF-mutant melanoma. Since clinical mechanisms of resistance (MoR) result in MAPK pathway reactivation, recent efforts have focused on combined targeting of RAF and MEK. The combination of D and T (D/T) increased progression-free survival and response rate compared with D alone (Flaherty et al, NEJM, 2012). The MoR to this combination remain unknown. Methods: To look for clinical MoR to combined RAF/MEK inhibition, we performed whole exome (WES) and whole transcriptome sequencing (RNASeq) on tumors from 4 patients (pts) with acquired resistance and 1 pt with intrinsic resistance to D/T. Pre-treatment and post-resistance tumors from all pts were analyzed for point mutations, insertions/deletions, copy number alterations, alternatively spliced transcripts, rearrangements, and expression changes. Results: In 2 of 4 pts with acquired resistance, WES identified mutations in MEK1 and MEK2 that were undetectable in the pre-treatment tumors. In the 3rd pt, RNASeq identified an alternatively spliced isoform of BRAF lacking exons 2-10, also undetectable in the pre-treatment tumor. In the 4th pt, no obvious MoR were seen, though multiple alterations were enriched in the post-resistance tumor. The pt with intrinsic resistance had several alterations in genes that conferred resistance to RAF/MEK inhibition when overexpressed in BRAF-mutant cell lines. Integration of WES and RNASeq data also identified several co-existing alterations that may synergize to increase resistance. Conclusions: Analysis of combined WES and RNASeq data from pt samples provides a more complete picture of clinical MoR to MAPK-targeted therapy. Post-resistance tumors from 3 of 4 pts with acquired resistance to D/T had alterations in MAPK genes not detectable in the pre-treatment tumors, suggesting that resistance involves reactivation of the MAPK pathway despite combined RAF/MEK inhibition. Alternative dosing of current agents, more potent RAF/MEK inhibitors, and/or inhibition of the downstream kinase ERK may be needed for durable control of BRAF-mutant melanoma.

A heuristic platform for clinical interpretation of cancer genome sequencing data

10502

Background: The ability to identify and effectively sort the full spectrum of biologically and therapeutically relevant genetic alterations identified by massively parallel sequencing may improve cancer care. A major challenge involves rapid and rational categorization of data-intensive output, including somatic mutations, insertions/deletions, copy number alterations, and rearrangements into ranked categories for clinician review. Methods: A database of clinically actionable alterations was created, consisting of over 100 annotated genes known to undergo somatic genomic alterations in cancer that may impact clinical decision-making. A heuristic algorithm was developed, which selectively identifies somatic alterations based on the clinically actionable alterations database. Remaining variants are sorted based on additional heuristics, including high priority alterations based on presence in the Cancer Gene Census, biologically significant cancer genes based on presence in COSMIC or MSigDB, and low priority alterations in the same gene family as biologically significant cancer genes. The heuristic algorithm was applied to whole exome sequencing data of clinical samples and whole genome sequencing data from a cohort of prostate cancer samples processed using established Broad Institute pipelines. Results: Application of the heuristic algorithm to the prostate cancer whole genome rearrangement data identified 172 (out of 5978) rearrangements involving actionable genes (averaging 2-3 events per tumor). Furthermore, two clinical samples processed prospectively were analyzed, yielding three potentially actionable alterations for clinical review. Conclusions: The heuristic model for clinical interpretation of next generation sequencing data may facilitate rapid analysis of tumor genomic information for clinician review by identifying and prioritizing alterations that can directly impact care. Our platform can also be applied to research data to prospectively explore clinically relevant findings from existing cohorts. Future analytical approaches using heuristic or probabilistic algorithms should underpin a robust prospective assessment of clinical cancer genome data.