Search

PURPOSE National Cancer Institute (NCI)–sponsored clinical trial network studies frequently require biopsy specimens for pharmacodynamic and molecular biomarker analyses, including paired pre- and post-treatment samples. The purpose of this meeting of NCI-sponsored investigators was to identify local institutional standard procedures found to ensure quantitative and qualitative specimen adequacy. METHODS NCI convened a conference on best biopsy practices, focusing on the clinical research community. Topics discussed were (1) criteria for specimen adequacy in the personalized medicine era, (2) team-based approaches to ensure specimen adequacy and quality control, and (3) risk considerations relevant to academic and community practitioners and their patients. RESULTS AND RECOMMENDATIONS Key recommendations from the convened consensus panel included (1) establishment of infrastructure for multidisciplinary biopsy teams with a formalized information capture process, (2) maintenance of standard operating procedures with regular team review, (3) optimization of tissue collection and yield methodology, (4) incorporation of needle aspiration and other newer techniques, and (5) commitment of stakeholders to use of guideline documents to increase awareness of best biopsy practices, with the goal of universally improving tumor biopsy practices.

e16056 Background: Colorectal cancer (CRC) is the 3rdmost common form of cancer in the US, responsible for over 50,000 death each year. Therapeutic options for advanced colorectal cancer are limited, and there remains an unmet clinical need to identify new therapies to treat this deadly disease. To address this need, we have developed a precision medicine pipeline that integrates high throughput chemical screens with matched patient-derived cell lines and patient-derived xenografts (PDXs) to identify new treatments for CRC. Methods: We used high-throughput chemical screens of 2,100 compounds across five low-passage, patient-derived CRC cell lines. These results were validated using dose-response IC50curves for CDK1, CDK2, CDK9 or CDK1/2/9 inhibitors and by siRNA-mediated knockdown of CDK9 with or without CDK2 inhibition. Cell cycle arrest analysis was performed by flow cytometry and anaphase catastrophe was analyzed by immunofluorescence staining. For in vivo studies, matched PDXs were treated with either CDK2, CDK9 or dual CDK2/9 inhibitors. Results: We identified the CDK inhibitor drug class as among the most effective cytotoxic compounds across all five CRC lines. Further analysis of the CDK inhibitor class revealed that combined targeting of CDK1, 2, and 9 was the most effective, with IC50 in the range of 110 nM to 1.2 μM. We further validated the efficacy of combined CDK2/9 inhibition using siRNA-mediated knockdown of CDK9 in the presence of a CDK2 inhibitor(CVT-313), and showed that CDK9 knockdown acted synergistically with CDK2 inhibition. Dual CDK2/9 inhibition led to significant G2/M cell cycle arrest and anaphase catastrophe. Finally, combined CDK2/9 inhibition in vivo synergistically inhibited PDX tumor growth as compared to single-agent CDK inhibitors. Conclusions: Our precision medicine pipeline revealed CDK2/9 dual inhibition as a combinatorial therapy to treat CRC and can also be used to identify new and novel therapies

e16079 Background: ICIs have revolutionized treatment for mRCC; however there are limited predictive biomarkers for response to ICIs. PD-L1 status is still controversial, demonstrating little predictive utility in mRCC. TMB is predictive for response to ICIs in melanoma and non-small cell lung cancer (NSCLC), but has not been validated in mRCC. Here, we assess the correlations between TMB and PD-L1 status with outcomes to ICI treatment in mRCC. Methods: 34 patients (pts) with mRCC who had previously received ICIs at Duke Cancer Institute were identified. Tumor samples were retrospectively evaluated using a Personal Genome Diagnostics Assay for somatic variants across > 500 genes, as well as TMB and microsatellite status. PD-L1 status was tested via the Dako 28-8 PD-L1 IHC assay. Deidentified clinical information was extracted from the medical record and tumor response was evaluated based on RECIST criteria. Results: Pts were grouped by overall response following ICI therapy into either progressive disease (“PD”, n = 18) or disease control group (“DC”, n = 16), defined as either stable disease, partial response, or complete response. Pts displayed a TMB range from 0.36 to 12.24 mutations/Mb with a mean score of 2.83 muts/Mb, with no significant difference between the PD and DC groups (mean 3.01 muts/Mb vs. 2.63 muts/Mb, p > 0.05). 9 of 32 evaluable samples were PD-L1 positive, with 4 in the PD group and 5 in the DC group. Notably, the DC group displayed a significant enrichment of mutations in genes affiliated with DNA repair (including BRCA1, BRCA2, FANCA, FANCB, FANCG, FANCM, MSH3, MSH6, RAD50, RAD51C, RAD51D, RAD54B, RECQL4, and SLX4; p = 0.0444). DNA damage gene mutations were found in 8/10 (80%) metastatic tumor specimens and 14/24 (58%) primary tumors. Conclusions: Overall, in this mRCC cohort, neither TMB nor PD-L1 correlated with patient outcomes or with ICI response. Furthermore, high TMB was not significantly associated with PD-L1 expression within the samples. The higher frequency of mutations in DNA repair genes in the DC group suggests potential use as a predictive signature for ICI response, warranting future prospective studies. Further studies with matched primary-metastatic samples would be beneficial to determine if DNA repair mutations occur more frequently in metastatic versus primary tumor specimens.

e14259 Background: Immune checkpoint inhibitors (ICIs) are standard of care for mRCC and mUC patients (pts). PD-L1 status is gaining importance as a predictive biomarker, particularly for cisplatin-ineligible mUC. PD-L1 positivity is defined differently by PD-L1 assay and tumor type, with limited concordance studies. Given real-world limitations in PD-L1 testing, assay concordance studies are needed to distinguish positive (pos)/negative (neg) results and treatment selection. We compared Dako 28–8 and Ventana SP142 assays in mRCC and Dako 22C3 and Ventana SP263 assays in mUC. Methods: 32 pts with mRCC and 18 pts with mUC who had received ICI therapy at Duke Cancer Institute were identified. FFPE archival tumor samples for pts with mRCC were evaluated with Dako 28–8 and Ventana SP142 PD-L1 immunohistochemistry (IHC) assays. For pts with mUC, FFPE archival tumor samples were evaluated with Dako 22C3 and Ventana SP263 PD-L1 IHC assays. Scoring was validated by two pathologists using the scoring system for each assay. PD-L1 status was subsequently correlated to best RECIST response (objective response rate (ORR) defined as stable disease or better). Results: Tissue was obtained from primary tissue in 72% of mRCC cases and in 61% of mUC cases, with remainder from metastatic biopsies. The majority of mRCC cases (29/32, 91%) were concordant between Dako 28-8 and Ventana SP142 assays (8 cases pos and 21 cases neg), with 3 discordant cases (1 case pos for Dako 28-8 but neg for Ventana SP142 and 2 cases neg for Dako 28-8 but pos for Ventana SP142), all from primary tissue. The majority of mUC cases (17/18, 94%) were also concordant between Dako 22C3 and Ventana SP263 assays (2 pos cases and 15 neg cases), with 1 indeterminate Dako 22C3 test on a metastatic biopsy due to background lymph node. In mRCC, the ORR for PD-L1 pos cases was 45% (5/11) versus 33% (8/24) for PD-L1 neg cases. In mUC, the ORR for PD-L1 positive cases was 50% (1/2) versus 31% (5/16) for PD-L1 neg cases. Conclusions: There was strong concordance between the clinically meaningful PD-L1 assays chosen for comparison in both mRCC and mUC. mUC results were limited by low PD-L1 expression in this cohort. Although PD-L1 status does not fully predict for response to ICIs, this suggests that PD-L1 testing could be used interchangeably for the majority of cases when selecting ICI treatment in mRCC and mUC.

62 Background: Cell proliferation is an important marker of survival in many tumors, and we hypothesized that this attribute could be related to response to immune checkpoint inhibitors (ICIs) in RCC. Previously we reported (SITC 2018) that moderately proliferative lung cancer have a much higher response rate than either poorly or highly proliferative tumors. Methods: 69 FFPE RCC tumor samples were evaluated by RNA-seq to measure transcript levels of 394 immune related genes. Cell proliferation was defined as the mean mRNA expression of 10 genes (BUB1, CCNB2, CDK1, CDKN3, FOXM1, KIAA0101, MAD2L1, MELK, MKI67, TOP2A) which was evaluated for association with ORR to ICIs by RECIST v1.1 criteria for both PD-L1 IHC positive and negative cases. Cell proliferation for each case was split into 3 tertiles of poorly ( < 33), moderately (33-66) and highly ( > 66) proliferative versus a reference population. Poorly and highly proliferative were grouped for comparison to moderately proliferative tumors. PD-L1 IHC was performed using DAKO 22C3 antibody and scored by FDA guidelines and considered positive if TPS ≥1% and negative if < 1%. Results: In our cohort of 69 patients, the overall ORR was 18.8. The majority, 91% of tumors were PD-L1 negative, with ORR of 14.8%, as opposed to ORR of 50% in PD-L1 positive cases. 62.2% of tumors were poorly proliferative, 8.7% were highly proliferative, and 29% were moderately proliferative. ORR in moderately proliferative tumors was 30% and 14.2% in poorly/highly proliferative tumors. In PD-L1 negative tumors the ORR in moderately proliferative tumors was 20% and 13% in poorly/highly proliferative tumors. For the 5 moderately proliferative, PD-L1 positive tumors, there were 2 PR and 1 CR, and in the 3 poorly/highly proliferative tumors there was 1 PR. Conclusions: Cell proliferation may play a crucial role in distinguishing RCC patients who may have a clinical benefit to ICI, including the important subgroup of PD-L1 negative tumors. Moderately proliferative tumors have a higher ORR than their poorly/highly counterparts.

61 Background: Cell proliferation is an important marker of survival in many tumors and we hypothesized that this attribute could be related to response to immune checkpoint inhibitors in RCC. Previously we reported (SITC 2018) moderately proliferative lung cancer has a much higher response rate than either poorly or highly proliferative tumors. Methods: 69 FFPE tumor samples of RCC were evaluated by RNA-seq to measure transcript levels of 394 immune related genes, including 10 related to cell proliferation (BUB1, CCNB2, CDK1, CDKN3, FOXM1, KIAA0101, MAD2L1, MELK, MKI67, TOP2A). Cell proliferation, defined as the mean mRNA expression of these 10 genes was evaluated for association with ORR to ICIs by RECIST v1.1 criteria for both PD-L1 IHC positive and negative cases. Cell proliferation for each case was split into 3 tertiles of poorly ( < 33), moderately (33-66) and highly ( > 66) proliferative compared to a reference population. Poorly and highly proliferative were grouped for comparison to moderately proliferative tumors. Tumors were inflamed or non-inflamed based upon RNA‐seq analysis of CD8 compared to a reference population of more than 500 cases of multiple tumors. Non-inflamed, or immune desert tumors, defined as the lower 25th percentile of rank for CD8+ T-cells, and greater than 75th percentile of rank as inflamed. Results: In our cohort of 69 patient the overall ORR was 18.8%. 15.9% of tumors were non-inflamed with an ORR of 9.1%. For 36.2% inflamed tumor the ORR was 32%. For cell proliferation 62.2% were poorly proliferative, 8.7% were highly proliferative, and 29% were moderately. ORR in moderately proliferative tumors was 30% versus 14.2% in poorly/highly proliferative tumors. In inflamed tumors, ORR in moderately proliferative tumors was 37.5% as opposed to 17.6% in poorly/highly proliferative tumors. In 11 non-inflamed tumors, there was only one responder, which was a poorly/highly proliferative tumor. Conclusions: Cell proliferation may play a crucial role in distinguishing RCC patients who may have a clinical benefit to ICI, including the important subgroup of inflamed tumors. Moderately proliferative tumors have a higher ORR than their poorly/highly counterparts.

2595 Background: ICIs are now standard of care for mRCC; however, there are few biomarkers to predict ICI response. Recent data from atezolizumab/bevacizumab trials in mRCC suggest tumors with high Teffhigh/PD-L1+ are more likely to respond to ICI. Here, we use two gene panels as well as other inflammation markers in the tumor microenvironment to correlate with ICI responses. Methods: This multicenter study evaluated 86 patients (pts) with mRCC treated with ICIs. FFPE tumor samples were evaluated by RNA sequencing for Teff status. Two gene panels were analyzed: a Teff Gene Panel (CD8, CD27, IFNG, GZMA, GZMB, PRF1, EOMES, CXCL9, CXCL10, CXCL11, CD274, CTLA4, FOXP3, TIGIT, IDO1, PSMB9, TAP1) and a 5-Gene panel (FOXP3, CCR4, KLRK1, ITK, and TIGIT) based on the gene expression pattern of tumors in our cohort. Objective response rates (ORRs, defined as CRs and PRs) were correlated with PD-L1 status (positivity was defined as ≥1% TPS based on Dako 22C3 IHC assay), and TMB (0-10, 10-20, ≥20 mut/Mb), and tumor inflammation (high CD8 expression compared to a large reference population). Best responses to ICI was determined by an expert radiologist using RECIST 1.1 criteria. Inflamed tumor status, Teff gene panel, 5-gene panel, PD-L1 status, and TMB were associated with ORR and tested using a chi-squared test with Yates’s continuity correction. Results: ORR was 50% (4/8) for PD-L1 positive pts and 14% (9/65) for PD-L1 negative pts (p = 0.042). The majority of tumors (95%, 82/86) had TMB < 10 mut/mb. 43 pts (50%) were classified as Teffhigh and 43 pts were classified as Tefflow. ORR was 23% (10/43) in the Teffhigh cohort and 12% (5/43) in the Tefflow cohort (p = 0.256). ORR was 31% (14/45) in the 5-Gene high cohort and 2% (1/41) in the 5-Gene low cohort (p = 0.001). Conclusions: TMB and tumor inflammation based on CD8 did not reliably predict for objective responses in this study of mRCC pts treated with ICIs. Gene expression signatures provide a more comprehensive evaluation of the tumor microenvironment and may lead to better predictive biomarkers for ICI response than individual biomarkers such as PD-L1, TMB, or CD8 expression.

400 Background: ICIs have expanded therapeutic options for mUC patients (pts); however, biomarkers such as PD-L1 have not served as reliable predictors of treatment efficacy. High tumor mutation burden (TMB) has been previously described as a potential biomarker for predicting ICI response in several indications, but its utility is still being explored in mUC. Here, we compare the genomic landscapes and clinical outcomes of mUC pts following ICI therapy using an investigational solid tissue-based next-generation sequencing assay to assess TMB and identify genetic correlates of ICI response. Methods: 20 pts with mUC treated with ICIs at Duke Cancer Institute were identified. Tumor samples were retrospectively evaluated with a Personal Genome Diagnostics Assay for somatic variants across > 500 genes, as well as TMB and microsatellite status. Tumor samples were also stained for PD-L1 status using the Dako 22C3 IHC assay. Deidentified clinical information was extracted from the medical record and tumor response was evaluated using RECIST 1.1 criteria. Results: Pts were grouped by overall response following ICI therapy as either responders (“R”, n = 6) or non-responders (“NR”, n = 13). Pts exhibited a wide range of TMB scores (0.7 to 30.4 mutations/Mb), with a mean TMB score of 9.60 vs. 3.87 mut/Mb in R vs NR groups, respectively; however, this difference was not statistically significant ( p = 0.284). 18 pts were evaluated for PD-L1 status, with only 2 positive samples (one in each group). Rs had significantly more mutations in histone methylation genes (KDM6A, KMT2C, and KMT2D), (67% vs 15% in NRs, p = 0.0039). FGFR3 mutations were also more frequent in R vs NR (67% vs 5%, p = 0.0339). Finally, there was a higher frequency of mutations in TP53 and BRCA1 in the NRs. Conclusions: In this mUC cohort, neither TMB nor PD-L1 correlated with response to ICI therapy. Histone modifying genes and FGFR3 mutations were more frequent in responders, whereas BRCA1 and TP53 mutations were enriched in non-responders, warranting future prospective studies to understand underlying mechanisms of ICI response and resistance in mUC.

e22149 Background: Although the use of genomic sequencing to guide cancer patients into clinical trials is being developed, much needs to be done before this becomes the standard of care. First, sm...

11036 Background: Src inhibition may augment sensitivity to chemotherapy, but in unselected patients (pts) with advanced solid tumors, src inhibitors have shown limited clinical activity. Biomarkers to predict benefit from src inhibitors in advanced solid tumors are not yet known. Methods: 22 pts (dose escalation cohort= 12 pts; colorectal cancer [CRC] expansion cohort= 10 pts) were enrolled in a phase I study to determine the safety and tolerability of the src inhibitor dasatinib with capecitabine, oxaliplatin, and bevacizumab (J Clin Oncol 29: 2011 [suppl; abstr 3586]). Src activation (src-a) was assessed in tumors from 16 evaluable pts. Src-a was measured by immunohistochemistry (IHC) in formalin-fixed, paraffin-embedded tumor samples using an antibody that selectively recognizes the active conformation of src (clone 28). A GI pathologist who was blinded to pt outcomes graded membranous src-a using a standard semi-quantitative method. The endpoint of this exploratory analysis was objective response rate ([ORR]= PR+CR). 2-sided Fisher’s Exact test was used to evaluate the association between ORR and src-a. Results: Across all tumor types, 8 tumors had no/faint src-a (IHC=0/1); 8 tumors had moderate/strong src-a (IHC≥2). Benign colonic epithelium had no src-a (IHC=0). The ORR was 75% (6/8) for pts with moderate/strong src-a versus (vs) 0% (0/8) for pts with no/faint src-a (p =0.007). In the CRC expansion cohort, the ORR was 83% (5/6) for patients with moderate/strong src-a vs 0% (0/2) for pts with no/faint src-a (p=0.107); progression free survival range was 7.9-24.4 months for pts with moderate/strong src-a. Conclusions: In this small phase I study, src-a is associated with benefit from the combination of dasatinib and oxaliplatin-based chemotherapy. Further evaluation of dasatinib in patients whose tumors demonstrate high levels of src-a may be warranted. Clinical trial information: NCT00920868.

68 Background: EC is commonly managed with concurrent chemoradiotherapy, with or without surgical resection. The optimal combination and dose of agents is the subject of continued investigation. This study examines chemotherapeutic agents with known efficacy in EC in combination with the EGFR inhibitor panitumumab. Methods: Eligible pts received RT (1.8 Gy qd to 50.4 Gy) combined with concurrent chemotherapy. Dose-level (DL) 1 was cape (625 mg/m2/bid RT days), ox (40 mg/m2 weekly X 6 weeks), and pmab (3.6 mg/kg, weeks 1, 3 and 5). Chemotherapy doses were escalated barring dose limiting toxicity (DLT). The primary endpoint was defining the maximally tolerated dose with this combination. Secondary endpoints included toxicity and radiographic/pathologic response rates. Results: Twenty-nine pts were enrolled. Twenty-five had adenocarcinoma, 24 (83%) were cN+ and 9 (31%) had M1a/b disease. DLT was not encountered in DL 1. Two of 6 patients at DL 2 (cape 825 mg/m2/bid RT days, ox 50 mg/m2 weekly, pmab 4.8 mg/kg, weeks 1, 3 and 5) developed DLT (one hospitalization due to dehydration; one with drug reaction requiring hospitalization). Twenty additional pts were enrolled at DL1. Primary toxicities were EGFR-rash, esophagitis, nausea/vomiting and fatigue. On repeat endoscopy, 16 (55%) had CR, 10 (35%) PR and 2 (7%) SD. Using PERCIST criteria, 12 (41%), 11 (38%), 2 (7%) and 3 (10%) had CR, PR, SD and PD response on restaging PET, respectively. Twenty pts underwent esophagectomy, revealing Gr 0 response (no residual disease) in 9 (45%), Gr 1 (single/microscopic cells) in 3 (15%), Gr 2 (fibrosis > gross disease) in 4 (20%) and Gr 3 (gross residual > fibrosis or no evident response) in 4 (20%). Seven pts (35%) experienced anastomotic leak (2 requiring reoperation and 3 stent placement). Conclusions: Concurrent chemoradiotherapy utilizing capecitabine, oxaliplatin, panitumumab is reasonably well-tolerated and associated with high rates of radiographic, endoscopic and pathologic response. Postoperative anastomotic leak rates were higher than expected. Further study of this regimen in the operative and nonoperative settings is warranted.