Background This study assessed the contrasting genomic profiles from the primary tumors (PTs), metastatic (MET) sites, and circulating tumor DNA (ctDNA) of patients with prostate cancer (PC). Methods A total of 1294 PC tissue specimens and 2462 ctDNA specimens underwent hybrid capture–based comprehensive genomic profiling (CGP). Specimens included tissue from PTs; MET biopsies from bone, liver (LIV), lung (LU), brain (BN), lymph node, and soft tissue sites; and ctDNA. Results Differences in alteration frequencies between PT, MET, and ctDNA specimens for selected genes were observed. TMPRSS2:ERG fusion frequencies were similar between PTs and MET sites (35% vs 33%) but varied among MET sites. Genomic alterations (GAs) in AR were lowest in PTs (2%) and highest in MET sites (from 24% in LU to 50% in LIV). BN had the highest genomic alterations/tumor (8) and enrichment for PTEN GAs. The BRCA2 GA frequency varied from 0% in BN to 15% in LIV. ERBB2 amplification was increased in MET sites in comparison with PTs. RB1 GAs were increased in LIV. Biomarkers potentially associated with an anti‐PD(L)1 response included CDK12 GAs (16% in LU) and a microsatellite instability–high status (29% in BN). Analyses of ctDNA featured a broad spectrum of GAs similar to those detected across MET sites. Conclusions CGP of PTs, MET sites, and ctDNA in PC exhibited differences most likely associated with tumor progression, clonal evolution, and exposure to systemic therapies; ctDNA can also capture a broad range of potential therapeutic opportunities for patients with PC.
Abstract ARIEL2 (NCT01891344) is a single-arm, open-label phase 2 study of the PARP inhibitor (PARPi) rucaparib in relapsed high-grade ovarian carcinoma. In this post hoc exploratory biomarker analysis of pre- and post-platinum ARIEL2 samples, RAD51C and RAD51D mutations and high-level BRCA1 promoter methylation predict response to rucaparib, similar to BRCA1 / BRCA2 mutations. BRCA1 methylation loss may be a major cross-resistance mechanism to platinum and PARPi. Genomic scars associated with homologous recombination deficiency are irreversible, persisting even as platinum resistance develops, and therefore are predictive of rucaparib response only in platinum-sensitive disease. The RAS, AKT, and cell cycle pathways may be additional modulators of PARPi sensitivity.
Abstract Introduction: De-regulation of the class I phosphoinositide 3’-kinase (PI3K) pathway has long been known to contribute to the development and progression of many tumors. However, the FDA only recently approved the first selective inhibitor of PIK3CA, the p110-alpha catalytic subunit of PI3K, specifically for use in treatment of a subset of PIK3CA-mutant breast carcinomas. Given the emergence of this therapeutic class of agents, we sought to identify other subsets of breast tumors that may be driven largely by PIK3CA mutations and which therefore could be candidates for these therapies. In breast fibroepithelial neoplasms, mutations in PIK3CA have been occasionally reported in borderline and malignant phyllodes tumors. In contrast, mutations in MED12 are common and recurrent across the entire spectrum of benign and malignant breast fibroepithelial tumors, and are enriched in borderline/malignant phyllodes cases that still have benign fibroadenoma-like areas. In the current study, we sought to define the histologic and molecular features of PIK3CA-mutant phyllodes tumors. Methods: From 2014 to 2019, we analyzed clinical tumor samples using comprehensive genomic profiling by a hybrid capture-based DNA sequencing platform. We searched our case archive to find breast phyllodes tumors with known or likely pathogenic alterations in PIK3CA and other known tumor-related genes. All cases were clinically advanced. We reviewed pathology reports, histopathology, and patient clinical data. Results: We identified 12 (16%) of 76 breast phyllodes tumors in our case archive as PIK3CA-mutant. Median patient age for PIK3CA-mutant phyllodes tumors was 56 years. Cases consisted of 6 primary tumors, 2 local recurrences, and 4 lung metastases. Primary tumor size measured from 38 to 220 mm (median 100 mm; mean 114 mm). Digital slides were available for histology review in 9 cases. Cases showed uniformly malignant histology, with no benign or fibroadenoma-like regions. 3 cases showed malignant heterologous elements. Compared to the rest of our breast phyllodes tumor cohort, PIK3CA-mutant cases showed significantly fewer pathogenic genomic alterations in MED12 (8% vs. 55%, p=0.0037) and TP53 (17% vs. 56%, p=0.0245), as well as a trend to fewer mutations in RB1 (0% vs. 22%, p=0.11). The other most frequently mutated genes in the PIK3CA-mutant group were TERTp (70%), CDKN2A (67%), and NF1 (50%). Conclusions: PIK3CA-mutant phyllodes tumors define a unique subset of tumors characterized by aggressive histologic features and largely lacking the MED12 mutations seen in many fibroepithelial neoplasms. These findings provide compelling rationale for comprehensive genomic profiling of advanced cases of this disease in an effort to inform therapeutic options including clinical trials of PI3K-targeted agents in this setting. Citation Format: Erik A Williams, Ethan S Sokol, Dean C Pavlick, Nikunj Shah, Julia A Elvin, Jo-Anne Vergilio, Jonathan K Killian, Nhu Ngo, Douglas Lin, Vincent A Miller, Jeffrey S Ross. PIK3CA-mutant breast phyllodes tumors show a uniformly aggressive histology and significant mutual exclusivity with MED12 mutation [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr P4-06-07.
5593 Background: The tumor suppressor gene ARID1A is an emerging target for new cancer treatment strategies including ATR inhibition. This study aimed to describe the landscape of ARID1A alterations ( ARID1Amut) in gynecologic malignancies. Methods: Patients (pts) with a diagnosis of ovarian (OC) or uterine cancer (UC) of different histologies and comprehensive genomic profiling (CGP) by Foundation Medicine Inc. were included in this study. CGP of solid tissue biopsies (Tbx; FoundationOneCDx) included evaluation of genomic loss of heterozygosity (gLOH; gLOH-high as ≥16% as validated in OC), microsatellite instability (MSI), and tumor mutational burden (TMB; high as ≥10 mutations/Megabase). CGP of peripheral whole-blood liquid biopsies (Lbx; FoundationOneLiquidCDx) included evaluation of MSI and tumor fraction (TF), a measure of the relative quantity of circulating tumor DNA (ctDNA). TF was calculated by measures of aneuploidy and allele frequency and binned as TF < 1%, TF 1 to < 10%, or TF ≥10%. Results: Tbx Cohort: 5,778/30,212 (19.1%) cases were ARID1Amut. Pts had a median age of 63 (range 21-89) years. ARID1Amut were observed across many subtypes and most frequently in endometrial endometrioid (n = 3,052, 57.7%) and ovarian clear cell (n = 982, 57.6%) but rarely in serous (OC, n = 12,258, 2.8%; UC, n = 2,682, 8.9%). Pts frequently harbored more than one ARID1Amut (2,360/5,778, 40.8%). Of the 8,484 observed ARID1Amut, 97.6% were short variants (SV), 0.5% were deletions, and 1.9% were inactivating rearrangements. SV were primarily frameshifts (68.5%) and nonsense mutations (27.6%). The most frequent alterations observed were frameshifts at the D1860 codon. SV were predicted to be homozygous in 11.9%, heterozygous in 65.3%, or unknown zygosity in 22.8%. Overall, 16.6% of ARID1Amut cases with SV had at least one homozygous alteration. 6.6% of pts with homozygous ARID1Amut were MSI-high (MSI-H), while 39.4% of pts with only heterozygous or unknown zygosity ARID1Amut were MSI-H (p < 0.0001). Overall, 1,905 (33.0%) of ARID1Amut cases were MSI-H, and 2,183 (37.8%) were TMB high. For ARID1Amut cases with evaluable gLOH (n = 4745), the median gLOH was 2.7%, and 5.9% pts were gLOH-high. The most frequently altered co-occurring genes were PTEN (62.2%), PIK3CA (54.2%), and TP53 (27.6%). 8.7% of ARID1Amut also harbored a BRCA alteration. Lbx Cohort: 59/967 (6.1%) cases were ARID1Amut. 17 (28.8%) were MSI-H. Frequency of ARID1Amut increased as TF increased, with a detected frequency of 1.3%, 6.7%, and 14.0% among Lbx with TF < 1%, TF 1 to < 10%, or TF ≥10% respectively. Conclusions: ARID1Amut are observed across a variety of Gynecologial cancer subtypes but are enriched in clear cell and endometrioid histologies and detected in both tissue and liquid biopsies. ARID1Amut were not associated with elevated gLOH but were often MSI-H and TMB ≥10mut/Mb. Clinical trials targeting ARID1A may wish to consider the context of co-occuring biomarkers.
726 Background: NF2 genomic alterations (GA) have been associated with aggressive behavior in RCC. Methods: FFPE tissues from 1,386 clear cell (ccRCC), 307 papillary (pRCC), 72 chromophobe (chRCC), 145 sarcomatoid (sRCC), 54 collecting duct (cdRCC),37 medullary (medRCC) and 134 unclassified (nosRCC) underwent hybrid-capture based CGP to evaluate all classes of genomic alterations (GA). Tumor mutational burden (TMB) was determined on up to 1.1 Mbp of sequenced DNA and MSI was determined on 114 loci. PD-L1 expression was determined by IHC (Dako 22C3). Results: 140 (7%) RCC featured NF2 GA which were predominantly short variant (SV) mutations. Gender and age were similar with male preponderance in all histologic subtypes. NF2 GA frequency was highest in cdRCC (20%) and sRCC (19%) and lowest in ccRCC (3%). The medRCC at 5% NF2 GA and chRCC at 0% NF2 GA were not further evaluated. VHL and PBRM1 GA were significantly more frequent in NF2 altered ccRCC than all other RCC (P < 0.001). Other mTOR pathway GA were uncommon. Potentially targetable kinase GA in NF2-mutated RCC included BRAF (2% of ccRCC), EGFR (3% of pRCC), ERBB3 (4% of sRCC) and PIK3CA (9% of cdRCC). No NF2 mutated RCC featured MSI -high status and both TMB and PD-L1 expression levels were extremely low in all subsets with exception of high PD-L1 staining in sRCC tumors. Conclusions: cdRCC, sRCC, pRCC and nosRCC are enriched in NF2 GA. Low PBRM1 GA, TMB and MSI- high predict resistance to immunotherapy in NF2 mutated RCC although the high PD-L1 expression in sRCC is noteworthy.[Table: see text]
A chemotherapy response score (CRS) system was recently described to assess the histopathologic response and prognosis of patients with tubo-ovarian high-grade serous carcinoma (HGSC) receiving neoadjuvant chemotherapy. The current study was performed as an independent assessment of this CRS system. We retrospectively identified advanced stage HGSC patients who received neoadjuvant chemotherapy and underwent interval debulking. If available, a hemotoxylin and eosin slide from the omentum and the adnexa was selected for the study. Slides were independently scored by 13 pathologists using the 3-tiered CRS system. Reviewers then received web-based training and rescored the slides. Overall survival and progression-free survival were estimated using the Kaplan-Meier method and compared using the log-rank test. A total of 68 patients with omental (n=65) and/or adnexal (n=59) slides were included in the study. Interobserver reproducibility was moderate for omentum (κ, 0.48) and poor for adnexa (κ, 0.40), which improved for omentum (κ, 0.62) but not for adnexa (κ, 0.38) after online training. For omental slides, a consensus CRS of 1/2 was associated with a shorter median progression-free survival (10.9 mo; 95% confidence interval, 9-14) than a CRS of 3 (18.9 mo; 95% CI, 18-24; P=0.020). In summary, a 3-tiered CRS system of hemotoxylin and eosin-stained omental deposits can yield prognostic information for HGSC patients receiving neoadjuvant chemotherapy, and web-based training improved reproducibility but did not alter determination of clinical outcomes. The CRS system may allow oncologists to identify potential nonresponders and triage HGSC patients for heightened observation and/or clinical trials.
451 Background: While renal AML is generally a benign tumor, malignant eAML is a rare form of renal malignancy of uncertain histogenesis featuring variable prognosis. Consequently, most series are limited in small sample sizes and lack comprehensive genomic analysis. We sought to survey the genomic landscape in malignant renal eAML and identify potential therapeutic targets. Methods: After reviewing our database containing over 400,000 advanced cancer samples we identified 34 cases of malignant eAML. Comprehensive genomic profiling (CGP) was performed using a hybrid capture technique to assess all classes of genomic alterations (GA). MSI status, gLOH, genomic ancestry and gene signatures were determined by CGP. TMB was measured in mutations/megabase of sequenced DNA. PD-L1 positivity was determined by IHC using the DAKO 22C3 tumor cell proportional score (0% = negative; 1-49% = low positive). Results: The primary malignant eAML was used for CGP in 14/34 (41.2%) and a metastatic site biopsy was used in 20 (58.8%) cases (10 liver, 3 lung, 4 retroperitoneum, 2 peritoneum and 1 psoas muscle). There were 19 (55.9%) female patients and a mean age of 60 years (median 53 years). In a subset 17 of cases when either MART1, MelanA or HMB45 IHC staining was performed all cases were positive for at least 1 marker. There were no MSI-high cases. The mean and median TMB was 1.6 mutations/Mb. Of the 4 malignant eAML cases tested, 3 (75%) were negative and 1 (25%) case was low PD-L1 expression positive. 25 (73.5%) of the cases featured short variant mutations in the TSC2 gene while the other 9 (26.5%) were TSC2 mutation negative. 4 (11.8%) of the malignant eAML featured germline GA in the MUTYH, CD36, FLCN, and FANCC genes of uncertain roles in the development of the disease. There were no TSC2 germline GA. Other GA were mostly not currently targetable and included the tumor suppressors TP53 at 29.4%, CDKN2A/B at 14.7%, ATRX at 11.8%, and RB1 at 8.8%. Aside from the frequent TSC2 alterations, other GA potentially implicating MTOR inhibitor use include GA in PTEN at 5.9% and NF2 at 2.9%. Additionally, BRCA1 was altered in 2.9% of cases, suggesting the possible utility of a PARP inhibitor. In a 28 case subset, the mean gLOH was 6.3% (range 0% to 37.9%) with 3 cases (10.7%) featuring a gLOH of > 16%. 26 (76.5%) of these patients were of EUR ancestry, 5 (14.7%) of AMR ancestry and 3 (8.8%) of AFR ancestry. No specific genomic signature characterized the malignant eAML cases. Conclusions: Renal malignant eAML, also known as malignant PEComa of the kidney, is an exceedingly rare malignant tumor. Our CGP identified that the majority of cases exhibit non-germline TSC2 mutations. Interestingly, other germline alterations were found in 4/34 cases which are of unknown significance. While there may be limited opportunities for targeted or immunotherapies aside from MTOR inhibition, CGP analysis may still provide guidance into identification of potential therapeutic targets.
11541 Background: GIST is the most common mesenchymal cancer of the digestive tract. Beyond surgery, treatment for GIST focuses largely on tyrosine kinase inhibitors (TKI), whose selection and potential resistance depend on select mutations. We present the molecular landscape of GIST utilizing tissue and liquid biopsies with emphasis on the clinical utility of liquid biopsy in advanced GIST. Methods: Liquid (FoundationOne Liquid CDx [F1LCDx]) and tissue (FoundationOne CDx) CGP was performed by hybrid capture, targeted NGS at Foundation Medicine Inc. Tissue and liquid samples from 2,198 and 147 patients, respectively, were analyzed. A cohort of 27 paired tissue and liquid samples were also evaluated. The levels of circulating tumor DNA (ctDNA) in liquid biopsies was quantified by tumor fraction (TF), with a TF algorithm incorporating aneuploidy, variant allele frequency, and canonical alterations detected on F1LCDx. Results: Tissue CGP (n = 2,198) revealed the following prevalence of primary driver alterations: KIT (77%), PDGFRA (8%), NF1 (6%), SDHA/B/C/D (SDHx, 3%) and BRAF (1%). Rates of molecular markers previously associated with worse prognosis included: CDKN2A (29%), RB1 (9%), TP53 (6%) and SETD2 (4%). 7% of cases had no reportable known pathogenic alterations in canonical GIST genes (wild-type GIST), while 2% of cases had a mutation in more than one driver. In a cohort of 147 liquid biopsies, TF was < 1% in 68.0%, 1-10% in 18.4%, > 10% in 13.6% of samples. In samples with elevated TF ( > 10%), the prevalence of targetable driver alterations in KIT (89%), PDGFRA (4%) , NF1 (4%) and BRAF (4%) was comparable to the tissue prevalence. In liquid, 58% (39/67) of samples with a KIT-driver mutation had a co-occurring imatinib-resistant KIT alteration. In addition, 4/147 patients (3%) were predicted to harbor a germline KIT mutation, including one patient (0.6%) with a potential imatinib-resistant KIT D820G germline mutation and another with clinical suspicion of germline KIT L576P mutation due to the presence of multiple primary GISTs, hyperplasia of myenteric plexus and dysplastic skin nevi. In paired tissue/liquid samples, liquid detected 2/2 driver mutations found in tissue when liquid TF was > 10%, and 5/6 in specimens with TF > 1%. In the overall cohort, the relative prevalence of KIT exon 9 and 11 driver alterations was comparable in tissue vs liquid, while imatinib-resistance KIT exon 13 and 17 mutations were enriched in liquid samples. Conclusions: Known driver and TKI-resistant mutations of both somatic and potential germline origin are identified in peripheral blood ctDNA of GIST patients. Liquid biopsy shows high concordance to tissue in identifying driver mutations in the presence of elevated TF and may exhibit TKI-resistant specific alterations. This study indicates that liquid biopsy may be useful in the molecular classification of GIST during the medical management of advanced GIST patients.
