Clinical laboratory mutation analysis performed on aggressive B cell non-Hodgkin lymphoma patient biopsies

e19561

Background: Results from comprehensive genomic analysis performed on aggressive B cell non-Hodgkin lymphomas (aBNHL) patient (pt) biopsies (bx) have identified mutations (mut) which may be predictive of pt survival. However, results from mut analysis (MA) performed on aBNHL pt bx in clinical laboratories are not well-described. Methods: Bx diagnostic of non-Burkitt aBNHL obtained from 7/2015-2/2021 with MA performed in the Penn Center for Personalized Diagnostics clinical laboratory using 3 consecutive versions of lymphoma sequencing panels were retrospectively analyzed. Mut tested varied by panel. Mut detected with variant allele frequency (VAF) ≥10% were included. Results: MA was performed on 293 aBNHL pt bx with a successful assay for 270 (7.8% failure rate). Request for MA was made by interpreting hematopathologist 93% and treating clinician 7%. Median turnaround time (TAT) from receipt of bx to result report was 18 (range 8-53) days. Notable pt characteristics included prior/concurrent indolent lymphoma (IL) 31% and prior systemic therapy received for IL and/or aBNHL 38%. Notable bx characteristics included bx type paraffin-embedded tissue 85% and bone marrow 12%, histologic classification diffuse large B cell lymphoma 86% and high grade B cell lymphoma 13%, cell of origin (COO) by Hans algorithm non-germinal center B (non-GCB) 51% and GCB 44%, known double expressor lymphoma 30%, known MYC rearrangement 18%, known double hit lymphoma 9% and any mut detected with VAF ≥10% 82%. Number and percentage of mut occurring in ≥10% of bx (minimum tested bx for mut ≥92) with associated COO are listed in the table. Subset analysis of 147 bx obtained from pts with newly-diagnosed aBNHL (15% with previously-untreated IL) had similar bx characteristics to the entire population were analyzed for the presence of mut reported in poor-prognosis COO-associated classifications: Cluster 5 for non-GCB and Cluster 3 for GCB (PMID: 29713087), double-hit signature (DHITsig) for GCB (PMID: 30523716) and molecular high grade (MHG) for GCB (PMID: 30523719). For 77 non-GCB bx, 19% had Cluster 5 mut. For 59 GCB bx, 56% had at least one of Cluster 3, DHITsig or MHG mut. Conclusions: Clinical laboratory MA performed on nearly 300 aBNHL pt bx was highly successful with a short TAT and demonstrated recurrent gene mut associated with aBNHL development. In the subset of bx from previously-untreated pts, MA identified mut associated with poor-prognosis genetic subtypes of aBNHL.[Table: see text]

Abstract 783: Impact of KRAS and TP53 co-mutations on outcomes following 1st-line therapy among patients with LKB1/STK11 mutated stage IV NSCLC

* Authors E. M. B. and M. E. M. contributed equally

Introduction: LKB1/STK11 is a tumor suppressor and a negative regulator of mTOR signaling. NSCLC patients with STK11 mutations (MT) commonly also have other co-mutations. We evaluated the impact of STK11 MT on outcomes following chemotherapy for metastatic NSCLC, and the role of common co-existing MTs in KRAS and TP53.

Methods: We conducted a retrospective review of patients (pts) with NSCLC and STK11 MT treated at the University of Pennsylvania. STK11 MT was identified through next generation sequencing (NGS) in tissue or using the Guardant 360TM platform for plasma. Four treatment groups were analyzed: STK11 alone (A), and 3 co-MT groups: STK11/KRAS (B), STK11/TP53 (C), and STK11/KRAS/TP53 (D). Chi-square analysis was used to assess differences in baseline characteristics between the MT groups. Cox proportional hazard models (HR) were used to determine the relationship of STK11 co-MT to survival. Kaplan-Meier analysis was used to estimate overall survival (OS) and progression-free survival (PFS).

Results: Between 2/14/13 and 12/1/16, 77 pts with STK11 MT who received systemic therapy were identified (56 tissue, 21 plasma): median age at diagnosis 66 yrs, 51.5% male, 85% ECOG PS of 0 or 1, 94% of pts received 1st-line chemotherapy. There were no significant differences in baseline characteristics among mutation groups. Pts with STK11/KRAS had a worse median PFS and OS vs. pts with STK11 alone (Table 1). Pts with STK11/TP53 had a significantly better OS compared to STK11/KRAS patients.

Conclusion: Among STK11 mutant NSCLC pts treated with chemotherapy, co-MT with KRAS was associated with a significantly worse PFS and OS compared to patients with STK11 alone. By contrast, co-MT with TP53 conferred a better prognosis. These results warrant further validation in a larger study.

Table 1: Outcomes by mutation statusA:STK11alone n=16 (24.2%)B:STK11/KRAS n=21 (31.8%)C:STK11/TP53 n=18 (27.3%)D:STK11/KRAS/TP53 n=11 (16.7%)Median PFS5.3 mo2.4 mo*9.9mo4.9 moMedian OS13.1 mo6.9 mo**22.2 mo***13.9 mo*compared to A, HR 2.8, 95% CI 1.1 to 6.9, p = 0.026**compared to A, HR 6.4, 95% CI 2.4 to 17.3, p<0.001***compared to B, HR 0.12, 95% CI 0.04 to 0.31, p<0.001

Citation Format: Erin M. Bange, Melina E. Marmarelis, Wei-Ting Hwang, Yu-Xiao Yang, Jeffrey C. Thompson, Joshua Bauml, Christine A. Ciunci, Evan W. Alley, Tracey L. Evans, Jennifer J. Morrissette, Roger B. Cohen, Corey J. Langer, Erica L. Carpenter, Charu Aggarwal. Impact of KRAS and TP53 co-mutations on outcomes following 1st-line therapy among patients with LKB1/STK11 mutated stage IV NSCLC [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 783.

Abstract 3691: Analysis of germline BRCA1/2 mutation associated breast and ovarian tumors reveals distinct pathways of immunosuppression

BRCA1 and BRCA2 are essential homologous recombination (HR) repair proteins; germline mutations confer elevated risk for breast and ovarian cancer. Defects in BRCA can lead to immune-related effects including depletion of peripheral T-cell pools and increased somatic mutational burden. We investigated the immunophenotypic properties of breast and ovarian cancers associated with germline BRCA1/2 mutations. We determined neoantigen load in 16 breast and 26 ovarian tumors with germline BRCA1/2 mutations and 50 non-BRCA tumors for each disease using whole exome sequencing data from the TCGA. Although neoantigen load (IC50 <500 nM) was variable across tumors, it did not differ between BRCA1/2 mutant and non-BRCA tumors. Leveraging RNAseq data from the TCGA, we investigated immune-related transcriptomic features within each disease. Cytolytic index, a measure of PRF1 and GZMA expression, was higher in non-BRCA tumors than BRCA1/2 mutant for breast (p<0.01) tumors and correlated more closely with neoantigen load than in non-BRCA tumors (Breast: BRCA r2=-0.15, non-BRCA r2=0.237). Class I HLA expression also correlated with cytolytic index (p<0.05) in non-BRCA tumors but not BRCA1/2 mutation associated tumors for both breast and ovarian cancers, suggesting more robust antigen-driven adaptive immunity in non-BRCA tumors. HLA expression was significantly lower (p<0.01) among BRCA mutant breast tumors, indicating immune escape by HLA down-regulation. We classified non-BRCA breast tumors by their level of HR deficiency (HRD score), as HRD by both BRCA and non-BRCA mediated mechanisms can lead to increased somatic mutational burden. Non-BRCA breast tumors irrespective of HRD score exhibited evidence of a more inflamed phenotype than BRCA1/2 mutant tumors, indicated by higher inferred immune cell infiltration, immune checkpoint expression, and chemokine expression (all p<0.01). To investigate potential mechanisms of immune escape in BRCA1/2 vs non-BRCA tumors, we compared canonical gene sets (MsigDb) across the two cohorts. Expression of extracellular matrix (ECM) proteins was higher in BRCA1/2 breast and ovarian tumors (q<0.02), suggesting that BRCA1/2 tumors may exclude immune cells via a dense ECM as observed for colorectal cancer and melanoma. Among Penn BRCA1/2 tumors (19 breast, 20 ovarian), neoantigen load did not differ from TCGA BRCA1/2 tumors. We directly measured immune cell infiltration via tumor immunohistochemistry in 12 breast and 14 ovarian Penn BRCA1/2 tumors. We found that CD3+ and CD8+ T cell infiltration was significantly lower in BRCA1/2 tumors with PTEN deletion (p<0.05), a known correlate of ECM remodeling in breast and ovarian cancer. Our work provides early evidence that BRCA1/2 breast and ovarian tumors mitigate intratumoral adaptive immunity by T-cell exclusion, so that despite higher mutational burden, they may not respond to checkpoint blockade.

Citation Format: Adam Kraya, Kara N. Maxwell, Brandon M. Wenz, Bradley Wubbenhorst, Nicole Lunceford, Amanda Barrett, Jennifer J. Morrissette, Michael D. Feldman, Susan M. Domchek, Robert H. Vonderheide, Katherine L. Nathanson. Analysis of germline BRCA1/2 mutation associated breast and ovarian tumors reveals distinct pathways of immunosuppression [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 3691. doi:10.1158/1538-7445.AM2017-3691

Use of targeted next generation sequencing (NGS) to assess mutational load in glioblastoma (GBM)

2027

Background: Trials of immune checkpoint inhibitors in GBM are ongoing. Evidence from other tumors suggests that high mutational load predicts response to these agents. We estimated mutational load in adults with newly diagnosed GBM using targeted NGS and assessed its association with other clinicopathologic parameters. Methods: NGS was performed for all patients diagnosed with isocitrate dehydrogenase wild-type (IDH-WT) GBM at our institution since September 1, 2016. Our panel performs targeted exome sequencing of 153 cancer-related genes (+/- 10 bp at intron/exon boundaries) and detects insertions, deletions, and single nucleotide variants at an allele frequency of 2%. Polymorphisms present in > 0.1% of the population according to the ExAC database and 1000 Genomes Project were filtered to reduce germline variants in the absence of matched normals. To avoid upward skewing of mutational load due to the panel’s preferential detection of genes recurrently mutated in cancer, alterations with known or likely functional disruption per the COSMIC database and/or our internal variant knowledge base were also not counted. The number of mutations counted was divided by the coding region target territory of the NGS panel (~0.5 Mb) to extrapolate mutational load to the whole exome. The Mann-Whitney U test was used to compare mutational load according to sex, age (≥65 vs. < 65), and MGMT methylation status (positive vs. negative). Results: Of 28 patients, median age was 65, 75% were male, and 37% were MGMT-methylated. Mean mutational load was 3.6 mutations/Mb (range 0-8, standard deviation 2.3) and was higher in males (4.3, 95% CI 3.3-5.3) compared to females (1.4, 95% CI 0.03-2.8) (p = 0.004). Mutational load did not differ according to age or MGMT methylation. Conclusions: In newly diagnosed IDH-WT GBM, targeted NGS revealed an estimated mutation rate similar to that reported in whole exome sequencing studies. In addition, we detected a higher mutational load in males compared to females. The latter finding adds to recent evidence suggesting sex-specific differences in the biology of GBM, and implies that future studies should account for sex when assessing mutational load as a predictive marker of immune checkpoint inhibition.

Differences in mutation rates between right- and left-sided colorectal adenocarcinoma

622

Background: Recent reports have demonstrated inferior outcomes for patients with right-sided colorectal cancer (CRC) compared to patients with left-sided disease (Schrag et al 2016, JCO 34 (suppl; abstr 3505)) as well as differences in treatment response based on disease sidedness (Venook et al 2016, JCO 34 (suppl; abstr 3504)). However, the biological and genetic underpinnings of these clinical differences are incompletely understood. Methods: We compared mutation rates among 38 genes in a retrospective review of next-generation sequencing data of CRC samples obtained in routine clinical practice at a single academic medical center. Primary location was identified via chart review. Right = cecum to transverse colon; Left = descending colon to rectum. Results: Among 293 samples (167 left-sided, 103 right, 23 synchronous or without clear primary), BRAF and CTNNB1 mutations were more prevalent in right-sided CRC. BRAF was mutated in 15.5% of right-sided CRC (95% CI: 8.5-22.5%) compared to 4.8% (CI: 1.6-8.0%) (p=0.003). CTNNB1 was mutated in 3.9% of right-sided CRC (CI: 0.2-7.6%) compared to no instances of CTNNB1 mutations in left-sided disease (p=0.01). Among right-sided CRC, there was a trend toward more KRAS mutations at 57.3% (CI: 47.7-66.8%) versus 44.9% (CI: 37.4-52.5%) and more PIK3CA mutations at 26.2% (CI:17.7-34.7%) versus 17.4% (CI: 11.6-23.1%), though these differences did not rise to the level of statistical significance. The overall rate of BRAF mutations in our sample (8.9%, CI: 5.5-12.3%) was consistent with the rate of BRAF mutations among large intestine adenocarcinomas in the COSMIC database (10.6%, CI: 10.4-10.8%), lending support to the external validity of these data. Conclusions: These differing mutation rates may implicate these genetic pathways in the mechanisms underlying the discrepant outcomes and treatment responses between right and left-sided disease described in prior studies. Further work is needed to more clearly elucidate genetic difference between right and left-sided CRC, as well as the mechanistic relationship between these mutations and prognosis.

Study of an ovarian sclerosing stromal tumor presenting as vaginal bleeding in a 7-month-old

This communication describes the histological, immunohistochemical, ultrastructural, and cytogenetic study of an ovarian sclerosing stromal tumor resected from a 7-month-old girl who presented with vaginal bleeding. The tumor is very rare, its pathogenesis is not clear, and its hormonal activity has been subject to debate. In addition, it has been rarely seen in children and never in infants, with the youngest patient reported being 10 years of age. Histological study of the tumor showed a process of multinodular asynchronous growth followed by gradual loss of cells, hyalinization, and eventual transformation into corpora albicantia-like structures, thus indicating that the process may be more akin to an ovarian nodular follicular hyperplasia than to a classical neoplasm. The study also documented an elevated proliferative MIB-1 index in the process, which had not been investigated in earlier reports, and illustrated the immunohistochemical reactivity of some of its stromal cells to progesterone receptors.