Homologous Recombination Abnormalities Associated With BRCA1/2 Mutations as Predicted by Machine Learning of Targeted Next-Generation Sequencing Data.

Background: Homologous recombination deficiency (HRD) is the hallmark of breast cancer gene 1/2 (BRCA1/2)-mutated tumors and the unique biomarker for predicting response to double-strand break (DSB)-inducing drugs. The demonstration of HRD in tumors with mutations in genes other than BRCA1/2 is considered the best biomarker of potential response to these DSB-inducer drugs. Objectives: We explored the potential of developing a practical approach to predict in any tumor the presence of HRD that is similar to that seen in tumors with BRCA1/2 mutations using next-generation sequencing (NGS) along with machine learning (ML). Design: We use copy number alteration (CNA) generated from routine-targeted NGS data along with a modified naïve Bayesian model for the prediction of the presence of HRD. Methods: The CNA from NGS of 434 targeted genes was analyzed using CNVkit software to calculate the log2 of CNA changes. The log2 values of various sequencing reads (bins) were used in ML to train the system on predicting tumors with BRCA1/2 mutations and tumors with abnormalities similar to those detected in BRCA1/2 mutations. Results: Using 31 breast or ovarian cancers with BRCA1/2 mutations and 84 tumors without mutations in any of 12 homologous recombination repair (HRR) genes, the ML demonstrated high sensitivity (90%, 95% confidence interval [CI] = 73%-97.5%) and specificity (98%, 95% CI = 90%-100%). Testing of 114 tumors with mutations in HRR genes other than BRCA1/2 showed 39% positivity for HRD similar to that seen in BRCA1/2. Testing 213 additional wild-type (WT) cancers showed HRD positivity similar to BRCA1/2 in 32% of cases. Correlation with proportional loss of heterozygosity (LOH) as determined using whole exome sequencing of 51 samples showed 90% (95% CI = 72%-97%) concordance. The approach was also validated in an independent set of 1312 consecutive tumor samples. Conclusions: These data demonstrate that CNA when combined with ML can reliably predict the presence of BRCA1/2 level HRD with high specificity. Using BRCA1/2 mutant cases as gold standard, this ML can be used to predict HRD in cancers with mutations in other HRR genes as well as in WT tumors.

Combining cell-free RNA with cell-free DNA in liquid biopsy for hematologic and solid tumors.

Current use of liquid biopsy is based on cell-free DNA (cfDNA) and the evaluation of mutations or methylation pattern. However, expressed RNA can capture mutations, changes in expression levels due to methylation, and provide information on cell of origin, growth, and proliferation status. We developed an approach to isolate cell-free total nucleic acid (cfDNA) and used targeted next generation sequencing to sequence cell-free RNA (cfRNA) and cfDNA as new approach in liquid biopsy. We demonstrate that cfRNA is overall more sensitive than cfDNA in detecting mutations. We show that cfRNA is reliable in detecting fusion genes and cfDNA is reliable in detecting chromosomal gains and losses. cfRNA levels of various solid tumor biomarkers were significantly higher (P < 0.0001) in samples from solid tumors as compared with normal control. Similarly, cfRNA lymphoid markers and cfRNA myeloid markers were all higher in lymphoid and myeloid neoplasms, respectively as compared with control (P < 0.0001). Using machine learning we demonstrate cfRNA was highly predictive of diagnosis (AUC >0.98) of solid tumors, B-cell lymphoid neoplasms, T-cell lymphoid neoplasms, and myeloid neoplasms. In evaluating the host immune system, cfRNA CD4:CD8B and CD3D:CD19 ratios in normal controls were as expected (median: 5.92 and 6.87, respectively) and were significantly lower in solid tumors (P < 0.0002). This data suggests that liquid biopsy combining analysis of cfRNA with cfDNA is practical and may provide helpful information in predicting genomic abnormalities, diagnosis of neoplasms and evaluating both the tumor biology and the host response.

Differential impact of cognitive computing augmented by real world evidence on novice and expert oncologists.

INTRODUCTION: Cognitive computing point-of-care decision support tools which ingest patient attributes from electronic health records and display treatment options based on expert training and medical literature, supplemented by real world evidence (RWE), might prove useful to expert and novice oncologists. The concordance of augmented intelligence systems with best medical practices and potential influences on physician behavior remain unknown. METHODS: Electronic health records from 88 breast cancer patients evaluated at a USA tertiary care center were presented to subspecialist experts and oncologists focusing on other disease states with and without reviewing the IBM Watson for Oncology with Cota RWE platform. RESULTS: The cognitive computing "recommended" option was concordant with selection by breast cancer experts in 78.5% and "for consideration" option was selected in 9.4%, yielding agreements in 87.9%. Fifty-nine percent of non-concordant responses were generated from 8% of cases. In the Cota observational database 69.3% of matched controls were treated with "recommended," 11.4% "for consideration", and 19.3% "not recommended." Without guidance from Watson for Oncology (WfO)/Cota RWE, novice oncologists chose 75.5% recommended/for consideration treatments which improved to 95.3% with WfO/Cota RWE. The novices were more likely than experts to choose a non-recommended option (P < .01) without WfO/Cota RWE and changed decisions in 39% cases. CONCLUSIONS: Watson for Oncology with Cota RWE options were largely concordant with disease expert judged best oncology practices, and was able to improve treatment decisions among breast cancer novices. The observation that nearly a fifth of patients with similar disease characteristics received non-recommended options in a real world database highlights a need for decision support.

Real-world economic value of a 21-gene assay in early-stage breast cancer.

OBJECTIVES: Value-based payment reforms shift cost-containment responsibilities to the physician. Although gene expression profiling (GEP) utilizing a 21-gene panel among patients with early-stage, axillary lymph node-negative, hormone receptor-positive, HER2/neu oncogene-negative breast cancer is able to identify a cohort that may achieve excellent outcomes without adjuvant chemotherapy, high up-front costs (list price, $4175) could dissuade usage. STUDY DESIGN: Retrospective review of consecutive patients with breast cancer treated at a single cancer center. METHODS: Chart review of 227 patients 70 years or younger with outpatient costs (ie, drug average sales price, reagent costs, physician charges) during first 6 months of treatment. RESULTS: Of these patients, 68% underwent GEP, with 52%, 43%, and 5% having low, intermediate, and high recurrence risk scores, respectively. Adjuvant chemotherapy was utilized less in genomically profiled cohorts (19% vs 29%; P = .08) and was consistent with recommendations of the recurrence scores. The mean 6-month outpatient costs were $24,955 with adjuvant chemotherapy and $2654 with hormonal therapy. Patients with stage II cancer undergoing GEP received adjuvant chemotherapy at a lower frequency (28.6% vs 86.7%), but patients with stage I cancer who underwent testing were slightly more likely to receive chemotherapy (15.8% vs 14%) because the test identified patients with higher-risk tumors. Universal GEP testing of patients with stage II cancer would have resulted in net savings of $11,494 per patient inclusive of test cost; stage I testing would have increased costs by $4505. Similar trends for grade 2/3 tumors (-$2394) and grade 1 tumors (+$6047) were noted. CONCLUSIONS: Universal GEP testing of women 70 years or younger with stage II or grade 2/3 lymph node-negative breast cancers would result in lower outpatient costs, inclusive of the diagnostic test, within the first 6-month episode of care.

Nonamplification ERBB2 genomic alterations in 5605 cases of recurrent and metastatic breast cancer: An emerging opportunity for anti-HER2 targeted therapies.

BACKGROUND: Activating, nonamplification ERBB2 mutations (ERBB2mut) are not detected by immunohistochemistry (IHC) or fluorescence in situ hybridization (FISH), but are detected by DNA sequencing and may predict clinical responses to human epidermal growth factor receptor (HER2)-targeted therapy. The authors queried 5605 advanced/metastatic breast cancers (mBC) to uncover the frequency of ERBB2mut genomic alterations. Clinical responses to anti-HER2 therapeutics were identified. METHODS: DNA was extracted from 40 µm of formalin-fixed paraffin-embedded (FFPE) sections. Comprehensive genomic profiling (CGP) was used to evaluate up to 315 genes (592× mean coverage depth). Results were analyzed for base substitutions, short indels, copy number changes, and selected rearrangements. RESULTS: Of 5605 cases, 698 (12.5%) featured ERBB2 alterations, including 596 (10.6%) ERBB2 amplifications (ERBB2amp) and 138 (2.4%) ERBB2mut; 38 cases (0.7%) had co-occurring ERBB2amp and ERBB2mut. ERBB2mut predominantly affected the kinase (124 cases; 90%) or extracellular (15 cases; 11%) domains. Both primary BC (52 cases; 38%) and metastatic site biopsies (86 cases; 62%) were found to harbor ERBB2mut, which were distributed across carcinoma not otherwise specified (NOS) (69 cases; 50%), invasive ductal carcinoma (IDC) (40 cases; 29%), invasive lobular carcinoma (ILC) (27 cases; 20%), and mucinous mBC (2 cases; 1%). Genes commonly coaltered with ERBB2 were tumor protein 53 (TP53) (49%); phosphatidylinositol 3-kinase catalytic subunit alpha (PIK3CA) (42%); cadherin 1, type 1 (CDH1) (37%); MYC (17%); and cyclin D1 protein (CCND1) (16%). CDH1 mutations were enriched in ERBB2mut mBC (P<0.0006) and associated with recurrent mBC. Selected patients with ERBB2mut, without ERBB2amp, who responded to anti-HER2 targeted therapies are presented herein. CONCLUSIONS: Within this large series, 1.8% of cases harbored ERBB2mut, which are undetectable by standard-of-care IHC or FISH tests. Metastatic BC driven by ERBB2mut respond to anti-HER2 targeted therapies, and expanding clinical trials designed to detect ERBB2mut by CGP and optimize targeted treatments are warranted. Cancer 2016. © 2016 American Cancer Society. Cancer 2016;122:2654-2662. © 2016 American Cancer Society.

Unusual viral infections (progressive multifocal leukoencephalopathy and cytomegalovirus disease) after high-dose chemotherapy with autologous blood stem cell rescue and peritransplantation rituximab.

Efforts to reduce relapse of non-Hodgkin lymphoma after autologous transplantation have included ex vivo stem cell selection and/or peritransplantation immunotherapy. The late infectious and immunologic consequences of these maneuvers are not well understood, although an increase in early cytomegaloviral disease after CD34(+) stem cell selection and an alteration in immunoglobulin and T-cell recovery after peritransplantation rituximab has been noted. We report the first 2 cases of progressive multifocal leukoencephalopathy caused by JC papovavirus after autologous peripheral blood stem cell transplantation and a case each of cytomegalovirus retinitis and pneumonitis. All 4 patients experienced significant impairment of CD4 T-cell recovery, placing them at risk for these unusual viral infections. The clustering of cases is concerning because all occurred shortly after the introduction of peritransplantation rituximab into treatment protocols (4 of 62 immunotherapy recipients compared with 0 of 276 without; z = 3.595; P <.001), although a direct association with this CD20 B-cell-directed therapy remains speculative.