Performance of the cobas EZH2 mutation test on clinical samples from non-Hodgkin lymphoma patients.

OBJECTIVE: To present the technical verification and clinical validation of the companion diagnostic assay, cobas® EZH2 Mutation Test (cobas EZH2 Test), targeting gain-of-function EZH2 mutations in follicular lymphoma (FL) and diffuse large B-cell lymphoma (DLBCL). The focus is on patient clinical samples proving that the test met the performance criteria required for FDA approval of a companion diagnostic test. DESIGN: Epizyme, Inc., Eisai Co., Ltd., and Roche Molecular Systems, Inc., collaborated to develop the cobas EZH2 Test on an RT-PCR platform. The assay design needed to detect the gain-of-function EZH2 mutations found in FL and DLBCL indications. Thus, the test was optimized for investigational purposes in a clinical trial setting. Part of its technical verification included testing of patient tumor samples with a documented diagnosis of FL and DLBCL procured from commercial vendors, and the clinical validation used patient samples from the Epizyme clinical study. Both the technical performance verification method correlation study (104 clinical commercially acquired samples) and the clinical validation accuracy study (341 patient samples from the therapeutic study) used next-generation sequencing as a reference method to establish true vs. false results by cobas EZH2 Test. The reproducibility study used a 15-member panel of DNA samples with varying EZH2 mutation status from procured clinical FL and DLBCL patient samples under multiple variables. RESULTS: Single and rare, infrequent double EZH2 mutations were detected in FL and DLBCL samples. Agreements between results from cobas EZH2 and sequencing were >98% from commercial clinical samples and from the therapeutic study clinical samples. The reproducibility study obtained 178 to 180 valid results for each panel member, with an overall invalid rate of 0.37%. The agreement for each per panel member was 100%. CONCLUSION: cobas EZH2 Test data demonstrated that the test is reliable and will perform well in a commercial customer environment.

Analytic performance studies and clinical reproducibility of a real-time PCR assay for the detection of epidermal growth factor receptor gene mutations in formalin-fixed paraffin-embedded tissue specimens of non-small cell lung cancer.

BACKGROUND: Epidermal growth factor receptor (EGFR) gene mutations identify patients with non-small cell lung cancer (NSCLC) who have a high likelihood of benefiting from treatment with anti-EGFR tyrosine kinase inhibitors. Sanger sequencing is widely used for mutation detection but can be technically challenging, resulting in longer turn-around-time, with limited sensitivity for low levels of mutations. This manuscript details the technical performance verification studies and external clinical reproducibility studies of the cobas EGFR Mutation Test, a rapid multiplex real-time PCR assay designed to detect 41 mutations in exons 18, 19, 20 and 21. METHODS: The assay's limit of detection was determined using 25 formalin-fixed paraffin-embedded tissue (FFPET)-derived and plasmid DNA blends. Assay performance for a panel of 201 specimens was compared against Sanger sequencing with resolution of discordant specimens by quantitative massively parallel pyrosequencing (MPP). Internal and external reproducibility was assessed using specimens tested in duplicate by different operators, using different reagent lots, instruments and at different sites. The effects on the performance of the cobas EGFR test of endogenous substances and nine therapeutic drugs were evaluated in ten FFPET specimens. Other tests included an evaluation of the effects of necrosis, micro-organisms and homologous DNA sequences on assay performance, and the inclusivity of the assay for less frequent mutations. RESULTS: A >95% hit rate was obtained in blends with >5% mutant alleles, as determined by MPP analysis, at a total DNA input of 150 ng. The overall percent agreement between Sanger sequencing and the cobas test was 96.7% (negative percent agreement 97.5%; positive percent agreement 95.8%). Assay repeatability was 98% when tested with two operators, instruments, and reagent lots. In the external reproducibility study, the agreement was > 99% across all sites, all operators and all reagent lots for 11/12 tumors tested. Test performance was not compromised by endogenous substances, therapeutic drugs, necrosis up to 85%, and common micro-organisms. All of the assessed less common mutations except one (exon 19 deletion mutation 2236_2248 > AGAC) were detected at a similar DNA input level as that for the corresponding predominant mutation. CONCLUSION: The cobas EGFR Mutation Test is a sensitive, accurate, rapid, and reproducible assay.