GENE MUTATIONS AND COPY NUMBER ALTERATIONS (CNA) PREDICT FOR EARLY FAILURE IN PATIENTS WITH DIFFUSE LARGE B‐CELL LYMPHOMA (DLBCL) TREATED WITH R‐CHOP

Introduction: Recently, we described a 23‐gene gene expression‐ based biomarker that was prognostic for overall survival (OS) in patients (pts) with locally extensive and advanced stage classical Hodgkin lymphoma treated with ABVD chemotherapy (Scott et al. J Clin Oncol2013). The observation that the results of PET scan after 2 cycles (PET2) of ABVD is highly prognostic of progression free survival (PFS) has led to the design of response‐adapted trials, where pts with a positive PET2 receive more intensive regimens aiming to improve outcomes. The purpose of this study was to test whether the 23‐gene assay could predict PET2 results, allowing rational selective escalation of treatment from the time of diagnosis. The US Intergroup S0816 trial was a response‐adapted trial in advanced stage (III and IV) classical Hodgkin lymphoma, with escalation to escBEACOPP where the PET2 was Deauville 4 or 5 (Press et al. J Clin Oncol 2016). Methods: The 23‐gene assay was applied to RNA extracted from the formalin‐fixed paraffin‐embedded (FFPE) diagnostic biopsies of 276 pts from the S0816 trial. Patients were categorized into low‐ and high‐risk groups using previously described assay thresholds. The primary endpoint was the comparison of the proportion of PET2 positive pts in the two risk groups. The secondary endpoint explored whether the risk groups were predictive of PFS in the context of response‐ adapted treatment. Results: Adequate gene expression was obtained in 217/276 (79%) biopsies – 119/164 (73%) from archival slides and 98/112 (88%) FFPE blocks. The patient characteristics of the 217 pts were not significantly different from the remainder of the total trial population of 336 pts. 142 (65%) pts were assigned to the low‐ and 75 (35%) to the high‐risk gene expression group. 38 (18%) pts had a positive PET2. The assay did not identify pts at greater risk of PET2 positivity with 21% and 11% of pts having a positive PET2 in the low‐ and high‐risk group, respectively. To test whether this was the result of a suboptimal threshold, we examined the area under the curve (AUC) of the receiver operator characteristic curve. The AUC of 0.41 (95% CI 0.31 – 0.51) demonstrates that the assay is not predictive of PET2 positivity. Furthermore, the assay was not predictive of PFS, with a 2‐year PFS of 80% and 84% in the low‐ and high‐risk groups, respectively (log rank P = 0.65). There were not sufficient events to test the prognostic power of the assay for OS. Conclusions: The 23‐gene assay, trained on the endpoint of OS, was not predictive of PET2 results. Furthermore, it was not predictive of PFS in this trial of response‐adapted treatment. These results provide strong evidence that this assay should not be used for risk stratification where response‐adapted therapy is utilized. Support: National Cancer Institute (NCI), National Clinical Trials Network (NCTN) grants CA180888, CA180819, CA180821, and CA180820; Lymphoma Research Foundation.