Primary Mediastinal B-Cell Lymphoma: Evaluation of Clinicopathologic Diagnosis Compared to Gene Expression Based Diagnosis in a Clinical Trial with CD30+ B-Cell Lymphomas

Background: PMBCL is a unique subtype of aggressive B-cell lymphoma representing about 5% of lymphoma cases. The diagnosis is generally based on a combination of clinical features (e.g., mediastinal mass) and pathological findings on tissue biopsy (e.g., large neoplastic B-cells with variable CD30 positivity by immunohistochemistry). However, the histopathologic diagnostic criteria are not well defined and the distinction between PMBCL and diffuse large B-cell lymphoma (DLBCL) or gray zone lymphoma (GZL) involving the mediastinum can be challenging. Most PMBCL trials use the traditional diagnostic criteria for study entry. Specific treatment approaches based on results of these trials are designed for patients with PMBCL. In this study, we hypothesized that a gene expression based assay that characterizes the molecular signature of PMBCL using formalin-fixed, paraffin-embedded (FFPE) tissue may improve the diagnostic criteria and allow more accurate interpretation of results for lymphoma patients enrolled in clinical trials.

Methods: This exploratory study compared the PMBCL diagnosis established by clinicopathologic criteria alone to the diagnosis assigned by a combination of clinicopathologic features and gene expression-based assay on FFPE tissue specimens of patients enrolled in a multisite phase I/II prospective trial using brentuximab vedotin (BV) in combination with rituximab - cyclophosphamide-hydroxydoxorubicin-prednisone (R-CHP) for CD30+ B-cell lymphomas (Svoboda, Blood 2017). The original diagnostic categories of PMBCL vs. DLBCL vs. GZL were assigned by investigators based on traditional clinicopathologic features. For exploratory Nanostring based diagnostic categorization, we used previously described and validated Lymph3Cx assay which consists of 64 probes with cut‐points defined at the 0.1 and 0.9 probability scores to distinguish between DLBCL and PMBCL (Mottok, Hematol Oncol 2017). The tissue was examined by a hematopathologist for adequate tumor content and nucleic acids were extracted from 10 mm FFPE scrolls or unstained slides. Survival curves were generated for PMBCL patients as categorized by investigator assessment alone and by investigator assessment plus molecular classification using STATA.

Results: We enrolled 31 treatment-naïve patients with CD30+ B-cell lymphomas between January 2014 and April 2017 (NCT01994850). Based on investigator assessment, patients were categorized as PMBCL (N=23), DLBCL (N=6), and GZL (N=2). As of June 15, 2018, we obtained and analyzed diagnostic FFPE tissue using the Lymph3Cx assay on 14 pts with all 3 subtypes of CD30+ B-cell lymphomas: PMBCL (N=11), DLBCL (N=2), and GZL (N=1). Of 11 pts with PMBCL by investigator assessment alone, 8 pts (73%) had Lymph3Cx probability scores > 0.9 which was consistent with a diagnosis of PMBCL by gene expression; 2 pts (18%) scored in the indeterminate category (0.1 to 0.9); 1 pt (9%) scored as DLBCL (< 0.1). All 8 pts with a concordant diagnosis of PMBCL by investigator assessment and gene expression assay achieved complete remission (CR) and remain progression free after completing BV+R-CHP with median follow-up of 18 months. However, 1 pt re-classified as DLBCL by Lymph3Cx and 1 of 2 pts with an indeterminate score by Lymph3Cx achieved only partial responses and ultimately progressed; 1 pt with an indeterminate score remains in CR. None of the non-PMBCL pts in our exploratory analysis (2 DLBCL; 1 GZL) as assessed by investigators were categorized as PMBCL by Lymph3Cx. The CR rate for patients categorized as PMBCL by investigator assessment alone was 82% compared to 100% in those categorized as PMBCL by both investigator and gene expression assay (Table 1). The reportable progression free survival would also be different for these two cohorts (Figure 1). We will complete Lymph3Cx testing of diagnostic tissue for all 31 enrolled patients with CD30+ B-cell lymphomas enrolled on our clinical trial by the meeting.

Conclusion: Preliminary results of this ongoing study suggest that a diagnosis of PMBCL by clinicopathologic assessment alone that is not supported by molecular classification may include non-PMBCL pts and affect treatment outcomes. We recommend that future clinical trials for PMBCL include gene expression based diagnostic assays to improve diagnostic accuracy and interpretation of results.

View largeDownload slide

View largeDownload slide

Close modal