Wright, George W.

Abstract Although follicular lymphoma (FL) typically follows an indolent course, patients with FL who experience early events, such as transformation or progression, have increased risk of death related to lymphoma. The FL24Cx is an algorithm based on a 45-target gene expression profiling (GEP) assay, which was developed and trained using 265 formalin-fixed, paraffin-embedded tissue samples on a reliable platform to predict, at the time of diagnosis, whether a patient will experience an event within 24 months. The modeling also confirmed and relied upon previously reported synergy between immune response (IR) gene expression signatures IR1 and IR2. Once locked, the 5-factor logistic regression FL24Cx model was independently validated in a retrospectively assessed cohort of 232 patients from 2 immunochemotherapy-treated arms of SWOG Cancer Research Network S0016 phase 3 clinical trial, in which it assigned 169 patients to the low-risk group with 29 events before 24 months (17.2%) and 63 patients to the high-risk group with 24 events before 24 months (38.1%). The relative risk of an event within 24 months after registration among patients who were classified into the high-risk group relative to patients who were classified into the low-risk group was 2.2 (95% confidence interval, 1.41 to 3.51). An up-front GEP biomarker, such as the FL24Cx, rigorously validated in a clinical laboratory and with a clinically relevant turnaround time, could identify and steer enrollment of patients at high risk for early events in clinical trials, thus enabling timely interpretation of such trials and increasing the pace of innovation.

Background: Anaplastic large cell lymphomas (ALCLs) represent a heterogeneous group of T-cell lymphomas that currently are classified by the presence or absence of ALK tyrosine kinase (TK) fusion genes (ALK+ or ALK−) and clinical presentation (systemic, cutaneous, or breast implant-associated). Two overarching molecular types of ALCL recently were discovered, defined by the presence (Type I) or absence (Type II) of a gene expression signature highly enriched for JAK-STAT3 activation. Fusions involving non-ALK TK genes occur in some ALK− cases, but the fusion landscape of ALCL remains incompletely characterized. Methods: Expert consensus pathology review was conducted in the Lymphoma/Leukemia Molecular Profiling Project (LLMPP). RNAseq was performed and fusions were identified using FusionCatcher. Here, we focused on recurrent in-frame coding fusions. Previously unreported fusions were validated by RT-PCR. Type I/II was assigned using a previously validated gene expression-based model. Genes with adjusted (adj) P<0.05 were considered differentially expressed. Overall survival (OS) was assessed for systemic ALCL, when available. Results: We evaluated 379 ALCLs (229M/150F; mean age, 56 y). Of 199 candidate fusions (excluding reciprocal events), 28 were recurrent and passed quality metrics. At least 1 of these 28 fusions was present in 150 cases (40%). ALK fusions were present in 106 ALK+ ALCLs (all Type I; P<0.0001). Of these, NPM1::ALK was seen in 68/106 (64%). Alternate partners (X::ALK) included ATIC (N=24) and CTLC, COL1A2, MSN, MYH9, RNF213, SATB1, TFH, TPM3, and TRAF1 (1-3 cases each). X::ALK was associated with older age (mean, 52 y) than NPM1::ALK (32 y; P<0.0001) and showed relative overexpression of 49 genes, including multiple activators of small GTPases such as CGNL1 (FC, 5.6; Padj=2.4×10-6), SRGAP1, ALS2, DOCK1, and NCKAP1. ALK expression was similar in cases with NPM1::ALK and X::ALK and there was no significant difference in OS. Non-ALK TK fusions were seen in 17/273 ALK− cases (6%), including TYK2 (N=9); JAK2 (N=6); and ROS1 (N=2). All were Type I ALCLs (P<0.0001). ALK− cases with non-ALK TK fusions overexpressed 38 genes compared to ALK− cases without TK fusions, including cytokines involved in the IL17 signaling pathway such as CXCL6 (FC, 12.5; Padj=4.2×10-3), CXCL1, and CSF3. We then assessed the relationship between ALK− cases with non-ALK TK fusions and ALK+ cases by separately comparing each subset to ALK− cases without TK fusions. FC values for ALK− cases with non-ALK TK fusions were significantly correlated with FC values for ALK+ ALCLs over the set of expressed genes (R=0.54; P<0.0001). At a median follow-up of 32 months, 0/4 systemic ALCL patients with non-ALK TK fusions had died, while 30% of those with ALK− ALCL without TK fusions had died, but this was not statistically significant (P=0.26). The other 13 ALK− cases with non-ALK TK fusions were either localized ALCLs or did not have outcome data available. Of 20 TP63 fusions, 17 (85%) were TBL1XR1::TP63 and 3 were X::TP63. As previously reported, TP63 fusions were associated with poor OS (median OS, 10 months; not reached for other ALK− ALCLs; P=0.007). Four ALCLs had a novel NCL::UBTF fusion involving genes encoding the nucleolar proteins nucleolin and nucleolar (upstream binding) transcription factor-1. Four cases each had PARG::BMS1 and TNK1::GPS2 (the latter seen only in Type I ALCLs). Conclusions: This large consortium-based analysis of coding fusions further elucidates the molecular distinction between Type I and Type II ALCLs. Both ALK and non-ALK TK fusions were seen exclusively in Type I ALCLs, which we have shown previously are enriched for JAK-STAT3 pathway genes and correlate strongly with pSTAT3Y705 positivity by immunohistochemistry. These fusions were not identified in Type II ALCLs, which previous data suggest are associated with epigenetic alterations. ALK+ ALCLs with NPM1::ALK and X::ALK showed differences in gene expression, suggesting biological differences. Furthermore, ALCLs with non-ALK TK fusions shared gene expression features with ALK+ ALCL. Clinical fusion detection could enhance molecular classification of Type I and Type II ALCLs and may guide precision therapy.

Abstract Rearrangements that place the oncogenes MYC, BCL2, or BCL6 adjacent to superenhancers are common in mature B-cell lymphomas. Lymphomas with diffuse large B-cell lymphoma (DLBCL) or high-grade morphology with both MYC and BCL2 rearrangements are classified as high-grade B-cell lymphoma with MYC and BCL2 rearrangements (“double hit”; HGBCL-DH-BCL2) and are associated with aggressive disease and poor outcomes. Although it is established that MYC rearrangements involving immunoglobulin (IG) loci are associated with inferior outcomes relative to those involving other non-IG superenhancers, the frequency of and mechanisms driving IG vs non-IG MYC rearrangements have not been elucidated. Here, we used custom targeted capture and/or whole-genome sequencing to characterize oncogene rearrangements across 883 mature B-cell lymphomas including Burkitt lymphoma, follicular lymphoma, DLBCL, and HGBCL-DH-BCL2 tumors. We demonstrate that, although BCL2 rearrangement topology is consistent across entities, HGBCL-DH-BCL2 have distinct MYC rearrangement architecture relative to tumors with single MYC rearrangements or with both MYC and BCL6 rearrangements (HGBCL-DH-BCL6), including both a higher frequency of non-IG rearrangements and different architecture of MYC::IGH rearrangements. The distinct MYC rearrangement patterns in HGBCL-DH-BCL2 occur on the background of high levels of somatic hypermutation across MYC partner loci in HGBCL-DH-BCL2, creating more opportunity to form these rearrangements. Furthermore, because 1 IGH allele is already disrupted by the existing BCL2 rearrangement, the MYC rearrangement architecture in HGBCL-DH-BCL2 likely reflects selective pressure to preserve both BCL2 and B-cell receptor expression. These data provide new mechanistic explanations for the distinct patterns of MYC rearrangements observed across different lymphoma entities.

Background: Anaplastic large cell lymphomas (ALCLs) are CD30-positive T-cell lymphomas that share pathologic features but differ in clinical presentation, outcome, and molecular features. The World Health Organization (WHO) and International Consensus Classification (ICC) classify ALCLs by presence or absence of ALK rearrangements (R) and clinical presentation (systemic, cutaneous [c], or breast implant-associated [BIA]). ICC, but not WHO, recognizes DUSP22-R as defining a new genetic subtype of ALK- ALCL. The classifications otherwise do not reflect additional molecular heterogeneity in genetics (e.g., TP63-R) or therapeutic vulnerabilities (e.g., JAK-STAT3 pathway activation). Methods: ALCLs (N=689) underwent expert consensus review (WHO/ICC) through the Lymphoma/Leukemia Molecular Profiling Project (LLMPP). All cases also underwent genetic subtyping (ALK, DUSP22, TP63, and triple-negative [TN]) using fluorescence in situ hybridization (FISH) and immunohistochemistry (IHC), as well as IHC for phospho-STAT3 Tyr705 (pSTAT3). RNAseq was performed and evaluable in 393 cases; the remaining cases had insufficient tissue, tumor content, RNA yield or quality, and/or sequencing data quality. Sequenced and non-sequenced sub-cohorts had similar demographics and subtype distribution. Results: Unsupervised gene expression profiling (GEP) identified 2 main molecular types of ALCL. Type I ALCLs predominantly included ALK+ ALCLs, BIA-ALCLs, and a subset of TN ALCLs (designated TN-I), whereas Type II ALCLs predominantly included ALCLs with DUSP22-R, TP63-R, or both (double-hit; DH), and the remaining TN ALCLs (TN-II). Type I ALCLs were strongly associated with pSTAT3 expression (74.2±26.4% positive malignant cells vs 9.9±22.7% for Type II; P<0.0001). An independently derived pSTAT3 staining threshold of 30% assigned Type I vs II with 91% accuracy. A third cluster of ALCLs, mostly cALCL, showed an epithelial GEP signature rather than a lymphoma signature; these cases were assigned to Types I or II based on pSTAT3 IHC. Distinct sub-signatures were identified for ALK+, DUSP22-R, TP63-R, and BIA ALCL (Fig. 1), but not ALK- ALCL or cALCL. Gene set enrichment analysis showed Type I ALCLs to be enriched for JAK-STAT3 signaling genes (normalized enrichment score [NES], 2.22; FDR<0.0001) and related pathways, such as TNFα-NFκB signaling (NES, 2.21; FDR<0.0001). In contrast, Type II ALCLs were enriched for cell cycle, DNA repair, epigenetic, and metabolic pathway genes, but not for major tyrosine kinase-mediated signaling pathway genes. Enriched epigenetic pathways included chromatin modifying enzymes (NES, -1.86; FDR=0.002) and histone methylation (NES, -1.71; FDR=0.002). EZH2 was the most overexpressed gene in Type II ALCLs (fold-change, 7.74; FDR=8.84×10 -306). At the protein level, EZH2 IHC H-scores were 275±45 in Type II and 168±70 in Type I ALCLs (P<0.0001); H3K27me3 H-scores were 170±81 and 76±63, respectively (P<0.001). The top metabolic gene set involved cholesterol biosynthesis (NES, -1.88; FDR=0.002). Overall survival (OS) data were available in 257 systemic ALCL patients (145 sequenced and 112 non-sequenced; cALCL and BIA-ALCL were excluded). Non-sequenced TN ALCLs were assigned to TN-I or TN-II using pSTAT3 IHC. The results supported earlier data indicating favorable prognosis of DUSP22-R ALCL (5 y OS, 95%) and ALK+ ALCL (87%), intermediate prognosis of TN ALCL (TN-I, 52% and TN-II, 38%; P=NS), and poor prognosis of TP63-R/DH ALCL (0%)(P<0.0001; Fig. 2). Conclusions: Two overarching molecular types of ALCL exist, predominantly associated with the presence (Type I) or absence (Type II) of the JAK-STAT3 signaling program. Distinct GEP signatures exist for ALK+ ALCL and BIA-ALCL (predominantly Type I), and DUSP22-R ALCL and TP63-R ALCL (predominantly Type II). TN ALCLs lacking ALK-R, DUSP22-R, and TP63-R can be stratified into TN-I and TN-II subtypes. pSTAT3 IHC has >90% accuracy as a surrogate for GEP-based subtyping. ALK- ALCL and cALCL cluster by molecular subtype rather than by defining GEP signatures. Type II ALCLs are enriched for targetable epigenetic and metabolic pathways, including EZH2/histone methylation and cholesterol biosynthesis. This molecular classification is diagnostically, prognostically, and potentially therapeutically relevant, and can be applied using FISH and IHC in routine practice and in the clinical trial setting.

Introduction Frontline therapy for pts with follicular lymphoma (FL) is not based on the molecular profile of the tumor. The PI3K pathway is central to FL biology, but the molecular profile of tumors most sensitive to PI3K inhibition (PI3Ki) is unknown. Copanlisib inhibits both PI3Kα and δ isoforms, and we hypothesized that most pts with treatment naïve FL will have PI3Ki-sensitive tumors and durable complete responses (CRs) may be achievable with short durations of targeted therapy. A gene expression profiling (GEP) predictor utilizes a combination of the expression of 45 selected informative genes and an additional 19 housekeeping genes to identify FL pts at high-risk for early progression within 24 months (Wright et al. ASH Annual Meeting2022). We also hypothesized that copanlisib would induce durable CRs in high-risk pts by the GEP assay. Here, we report a preliminary analysis of an ongoing “window of opportunity” study of copanlisib followed by response-adapted copanlisib and rituximab in treatment-naïve pts with FL (NCT03789240). Methods Pts with untreated grade 1-2, 3A FL, ≥stage 2 are eligible if systemic therapy is indicated: symptoms, progressive lymphadenopathy, or organ compromise. Prior radiation is permitted. Eligibility includes age ≥18 and adequate organ function unless due to lymphoma. HIV, CMV, Hep B or C, and autoimmune conditions are excluded. Pts first receive copanlisib 60mg on days 1, 8, and 15 of a 28-day “window of opportunity” to explore the activity of copanlisib. Following the window, pts receive 6 cycles of copanlisib 60mg on days 1, 8, and 15 of a 28-day cycle along with rituximab 375mg weekly x 4 then on day 1 of each cycle and response is assessed with FDG-PET and CT scans. Pts without response after C6 are taken off study and those with a partial response (PR) can continue for an additional 6 cycles and then stop therapy. Pts who achieve a CR after 6 cycles stop therapy. No maintenance is given. Supportive care includesPCP prophylaxis. The primary endpoint is the CR rate with secondary endpoints of safety, duration of CR, and PFS. Exploratory objectives include identification of a predictive signature of PI3Ki response and the response rate in high-risk pts identified by the GEP assay. Results Twenty-eight pts have enrolled. Median age was 56y (range, 24-80). Seventeen (61%) and 12 (43%) pts had high-risk FLIPI and FLIPI-2 scores ≥3, respectively; 6 (21%) pts had Grade 3A FL, and 17 (61%) pts met two or more GELF criteria. After copanlisib monotherapy in the window, 27 (96%) pts had tumor reductions by CT with a median change of -38% (-62% to +16%) (Figure 1). Among 22 pts who completed 6 cycles, 21 (96%) responded, including 9 (41%) pts with PR and 12 (55%) pts with CR. Of 9 pts with PR who received an additional 6 cycles of therapy, only 1 (11%) converted to a CR bringing the overall CR rate to 59% (13/22 pts). After a median follow-up of 25.8 months, the 2-year PFS was 43.2% (95% CI, 20.7-64.0%). The median duration of CR was 29.7 months (95% CI, 6.3 months - not estimable) with 8 (62%) CRs ongoing (Figure 2). Toxicity was evaluated in 28 pts across 220 cycles. Most common were rash (64%), mucositis (50%), and nausea/vomiting (50%), all G1-2 except 1 (4%) pt with G3 mucositis. The most common ≥G3 toxicities were neutropenia and infections in 5 (18%) pts each. ≥G3 neutropenia occurred in 10 (5%) cycles, responded to therapy interruption and/or growth factors, and did not require dose reduction. Two (8%) pts had copanlisib dose reduced, 1 (4%) each due to rash and mucositis. Therapy was prematurely discontinued in 5 pts: 3 due to COVID-19, 1 for hepatotoxicity that resolved, and 1 due to infusion reaction to rituximab. Nine of 24 (38%) pts were poor risk by the GEP assay. Complete responses were observed in 71%, 67%, and 57% of pts categorized as high-risk by FLIPI, FLIPI-2, and GEP assay, respectively. The CR rate was not affected by bulky disease (any LN >7cm; p=1.00), high-risk FLIPI (p=0.19), high-risk FLIPI-2 (p=0.67), poor-risk GEP (p=1.00), or metabolic tumor volume (p=0.33). Conclusion Nearly all pts with treatment naïve FL have PI3Ki-responsive tumors, and the combination of copanlisib with rituximab achieved a CR rate of 59% at the end of induction therapy. The median duration of CR is nearly 30 months, including ongoing remissions without maintenance. CRs were observed in pts deemed high-risk by the GEP assay and FLIPI score. The safety profile is excellent, with mostly G1-2 toxicities manageable with supportive care.

PURPOSE Peripheral T-cell lymphoma (PTCL) includes heterogeneous clinicopathologic entities with numerous diagnostic and treatment challenges. We previously defined robust transcriptomic signatures that distinguish common PTCL entities and identified two novel biologic and prognostic PTCL-not otherwise specified subtypes (PTCL-TBX21 and PTCL-GATA3). We aimed to consolidate a gene expression–based subclassification using formalin-fixed, paraffin-embedded (FFPE) tissues to improve the accuracy and precision in PTCL diagnosis. MATERIALS AND METHODS We assembled a well-characterized PTCL training cohort (n = 105) with gene expression profiling data to derive a diagnostic signature using fresh-frozen tissue on the HG-U133plus2.0 platform (Affymetrix, Inc, Santa Clara, CA) subsequently validated using matched FFPE tissues in a digital gene expression profiling platform (nCounter, NanoString Technologies, Inc, Seattle, WA). Statistical filtering approaches were applied to refine the transcriptomic signatures and then validated in another PTCL cohort (n = 140) with rigorous pathology review and ancillary assays. RESULTS In the training cohort, the refined transcriptomic classifier in FFPE tissues showed high sensitivity (> 80%), specificity (> 95%), and accuracy (> 94%) for PTCL subclassification compared with the fresh-frozen–derived diagnostic model and showed high reproducibility between three independent laboratories. In the validation cohort, the transcriptional classifier matched the pathology diagnosis rendered by three expert hematopathologists in 85% (n = 119) of the cases, showed borderline association with the molecular signatures in 6% (n = 8), and disagreed in 8% (n = 11). The classifier improved the pathology diagnosis in two cases, validated by clinical findings. Of the 11 cases with disagreements, four had a molecular classification that may provide an improvement over pathology diagnosis on the basis of overall transcriptomic and morphological features. The molecular subclassification provided a comprehensive molecular characterization of PTCL subtypes, including viral etiologic factors and translocation partners. CONCLUSION We developed a novel transcriptomic approach for PTCL subclassification that facilitates translation into clinical practice with higher precision and uniformity than conventional pathology diagnosis.

Abstract Introduction: Diffuse large B-cell lymphoma (DLBCL) is an aggressive non-Hodgkin’s lymphoma (NHL) comprising 3 molecular subtypes: germinal center B-cell (GCB), activated B-cell (ABC) and class III. ABC patients (pts) have a poor prognosis. The immunomodulatory drug lenalidomide (Len) produces durable responses in pts with aggressive NHL (Witzig 2011), with preferential activity reported in non-GCB DLBCL (Hernandez-Ilizaliturri 2011). Methods: This randomized, multicenter, open-label, phase 2/3 study was conducted to determine the efficacy and safety of single-agent Len vs single-agent investigator’s choice (IC) in relapsed/refractory DLBCL pts who received ≥2 prior therapies, or were ineligible for stem cell transplantation or further combination chemotherapy. DLBCL subtype (GCB vs non-GCB) was determined by a central pathology lab using immunohistochemistry (IHC) per the Hans method (Hans 2004). Pts were stratified by subtype, then randomized 1:1 to receive Len (25 mg/day, 21 days of 28-day cycle) or IC (gemcitabine, rituximab, etoposide, or oxaliplatin) until progressive disease (PD), unacceptable toxicity, or voluntary withdrawal. In the event of radiologically confirmed PD, pts in the IC arm were allowed to cross over to Len. The primary endpoint for Stage 1 was overall response rate (ORR), as determined by an Independent Response Assessment Committee. Progression-free survival (PFS), overall survival (OS) and subtype analysis using gene expression profiling (GEP) were exploratory endpoints. Concordance of GEP and IHC was evaluated from 3 separate laboratories. Results: IHC subtyping agreement rate among the 3 laboratories ranged from 87.5%-97.9%, and sensitivity of IHC to detect ABC or GCB subtypes vs GEP ranged from 92.3%-100.0%. By IHC, 102 DLBCL pts (GCB, n=48; non-GCB, n=54) were treated with ≥1 dose of Len or IC (modified intent-to-treat population) in Stage 1. In this heavily pretreated population, >90.0% of pts received ≥2 prior systemic chemotherapies; 25 pts in Len and 32 pts in IC received ≥3 prior systemic chemotherapy regimens. Median age was 65 y in the IC arm vs 69 y in the Len arm. Twenty-nine pts crossed over from IC to Len after confirmed PD. All pts, regardless of subtype or therapy group, experienced ≥1 treatment-emergent adverse event, with neutropenia, anemia, and thrombocytopenia being the most common. Efficacy data are presented in the Table. Pts with GCB or non-GCB DLBCL (per IHC) treated with Len had similar ORR, but the data suggested greater improvements in PFS and OS with Len vs IC in the non-GCB pts. In an exploratory analysis of pts subtyped by GEP, ABC pts treated with Len vs IC-treated showed greater improvements in ORR, PFS, and OS compared with GCB pts. Prespecified criterion to advance to Stage 2 was a 2-sided 15% significance level in ORR in favor of Len based on IHC-defined subtype. The data did not fulfill this requirement, and Stage 2 was not opened. Conclusion: Len monotherapy showed clinical activity in heavily pretreated pts with DLBCL. The data suggest improved ORR, PFS, and OS with Len vs IC in the non-GCB population as defined by IHC, and the difference appears to be more pronounced in the ABC population as defined by GEP. Subtyping by GEP is warranted in further studies of Len in DLBCL. Abstract 628. Table 1.Table. Efficacy DataBy IHCBy GEPOverallGCBNon-GCBGCBABCLen(n=51)IC(n=51)Len(n=23)IC(n=25)Len(n=28)IC(n=26)Len(n=14)IC(n=16)Len(n=11)IC(n=16)ORR, % (95% CI)27.5 (15.9-41.7)11.8 (4.4-23.9) 26.1 (10.2-48.4)12.0 (2.5-31.2)28.6 (13.2-48.7)11.5 (2.4-30.2)21.4 (4.7-50.8)12.5 (1.6-38.3)45.5 (16.7-76.6)18.8 (4.0-45.6)P Value .079 .279 .179 .642 .206PFS, med wk (95% CI)13.6 (8.6-17.7)7.9 (6.3-9.0) 10.1 (8.3-22.3)29.0 (6.3-20.6)15.1 (8.3-24.1)7.1 (5.3-8.4)13.2 (8.3-24.9)7.1 (6.0-20.6)82.0 (7.3-NA)6.2 (4.3-10.1)P Value .041 .550 .021 .506 .105HR (95% CI) 0.64 (0.41-0.99) 0.82 (0.43-1.57) 0.50 (0.27-0.92) 0.77 (0.35-1.68) 0.44 (0.15-1.23)OS, med wk (95% CI)31.0 (16.6-41.3)24.6 (12.7-33.9) 30.0 (14.9-44.4)24.9 (13.7-58.3)32.3 (15.9-48.1)20.4 (10.3-33.9)30.0 (18.0-34.6)20.1 (13.7-36.9)108.4 (9.6-108.4)18.6 (6.6-48.0)P Value .673 .526 .253 .767 .144HR (95% CI) 0.91 (0.59-1.41) 1.23 (0.65-2.34) 0.70 (0.38-1.30) 1.12 (0.52-2.42) 0.47 (0.17-1.33) Abbreviations: CI, confidence interval; HR, hazard ratio; med, median; NA, not applicable/not available. Disclosures Czuczman: Celgene: Consultancy. Off Label Use: This abstract describes a clinical trial of lenalidomide, which is an orally-available immunomodulatory agent under investigation for treating patients with diffuse large B-cell lymphoma.. Davies:GlaxoSmithKlein: Research Funding; Hoffman La Roche: Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding; Celgene: Research Funding. Wagner-Johnston:Celgene: Research Funding. Gascoyne:Celgene: Consultancy, Research Funding. Salles:Pfizer: Honoraria; Gilead: Honoraria; Jansen: Honoraria; Hoffman La Roche: Honoraria; Celgene: Honoraria; Mundipharma: Honoraria. Witzig:Celgene Corporation: Membership on an entity's Board of Directors or advisory committees, Research Funding. Zinzani:Mundipharma: Honoraria; Pfizer: Honoraria; Gilead: Honoraria, Membership on an entity's Board of Directors or advisory committees; Roche: Honoraria; Millennium Takeda: Honoraria; Celgene: Honoraria; Teva: Membership on an entity's Board of Directors or advisory committees. Wright:Celgene: Research Funding. Staudt:Celgene Corporation: Research Funding. Repici:Celgene: Employment. Song:Celgene: Employment. Manzke:Celgene: Employment.

Abstract Primary mediastinal large B-cell lymphoma (PMBCL) is recognized as a distinct entity in the World Health Organization classification. Currently, the diagnosis relies on consensus of histopathology, clinical variables, and presentation, giving rise to diagnostic inaccuracy in routine practice. Previous studies have demonstrated that PMBCL can be distinguished from subtypes of diffuse large B-cell lymphoma (DLBCL) based on gene expression signatures. However, requirement of fresh-frozen biopsy material has precluded the transfer of gene expression–based assays to the clinic. Here, we developed a robust and accurate molecular classification assay (Lymph3Cx) for the distinction of PMBCL from DLBCL subtypes based on gene expression measurements in formalin-fixed, paraffin-embedded tissue. A probabilistic model accounting for classification error, comprising 58 gene features, was trained on 68 cases of PMBCL and DLBCL. Performance of the model was subsequently evaluated in an independent validation cohort of 158 cases and showed high agreement of the Lymph3Cx molecular classification with the clinicopathological diagnosis of an expert panel (frank misclassification rate, 3.8%). Furthermore, we demonstrate reproducibility of the assay with 100% concordance of subtype assignments at 2 independent laboratories. Future studies will determine Lymph3Cx’s utility for routine diagnostic purposes and therapeutic decision making.