Clinical and Pathological Features That Predict High-Grade B-Cell Lymphomas (HGBCLs) with MYC and BCL2 or BCL6 Translocations (Double-Hit Lymphoma): A Systematic Review and Meta-Analysis Authors: Ernest Lee, Wai Ying Katherine Wong, Han-Chieh Yang, Elizabeth J. Soilleux Published: 2026 ABSTRACT Background/Objectives: High-grade B-cell lymphomas (HGBCLs) with MYC and BCL2 or BCL6 translocations, colloquially referred to as double-hit lymphomas (and abbreviated here to DHL), are aggressive malignancies. Differentiating DHL from non-DHL HGBCLs is important, as DHL patients may benefit from more intensive treatment regimes. We aimed to identify predictive clinicopathological, morphological, and immunophenotypic features that could guide selection of HGBCLs for fluorescence in situ hybridization (FISH), which is expensive and less accessible in some centers. Methods: We conducted a PRISMA systematic review and meta-analysis on 29 studies identified from four databases (PubMed (MEDLINE), Ovid (Embase), Web of Science, and Scopus). We calculated risk ratios (RRs) to compare features between DHL and non-DHL HGBCL and between MYC/BCL2 and MYC/BCL6 DHL patients. Results: DHL patients were associated with higher Ann Arbor stage (RR 1.15, p = 0.028, I2 = 38.7%), International Prognostic Index (IPI) score (RR 1.27, p = 0.047, I2 = 37.9%), elevated lactate dehydrogenase (RR 1.26, p = 0.012, I2 = 34.0%), and germinal center B-cell-like (GCB) immunophenotype (RR 1.21, p = 0.043, I2 = 35.8%) compared to non-DHL HGBCL patients. c-Myc immunopositivity, extranodal disease, and bone marrow involvement were more likely in DHL, albeit not reaching statistical significance. Extranodal disease (p = 0.015, I2 = 0.0%), central nervous system involvement (p = 0.044, I2 = 0.0%), and non-GCB immunophenotype (p = 0.016, I2 = 71.1%) were more likely in MYC/BCL6 compared to MYC/BCL2 DHL patients. BCL2 immunopositivity, CD10 immunopositivity, and MUM1 immunonegativity were more likely in MYC/BCL2 DHL, although the differences were not statistically significant. Conclusions: Our results have associated DHL with features of aggressive disease and found GCB immunophenotype as a histopathological feature with statistically significant predictive value for MYC/BCL2 DHL. Heterogeneity within the non-DHL HGBCL group and variation in immunohistochemical cut-off values between studies limited identification of other predictive features. Larger, consistently designed, prospective cohort studies could provide further evidence for a screening strategy for DHL. 1. Introduction High-grade B-cell lymphoma (HGBCL) harboring MYC (8q24.2) and BCL2 (18q21.3) and/or BCL6 (3q27.3) rearrangement, also known colloquially as “double-hit lymphoma” (DHL), was introduced in the 2016 4th edition of the World Health Organization (WHO) Classification of Haematolymphoid Tumors (WHO-HAEM4) [1]. These entities account for 5–10% of all HGBCL, with MYC/BCL2 being the most common type (65–72% and 0.3–0.8 cases per 100,000 person-years), followed by MYC/BCL6 (10–14%) [2,3,4,5,6]. HGBCL with MYC/BCL6 rearrangements have since been removed from a specific “double-hit” category in the 2022 5th Edition WHO-HAEM5, recognizing that these malignancies have more heterogeneous gene expression profiles and mutational landscapes compared to MYC/BCL2-rearranged HGBCL [7]. These malignancies more commonly harbor mutations such as PIM1 and CD79B, linking them to the Activated B-Cell-like (ABC)-diffuse large B-cell lymphoma (DLBCL) subtype, whereas MYC/BCL2-rearranged HGBCLs possess more frequent epigenetic regulator gene mutations and cluster with follicular lymphoma and EZB-DLBCL [7]. In WHO-HAEM5, MYC/BCL6-rearranged HGBCL have now been reclassified as genetic subtypes of DLBCL, not otherwise specified (NOS), or of HGBCL, NOS [7]. The differentiation between these two groups is based on differences in morphology, with the former characterized by medium-sized to large B cells with a diffuse growth pattern that do not meet diagnostic criteria for specific large B-cell lymphomas, and the latter by medium-sized or blastoid cells that do not meet the diagnostic criteria for other defined lymphoma entities [7]. MYC/BCL6-rearranged HGBCL remains a provisional entity within the “double-hit HGBCL” category according to the 2022 International Consensus Classification [8]. Consistent with this, the WHO-HAEM5 classification continues to recommend reporting MYC/BCL6-rearranged cases to facilitate patient recruitment into research studies, given the incomplete biological and clinical characterization of these neoplasms [7]. Differentiating between DHL and non-DHL HGBCL (abbreviated here as non-DHL) is crucial, as DHL patients (possessing either MYC/BCL2 or MYC/BCL6 rearrangements) often present with a more aggressive clinical course and are known to have poorer prognosis when compared to, for example, single-hit DLBCL, which have MYC or BCL2 rearrangements alone or other molecular aberrations [9,10,11,12]. As a result, DHL patients often require more intensive chemoimmunotherapy than the conventional rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP) regimen [13]. These include a dose-adjusted regimen of etoposide, prednisone, vincristine, cyclophosphamide, and doxorubicin (DA-EPOCH), hyperfractionated cyclophosphamide, vincristine, doxorubicin, and dexamethasone (Hyper-CVAD), and cyclophosphamide, vincristine, doxorubicin, and cytarabine with methotrexate/etoposide, ifosfamide, and cytarabine (CODOX-M/IVAC) [13]. Accurate and timely diagnosis of MYC/BCL2- or MYC/BCL6-rearranged DHL, by identification of MYC, BCL2 or BCL6 gene rearrangement, is imperative to guide optimal treatment decisions. Fluorescence in situ hybridization (FISH) for MYC, BCL2 and BCL6 translocations is the current “gold standard” test in clinical practice for detecting the presence of these translocations. The WHO-HAEM5 guidelines advise a sequential strategy starting with evaluation of MYC translocation followed by the evaluation of BCL2 (and BCL6) rearrangement status and MYC translocation partner [7]. Targeted next-generation sequencing techniques have been suggested as the next step of screening to more precisely identify genetic alterations (such as single-nucleotide variants or copy number variations), as well as MYC translocation partners, which have clinical prognostic implications and may not be easily identifiable with routine interphase FISH [14,15]. There are currently no universally agreed-upon strategies or guidelines for selecting which HGBCLs should undergo FISH testing [7]. Proposed immunophenotypic strategies to identify DHL include germinal center B-cell-like (GCB) cell-of-origin (COO) immunohistochemistry [4,16], GCB COO immunohistochemistry testing together with c-Myc and BCL2 dual expression [4], and moderate to strong c-Myc immunostaining in >70% of neoplastic cells [17,18,19]. While pragmatic and cost-saving, these approaches have limitations, including missing a proportion of DHL cases and lacking specificity for identifying DHL. For example, some HGBCLs with high c-Myc expression do not carry MYC translocations [19]. The lack of reliable strategies may be attributable to overlapping biological features between DHL and non-DHL cases, as well as between the MYC/BCL2 and MYC/BCL6-rearranged DHL (abbreviated here to MYC/BCL2 and MYC/BCL6 DHL, respectively) [12]. Furthermore, our knowledge of DHL comes predominantly from studies with small sample sizes reflecting the relative underutilization, until recently, of FISH for diagnosing DHL, resulting in its under-diagnosis [20]. Without reliable stratification strategies, many centers have resorted to routine FISH screening strategies for all HGBCL cases. However, testing is typically performed in centralized cancer centers with turnaround time ranging from days to weeks [21]. Access in more resource-limited settings may be limited, delaying treatment decisions [20,22]. In this review, we aimed to collate results from multiple studies to identify predictive clinical and histopathological features to screen potential DHL patients for FISH testing. 2. Materials and Methods 2.1. Search Strategy This systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines (Supplementary Table S1) [23]. A literature search was performed from conception to July 2025 using the following databases [24]: (1) PubMed (MEDLINE), (2) Ovid (Embase), (3) Web of Science, and (4) Scopus. The search strategy is shown in Table 1. This review was registered in PROSPERO (CRD42021269567) [25] and included up-to-date studies until July 2025. A manual bibliography search was not performed. All studies identified by the search were imported into the Rayyan website and deduplicated [26]. Title, abstract, and full-text screening were independently performed by E.L. and W.Y.K.W. Disagreements were resolved by consensus or through discussion with a third reviewer (E.J.S.). 2.2. Inclusion and Exclusion Criteria We included eligible studies available in full-text English with clinicopathological, morphological, immunophenotypic, and cytogenetic features of DHL patients. Corresponding data from a comparison group, defined as any large B-cell lymphomas or HGBCL without double-hit (DH) translocations, were also extracted. The inclusion and exclusion criteria are shown in Table 2 using the Population, Intervention, Comparison, Outcome, Study type (PICOS) model as a guide [27]. 2.3. Data Extraction Data extraction was independently performed by E.L. and W.Y.K.W. A data extraction strategy was agreed upon amongst reviewers prior to data extraction. Discrepancies were resolved by consensus or through discussion with a third reviewer (E.J.S.). A template of the data extraction form is provided in Supplementary Table S2. Data were extracted from each included study onto an Excel spreadsheet, when available, including:Study characteristics and subject demographics.Clinical features: Age, Sex, Ann Arbor stage, International Prognostic Index (IPI) score, serum lactate dehydrogenase (LDH) level, Eastern Cooperative Oncology Group (ECOG) performance status (PS), extranodal involvement and affected sites, and the presence of B symptoms, BM, and CNS involvement.Histopathological features: immunophenotype (immunohistochemistry and flow cytometry), cell-of-origin (COO) (germinal center B-cell (GCB) versus non-germinal center B-cell-like (non-GCB)), and morphology.Cytogenetic and molecular features: fluorescent in situ hybridization (FISH) and Epstein–Barr virus (EBV) status. Study characteristics and subject demographics. Study characteristics and subject demographics. Clinical features: Age, Sex, Ann Arbor stage, International Prognostic Index (IPI) score, serum lactate dehydrogenase (LDH) level, Eastern Cooperative Oncology Group (ECOG) performance status (PS), extranodal involvement and affected sites, and the presence of B symptoms, BM, and CNS involvement. Clinical features: Age, Sex, Ann Arbor stage, International Prognostic Index (IPI) score, serum lactate dehydrogenase (LDH) level, Eastern Cooperative Oncology Group (ECOG) performance status (PS), extranodal involvement and affected sites, and the presence of B symptoms, BM, and CNS involvement. Histopathological features: immunophenotype (immunohistochemistry and flow cytometry), cell-of-origin (COO) (germinal center B-cell (GCB) versus non-germinal center B-cell-like (non-GCB)), and morphology. Histopathological features: immunophenotype (immunohistochemistry and flow cytometry), cell-of-origin (COO) (germinal center B-cell (GCB) versus non-germinal center B-cell-like (non-GCB)), and morphology. Cytogenetic and molecular features: fluorescent in situ hybridization (FISH) and Epstein–Barr virus (EBV) status. Cytogenetic and molecular features: fluorescent in situ hybridization (FISH) and Epstein–Barr virus (EBV) status. 2.4. Statistical Analyses Meta-analyses were conducted using RStudio version 5.5.1. For binary outcomes, a random-effects model for risk ratios (RR), between DHL and non-DHL cases and between MYC/BCL2 and MYC/BCL6 DHL, was implemented using the Mantel-Haenszel method within the metabin package [28]. The Paule-Mandel estimator for tau-squared, and Higgins and Thompson’s I-squared statistic were used to assess heterogeneity [29]. Pooled proportion effect sizes were calculated with the metaprop package [28] from logit transformed values of the raw proportions, which were used to build a generalized linear mixed-effects model. p-values < 0.05 were considered to be statistically significant. Forest plots were used for graphic presentation of meta-analyzed results. Comparison between RR and pooled proportions served as a sensitivity analysis. An available-case analysis approach was adopted for handling missing data from studies [30]. There is debate on when and how adjustments for multiple testing should be performed, with some viewing that exploratory analyses should be performed without multiplicity adjustment [31,32]. In this meta-analysis, we compared pre-defined clinicopathological features between DHL and non-DHL groups. These features had been pre-selected based on their previously reported use as proposed pre-screening features for FISH testing and constitute distinct clinicopathological features. Furthermore, each DHL versus non-DHL comparison was performed on different subsets of patients across the studies, reflecting variability in the reporting of clinicopathological features. Given these considerations, adjustment for multiple testing was not performed in this study. The Tau-squared statistic was used to estimate between-study heterogeneity, calculated with the maximum-likelihood estimator. Similarly, I2 values were calculated to measure effect size variance across studies, interpreted as low (0–25%), moderate (25–50%), substantial (50–75%), and critical (>75%) [29]. Meta-analyses were only conducted when at least three studies were available, with each study contributing at least three or more observations. For the comparison between MYC/BCL2- and MYC/BCL6 DHL, the latter was used as the control group. Non-meta-analyzed parameters were analyzed using descriptive statistics. Publication bias was determined by the degree of symmetry of funnel plots and Egger’s test, on parameters with greater than or equal to ten studies. 2.5. Risk of Bias Assessment Quality assessment was performed independently by E.L. and W.Y.K.W. using the National Institute of Health/National Heart, Lung, and Blood Institute (NIH/NHLBI) Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies and Case Series Studies. Disagreements were resolved by consensus or through discussion with a third reviewer (EJS). The summary of the overall results of the risk of bias assessment is presented in Supplementary Tables S3 and S4. 3. Results 3.1. Study Selection and Characteristics A total of 3379 studies were identified from the literature search, with 1921 remaining after deduplication for title and abstract screening (Figure 1). Eighty-eight studies were selected for full-text screening based on the inclusion and exclusion criteria. Twenty-eight retrospective studies and one prospective non-randomized study were included (Table 3) [11,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60]. A total of 773 DHL cases, of which 471 and 246 were classified as MYC/BCL2 and MYC/BCL6 DHL, respectively, were included. The median age of DHL patients ranged from 48.0 to 72.3 years. Males accounted for 14.3–100% of cases. Fourteen studies included comparative non-DHL groups, totaling 1237 patients. 3.2. Clinical Features Compared to non-DHL patients, DHL patients were more likely to exhibit higher-stage disease (RR 1.15, p = 0.028), higher IPI scores (RR 1.27, p = 0.047), and elevated LDH levels (RR 1.26, p = 0.012) (Figure 2). The presence of extranodal disease (RR 1.12, p = 0.078) and BM involvement (RR 1.34, p = 0.16) was also more likely in DHL patients, although these associations were not statistically significant. There was no significant difference between the presence of B symptoms between DHL and non-DHL patients (RR 0.90, p = 0.55). Pooled proportions of clinical features amongst DHL patients are shown in Table 4. Across eleven studies (374 patients), the pooled proportion of DHL patients with bone marrow (BM) involvement was 38.8%. Central nervous system (CNS) involvement in DHL patients, pooled from ten studies, was 14.8%. Disease localization was reported in seven studies, one of which included cases that all had oropharyngeal involvement [49]. The most common presenting sites were lymph nodes (26.5%), followed by the gastrointestinal tract (13.6%). Meta-analysis of CNS involvement was not performed between DHL and non-DHL due to insufficient studies reporting this in both groups. Compared to MYC/BCL6 DHL patients, MYC/BCL2 DHL patients were less likely to have extranodal disease (RR 0.69, p = 0.013) and CNS involvement (RR 0.43, p = 0.044). ECOG PS was lower in MYC/BCL2 than MYC/BCL6 patients, although not reaching statistical significance (RR 0.75, p = 0.21). There was no significant difference in disease staging, IPI score, the presence of LDH elevation, BM involvement, and B symptoms (Figure 3). Pooled proportions of clinical features amongst MYC/BCL2 and MYC/BCL6 patients are shown in Table 5 and Table 6. 3.3. Histopathological Features c-Myc protein expression was more likely in DHL than non-DHL patients (RR 1.41, p = 0.051) and approached but did not reach statistical significance. There were also no significant differences in BCL2, BCL6, CD10, and MUM1/IRF4 immunostaining (Figure 2). Based on the Hans classification for cell-of-origin (COO) determination, GCB phenotype was significantly more likely in DHL patients (RR 1.21, p = 0.043). Pooled proportions of immunohistochemical expression and COO of DHL patients are shown in Table 4. Proportions of other IHC (CD19, PAX5, CD79a, TdT) and flow cytometry markers (CD19, CD20, CD10, CD45, CD34, TdT, and light chain) are shown in Supplementary Tables S5 and S6. Compared to MYC/BCL6, MYC/BCL2 DHL patients were more likely to be of GCB phenotype (RR 1.40, p = 0.016) (Figure 3). CD10 immunopositivity (RR 1.46, p = 0.098) and MUM1 immunopositivity (RR 0.49, p = 0.095) were more likely in MYC/BCL2 DHL patients although not reaching statistical significance. BCL6 negative (RR 0.91, p = 0.23) and BCL2 positive (RR 1.34, p = 0.071) immunostaining was more likely in MYC/BCL2 DHL patients but this difference was not statistically significant. Expression of c-Myc protein and CD20 also did not differ between the two subtypes (Figure 3). Three of the five studies that evaluated Ki-67 proliferation index between DHL and non-DHL patients defined high Ki-67 expression as ≥90% [11,43,58], one used ≥80% [46] and the other ≥70% [50] (Table 7). No significant difference in Ki-67 expression levels between the two groups was observed. Out of the six papers which compared Ki-67 in the MYC/BCL2 versus MYC/BCL6 groups, three studies used a cut-off of ≥70% [38,48,50], one study used a cut of ≥90% [39], and two studies reported individual Ki-67 values [36,54] (Table 7). As most studies used a ≥70% Ki-67 positivity threshold or reported individual Ki-67 values to which this threshold could be applied, ≥70% was used as the positivity cut-off value for the MYC/BCL2 versus MYC/BCL6 Ki-67 comparison, with no significant difference observed between these two groups. DHL morphology was reported in 16 studies. Seven studies which preselected DHL based on morphology were excluded from this analysis. The remaining nine studies comprised 252 patient samples, which had undergone FISH testing to identify DHL cases and were subsequently evaluated for morphology. We recognize that the morphological subtypes of HGBL are poorly defined and demonstrate limited reproducibility. “DLBCL NOS” morphology describes morphologies that simply do not meet the diagnostic criteria for specific large B-cell lymphoma neoplasms [7]. Morphological classifications have also evolved over time. For example, “BCLU” introduced in the 2008 WHO classification, was removed in WHO-HAEM4, and reclassified as HGBCL, NOS, or HGBCL with MYC and BCL2 and/or BCL6 rearrangements [7]. Nonetheless “DLBCL” was the most common morphology (range 28.6–81.8%), followed by “BCLU” (range 0.0–66.7%) (Supplementary Table S7). biomedicines-14-01375-t004_Table 4Table 4Summary estimates (proportion as effect size, and pooled) of clinicopathological features of overall DHL cases. N /A = Not applicable. No. of StudiesNo. of CasesProportion [95% CI] (%)I2 Statistic [95% CI] (%)τ2 Likelihood Ratio Test (LRT)Egger’s Test p-Value Clinical features Ann Arbor Stage 3-42157175.9 [68.1; 82.3]45.8 [9.6; 67.5]0.344<0.00010.208IPI Score 3-51645667.2 [54.5; 77.7]69.9 [49.8; 82.0]0.628<0.00010.867Elevated LDH1341276.5 [65.2; 85.1]3.5 [0.0; 58.1]0.4820.00040.402Extranodal involvement1225567.1 [60.3; 73.2]0.0 [0.0; 58.3])00.1560.0071Extranodal involvement ≥2 sites631043.7 [31.6; 56.6]51.2 [0.0; 80.6]0.07850.0556N/ABM involvement1137438.8 [29.2; 49.4]48.1 [0.0; 74.1]0.1680.00600.247CNS involvement1025014.8 [10.4; 20.6]0.0 [0.0; 62.4]00.1600.814B symptoms613643.0 [27.4; 60.1]39.1 [0.0; 75.8]0.15540.0751N/AECOG PS 2-4712238.2 [19.8; 60.8]52.1 [0.0; 79.6]0.53550.0023N/A Immunophenotypic features BCL21336381.7 [74.3; 87.4]29.6 [0.0; 63.6]0.1670.03400.725BCL61431287.0 [80.0; 91.8]0.0 [0.0; 55.0]0.1890.02480.330c-Myc protein1021576.6 [55.3; 89.7]56.7 [12.2; 78.6]1.32<0.00010.990CD20611397.4 [85.5; 99.6]0.0 [0.0; 74.6]0.5910.170N/ACD101439581.2 [66.7; 90.3]74.7 [57.3; 85.0]1.156<0.00010.0510MUM11117251.8 [38.4; 64.9]27.4 [0.0; 64.1]0.2580.00730.287Ki-67 ≥70%913281.1 [69.5; 89.0]1.8 [0.0; 65.4]0.1040.197N/AKi-67 ≥90%413540.7 [18.2; 68.0]64.0 [0.0; 87.8]0.21730.028N/AGCB1949575.4 [65.5; 83.3]57.2 [28.6; 74.3]0.631<0.00010.719 Summary estimates (proportion as effect size, and pooled) of clinicopathological features of overall DHL cases. N /A = Not applicable. biomedicines-14-01375-t005_Table 5Table 5Summary estimates (proportion as effect size, and pooled) of clinicopathological features of MYC/BCL2 DHL cases. N/A = Not applicable. No. of StudiesNo. of CasesProportion [95% CI] (%)I2 Statistic [95% CI] (%)τ2Likelihood Ratio Test (LRT)Egger’s Test p-Value Clinical features Ann Arbor Stage 3-41432480.8 [71.8; 87.4]24.1 [0.0; 59.7]0.1530.02750.243IPI Score 3-51128570.5 [59.9; 79.4]32.4 [0.0; 66.8]0.1420.04660.505Elevated LDH925780.9 [74.6; 86.0]0.0 [0.0; 64.8]00.0356N/AExtranodal involvement512260.7 [48.0; 72.1]0.0 [0.0; 79.2]00.616N/AExtranodal involvement ≥2 sites419439.6 [15.3; 70.4]58.4 [0.0; 86.2]0.2020.0334N/ABM involvement825147.0 [39.7; 54.5]0.0 [0.0; 67.6]00.624N/ACNS involvement616110.9 [3.8; 27.6]0.0 [0.0; 74.6]00.105N/AB symptoms33647.2 [17.5; 79.0]34.5 [0.0; 78.7]00.133N/AECOG PS 2-465039.1 [17.2; 66.5]0.0 [0.0; 74.6]0.3870.0386N/A Immunophenotypic features BCL21124090.0 [84.8; 93.6]0.0 [0.0; 60.2]00.1820.319BCL61118786.4 [77.4; 92.1]0.0 [0.0; 60.2]0.08820.2080.601MYC812289.8 [53.7; 98.5]23.0 [0.0; 64.6]2.69170.0003N/ACD2044695.7 [68.8; 100.0]0.0 [0.0; 84.7]00.2929N/ACD101125992.3 [83.7; 96.6]28.3 [0.0; 64.7]0.5670.02890.996MUM187135.2 [23.2; 49.4]0.0 [0.0; 67.6]00.861N/AKi-67 ≥70%97779.4 [62.3; 90.0]0.0 [0.0; 64.8]0.2960.0936N/AGCB1438093.7 [87.1; 97.0]25.2 [0.0; 60.4]0.80120.00590.5925 Summary estimates (proportion as effect size, and pooled) of clinicopathological features of MYC/BCL2 DHL cases. N/A = Not applicable. biomedicines-14-01375-t006_Table 6Table 6Summary estimates (proportion as effect size, and pooled) of clinicopathological features of MYC/BCL6 DHL cases. N/A = Not applicable. No. of StudiesNo. of CasesProportion [95% CI] (%)I2 Statistic [95% CI] (%)τ2Likelihood Ratio Test (LRT)Egger’s Test p-Value Clinical features Ann Arbor Stage 3-41216776.0 [61.9; 86.1]39.7 [0.0; 69.5]0.4830.00390.496IPI Score 3-5911365.3 [40.9; 83.7]64.0 [26.3; 82.4]1.043<0.0001N/AElevated LDH79072.7 [49.3; 88.0]0.0 [0.0; 70.8]0.5420.0057N/AExtranodal involvement56874.7 [50.3; 89.6]0.0 [0.0; 79.2]0.06240.174N/AExtranodal involvement ≥2 sites37048.6 [25.2; 72.6]0.0 [0.0; 89.6]00.593N/ABM involvement67837.3 [14.2; 68.1]50.2 [0.0; 80.2]0.8270.006N/ACNS involvement57617.1 [8.1; 32.5]0.0 [0.0; 79.2]00.780N/AB symptoms45445.1 [20.3; 72.7]23.2 [0.0; 88.2]0.0290.220N/AECOG PS 2-442948.3 [22.2; 75.3]0.0 [0.0; 84.7]00.636N/A Immunophenotypic features BCL2511168.5 [55.2; 79.3]5.0 [0.0; 80.2]00.347N/ABCL6511292.0 [81.3; 96.8]0.0 [0.0; 79.2]00.139N/Ac-Myc protein48576.8 [40.4; 94.2]69.1 [10.6; 89.3]0.6110.0149N/ACD203561.0 [0.0; 1.0]0.0 [0.0; 89.6]01N/ACD10812754.4 [38.9; 69.0]19.0 [0.0; 61.7]0.1360.152N/AMUM179570.5 [58.0; 80.6]0.0 [0.0; 70.8]00.295N/AKi-67 ≥70%65383.1 [64.1; 93.1]0.0 [0.0; 74.6]0.01460.164N/AGCB1218860.0 [46.3; 72.2]0.0 [0.0; 58.3]0.3720.00130.219 Summary estimates (proportion as effect size, and pooled) of clinicopathological features of MYC/BCL6 DHL cases. N/A = Not applicable. biomedicines-14-01375-t007_Table 7Table 7Diagnostic criteria, FISH methodology, and key IHC thresholds used in each study. This table summarizes the diagnostic criteria of “DHL” according to (1) the 2008 WHO Classification, (2) WHO-HAEM4, and (3) WHO-HAEM5. 2008 WHO Classification: B-cell lymphoma unclassifiable, intermediate between diffuse large B-cell lymphoma and Burkitt lymphoma (BCLU), with MYC and BCL2 translocations [61]. WHO-HAEM4: all large B-cell lymphomas with MYC and BCL2 and/or BCL6 rearrangements except those fulfilling the diagnostic criteria for a follicular lymphoma or B-lymphoblastic leukemia/lymphoma (B-LBL) [62]. WHO-HAEM5: aggressive B-cell lymphomas with concurrent MYC and BCL2 rearrangements, with or without BCL6 rearrangement [7]. FP = fusion probe; BAP = break apart probe. N/A = Not available.StudyDiagnostic CriteriaFISH MethodologyFISH Probe Used, Yes (Y)/No (N)Threshold (%)IGH/MYC FPIGH/BCL2 FPMYC BAPBCL2 BAPBCL6 BAPc-MycBCL2BCL6Ki-67Wu et al., 2010 [33]2008 WHO classificationN/AN/AN/AN/AN/AN/AKobayashi et al., 2012 [34]2008 WHO classificationBCL6 break apart probe, IGH/MYC fusion probe, IGH/BCL2 fusion probeYYNNYN/AN/AN/AN/APerry et al., 2013 [35]2008 WHO classificationMYC, BCL6, IGK, IGL break apart probes, IGH/MYC fusion probe, IGH/BCL2 fusion probeYYYNYN/AN/AN/AN/AGebauer et al., 2013 [36]2008 WHO classificationMYC, BCL2, BCL6, IGH break apart probesNNYYYN/AN/AN/AValuesYoshida et al., 2015 [37]2008 WHO classificationMYC, BCL6 break apart probe, IGH/BCL2 fusion probeNYYNYN/A3030ValuesYe et al., 2016 [38]The applicable WHO classification between 1998 and 2010MYC, BCL2, BCL6, IGH break apart probes, IGH/MYC fusion probeYNYYYN/AN/AN/A70Landsburg, Petrich et al., 2016 [39]2008 WHO classificationN/AN/AN/AN/AN/A90Li et al., 2016 [40]WHO-HAEM4MYC, BCL6 break apart probes, IGH/BCL2 fusion probeNYYNY40503070Landsburg, Falkiewicz et al., 2016 [11]2008 WHO classificationN/AN/AN/AN/AN/A90Roth et al., 2016 [41]2008 WHO classificationN/AN/AN/AN/AN/AN/AMoench et al., 2016 [42]2008 WHO classificationMYC, BCL2, BCL6 break apart probes, IGH/MYC fusion probe, IGH/BCL2 fusion probeYYYYYN/AN/AN/AValuesOliveira et al., 2017 [43]2008 WHO classificationMYC, BCL2, BCL6 break apart probes, IGH/MYC fusion probeYNYYY50705090Miyaoka et al., 2018 [44]WHO-HAEM4MYC, BCL2, BCL6 break apart probes, IGH/MYC fusion probeYNYYY503030ValuesHancock et al., 2018 [45]WHO-HAEM4N/AN/AN/AN/AN/AN/AMa et al. 2019 [46]WHO-HAEM4N/AN/AN/AN/AN/A80Zhang et al., 2020 [47]WHO-HAEM4MYC, BCL6 break apart probes, IGH/MYC fusion probe, IGH/BCL2 fusion probeYYYNY405030N/AMa et al., 2020 [48]2008 WHO classificationMYC, BCL2, BCL6 break apart probesNNYYYN/AN/AN/A70Rodrigues-Fernandes et al., 2021 [49]WHO-HAEM4MYC, BCL6 break apart probes, IGH/BCL2 fusion probeNYYNY405030ValuesTsai et al., 2021 [50] WHO-HAEM4MYC, BCL2, BCL6 break apart probesNNYYY40503070Leskova et al., 2022 [51]WHO-HAEM4MYC, BCL2, BCL6 break apart probesNNYYYN/AN/AN/AN/AAlmeida et al., 2022 [52]The applicable WHO classification between 2010 and 2020MYC, BCL2, BCL6 break apart probesNNYYY405030N/AAlsuwaidan et al., 2022 [53]WHO-HAEM4MYC, BCL6 break apart probes, IGH/MYC fusion probe, IGH/BCL2 fusion probeYYYNY40N/AN/AN/ALiu et al., 2022 [54]WHO-HAEM4MYC, BCL2, BCL6, IGH break apart probes, IGH/MYC fusion probe, IGH/BCL2 fusion probeYYYYY405030N/AThirunavukkarasu et al., 2022 [55]WHO-HAEM4MYC, BCL2, BCL6 break apart probesNNYYYN/AN/AN/AN/ABlomme et al., 2024 [56]WHO-HAEM5MYC, BCL2, BCL6 break apart probes, MYC/BCL6 fusion probeNNYYYN/AN/AN/AN/AMarianini et al., 2024 [57]WHO-HAEM4N/AN/AN/AN/AN/AN/ATaybi et al., 2024 [58]The applicable WHO classification between 2013 and 2022MYC, BCL2, BCL6 break apart probesNNYYYN/AN/AN/A90Sebastian et al., 2024 [59]WHO-HAEM4N/AN/AN/AN/AN/AN/AKim et al., 2025 [60]The applicable WHO classification between 2012 and 2021MYC, BCL6 break apart probes, IGH/BCL2 fusion probe, MYC/BCL6 fusion probeNYYNY405030N/A Diagnostic criteria, FISH methodology, and key IHC thresholds used in each study. This table summarizes the diagnostic criteria of “DHL” according to (1) the 2008 WHO Classification, (2) WHO-HAEM4, and (3) WHO-HAEM5. 2008 WHO Classification: B-cell lymphoma unclassifiable, intermediate between diffuse large B-cell lymphoma and Burkitt lymphoma (BCLU), with MYC and BCL2 translocations [61]. WHO-HAEM4: all large B-cell lymphomas with MYC and BCL2 and/or BCL6 rearrangements except those fulfilling the diagnostic criteria for a follicular lymphoma or B-lymphoblastic leukemia/lymphoma (B-LBL) [62]. WHO-HAEM5: aggressive B-cell lymphomas with concurrent MYC and BCL2 rearrangements, with or without BCL6 rearrangement [7]. FP = fusion probe; BAP = break apart probe. N/A = Not available. Three B-lymphoblastic leukemia/lymphoma (B-LBL) cases across two studies were subsequently found to harbor MYC/BCL2 translocations [37,42]. B-LBL falls under the category of precursor lymphoid cells committed to the B-cell lineage, as opposed to DHL, which are mature B-cell neoplasms [7]. Two cases were excluded from the meta-analysis due to THL rearrangements [42], lack of testing for BCL6 rearrangement for possible THL [37], and absence of IHC or flow cytometry for terminal deoxynucleotidyl transferase (TdT), CD34, or CD20 to confirm precursor or mature B-cell states [37]. The third case highlighted a rare situation, whereby immunoprofiling showed an immature phenotype [37], a recognized phenotype of DHL in WHO-HAEM5 [7]. Meta-analysis of morphological features between MYC/BCL2 and MYC/BCL6 DHL was not performed, as most morphological data (90%) was available for MYC/BCL2 DHL cases only. 3.4. Cytogenetic and Molecular Features FISH methodology for each study is summarized in Table 7. MYC, BCL2, and BCL6 translocation partner genes were reported in 32, 12, and three cases, respectively, across five studies [34,35,42,44,54]. As expected, IGH was a more common translocation partner than non-IGH genes for MYC and BCL2, accounting for 30.0–85.7% and 85.7–100.0% of cases, respectively [34,35,42,44,54] (Supplementary Table S8). It should be noted that apart from IGH/MYC probes, break-apart, rather than dual-fusion probes, were used, which preclude the identification of non-IGH translocation partners, such as IGL and IGK. Likewise, the exclusive use of BCL6 break-apart probes and the overall paucity of data limited the interpretation of BCL6 translocation partners. We could not perform a meta-analysis of EBV status, as it was only reported in a few studies and was often missing in the non-DHL group. EBV status was reported in seven studies using in situ hybridization (ISH) and IHC (Supplementary Table S9) [36,42,47,48,49,50]. One study was excluded from the meta-analysis as it included a mixed group of DHL and THL in the control group [54]. The remaining six studies included 96 patients [36,42,47,48,49,50]. Only one study detected any EBV-positive DHL, at a frequency of 6.7% (3/45 patients) [47]. 3.5. Risk of Bias Risk of bias assessment was performed using the National Institute of Health/National Heart, Lung, and Blood Institute (NIH/NHLBI) Quality Assessment Tool. The overall quality was fair in most included studies, except in two, which were considered poor [42,59] (Supplementary Tables S2 and S3). Eleven studies did not clearly define their case selection criteria, potentially creating selection bias. There was likely a degree of information or reporting bias because the pathologists assessing the immunophenotype were not blinded to translocation status in most studies. A small degree of measurement bias was likely in a small number of studies that did not report the threshold used to define positivity for IHC markers. Most studies did not report measures to adjust for confounding factors. Funnel plots and Egger’s test showed no significant publication bias in most meta-analyzed parameters (Table 4, Table 5 and Table 6, Supplementary Figure S1). 4. Discussion This review aimed to collate and meta-analyze evidence from multiple studies to identify clinical and histopathological predictive features of DHL in HGBCL, which may inform patient stratification strategies for further FISH testing. Twenty-nine studies reporting routine diagnostic workup features for mature B-cell lymphomas and features proposed as a pre-screening strategy for FISH testing were included (Table 3). 4.1. Clinical Features DHL patients were associated with indicators of more aggressive disease, compared to non-DHL patients, including IPI scores and LDH elevation, which is consistent with prior studies [3,7,13,63,64]. Elevated IPI scores in DHL patients are thought to partly be attributable to extranodal involvement, including CNS and BM involvement [7,65,66]. We found that DHL was not significantly associated with extranodal involvement compared to non-DHL, although most DHL patients (67.1%) had extranodal disease. Insufficient numbers of studies reporting CNS involvement in DHL and non-DHL groups limited meta-analysis. Across ten studies, we found CNS involvement in 14.8% of DHL cases, which is within the reported range of 9–50% in previous studies [3]. BM involvement was present in 38.8% of DHL patients. We found a non-significant positive association between BM involvement and DHL compared with non-DHL patients. MYC/BCL6 DHL patients were notably more likely to have extranodal disease than those with MYC/BCL2 rearrangement. CNS involvement was also more likely in MYC/BCL6 compared to MYC/BCL2 DHL cases, which has not been reported previously. Some studies have shown an increased risk of CNS disease in non-GCB DLBCL [67,68]. It is possible that the increased risk of CNS involvement was related to the propensity of the non-GCB subtype in MYC/BCL6 DHL. Further studies that adjust for COO, a potential confounding factor, are needed to evaluate the association between MYC/BCL6 DHL and CNS involvement. 4.2. Histopathological Features We found that the GCB subtype was significantly more likely in DHL compared to non-DHL patients (RR 1.21, p = 0.043), in keeping with previously published findings [69,70]. This could be related to MYC/BCL2 being the predominant subtype of DHL, which we confirmed to be more likely of GCB subtype (RR 1.40, p = 0.016) compared to MYC/BCL6 using the Hans algorithm [16]. Interestingly, by individually considering COO determinants according to the Hans algorithm between the DHL and non-DHL groups, and between the two DHL subtypes, we identified non-significant trends for CD10 positivity, MUM1/IRF4 negativity, and BCL6 positivity. This may reflect small sample sizes when meta-analyzing the individual COO determinants, reducing statistical power to detect significant differences. We identified a near-significant trend for c-Myc protein positivity in DHL (RR 1.41, p = 0.051), which has previously been reported to predict MYC rearrangement with variable accuracies [71,72]. One reason why only a near-significant difference was observed may be due to the overlapping immunophenotypic profiles between DHL and other non-DHL HGBCLs. For instance, double-expressor lymphoma (DEL), which is more prevalent than DHL, is characterized by concurrent c-Myc and BCL2 IHC overexpression [2,73]. DEL inclusion in the non-DH HGBCL group may have attenuated the pooled effect size of c-Myc expression in DHL. There was no significant difference in c-Myc expression between the MYC/BCL2 and MYC/BCL6 DHL subtypes. Previously published studies have highlighted the limited utility of using high Ki-67 in predicting DHL [2]. We similarly observed a positive, non-significant association between increased Ki-67 and DHL compared with non-DHL (RR 1.36, p = 0.71). Limited availability of Ki-67 data in non-DHL comparison groups restricted meta-analysis to three studies using a ≥90% Ki-67 positivity threshold. Prior studies using the ≥90% Ki-67 threshold have reported missing a significant proportion of DHL [2]. Further studies using a lower ≥70% threshold, which was commonly applied in the studies evaluating Ki-67 between MYC/BCL2 and MYC/BCL6 DHL, may improve sensitivity for finding DHL cases among HGBCLs. The WHO-HAEM5 guidelines recommend a sequential strategy of assessing morphology first, followed by testing for MYC and BCL2 rearrangements when diagnosing DHL, recognizing that MYC/BCL2-rearranged HGBCL may arise within several morphological patterns (including DLBCL, NOS, Burkitt lymphoma, and HGBCL, NOS with blastoid or Burkitt-like features) [7]. Although we observed that the so-called “DLBCL morphology” was most common (range 28.6–81.8%), followed by BCLU (range 0.0–66.7%), we could not determine whether DHL was significantly associated with any specific morphology. This was partly confounded by variation in definitions and terminology for HGBCL morphology over time. Additionally, finding a clearly defined, non-DHL control group was challenging as we were limited to the non-DHL cases reported in the included studies. In these studies, non-DHL groups had been pre-selected based on morphology and, consequently, comparator groups comprised a variety of different non-DHL HGBCL with different morphologies and genetic aberrations. Morphology-controlled analyses between DHL and non-DHL cases were considered (for instance, by restricting analyses to “DLBCL, NOS” morphology only). However, as previously discussed, these morphological subtypes are poorly defined, demonstrate limited reproducibility, and, in the case of DLBCL, NOS, encompass heterogeneous entities, precluding morphology-controlled comparisons between DHL and non-DHL. Future prospective studies using the standardized morphological definitions (from WHO-HAEM5) with systematic FISH DHL verification may help confirm these morphological associations. However, as written, even these morphological definitions are likely to be challenging to apply with any consistency. It should be noted that a case initially classified as B-LBL based on morphology and immunoprofile was subsequently found to harbor a MYC/BCL2 translocation [42]. WHO-HAEM5 recognizes that around 2% of DHL cases have reported immature immunophenotypes (TdT immunopositivity, weak/negative CD20, negative BCL6, and/or light chains), which was seen here [7,42]. Importantly, this is distinct from B-LBL, a bona fide precursor lymphoid cell neoplasm, in which genetic abnormalities, including MYC/BCL2 rearrangements, constitute “desirable” exclusion criteria [7]. The overlapping features between DHL with an immature phenotype and B-LBL highlight a known diagnostic challenge and underscore the need for FISH testing to differentiate these two entities, where DHL is clinically suspected. Because cases of DHL with immature phenotype are rare and may require different treatment regimens compared to B-LBL [74], larger multi-center studies, beyond existing case reports, are needed to determine its clinical characteristics. Our results corroborate previously identified features but do not identify any new features to add to an algorithm. Our analysis indicates there is considerable overlap between DHL and non-DHL for many parameters and shows that testing has, to date, varied between centers and even between individual cases. In addition, the observed associations were of modest effect size, meaning features we have identified should only ever be used to identify which cases should be screened by FISH, rather than to predict their translocation status independent of FISH. Thus, there is a need for integrated, standardized molecular testing, most likely comprising a universal HGBCL MYC FISH test, followed by BCL2 FISH testing for MYC-rearranged cases to satisfy WHO-HAEM5 classification criteria, or BCL2 and BCL6 FISH testing for MYC-rearranged cases to satisfy ICC classification criteria [7,8]. The cost-effectiveness of such a universal FISH strategy will require analysis. While we considered a range of parameters, including currently assessed histopathological and immunophenotypic parameters, the implications of the application of artificial intelligence (AI) to digitally scanned histopathological images remain to be determined. Such an AI approach might identify predictive parameters, or combinations of parameters, which are not currently identified by pathologists. Ultimately, the identification of emerging targeted therapies, including additional-agent venetoclax and PROteolysis TArgeting Chimera (PROTAC) BCL6 degraders [75,76], will make identification of MYC/BCL2 and MYC/BCL6-rearranged lymphomas essential. 4.3. Strengths and Limitations Our study has several strengths. Through our literature search, we comprehensively incorporated evidence from multiple geographical regions and captured DHL cases in a more representative demographic context. We also applied stringent criteria to select only studies that performed MYC, BCL2, and BCL6 FISH testing to confirm DHL diagnosis. No significant uniform publication bias across the evaluated markers was shown, as demonstrated by relative symmetry in funnel plots. Furthermore, to our knowledge, this is the first meta-analysis evaluating the clinical, histological, and immunophenotypic features of DHL, including the MYC/BCL2 and MYC/BCL6 DHL subtypes. There are relatively few studies reporting MYC/BCL6 DHL due to its relatively low prevalence (8–20% of cases) and previous inclusion within the same category as MYC/BCL2 DHL or THL [3,77]. These results, however, need to be interpreted cautiously. While GCB was identified as a predictive feature, the reproducibility of immunohistochemically assigned COO compared with gene expression profiling is, itself, limited, with 72–83% concordance [78,79]. This is due to a combination of innate lymphoma biology, the challenges of interobserver variability, and different pre-analytical processing of the sample [16,79,80]. Furthermore, protein expression patterns of c-Myc, BCL2, and BCL6 can identify an additional group of HGBCLs above and beyond those identified by FISH, indicating that rearrangement-driven and expression-driven aggressive HGBCL are only partly overlapping groups [69]. Results should be interpreted with consideration of inter-study heterogeneity. We used a random effects model, and used I2 values (the percentages of total variation in effect estimates due to heterogeneity between studies rather than chance) to estimate statistical heterogeneity [29], which was low to moderate for several immunotypic marker comparisons (c-Myc, MUM1, GCB), but substantial to critical for other markers (BCL2, CD10, Ki-67). Similarly, tau2 values (estimating variance across all studies) were predominantly zero or low but higher for several markers (BCL2, CD10, Ki-67). Inter-study heterogeneity was likely driven by heterogeneity in the control groups used in different studies. As previously mentioned, we were limited by the non-DHL cases reported in the included studies, which included a variety of different non-DHL HGBCL, with different morphologies and genetic aberrations. These entities may have overlapping features with DHL, thus reducing the overall pooled effect size. It should also be noted that all but one of the included studies were retrospective in nature, hence were subject to bias in selection and the presence of missing data. Our handling of missing data, using the available-case analysis approach, may also introduce bias in effect estimates if data were not missing at random [30]. For instance, we were not able to comment on MYC translocation partners, although the presence of MYC rearrangements involving non-IGH versus IGH translocation partners is now known to have different prognostic implications [81]. Seven out of the 21 studies with available FISH methodology data solely used MYC, BCL2, and BCL6 break apart probes to identify MYC, BCL2, and BCL6 gene rearrangement. The remaining studies utilized IGH/MYC and/or IGH/BCL2 fusion probes, which will only identify translocation to IGH. Only five studies reported gene translocation partners (Supplementary Table S7). As routine interphase FISH is unable to detect some cryptic MYC or BCL2 rearrangements, MYC/BCL2 rearrangements will inevitably have been missed in a small subset of HGBLs [81]. In a small proportion of cases, complex karyotypes may also have proved difficult to interpret [64]. Results should also be interpreted with consideration of small-study effects due to the rarity of DHL cases. Small study sizes can result in wide confidence intervals around measured outcomes in individual studies and increase the uncertainty of the overall effect sizes when pooling different studies together. FISH has not historically been performed on all DLBCL or other HGBCL. Limited sample sizes also limited broader interpretation of results, including adjustment for potential clinical confounders such as patient age, sex, and geographic location. The WHO-HAEM5 guidelines, based on MYC/BCL2 translocation, should facilitate the development of larger, more consistently defined cohorts for future studies. Our meta-analysis also highlighted the lack of standardized IHC thresholds (Table 7). For the c-Myc protein, eight and two studies selected thresholds for labeling the case “positive” of 40% and 50% of cells, respectively. BCL2 protein expression cut-off values of 50%, 30%, and 70% of cells were selected in seven, two, and one studies, respectively, corroborating recognized variability in defining IHC positivity [2,82]. Establishing uniform inter-study thresholds, perhaps of 40% and 50% for c-Myc and BCL2, respectively, might improve reproducibility in IHC interpretation and enable more reliable comparisons between DHL and non-DHL, and between DHL subtypes. This is pragmatic because 40% and 50% thresholds for c-Myc and BCL2 are in common use for assessing so-called “double-expresser” and COO status [6,83]. Lastly, although this review identified clinical and pathological features associated with DHL, compared to non-DHL, amongst all HGBCLs, it is important to note that the purpose of identifying these associations is to help select a subset of patients for FISH testing to confirm the diagnosis of DHL, rather than attempting to make a diagnosis of DHL on the basis of predictive features without undertaking FISH. 5. Conclusions This review correlates DHL with features of a more aggressive clinical disease course, including a higher IPI score and LDH. We found GCB immunophenotype as a histopathological feature with statistically significant predictive value for DHL. When comparing MYC/BCL2 DHL predictors with MYC/BCL6 DHL predictors, we additionally identified extranodal disease, CNS involvement, and non-GCB immunophenotype. However, none of these features could identify 100% of DHL, nor completely distinguish between MYC/BCL2 and MYC/BCL6-rearranged cases. Individual features that may warrant further investigation include c-Myc immunostaining, which just failed to reach statistical significance. Further studies that adjust for COO, a potential confounding factor, are needed to evaluate whether CNS involvement can predict MYC/BCL6 DHL rather than MYC/BCL2 DHL. We also recommend standardizing IHC threshold values, perhaps utilizing existing widely adopted thresholds for c-Myc (40%), BCL2 (50%), and Ki-67 (≥70%). Such standardization will make the pooling of cohorts into larger studies more feasible, giving clearer answers to the questions that we tackled in this study.