Ancillary Studies in the Diagnostic Evaluation of Large B-Cell Lymphoma

An example of a DLBCL is presented in Figure 1, A through G. The core biopsy sample shows a cellular infiltrate composed of haphazardly arranged, large lymphocytes with elongated ovoid nuclei, speckled chromatin, and multiple small nucleoli. There are atypical mitoses and admixed small lymphocytes and histiocytes. Immunostains show that the lymphoma cells are positive for CD20 and negative for CD3. Flow cytometry revealed lymphoma cells with high forward scatter, expressing CD19, CD20, and λ light chain.

DLBCL, NOS is a mature B-cell neoplasm characterized by a proliferation of large lymphoid cells with variable heterogeneous cytology. Morphologic variants include centroblastic, immunoblastic, and anaplastic and are shown in Figure 2, A through C. The lymphoma cells are positive for B-cell markers including CD19, CD20, CD79a, and PAX-5. In the molecular era, it is a diagnosis of exclusion.

In 2019, the standard of care for ancillary studies in this differential diagnosis consists of immunophenotyping for conventional B-cell antigens to include reporting CD20 status and assessment of CD5; cell of origin studies; cytogenetic analysis for MYC and if positive, BCL2 and/or BCL6; and EBV in situ hybridization studies.³  Some authors additionally recommend Ki-67 and MYC and BCL2 immunohistochemistry on all cases.³  DLBCL, NOS expresses strong B-cell antigens typically; therefore, underexpression or lack of conventional B-cell markers may narrow the differential diagnosis to specific lymphomas (eg, primary effusion lymphoma or plasmablastic lymphoma). CD20 status is needed for anti-CD20 immunotherapy. CD5 expression should be evaluated to identify CD5⁺ DLBCLs, which may be seen in transformed chronic lymphocytic leukemia/small lymphocytic lymphoma (CLL/SLL) or de novo CD5⁺ DLBCL, an entity associated with aggressive clinical features, and to exclude the pleomorphic variant of mantle cell lymphoma. For this reason, CD5 positivity in a LBCL requires correlation with cyclin D1 immunohistochemistry and clinical history. The importance of other ancillary studies is further reviewed.

Three categories were originally defined on the basis of gene expression profiling, including germinal-center B-cell type (DLBCL-GC), activated/postgerminal center B-cell type (DLBCL-ABC), and unclassified.²  Based on testing accessibility, although imperfect, cell of origin testing is more commonly defined by surrogate immunohistochemistry algorithms, such as the Hans algorithm.⁴  According to the Hans algorithm, DLBCL-GC shows greater than 30% CD10 expression and/or BCL6 expression without evidence of MUM-1/IRF (<30%), and DLBCL-ABC shows greater than 30% staining of the postgerminal center marker MUM-1/IRF4, but lacks CD10 expression (<30%). Limitations of the Hans algorithm include an 87% positive predictive value for DLBCL-GC and 73% positive predictive value for DLBCL-ABC, no unclassified category, staining quality, interobserver variability, and failure to incorporate flow cytometry findings.⁵ 

These classifications represent biologically distinct entities and are associated with differences in survival. In general, DLBCL-GC is associated with a better prognosis with standard R-CHOP (rituxan, cyclophosphamide, doxorubicin, vincristine, and prednisone) chemotherapy and is more often associated with BCL2 rearrangements,^2,6  whereas DLBCL-ABC is associated with worse overall survival and with mutations of the activating B-cell receptor NF-κB pathways, including MYD88, CD79A, CARD11, and TNFAIP3.^6,7  Therefore, cell of origin testing should be performed on all DLBCLs, NOS, for use as a predictive marker. Designating cell of origin is also helpful in identifying patients who may benefit from available novel therapies. More recent therapeutic options, such as the use of lenalidomide, an immunomodulatory drug, have shown improved overall survival in DLBCL-ABC, virtually eliminating the negative prognosis.⁸ 

Returning to our initial presented example, as shown in Figure 3, A through C, this lymphoma is negative for CD10 and positive for BCL6 and MUM-1/IRF, consistent with a DLBCL-ABC.

The preferred assessment method for EBV in tissues is in situ hybridization for EBV-encoded small RNAs.⁹  This stain has largely replaced stains for Epstein-Barr virus nuclear antigen-2 and latent membrane protein-1 as standard of care. The number of neoplastic cells staining positive varies by lymphoma type, given the variable number of neoplastic large cells; however, in all of the EBV-associated LBCLs, most neoplastic large B-cells should stain positively.

As previously mentioned, several of the LBCL subtypes are associated with EBV, including EBV-positive DLBCL, NOS; DLBCL associated with chronic inflammation; lymphomatoid granulomatosis; EBV-positive mucocutaneous ulcer; immunodeficiency-associated lymphoproliferative disorders; plasmablastic lymphoma; and primary effusion lymphoma. (The latter 3 lymphoma subtypes show EBV positivity in most cases, but viral antigen expression is not required for diagnosis of these lymphomas.)² Given this, EBV-positive DLBCL is a diagnosis of exclusion. It represents a rare subtype of DLBCL, accounting for 1% to 4% of DLBCLs in the United States and 8% to 10% of DLBCLs in Asia and Latin America.²  It was renamed in the 2016 WHO classification from the previous term of EBV-positive DLBCL of the elderly. Most patients are older than 50 years when diagnosed and immunosenescence is thought to be the main risk factor. Histologically, there are variable numbers of large, atypical lymphoma cells, some of which may resemble Reed-Sternberg cells, within a variable background of inflammatory cells and necrosis (Figure 4). Lymphomas related to other immunodeficiency states such as transplant and immunomodulatory therapy are excluded from the EBV-positive DLBCL category.² 

EBV-positive DLBCLs are a separate category in the WHO classification because they represent a more clinically aggressive subtype with poorer overall survival than DLBCL, NOS.¹⁰  Compared to DLBCL, NOS cases, the EBV-positive DLBCLs show increased CD30 expression in lymphoma cells, are subgrouped as activated B-cell type, have higher rates of bone marrow involvement and higher international prognostic indices; moreover, patients tend to be older with higher performance statuses and more B symptoms.^10–12  Interestingly, younger patients (<45 years old), including children, tend to have T-cell/histiocyte-rich morphology for this entity, which has also been associated with a better prognosis.¹³  Such cases should be classified as EBV-positive DLBCLs according to the WHO and not T-cell/histiocyte-rich LBCLs.² 

It is important to recognize a histologically similar entity, namely, EBV-positive mucocutaneous ulcer (EBVMCU), and separate it from EBV-positive DLBCL, given significant differences in prognosis (Figure 5, A through D). Both lymphoproliferative disorders can have similar morphology, including a heterogeneous lymphoid infiltrate with variable neoplastic EBV-positive atypical cells, some of which may resemble Reed-Sternberg cells. Both entities can show sheets of large, atypical, EBV-positive cells, which can also be confused with classic Hodgkin lymphoma. EBVMCU occurs at mucosal sites, such as the oral cavity and gastrointestinal tract, as ulcerated lesions, and is usually seen in immunocompromised patients.^12,14,15  CD20 may be variably expressed in the neoplastic cells.¹⁶  Patients with EBVMCU usually respond to withdrawal of immunosuppression or anti-CD20 immunomodulatory therapy.^10,14 

MYC rearrangements can be identified in 5% to 15% of DLBCLs.⁶  There have been several studies looking at the effect of isolated MYC rearrangements on prognosis. In general studies have shown MYC rearrangements to be an adverse prognostic finding, including lower response to standard R-CHOP therapy.¹⁷  While there is a worse prognosis associated with MYC rearrangements in DLBCL, the prognosis is not as poor as when MYC is seen in association with either BCL2 or BCL6 rearrangement (“double-/triple-hit” lymphomas).³  Investigation on the rearrangement partner involved in MYC translocations has shown that MYC rearrangements involving immunoglobulin (Ig) genes (IgH, IgK [Ig kappa], or IgL [Ig lambda]) are more likely to be associated with this adverse prognostic risk than rearrangements involving other non-Ig gene partners.¹⁸  Currently most MYC fluorescence in situ hybridization (FISH) analyses use break-apart probes that do not identify the rearrangement partner, and given these findings the use of IgH/MYC, IgK/MYC, IgL/MYC fusion probes for prognostication may be beneficial to patients in the future. However, currently kappa and lambda immunoglobulin FISH probes are not routinely available at most institutions.

MYC protein expression has been identified in 30% to 50% of DLBCLs by immunohistochemistry, and coexpression with BCL2 at high levels (so-called double expressers) has been associated with an intermediate prognosis between DLBCL, NOS, lacking MYC/BCL2 protein expression and DLBCL with 2 or more gene rearrangements involving MYC, BCL2, and BCL6 (see High-Grade B-Cell Lymphoma below).⁶  Double-expresser DLBCL is defined by many investigators as having more than 40% of cells that are MYC⁺ and more than 50% to 70% of cells that are BCL2⁺, although a standardized algorithm for exact cutoffs has not been established.¹²  Associations with an increased risk of central nervous system relapse and lower rates of overall survival have been reported.^17,19  Given the lack of a standardized algorithm to define this double-expresser subgroup, the clinical utility of MYC and BCL2 expression by immunohistochemistry is uncertain and likely depends on institutional validation.¹¹ 

While polymerase chain reaction (PCR) clonality assessment is not routinely required for the diagnosis of most DLBCLs, a pitfall in this type of analysis includes the presence of clonal T-cell receptor (TCR) rearrangements in up to 33% of DLBCLs due to lineage infidelity.²⁰  Therefore, caution should be used when interpreting the results of TCR and IgH PCR clonality testing and should not be used in isolation to assign lineage.

Most DLBCLs can be identified by flow cytometry, though sample size and tumor viability may cause poor recovery.²¹  Lymphoma cells show expression of CD19, CD20, CD22, and bright CD45 most commonly. Most DLBCLs show monotypic light chain expression by flow cytometry; however, nearly a quarter of cases lack both κ and λ surface staining despite multiple light-chain antibody tubes.^7,22  CD5 and CD10 expression are observed in 5% to 10% and 30% to 60% of DLBCLs, respectively. Transformed CLL/SLL cases show similar immunophenotypic characteristics to CLL/SLL.²³ 

In the 2016 WHO classification, 2 new, cytogenetically defined categories were introduced that have overlapping morphologic and immunophenotypic features with DLBCL, NOS. A brief discussion of each type is presented.

This entity replaces the so-called aggressive double-/triple-hit lymphomas, which were not previously categorized in the 2008 WHO classification. These lymphomas have variable cytology and are characterized by gene rearrangements of MYC in association with either a BCL2 or BCL6 gene rearrangement or both, with the exclusion of high-grade follicular lymphomas and lymphoblastic lymphomas (Figure 6, A and B).⁶  Routine FISH analysis is the most sensitive method for identifying most of these rearrangements and is superior to conventional cytogenetics owing to higher sensitivity and its ability to be performed on either fresh or formalin-fixed, paraffin-embedded tissue.³  Studies looking at MYC and/or BCL2 immunohistochemistry have shown a correlation between higher protein expression and the presence of MYC and BCL2 gene rearrangements; however, immunohistochemistry does not consistently identify MYC- and/or BCL2-rearranged lymphomas and should not be used as a surrogate to FISH analysis.^3,6  In general, co-occurring MYC and BCL2 rearrangements are seen more commonly in DLBCL-GC, and co-occurring MYC and BCL6 gene rearrangements are more commonly seen in DLBCL-ABC.³  Although the above associations are noted, many institutions perform initial MYC FISH analysis on all DLBCL, NOS cases with reflex testing for BCL2 and BCL6 if positive. These lymphomas may show variable morphology, including Burkitt-like features, and are generally associated with a poor prognosis with low overall survival.²  Of note, relapsed DLBCL may benefit from FISH analysis for these gene rearrangements as well, given their negative impact on prognosis in this clinical setting.¹⁷ 

This is a new category in the 2016 WHO classification that defines a group of predominantly pediatric large B-cell lymphomas of germinal center B-cell origin that show strong expression of MUM-1/IRF4 by immunohistochemistry and are characterized by IRF4 gene rearrangements (Figure 6, C and D).^2,6  IRF4 is a transcription factor located on the short arm of chromosome 6. The most common translocation partner is the immunoglobulin heavy chain; however, other translocations have been observed.²⁴  LBCL-IRF4 are rare and often present as a solitary enlarged lymph node in the head and neck, including involvement of Waldeyer ring. These lymphomas can show variable growth patterns, including diffuse patterns resembling DLBCL, NOS, and follicular patterns, that resemble high-grade follicular lymphomas; they demonstrate a germinal center B-cell immunophenotype, including BCL6 and often CD10 expression. Given the overlapping age range, clinical presentation, and morphology, LBCL-IRF4 must be differentiated from pediatric follicular lymphoma. Pediatric follicular lymphoma lacks IRF4 expression, shows enlarged serpiginous follicles, and can be managed with local excision in most cases. However, LBCL-IRF4 show worse prognosis and require management with chemotherapy with or without radiation.²  Although LBCL-IRF4 have a worse prognosis than pediatric follicular lymphomas, they generally respond well to therapy and behave better than follicular lymphomas in adults that are IRF4⁺ but lack the defining gene rearrangement.⁶  Therefore, it is recommended that large B-cell lymphomas that show coexpression of CD10, BCL6, and IRF4 by immunohistochemistry be evaluated for IRF4 gene rearrangement by FISH.³