Real-world impact of differences in the WHO and ICC classifications of non-Hodgkin lymphoma: a LEO cohort study analysis Free

TO THE EDITOR:

The use of standardized, evidence-based classification systems is crucial for the accurate diagnosis and treatment of hematological diseases. Moreover, standardized classification facilitates consistency in clinical trial enrollment, interpretation of translational research findings, epidemiologic studies, and communication among health care professionals. From 2016 to 2022, the revised fourth edition of the World Health Organization (WHO) classification of hematopoietic neoplasms (WHO-HAEM4R) was the global standard for the diagnosis and classification of lymphoid malignancies.¹ In 2022, two new classification systems were published: the fifth edition of the WHO classification (WHO-HAEM5) and the International Consensus Classification (ICC).^2,3 Both WHO-HAEM5 and ICC reflect the emergence of new data and maintain a shared fundamental concept of disease classification that integrates clinical, pathologic, and molecular data. However, the 2 classifications differ on nomenclature, establishment of new entities, and/or diagnostic criteria for some lymphoma subtypes. The extent of the challenges associated with the existence of 2 lymphoma classifications is an unresolved topic of debate at many centers. Here, we evaluated the anticipated impact of differences between WHO-HAEM5 and ICC on the real-world diagnosis of non-Hodgkin lymphoma (NHL) based on the classification of NHLs in the Lymphoma Epidemiology of Outcomes (LEO) cohort study (NCT02736357).

Initiated in 2015, LEO is an ongoing prospective observational study of patients aged ≥18 years with newly diagnosed NHL. Patients are enrolled at 8 major US medical centers, with a distribution of NHL subtypes similar to Surveillance, Epidemiology and End Results Program (SEER) data.⁴ Clinical, epidemiologic, pathologic, and treatment data are abstracted at baseline, and active follow-up is conducted for all patients. To classify lymphoma subtype, each case undergoes expert hematopathology rereview of the diagnostic pathology slides, pathology report, and biomarker data. The clinical diagnosis is abstracted from the medical record, and any ambiguity or discrepancy between the coding of the clinical and pathological diagnoses is resolved by consensus review between the hematopathologist and hematologist/oncologist at each site. Here, we studied all LEO patients diagnosed between 1 June 2016 (when LEO adopted WHO-HAEM4R) and 13 February 2023. The study was conducted with institutional review board approval, in accordance with the Declaration of Helsinki. WHO-HAEM4R diagnoses and additional clinicopathologic data were reviewed by 3 LEO hematopathologists (B.A., D.L.J., and A.L.F.) to assign cases by consensus to the corresponding WHO-HAEM5 and ICC diagnoses. Differences between WHO-HAEM5 and ICC were annotated for each case as follows: none, minor (nomenclature difference only), major (difference in disease definitions, resulting in patients being categorized differently in the 2 classifications), or unevaluable (insufficient data for classification). Similarly, we annotated differences between WHO-HAEM4R and WHO-HAEM5 and between WHO-HAEM4R and ICC.

LEO enrolled 6143 patients during the study period. Comparisons among WHO-HAEM4R, WHO-HAEM5, and ICC were evaluable in 5702 (93%; Table 1; supplemental Table 1, available on the Blood website). Unevaluable cases included those without a specific WHO-HAEM4R NHL diagnosis (n = 412; eg, low-grade B-cell lymphoma that could not be subclassified) and those with a WHO-HAEM4R diagnosis but with insufficient data to assign WHO-HAEM5/ICC diagnoses (n = 29). Of the evaluable cases, 5328 (93.4%) showed no difference between WHO-HAEM5 and ICC, 331 (5.8%) showed minor differences (supplemental Table 2), and 43 (0.8%) showed major differences (Table 2). Twenty cases with major differences (46.5%) were B-cell lymphomas with MYC and BCL6 rearrangements; 21 (48.9%) were related to classification of splenic/leukemic B-cell lymphomas; and 2 (4.6%) involved classification of B-cell lymphomas occurring at specific anatomic sites. Major differences were observed between WHO-HAEM4R and WHO-HAEM5 in 128 cases (2.2%) and between WHO-HAEM4R and ICC in 91 (1.6%; supplemental Tables 3 and 4). Most differences between WHO-HAEM4R and the 2 newer classifications reflected revisions in the diagnostic criteria for B-cell NHLs, with fewer differences for peripheral T-cell lymphomas (PTCLs).

In a large, prospective lymphoma cohort reflecting real-world clinical practice at 8 major US medical centers, major diagnostic differences in the classification of NHL using WHO-HAEM5 or ICC classification criteria were seen in 0.8% cases, whereas 99.2% of diagnoses were the same or showed differences in nomenclature only. This finding does not negate the concern that the existence of 2 concurrent classification systems presents potential for discrepancies in pathologic diagnosis, clinical practice, clinical trials, cancer registries, and other lymphoma research. Nevertheless, our findings predict that the proportion of patients with NHL affected will be small in real-world practice and public health settings, in part because major differences between the 2 classifications predominantly affect very rare entities, whereas there is general concordance on the most common NHL subtypes. Among newer B-cell entities, the most common diagnoses that reflect this concordance are as follows: the recognition of primary large B-cell lymphoma of the testis as an entity distinct from diffuse large B-cell lymphoma, not otherwise specified (NOS)^9,10; the distinction of primary cutaneous marginal zone lymphoma from other extranodal marginal zone lymphomas^11,12; and the development of diagnostic criteria for mediastinal gray zone lymphoma as a new entity.^13,14 Both classifications discontinue the WHO-HAEM4R category of “high-grade B-cell lymphoma with MYC and BCL2 and/or BCL6 rearrangements” based on data showing clinicopathological differences between BCL2 and BCL6 rearranged cases.^15-18 However, WHO-HAEM5 regrouped cases with MYC and BCL6 rearrangements into either diffuse large B-cell lymphoma, NOS or high-grade B-cell lymphoma, NOS (depending on morphology), whereas ICC created a new provisional entity for these cases.^2,19 The largest group of cases that reflected different approaches by the 2 classification systems were the splenic and leukemic B-cell lymphomas, in which the creation of splenic B-cell lymphoma/leukemia with prominent nucleoli in WHO-HAEM5 results in the reclassification of a number of tumors, including hairy cell leukemia variant, that remain largely unchanged in ICC.^3,20 This difference could affect management in some cases, because hairy cell leukemia variant typically does not respond well to single-agent rituximab, which is still used initially for other splenic B-cell neoplasms included under splenic B-cell lymphoma/leukemia with prominent nucleoli.^5,21 Our data set did not allow for the evaluation of all differences in subclassification and grading of follicular lymphoma (eg, BCL2 rearrangement (BCL2-R)-negative and CD23⁺ follicle center lymphoma in ICC),²² and further study is required to assess their impact. Few PTCLs were identified in any of the comparisons, because they are less frequent than B-cell lymphomas. Both WHO-HAEM5 and ICC recognize an Epstein-Barr virus-positive form of nodal PTCL as a new entity^23,24; however, PTCLs for which the 2 classifications differ are rare and were not observed in our cohort. Because of LEO enrollment criteria, our cohort did not include Hodgkin lymphoma or patients aged <18 years. Pediatric NHLs are relatively infrequent and are unlikely to contribute significantly to the overall rate of diagnostic discrepancies. An interesting conceptual difference is grouping nodular lymphocyte predominant Hodgkin lymphoma (WHO-HAEM4R) with the NHLs in ICC (as nodular lymphocyte predominant B-cell lymphoma), based on clinicopathologic, genetic, and biologic similarities to NHLs.^25,26 It remains nodular lymphocyte predominant Hodgkin lymphoma in WHO-HAEM5. Although diagnostic criteria remain the same, this classification change may affect clinical trial eligibility and epidemiologic studies that depend on consistency in the distinction between Hodgkin lymphoma and NHL. Similarly, it is possible that, with the refinement of new diagnostic criteria, some cases previously diagnosed as classic Hodgkin lymphoma might be reclassified as mediastinal gray zone lymphoma, and further studies are warranted.

In summary, both WHO-HAEM5 and ICC introduce a number of changes to lymphoma classification that are largely concordant and reflect advancing knowledge in the field. Although the 2 systems have real differences in their approaches to some entities, these tend to affect mostly rare lymphoma types. In the real-world clinical practice setting, the coexistence of 2 classification systems will likely affect <1% of NHL diagnoses and pose only infrequent clinical management challenges, although treatment and outcome data for the small number of new entities are needed. Recognition of common ground and additional data on rare entities that are now controversial should facilitate further discussion and return to a single classification system in the future.