Concordance between flow cytometry CLL scores

Abstract

Introduction Multiple flow cytometry scores/diagnostic systems for the classification of leukemic lymphoproliferative disorders (LPD) have been published but few have been compared between them. Patients and Methods We classified a cohort of leukemic LPD based on eleven published flow cytometry scores/diagnostic systems and compared their classification as chronic lymphocytic leukemia (CLL) or non‐CLL LPD. Results 329 patients were included. Patients classified as CLL ranged from 46% to 73%, depending on the score/diagnostic system used. All eleven scores/diagnostic systems agreed in 184/324 (57%) of patients while in 58/324 (18%) at least two scores/diagnostic systems classified the patient differently (from the majority). Fleiss kappa was 0.74, but pairwise agreement was variable (Cohen's kappa: 0.48 to 0.87). Conclusion This study found a suboptimal agreement between published flow cytometry scores/diagnostic systems for the classification of LPD.

Treatment of Chronic Lymphocytic Leukemia in the Personalized Medicine Era

“…Exploring non-score-based systems, such as the “full phenotype” system, appears to be a worthwhile consideration. Finally, the study underscores the importance of evaluating the reproducibility of the integrated diagnosis of leukemic LPD, encompassing not only flow cytometry but also cytology, cytogenetics, and molecular biology [ 57 ].…”

Flow cytometry in leukaemic B cell lymphoproliferative disorders. New scores, same old concerns

“…A nonstructured search revealed 10 proposals including commonly used markers (seven of them published in or after 2016). 1 Additionally, there is a larger number of them that include less commonly used markers or that evaluate the diagnostic performance of a single marker. In May 2021, we analysed the agreement between the 10 aforementioned classification proposals and we reported suboptimal agreement between them, with an overall kappa of 0.74 and pair-wise kappa indices between 0.48 and 0.87.…”

“…In May 2021, we analysed the agreement between the 10 aforementioned classification proposals and we reported suboptimal agreement between them, with an overall kappa of 0.74 and pair-wise kappa indices between 0.48 and 0.87. 1 This datapoint may be surprising at first, but it is perhaps less so when placed in light of the fact of how discordant the diagnostic performance of the classic Moreau score-the modified Matutes score (mMS)-is when applied by different groups or investigators. This score, published in 1997, 2 is reported to have a sensitivity and specificity for the diagnosis of chronic lymphocytic leukaemia (CLL) ranging from 0.8 to 1 and 0.54 to 1 (Table 1), respectively.…”

“…We have long defended the importance of finding a true, objective, reference standard that defines CLL that leads the community to greater inter-observer agreement. 1 However, whilst this eludes us, we call on investigators to provide external rather than the internal, almost perfunctory, validation efforts that have become common in these types of studies. We also call on journal editors and reviewers to ensure that rigorous methods sections are provided and adequate statistical review is applied.…”

Overlap between CD23 and CD200 in leukemic lymphoproliferative disorders

“…We have previously made a case against the use of FC scores based on a number of considerations, including the loss of information due to dichotomization of biological variables, awarding a similar number of points for features with likely different relevance, inter-observer interpretive differences and the idea that scores reduce an entire phenotype to a single value. 9,10 We favor a relatively comprehensive phenotypic analysis 10,11 and are not persuaded by cost arguments, given the small cost of each additional antibody when overwhelmingly more expensive molecular biology testing is becoming common and when increasing therapeutic options making accurate biologic characterization of disorders ever more important. However, if a score was to be used, current data seem to suggest that two markers are sufficient in most instances and that only one of CD200 or CD23, likely the former, is required.…”