Holding on to the Matutes score while dropping FMC7: new opportunity from standardised approaches in multiparameter flow cytometry
Year: 2020
Type: article
Abstract: Despite significant progress in cytogenetics and molecular biology, morphology and immunophenotype remain the 'gold standard' for the diagnosis of mature B-cell lymphoproliferative disorders (B-LPD).1 Chronic lymphocytic leukemia (CLL) represents the most frequent form of B-LPD and considering its specific phenotype, the British group from the Royal Marsden Hospital proposed in 1994 a scoring system based on the expression of five membrane markers, including FMC7, the Matutes score.2, 3 Other markers have been shown to contribute to the differential diagnosis between CLL and other CD5+ B-LPD: CD43, CD200 and more recently ROR-1.4, 5 With the development of multiparameter flow cytometry (MFC), new and wider panels for accurate diagnosis of B-LPD have been suggested. Cooperative groups like the EuroFlow consortium5 or the ERIC & ESCCA harmonization cooperative project4 did not select FMC7 as a required marker for CLL differential diagnosis in order to avoid redundancy and reduce the number of analyzed tubes. As FMC7 has been shown to be an epitope of CD20, both antibodies show mutual inhibition and should not be mixed in the same tube, requiring a specific tube to analyze FMC7 expression.6 According to sales figures from the public reagents market (personal and confidential data), one volume of the FMC7 antibody is sold for 1·38 volume of CD23, confirming that despite the addition of new markers in B-LPD panels, FMC7 was not suppressed and that Matutes score is still strongly implemented in practice. At a time when our cytometers are rising from eight to 10 to 12–13 colours, allowing combination of more informative antigens and limitation of redundancy between tubes, we have wondered if we could keep using the Matutes score replacing the FMC7 percentage (FMC7%) by the expression of CD20. This question has already been raised by others, with conflicting results,7,8 but that was before the implementation of standardised operating procedures (SOP) in flow laboratories.9 In the present work, we investigated the value of using CD20 expression instead of FMC7% in the Matutes score for CLL diagnosis in a context of SOP for multiparameter flow cytometry. A cohort of 508 patients with B-LPD issued from two centres (Ambroise Paré and Saint Louis, Paris, France) was included in this multicentre retrospective study. All blood samples were immunophenotyped according to EuroFlow SOP within the FranceFlow Quality Assessment Programme for the Lymphoid Screening Tube (LST) tube10 and to local procedures for the tube including FMC7, using eight-colour immunophenotyping panels on Canto II (BD Biosciences, San Jose, CA, USA; n = 458) and Lyric (BD Biosciences; n = 50) flow cytometers. Morphological, immunophenotypic, cytogenetic and molecular analyses, as well as clinical data, were collected to establish consensual integrated diagnoses according to the criteria of the World Health Organization (WHO) classification.1 Matutes score was calculated as previously published.3 All analyses were performed using Stata Statistical Software: release 14 (Stata Corp., LLC, College Station, TX, USA). Correlation matrix was performed using the R Corrplot package (R Foundation for Statistical Computing, Vienna, Austria). The clinical and biological characteristics of the study population are reported in Tables SI and SII, and Fig S1. We first analysed the correlation between CD20 expression, CD20 mean fluorescence intensity (MFI), FMC7%, FMC7 MFI and FMC7 ratio, i.e. the MFI of B cells normalised to the MFI of non-B lymphocytes, as cytometers were not harmonised for this parameter contrary to CD20. Overall, the best correlation was found between CD20 MFI and FMC7% (ρ = 0·620, P < 10−4), FMC7 MFI and FMC7 ratio showed lower correlations with CD20 MFI (ρ = 0·462 and ρ = 0·472, P < 10−4, respectively) Fig 1A. Using receiver operating characteristic (ROC) curves we compared the diagnostic performance of CD20 MFI, FMC7%, FMC7 MFI and FMC7 ratio. Surprisingly, CD20 MFI showed the highest diagnostic performance. Using the Youden index, the discriminating CD20 MFI threshold was defined as 6000. With this cut-off, the sensitivity for the diagnosis of CLL was 84·7% and specificity 94·9% (Fig 1B), the positive predictive value (PPV) and negative predictive value (NPV) were 95·1% and 84·1%, respectively. Using the classical FMC7 threshold of 30%,2 the sensitivity for the diagnosis of CLL was 80·4% and specificity 88·7% (Fig 1C), FMC7 MFI and FMC7 ratio showed lower performances (Figure S2). As our cytometers were not harmonised for FMC7 expression, we hypothesised that using FMC7 ratio rather than the raw MFI could give better results. This was not the case, illustrating well that standardisation of cytometers between centres and over time is crucial to get the most of the MFI. Interestingly, a small set of data issued from a Lyric cytometer showed fully comparable results with the whole cohort (data not shown). In the last part of the present work, we compared the Matutes score obtained as historically described and the Matutes score calculated using CD20 MFI instead of FMC7% with 1 point for the MFI values of <6000 (Table SIII). All non-CLL patients remained with a Matutes score of ≤3; in CLL, no decrease and an increase in 8·6% of patients were observed (Fig 2). The diagnostic performance of this new calculation of the Matutes score was then compared with that of the historical one and a score without FMC7. The ROC curves (Fig S3) confirmed that the Matutes score using CD20 MFI had equal performance to the Matutes score using FMC7%, with sensitivity remaining equal to 100% and specificity slightly increasing from 97·1 to 98·1%. Similarly, the NPV remained equal to 100% and the PPV increased from 96·7% to 98·0%. In all cases, the diagnostic performance was very high and the strict omission of FMC7/CD20 MFI could be discussed. Nevertheless, this omission would result in a modification of the threshold for CLL diagnosis, which could be very confusing for physicians. Due to worldwide implementation of the Matutes score in practice, we consider it preferable and easier to replace FMC7% by CD20 MFI in this score in the context of a standardised multiparameter flow cytometry approach. The authors declare no conflict of interest. Jaja Zhu analysed the data, performed the statistical analyses and wrote the paper. Anna Raimbault analysed the data and wrote the paper. Elise Sourdeau and Agathe Maillon collected and analysed the data, Stéphanie Mathis, Claude Capron, Pierre Lemaire and Emily Ronez collected the biological data. Hannah Moatti and Katayoun Jondeau collected the clinical data. Sylvain Clauser and Valérie Bardet designed the research study, analysed the data and wrote the paper. Table SI. Basic characteristics of the study population. Table SII. Distribution of FMC7 and CD20 on the cohort. Table SIII. Matutes score calculated as historically described (left part of each column) and using CD20 MFI instead of FMC7% (right part of each column in bold). Fig S1. Vioplots showing expression of CD20 MFI (A), FMC7% (B), FMC7 MFI (C) and FMC7 ratio (D) in atypical chronic lymphocytic leukemia (aCLL), chronic lymphocytic leukemia (CLL), diffuse large B-cell lymphoma (DLBCL), follicular lymphoma (FL), hairy cell leukaemia (HCL), mantle cell lymphoma (MCL), marginal zone lymphoma (MZL), splenic diffuse red pulp small B-cell lymphoma (SDRPL) and Waldenström macroglobulinaemia (WM). CLL patients (in blue) harbour a typical low expression of CD20 and FMC7. Fig S2. Receiver operating characteristic (ROC) curves for FMC7 MFI (A) and FMC7 ratio (B) and CLL diagnosis. FMC7 MFI and FMC7 ratio show lower performances (area under the curve [AUC] 0.82, 95% CI 0.79–0.86; and 0.83, 95% CI 0.80–0.87, respectively) for a threshold of 530 and 4.5, respectively compared to CD20 MFI and FMC7% (shown in Fig 1). Fig S3. Receiver operating characteristic (ROC) curves of Matutes score calculated using either A) FMC7% (area under the curve [AUC] 0.9994, 95% CI 0.999–1), B) CD20 MFI (AUC 0.9995, 95% CI = 0.999–1), or C) without FMC7 (AUC 0.996, 95% CI 0.993–0.999). For a threshold of 4 (A and B), sensitivity for the diagnosis of CLL was 100% (95% CI 98.4–100) and 100% (95% CI 98.5–100) for A and B, respectively. Specificity was 97.1% (95% CI 94.3–98.7) and 98.1% (95% CI 95.7–99.3) for A and B, respectively. For a threshold of 3 (C), sensitivity for the diagnosis of CLL was 97.6% (95% CI 94.8–99.1) and specificity 98.9% (95% CI 96.9–99.8). Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.
Source: British Journal of Haematology
DOI: 10.1111/bjh.16870
Language: English
Subfield: Hematology
Field: Medicine
Domain: Health Sciences
Open Access status: bronze
OpenAlex ID: W3036499950