A novel scoring system combining expression of CD23, CD20, and CD38 with platelet count predicts for the presence of the t(11;14) translocation of mantle cell lymphoma


  • Patrick G. Medd,

    1. Department of Haematology, Oxford Radcliffe Hospitals NHS Trust, John Radcliffe Hospital, Headington, Oxford, OX3 9DU, United Kingdom
    Search for more papers by this author
  • Nithiya Clark,

    1. Department of Haematology, Oxford Radcliffe Hospitals NHS Trust, John Radcliffe Hospital, Headington, Oxford, OX3 9DU, United Kingdom
    Search for more papers by this author
  • Kevin Leyden,

    1. Department of Haematology, Oxford Radcliffe Hospitals NHS Trust, John Radcliffe Hospital, Headington, Oxford, OX3 9DU, United Kingdom
    Search for more papers by this author
  • Susan Turner,

    1. Department of Haematology, Oxford Radcliffe Hospitals NHS Trust, John Radcliffe Hospital, Headington, Oxford, OX3 9DU, United Kingdom
    Search for more papers by this author
  • Jennifer A. Strefford,

    1. Department of Cytogenetics, Oxford Radcliffe Hospitals NHS Trust, Churchill Hospital, Headington, Oxford, OX3 7LJ, United Kingdom
    Search for more papers by this author
  • Caroline Butler,

    1. Department of Haematology, Oxford Radcliffe Hospitals NHS Trust, John Radcliffe Hospital, Headington, Oxford, OX3 9DU, United Kingdom
    Search for more papers by this author
  • Graham P. Collins,

    1. Department of Haematology, Oxford Radcliffe Hospitals NHS Trust, John Radcliffe Hospital, Headington, Oxford, OX3 9DU, United Kingdom
    Search for more papers by this author
  • David J. Roberts,

    1. NHS Blood and Transplant Oxford Centre, John Radcliffe Hospital, Headington, Oxford, OX3 9DU, United Kingdom
    Search for more papers by this author
  • Wale Atoyebi,

    1. Department of Haematology, Oxford Radcliffe Hospitals NHS Trust, John Radcliffe Hospital, Headington, Oxford, OX3 9DU, United Kingdom
    Search for more papers by this author
  • Chris S. R. Hatton

    Corresponding author
    1. Department of Haematology, Oxford Radcliffe Hospitals NHS Trust, John Radcliffe Hospital, Headington, Oxford, OX3 9DU, United Kingdom
    • Department of Haematology, Cancer and Haematology Centre, Churchill Hospital, Old Road, Headington, Oxford, OX3 7LJ, United Kingdom
    Search for more papers by this author

  • How to cite this article: Medd PG, Clark N, Leyden K, Turner S, Strefford JA, Butler C, Collins GP, Roberts DJ, Atoyebi W, Hatton CSR. A novel scoring system combining expression of CD23, CD20, and CD38 with platelet count predicts for the presence of the t(11;14) translocation of mantle cell lymphoma. Cytometry Part B 2011;80B: 230–237.



Both chronic lymphocytic leukemia (CLL) and mantle cell lymphoma (MCL) are CD5/19 positive. The t(11;14) MCL translocation is identified by fluorescent in situ hybridization (FISH) and can distinguish the two disorders. We attempted to identify flow cytometric and other markers predictive of a positive FISH test.


We examined 100 atypical CLL/MCL cases for demographic, hematological, and cytometric variables, 96 were FISH tested for t(11;14) and four were known MCL.


Twenty-two cases were confirmed as MCL. Multivariate analysis identified four variables associated with MCL: Thrombocytopenia (taken as Plt < 150 × 109/L), CD23 negative, CD20 strong, and CD38 positive, with these variables a four-point score was devised. By ROC analysis, the MCL score was superior in differentiating MCL to the Marsden CLL score (AUC 0.95 vs. 0.78). MCL score ≥2 showed sensitivity 1, specificity 0.66, positive predictive value (PPV) 0.49, and negative predictive value (NPV) 1 for MCL. The score was then prospectively validated on an independent cohort of 44 cases of atypical CLL/MCL. No MCL had a score <3. Validation PPV/NPV of score ≥3 were 0.5/1. Overall survival in MCL was shorter compared to t(11;14) negative patients (median 3.3 vs. 4.2 years, HR 2.2, 95% CI 0.87–5.5, P = 0.1).


The score described can be used to identify cases of CD5/19 positive lymphoproliferative disorders likely to be t(11;14) positive MCL. © 2011 International Clinical Cytometry Society

Immunophenotyping by flow cytometry is central to the diagnosis of lymphoproliferative disorders (LPD). The commonest LPD is chronic lymphocytic leukemia (CLL), which characteristically expresses CD5 and CD19 on the cell surface. The differential diagnosis of a CD5/19 dual positive LPD lies between CLL and mantle cell lymphoma (MCL) (1–4). MCL has a significantly poorer prognosis than CLL, often requiring more aggressive treatment (5); thus it is of clinical importance to distinguish the two.

Classical CLL is easily distinguished from MCL on immunophenotype alone, as it usually expresses CD23, has weak cell surface expression of immunoglobulin (sIg), and lacks expression of FMC7 and CD22. These markers form the basis of the CLL score developed by Matutes et al. (1) and subsequently revised (with replacement of CD22 by CD79b) by Moreau et al. (6). In contrast, MCL most commonly has the phenotype CD5+, CD23−, FMC7+, CD79b+, and sIg strong (7, 8). In addition CD20 expression is strong in MCL and weak in CLL when compared to normal B lymphocytes (9). While the revised CLL score enables the identification of the majority of classical CLL, there remains a grey area of ‘atypical CLL’ that scores less than 4 on the revised CLL score (10). It has been suggested that absence of the low affinity immunoglobulin E receptor CD23 precludes the diagnosis of CLL and indicates MCL (8, 11, 12). These data, however, were largely extrapolated from immunohistochemistry on paraffin embedded tissue sections. With flow cytometry CD23 positive MCL is a much more frequent finding: estimates of prevalence range from 15% to 55% (13–16). The difference between CD23 expression on MCL in lymph nodes and that in circulation may also be because of a variation in LPD behavior. Some data suggest that CD23 positive MCL may be a biologically distinct condition with a greater tendency towards leukemic phase (16) and less aggressive clinical course than CD23− MCL (14). In addition to CD23+ MCL, cases of CLL clearly negative for CD23 are also well described (10, 17). Thus CD23 alone cannot be relied upon to separate these two conditions. Indeed variation from the classical immunophenotype may be present in up to 40% of MCL (18).

The primary genetic event leading to the development of MCL is overexpression of the cyclin D1 gene (CCND1) (19). This is nearly always because of the t(11;14)(q13;q32) chromosomal translocation, which places the gene under the control of the Ig heavy chain promoter (20). This translocation is present in the vast majority of MCL (2, 21–24) and can be readily detected by fluorescent in situ hybridization (FISH). FISH appears to offer the greatest sensitivity for detection of the underlying genetic driver of MCL (21, 24) and can therefore be regarded as approaching a gold standard for diagnosis of MCL. A study of 28 cases with morphology and immunophenotype typical for MCL demonstrated that while 57% did possess t(11;14) on FISH, 32% harbored translocations associated with the diagnosis of CLL (17). These cases—indistinguishable on morphology and conventional flow cytometry—indicate that there is clear overlap between MCL and atypical CLL phenotypes. In this paper, we analyze the hematological and immunophenotypic characteristics of cases of atypical CLL/MCL referred for FISH t(11;14) testing to try and identify variables associated with the MCL translocation.



The derivation cohort comprised patient peripheral blood (n = 90) and bone marrow (n = 10) samples referred for morphology and immunophenotyping between January 2005 and December 2007 inclusive. Samples were considered consistent with a diagnosis of CLL or MCL if they demonstrated dual positive expression of surface CD5 and CD19. Samples considered atypical for CLL on the basis of morphology and immunophenotype were referred for FISH t(11;14) testing. In two cases, CD5 negative samples were also considered possible atypical CLL/MCL. In addition cases of known MCL, previously diagnosed on the basis of t(11;14) FISH and/or bone marrow/lymph node immunohistology underwent immunophenotyping during this period. These cases were included with t(11;14) positive cases as confirmed MCL. The validation cohort comprised consecutive cases of CD5/19 dual positive atypical CLL referred for immunophenotyping between January 2008 and June 2009.

Flow Cytometry Immunophenotype Analysis

Whole blood or bone marrow samples were subjected to flow cytometric immunophenotype analysis. A combination of two- and three-color immunophenotype analyzes were performed using a FACSCalibur flow cytometer (BD Biosciences, Franklin Lakes, NJ, USA). All patients had full blood counts performed, and peripheral blood or bone marrow smears were made for morphological assessment under May-Grunwald-Giemsa staining.

Red blood cells were lysed with 10% (w/v) ammonium chloride for 10 minutes in the dark. The cell suspension was then spun at 2,500 rpm for 3 minutes, supernatant decanted, and the remaining cell population washed twice with phosphate buffered saline bovine serum albumin (PBSA), pH 7.2–7.4. The cell population was resuspended to the desired cell concentration (4–9 × 109 cells/L) in PBSA. Then 50 μL of these cell suspensions were incubated with the following antibody combinations, concentrations varying according to each manufacturer's recommendations: IgG1 FITC/IgG1 PE (BD/CD45 PerCP (BD), CD19 FITC/CD5 PE (Immunotech)/CD45PerCP (BD), Kappa FITC/Lambda PE/CD19 PerCP Cy5.5 (Dako, Glostrup, Denmark), FMC7 FITC (Dako)/CD23 PE (Dako), CD79b FITC (Dako)/CD19 PE (Dako), CD20 FITC (Dako)/IgM (Dako), IgM FITC (Dako)/CD38 PE (Dako).

All incubations were performed at room temperature in the dark for 20 minutes. Thereafter, 3 mL of PBSA was added to each tube that was centrifuged at 2,500 rpm for 3 minutes and pellets resuspended in 1% (v/v) formaldehyde solution.

Throughout the study we used CaliBRITE™ beads to set up the flow cytometer in accordance with the manufacturer's recommendation (BD Biosciences, Franklin Lakes, NJ, USA). A total of 10,000 lymphocyte events were collected per tube. In brief, using a three-color combination, the lymphocytes were identified using a CD45 vs. SSC (side scatter) strategy (lymphocytes expressing bright CD45). Clonality was determined using CD19 vs. SSC to determine B cell population. Subsequently, tubes containing two fluorochrome antibody combinations using forward scatter (FSC) vs. SSC analyzes were used to gate the lymphocytes. Two controls were used: samples from each patient had an isotype control and patients' T cells (CD5+) from cell populations were used as an internal control as guidance for analysis. A cut-off limit of 30% of lymphoid cells was used to denote a positive result with a given antibody using computer software FACSCalibur™ Quad stat analysis. The 30% cut-off was selected as recommended by the British Committee for Standards in Hematology guideline (25). Intensity of expression of surface CD20 and immunoglobulin were determined from the scatter plot display by comparison to the control without antibody. Intensities were defined as follows: weak expression when the majority of the population was in the first log percentile above control, strong expression when the majority of the population was in the second log percentile or higher above control. The CLL score of Moreau et al. was applied to all cases, this comprises 1 point each for expression of CD5, strong sIg and CD23, and lack of expression of CD79b and FMC7, the majority of CLL will score 4 or 5 on this scale (6).

Fluorescent In Situ Hybridization (FISH) Testing

From each patient, bone marrow or peripheral blood cells were smeared onto a glass slide, fixed in 3:1 methanol:glacial acetic acid for 2 minutes and allowed to air dry. Probe mixture (1μL of dH20, 3.5μL LSI hybridization buffer, 0.5μL of LS1 IGH/CCND1 Dual color dual fusion translocation probe, Abbott Diagnostics, Maidenhead, UK) was then applied to each slide. The slide was sealed and denatured (75°C for 2 minutes) prior to an overnight hybridization at 37°C (Vysis Hybrite™, Abbot Diagnostics). The next day, the sealant was removed and the slide was soaked in 2 × SSC for 2 minutes. Subsequent to a post hybridization wash (0.4 × SSC/0.3%Triton X, VWR International) at 70°C for 3 minutes, slides were taken through an ethanol series (50%, 90%, 100%, 1 minute each). Slides were mounted with vector shield/DAPI (4′6-Diamidino-2-phenylindole) solution (Vector Laboratories) and viewed under a fluorescent microscope (Eclipse 50i, Nikon Corp., Tokyo, Japan). For each case, up to 200 cells were analyzed by two independent analysts.

Statistical Analysis

Differences between groups were assessed using Fisher's exact test for categorical data or unpaired t-test for continuous variables. Overall survival was defined as time from immunophenotyping until time of last followup or death from any cause. Survival analysis was performed according to the method of Kaplan and Meier (26) and survival curves were compared using the log-rank test. These analyzes were performed using GraphPad Prism version 5.03 for Windows (GraphPad Software, San Diego, CA). Multivariate analysis was performed by logistic regression using SPSS software (SPSS Inc., Chicago, IL). A P value of <0.05 was taken as the level of significance.


Characterization of the Derivation Cohort

We analyzed 346 consecutive clonal LPD samples referred at first diagnosis that expressed CD5 and CD19. Of these 94 were considered to have an immunophenotype atypical for CLL on the basis of CLL score ≤3 (78 cases) or CLL score 4 but lacking CD23 expression (five cases) or CLL score 4 but with morphology atypical for CLL (11 cases). These 94 and 2 additional CD5 negative cases with immunophenotype and morphology consistent with possible MCL were referred for FISH t(11;14) testing. The translocation was detected in 18 cases, in addition four cases of known, previously diagnosed, MCL that were undergoing immunophenotyping were examined. Thus a total of 22 cases of confirmed MCL were identified from the 100 cases where a diagnostic test for MCL had been performed. The remaining 78 cases formed the comparator t(11;14) FISH negative group. Case characteristics are shown in Tables 1 and 4.

Table 1. Characteristics of the Derivation Cohort
Characteristict(11;14) FISH negativeConfirmed mantle cell lymphomaOR (95% CI)P
  Male41195.7 (1.6–20.9)
 Median age/years (range)68 (43–94)70 (44–83)0.79
Hematological parameters
 Median WCC × 109/L (range)18.9 (3.6–226)15 (3.7–354)0.31
 Median Hb/g/dL (range)13.2 (5.7–16.5)11 (7.1–15.4)0.02
 Median platelet count × 109/L (range)219 (2–624)138 (12–317)0.0002
 Platelet count <150 × 109/L15/7115/218.6 (2.9–25.8)0.0001
 Surface immunogloblin strong28/7818/228.0 (2.5–26.1)0.0002
 CD23 negative42/7820/228.6 (1.9–39.2)0.001
 CD20 strong21/7719/2216.9 (4.5–63)<0.0001
 FMC7 positive50/7820/225.6 (1.2–26)0.02
 CD38 positive15/7316/2210.3 (3.4–30.8)<0.0001
 CD79b positive64/7819/221.4 (0.36–5.3)0.75
 Mean CLL score (range)2.6 (0–4)1.5 (1–3)<0.0001

Cases negative for the t(11;14) translocation and confirmed MCL were compared (Table 1). On univariate analysis, the following characteristics were significantly associated with MCL: male sex, lower hemoglobin and platelet count, platelet count below normal range, strong sIg expression, lack of CD23 expression, strong CD20 expression, FMC7 expression, CD38 expression, and lower CLL score. Median age and CD79b expression did not significantly differ between the two groups. No single immunophenotypic marker reliably identified all MCL. In particular, CD23 did not demonstrate markedly better discriminatory activity than other associated markers. We therefore assessed combinations of markers for discrimination between t(11;14) negative cases and MCL.

Figure 1A shows the distribution of CLL scores of t(11;14) negative cases and of confirmed MCL. Of t(11;14) negative cases, 60% had a CLL score of 3 or 4, while in 73% of MCL CLL score was 1. No MCL had a CLL score >3. The CLL score does not include other variables that we found to be significantly associated with MCL in univariate analysis. We therefore attempted to develop an MCL score from these variables.

Figure 1.

Histograms illustrating the distribution of CLL (A) and MCL (B) scores for confirmed MCL cases and t(11;14) FISH negative LPDs in the derivation cohort.

Development of a Mantle Cell Score

Multivariate analysis was used to identify variables independently associated with MCL. Covariates included in the regression model were: sex, hemoglobin concentration, platelet count, surface IgM and CD20 expression (strong vs. weak), light chain restriction and CD23, FMC7, CD38 and CD79b expression. Table 2 shows variables independently associated with MCL, and these were: platelet count <150 × 109/L, absence of CD23, strong CD20, and expression of CD38. Giving a point for each of these four variables, an MCL score was constructed that could be used to try and predict which cases are likely to possess the t(11;14) translocation. Data were missing for one of the four points in 13 FISH t(11;14) negative cases and one MCL case. The MCL score was applied to the remaining 86 derivation cohort cases alongside the CLL score (applied to all cases), Figure 1B.

Table 2. Multivariate Analysis to Derive an MCL Score
VariableOdds ratio for detecting MCL (95% CI)P
Platelets <150 × 109/L12.97 (8.07–17.87)0.005
CD23 negative8.44 (4.32–12.56)0.004
CD20 strong17.85 (12.52–23.18)0.004
CD38 positive3.39 (1.81–4.97)0.007

Figure 2 shows receiver operator characteristic (ROC) curves for MCL and CLL scores. In ROC analysis a perfect screening test would achieve an area under the curve (AUC) of 1, while a test no better than random guessing would achieve an AUC of 0.5. The MCL score has a greater AUC than the CLL score (0.95 vs. 0.79), suggesting that this new score has a greater discriminatory value for identification of MCL. Table 3 shows the characteristics of the two scores. MCL score ≥2 would correctly identify 100% of MCL with a positive predictive value of 49%. The specificity of MCL score ≥2 was 66%. This would exclude 44 of the 78 t(11;14) negative cases in the derivation cohort (negative predictive value 100%). Using MCL score ≥3 improves specificity to 92% but this cut-off would give four false negative results amongst the 22 confirmed cases of MCL (sensitivity 81%).

Figure 2.

ROC curves of CLL (A) and MCL (B) score. Black lines indicate the values of sensitivity and 1 - specificity for discriminating confirmed MCL from t(11;14) FISH negative LPDs at score values of 0, 1, 2, and 3 in CLL and MCL scores. Broken grey line indicates the ROC curve of hypothetical ‘perfect’ screening test with AUC = 1, while solid grey line indicates path of the ROC curve for a test which is no better than random guessing (AUC = 0.5).

Table 3. Characteristics of Mantle Cell and CLL Scores from Derivation Cohort
CLL score≤310.210.261
MCL score≥110.250.31

Validation of the MCL Score

To confirm the utility of the MCL score in a patient group other than the one it was derived from, the score was prospectively applied to a FISH t(11;14) tested validation cohort. This comprised 42 cases referred at diagnosis and considered to be atypical CLL on the basis of CLL score ≤3 (32 cases), CLL score 4 but CD23 negative (2 cases), and CLL score 4 but with morphology atypical for CLL (eight cases). Of these five were found to be t(11;14) positive and an additional two cases of known MCL were analyzed—seven MCL in total. Derivation and validation cohorts are compared in Table 4, and they are comparable other than a significantly higher proportion of males in the validation cohort (82% vs. 60%, P = 0.01). The MCL score was applied to this cohort in a blinded manner. The AUC of the ROC curve for this validation cohort was 0.92, similar to that of the derivation cohort. Table 5 illustrates the characteristics of the MCL score in the validation cohort. No case of confirmed MCL achieved an MCL score of less than 3. This increase in sensitivity in the validation cohort is mirrored by a reduced specificity with specificity of 38% for an MCL score ≥2. The positive predictive value of MCL score ≥2 is 23% in the validation cohort. Negative predictive value is high at 100% for any MCL score up to 3.

Table 4. Comparison of Derivation and Validation Cohorts
CharacteristicDerivation cohortValidation cohortP
 Median age/years (range)69 (43–94)68 (49–88)0.61
Hematological parameters
 Median WCC × 109/L (range)18.7 (3.6–354)16.5 (1.4–202)0.96
 Median Hb/g/dL (range)12.9 (5.7–16.5)13 (8.7–16.6)0.36
 Median platelet count × 109/L (range)202 (2–624)164 (15–430)0.13
 Platelet count <150 × 109/L30/9318/440.34
 Surface immunogloblin strong46/10020/441
 CD23 negative62/10031/440.35
 CD20 strong40/9923/440.21
 FMC7 positive70/10032/440.84
 CD38 positive31/9516/440.7
 CD79b positive83/10038/441
 Mean CLL score (range)2.4 (0–4)2.3 (1–4)0.75
Table 5. Characteristics of Mantle Cell Score Applied to the Validation Cohort
MCL scoreSensitivitySpecificityPPVNPV
  1. Abbreviations: PPV, positive predictive value; NPV, negative predictive value.


Overall Survival

The overall survival of patients with confirmed MCL was compared to that of FISH t(11;14) negative patients. Followup data were available for 131 patients from both derivation and validation cohorts (109 FISH negative and 22 MCL). Patients with MCL had a shorter median survival than those classified as atypical CLL (3.3 vs. 4.2 years, Figure 3), although this difference did not reach a level of statistical significance (HR 2.2, 95% CI 0.87–5.5, P = 0.1).

Figure 3.

Cumulative overall survival of MCL (grey) and atypical CLL (black) patients. Kaplan-Meier estimates of overall survival proportions from time of initial diagnostic immunophenotyping.


In this study we have examined a large series of LPD cases to identify markers pointing to the diagnosis of MCL. MCL was defined either on the basis of histological sections stained for immunophenotypic markers and the presence of cyclin D1 (known MCL cases) or by the disease driving translocation t(11;14). t(11;14) approaches a definitive diagnostic test (2, 21). In their series of 80 t(11;14) positive LPDs Orchard et al., after extensive morphologic, immunophenotypic, and immunohistochemical studies could find no evidence that the diagnosis in all cases was anything other than MCL (27).

FISH t(11;14) testing is a reasonably time-consuming and expensive test and it would be unfeasible and unnecessary to screen all CD5/19 dual positive LPDs for the presence of the MCL translocation. In this study, we have assessed the utility of the revised Matutes/Marsden CLL score (6) in identifying MCL. While no MCL scored 4 or 5 on the CLL score, the majority of score ≤3 or less were not confirmed as MCL. Previous studies have frequently identified the absence of CD23 as the most reliable immunophenotypic marker for distinguishing CLL and MCL (8, 11, 12). Our data suggest that although CD23 is one of the three most reliable immunophenotypic markers for making this distinction it is not sufficient alone. CD23 expression on LPDs carrying the t(11;14) translocation has previously been described (16, 28) and we found three (of 29) cases of CD23 positive MCL in these series. In contrast the majority (68 of 94) of CD23 negative LPDs were not confirmed as MCL.

In addition to CD23, we find that CD38 and strong CD20 expression are associated with MCL. Strong CD20 expression has been described as discriminatory for MCL (9). CD38 is an ADP-ribosyl cyclase expressed on a range of hematopoietic cells, and its ligation stimulates intracellular signaling which has multiple effects on cell activation and proliferation (reviewed in Ref.29). CD38 came to prominence in CLL as a prognostic marker (30). Although CD38 expression is therefore well recognized in CLL our data suggest a stronger association with MCL. There are data to suggest that CD38 expression in CLL may vary during the course of the disease (31), it is tempting to speculate that its expression in MCL is more robust. CD38 expression in MCL has been suggested as a marker that partially differentiates nodal and non-nodal MCL, with 94% of nodal MCL being CD38 positive in comparison to 48% of non-nodal disease (27). We have not determined what proportion of our patients had underlying nodal disease; however in our cohort of patients mostly with peripheral blood involvement by MCL, we found CD38 expression in 22 of 29 cases (76%).

Using multivariate analysis three immunophenotypic markers independently associated with MCL. Of note CD79b, one of the components of the CLL score, demonstrated no discriminatory activity between MCL and CLL. In addition to immunophenotypic markers, the other parameter independently associated with MCL was thrombocytopenia, as defined as Plt < 150 × 109/L. Univariate analysis indicated that hemoglobin concentration and thrombocytopenia were significantly associated with MCL, however platelet count showed the stronger association. Thrombocytopenia has previously been found to be an association of MCL (15). It is perhaps significant that one of the markers associated with MCL is also the one that demonstrates the more aggressive nature of MCL in comparison to CLL, indicating disease suppressing normal hematopoiesis in a significant proportion of these patients.

One potential weakness of our study is that, with the exception of 2 t(11;14) FISH negative cases in the derivation cohort, we have made an assumption that CD5 expression is a condicio sine qua non for the diagnosis of MCL. While others have defined MCL on the basis of obligate CD5 expression (12, 16, 24), there are reasonably extensive data describing CD5 negative MCL with incidence estimates of 10%–30% of MCL (1, 15, 18, 32). There is some suggestion that CD5 expression in MCL may vary between bone marrow disease and MCL found in the peripheral blood (33) or between classical and blastic variants of MCL (2). As our scoring system was derived on the assumption of CD5 expression, we cannot extrapolate its use to CD5 negative cases. A worthwhile further study would be to apply the score to CD5 negative cases considered on the basis of immunophenotype and or morphology to be possible MCL. This would also allow the opportunity to examine the significance of CD148 expression, the absence of CD11c expression and nuclear SOX11 expression—all phenotypes that have been recently suggested to be associated with MCL—and which were not examined in our study (34–36).

With any screening test based on a scoring system, a cut-off point needs to be selected in determining whether to proceed to a diagnostic test. This decision will be influenced by the clinical implications of missing the diagnosis. Others have previously reported that the overall survival of atypical CLL and MCL cases is comparable (10), suggesting that the distinction may not be clinically relevant (although treatment for the two conditions is likely to vary). MCL is considered to be a low grade lymphoma but one characterized by an aggressive clinical course (2). Here we demonstrate that overall survival was diminished in the the MCL group compared to t(11;14) FISH negative cases, although not reaching statistical significance, we suggest that a trend in the survival curves is apparent. Although we have superficially labeled the group of t(11;14) FISH negative cases as ‘atypical CLL’, this is likely not to be a pure population of a single lymphoproliferative disorder, in particular some of the low CLL score t(11;14) FISH negative cases may be splenic lymphoma with villous lymphocytes or lymphoplasmacytoid lymphoma (1, 37–39). Despite this probable heterogeneity the survival data suggest that MCL carries a poorer prognosis than this group and should be definitively identified. In both derivation and validation cohorts an MCL score of ≥2 identified all cases of MCL, the majority of cases scored ≥3. A cut-off of ≥2 should safely identify all MCL. As further experience with the score develops, it may be possible to improve specificity by choosing a higher cut-off. On the basis of this study, we have revised our practice on referring samples for FISH testing. We now refer for FISH testing any CD5 positive LPD with a score of 2 or more on the MCL score described. This has reduced our use of FISH testing as it sets clear criteria for the selection of cases which are more likely to yield a positive result and ensures that t(11;14) cases are not missed.

In conclusion, we present data here suggesting that in trying to differentiate MCL from CLL amongst CD5/19 dual positive populations of cells greatest weight should be given to absence of CD23, presence of CD38, strong CD20 expression, and thrombocytopenia. A combination of two or more of these markers will identify a group that comprises 23%–49% MCL and which should be referred for definitive diagnostic testing by FISH t(11;14). The median overall survival of the group of identified MCL patients is diminished compared to those patients with LPDs lacking the t(11;14) translocation.


The authors would like to thank Estelle Glasper, Ruth Coster and Pandora Ford for assistance with collection of survival data.