Employing a multicolour flow cytometry assay, 133 B-chronic lymphocytic leukaemia (B-CLL) cases were analysed for surface expression of CD38. Based on a cut-off value of 20%, CLL patients were categorised into a CD38-positive (20%, n = 56) and a CD38-negative subgroup (<20%, n = 77) and separately analysed for clinical and laboratory parameters. Patients in the CD38-positive cohort were characterised by an unfavourable clinical course with a more advanced disease stage, poor responsiveness to chemotherapy, short time to initiation of first treatment and shorter survival. In contrast, the CD38- negative group required minimal or no treatment, remained treatment-free for a longer time period and had prolonged survival (P < 0.05). CD38 expression was a robust marker in the majority of patients in that it was stable over time and not significantly influenced by chemotherapy. In conclusion, our data confirm recent studies suggesting a role of CD38 as a predictor of clinical outcome in patients with B-CLL.

B cell chronic lymphocytic leukaemia is a heterogenous disease with a highly variable clinical course. Staging systems devised by Rai et al¹ and Binet et al² are useful methods for predicting survival and treatment requirements in patients with CLL. However, these staging systems are of limited prognostic value in early stages of the disease (Binet A or Rai stage 0 to 2), which include most of the patients at diagnosis.^3,4 Therefore, a number of studies have focused on identifying novel prognostic markers, which may help define patient subgroups with favourable vs poor clinical outcome in early CLL.^5,6 Recently, two independent studies by Damle et al⁷ and Hamblin et al⁸ have demonstrated that B-CLL may arise from either an immature pregerminal centre B cell with unmutated immunoglobulin (Ig) variable heavy chain (V_H) genes or from a more mature post-germinal-centre B cell with somatically mutated Ig V_H genes. Moreover, Damle et al⁷ found a strong correlation between the Ig V_H gene mutation status, CD38 surface expression of the respective B-CLL clone and clinical outcome in individual patients. B-CLL cases with mutated Ig V_H genes and low numbers of CD38-positive cells exhibit a favourable clinical course, while B-CLL patients with unmutated Ig V_H genes are characterised by a poor outcome in terms of reduced survival and responsiveness to chemotherapy. However, in a follow-up study no association between CD38 expression and Ig V_H gene mutation status was found, although the independent prognostic impact of both Ig V_H gene mutation and CD38 positivity was confirmed.⁹ By contrast, Thunberg et al¹⁰ could not find any prognostic significance of CD38 expression in their CLL patient cohort.

CD38 is a single-chain type II transmembrane glycoprotein that is expressed by a variety of haematologic cells in an activation- and differentiation-dependent manner.¹¹ Its cellular functions include a complex ectoenzymatic activity and the ability to transduce signals involved in the regulation of cell proliferation and survival.¹² Furthermore, it mediates a selectin-like binding to endothelial cells, thus functioning as an adhesion molecule.¹³ In normal human B cell development, CD38 exhibits a discontinuous expression pattern, where the molecule is detected at high levels in B cell precursors, germinal centre and plasma cells, while circulating peripheral blood and tonsillar B cells have markedly lower CD38 surface expression.¹² The potential role of CD38 in CLL pathophysiology is presently unknown. However, it is tempting to speculate that differences in CD38–ligand interaction between CLL clones may influence their proliferative behaviour and chance of survival.

Unlike analysis of the Ig V_H gene mutation status, flow cytometric detection of CD38, expression can be conveniently performed in most clinical laboratories and may prove a valuable adjunct in the current staging system for predicting the clinical outcome in B-CLL patients. To explore this possibility further, we studied CD38 expression in 133 B-CLL patients and correlated the results with clinical and laboratory parameters.

Patients

Between May 1994 and December 2000, 133 consecutive patients with chronic lymphocytic leukaemia from a single institution were enrolled in this study and analysed for several biological and clinical characteristics: age, sex, Binet stage, white blood cell count, haemoglobin level, platelet count, lactate dehydrogenase, thymidine kinase, immunoglobulin A (IgA) serum-concentration, survival, treatment history and time from diagnosis to first treatment. In each patient morphologic diagnosis of B-CLL was confirmed flow cytometrically¹⁴ revealing a typical CD19,20,5,23-positive Ig light chain, (κ or λ light chain) restricted immunophenotype. Patient characteristics are shown in Table 1. The mean follow-up time was 64.3 months (range 0–299).

Cell surface staining and flow cytometry

Fresh heparinised peripheral blood (PB, n = 91) and bone marrow (BM, n = 42) samples were prepared for flow cytometry by ammonium chloride erythrocyte lysis (Ortho-mune Lysing Reagent; Ortho Diagnostic Systems, Raritan, NJ, USA). The immunophenotype was characterised using the following panel of fluorochrome-labeled monoclonal antibodies employing a standard three-colour flow cytometry approach (Figure 1):¹⁴ CD45-fluorescein isothiocyanate (FITC)/CD14-phycoerythrin (PE)/CD20-peridinin chlorophyll (PerCP); CD4 (FITC)/CD8 (PE)/CD3 (PerCP); κ light chain (FITC)/CD19 (PE)/CD5-phycoerythrin-cyanin (PeCy5); λ light chain (FITC)/CD19 (PE)/CD5 (PeCy5); IgM (FITC)/CD23 (PE)/CD19 (PECy5); CD10 (FITC)/CD38 (PE)/CD19 (PECy5). Antibodies were purchased from DAKO (Glostrup, Denmark; CD19, CD10, IgM, κ and λ light chains), Immunotech (Marseille, France; CD5) and Becton Dickinson (Heidelberg, Germany; CD38, CD4, CD8, CD3). Negative isotype-matched controls (Becton Dickinson) were used to define the threshold line separating surface marker positive and negative cells such that less than 1% of isotype-positive cells were present to the right of the line (Figure 1). A CLL population was considered CD38-positive when more than 20% of the gated population (CD19⁺/CD5⁺ cells) expressed it. To minimize potential contamination with coexisting normal B cells, only B-CLL cases in which >90% of CD19⁺ cells co-expressed CD5 were included in the study (Figure 1). In all experiments, a minimum of 10000 cells was analysed. The flow cytometer (FACScan; Becton Dickinson) was calibrated with CAliBRITE-3 beads (Becton Dickinson) and FACSComp Software (Becton Dickinson). Data acquisition and analysis was performed using Attractors and CellQuest software (Becton Dickinson). The same method of sample preparation and three-colour staining was used throughout the entire study period.

Representative flow cytometry dot plot analyses of CD38 expression in patients with B-CLL. Three-colour flow cytometry as detailed in Materials and methods was used to analyse CD38 surface expression on lymphocyte/monocyte gated CD19⁺ B cells co-expressing CD5. (a) Sample of a patient positive for CD38 (20% or more). (b) Sample of a patient negative for CD38 (less than 20%). Numbers are percentages of CD19⁺ B CLL cells expressing CD38.

Full size image

Statistical analysis

Survival times and censored waiting times measured from the time of diagnosis were plotted by the Kaplan–Meier method and compared using the log-rank test. Comparison of clinical and laboratory parameters between the CD38-positive and CD38-negative subgroups was performed using the Mann–Whitney U test (for quantitative variables) and chi-square test (for categorial variables). The Cox proportional model was used for multivariate analyses on overall survival. Significance was defined as P = 0.05, as determined by the two-tailed test.

CD38 expression in B-CLL

We evaluated the surface expression of CD38 in 133 cases of B-CLL employing a three-colour flow cytometry approach with directly conjugated monoclonal antibodies (Figure 1). As per current convention, a given leukaemic population was considered positive for CD38 when 20% of the B-CLL cells expressed the membrane marker.^15,16 Figure 2 shows the distribution of CD38 in the whole patient cohort. Based on the 20% cut-off value, 56 patients (42%) were defined as CD38-positive and 77 patients (58%) as CD38-negative, respectively.

Distribution of CD38 expression in the B-CLL study population (n = 133). CD38 expression was analysed employing a multicolour flow cytometry assay as described in Materials and methods. The dotted vertical line represents the 20% cut-off used to separate CD38-negative from CD38-positive patients. Data are percentages of CD38⁺/CD19⁺ B-CLL cells co-expressing CD5.

Full size image

Twenty-seven patients (13 CD38⁺ and 14 CD38⁻ patients) were studied at two or more time-points (range 2 to 4). As illustrated in Figure 3, CD38 expression in individual patients was relatively stable over time and not substantially influenced by chemotherapy (five patients of the CD38⁻ and 10 patients of the CD38⁺ subgroup received chemotherapy in the observation period) in the majority of patients analysed. Variations in CD38 staining over time were particularly high in CLL cases with CD38 levels around 50%. However, in none of the cases studied did we observe a cross-over of a CD38-negative patient to the CD38-positive cohort and vice versa (Figure 3).

CD38 expression in B-CLL over time. Twenty-seven patients were flow-cytometrically analysed for CD38 expression at two or more time-points (range 2 to 4). Each line in the graph represents an individual patient.

Full size image

In eight CLL patients, comparative analysis of CD38 in bone marrow and peripheral blood samples obtained at identical time-points yielded comparable results (Table 2) in both groups, suggesting that CD38 expression levels on B-CLL cells may be largely independent of the surrounding cellular microenvironment.

Correlation of CD38 expression with clinical and laboratory data

The treatment histories of CD38⁺ and CD38⁻ patients differed significantly (Table 3) in that the former group required more intensive chemotherapy over longer time periods than CD38⁻ patients (P = 0.004). The higher treatment intensity in CD38⁺ patients correlated with a more advanced disease stage observed, both at diagnosis and study enrollment (Table 1). Furthermore, we found highly significant differences in disease progression as indicated by the treatment-free interval (Figure 4). The mean treatment-free interval (Table 1) was longer in the CD38⁻ group (120 months) than in the CD38⁺ patient cohort (32 months, P = 0.00008). Finally, we compared overall survival among the two groups and again, observed statistically highly significant differences (Figure 5). The median survival of patients in the CD38⁺ cohort was 121 months, whereas the median survival for the CD38⁻ group was not reached for the duration of follow-up (P = 0.00655).

Probability of disease progression, as indicated by the treatment-free interval. Kaplan–Meier plot comparing time periods from diagnosis to initiation of chemotherapy in CD38-positive to CD38-negative CLL patients. Mean time from diagnosis to first treatment (months) was 120 vs 32 months in the CD38-negative and CD38-positive group, respectively. Statistical analysis was performed using the log-rank test.

Full size image

Overall survival in CD38-negative and CD38-positive B-CLL patients. Kaplan–Meier plot comparing survival based on CD38 expression. The mean survival of patients in the CD38⁺ cohort was 121 months, whereas the median survival for the CD38⁻ group was not reached for the duration of follow-up. Statistical analysis was performed using the log-rank test.

Full size image

A comparison of further clinical and laboratory parameters among the two groups is shown in Table 1. Notably, significant differences were found for Binet stage, white blood cell count, platelet count, lactate dehydrogenase, β²- microglobulin and bone marrow histology (P < 0.05).

Univariate analysis of risk factors

Univariate Cox regression analysis was used to assess associations between survival time and potential risk factors. Binet stage (during follow-up), haemoglobin levels, platelet counts and CD38 expression and IgA serum concentration were identified as significant factors influencing survival (Table 4).

Multivariate analysis

The following patient characteristics found to impact significantly on survival in the univariate analysis were included in the Cox regression model (complete case analysis, n = 102): CD38, Binet stage (during follow-up), platelet count, haemoglobin and IgA (Table 5). In multivariate analysis platelet count, haemoglobin and IgA, but not CD38 and Binet stage influenced overall survival.

This retrospective study was performed to evaluate the potential of CD38 expression as a prognostic indicator in B-CLL. In line with previous data,⁷ our results demonstrate that in the majority of patients the proportion of CLL cells co-expressing CD38 as determined by standard three-colour flow cytometry is relatively stable over time and does not appear to be influenced by chemotherapy. However, in a small subset of patients characterised by CD38 expression levels around 50%, we observed substantial variations in CD38 surface immunoreactivity in sequential samples obtained at different time-points. This finding may be due to technical problems encountered in the gating procedure used to define CD38 expression levels, ie small ‘random’ deviations in cursor positioning impact more profoundly on CLL populations with intermediate CD38 expression than on leukaemic populations with either high or low CD38 expression levels. However, due to the comparably small number of patients analysed in this subset of experiments, we cannot exclude the possibility that the observed differences indeed reflect fluctuations in CD38 expression in a minority of patients. Furthermore, intraindividual comparisons of specimens obtained simultaneously from peripheral blood and bone marrow yielded similar results. Taken together, these observations suggest that CD38 may be a robust marker that could be used reliably in most routine flow cytometry laboratories. Clearly, these data need to be confirmed in a larger patient cohort.

More importantly, we found that CLL cases displaying a high percentage of CD38 (20%) are characterised by an unfavourable clinical course as compared with the CD38 negative patients (CD38 <20%). In particular, our data show significant differences in terms of overall survival and treatment requirements between the two groups confirming the work of Damle et al⁷ and Hamblin and colleagues⁹ and extending it to a larger patient cohort. Furthermore, the time from diagnosis to initiation of chemotherapy, used as a surrogate marker for progression-free survival,¹⁷ was significantly reduced in the CD38-positive, as compared to the CD38-negative patient cohort.

These findings contrast those of a recent report proposing that CD38 is a poor predictor of prognosis in CLL.¹⁰ This discrepancy may be at least partly explained by technical differences between the studies, ie the latter authors used a two-colour rather than a three colour flow cytometric assay and thus may have included normal residual B cells in their analysis, also the patient number evaluated for CD38 (n = 44) was relatively small.

The CD38 cut-off level employed for risk stratification in CLL patients remains a matter of controversy.¹⁰ In the initial study published by Damle et al⁷ a 30% cut-off value was chosen empirically based on a plot of CD38 expression vs Ig V_H gene mutation status. Unmutated cases were found to exhibit a higher percentage of CD38 expression than mutated (post-germinal centre) B-CLL clones. However, while the prognostic significance of CD38 expression has been confirmed in two follow-up studies^9,18 and now by the data presented here, the association between CD38 expression and Ig V_H gene mutation remains conjectural.^7,9,10 Thus, with the current state of knowledge, Ig V_H gene mutation status and CD38 expression appear to be independent prognostic factors in B-CLL. The 20% CD38 cut-off value used in the present report was selected by reference to previously published immunophenotyping studies of haematologic malignancies and the recent proposals of the EGIL group.^15,16 To better compare our data with those of previous studies^7,9,18 survival analyses were recalculated using the 30% cut-off value employed by other authors.^7,9,18 Interestingly, the resulting survival curves did not differ significantly from those previously obtained with the 20% cut-off value (data not shown). Furthermore, the median survival of our CD38-positive group (121 months) was very similar to that reported by Damle et al⁷ (120 months) and Hamblin and colleagues⁹ (105 months).

In conclusion, our data confirm previous studies showing that CD38 expression is a novel prognostic marker in B-CLL. It will be important to determine whether this parameter in conjunction with other established prognostic factors can improve risk stratification in the routine diagnostic work-up of CLL patients.

This work represents a part of MNs MD thesis. We are indebted to numerous colleagues for generously contributing information on the clinical course and treatment histories of the study patients.

Affiliations

1. Department of Haematology, Medical Faculty, University of Essen, Essen, Germany

   • J Dürig
   • , M Naschar
   • , U Schmücker
   • , A Hüttmann
   •  & U Dührsen

2. Institute of Medical Informatics, Biometry and Epidemiology, Medical Faculty, University of Essen, Essen, Germany

   • K Renzing-Köhler
   •  & T Hölter

Corresponding author

Correspondence to J Dürig.

To obtain permission to re-use content from this article visit RightsLink.

Publication history

DOI

https://doi.org/10.1038/sj.leu.2402339