Dr T. Mainou-Fowler, University Department of Haematology, School of Clinical and Laboratory Sciences, Royal Victoria Infirmary, Queen Victoria Road, Newcastle upon Tyne NE1 4LP, UK. E-mail: Tryfonia.Mainou-Fowler@ncl.ac.uk
Summary. Recent studies have shown that CD38 expressed as a percentage of the antigen positivity can predict prognosis and disease progression in patients with B-cell chronic lymphocytic leukaemia (B-CLL). The present study showed that quantification of CD38 expressed as antibody-binding capacity (ABC) improves the prognostic value of the percentage of CD38 positivity in B-CLL. In a cohort of 81 patients with B-CLL, a level of CD38 expression of ≥ 30% and an ABC value of 250 proved statistically valid cut-off points to predict disease progression (% CD38: P=0·0027; ABC: P < 0·0001). There was a positive and significant correlation between the percentage of CD38 expression and ABC (r=0·7; P < 0·0001). There was a better discrimination of survival using ABC rather than percentage CD38 positivity (P < 0·0001 compared with P=0·0027). Only ABC predicted for survival in patients under 60 years of age (P=0·0076) or with stage A disease (P=0·0195). Both percentage CD38 and ABC discriminated between time to first treatment for all patients but only ABC predicted time to treatment for stage A patients (P=0·0004). In conclusion, CD38 positivity is an important prognostic factor in B-CLL. However, quantification of CD38 is superior to the percentage positivity and should be used clinically in conjunction with other variables of predictive value to identify B-CLL patients that are likely to progress.
B-cell chronic lymphocytic leukaemia (B-CLL) is a malignant B-cell proliferation with a wide clinical spectrum of disease, course and prognosis. Initial treatment is usually with oral chemotherapy (O'Brien et al, 1995). The efficacy of this treatment decreases with time and most patients die from causes related to B-CLL. Therefore, it would be an advantage to identify the group of patients that are likely to progress rapidly and could benefit from early, more intensive chemotherapy. Although the Rai and Binet staging systems are useful tools to assess the prognosis in B-CLL, both fail to identify the subsets of patients whose disease may progress rapidly.
A large number of prognostic factors are already recognized in B-CLL (Zwiebel & Cheson, 1998). Recently, the mutation status of the immunoglobulin (Ig) variable region (V) genes has been shown to be a powerful predictor of disease outcome. Patients with unmutated VH genes are more likely to have advanced stage and shorter survival (Damle et al, 1999; Hamblin et al, 1999). Recently, Damle et al (1999) have reported that patients with unmutated VH gene express the CD38 antigen in 30% or more of their cells and that this was associated with more aggressive disease.
CD38 was originally described as a T-cell differentiation antigen (Reinherz et al, 1980). A wider antigen distribution is now recognized and includes medullary thymocytes, lymph node germinal centre lymphoblasts, plasma cells, resting natural killer cells, monocytes and endothelial cells (Malavasi et al, 1994). Tissue distribution of CD38 appears to be dependent on the stage of cell maturation and activation (Malavasi et al, 1994; Oertel et al, 1996). Although the function of the CD38 antigen remains unclear, its expression has been used not only as a marker to study lymphocyte activation and differentiation but also in the systematic classification of T- and B-cell malignancies (Reinherz et al, 1980). It has been reported that leukaemias which result in immortalization of early differentiation stages are CD38+, while differentiated phenotypes are CD38– (Malavasi et al, 1994). Recently, the prognostic significance of CD38 expression in many different types of leukaemias including B-CLL has been the subject of many investigations (Omede et al, 1990; Damle et al, 1999; Keyhani et al, 1999). With respect to B-CLL, there has been some controversy about the value of CD38 expression as a prognostic indicator. While some investigators have found no prognostic significance (Thunberg et al, 2001), others have observed a significant association between CD38 positivity and presence of known adverse prognostic factors including VH gene rearrangements (Damle et al, 1999; Benet et al, 2000; Del Poeta et al, 2000, 2001; Ibrahim et al, 2001).
The present investigation re-assesses the prognostic significance of CD38 in B-CLL and examines whether quantification of CD38 expression improves its prognostic value.
Patients and methods
Patient population. Eighty-one patients with B-CLL from the former Northern Health Region of the UK were included in this study. Patient selection was based on the availability of detailed clinical histories and cryopreserved cells. All patients were diagnosed as having B-CLL on the basis of clinical examination, peripheral blood (PB) count, absolute lymphocyte count, immunophenotyping (CD3, CD5, CD19, CD10, CD23, FMC7) and monoclonality of the j or λ light immunoglobulin chain expression. The clinical outcomes for all the patients were obtained from the patients records. Fifty patients had received treatment before entry to this study. Thirty-one patients had received no treatment prior to study. All patients gave written consent for their study which had been approved by the local ethical committee.
Cells. Peripheral blood mononuclear cells were isolated by centrifugation on Ficoll-Hypaque gradient (Nycomed, Sheldon, Birmingham, UK) and cryopreserved in tissue culture medium [Roswell Park Memorial Institute (RPMI)-1640 medium; Invitrogen Life Technologies, Paisley, UK] containing 10% dimethylsulphoxide (NBS Biologicals, Huntingdon, Cambs, UK) and 20% heat-inactivated fetal calf serum (FCS, Invitrogen Life Technologies). The cryopreserved cells were stored at −140°C for a maximum of 6 years.
CD38 expression. Cryopreserved cells were thawed and washed twice with phosphate-buffered saline (PBS) containing 1% FCS before being stained for CD38 expression. Briefly, cells (1 × 106) were stained for 60 min in the dark and at room temperature with peridinine chlorophyll protein (perCP)-labelled α-CD19, fluorescein isothiocyanate (FITC)-labelled α-CD5 and phycoerythrin (PE)-conjugated α-CD38 monoclonal antibodies or with PE-labelled control of the same isotype as the CD38 antibody (IgG1) (Becton Dickinson, Cowley, Oxford, UK).
After washing the cells twice with cold PBS 1% FCS, the cell-associated fluorescence was measured on a facscan flow cytometer (Becton Dickinson). The data was registered on the logarithmic scale on the FL2 channel. Cell debris was excluded by appropriately raising the forward scatter threshold. Each time, 5 × 103α-CD19+- and α-CD5+double positive-lymphocyte-gated events were acquired and analysed for CD38 expression (FL2 channel; cellquest software, Becton Dickinson). All measurements were done under the same instrument settings. In order to validate the method of CD38 expression, dual samples from seven patients were thawed at different time intervals and the cells were stained for CD38 expression as described above. Statistical analysis using Student's paired t-test showed that the pairing was effective (the correlation coefficient was 0·9631; P=0·8245 (two-tailed); 95% confidence intervals (CI): −4·955–4·098).
Quantification of CD38 expression. The expression of the CD38 antigen was quantified by flow cytometry using microbeads of specific antibody binding capacity (ABC) (Quantum Simply Cellular kit; Sigma-Aldrich, Poole, Dorset, UK). When the microbeads are labelled with a specific antibody (i.e. α-CD38), they can serve as a set of standards to calibrate the fluorescence scale of the flow cytometer in units of ABC which is the number of antibodies bound per cell or microbead. The Quantum Simply Cellular kit is a mixture of four highly uniform microbead populations of known antibody binding capacities.
The microbeads are labelled under the same conditions and with an equal amount of the test antibody as the experimental samples. The median values of the fluorescence intensity of the four peaks corresponding to the four microbead populations are used to construct a calibration curve. The quick cal data program, a software program provided with the kit (Sigma-Aldrich), automatically constructs the calibration curve of the test antibody and derives ABC values for all the cell unknowns. Using the Quantum Simply Cellular kit we obtained ABC values for CD38 expression for all samples under investigation. Validation of the ABC method using dual samples as described previously gave a P-value of 0·3796 (CI: −565·3–248·1; correlation coefficient: 0·6836).
Statistical analyses. Fisher's Exact test was used to analyse and identify the level of significance of CD38 expression that differentiated between the group of patients that survived and those who died, or between the different Binet stages. The non-parametric Spearman rank correlation test was used to identify the association between the percentage CD38 expression and ABC. The Kaplan–Meier method was used to construct the survival curves using GraphPad Prism version 2 (San Diego, CA, USA, http://www.graphpad.com). Student's paired t-test was used to validate the expression of CD38.
In the cohort of 81 patients used in the present study, 47 were men and 34 were women, with a median age of 61 years (range: 31–81). For full patient characteristics see Table I. At diagnosis, 62 patients were at Binet stage A including 49 stage AO of disease progression, 11 at stage B and 8 at stage C. All patients had the full diagnostic criteria of CLL including clinical examination, absolute lymphocytosis and immunophenotyping to confirm B-cell clonality. Median follow up was 6·5 years (range; 6 months to 21 years 7 months). There were 20 deaths in the patient cohort, 15 from CLL, four from ischaemic heart disease and one from metastatic malignant melanoma. The median age at diagnosis of the > 30% CD38+ cases was 63 years (range: 38–76) and of the < 30% was 61 years (range: 31–81). The median ages of the > 250 ABC (median: 62; range: 31–81) cases were similar to those of < 250 ABC (median: 61 years; range: 31–81).
Table I. Patient characteristics.
CD38 < 30
CD38 > 30
ABC < 250
ABC > 250
Median age at diagnosis, years (range)
Median length of follow-up, years (range)
6·5 (6 months to 21 years and 7 months)
Binet stage A0
Binet stage A
Binet stage B
Binet stage C
We confirmed that a level of CD38 expression of > 30% (or of > 20%) was a statistically valid cut-off point to predict disease progression and outcome in our cohort of patients with B-CLL (Table II). We found that in terms of density of CD38 expression, an ABC value of > 250 was a statistically valid cut-off level in predicting patient survival and disease progression (Table II). Regression analysis showed a positive and significant correlation between the percentage of CD38 expression and ABC (r=0·7; P < 0·0001) (data not shown).
Table II. Levels of significance of CD38 expression that identified disease progression and outcome in patients with B-CLL.
With respect to disease stage, a level of CD38 expression of > 30% identified 6/11 and 5/8 patients who presented with Binet stage B and C respectively (data not shown). The cut-off point of ABC > 250 identified 8/11 and 5/8 patients with Binet stage B and C respectively (results not shown).
Time to treatment
During the follow up in this study, 57 patients required therapy. As initial treatment, 49 received chlorambucil, five received prednisolone as a single agent, one received CHOP (cyclophosphamide, hydroxydoxorubicin, oncovin, prednisone) (Fisher et al, 1993), one patient CVP (cyclophosphamide, vincristine, prednisone) (Kennedy et al, 1978) and one fludarabine. Both percentage CD38 binding and ABC discriminated between time to first treatment for the entire patient cohort; ABC (> or < 250) gave median time to treatment of 14 months vs 59 months (P < 0·0001), respectively, and percentage CD38 (> or < 30) was slightly less predictive (13 months vs 44 months; P=0·0025) (Fig 1). Only ABC achieved significance for predicting time to treatment for stage A patients; 23 months vs 60 months, P=0·0004),% CD38 binding did not distinguish between time to treatment in these early patients (Fig 2).
Both percentage CD38 binding and ABC were able to predict for disease-specific survival of all patients (Fig 3). However, when stage A patients were analysed separately, only ABC predicted for survival (P=0·0195), whereas percentageCD38 did not (Fig 4). In patients aged < 60 years, only ABC predicted for survival (P=0·0076 vs P=0·1853 for percentage CD38) (Fig 5), while for patients over 60 years of age, both ABC and CD38 positivity predicted for poorer survival (Fig 5).
The present study confirms previous findings that high CD38 expression on malignant cells (i.e. > 30%) predicts disease progression and outcome in patients with B-CLL. This study is the first to report that variation of CD38 site intensity has a greater predictive value in determining disease progression and survival than the percentage of cells that are CD38 positive.
The mutational status of the IgVH genes is a powerful prognostic predictor for B-CLL patients (Damle et al, 1999; Hamblin et al, 1999). Damle et al (1999) have reported that there is a strong and inverse relationship between IgVH gene mutations and the expression of the CD38 antigen by the malignant cells in patients with B-CLL. Based on the degree of CD38 positivity and IgVH gene mutations, these authors showed that the cut-off point at 30% of cellular CD38 expression could identify two distinct prognostic groups. Patients with high CD38 cellular expression (i.e. > 30%) experienced a worse clinical course while those with low CD38 levels had a more indolent disease.
Our findings showed that CD38 positivity expressed as antibody-binding capacity (ABC) > 250 was a statistically valid threshold for predicting disease progression and survival. This was particularly true in younger patients (aged < 60 years) and those with early stage disease. Due to small numbers we were unable to detect any significant difference in survival in more advanced stage patients although there was a trend for the ABC to be predictive in these patients also (data not shown).
In conclusion, the present study confirms that CD38 expression > 30% predicts disease progression and survival in B-CLL. We are the first to report that the density of CD38 positivity expressed as ABC is a better predictor of disease progression and outcome in B-CLL than the percentage positivity alone. We would suggest that measurement of the density of CD38 expression by flow cytometry should be used as a routine test in assessing B-CLL patients. As neither the density nor the percentage of CD38 expression can predict time to treatment or survival in all patients, these parameters should be used clinically in conjunction with other variables of predictive value in the evaluation of patients with B-CLL.
This work was supported by the Tyneside Leukaemia Research Association. We wish to thank Lisa Thomas for help with the flow cytometry.