Clinical utility of assessing ZAP‐70 and CD38 in chronic lymphocytic leukemia

Abstract

The clinical course of patients with chronic lymphocytic leukemia (CLL) is diverse (1-3). While some patients have progressive disease requiring therapy within a relatively short time after diagnosis, others have indolent, asymptomatic disease that might not require therapy for many years after diagnosis. Because of the difficulty in identifying such patients at diagnosis, only patients with progressive and/or symptomatic disease currently are recommended for therapy. Early treatment of the latter group could place patients at risk for therapy-related complications that may compromise their quality of life and/or survival (4-6). The detection of specific markers that can stratify patients into groups with good or poor prognosis can be a vital tool to predict their response to treatment (7). Several markers have been identified that can segregate patients into groups that differ in their relative propensities for disease progression. One is the expression of CD38, which is associated with more aggressive disease (8-15). Also found to predict more aggressive clinical behavior is the expression of unmutated immunoglobulin heavy chain variable region genes (IgVH) (8, 9, 16-18). A recent popular prognostic marker has been the zeta-chain associated protein of 70 kD (ZAP-70), an intracellular tyrosine kinase, which is involved in T cell-receptor signaling (19, 20) and is highly expressed by the B cells of CLL patients that also express nonmutated IgVH genes (21). Studies have revealed that ZAP-70 is associated with enhanced B-cell receptor signaling (22, 23) and that measurements of this intracellular protein can be used as a surrogate marker for IgVH mutational status (24-28). Although CD38 can be easily measured by clinical laboratories, there has been considerable controversy of its use as a prognostic marker for CLL. The expression of CD38 expression has been shown to be unstable over time in some patients and not in others (8-10, 12, 14, 29-34) and there has been disagreement about which cut-off value to use for CD38 positivity for defining prognosis. Some investigators have used 7% (14, 15, 18, 28), others 20% (10, 13), and still others 30% (8, 9, 11, 12, 31). Thornton et al. and Ghia et al. reported that CD38 expression did not change over an interval of greater than 8 years (15, 32, 35). In addition, Gentile et al. reported stable CD38 expression during the course of the disease in 94–100% of their patients (14). In our laboratory (Kipps) we have found that the expression of CD38 is stable (36). Our investigation of 10 patient samples over a period of 4 years confirms previous analyses that CD38 expression is stable over time (8, 10, 13-15, 31, 35, 37 Appendix B). It has also been suggested that the presence of a distinct CD38 positive population within the leukemic clone (bimodal expression), rather than the numerical cut-off value, correlates with IgVH mutational status and progressive disease (35, 37). We found 18% (56/307) of the patients' leukemic B cells exhibited a bimodal CD38 pattern. This is similar to other investigators who have detected 14.5 and 27% of their patient samples so exhibit a bimodal CD38 pattern (35, 38). As observed by other investigators (35, 38), the presence of bimodality in CD38 expression in our cohort did not alter disease progression. We did not detect any differences between peripheral blood and bone marrow obtained from the same patient, indicating that these tissues can be used interchangeably for CD38 assessment as previously observed (9, 31, 36 Appendix B). The cut-off value for CD38 used by various investigators has been decided according to the survival advantage in CD38 negative patients in the different patient populations analyzed. In studies by Krober et al., Thornton et al., Weistner et al., and Gentile et al., this value was 7% (14, 15, 18, 28). Whether Del Giudice et al. used the 30 or 7% cut-off point for CD38 they detected a statistically significant correlation between CD38 positivity and ZAP-70 positivity (39). Hus et al. did not find a significant difference in their results whether they used 20 or 30% cut-off values for CD38 positivity (40). Ghia et al. suggests that it is not whether CD38 expression changes over time but rather whether such fluctuations may modify the prediction of the patient's prognosis at any given time (i.e., whether a case previously defined positive becomes negative and vice versa). They compared CD38 expression from 111 CLL patients with a minimum of 6 months to a maximum of 113 months interval. They did not detect any CD38 fluctuations in 99/111 in both untreated (76 cases) and treated (23 cases) patients and there was no change in the prognostic evaluation of these patients. They analyzed their patients using several of the proposed cut-off values for CD38 expression. Only 5/111 cases would change prognostic class if using the classic 30 or 20% thresholds and none considering the 7% cutoff. They indicate that even using the most stringent cutoffs, CD38 expression would be unreliable only in 5/111: 4.5% of cases (32). In addition, Ghia et al. suggests that similar levels of infidelity also occurs when the prognosis is based on the IgVH gene mutational status for the cases with a percentage of somatic mutations between 97 and 98% (3.1% of their cases) (32). Rassenti et al. used the log rank test to compare time to initial therapy using three different cut-off points for CD38 positivity: 11.5, 30, and 34%, which represented their data set, a previously published value, and the percent derived by recursive partitioning respectively. For each set of curves, they observed a significant difference between the median time to initial therapy for the low CD38 and high CD38 groups regardless of which cut-off value was used for CD38 positivity (P < 0.0001). The median time to initial therapy for the low CD38 and high CD38 groups for the 11.5, 30, and 34% cut-off values ranged from 8.25 to 3.42 years respectively (36). The variability of CD38 expression observed by different laboratories could be due to the specific CD38 monoclonal antibody used. It would be more accurate to compare the results from these laboratories if they all used the same CD38 monoclonal antibody clone with the same flurochrome. A consensus protocol on the clinical use of CD38 would be welcomed. There has been considerable controversy on which method to use to standardize Zap-70 expression in CLL (24-28). This has been primarily related to the use of internal controls such as T/NK cells versus the use of normal B cells or the use of an isotype control and the particular ZAP-70 clone and conjugated dye used. Crespo et al. assessed ZAP-70 expression in the B cells of 56 CLL patients by flow cytometry and correlated to the IgVH mutational status. They labeled intracellular ZAP-70 with an unconjugated anti-ZAP-70 antibody and then performed a secondary stain with goat antimouse immunoglobulin fluorescein isothiocyanate (FITC). They measured ZAP-70 expression gated on T and natural killer cells (CD3+CD56+). Subsequently, CLL cells (CD5+CD19+) were displayed (24). They concluded that high ZAP-70 expression correlated with unmutated IgVH gene expression and that ZAP-70 expression can be used as a surrogate for IgVH gene expression to determine disease progression and survival in CLL (24). In a retrospective study, Durig et al. confirmed the method of Crespo et al. Rassenti et al. (Appendix A) used the ZAP-70 clone 1E7.2 conjugated to the Alexa-488 dye. They determined the ZAP-70 expression by setting the gates such that 0.1% of the total normal B lymphocytes were in the upper right-hand quadrant. The expression of ZAP-70 in their 300 CLL samples was measured by calculating the percentage of CD19+CD3− cells that was above this gating threshold. They also concluded that high ZAP-70 expression correlated with unmutated IgVH gene expression. However, when they analyzed discordant cases for ZAP-70 and IgVH mutational status, they were able to conclude that ZAP-70 is a stronger predictor for the need for treatment in CLL than IgVH mutational status (Appendix A). Gibbs et al. confirmed the sensitivity of the Alexa Flur 1E7.2 conjugate and the optimal fixation and permeabilization protocol. Orchard et al. also found high concordance between ZAP-70 protein expression and IgVH mutational status in 167 CLL patients (26). For their flow cytometry technique they used the unlabeled ZAP-70 clone 2F3.2 followed by a secondary antibody (sheep-anti-mouse FITC). The T and NK cells were gated and this value was subtracted from the total ZAP-70 positive population to give a value for the CLL cells. Using this technique they also found concordance for ZAP-70 expression in both fresh and frozen CLL cells. They concluded that ZAP-70 protein, which can be measured by flow cytometry, is a reliable marker for CLL, equivalent to that of IgVH mutational status (26). Del Principe et al. used the ZAP-70 clone 1E7.2 conjugated to the Alexa-488 dye in their cohort of 289 CLL patients. They gated on T and natural killer cells (CD3+CD56+) vs. CLL cells (CD5+CD3−CD56−) and used mouse IgG1 isotypic antibody conjugated to Alexa Fluor as a control marker for ZAP-70 positivity. This technique allowed them to conclude (as by Rassenti et al. (27, 36)) that ZAP-70 is stable over time, and that increased ZAP-70 expression in CLL B cells is a more significant predictor of disease progression than the presence of CD38 and sCD23. In addition, they suggest that ZAP-70 has a superior prognostic role over CD38 and IgVH mutational status. They also remark that further studies are needed to develop a standardized flow cytometry protocol for ZAP-70 (41). Gibbs et al. compared the measurement of ZAP-70 in 33 CLLs by using two different antibodies and staining protocols. They used both the Upsate and Caltag anti-ZAP-70 antibodies and two cell fixation and permeabilization methods using only fresh whole blood. The Upstate protocol requires an indirect labeling method, whereas the Caltag uses a direct method. They concluded that the Caltag ZAP-70 antibody (Clone 1E7.2 conjugated to Alexa-488) and their Fix and Perm kit were the easiest to use and were the most sensitive, with 91% concordance between ZAP-70 expression and IgVH mutational status. However, they did not find any correlation between CD38 and either ZAP-70 expression or IgVH mutational status (42). The 20% cut-off value for ZAP-70 expression seems to be universally accepted among laboratories as a true biological numerical definition for this predictor of disease progression in CLL. Previous studies have shown a significant split in the Kaplan-Meier curves at 20% when ZAP-70 expression was related to time to initial treatment (24-28, 39, 41). At this time is unclear blood and bone marrow samples can be used interchangeably for ZAP-70 assessment as was previously observed for CD38. Although Schroers et al. have shown stability of ZAP-70 levels over time, this needs to be studied more widely (43). It is clear from the work of Del Principe et al. that both progression free survival and overall survival are shorter for ZAP-70 positive patients while ZAP-70 negative patients achieve a higher complete remission. And ZAP-70 remains useful in understanding the discordance between CD38 and or mutational status and allows for a better understanding of the heterogeneity of patients with high risk cytogenetics. Hus et al. also show that low ZAP-70 expression correlates with a better response to fluradarbine treatment. In summary, even though there is a need for inter-laboratory harmonization for the ZAP-70 protocol, ZAP-70 stands as a very strong prognostic marker for CLL. Rassenti et al. found in their cohort of 300 CLL patients that low CD38 expression was significantly associated with absence of ZAP-70 expression and with presence of mutations in the IgVH genes (P < 0.0001 for each assessment by the Fisher exact test). The median level of CD38 among cases that used mutated IgVH was 4.0%, when compared with 27.4% among cases that used unmutated IgVH (P < 0.0001). The median level of CD38 among ZAP-70-negative cases was 5.3%, when compared with 27.2% among ZAP-70-positive cases (P < 0.0001) (36). This is consistent with other investigators who have used CD38 and ZAP-70 expression to predict the time to first treatment (8, 9, 11, 18, 33, 39). Rassenti et al. additionally assessed the value of CD38 as a stand alone marker to predict disease progression in patients with CLL for clinicians who do not have the means to asses ZAP-70 and/or IgVH mutational status and how the value of CD38 changes if IgVH mutational status and ZAP-70 expression can be subsequently determined. They concluded that the expression of CD38 is stable in CLL B cells, and that CD38 is an advisable autonomous prognostic parameter to use in the practice of CLL when it is not possible to measure ZAP-70 and/or IgVH mutational status. However, CD38 becomes secondary importance if either the IgVH mutational status or expression of ZAP-70 are subsequently measured and that it has no value if both ZAP-70 expression and IgVH mutational status are additionally assessed (36 Appendix A and B). Del Giudice et al. suggest that both CD38 and ZAP-70 should be used to predict time for the treatment in early stage CLL patients. They used the method of Crespo et al., on fresh peripheral blood and the ZAP-70 clone from Upsate Biotech 2F3.2. They performed a multivariate analysis of 201 CLL patients and found a hazard ratio for CD38 of 1.89 for their cohort with a P value of 0.004 using the 7% cut off for CD38 positivity. When they combined CD38 expression and Zap-70 expression, they obtained more informative information, and suggest that such a flow cytometry based evaluation may overcome the need for the IgVH analysis and FISH. Schroers et al. examined 252 patients with B-CLL and found that ZAP-70 expression of 20% or more and CD38 expression of 30% or more were associated with an unfavorable clinical course (43). The levels of ZAP-70 and CD38 did not change over time. The median treatment-free survival times in patients whose leukemic cells were CD38+/ZAP-70+ were 30 months when compared with 130 months in patients with a ZAP-70−/CD38− status. In patients with discordant ZAP-70/CD38 results, the median treatment-free survival time was 43 months. They suggest that ZAP-70 and CD38 expression provided complementary prognostic information identifying three patient subgroups with good, intermediate, and poor prognosis. They supported their findings by microarray-based gene expression profiling in a subset of 35 patients. They identified 37 genes significantly differentially expressed between the three groups defined by their expression of ZAP-70 and CD38. They summarize that CD38 further refines the clinical relevance of ZAP-70 as an adverse prognostic factor and it allowed them to define three different groups of patients (43). They advise a combined analysis of CD38 and ZAP-70 that will allow the identification of three patient groups: good, intermediate, and poor prognosis (39). Since flow cytometry can be used for both CD38 and ZAP-70, this is more suitable for applications in clinical laboratories than IgVH mutation analysis. However, they indicate that further studies are necessary to universally standardize these protocols (39, 43). One such study suggests that the use of a calibration curve may be one way of achieving standardization (44 Appendix B). Hus et al. analyzed ZAP-70 and CD38 expression on the CD19+/CD5+ B cells of 156 CLL patients and correlated the results with the clinical outcome and risk factors such as the stage of the disease, the lymphocyte count, the lactate dehydrogenase (LDH), and β2-microglobulin levels. They found that the combination of ZAP-70 and CD38 increased the prognostic power of both of the factors. However, they found more significant differences between ZAP-70 positive and ZAP-70 negative patients in CLL poor prognosis parameters (stage of disease, WBC, lymphocytosis, and LDH levels) than between CD38 positive and negative patients. In addition, in their study, ZAP-70 expression appeared to be more relevant than CD38 expression in defining cases of CLL responding and not responding to first line chemotherapy (40). Immunophenotyping for ZAP-70 is performed as previously described (27). The peripheral blood mononuclear cells (PBMC) are stained for 20 min at 4°C with CD19 and CD3 specific mAbs conjugated with allophycocyanin (APC) and phycoerythrin (PE), respectively (PharMingen, San Diego, CA). The cells are washed twice and fixed with 4% paraformaldehyde in PBS and then permeabilized with Saponin in HBSS for 5 min at 4°C. The cells are washed and then stained with Alexa-488-conjugated mAb specific for ZAP-70 (Clone 1E7.2) for 45 min at 4°C. Samples are washed and analyzed via flow cytometry using a FACS-Calibur® (Becton Dickinson, San Jose, CA) and Flow Jo 2.7.4 software (Tree Star, San Carlos, CA). We gate on lymphocytes using their forward-angle light-scatter (FSC) and side-angle light-scatter (SSC). Quadrants are set on gated cells such that 0.1% of the total normal lymphocytes are in the upper right-hand quadrant. This gating is used on all the subsequent samples in the experiment. The expression of ZAP-70 is measured by calculating the percentage of CD19+CD3− cells that is above this gating threshold. In each experiment we use control cells from healthy donors, CLL cells from one patient with high-levels of ZAP-70, and CLL cells from another patient with low-levels of ZAP-70. CLL cells also are analyzed for CD19, CD20, and CD23, using mAbs conjugated to APC, PerCp, and FITC, respectively (BD PharMingen, La Jolla, CA), as previously described (3). Fluorochrome-conjugated, isotype control mAbs of irrelevant specificity were used in all experiments to monitor for nonspecific staining. PBMCs were stained for 20 min at 4° with the following antibodies: PE-labeled anti-CD38 (clone HB7, Becton Dickinson, San Jose, CA), APC-labeled anti CD5, and Peridinin Chlorophyll Protein(PerCP)-labeled anti CD19 (Becton Dickinson, Pharmingen, La Jolla, CA). Samples were washed and analyzed via flow cytometry using a FACS-Calibur (Becton Dickinson, San Jose, CA) and Flow Jo 2.7.4 software (Tree Star). Each sample was run with the appropriate isotype control and this was used to define the negative population. CD38 positivity was defined for any sample where the quotient of the mean fluorescence intensity (MFI) of the CD38 staining divided by the MFI of the isotype control was greater than 1. A minimum of 10,000 cells were analyzed. The negative isotype-matched controls 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 at the right side of the line, as previously described (13). The dot-blot of each sample was gated on the lymphoid gate on the side-forward scatter (SSC-FSC) plot. Within this gate the markers were set on the isotype control to define the negative population. The positive CD38 cells were detected outside this marker. The tumor population was defined by gating on the lymphoid population on the SCC-FSC plot, followed by gating on the CD5+CD19+ population. The percentage of CD38+ cells in this gate was then determined. All immunophenotyping for CD38 expression was performed on cryopreserved cells after having assessed that fresh and cryopreserved cells did not yield different results as previously observed (10, 35).