CD38 Variation in Chronic Lymphocytic Leukemia

Chronic lymphocytic leukemia (CLL) has a highly variable clinical course. Some patients require treatment early while others can be monitored without therapy. CD38 expression has been shown in multiple cohorts to have prognostic significance. An elevated percentage of CD38 positive CLL lymphocytes at the time of diagnosis is correlated with a more rapid need for therapy and a shorter overall survival. The extent to which CD38 varies during the course of CLL, including after therapy, has only been evaluated in a limited fashion.

From a cohort of over 500 CLL patients at the Duke University and Durham VA Medical Centers, we selected 136 patients in whom we had measured CD38 expression by flow cytometry on two or more occasions. We determined the first, maximum, minimum, and range (maximum – minimum) CD38 values. We compared these values to other molecular prognostic markers using Wilcoxon tests and assessed the prognostic significance of these values using Cox proportional hazard models and Kaplan-Meier analyses.

Of the 136 patients, 70% were male and 88% Caucasian, with a median age of 60. The majority had low clinical stage at diagnosis—either Rai stage 0 (68%) or 1 (19%). Molecular prognostic markers were also generally favorable. Eighty-two (67%) patients had mutated IGHV status, 69 (51%) were ZAP70 negative, and 76 (63%) had either 13q deletion or normal cytogenetics, determined by fluorescent in situ hybridization. CD38 expression was measured a median of 5.5 times (2 – 19). The median time between the first and last CD38 measurements was 1206 days (81 – 4109). The median values were 6% (0.6 – 99) for maximum CD38, 1.5% (0 to 84.5) for minimum CD38, and 4.9% (0.2 to 95.3) for CD38 range. Maximum, minimum, and CD38 range were significantly lower in patients with mutated compared to unmutated IGHV status (p < 0.005 for all parameters, Wilcoxon rank sum test). Elevated maximum and CD38 range were significantly associated with a more rapid time to therapy (TTT) and shorter overall survival (OS) in a univariate Cox proportional hazards model (p < 0.03 for all, Wald test). In a multivariate Cox proportional hazards model including first CD38 and maximum CD38 values, only maximum CD38 remained statistically significant. We found that patients with high CD38 variation (CD38 range greater than the median) had significantly shorter TTT and OS than patients with low CD38 variation (p = 0.002 for both, log rank test). Using receiver operator characteristic analyses, we determined that the best cut-off for dichotomizing the first CD38 according to TTT and OS in the entire Duke/Durham VA CLL cohort was 11%. Using this cut-off, 15 patients (11%) converted from CD38 negative to CD38 positive. Using the standard 30% cut-off, 14 patients (10%) converted from CD38 negative to CD38 positive. Patients with a first CD38 measurement less than 11% and subsequent measurements above 11% had a favorable OS, similar to patients with low CD38 for all measurements (p = 0.002, log rank test). However, patients with a first CD38 measurement less than 30% who had subsequent measurements above 30% had an inferior OS, similar to patients with high CD38 for all measurements (p = 0.006, log rank test). Lastly, among 24 patients with CD38 measurements before and after first therapy, the percentage of CD38 positive cells increased in 19 patients (79%), with a median value of 3.2% before to 6.9% after therapy (p = 0.005, Wilcoxon signed rank test).

CD38 values vary as patients transition across the disease trajectory. This variation appears to have prognostic significance, with high variation associated with faster time to first therapy and shorter overall survival. Additionally, in our cohort, a patient's maximum CD38 value had more prognostic significance than a single initial measurement. Thus, longitudinally measuring CD38 throughout the clinical course of CLL could aid in the management of CLL patients, refining the initial prognostic assessment, and improving patient counseling and decision making.

No relevant conflicts of interest to declare.