Predictors of Survival in CLL: Multivariate Analyses Reveal Significant Correlation with IgVH Gene Mutation Status, Clinical Stage, and Cellular CD38 (but Not Zap-70) Expression.

Abstract

One of the complexities of chronic lymphocytic leukemia (CLL) is that some patients die early from the disease, while others identically staged by conventional means may live more than two decades. This has led to a search for prognostic indicators that correlate better with eventual survival. Candidate markers include Rai stage, Binet stage, lymphocyte doubling time (DT), immunoglobulin heavy chain V region (IgVH) mutation status, and leukemia cell CD38 and Zap-70 expression. To determine which of these markers correlate best with survival in univariate and multivariate analyses, we examined these parameters in 150 CLL patients from a VA medical center and a university medical center. CLL cells were purified from patients not receiving active treatment with WBC >20,000/uL by deleting T cells, monocytes, neutrophils, and erythrocytes using antibodies against CD2, CD3, CD16, CD36, CD56, and glycophorin A, and centrifugation over ficoll-Hypaque (CLL purity > 97%). In univariate analyses, patients with advanced stage on presentation had shorter survival than those with low stage [14.0 yr for high modified Rai stage vs. 25.3 yr for low stage (p = 0.013), and 13.8 yr, 13.4, and 25.3 yr for Binet stages C, B, and A, respectively (p = 0.033)]. Patients with a lymphocyte DT >3 yr survived twice as long (22.6 yr) as those with a DT <3 yr (11.3 yr) (p = 0.004). There was a strong relationship of IgVH mutation status with survival (median survival 22.6 yr for mutated, 9.4 yr for unmutated; p = 0.0006). Likewise, cellular CD38 expression significantly correlated with survival (CD38 neg 25.3 yr, CD38 pos 13.4 yr, p = 0.008). Zap-70 expression determined by quantitative immunoblot [Zap-70/actin ratio (anti-Zap-70 antibody from Upstate; clone 2F3.2)] was not significantly related to survival (Zap-70 pos 20.8 yr, Zap-70 neg 22.6 yr). Other parameters found not to be related to eventual survival were WBC on presentation, maximum WBC reached before treatment, and time from diagnosis to initial treatment. IgVH mutation status was significantly related to lymphocyte DT, CD38 expression, and Zap-70 expression. In multivariate analyses using the proportional hazards model, IgVH mutation status (p = 0.001), cellular CD38 expression (p = 0.025), and Binet stage (p = 0.026) were significantly associated with survival (with p = 0.0002 overall for these three parameters in the model), but Zap-70 had no significant relationship to survival in the model. Thus, in contrast to other reports, cellular Zap-70 expression does not correlate with various parameters of disease severity (including survival) in our cohort of CLL patients. In summary, our univariate analyses show significant correlations of advanced clinical stage, short lymphocyte doubling time, high CD38 expression, and unmutated IgVH with short survival. However, multivariate analyses show that only IgVH gene mutation status, clinical stage at presentation, and cellular CD38 expression correlate significantly with survival.