Definition of a Prognostic Scoring System for Predicting Clinical Outcome in B-Cell Chronic Lymphocytic Leukemia.

Patients whose cells utilize unmutated Ig VH region genes and/or express ZAP-70 or CD38 have a more aggressive course than patients whose cells have mutated Ig VH genes and/or do not express ZAP-70 or CD38. We have conducted here a study on 500 patients characterized for marker expression. By ROC curve analysis, we found 30% as the best cut-off value of CD38 which discriminates between mutated and unmutated cases in CLLs. CD38 expression, had low sensitivity (66%), but relatively high specificity (80%), with a positive and a negative predictive value respectively of 68% and 78% in anticipating VH mutational status. Moreover, the agreement between CD38 expression and VH mutational status was low although significant (K=0.46, p < 0.001). ZAP-70 showed high sensitivity (76%) and specificity (75%), high negative (89%) but a low positive (54%) predictive value and a low, although significant, K statistic (0.45, p < 0.001). Furthermore, we combined CD38 and Zap-70 expression to evaluate whether both variables provided more precise information in estimating VH mutational status. We found sensitivity, 42%; specificity, 97%; positive predictive value, 90%; negative predictive value, 72%; K statistic 0.43, p < 0.001. In conclusion neither CD38 nor ZAP-70 by themselves or in combination were able to anticipate VH mutational status, meaning that CD38 and/or Zap-70 expression could surrogate the VH mutational status. In the second part of this study we wanted to validate this findings on clinical ground. Clinical information was available for 150/500 CLL cases investigated. After a median follow-up of 38 months, 83 cases remained untreated, while 67 cases received treatment. We show that these markers predict the clinical course by using time to first treatment (TTT) as a measure for disease progression. Each of the three markers was capable of discriminating two distinct groups of patients (p < 0.0001 for CD38, p < 0.00001 for ZAP-70 and Ig VH mutations) with different clinical behavior, although marker combinations provided a more precise definition of prognosis. Although many patients expressed all favorable or all unfavorable markers, there also were patients with different marker combinations. We devised a scoring system that subdivides patients based on the absence (score 0) or presence of 1 (score 1), 2 (score 2), or 3 (score 3) unfavorable prognostic markers. Using this scoring system, we have identified 3 groups with significantly different clinical courses: i.e., low- (score 0), intermediate-risk (score 1) and high-risk (score2–3) patients. This scoring system has potential utilization for prognostic stratification of CLL in designing prospective clinical trials.