Multivariate Analysis of Prognostic Factors in CLL: A Study on 431 Patients Showing Usefulness of Novel Biological and Old Clinical Parameters in Predicting Shorter Survival: An Italian Multicentric Study

Abstract

B-cell chronic lymphocytic leukaemia (B-CLL) is a heterogeneous disease with highly variable clinical courses. Recently, besides classical clinical parameters, including Rai/Binet's staging system, novel biological factors acquired prognostic relevance especially for risk-adapted therapeutic strategies: IgVH mutational status, ZAP-70 and CD38 expression, chromosome aberration (del13q, trisomy 12, del11q, del17p) detected by FISH analysis. In a recent report, Wierda et al constructed a prognostic nomogram predictive for overall survival based on clinical and routine laboratory characteristic of 1674 CLL patients: age, beta2 microglobulin, absolute lymphocyte count, sex, Rai stage and number of involved lymph node groups. Biological prognostic features were not considered in this model but the authors concluded that “future works will focus on incorporation of new prognostic factors into this prognostic model. Actually, unlike other haematological malignancies, there is no a standard Prognostic Index that can be used to group CLL patients according clinical and new biological prognostic parameters and to guide the choice of treatment. To assess which factors give independent contribution in predicting shorter survival we collected 784 patients data from 4 laboratories. All variables were measured at diagnosis or before start of therapy; they included: age above 65 years, male sex, Rai stage >0, beta2 microglobulin >2.5 mg/L, absolute lymphocyte count (ALC), involvement of 3 or more nodal groups, high risk FISH cytogenetics (17p- or 11q-), IgVH mutational status, ZAP-70 and CD38 expression. Survival was measured from diagnosis until death from any cause. A sample of 431 patients was obtained after deleting cases with missing values. Median follow-up was 60 months (54–65). There were 63 deaths (15%) during follow-up, with a not reached median survival at 250 months. During follow-up 43% of patients were treated. Cox proporzional hazard model was used for univariate and multivariate analysis. P <0.05 was considered to be statistically significant. Multivariate model was developed including those variables statistically significant in univariate analysis and further variables were eliminated from the final model by backward selection. IgVH, ZAP-70, CD38, ALC, male sex were not significant in univariate analysis. In multivariate analysis the following variables predicted for shorter survival (table 1): age>65 years, Rai stage>0, beta2>2.5mg/L, high risk FISH. In a larger subset of patients (568), selected with all variables except beta2 microglobulin, the above results were confirmed, in particular no independent contribution was given by IgVH, ZAP-70 and CD38, even if they turned out significant in univariate analysis. Beta2 microglobulin and high risk FISH or only high risk FISH retained prognostic power respectively in a subset of 198 patients under 65 years of age and in a subset of 252 patients with Rai stage 0. In conclusion age>65 years, B2 microglobulin> 2,5 mg/L, Rai stage >0, and high risk FISH abnormalities (del 17p or 11q) will be considered by clinician because of independent prognostic factor of OS. A nomogram will developed to predict for survival using the independent covariates identified in multivariate analysis. This weighted prognostic model will permit to obtain a total point score predictive for median survival of each patient, based on clinical and biological characteristic.Table 1: MULTIVARIATE ANALYSIS (COX MODEL, p = Wald test)Variableshazard ratio (confidence interval)p-valueAge>652.19 (1.26–3.79)0,0052Nodal sites involved >= 3rrBeta-2 microglobulin >2.5mg/L1.87 (1.06–3.28)0,03Rai stage > 02.13 (1.26–3.62)0,0051FISH 17p- or 11q-2.41 (1.42–4.08)0,0011r = removed by bacward selection