BOTH MD Anderson Cancer Center MODEL and German Score Work in Predicting Time to First Treatment in EARLY Chronic Lymphocytic Leukemia: Results of an External Validation Analysis

Abstract 3934

In chronic lymphocytic leukemia (CLL) there has been a limited published work on the development of models predicting time to first treatment that incorporate traditional and novel prognostic factors. We wondered whether the nomogram recently proposed by M D Anderson Cancer Center (MDACC) investigators that includes clinical and newer prognostic variables (i.e., three or more involved lymph node sites, increased size of cervical lymph nodes, presence of 17p deletion or 11q deletion by FISH, increased serum LDH, and unmutated IGHV mutation status) could work when applied to a prospective community-based CLL patient cohort.

We considered suitable for the purpose of this study 328 previously untreated patients with newly diagnosed asymptomatic Binet stage A CLL from different primary hematology centers registered on a prospective basis during the period 2006–2010 on an observational database of Gruppo Italiano Studio Linfomi (GISL).

Using the formula proposed by Wierda et a total point score was calculated for each patient. A value ranging between 0 and 69.5 (median, 19.9) - therefore similar to total point scores reported by the MDACC group (median, 21.0; range, 0 – 87.4) – was found. As a continuous variable (i.e., measuring the risk of each point increase), total point score was associated with time to first treatment (HR 1.04; 95% CI 1.02–1.05; P < 0.0001). Receiver Operating Characteristic analysis identified a point score of 25 [area under curve (AUC) 0.64, sensitivity 61.5 specificity 72.1; P<0.0001)] as the best threshold capable of separating patients who needed therapy from patients who did not (HR, 3.27; 95% CI, 2,07–5.18; P<0.0001). The prognostic index category remained a predictor of time to first treatment also when analysis was limited to patients in Rai stage 0 (HR,4.05; 95% CI: 2.25–7.52; P<0.0001). Finally, the goodness of fit showed that the nomogram model had significant good fit at 2 years (r²=0.966; P=0.002).

We next wondered whether the German score based on 7 prognostic variables (i.e., age, gender, b2-microglobulin, sTK, IGVH mutational status, ECOG performance status and hierachic type of FISH abnormalities) and originally designed for predicting overall survival could forecast time to first treatment in patients with early disease. Since sTK was not available a slightly modified version of German model based on 6 prognostic variables was used. The risk score calculated in each patient by summing single ratings of adverse factors, allowed the identification of two different risk categories scoring respectively 0–2 and 3–5 whose clinical outcomes were different with respect to time to first treatment (HR 2.19, 95% CI: 1.15–2.20; P=0.005).

In conclusion, this analysis shows that both models proposed by MDACC investigators and German group provide interchangeable tool for assessing time to first treatment. Interestingly, in this validation analysis the pattern of distribution of patients into a low and high-risk category revealed a high degree of concordance between models (P<0.0001; Mann-Whitney test).