Prognostic indexes have been developed recently to assess time to first treatment in early-stage chronic lymphocytic leukemia (CLL) patients. However, none of five indexes evaluated in a study showed more than a moderate prognostic value or were precise enough to permit clinical decisions to be made, according to a report by Spanish researchers.
Their study, published in Clinical Lymphoma, Myeloma and Leukemia, examined the comparative prognostic value of five prognostic indexes – the CLL-IPI, the Barcelona-Brno, the IPS-A, the CLL-01, and the Tailored approach – on evaluating 428 Binet A CLL patients from a multicenter Spanish database which contained the relevant necessary clinical and biological information. The predictive value of the scores was assessed with Harrell´s C index and receiver operating characteristic curve (area under the curve, AUC).
The researchers found a significant association between time to first treatment and risk subgroups for all the indexes used. The most accurate index was the IPS-A (Harrell´s C, 0.72; AUC, 0.76), followed by the CLL-01 (Harrell´s C, 0.69; AUC, 0.70), the CLL-IPI (Harrell´s C, .69; AUC, 0.69), the Barcelona-Brno (Harrell´s C: 0.67, AUC, 0.69) and the Tailored approach (Harrell´s C, 0.61 and 0.58, AUCs, 0.58 and 0.54).
However, the concordance between four of the five indexes (the Tailored approach was not included for technical reasons) compared was low (44%): 146 cases were classified as low risk with all four indexes tested, 36 as intermediate risk, and 4 as high risk. In the remaining 242 patients (56%) at least one discrepancy was detected in the allocation among prognostic subgroups between the indexes. However, only 12 patients (3%) were allocated as low and high risk at the same time with different indexes, showing the extremes of the discordance.
These data suggest that, although all of these indexes “significantly improve clinical staging and help physicians in routine clinical practice, it is necessary to harmonize larger cohorts of patients in order to define the best index for treatment decision making in the real world,” the authors stated.
“All the models had a moderate prognostic value to predict time to first therapy. ... None of them was precise enough to allow clinical decisions based exclusively on it,” the authors concluded.
The study was supported by grants from the Spanish government and a variety of nonprofit institutions. The authors reported no commercial disclosures.
SOURCE: Gascon y Marín IG et al. Clin Lymphoma Myeloma Leuk. 2020 Apr 13. doi: 10.1016/j.clml.2020.03.003.