Improving outcome prediction in indolent ATL by subdividing intermediate risk into low- and high-intermediate groups.

Abstract

The indolent adult T-cell leukemia-lymphoma prognostic index (iATL-PI) uses soluble interleukin-2 receptor (sIL-2R) levels of 1,000 and 6,000 U/mL as a cut-off. The disadvantage of the iATL-PI is that approximately half of the patients are classified as intermediate risk (1,000 ≤ sIL-2R < 6,000). Here, we aimed to develop a novel prognostic model for indolent ATL using a prospectively registered database. We identified 375 patients with indolent ATL. Median age was 61 years. The median follow-up of surviving patients was approximately 62 months. In multivariate analysis for overall survival (OS), sIL-2R level (as log(sIL-2R), HR 4.58), male sex (HR 1.41) and LDH > ULN (HR 1.68) were independent prognostic factors. The best cut-off value for sIL-2R to predict OS was 2,870 U/mL. Including the cut-off of 3,000 U/mL in the original iATL-PI, the probabilities of 5-year OS were 88.9% in the low group (sIL-2R < 1,000), 67.5% in the low-intermediate group (1,000 ≤ sIL-2R < 3,000), 34.0% in the high-intermediate group (3,000 ≤ sIL-2R < 6,000) and 17.0% in the high group (sIL-2R ≥ 6,000). The c-index for predicting OS was higher with the modified iATL-PI than with the original iATL-PI (0.749 vs. 0.725). The addition of sex and LDH levels did not improve the c-index. In conclusion, we developed a novel prognostic index of indolent ATL.