Is the International Prognostic INDEX for CLL (CLL-IPI) Useful to Predict Time to First Treatment of Patients with EARLY Disease? Results of a Prospective Multicenter Analysis

BACKGROUND: The CLL-IPI  score is  a large  cooperative effort in which clinical data collected from 8 randomized trials were used to develop an internationally applicable prognostic index for CLL patients.   The model includes 5 independent parameters that predict for overall survival such as  age, clinical stage, del(17p) and/or TP53 mutation, IGHV mutation status and β2-microglobulin (B2M) level. A potential limitation for an extensive use of CLL-IPI is represented, however, by the fact that only 20% of patients included in the full analysis set had early  disease.

PATIENTS: The present analysis based on an observational multicenter CLL database  including 337 Binet stage A patients (O-CLL1 protocol, clinicaltrial.gov

identifier NCT00917540) was designed to assess the utility of the CLL-IPI score to predict time to first treatment (TTFT) in patients with early disease.

RESULTS: Patients were followed up for a total of 2038 person-years (median, 42 months; range, 1–82 months), during which 91 (26.9%) experienced disease-progression requiring therapy according to 1996 IWCLL guidelines. The CLL-IPI score enabled Binet stage A patients to be divided into three subgroups [i.e.,  score 0-1, low-risk (n=229); score 2-3, intermediate-risk(n=99); score  4 or higher, high-risk (n=9)]   that differed with respect to TTFT (P<0.0001). A comparative performance analysis between CLL-IPI and 2007 MD Anderson Cancer Center (MDACC) score, barely based on traditional clinical parameters (i.e., age, gender, Rai substage, absolute lymphocyte count, number of involved

nodal groups and B2M), revealed that prediction of the TTFT was more accurate with the former. The c-statistic of the MDACC model was 0.62 (95% CI: 0.49-0.75)  a level below than that of the CLL-IPI (c=0.70; 95% CI:0.58-0.81) and below the accepted 0.7 threshold necessary to have value at the individual patient level. These results are in keeping with the change in area under the receiver operating characteristic (ROC) curve (AUC) which increased from 0.646 (95% CI: 0.578-0.714) to 0.720 (95%CI:0.658-0.783) when moving from MDACC model to CLL-IPI score .

Since the CLL-IPI score was originally derived from patients with active CLL enrolled in phase 3 trials we sought for different cut-off scores that  better predict for TTFT  in our patient cohort  of early CLL. According to the recursive partitioning (RPART) analysis, a classification tree was built that identified three subsets of patients who scored  0 (low- risk,n=139), 1(intermediate-risk,n=90) and >1 (high-risk,n=108), respectively.  The probability of remaining free from therapy at 5 years was 85% in the low-risk group, 68% in the intermediate-risk group and 47% in the high-risk group (P<0.0001)(Fig 1). Our revised IPI score remained a predictor of TTFT also when analysis was limited to 262 Rai stage 0 (P<0.0001) and 99 clinical monoclonal B-cell lymphocytosis (cMBL) cases (P=0.006).

CONCLUSIONS: The results of this study confirm the utility of CLL-IPI score for predicting TTFT in a prospective cohort of community-based patients with early  CLL at presentation. Our effort to adapt CLL-IPI score to patients with early disease meets the need to separate Binet stage A patients into different prognostic groups suitable for individualized follow-up programmes and possibly for early therapeutic interventions.