Portal de Periódicos da CAPES

Comparison between the CLL-IPI and the Barcelona-Brno prognostic model: Analysis of 1299 newly diagnosed cases

2017; Wiley; Volume: 93; Issue: 2 Linguagem: Inglês

10.1002/ajh.24960

1096-8652

Massimo Gentile, Tait D. Shanafelt, Francesca Romana Mauro, Luca Laurenti, Davide Rossi, Stefano Molica, Iolanda Donatella Vincelli, Giovanna Cutrona, Giuseppina Uccello, Sara Pepe, Ernesto Vigna, Giovanni Tripepi, Kari G. Chaffee, Sameer A. Parikh, Sabrina Bossio, Anna Grazia Recchia, Idanna Innocenti, Raffaella Pasquale, Antonino Neri, Manlio Ferrarini, Gianluca Gaïdano, Robin Foà, Fortunato Morabito,

Gastrointestinal Tumor Research and Treatment

To the Editor: In the last two decades, numerous clinical, serological, and biologic markers have been identified that are significantly associated with the prognosis of chronic lymphocytic leukemia (CLL) patients.1 Recent research has focused on the development of scoring systems capable of integrating the major prognostic parameters, applicable in the daily clinical practice, and of improving risk stratification. The CLL International Prognostic Index (CLL-IPI), based upon TP53 deletion and/or mutation, IGHV mutational status, β2M plasma concentration, clinical stage, and age led to the identification of four patients subsets (low-, intermediate-, high-, and very high-risk) with different overall survival (OS) and time to first treatment (TTFT).2 This score has been externally validated (Supporting Information Table S1).3, 4 Recently, a Barcelona-Brno CLL consortium has proposed a prognostic model which is comprised of two biomarkers only (IGHV mutational status and FISH cytogenetics) with the aim of simplifying the CLL-IPI (Supporting Information Table S1).5 This led to the identification of three patients categories (low-, intermediate-, and high-risk) with a different prognostic likelihood.5 Herein, we compared the CLL-IPI score and that of the Barcelona-Brno CLL consortium in an Italian-American cohort of patients whose prognostic parameters were obtained at diagnosis. A total of 1299 untreated CLL patients were included in this analysis (Supporting Information Appendix and Figure S1). The median age was 63 years (range 27–92); 61.3% were males and the majority of patients presented at Rai stage 0 (57.9%; Supporting Information Table S2 lists baseline patient features). After a median follow-up of 6.83 years (range, 3 months to 27.5 years), 283 patients have died and 510 received chemotherapy [194 (38%)] or chemoimmunotherapy [316 (62%)] as first-line treatment. We first evaluated the capacity of the CLL-IPI score to predict OS. Due to missing data regarding TP53 mutations, del17p was used as the sole marker of TP53 status. All of the selected markers showed an independent prognostic impact on OS (Supporting Information Table S3). According to the CLL-IPI, 51.3% of patients (n = 667) were classified as low-, 28.7% (n = 373) as intermediate-, 16.2% (n = 210) as high-, and 3.8% (n = 49) as very high-risk. Stratification of patients according to the CLL-IPI showed significant differences in terms of OS (Figure 1A). The median, 5-year and 10-year OS rates by prognostic index category in our cohort are quite similar to those observed in the original study,2 suggesting that the survival estimates provided by the CLL-IPI are reproducible. The C-statistic was 0.73 (P < .0001) for OS prediction, exceeding the 0.70 threshold and underscoring the prognostic utility at the individual patient level. OS of the entire population of 1299 CLL patients according to the CLL-IPI score (A), and to the MDACC prognostic index (B) Moreover, the CLL-IPI score was also useful to predict TTFT (Supporting Information Figure S2A), with a C-statistic of 0.73 (P < .0001) confirming the validity of the CLL-IPI in predicting progression for individual patients.2 According to the Barcelona-Brno prognostic model, 58.1% of our patients (n = 755) were classified as low-, 31.8% (n = 413) as intermediate-, and 10.1% (n = 131) as high-risk (Figure 1B). In addition, all parameters of the Barcelona-Brno prognostic model (IGHV mutational status, del17p and del11q) showed an independent predictive power on survival (Supporting Information Table S3). The Harrell C-index of this prognostic model was lower than that of the CLL-IPI score [0.65 vs. 0.73] (P < .001) Furthermore, the Akaike information criterion (AIC) showed the superiority of the CLL-IPI compared with the Barcelona-Brno prognostic model in predicting OS (CLL-IPI, AIC = 3432.167 versus the Barcelona-Brno prognostic model, AIC = 3549.492; the lower AIC value represents the better fit). Accordingly, the explained variation in mortality provided by the CLL-IPI was 42% (P < .001), a value higher than that of the Barcelona-Brno prognostic model (21%, P < .001), indicating that the CLL-IPI had a higher prognostic accuracy for mortality compared with that of the biomarkers-only prognostic model (Supporting Information Table S4). Next, we compared the ability of the two scores to predict TTFT. The Harrell C-index of the Barcelona-Brno prognostic model was lower than that of the CLL-IPI score (0.70 vs. 0.73, P < .001; Supporting Information Figure S2B). The AIC values also demonstrated that CLL-IPI was superior to the Barcelona-Brno prognostic model in predicting TTFT (AIC = 5960.503 vs. AIC = 6010.929). Accordingly, the explained variation provided by the CLL-IPI was 33% (P < .001), a figure higher than that achieved by the Barcelona-Brno prognostic model (28%, P < .001) (Supporting Information Table S4). Our results confirm the value of the CLL-IPI score for survival prediction of previously untreated CLL patients. The Barcelona-Brno prognostic model has some prognostic utility, but shows C-statistic values below the accepted 0.7 threshold required for significance at the individual patient level. In addition, the CLL-IPI score also seems to be better suited to predict progression than the Barcelona-Brno prognostic model in our cohort. The additional variables included in CLL-IPI and not in the Barcelona-Brno model comprise universally available clinical characteristics (age and stage) along with an inexpensive and near universally available serum marker (beta-2-microglobulin). Although our analyses demonstrate the superiority of the CLL-IPI score compared with the Barcelona-Brno prognostic model, some limitations must be considered. First, the comparative study could be carried out on 1299 of the 4491 patients available, that is, in one third of cases recorded in our cohort, because of the absence of the required laboratory data in the cohort initially considered. However, the patients included were representative of the whole cohort and were similar for age, sex, and Rai stage distribution. Second, the length of this study spans over a 30-year period. In consideration of the therapeutic advances in recent years, the value of the proposed score methodology will have to be reassessed in prospective studies where the new drugs are utilized to treat patients. Finally, when del11q was forced in a multivariate model together with all parameters of CLL–IPI in our cohort (data not shown), this marker showed an independent prognostic value. This is probably due to the fact that roughly 40% of treated patients in our series received chemotherapy, rather than chemo-immunotherapy, which overcomes the negative impact of del11q.6 Despite these limiting factors, our study has important strengths, such as the large sample size, the community-based cohort of patients and the method for patient recruitment, which prevented biases related to patients' selection. The data so far collected points to CLL-IPI as the most suitable method for CLL prognostication. The authors have no conflict of interest to disclose. All authors read and approved the final version of the manuscript. Designed the study: Massimo Gentile, Tait D. Shanafelt, Francesca R. Mauro, Luca Laurenti, Davide Rossi, Stefano Molica, Antonino Neri, Manlio Ferrarini, Gianluca Gaidano, Robin Foà, Fortunato Morabito Analyzed and interpreted data: Massimo Gentile, Tait D. Shanafelt, Francesca R. Mauro, Luca Laurenti, Davide Rossi, Stefano Molica, Antonino Neri, Manlio Ferrarini, Gianluca Gaidano, Robin Foà, Fortunato Morabito Wrote the manuscript: Massimo Gentile, Tait D. Shanafelt, Francesca R. Mauro, Luca Laurenti, Davide Rossi, Stefano Molica, Antonino Neri, Manlio Ferrarini, Gianluca Gaidano, Robin Foà, Fortunato Morabito Performed statistical analysis: Massimo Gentile, Giovanni Tripepi, Kari G. Chaffee, Fortunato Morabito Performed central laboratory tests: Giovanna Cutrona, Sabrina Bossio, Anna Grazia Recchia, Antonino Neri, Manlio Ferrarini. Provided the patients and collected clinical data: Giuseppina Uccello, Iolanda Vincelli, Ernesto Vigna, Sameer A. Parikh, Raffaella Pasquale, Idanna Innocenti, Sara Pepe. Additional Supporting Information may be found online in the supporting information tab for this article. Supporting Information Figure 1 Supporting Information Figure 2 Supporting Information Table 1 Supplementary Appendix Contents Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.