Long-Term Prognostic Value of Hematological and “Old” Biological Parameters in CLL. Reflexions on 156 Patients with a Median Follow-Up Time More Than 6.5 Years.

Waiting for a standardization of “modern” biological prognostic factors in CLL, such as mutational status, ZAP-protein, CD38… we review with a long median delay of survey, the prognostic value of classical hematological data and “old” biological parameters such as seric LDH, β2-microglobuline, sCD23 and sCD25. At time of diagnosis our 156 pts were aged from 39 to 80 yrs (median=66 yrs) and sex-ratio shows as usually a male predominance (M/F=1.40). According to Binet’s system we note 131 stages A (95 stages A0), 14 stages B and 11 stages C. With a median follow-up time more than 6.5 yrs, median survival time is 9.5 yrs and median “corrected” (i.e., without not CLL-related deaths including those due to a solid tumor because this complication is independent of initial prognostic factors of CLL) survival time is 10.2 yrs. When looking at “corrected” survival, best cut-off values for “old” biological factors are 1.25 N for LDH, 3 mg for β2-m, 33 U (=10 N) for sCD23 and 2282.5 U (=1.25 N) for sCD25. Cox model with these 4 pejorative factors (PF) retains sCD23 (p=0.00004), β2-m (p=0.00002) and LDH (p=0.001). When adding these 3 PF, we can build a very strong prognostic system: median survival time of the 117 pts with 0 or 1 PF is more than 11 yrs versus 4 yrs for the 39 pts with 2 PF or more (p<10⁻⁶). Among hematological parameters, Cox model retains 3 of them: platelets level at 2 cut-off values 150 and 100 G (p=0.000005 and 0.02); 3 or more lymphoid area according to Binet’s classification (p=0.0002) and hemoglobin level at cut-off value 120 g (p=0.03). DLT is also a strong prognostic factor (p= 0.002) but this value was missing for 47 pts. However Cox model done with DLT retains 3 factors: DLT (p=0.0003); hemoglobin at cut-off value 100 g (p=0.0003) and platelets at cut-off value 150 G (p=0.001). Cox model including “old” biological PF and classical hematological parameters without DLT, retains sCD23 (p=0.0003), platelet at cut-off value 150 G (p=0.0005) and 3 or more lymphoid areas (p = 0.003). Finally we test 3 clinical classifications (Rai, Binet and our local classification based on clinical examination, hemoglobin and platelet levels, and medullary examination) and 2 biological classifications (LDH+β2-m system and this new PF system). Cox model retains the sCD23+LDH+β2-m system (p=0.00009) and our local system (p=0.003) and when looking at global survival our prognostic system seems more performing than PF system: p=0.0007 and 0.001. Addition of LDT does not modify our results but this parameter is kept in Cox model too.

We conclude on the necessity to clearly determine the best cut-off values for all prognostic parameters, classical and “old” and “new” biological ones. We also claim the usefulness to compare all prognostic factors of CLL together because “old” clinical system and DLT are perhaps more useful for clinical practice when patients are out of a prospective therapeutic trial.