A Comprehensive Progression Risk Score to Predict Treatment Free Survival for Early Stage Chronic Lymphocytic Leukemia Patients

Background: Over the last two decades several phenotypic, molecular, and chromosomal markers have been identified that are significantly associated with the prognosis of CLL patients. Therefore, clinicians managing CLL patients would benefit from a simplified prognostic index.

Methods: We analyzed prospectively collected data from 337 Binet A CLL patients enrolled in Italian the O-CLL1-GISL protocol with the aim of developing scores capable of predicting treatment free survival (TFS). Factors independently associated with TFS were included in the prognostic indexes. To account for differences in the magnitude of the association between the individual independent factors and TFS, we assigned a weighted risk score to each factor based on ranges of their corresponding hazard ratios (HRs) (i.e., 1 point for HR 1.1-1.9; 2 points for HR 2.0-2.9, etc.). The total risk score was then calculated by the sum of the ratings of each factor on individual basis. Risk groups were identified combining risk categories with a non-statistically different TFS.

Results: We developed two scores based on weighted multivariable models: the first included clinical and laboratory parameters [clinical score (c-score)], while the second was based on biological markers [biological score (b-score)] (Table 1). The c-score allowed to predict the TFS of patients through the combination of Rai stage, b2-microglobulin and absolute lymphocyte count (ALC), while the b-score predicted TFS by IGHV mutational status and CD38 expression. The c-score showed a C-statistic of 0.72, while the b-score was 0.67, although cases stratified according to the b-score showed a more specific mRNA/microRNA profile. When the two scores were forced in a multivariate analysis, both showed an independent predictive value on TFS with a similar HR, demonstrating their complementarity. Thus, we attempted to integrate the two scores performing a further multivariate analysis in which all parameters, significantly associated with TFS at univariate analysis, were tested (Table 1). ALC, Rai stage, b2-microglobulin together with IGHV mutational status, resulted independently associated with TFS. We constructed a weighted score [comprehensive score (co-score)], including all the above 4 variables, which allowed the identification of 3 different risk groups with significantly different TFS (Figure 1). The C-statistic of the g-score was 0.75, showing a better concordance than the other two scores. Moreover, its validity was externally validated in a series of 297 newly diagnosed Binet A CLL patients from the Mayo Clinic, USA.

Conclusions: Using this multistep process and external validation, we developed a score with high discriminatory power and predictive significance on the individual patient level.

Table 1. Univariate and multivariate Cox proportional Hazards Models

Variable	Univariate analysis		Multivariate analysis
			Clinical model			Biological model			Comprehensive model
	HR (95% CI)	P	HR (95% CI)	P	score	HR (95% CI)	P	score	HR (95% CI)	P	score
Age (years) <60/≥60	1.12 (0.73-1.74)	0.59	-	-	-	-	-	-	-	-	-
Sex Male/Female	0.93 (0.6-1.44)	0.93	-	-	-	-	-	-	-	-	-
Rai stage 0/I-II	2.30 (1.47-3.50)	<0.0001	2.13 (1.24-3.03)	0.004	0/2	-	-	-	1.76 (1.11-2.78)	0.015	0/1
ALC (109/L) <10/≥10	3.43 (1.99-5.92)	<0.0001	2.91 (1.85-5.20)	<0.0001	0/2	-	-	-	2.70 (1.54-4.72)	0.001	0/2
b-2 microglobulin normal/elevated	3.04 (1.96-4.70)	<0.0001	2.78 (1.79-4.30)	<0.0001	0/2	-	-	-	2.65 (1.66-4.21)	<0.0001	0/2
LDH normal/elevated	1.25 (0.57-2.71)	0.57	-	-	-	-	-	-	-	-	-
CD38 negative/positive	3.22 (2.06-5.02)	<0.0001	-	-	-	2.11 (1.15-3.16)	0.02	0/2	1.40 (0.80-2.42)	0.24	-
ZAP-70 negative/positive	2.34 (1.51-3.61)	<0.0001	-	-	-	1.21 (0.70-2.16)	0.485	-	1.0 (0.98-1.01)	0.72	-
IGHV mutated/unmutated	3.57 (2.32-5.50)	<0.0001	-	-	-	2.10 (1.12-3.90)	0.019	0/2	2.39 (1.27-4.50)	0.007	0/2
FISH risk low+int/high	2.93 (1.46-5.90)	0.002	-	-	-	1.65 (0.76-3.34)	0.216	-	1.80 (0.84-3.88)	0.13	-

Abbreviations: ALC: absolute lymphocyte count; CI: confidential interval; HR: hazard ratio.

Figure 1. TFS according to comprehensive progression risk score.