Chronic Lymphocytic Leukemia: which prognostic factor to choose?
Stamatopoulos B, Meuleman N, Haibe-Kains B, Duvillier H, Massy M, Martiat P, Bron D, Lagneaux L
Background:
Chronic Lymphocytic Leukemia (CLL) has an extremely variable clinical
course with overall survival time ranging from months to decades. For
some patients, the disease runs an indolent clinical course and life
expectancy is not shortened; for others, the disease is aggressive,
progresses rapidly and survival after diagnosis is decreased to 2-3
years. Therefore it is very important to identify factors that can
predict poor prognostic and also identify patients who will benefit
from intense therapy in an early stage. These two different groups in
terms of overall survival and clinical characteristics were classified
for a long time on Binet Stage and more recently on the IgVH mutational
status that seems to be one of the most robust biological prognostic
factors. However, this costly analysis is very laborious and
time-consuming. Therefore, many surrogate markers have been
investigated. Finally, among all these factors, one question remains:
which prognostic factor to choose?
Methods: We compared the most
commonly used prognostic factors (Binet Stage, IgVH mutational status,
Zap-70, CD38 and LPL expression) in a cohort of 113 patients with a
median follow-up of 82 months to evaluate their association with
overall survival (OS) and treatment-free survival (TFS). Flow cytometry
(FC) and quantitative PCR (qPCR) on purified CD19+ cells were used.
Association of surrogate markers with IgVH mutational status (using
chi_ Pearson and Cramer's V statistic), optimal cut-off values of
Zap-70, LPL and CD38 that best distinguished between mutated and
unmutated cases (evaluated with ROC curve analysis), power of
prognostic marker after one and two years after diagnosis (evaluated
with ROC time-dependent curves), OS and TFS distributions (using
Kaplan-Meier estimates and the log-rank test) and finally the impact of
the different prognostic factors on TFS and/or OS (evaluated with
univariate and multivariate Cox regression analysis with binarised
date) were performed.
Results: All prognostic factors
tested were associated with IgVH mutational status but Zap-70 measured
by qPCR [P<0.0001] was characterised by the higher Cramer's V
statistic (0.72) indicating a very strong relation. This method also
presents the best 87.8% sensitivity, 85.7% specificity, 87.5% positive
predictive value and 86% negative predictive value. The concordance
rate between Zap-70 and IgVH mutational status were largely higher than
other factors (78% and 86% respectively for Zap-70 by FC and qPCR). All
prognostic factors were significant TFS predictor (regarding log-rank
test and univariate Cox regression) but only IgVH mutational status
[P=0.0034] and Zap-70 [by both methods: FC, P=0.0006; qPCR, P=0.0021]
were significant OS predictors. For example, Zap-70-positive patients
had a significantly shorter median TFS (24 months) than Zap-70-negative
patients (157 months). Moreover, in case of discordance with IgVH
mutational status, only Zap-70 by qPCR was associated with TFS
[P=0.0395]. Multivariate Cox regression including Zap-70 (by qPCR
or by FC, LPL by qPCR, mutational status and CD38 expression indicated
also that Zap-70 [by qPCR:P=0.038, by FC:P=0.005] was more powerful to
predict TFS than the classical mutational status and the other markers
tested. ROC time-dependent curves were also generated to evaluate the
power of all markers tested at one and two years after diagnosis:
Zap-70 expression (by both methods) Area Under the Curve (AUC) was
higher than the other prognostic factors including IgVH mutational
status. For example, 2 years AUC was 0.83 for Zap-70 by qPCR, 0.84 for
Zap-70 by FC while this values was 0.77, 0.69, 0.66 respectively for
IgVH mutational status, LPL and CD38 expression.
Conclusions: Regarding all
these analysis, we conclude that Zap-70 is the most powerful prognostic
factor and the best surrogate of IgVH mutational status among all
factors tested. The choice of the method to measure Zap-70 is more
complicated but the qPCR method is more accurate, can offset FC
limitations, is strongly associated with IgVH mutational status,
prevalent on this status in case of discordance, and in case of
discordance with Zap-70 by FC, Zap-70 by qPCR shows a clear trend to be
prevalent. Therefore we recommend the use of Zap-70 measured by qPCR as
prognostic factor.