In this issue of Blood, Pflug et al describe the new prognostic score published by the German Chronic Lymphocytic Leukemia Study Group (GCLLSG).1
Accurate prognostication is essential in modern medicine. For nearly 40 years, prognostication for patients with chronic lymphocytic leukemia (CLL) used simple measures of disease bulk and residual normal bone marrow function, such as the Rai2 and Binet stage. In order to provide improved discriminatory power for patients within these traditional disease stages, an expanding and at times bewildering list of novel prognostic markers in CLL has emerged.3 Many of these markers were developed and validated in small, retrospective, and heterogeneously treated cohorts of patients, often without consideration of potential association or interaction with other markers.
The new prognostic score published by the GCLLSG1 took these new developments into account by exploring the relative contributions of disease stage, disease biology, and patient-related factors such as age and fitness in a large cohort of patients enrolled onto 3 prospective studies with mature follow-up. In total, 1223 treatment-naïve patients with complete risk profiles (including immunoglobulin heavy chain [IgHV] mutation status and fluorescent in situ hybridization [FISH] studies4 ) were selected from 2068 patients enrolled onto an early intervention protocol (CLL1: Binet stage A CLL, in which patients with high-risk features were randomly assigned to observation or fludarabine) and two treatment protocols (CLL4: fludarabine vs fludarabine-cyclophosphamide [FC] or CLL8: FC vs FC-rituximab). A multivariate model identified 8 factors as being independently associated with inferior survival from study entry. The hazard ratio for each of these factors then formed the basis for a new prognostic score (see figure). The authors acknowledged that this score was derived from a mixed population of patients not requiring treatment (CLL1) and those about to embark on initial therapy (CLL4 and CLL8), and they undertook additional analyses to show that the same prognostic factors applied to both populations. Finally, the prognostic score was validated in an independent cohort of newly diagnosed patients managed at the Mayo Clinic.
The elimination of Rai and Binet stage from the prognostic model underscores how far our understanding of CLL biology has evolved. Rather than determining survival by measuring disease bulk, survival in CLL is now determined by molecular genetic features (p53 or 17p deletion, IgHV unmutated status, and 11q deletion4 ), serum markers of disease proliferation (thymidine kinase and β2-microglobulin5 ), and host factors (age, ECOG performance status, and male sex). In clinical practice and treatment decision making, the most important subgroups are the 25% of patients in the low-risk group, with 5-year survival of 95% and median time to first therapy of >10 years, and the 4% of patients in the very-high-risk group, with a dismal 5-year survival of 19%. Of note, the presence of 17p deletion is required for patients to be classified as very high risk, but even within this p53-deleted subgroup, patients with relatively favorable survival can be identified (eg, those patients without concomitant adverse features and those who do not require immediate therapy). Thus, much of the adverse impact of del(17p) is due to the relative ineffectiveness of current (fludarabine-based) therapy in combating this form of CLL.
What are the caveats against and barriers to the immediate adoption of this prognostic score into clinical practice? First, the score depends on the ability to assay serum thymidine kinase,5 a prognostic marker that is not routinely measured outside of Germany. Modern methods such as chemiluminescence-based assays6 may facilitate the adoption of this test by diagnostic and commercial laboratories. Second, the understanding of the genomic landscape of CLL is rapidly evolving, and as data regarding novel mutations such as those in NOTCH1, SF3B1, and BIRC3 accumulate,7 their roles will need to be re-evaluated in the context of this prognostic model. Additionally, most diagnostic and commercial laboratories test for 17p deletion by FISH, but few currently have the ability to screen for p53 mutations. As the availability of p53 mutation screening increases in clinical practice, the prognostic score will require recalculation to determine whether p53 mutations have an adverse impact as substantial as that of 17p deletion.8 The authors also acknowledge a degree of patient selection and underrepresentation of older patients from the studies used to generate and validate this index. Finally, the treatment landscape of CLL is changing with the advent of highly active oral small molecules which can be delivered effectively, even in older patients and in those with p53-aberrant CLL9,10 ; in time, the applicability of the prognostic score in the era beyond chemotherapy will need to be re-evaluated.
Conflict-of-interest disclosure: The authors declare no competing financial interests.