Prognostic and Predictive Factors in Patients With Chronic Lymphocytic Leukemia: Relevant in the Era of Novel Treatment Approaches?

Abstract

In modern oncology, we are experiencing a paradigm shift toward personalized medicine. Conceptually, this means that patients will be treated with compounds that specifically target the tumor on the basis of its individual molecular characteristics. Personalized targeted therapy requires the identification of (bio)markers for at least two reasons: first, patients will need to be stratified according to their potential to respond to targeted therapy, and second, response to therapy will need to be assessed by analyzing the molecular target of the therapy. At present, arguably the most direct and frequently used biomarkers are the underlying genetic aberrations of a tumor. In chronic lymphocytic leukemia (CLL), genetic markers are well known for their pathogenic and prognostic relevance. CLL is suitable as a model disease for personalized medicine for several reasons: the clinical course is heterogeneous in different patients; malignant cells are accessible; the disease is prevalent, albeit only in the Western population; large international consortia have comprehensively characterized the genetic aberrations at high resolution; and patient characteristics like clinical parameters (eg, age, comorbidity) and biologic parameters (eg, genomic aberrations, mutations) are diverse. With respect to the heterogeneous clinical course, some patients die within 2 to 3 years with refractory disease, whereas others live for decades after diagnosis without need for therapy. To address this heterogeneity and predict the prognosis of patients, Rai and Binet developed a staging system in the 1970s that is still one of the mainstays of clinical management. One of the first achievements to robustly complement the Binet/Rai classification scheme was the development of interphase fluorescent in situ hybridization, which allowed the routine detection of genetic aberrations in noncycling CLL cells (as reviewed by Oscier ). These different genetic aberrations could stratify patients with CLL into a hierarchy of five prognostic subgroups [del(13q), normal karyotype, tris(12), del(11q), and del(17p)] with decreasing survival times, a paradigm for the role of genetic lesions in the clinical course of a malignancy. Another genetic marker is the physiological rearrangement and somatic hypermutation of the variable region of the immunoglobulin heavy chain genes (IGHV, mutated and nonmutated). CLLs with mutated and nonmutated IGHV not only reflect a possibly different cellular origin but also display a significant difference in prognosis. This suggests that B-cell receptor (BCR) signaling is central to CLL pathogenesis, which has opened up novel therapeutic options (Fig 1). Current treatment includes DNA-damaging chemotherapy that results in the development of resistance in virtually all patients on several treatment courses and actually selects for high-risk genetic alterations. The current paradigm is therefore to treat patients only when they are symptomatic (with so-called active disease). Resistance against chemotherapy frequently correlates with loss of function of the DNAdamage response pathways, in CLL as well as in most other cancers with TP53 and ATM genes localized in 17p and 11q, respectively (Fig 1). Thus, TP53 is the only biomarker in CLL that currently drives treatment decisions. Functionality of TP53 can be lost either by genomic deletion or by point mutation, highlighting the important role of single base pair changes. The prevalence of CLL and the gap in understanding of the underlying pathogenesis led to prioritization of the disease in the search for point mutations in the International Cancer Genome Consortium. In this combined effort, point mutations have been identified in CLL, the vast majority of which are not recurrent, however. The suitability of CLL as a model for the development of genetic biomarkers is also the relative genetic stability of the disease. The most common genetic mutations were detected in NOTCH1, SF3B1, and MYD88. The presence of these mutations is of prognostic value with many but not all treatment modalities. Therefore, it has been suggested that these point mutations should be incorporated into the hierarchic cytogenetic model. In fact, in a recent landmark study, the mutational heterogeneity itself, along with the presence of subclonal mutations that are indicative of genomic instability, has been proposed to be used as a biomarker. Last but not least, soluble serum markers like thymidine kinase and s2microglobulin levels have repeatedly been shown to correlate with prognosis in CLL. In addition, two novel classes of biomarkers have been identified in CLL that are also relevant other tumor entities: first, the activity of microRNA (miRNA) genes, and second, epigenetic aberrations. CLL has been the first entity for which miRNAs have been described as tumor suppressors (miR15a and miR16-1). Arguably one of the most interesting tumor-associated miRs (not only in CLL) is miR34a. Although the miR15/16 family are probably causative in CLL at least in a subset of patients, miR34a is of interest with respect to a large number of tumor entities because of its functional connection with JOURNAL OF CLINICAL ONCOLOGY E D I T O R I A L VOLUME 32 NUMBER 9 MARCH 2