Molecular Profiling in CLL

Chronic lymphocytic leukemia (CLL) has a highly variable clinical course. Approximately 25% of patients require therapy at diagnosis due to bone marrow failure or to symptoms such as bulky adenopathy, organomegaly, fatigue or B-symptoms such as fevers, night sweats, weight loss or extreme fatigue. There is a wide range of initial presenting features, most commonly painless lymphadenopathy, followed by splenomegaly and or hepatomegaly. Only 5% of patients present with lymphadenopathy without evidence of leukemic infiltration, and in this situation the disease is known as small lymphocytic lymphoma (SLL). CLL is increasingly diagnosed in asymptomatic patients when a lymphocytosis is found at the time of a routine blood count. The international workshop on Chronic Lymphocytic Leukemia (iwCLL) revised guidelines require a lymphocytosis of greater than 5000/μl maintained for more than 3 months with the cells expressing the diagnostic immunophenotype for the diagnosis of CLL.¹ The presence of fewer than 5000 B-lymphocytes/μL in the absence of lymphadenopathy is defined as “monoclonal B-lymphocytosis” (MBL). Individuals diagnosed at an early stage may remain asymptomatic for the rest of their lives, with their lifespan unaffected by CLL. Others may progress to aggressive disease over time. More than 2000 patients with early disease have been enrolled in trials of immediate versus deferred chemotherapy, all performed using alkylating agents; meta-analysis of these studies showed no survival benefit of early versus deferred therapy.² Treatment guidelines state that therapy should be reserved for those with advanced, symptomatic or progressive disease (Table 1 ),¹ with treatment being considered palliative due to the incurable nature of the disease with conventional chemotherapeutic agents and the often advanced age of the patient. There has been significant improvement over the past decade in the results of treatment of CLL with the use of combination chemotherapy and chemo-immunotherapy (Table 2 ). The detailed results of the German CLL Study Group (GCLLSG)–led study comparing the use of fludarabine and cyclophosphamide with rituximab, fludarabine and cyclophosphamide are awaited, but the trial has successfully met its primary endpoint since analysis has shown that patients treated with chemoimmunotherapy achieved a significant improvement in progression-free survival (PFS) compared to patients treated with chemotherapy using fludarabine and cyclophosphamide alone.

In parallel with this improvement in treatment outcome there has been dramatic progress in the understanding of the basic biology of CLL and the development of a number of prognostic factors, which have largely been validated as powerful determinants of time from diagnosis to time to requirement for treatment (Table 3 ). The Rai and Binet staging systems were based upon their prognostic significance and the stage of disease remains perhaps the most important prognostic factor in CLL. Although there is strong evidence linking prognosis with biologic markers, it is by no means clear how we should use these factors in the management of patients with CLL.

Understanding the pathogenesis of CLL/SLL remains a major challenge. Unlike many of the other low-grade B cell malignancies, non-random reciprocal chromosomal translocations occur rarely in CLL/SLL. However, a number of cytogenetic abnormalities have been identified. Using fluorescent in situ hybridization (FISH) techniques, one or more of these cytogenetic abnormalities can be found in more than 80% of patients. In a comprehensive study chromosomal abnormalities were detected by FISH in 268 of 325 CLL patients studied (82%).³ This study demonstrated convincingly that genomic aberrations in CLL are important independent predictors of disease progression and survival. The most common recurrent chromosomal abnormalities observed include deletion of the long arm of chromosome 13 (del 13q), del 11q, trisomy 12, del 17p and del 6q.^3,4 The most common abnormality is del 13q.14, which occurs in more than 50% of cases. The first report linking microRNAs to cancer was in CLL, where it was demonstrated that two microRNA clusters, mir-15a and mir16-1,⁵ were located within the deleted region at 13q14.⁶ The next most common cytogenetic abnormality is del 11q, seen in up to 20% of cases of CLL. This deletion is associated with a distinct clinical presentation including younger age, male sex, bulky lymphadenopathy and poor prognosis. The ATM gene is located within the minimal region of loss at 11q23, suggesting that alterations in this gene may be involved in the pathogenesis of the disease. This is further supported by the finding that mutations in the ATM gene are associated with poor prognosis.⁷ Trisomy 12 occurs in up to 20% of cases of CLL/SLL, but the molecular mechanism by which this genetic abnormality contributes to leukemogenesis is unknown. Although less common and occurring in less than 10% of patients at diagnosis, del 17p is associated with rapid progression of disease, poor response to therapy and short survival. The deletion involves the p53 locus at 17p13, and it is clear that mutations in the p53 gene can contribute to disease progression and alter the sensitivity of CLL cells to chemotherapy agents. Many patients’ CLL cells exhibit multiple cytogenetic abnormalities and there is a hierarchical structure.³ The detection of del 17p or del 11q is associated with poor risk, while del 13q as a sole abnormality is associated with good-risk disease. Ongoing studies are assessing the impact of specific cytogenetic abnormalities in response to particular therapeutic approaches.

An important advance in the understanding of CLL was made with the demonstration that 50% of CLL cases have somatic hypermutation in the variable regions of the genes and that this has prognostic significance, with cases with somatic hypermutation having a more indolent clinical course and longer survival than those without somatic hyper-mutation.^8,9 The levels of somatic hypermutation in particular B cells is evaluated by comparison of the sequence of the rearranged variable region gene with germline sequences, and guidelines have been reported for analysis of IgVH rearrangements from the working group of European Research Initiative in CLL guidelines.¹⁰ Sequences with less than 98% homology to germline are considered to have undergone somatic hypermutation. The finding that CLL cases can be divided into mutated and unmutated groups implied that the two groups may be diseases that arise from different normal cellular counterparts, but this was not supported by subsequent gene expression profiling studies.^11,12 

Analysis of variable region sequences demonstrated that CLL cells utilize a biased repertoire of V genes characterized by over-representation of selected Ig gene segments, in particular IGHV1-69, IGHV4-34, IGHV3-7, and IGHV3-21.^13,14 Somatic hypermutation does occur uniformly among IGHV genes: for example, IGHV1-69 consistently carries very few mutations as opposed to the typically mutated IGHV3-7, IGHV3-23, and IGHV4-34 genes. An apparent exemption to the generalization that mutated CLL cases have good prognosis is in the subgroup of patients with CLL cells that use IGHV3-21 since these patients have relatively aggressive disease even when the expressed IGHV3-21 is mutated.¹⁵ 

Not only is the immunoglobulin gene repertoire expressed by CLL cell biased, it is also notable for the existence of subsets with near identical (stereotyped) B-cell receptors (BCRs), implying the recognition of structurally similar epitopes, likely selecting the leukemic clones.¹⁶ The nature of the antigens that these BCRs might be recognizing and whether these are important in driving the pathogenesis of CLL remains unknown. The presence of such stereotypic rearrangements may also have prognostic significance.^17,18 

It is not possible to perform IgVH mutational status on a routine basis in clinical laboratory laboratories and attempts have therefore been made to identify surrogate markers for mutational status. In particular, expression of two proteins, ZAP70 and CD38, which both have prognostic significance, have been examined. CLL cells demonstrate a continuum of expression of these proteins, and it is necessary to determine a cut-off point at which a case is deemed to be positive or negative, leading to difficulties in standardization since different laboratories assess individual cases as being positive or negative for expression.

When gene expression profiles were analyzed comparing mutated and unmutated cases of CLL,^11,12 only a small number of genes were found to be differentially expressed, the most specific being the gene encoding the 70 kDa zeta-associated protein (ZAP70).¹⁹ Most mutated cases are ZAP-70 negative and unmutated cases ZAP70 positive. ZAP70 expression can be measured by number of methods including western blotting, reverse transcriptase-PCR, immunohistochemistry, and flow cytometry.^20,–23 Levels of expression are higher in T cells and NK cells than in CLL cells, and it is important that effective gating strategies are used to ensure that expression is being measured in the CLL cells. ZAP70 expression appears to be stable over time.^22,24 Studies have demonstrated that there is not an absolute relationship between ZAP70 expression and IgVH mutational status, with discrepant cases ranging from 8% to 25%.^22,24 These discordant cases may have other biological features with poor prognostic implications such as del 17p, del 11q or use of IGHV3-21.²⁵ Some studies have suggested that ZAP70 status is more useful as a predictor of time to progression than mutation status,^22,26 but this remains controversial.

CD38 is a surface marker associated with CLL, and easily determined using standard flow cytometric methods. It was initially found to correlate with IgVH mutation status 9; however, the relationship is not absolute, and CD38 expression may vary over time.^27,28 The field is somewhat confused by a variety of cut-offs ranging from 5% to 30% being used in different series to define a case as being CD38,^28,–30 and it has been suggested that CD38 should be evaluated by its modal expression, by flow cytometry, or by antigen density. Other surrogates of mutation status have been suggested including expression of thymidine kinase, activation-induced cytidine deaminase, lipoprotein lipase A and ADAM29.^31,–33 MicroRNA arrays have revealed a 13-gene signature found to correlate with ZAP70 status and unmutated IgVH expression³⁴ and disease progression.³⁵ Recent work has suggested that this altered microRNA expression regulates expression of genes regulating apoptosis and cell-cycle progression.³⁶ 

The molecular profile of CLL provides insight into the underlying pathogenesis of the disease and provides predictors of time to progression to need for therapy and overall survival. High-risk features include the high-risk cytogenetic features del 11q and del 17p, IgVH unmutated status, use of the IGHV3-21 gene segment, expression of ZAP70 and expression of CD38. It is tempting to speculate that these markers can now be useful in clinical practice, but a number of questions remain (Table 4 ). There is no evidence to date that if a patient presents with “high-risk” disease features that there is any benefit in offering therapy earlier. This question is now being addressed in ongoing and planned clinical trials; until the results of these studies are available, patients should not be offered treatment on the basis of any molecular marker until the standard criteria for treatment are reached.¹ 

With the finding that a number of molecular markers have prognostic significance it is not surprising that many of these factors are correlated. However, there are discrepancies, with many cases having some high-risk and other low-risk molecular features, and more than 50% of IgVH unmutated cases have no unfavorable cytogenetics.²⁵ There is an association between unfavorable cytogenetic aberrations (del 17p and del 11q) and unmutated CLL, while 13q- is more frequent in mutated CLL. Multivariate analysis identified IgVH mutational status, poor-risk cytogenetic abnormalities, white blood count and lactate dehydrogenase were identified as independent risk factors, and, when these factors were incorporated into models, clinical staging (using either the Rai and Binet staging systems) loses independent prognostic value.⁴ A recent study involving more than 1000 CLL patients examined the relative value of ZAP70, CD38 and IgVH mutation status and found that ZAP70 expression was the strongest predictor of time from diagnosis to requirement for treatment.³⁴ It may be that high-risk cytogenetics, IgVH mutational status, ZAP70 and CD38 provide complementary prognostic information, with expression of both markers conferring a poor prognosis, lack of both a good prognosis, and discordant expression an intermediate prognosis.^22,25,37 

Most of the modern prognostic markers were validated by retrospective analysis, often from single-center studies, but have now been applied to prospective randomized clinical trials. These studies suggest that the same molecular markers that identify patients with more aggressive disease also impact on outcome after treatment. This finding is not surprising since these same factors have been predictive of overall survival in retrospective studies, where it would have been expected that the same treatment options would have been offered to patients with and without risk factors. As shown in Table 5 , three studies have been published examining the impact of these factors on response in prospective randomized trials in previously untreated patients with CLL,^38,–40 and these results have been confirmed in a number of other studies that have been reported in abstract format only. These findings suggest poor-risk features for CLL are largely also predictive of poor response. There is not yet sufficient evidence to alter therapy based upon molecular features, but the one exclusion from this is the group of patients who present with del 17p. These patients have poor response to chemotherapy and impaired survival. Although this represents only a small group of previously untreated patients, these patients should be treated preferably in clinical trials, examining agents that have efficacy in patients without functional p53.

Finally, does the practicing clinician benefit from obtaining these tests in routine clinical practice? Several of the factors cannot readily be obtained including IgVH mutational status and the current assays for expression of ZAP70 are inconsistent, with no clear guidelines on the established methodology or where the cut-off should be for designation as ZAP70 positive or negative. There is considerable ongoing debate regarding the clinical utility of CD38 expression and its stability over time. Cytogenetics assessed by FISH appears to be robust and reproducible and provides prognostic information for the clinician and patients. Since cytogenetic abnormalities also change over time with the evolution of new genetic changes, it may make more sense to perform this analysis at the time of institution of therapy. Clinicians may be better served using cheaper and more established markers of disease including lactate dehydrogenase and beta-2 microglobulin, which can be incorporated into nomograms to assess risk of progression.⁴¹ Therefore, decision making in clinical practice should be made on the basis of clinical features of the disease, and the use of molecular profiles in the management of CLL remains a clinical trial question only. The only clear exception to this at present is in symptomatic patients with 17p del, in whom efforts should be made to treat these patients with agents that may act independently of p53.