Predicting response in CML

Predicting response in chronic myeloid leukemia (CML) patients who are offered tyrosine kinase inhibitor (TKI) second-line therapy is vitally important. For younger patients, allogeneic transplantation is still a therapeutic option Thus, physicians need new models to predict the outcome of their patients.

In this issue of Blood, Jabbour et al propose a simple scoring system easily applicable in the clinic.¹  The analysis was based on a group of 123 patients with chronic-phase CML who failed imatinib therapy and who were switched to dasatinib or nilotinib. Their model identified 2 variables that were significantly associated with event-free survival. In a multivariate analysis, the lack of any cytogenetic response to previous imatinib therapy and Eastern Cooperative Oncology Group performance status of 1 or more at the start of second-generation TKI therapy were identified as independent adverse factors associated with poor event-free survival. The model thus subdivides patients into 3 categories: good risk (0 factor), intermediate risk (1 factor), and poor risk (factors) associated with 2-year, event-free survival of 78%, 49%, and 20%. Of note, the 2-year overall survival was also significantly associated with the risk categories. Moreover, Jabbour et al performed additional multivariate analyses using a 12-month landmark. They stated that the lack of major cytogenetic response at 12 months after the onset of the second TKI is an adverse independent factor for event-free survival. Although the definition of “event”-free survival has to be carefully considered, the general finding is in accordance with the definition of the response to the second-generation TKIs dasatinib and nilotinib as second-line therapy of patients with imatinib-resistant CML–chronic phase (CML-CP) which was proposed by the new European Leukemia Net recommendation.² 

Less-than-major molecular response at 12 months was considered (at least provisionally) a suboptimal response and less-than-partial cytogenetic response as failure. Using this approach, the authors confirmed that a performance status of 1 or more is a poor prognosis factor for the survival. However, other clinical and/or biological parameters could still be of interest. First, prognostic scores, such as the Sokal score, which relates to the pathophysiology of the disease at diagnosis, may still be of some value even later on the course of their disease. For example, intermediate- and high-risk patients are at higher risk of molecular relapse after cessation of imatinib.³  The Hammersmith group recently proposed a score including 3 factors associated with the achievement of responses after second TKI therapy: that is, the best cytogenetic response on imatinib, the occurrence of neutropenia, and the Sokal score at diagnosis.⁴  Second, mechanisms of resistance to imatinib are not unique. Consequently, some parameters may be predictive of resistance to second generation of TKIs in specific subgroups of patients and not in others.

Early retrospective data showed a high incidence of imatinib noncompliance in CML patients which could lead to undesired clinical outcomes. The ADAGIO (adherence assessment with gleevec: indicators and outcomes) study⁵  evaluated adherence to imatinib in 169 CML patients and found that during the initial 90-day period of imatinib treatment, one-third of patients were considered to be non adherent. Only 14.2% of patients were compliant with all prescribed doses of imatinib. Thus, nonadherences could be an issue with second TKI. Chronic adverse events of even grade 1 or 2 must also to be considered, essentially because these side effects may explain non adherence to treatment. In these 2 situations, predictors of response to second TKI may be somewhat different from those of patients who received imatinib using prescribed dosages.

Point mutations in the BCR-ABL kinase domain, which are frequently involved in TKI resistance, may be an important determinant in clinical decisions. Several recent reports suggested that routine mutation screening could provide valuable information regarding the selection of the optimal TKI and could also identify patients at high risk of disease progression.⁶  Thus, such biologic abnormalities should be considered in parallel with the new score.

Clonal cytogenetic abnormalities and elevation in BCR-ABL transcript levels could be investigated for further models because some studies showed that elevations in BCR-ABL transcript levels might indicate a potential for BCR-ABL gene mutations and emergence of TKI resistance.⁷ 

Identification of biomarkers to predict resistance to TKIs is currently in progress. For example, gene array data on blast cells or the CD34-enriched cell population from chronic phase have provided interesting information.^8,9  Although a pretreatment molecular signature has been identified for imatinib-treated patients, such a signature could serve as a molecular biomarker for stratifying patients treated with any other TKI into risk group. Moreover, recent data provided evidence that BCR-ABL mutation leads to kinase activation, suggesting that this mechanism may extend beyond activation loop mutations.¹⁰  Finally, it would be helpful if several international CML groups could select a large independent cohort of patients to validate these new scoring systems and select the one that would produce the most accurate information for the monitoring of CML patients.