European LeukemiaNet criteria for failure or suboptimal response reliably identify patients with CML in early chronic phase treated with imatinib whose eventual outcome is poor

Abstract

The majority of patients with chronic myeloid leukemia in chronic phase gain substantial benefit from imatinib but some fail to respond or lose their initial response. In 2006, the European LeukemiaNet published recommendations designed to help identify patients responding poorly to imatinib. Patients were evaluated at 3, 6, 12, and 18 months and some were classified as "failure" or "suboptimal responders." We analyzed outcomes for 224 patients with chronic myeloid leukemia in chronic phase treated in a single institution to validate these recommendations. Patients were followed for a median of 46.1 months. At each time point, patients classified as "failure" showed significantly worse survival, progression-free survival, and cytogenetic response than other patients; for example, based on the assessment at 12 months, the 5-year survival was 87.1% versus 95.1% (P = .02), progression-free survival 76.% versus 90% (P = .002), and complete cytogenetic response rate 26.7% versus 94.1% (P < .001). Similarly, the criteria for "suboptimal response" at 6 and 12 months identified patients destined to fare badly, although criteria at 18 months were less useful. The predictive value of some other individual criteria varied. In general, the LeukemiaNet criteria have useful predictive value, but a case could now be made for combining the categories "failure" and "suboptimal response."

Figure 1

Classification of 224 patients according to criteria for failure at 3, 6, 12, and 18 months and their final outcome. Numbers underlined (3 and 9) reflect patients with a follow-up of less than 18 months. Numbers in italics represent patients who changed categories.

Figure 2

Probabilities of PFS and CCyR according to the criteria for failure at 3, 6, 12, and 18 months. Failure patients are represented by the dashed line and nonfailure patients by the continuous line. PFS curves start from 100%, and CCyR curves start from 0%. The number of patients in each category, the precise numerical values for the probabilities of PFS and CCyR, and the P values are shown in Table 2. Vertical lines represent censored patients (but the corresponding lines are not shown in the CCyR curves). At 18 months, 100% of the patients in the nonfailure group were in CCyR (by definition); thus, the curve is not shown.

Figure 3

PFS and probability of CCyR for patients who met criteria of failure at 3, 6, 12, or 18 months and for patients who met criteria of failure and suboptimal response at 3, 6, or 12 months. (A) The PFS and probability of CCyR for patients who met the criteria of failure at 3, 6, 12, or 18 months (dashed line) compared with those patients who never met criteria for failure (continuous line); the 5-year PFS was 63.8% versus 90.8% (P < .001), and the 5-year probability of CCyR was 46.3% versus 100% (P < .001). (B) The PFS and probability of CCyR for patients who met either criteria of failure or suboptimal response at 3, 6, or 12 months (dashed line) compared with those patients who did not meet criteria for failure or suboptimal response at 3, 6, and 12 months; the 5-year PFS was 70.4% versus 95.9% (P < .001), and the probability of CCyR was 51.4% versus 100% (P < .001). Vertical lines represent censored patients (but the corresponding lines are not shown in the CCyR curves). In the 18- and 12-month values, 100% of the patients in the responding groups were in CCyR (by definition).

Figure 4

Twelve- and 18-month landmark analysis for loss of CCyR according to the level of molecular response. Vertical lines represent censored patients.