Adherence is the critical factor for achieving molecular responses in patients with chronic myeloid leukemia who achieve complete cytogenetic responses on imatinib

Abstract

Purpose: There is a considerable variability in the level of molecular responses achieved with imatinib therapy in patients with chronic myeloid leukemia (CML). These differences could result from variable therapy adherence.

Methods: Eighty-seven patients with chronic-phase CML treated with imatinib 400 mg/d for a median of 59.7 months (range, 25 to 104 months) who had achieved complete cytogenetic response had adherence monitored during a 3-month period by using a microelectronic monitoring device. Adherence was correlated with levels of molecular response. Other factors that could influence outcome were also analyzed.

Results: Median adherence rate was 98% (range, 24% to 104%). Twenty-three patients (26.4%) had adherence <or= 90%; in 12 of these patients (14%), adherence was <or= 80%. There was a strong correlation between adherence rate (<or= 90% or > 90%) and the 6-year probability of a 3-log reduction (also known as major molecular response [MMR]) in BCR-ABL1 transcripts (28.4% v 94.5%; P < .001) and also complete molecular response (CMR; 0% v 43.8%; P = .002). Multivariate analysis identified adherence (relative risk [RR], 11.7; P = .001) and expression of the molecular human organic cation transporter-1 (RR, 1.79; P = .038) as the only independent predictors for MMR. Adherence was the only independent predictor for CMR. No molecular responses were observed when adherence was <or= 80% (P < .001). Patients whose imatinib doses were increased had poor adherence (86.4%). In this latter population, adherence was the only independent predictor for inability to achieve an MMR (RR, 17.66; P = .006).

Conclusion: In patients with CML treated with imatinib for some years, poor adherence may be the predominant reason for inability to obtain adequate molecular responses.

Fig 1.

Six-year probability of major molecular response (MMR), 4-log reduction in transcript levels, and complete molecular response (CMR) in the 87 enrolled patients according to the measured adherence rate. The probability of MMR for the 23 patients with an adherence rate ≤ 90% was 13.9%, whereas the probability was 93.7% for the 64 patients with an adherence rate greater than 90% (P < .001). Similarly, the probability of a 4-log reduction was 4.3% versus 76% (P < .001), and the probability of CMR was 0% versus 43.8% (P = .002).

Fig 2.

Six-year probability of major molecular response (MMR) and 4-log reduction in the transcript level in the 32 patients who had their dose of imatinib increased according to the measured adherence rate. Higher adherence rates were associated with achievement of MMR (relative risk [RR], 1.105; P = .008) and 4-log reduction (RR, 1.095; P = .026). Only three patients receiving imatinib 600 mg achieved complete molecular response (CMR), so we did not perform an analysis for this outcome. The 14 patients with an adherence rate ≤ 90% had a lower 6-year probability of MMR and 4-log reduction in transcript levels than the 18 patients with greater than 90% adherence (7.1% v 80% [P = .002] and 0% v 36.3% [P = .01]). Similar results were achieved when we considered other cutoff points for the adherence rate. The 6-year probabilities of MMR were 18.6% for the 17 patients with an adherence rate ≤ 95% versus 88.2% for the 15 patients with a rate greater than 95% (P < .001); 9.1% for the 11 patients with an adherence rate ≤ 85% versus 67.9% for the 21 patients with a rate greater than 85% (P = .006); and 0% for the seven patients with an adherence rate ≤ 80% versus 60.7% for the 25 patients with a rate greater than 80% (P = .02).

Fig A1.

Comparison of the three methods utilized to measure adherence. MEMS, microelectronic monitoring systems.