Trough plasma imatinib threshold for major molecular response (MMR).(A) Receiver operating characteristic (ROC) curve analysis. Regarding trough imatinib plasma levels and their discrimination potential for MMR, the area under the ROC curve (AUC) was 0.775, with best sensitivity (77%) and specificity (71%) at a plasma threshold of 1002 ng imatinib/mL. (B) Box plots of trough imatinib plasma levels. The graph shows the dispersion around the median for patients with MMR (n = 34; median = 1350 ng/mL) and those without (n = 34; median = 885 ng/mL). The line across each box is the median; the bottom edge is the first quartile and the top edge is the third quartile; the error bars represent minimal and maximal values; the bottom line shows the 493.6 ng/mL (ie, 1 μmol/L) target concentration required to result in BCR-ABL–positive cell death in vitro; the top line shows the 1002 ng/mL efficient plasma threshold for trough imatinib levels in vivo. Of the 34 patients with MMR, 26 (76%) had trough imatinib plasma levels exceeding the 1002 ng/mL threshold. Of the 34 patients without MMR, 24 (71%) had trough imatinib plasma levels below the 1002 ng/mL threshold, whereas 27 patients (79%) had trough imatinib plasma levels exceeding the initially described target concentration (493.6 ng/mL) required to result in BCR-ABL–positive cell death in vitro.^8,15,16  This initial target was validated in phase 1 to 2 studies on the basis of cytogenetic criteria, but it is not always sufficient to achieve MMR in vivo. Indeed, using a sharper criterion than CCR such as a 3 log reduction of BCR-ABL transcript levels, the efficient plasma threshold for trough imatinib levels should be set above 1002 ng/mL in vivo.