A better response was indeed observed in accelerated and blast phases of CML with 600mg daily (Talpaz et research only al, 2002), and a 800mg daily regimen allowed a longer progression-free survival in GIST patients (Verweij et al, 2004), whereas this was not the case with 600mg (Blanke et al, 2008). The inverse relationship initially observed in our PK�CPD analysis (for both CML and GIST patients) between Dose/AUC and therapeutic response could be considered paradoxical. However, as our study was purely observational, we were in the presence of good responders selected to receive low doses and bad responders high doses, but without apparent advantage. In GIST patients, Dose was indeed highly correlated with AUC and CL, confirming the presumption of such a bias.
Conversely, in the CML sub-population, the lower the clearance of the unbound drug, the better was the response, suggesting that CLu was a better predictor of effect than AUC/AUCu. Most CML patients were apparently exposed to sufficient drug amounts to achieve a haematological response (i.e. ceiling of the concentration�Ceffect curve), making them partly obscure the PK�CPD relationship. It has indeed been reported that imatinib doses of 350mg (corresponding to a trough plasma concentration, TPC of 570��gl?1) already ensure an optimal haematological response in CML (Peng et al, 2004). Such an amount could, however, not be sufficient for a cytogenetic or molecular response, which appears to require TPC as high as 1000��gl?1 (Picard et al, 2007; Larson et al, 2008).
Moreover, the design of our study wherein AUC derived from sparse measurements were used as an index of exposure may have prevented us from observing similar results as in the IRIS study (steady-state imatinib TPC at initiation of therapy in patients on 400mg QD predicts long-term complete cytogenetic and major molecular responses) (Larson et al, 2008). As the PK�CPD relationship for a targeted agent such as imatinib may be confounded by genotypic heterogeneity of intracellular pharmacological targets (BCR-ABL and c-KIT, respectively), the mutational status of BCR-ABL was also assessed in our CML population by DNA sequencing. However, no point mutations known to confer resistance were observed (data not shown).
Conversely, focusing on GISTs allowed us to uncover a relationship between free drug exposure and response when integrating the target mutation profile (with higher Batimastat drug exposure predicting better response, and being a superior predictor than the mutation status). Of importance, the inclusion of SD in the OR score did not significantly affect the correlations observed. Imatinib-free plasma levels thus appeared a better predictor of drug effect than total levels. This is in line with previous data showing that the total plasma concentration of imatinib is a poor marker of imatinib clinical effect (Delbaldo et al, 2006).