Relationship of imatinib-free plasma levels and target genotype with efficacy and tolerability

Abstract

Imatinib has revolutionised the treatment of chronic myeloid leukaemia (CML) and gastrointestinal stromal tumours (GIST). Using a nonlinear mixed effects population model, individual estimates of pharmacokinetic parameters were derived and used to estimate imatinib exposure (area under the curve, AUC) in 58 patients. Plasma-free concentration was deduced from a model incorporating plasma levels of alpha(1)-acid glycoprotein. Associations between AUC (or clearance) and response or incidence of side effects were explored by logistic regression analysis. Influence of KIT genotype was also assessed in GIST patients. Both total (in GIST) and free drug exposure (in CML and GIST) correlated with the occurrence and number of side effects (e.g. odds ratio 2.7+/-0.6 for a two-fold free AUC increase in GIST; P<0.001). Higher free AUC also predicted a higher probability of therapeutic response in GIST (odds ratio 2.6+/-1.1; P=0.026) when taking into account tumour KIT genotype (strongest association in patients harbouring exon 9 mutation or wild-type KIT, known to decrease tumour sensitivity towards imatinib). In CML, no straightforward concentration-response relationships were obtained. Our findings represent additional arguments to further evaluate the usefulness of individualizing imatinib prescription based on a therapeutic drug monitoring programme, possibly associated with target genotype profiling of patients.

Figure 1

Relationship between free drug exposure (AUC_u) and toxicity in CML (upper part) and GIST patients (lower part). Left panel: scatter plot of AUC_u according to side effects score (0=no side effects, 1=1 side effect, 2=2 side effects and 3=3 or more side effects). Right panel: probability of side effects according to the per-sample PK–PD analyses. The histograms represent the percentages observed for the three types of response at three typical AUC_u range values (side effects score: white box=0; light grey box=1; grey box=2; dark grey box=3). The curves, modelled by a four-level ordered logistic regression, show the probability of side effects according to AUC_u.

Figure 2

Relationship between free drug exposure (AUC_u) and response in GIST patients. Upper part: exon 11 KIT genotype; lower part: exon 9 or wt KIT genotype. Left panel: scatter plot of AUC_u according to RECIST score; white box=PD+SD (score 0; n=23 for exon 9/wt, 46 for exon 11); grey box=OR, OR=CR+PR (score 1; n=10 for exon 9/wt, 32 for exon 11). Right panel: probability of response according to the per-sample PK–PD analysis for both main genotypes of GIST patients. The histograms represent the percentages observed for the two types of response at three typical AUC_u range values. The curve, modelled by a two-level ordered logistic regression, shows the probability of response according to AUC_u.