Objectives: Trough total imatinib (t-IM) concentrations have been reported to be associated with therapeutic and toxic responses in patients with chronic myelogenous leukemia (CML) and gastrointestinal stromal tumor (GIST). Little is known about the relationships between effects and concentrations of either unbound imatinib (f-IM) or imatinib's major metabolite, N-desmethyl imatinib (NDI). In part, this is because of the lack of a single, validated, well-described clinically useful assay for these measurements. The authors report the development and application of such an assay.
Materials and methods: A single liquid-chromatography tandem-mass-spectrometry assay was used to monitor t-IM, f-IM, and t-NDI concentrations in CML and GIST patients treated at a tertiary German teaching hospital. The assay was also validated for measuring other kinase inhibitors, including t-nilotinib, sunitinib, and erlotinib. Ultrafiltration assays were validated and used to measure f-IM and to compare free fractions to plasma α1-acid glycoprotein concentrations (AGP).
Results: The assays were linear over a working range (in micrograms per liter) of 8.4-8370, 8.3-4165, and 1.0-250 and had within- and between-run coefficient of variance of <7%, <12%, and <9% for t-IM, t-NDI, and f-IM, respectively. The f-IM assay was reproducible despite high (25.2%-31.6%) but concentration-independent binding to ultrafiltration devices. Clinically relevant results, such as nondetectable (ND) t-IM (<8.4 μg/L) in non-responders and >1500 μg/L in patients with major toxicity, were found. Of 156 total samples from 68 adult CML patients and 127 total samples from 42 adult GIST, only 48 samples from 22 CML patients and 40 samples from 20 GIST patients were trough samples with adequate dosing and collection information. More than half (27 of 48 CML and 24 of 40 GIST) had t-IM concentrations ≥10% below recommended target concentrations (1002 μg/L for CML and 1100 μg/L for GIST). Concentrations >50% over targets were also found in 6 of 48 CML and 4 of 40 GIST samples. Wide variations in concentrations of t-IM (range, ND to 2973 μg/L), t-NDI (range, ND to 659 μg/L), f-IM (range, 8.3-262 μg/L), and t-IM:f-IM ratios (range, 2.6%-14%) were found both between and within patients. A statistically significant association (Spearman correlation coefficient and P value for all samples, r = 0.290 and P = 0.023; for trough only, r = -0.585 and P = 0.028) was found between AGP and f-IM concentrations but wide interpatient and intrapatient variations made individual predictions unreliable.
Conclusions: The liquid-chromatography tandem-mass-spectrometry methods developed provided information useful to understand individual responses to therapy even though necessary sampling and dosing information was often not available. Wide unpredictable variations in t-IM, t-NDI, and f-IM were found. Clinical outcome trials are needed to examine whether f-IM or NDI monitoring can improve the ability to predict individual responses.