Introduction

Imatinib mesylate (Glivec®/Gleevec®/Imatinib, Novartis Oncology, East Hanover, NJ) is an orally bioavailable tyrosine kinase inhibitor that acts on BCR-ABL, as well as the DDR, KIT, and PDGFR kinases [1–3]. Imatinib has markedly improved survival outcomes and become a treatment of choice in patients with advanced, unresectable gastrointestinal stromal tumors (GIST) [4–7]. The standard dose of imatinib is 400 mg once daily, but an increase to 800 mg/day has shown additional clinical benefits in patients with disease progression on 400 mg/day [8–10].

Imatinib is rapidly and almost completely (98%) absorbed from the gastrointestinal tract, with most (>95%) binding to plasma proteins such as albumin and α1-acid glycoprotein (AGP) [11]. Imatinib is metabolized by cytochrome P450 (CYP) enzymes, primarily CYP 3A4 and 3A5, with a number of influx and efflux transporters thought to be responsible for intracellular drug concentration [11]. Pharmacokinetic (PK) studies in patients with chronic myeloid leukemia (CML) and GIST have found that variables significantly affecting or associated with the plasma concentration of imatinib include white blood cell (WBC) and granulocyte counts; hemoglobin and AGP concentrations; age; body weight and body surface area (BSA); and previous major gastrectomy (i.e., subtotal or total gastrectomy) [12–17].

Satisfactory clinical outcomes in patients with CML and GIST require sufficient exposure to imatinib [18–20]. In patients with GIST, imatinib plasma trough concentrations (Cmin) greater than 1,100 ng/mL were correlated with higher rates of objective response and longer progression-free survival [20].

An early population PK study of patients with GIST showed that imatinib clearance was 33% higher after about 12 months of therapy than on day 1, suggesting that exposure to imatinib decreases with extended administration [12]. If this is true, since median progression in GIST patients during first line treatment with imatinib occurs at about 2 years [6, 8], disease progression in patients initially controlled on imatinib may be at least in some patients due to a gradual decrease in imatinib exposure over time. Moreover, frequent monitoring of plasma imatinib concentration and fine tuning of imatinib dosage may be necessary to maintain the long-term effects of imatinib. In a previous study, however, we did not observe a correlation between duration of exposure to imatinib and imatinib Cmin [16]. Because that study was limited by its cross-sectional design, we assessed changes in imatinib Cmin during long-term exposure.

Materials and methods

Patients

Since October 2008, steady-state imatinib Cmin has been measured in patients ≥18 years of age at Asan Medical Center, Seoul, Korea, with histologically and/or molecularly documented GIST, who showed good compliance and were treated for at least 2 weeks with a constant dose of imatinib. Of patients whose plasma concentration of imatinib had been measured (initial PK tests), we selected those who received the same dose of imatinib for at least 9 months and performed follow-up (FU) measurements of plasma imatinib concentration. Patients with any serious comorbidity or who were administered any concomitant medications that could inhibit or induce CYP 3A enzymes were excluded. Before the initial and FU PK measurements, patients were asked to take the same dose of imatinib at a fixed time for at least 7 days and to keep a drug-use diary. Patients who did not comply with this request were rescheduled for PK tests. Body weight, body surface area (BSA), complete blood counts, serum chemistry and serum creatinine concentration were measured within 7 days of each PK assessment. Creatinine clearance was estimated using the Cockcroft-Gault formula: estimated creatinine clearance = (140-age) × (weight in kilograms) × (0.85 if female)/(72 × serum creatinine) [21]. Patient medical records were retrospectively reviewed to collect clinicopathologic data. The study protocol was approved by the Institutional Review Board of the Asan Medical Center, and all patients provided written informed consent.

PK data collection

Within 22–26 h after the previous dose of imatinib, blood samples (at least 4 mL) were collected into heparinized tubes, centrifuged at 3,000 rpm for 10 min at room temperature and stored at −20°C within 1 h of collection. Plasma concentrations of imatinib were measured by liquid chromatography-tandem mass spectrometry at the Seoul Clinical Laboratories (Seoul, Korea), with the kind support of Novartis Korea [22]. Two measurements of imatinib Cmin were obtained on two different days, with the average value used for analysis; if, however, imatinib Cmin was measured once, that value was used.

Statistical analysis

Imatinib Cmin was dose-adjusted and log-transformed. The Wilcoxon rank test was used to compare the means of initial and FU imatinib Cmin, and one sample t-tests were performed to determine whether initial and FU imatinib Cmin differed within each individual patient. Linear regression analysis was used to compare the ratios of FU to initial imatinib Cmin (FU:initial) with covariates to find correlative factors for changes in imatinib Cmin over time. In per-sample analysis, the correlations between imatinib Cmin and covariates, such as age, sex, body weight, BSA, duration of imatinib treatment, hemoglobin concentration, WBC, absolute neutrophil count (ANC), platelet count, albumin concentration, creatinine clearance and previous major gastrectomy, were analyzed using linear regression. Potential correlative covariates with p < 0.2 in univariate analyses were assessed in multivariate analysis using a multiple linear regression model. All tests were two-sided and a P-value <0.05 was considered statistically significant. All statistical analyses were performed using SPSS 18.0 (SPSS Inc., Chicago, IL, USA).

Results

Patient characteristics

Between November 2009 and May 2010, 65 patients who met the inclusion criteria were enrolled. Seven patients who had been treated with imatinib for over 2 years (range, 29–67 months) at the time of initial PK testing were excluded because of potential confounding effects on time-dependent PK changes. From the remaining 58 patients, Cmin was initially measured in 114 blood samples between October 2008 and May 2009 and at FU, Cmin was measured in 104 blood samples.

Baseline characteristics of the 58 included patients are shown in Table 1. Median patient age was 53 years (range, 28–76 years) and 37 (63.8%) patients were male. Seventeen (29.3%) patients received imatinib as adjuvant therapy after curative resection, whereas 41 (70.7%) received imatinib as primary therapy for unresectable or metastatic disease. Fifty-four (93.1%) patients were treated with 400 mg/day and 4 (6.9%) with 300 mg/day imatinib. The median time from initiation of imatinib to initial PK testing was 5.5 months (range, 0.5–24.0 months), the median time between initial and FU PK testing was 13.0 months (range, 9.6–17.9 months), and the median time from initiation of imatinib to FU PK testing was 19.4 months (range, 11.6–38.8 months).

Table 1 Baseline clinicopathologic characteristics
Full size table

Imatinib Cmin at initial and FU PK testing

We found that the FU mean ± standard deviation (SD) imatinib Cmin (1,370 ± 661 ng/mL) was significantly higher than the mean ± SD initial Cmin (1,171 ± 573 ng/mL; p = 0.003; Fig. 1). The distribution of FU:initial imatinib Cmin ratios, representing individual changes in imatinib Cmin over time, is presented in Fig. 2. The median (range) of FU:initial imatinib Cmin ratios was 1.14 (0.50–3.35), indicating that imatinib Cmin was a median 14% higher at FU than initially (p = 0.001). When compared with initial Cmin, FU Cmin was lower in 22 (38%) patients and higher in 36 (62%), with median changes of 13% and 32%, respectively. The absolute change from initial to FU Cmin was ≤30% in 35 patients, >30% to ≤60% in 14 patients, and >60% in 9 patients. The mean inter- and intra-subject variabilities were 47.7% and 20.9%, respectively, at initial measurement and 45.2% and 19.4%, respectively, at FU.

Fig. 1
figure 1

Initial and follow-up imatinib Cmin, reported as mean ± standard deviation

Full size image
Fig. 2
figure 2

Ratios of follow-up to initial imatinib Cmin

Full size image

Covariates associated with changes in imatinib Cmin over time

To investigate the covariates correlating with changes in imatinib Cmin over time, we compared FU:initial ratios of covariates and imatinib Cmin. Univariate analysis showed that the FU:initial ratios of hemoglobin (p = 0.02) and albumin (p < 0.001) concentrations were inversely correlated with those of imatinib Cmin (Fig. 3 and Table 2). However, we observed no correlation between the FU:initial ratios of imatinib Cmin and other variables, such as BSA, platelet count, ANC, creatinine clearance and time from initiation of imatinib treatment to initial PK testing. In multivariate analysis, only changes in albumin concentration were significantly correlated with changes in imatinib Cmin (Table 2).

Fig. 3
figure 3

Correlation between follow-up (FU):initial imatinib Cmin ratio and FU:initial ratios of hemoglobin and albumin

Full size image
Table 2 Correlation between follow-up/initial ratios of imatinib Cmin and covariates
Full size table

Per-sample analysis of imatinib Cmin

Per-sample PK analysis showed that age, BSA, hemoglobin and albumin concentrations, creatinine clearance, and previous major gastrectomy were associated with imatinib Cmin in a univariate linear regression model. In multivariate analysis, hemoglobin concentration, creatinine clearance, and previous major gastrectomy remained significant, independent covariates associated with imatinib Cmin (Table 3). In contrast, duration of treatment with imatinib before measurement of plasma imatinib concentration was not associated with imatinib Cmin (p = 0.55).

Table 3 Per-sample analysis for correlation with imatinib Cmin
Full size table

Discussion

We have shown here that imatinib Cmin did not decrease significantly over time in most patients with GIST. When compared with Cmin measurements performed at a median 5.5 months after initiation of imatinib, Cmin measurements after a median 13 months of further treatment increased in about 50% of patients. Since the median increase was only 14%, however, it may not be clinically meaningful, being due to the large intra-patient variability. Therefore, our results suggest that, in patients on long-term, chronic imatinib therapy, steady-state imatinib Cmin remains stable over time without significantly declining. Although this finding may conflict with previous results, which showed a time-dependent decrease in imatinib PK levels [12], differences in study design and PK parameters make it difficult to directly compare these results with ours. The strength of the earlier study was that it measured all major PK parameters at fixed time points (day 1, day 29 and after 1 year) [12]. PK results at days 1 and 29 were drawn from sufficient patients to allow conclusions to be drawn regarding the potential effects of changes in activities of metabolizing enzymes and plasma binding proteins during the early course of imatinib treatment. However, PK tests after 1 year of imatinib treatment were performed in only about half of initially enrolled patients. This may have limited the ability of these data to show changes in imatinib PK in GIST patients receiving long-term treatment. In contrast to that study, we assessed Cmin data at two time points (initial and FU) in all of our patients, allowing us to determine both overall and individual changes in imatinib Cmin. Because initial PK was measured at a median 5.5 months after initiation of imatinib therapy, adjustments in metabolizing enzyme activity and plasma binding proteins may already have occurred in most of our patients at the time of first PK sampling. Thus, our initial PK measurements likely reflect the status of patients after chronic exposure to imatinib. Accordingly, our results indicate that imatinib plasma levels remained stable during long-term treatment in most patients, not during the early stage of treatment. In addition, the median 19 months of imatinib therapy at FU PK testing in the present study seems sufficient to test the hypothesis that secondary imatinib resistance, which occurs around a median 2 years of imatinib treatment, may be the result of a reduction in imatinib levels [23, 24].

We performed additional subgroup analyses according to time from initiation of imatinib to initial PK testing (<3 months vs 3–12 months vs >12 months) and from initial to FU PK testing (9–12 months vs >12 months). In each group, mean FU Cmin was not lower than initial Cmin. Moreover, there were no between group differences in the numbers of patients with decreased Cmin, although these findings were not conclusive due to the small number of patients in each group (data not shown).

Although treatment with imatinib has remarkably improved clinical outcomes of patients with GIST, over half of these patients show disease progression within 2 years [6, 8]. There have been concerns that tumor progression may be due to suboptimal exposure to imatinib during long-term treatment. Clinically, however, our results suggest that decreased exposure to imatinib due to increased clearance is not a major cause of disease progression in these patients on long-term treatment. Rather, disease progression during chronic treatment is more likely due to acquired resistance including secondary gene mutations [24]. Therefore, our findings indicate that monitoring of imatinib Cmin is not required unless patients appear to be exposed beyond the appropriate range, such as non-compliance, unusual severe adverse events, or initiation or discontinuation of concomitant medications that considerably affect the activity of CYP 3A enzymes.

Drug exposure to imatinib during treatment may be affected by various factors, especially by changes in covariates. We found that changes in albumin were significantly and inversely associated with changes in imatinib Cmin [16, 20]. This finding suggests that increased imatinib Cmin might be associated with decreased albumin concentration, which could be affected by various clinical occurrences, rather than a real decline in imatinib clearance. Changes in other variables, such as hemoglobin concentration, WBC, creatinine clearance and treatment duration, did not correlate with changes in imatinib Cmin.

Although it was not a primary objective of this study, we performed per-sample analysis to verify the covariates correlated with imatinib Cmin in this patient population. Our results were consistent with those of previous studies, which found that hemoglobin concentration, creatinine clearance and previous major gastrectomy were significantly associated with imatinib Cmin [12–17]. Our finding on the association between major gastrectomy and imatinib Cmin is in good agreement with previous results on the impact of anatomical defects in the stomach on exposure to imatinib [16]. This result, however, requires further validation, because 47 (81.0%) patients in this study were included in the previous study [16].

Other factors may also affect exposure to imatinib. Because imatinib is metabolized primarily by CYP 3A enzymes and in the blood, and is bound to AGP as well as to albumin [11], variations in CYP 3A activities and AGP concentrations may affect imatinib Cmin. Therefore, our lack of data on CYP 3A enzymes and AGP limits our ability to explain the variation in imatinib Cmin. Although we measured only Cmin, because Cmin of imatinib in patients with GIST has been shown to be highly correlated with peak concentration (Cmax) and area under the plasma concentration-time curve (AUC), it may not limit our findings in the present study [20].

In conclusion, steady-state imatinib Cmin did not decrease significantly but remained stable or increased slightly in most patients during long-term treatment. Changes in imatinib Cmin were correlated with changes in serum albumin concentration. Monitoring of imatinib Cmin only due to concerns about time-dependent increased clearance of imatinib is not necessary.