Imatinib trough levels in chronic myelogenous leukemia: does one dose fit all?

The excellent outcomes associated with imatinib (IM) treatment for chronic myeloid leukemia (CML), especially in chronic phase (CP-CML), are well documented. The 8-year follow-up of the IRIS trial (International Randomized Study of Interferon vs. STI571) shows an overall survival of 85%, and survival of 93% when only CML related deaths are considered [1]. However, only approximately 60% of patients will remain on IM, with a significant proportion stopping the drug either for intolerance or resistance. Despite this, IM 400 mg QD remains the current standard of care for frontline tyrosine kinase inhibitor (TKI) therapy in CP-CML. Therefore, the challenge is to prospectively identify patients who are unlikely to achieve optimal response, and to improve outcomes in these patients. The search for markers predictive of disease response is a natural extension to clinical practice. The currently known risk stratification markers include clinical parameters (Sokal and Hasford scores), biomarkers correlated with intracellular IM concentration (organic cation transporter-1 [OCT-1] activity) [2], and IM therapeutic drug monitoring (IM-TDM).

Imatinib has favorable pharmacokinetic properties, with rapid enteral absorption and high bioavailability irrespective of dosage. It is bound and transported by plasma proteins, mainly albumin and α1-acid glycoprotein. Its half-life of 18 h enables convenient once-daily dosing. It is metabolized by CYP3A4/5, with the major active metabolite being CGP74588 [3]. After enteral absorption, plasma IM then crosses into the cytosol to act on BCR–ABL. This process is highly dependent on cell membrane transporters such as OCT-1 [2]. Efflux mechanisms such as MDR1 (multidrug resistance gene) play a lesser role in affecting the intracellular drug concentration in the majority of patients [4].

IM-TDM is measured using high performance liquid chromatography (HPLC) or liquid chromatography with mass spectrometry (LC-MS), but it is not yet mandated as the standard of care. The currently available data to support its role in prognosticating outcome rest on several retrospective analyses. These studies demonstrate a surprisingly wide inter-patient variability in drug levels. For instance, Picard and colleagues found trough IM plasma levels ranging from 181 to 2947 ng/mL in 68 French patients taking 400–600 mg of IM QD, with the mean ± SD of 1058 ± 557 ng/mL and 1444 ± 710 ng/mL for the 400 mg and 600 mg groups, respectively. IM trough levels were found to correlate with the achievement of major molecular response (MMR, defined as < 0.1% international scale [IS] BCR–ABL by real-time quantitative polymerase chain reaction [RQ-PCR]), with patients achieving IM trough levels of <1002 ng/mL significantly more likely to achieve MMR (odds ratio 7.80, 95% confidence interval 2.64–23.03) [5]. This evidence is surprising for a drug with such favorable pharmacokinetic parameters and raises concerns regarding the 400 mg ‘dose for all’ strategy commonly employed.

A retrospective cohort of 78 patients from Canada offers a different view. 51/78 patients achieved MMR with median IM trough levels of 1067 ± 473 ng/mL, whilst 27/78 who failed to achieve MMR had IM trough levels of 1063 ± 642.5 ng/mL (p = 0.74). The inter-patient variability was equally significant. These measurements, like those of Picard, were performed on patients well established on therapy [6].

These retrospective studies are likely to suffer from selection bias, as patients deriving the least benefit are most likely to have discontinued IM therapy, their IM-TDM thus being excluded from analysis. Information is available from the IRIS and TOPS (Tyrosine Kinase Inhibitor Optimization and Selectivity) trials on the prognostic value of IM-TDM performed soon after therapy commencement. In 351 patients on the IRIS trial taking 400 mg of IM QD, the inter-patient variability observed was similarly large, with the plasma IM trough level 29 days after initiation of therapy ranging from 153 to 3910 ng/mL (979 ± 530) [7]. By dividing IM trough levels into quartiles, a clear relationship between levels and the likelihood of achieving complete cytogenetic response (CCyR) and MMR was observed. Patients with trough levels in the lowest quartile (Q1: mean 490 ng/mL) demonstrated a 12-month MMR rate of 43%, compared to 56% and 55% for patients in Q2–3 (mean 889 ng/mL) and Q4 (1661 ng/mL), respectively. Patients who achieved CCyR had a mean trough IM level of 1009 ng/mL, compared to 812 ng/mL in those with no CCyR. This is concordant with Picard's work, and 1000 ng/mL has thus become a putative therapeutic target. Interestingly, there was no statistically significant correlation between anthropomorphic parameters (such as body surface area, weight, and age) and IM trough level, and race was of no apparent consequence [8]. In the TOPS study comparing 400 mg QD of IM to 800 mg QD, patients with the lowest quartile IM trough level (< 1165 ng/mL) were also less likely to achieve MMR at 12 months. It may be counterintuitive that traditional predictors of pharmacokinetics make little difference to the IM trough levels, but the contributions from these factors are likely masked by the otherwise large inter-patient variation [9].

In this issue of Leukemia and Lymphoma, Liu and Artz demonstrate the complexity of using IM in a real clinical setting by describing a patient with unexpected IM levels from significantly altered intestinal transit time after gastrointestinal bypass surgery [10]. The low trough drug levels over a period of months may have contributed to the patient's suboptimal disease control and failure to achieve MMR. We know that many factors contribute to variations in IM plasma levels in different patients. Compliance is a problem whenever patients are asked to take medications regularly for long periods of time, and several studies have shown that patients often under-report missed doses to their physicians, as judged against records from electronic monitoring devices [11]. Concomitant medications may enhance or impede absorption as well as the metabolism of IM. The list of implicated drugs includes many commonly used over-the-counter medications such as analgesics (e.g. paracetamol), antihypertensives (verapamil), antidepressants (venlafaxine), sedatives (oxazepam), and proton pump inhibitors. Altered gastrointestinal transit time, as illustrated in this report, can also plausibly lead to unpredictable absorption. As gastrointestinal side effects from IM such as diarrhea and vomiting are common and themselves unpredictable, IM plasma levels may further fluctuate, and as demonstrated in this report additional diarrhea and vomiting frequently associated with gastric surgery itself further confound the ability to achieve good plasma levels.

Patients rarely present for blood tests at exactly the right time (i.e .24 h post-dose in once-daily regimens or 12 h post-dose in twice-daily regimens), and this is a further impediment to valid TDM interpretation. Through mathematical modeling, nomograms and an elimination constant have recently been published to aid comparisons between drug levels at known intervals after drug ingestion and true trough levels [12]. However, the accurate interpretation even in this setting will be dependent on the time of last drug ingestion being recorded.

It is important to emphasize that IM-TDM is itself a surrogate marker, and is only one of many factors that can influence disease outcome. A more critical factor may be the intracellular IM concentration and the degree of BCR–ABL kinase inhibition achieved, which not only is dependent on plasma IM level, but is more tightly linked to the transporter OCT-1. This has been shown to be of prognostic value when assessed on diagnostic samples. Our group has shown that when comparing outcomes at 60 months for patients with high OCT-1 activity (defined as > 7.2 ng/200 000 cells) versus low OCT-1 activity, the former are more likely to achieve MMR (89% vs. 55%), and have high overall survival (96% vs. 87%) as well as event-free survival (74% vs. 48%). Kinase domain mutations are also less common in patients with high OCT-1 activity (21% vs. 4%) [2]. In addition, the prognostic significance of OCT-1 activity is compounded by plasma levels. In the TOPS cohort, patients in the lowest cohort for both OCT-1 activity and IM level had significantly worse outcomes, compared to patients who had either low OCT-1 or plasma level alone [13]. There is suggestion that a higher IM dose may overcome the effect of low OCT-1 on prognosis [14]. Though powerful, the OCT-1 assay requires a specialized laboratory, and is not as yet widely available.

Finally, it is important to continuously re-evaluate treatment response. A patient with suboptimal response who fails goals as set out in the European LeukemiaNet criteria will potentially benefit from therapeutic intervention. Studies have confirmed that disease response after 3 months of IM treatment is highly correlated with MMR at 12 months [15]. After excluding poor adherence, interfering concomitant medications, and low trough IM levels, a second-generation TKI should be considered. If present, tyrosine kinase domain mutations may influence the choice of the second-generation TKI. The role of TDM with these TKIs is not yet established.

The IM trough level is an important tool for optimizing patient outcome. In anticipation of data from prospective clinical trials, the current opinion is that a minimum IM trough level of 1000 ng/mL is associated with the achievement of MMR. Potentially beneficial interventions to achieve this pharmacokinetic goal include regular IM-TDM, encouraging adherence, optimizing the IM dose, checking concomitant medications, and noting gastrointestinal anatomical or physiological variations. OCT-1 activity, where available, should be used in conjunction to enhance prognostication. Finally, the IM trough level should be interpreted holistically and within the context of overall disease response. In the case described by Liu and Artz, IM-TDM provided valuable insight into the underlying cause of poor treatment response in a patient with CML undergoing gastric bypass surgery.