A 63-year-old man was referred to our clinical pharmacology service due to disease progression on IMA therapy. The patient was diagnosed with abdominal GIST by biopsy of liver metastases. Immunohistochemical evaluation of the biopsy specimen revealed c- KIT expression, but no data on the molecular features of the tumour were available. At diagnosis, first-line treatment with IMA 400 mg OD (Gleevec, Novartis) was started, administered continuously, and achieved a partial response as the best clinical response in almost 6 years. When CT showed evidence of progressive disease in the abdominal lesion, pharmacologic counselling was requested.

Blood was drawn from the patient after signing a written informed consent form, and the plasma concentration of the drug was determined using an internally validated LC-MS method (more details can be found in [10]). TDM analysis revealed a IMA Cmin of 406 ng/mL and NOR Cmin of 213 ng/mL. The possible causes for this low exposure were investigated. Treatment adherence was considered adequate based on patient reports. To investigate DDIs of potential clinical significance, the patient’s medical treatments were reviewed. This revealed that the patient also had epilepsy, which had been controlled with carbamazepine 200 mg daily for nearly 30 years. In addition, he was taking pantoprazole for gastroesophageal reflux and ramipril/amlodipine to control hypertension. Potential DDIs were investigated using Lexicomp and the summary of characteristics of each co-administered drug (EMA).

Although pantoprazole can increase the exposure of IMA and amlodipine concentration is increased by IMA [11], the most significant DDI was identified between IMA and carbamazepine, which is known to be a potent inducer of CYP3A4 and P-gp [12, 13]. The observed NOR /IMA ratio in the patient sample was 52.5% (normally in the range of 20-25%), confirming an induction of metabolism. Meanwhile, as a consequence of the patient’s disease progression, the dose of IMA was increased to 400 mg twice daily (BID) and plasma concentration was re-monitored once it reached the new steady state to determine if this dose increase would have been sufficient to raise the plasma concentration of IMA to the target value of 1100 ng/mL.

The first sampling after dose escalation showed an IMA Cmin of 892 ng/mL, with a NOR /IMA ratio of 51.9%. Although the concentration of IMA increased slightly, it was still below the plasma levels associated with efficacy, and the NOR /IMA ratio was unchanged. Because no other options for dose escalation were feasible, the patient was informed and asked to talk with his neurologist about switching to a non-EIAED drug. Switching to another drug would have been beneficial for second-line therapy of GIST because sunitinib is also a substrate of CYP3A4 and thus a victim of CYP3A4 induction by carbamazepine.

The patient’s neurologist established a planned switch over four months in which carbamazepine was de-escalated and lacosamide, a non-EIAED, was gradually increased. Two weeks before completion of the switch (i.e., while the patient was receiving 50 mg carbamazepine and 100 mg lacosamide), the IMA level was measured again, revealing a markedly increased IMA concentration of 4298 ng/mL (NOR =1375 ng/mL) and a partial normalization of the NOR /IMA ratio (31.9%), synonymous with reversal of the effect of carbamazepine. Because of this high concentration, the oncologist decided to decrease the dose of IMA to 400 mg OD (to reduce the risk of adverse toxicity to the patient) and to reassess the concentration of IMA after 1 month, an appropriate time period since the reversal of carbamazepine induction can be considered complete within 2 weeks after cessation of carbamazepine therapy [14]. The concentration of IMA was 2586 ng/mL, which was lower than the previous determination but still within the therapeutic range (Fig. 1) [15].

Graphical representation of patient’s clinical history time-line with variation of IMA, NOR and ctDNA concentrations levels. Dashed grey line represents the Cmin threshold of 1100 ng/mL recommended for GIST patients. Grey backgrounds represent treatment at 400 mg OD, while white background identifies IMA treatment at 400 mg BID. Dashed black vertical lines indicate disease progression and switch to lacosamide, respectively. Abbreviations: IMA: imatinib, NOR: norimatinib, ctDNA: circulating tumor DNA, OD: once daily, BID: bis in die (twice daily), non-EIAED: non enzyme inducing antiepileptic drug, GIST: gastrointestinal stromal tumors

The analysis included the study of polymorphisms affecting the main IMA metabolizing enzymes (CYP3A4 and CYP3A5) and transcellular transporters (ABCB1 and ABCG2). No significant pharmacogenetic markers were identified that could potentially affect patients’ exposure to the drug (Table 1).

To elucidate potential alternative causes of tumor resistance to IMA, ctDNA was studied during treatment. Specifically, a panel of GIST-related genes in patient ctDNA was analysed (Supplementary Table 1) to investigate somatic changes potentially responsible for imatinib resistance. Two c-KIT mutations, c.1621A > C and c.2586G > C (Supplementary Table 2), were detected simultaneously during the course of IMA 400 mg BID, both with an allele frequency of 11% (Fig. 1). Although not associated with the development of IMA resistance, the presence of c-KIT mutations, tracking tumor DNA in plasma, further supports the clinical diagnosis of disease progression.

The applied TDM approach combined with pharmacogenetic analysis and comprehensive review of the DDI identified a case of severe imatinib underexposure after several years of concomitant treatment with the interacting drug carbamazepine. With the support of the neurologist, carbamazepine was gradually switched to lacosamide, a non-EIAED drug. Plasma levels of imatinib rose back above the efficacy threshold. Disease progression was confirmed by an increase in plasma ctDNA levels. Unfortunately, the patient’s disease progressed steadily during this time and resulted in death.