Phase I Study of Oral Gemcitabine Prodrug (LY2334737) Alone and in Combination with Erlotinib in Patients with Advanced Solid Tumors

Gemcitabine [(2′,2′-difluorodeoxycytidine (dFdC)] is an anticancer agent approved for the treatment of a variety of solid tumor types, including pancreatic, non–small-cell lung, ovarian, bladder, and breast cancer. The anticancer activity of gemcitabine is mediated through its effects on DNA synthesis in rapidly dividing cells. Gemcitabine must be phosphorylated by deoxycytidine kinase and other intracellular kinases to produce the active forms gemcitabine diphosphate and triphosphate (dFdC-DP and dFdC-TP). Incorporation of dFdC-TP into DNA during S-phase of cell cycle results in termination of DNA synthesis, single-strand breaks, and eventually cell death (1, 2). Gemcitabine is rapidly metabolized into the active metabolite 2′,2′-difluorodeoxyuridine (dFdU) by cytidine deaminase that is present at high levels in plasma, red blood cells, and liver (3, 4).

Gemcitabine activity and toxicity are highly dependent on the frequency and duration of administration. In previous phase I and II studies (5–8), lower gemcitabine concentrations over longer infusion times were clinically active. As a surrogate marker for tumor uptake and activation of gemcitabine, levels of dFdC-TP were measured in patients' peripheral blood mononuclear cells (PBMC). Prolonged gemcitabine infusion resulted in enhanced accumulation of dFdC-TP in PBMCs (9) and in leukemia cells (5, 10). Taken together, these studies indicate that the antitumor effect of gemcitabine could be schedule dependent and that lower doses given over longer exposure times might be efficacious (8). However, pancreatic cancer was not found to be a sensitive tumor type to prove this concept, as prolonged infusion times did not result in improved overall and progression-free survival (11). Nevertheless, preclinical studies in mice with daily administration of gemcitabine showed antitumor activity in human colon, lung, and prostate tumor xenograft models. In addition, a colon xenograft model, LY2334737, resulted in higher dFdC incorporation into tumor DNA than gemcitabine (data on file at Lilly).

In view of this, daily and every other day oral administration of gemcitabine was studied in a previous clinical trial (12). The pharmacokinetic data obtained in this trial revealed that this approach was not feasible because of lack of bioavailability. The poor bioavailability was attributed to the extensive first-pass metabolism by deamination of gemcitabine into dFdU. Hence, a new chemical entity, LY2334737, was developed (13). LY2334737 is a prodrug of gemcitabine, in which the (metabolic) unstable amine group is covalently bound to valproic acid. When LY2334737 is administered orally, its amide bond is slowly hydrolyzed and gemcitabine and valproic acid are released systemically. It is postulated that following LY2334737 administration, gemcitabine will be protected from extensive presystemic deamination, resulting in lower conversion into dFdU and thereby significant exposure to gemcitabine in plasma after oral administration.

Moore and colleagues showed that the combination of epidermal growth factor receptor (EGFR) inhibitor erlotinib and gemcitabine resulted in a modest albeit statistically significantly improved survival in patients with advanced pancreatic cancer when compared with gemcitabine alone (14). Because erlotinib apparently has no overlapping toxicity or pharmacokinetic interactions with gemcitabine (14), the feasibility of the combination of LY2334737 with erlotinib was investigated in this study as well.

The aim of this study was to determine a dose of LY2334737 to be recommended for phase II studies that may be safely administered to patients with cancer as monotherapy and in combination with the EGFR inhibitor erlotinib. During this study, 2 different formulations of LY2334737 were tested (nonregistration formulation and registration formulation designed as NRF and RF, respectively). The RF formulation had improved pharmaceutical characteristics relative to the NRF in terms of shelf-life conditions and loading capacity of the carrier, resulting in smaller capsules. After the determination of the recommended dose (using the NRF) for testing in phase II studies, this study was extended with a bioequivalence study to investigate the RF formulation of LY2334737.

Patients with histologically or cytologically proven cancer (solid tumors only) for whom no treatment of proven benefit existed were eligible. Other eligibility criteria were as follows: age 18 years or older, performance status 2 or less [Eastern Cooperative Oncology Group (ECOG)], and an estimated life expectancy of 12 weeks or more. Previous therapies for cancer had to be discontinued for at least 30 days before study entry and 6 weeks in case of mitomycin C or nitrosoureas. Patients had to have adequate bone marrow function, defined as absolute neutrophil count of 1.5 × 10⁹/L or more, platelets 100 × 10⁹/L or more, and hemoglobin 9 g/dL or more and adequate renal and hepatic function defined as serum creatinine 1.5 or less the upper limit of normal (ULN), bilirubin 1.5 or less the ULN, and alanine transaminase (ALT) and aspartate transaminase (AST) 2.5 or less the ULN. In case of metastases, AST and ALT ≤ 5 times ULN were acceptable. Exclusion criteria were as follows: experimental therapy received within the last 30 days before study entry, symptomatic central nervous system malignancy or metastasis, gastrointestinal disease that may interfere with adequate oral absorption, and patients who had previous diagnosis of liver cirrhoses, chronic hepatitis, and a history of active alcohol abuse and acute or chronic leukemia. The study was approved by the local medical ethics committee of each hospital, and all patients had to give written informed consent. The study was conducted at the Netherlands Cancer Institute and the University Medical Center Utrecht.

The study is a nonrandomized, open-label, dose escalation, 3-arm phase I study of LY2334737 as monotherapy (arm A) and in combination with erlotinib (Tarceva; arm B). During the study, an additional arm (arm C; monotherapy crossover replicate) was added by amendment to determine the bioequivalence of a novel LY2334737 drug formulation: RF (Fig. 3). Arm B started when at least 15 patients were followed up in arm A (LY2334737 monotherapy) and after at least 1 treatment-related grade 2 toxicity was observed. These conditions were implemented to reduce the number of patients treated with low, possibly inactive, doses of LY2334737. After completion of arm A, arm C was opened. This part of the study was conducted at the recommended dose determined in arm A.

Safety was determined according to the Common Terminology Criteria for Adverse Events (CTCAE), version 3.0, and preliminary antitumor activity was determined using Response Evaluation Criteria in Solid Tumors (RECIST), version 1.0 (15). Tumor biomarkers were determined to explore any preliminary antitumor activity.

The primary objective of arms A and B of this study was to determine the recommended dose for phase II studies of LY2334737 alone or in combination with erlotinib. The recommended dose was defined as the maximum tolerated dose (MTD), which is the highest dose of LY2334737 monotherapy or in combination with erlotinib that had no more than 33% probability of causing a dose limiting toxicity (DLT). A DLT was defined as an adverse event observed during the first 21-day cycle of LY2334737 therapy according to any of the following criteria: (i) grade 4 neutropenia lasting 5 days or more, or neutropenic fever; (ii) grade 4 thrombocytopenia, (iii) any grade 3 or higher nonhematologic toxicity; and (iv) interrupted treatment due to toxicity.

After enrollment in the study, patients assigned to arms A and B received one 5-mg (nontherapeutic) dose of gemcitabine (Gemzar) via an intravenous push to determine the individual gemcitabine clearance and thereafter the bioavailability of gemcitabine after LY2334737 oral administration. Subsequently, a 4- to 7-day washout period was required before initiating LY2334737. The intravenous administration together with the washout period was designated as cycle 1.

In cycle 2, the first cohort of 3 patients received 5 mg LY2334737 every other day for 14 days followed by 7 days of rest. This starting dose was selected on the basis of preclinical safety studies, and data were generated with oral gemcitabine (12).

On the condition that no drug-related grade 2 (or greater) hematologic toxicity and/or nonhematologic toxicity (excluding nausea and vomiting without treatment and alopecia) were observed in the first cohort, subsequent cohorts were administered LY2334737 daily for 14 days, followed by 7 days of rest.

The same LY2334737 treatment schedule was applied in arm B with the addition of daily 100 mg erlotinib. This dose was selected, as this is the approved erlotinib dose in combination with gemcitabine for the treatment of pancreatic cancer.

Dose escalation followed the continual reassessment method (CRM), a Bayesian model to efficiently estimate the MTD. The incidence of DLTs across all investigated dose levels is taken into account. Subsequently, a probability distribution of DLT versus the LY2334737 dose is generated. The MTD is determined by the dose at which the probability of DLT is 33%. The CRM potentially escalates the dose too rapidly, using single-patient cohorts (16). Modifications based on observed toxicity that limit dose escalations were implemented. This modified CRM has proven to provide greater efficiency than standard dose escalation schemes (16, 17).

While recruitment in arms A and B was ongoing, the RF formulation of LY2334737 was developed. In arm C, the objective was to compare the relative bioavailability of RF with the NRF evaluated in arms A and B. This was investigated using a 3-period replicate crossover design. Thirty evaluable patients were planned to enroll into 2 groups of 15 patients each. Patients received the RF or NRF formulation on days 1, 3, and 5 in the following order: group 1: RF, RF, and NRF; and group 2: NRF, RF, and NRF. After pharmacokinetic assessments, patients received on 10 consecutive days LY2334737 (starting day 6) as NRF at the recommended phase II dose followed by 7 days of rest. Cycle 1 was defined as the first 5 days of bioequivalence assessment, and cycle 2 was defined as the subsequent 10 days daily treatment and 7 days rest period. In cycle 3 (and beyond), patients received the NRF of LY2334737 as daily dosing for 14 days followed by 7 days of rest.

The 2 formulations were considered equivalent if the 90% confidence limits ratios of exposure were within the 0.7 to 1.43 interval (data were analyzed on the log scale). These boundaries are wider than the standard bioequivalence boundaries set forth by the Food and Drug Administration: 0.8 to 1.25 (18). However, as LY2334737 is not a marketed drug and still in phase I clinical testing, a formal bioequivalence trial is not required.

Intravenous gemcitabine (Gemzar; Eli Lilly and Company) is commercially available. The NRF formulation of LY2334737 (Eli Lilly and Company) was supplied as 5-, 15-, and 30-mg capsules for oral consumption, and the RF of LY2334737 (Eli Lilly and Company) was supplied as 15- and 20-mg capsules. Two substantial changes have been made in the composition of the RF from the NRF. To facilitate the manufacturing process, the enteric coating is changed from hydroxypropyl methylcellulose acetate succinate to Eudragit. The second change was made to decrease capsule size, and the amount of LY2334737 loaded onto the carrier beads increased from 10% (w/w) in NRF to 25% in RF. Neither of these 2 changes is expected to cause significant increase or decrease in exposure between NRF and RF. Erlotinib (Tarceva; OSI Pharmaceutical, Genentech, and Roche) is commercially available.

Pretreatment evaluation included a complete medical history, physical examination, electrocardiography, chest X-ray analysis, vital signs, assessments of adverse events using CTCAE version 3.0, the use of concomitant medications, urine or serum pregnancy test, and laboratory tests of hematology [hemoglobin, leukocytes, platelets, neutrophils (ANC), lymphocytes, monocytes, eosinophils, and basophils] and serum chemistry (total bilirubin, alkaline phosphatase, AST, ALT, creatinine, calcium, glucose, sodium, phosphorus, and potassium). Before each cycle, a physical examination, assessment of adverse events, and notation of concomitant medication were repeated and hematology and serum chemistry were checked. If any grade 2 or 3 toxicities were seen with laboratory tests, hematology and serum chemistry were repeated every 2 to 3 days.

The pharmacokinetics of LY2334737, gemcitabine, dFdU, valproic acid, and, if applicable, erlotinib were monitored during the study. The pharmacokinetic sampling scheme of arms A and B consisted of sampling cycle 1 after the fourth gemcitabine dose on day 1 (predose, end of infusion, 0.5, 1, and 2 hours after infusion) and cycle 2 after LY2334737 dose on day 1 (predose, 0.5, 1, 2, 4, 6, 8, and 24 hours after dosing) and day 14 (predose, 0.5, 1, 2, 4, 6, 8, 24, 48, and 168 hours after dosing). Blood samples for LY2334737 and metabolites were drawn in lithium heparinized tubes containing 0.075 mL of 10 mg/mL tetrahydrouridine (inhibitor of cytidine deaminase). The pharmacokinetic sampling scheme of arm C consisted of sampling cycle 1 on days 1, 3, and 5 (predose, 0.5, 1, 2, 4, 6, and 8 hours after dosing). The intracellular phosphorylated gemcitabine metabolite levels (dFdC-TP) were monitored on day 1 of cycle 2 (arms A and B) at 1, 4, 8, and 24 hours after dosing and day 14 (cycle 2; arms A and B) at 1, 4, 8, 24, and 168 hours after LY2334737 dosing. LY2334737, metabolites, and erlotinib concentrations were measured using validated [liquid chromatography/tandem mass spectrometry (LC/MS-MS)] assays. For the determination of dFdC-TP, approximately 15 mL of blood was drawn into sodium heparinized tubes. The tubes were centrifuged at 4°C for 5 minutes at 1,500 × g. PBMCs were isolated and dFdC-TP levels were determined, as described previously, using a validated LC/MS-MS assay (19). The following pharmacokinetic parameters were determined by noncompartmental analysis using WinNonLin version 5.2 (Pharsight Corporation): maximum plasma concentration (C_max), time to reach C_max (t_max), area under the plasma concentration–time curve from time 0 to 24 hours (AUC_{0–24 h}), AUC from time zero to infinity (AUC_0–∞), and terminal half life (t_1/2). The geometric mean and coefficient of variation (CV) and the accumulation index (AUC_day14/AUC_day1) are provided.

The incorporation of gemcitabine into genomic DNA in PBMCs was determined in arm C. One blood sample (10 mL using EDTA tubes) was collected on day 3 of cycle 1 and days 3, 6, and 10 of cycle 2. After the DNA had been isolated and completely hydrolyzed, the fluorinated nucleotide (dFdC) was detected using LC/MS-MS (20).

In arms A and B, blood samples were drawn for the determination of antiangiogenesis biomarkers: circulating endothelial cells (CEC) and VEGF. Blood samples for the determination of CECs were collected on 5 occasions; twice before start (mean baseline value) and subsequently once on days 7, 14, and 21 of cycle 2. Mononuclear cells were isolated and CECs were detected using flow cytometry as previously described. (21) In addition, VEGF was determined in plasma at baseline and days 7 and 14 of cycle 2. Analysis was carried out using a commercially available ELISA kit (R&D Systems).

A total of 65 patients with histologically or cytologically confirmed advanced solid tumors were treated in this study. Thirty-two patients were entered in arm A, 10 patient in arm B, and 23 patients in arm C. The characteristics of these patients are summarized in Table 1.

The LY2334737 dose in the monotherapy arm (arm A) was rapidly escalated from 5 mg every other day up to 40 mg daily with dose increments of 100%. Because moderate toxicities were observed in the 3 patients treated at the 40-mg dose level, the magnitude of dose escalation was reduced and the next highest dose level was set at 50 mg daily. Seven patients were treated with 50 mg LY2334737 monotherapy, of whom 4 experienced DLTs. The first patient after 11 days of LY2334737 treatment developed grade 3 pyrexia (drug-related fever) and grade 4 thrombocytopenia, which both completely recovered after interruption of treatment. A second patient developed grade 3 fatigue after 9 days, for which treatment was discontinued. Severe hyponatremia, fatigue, and pulmonary embolism were the DLTs of the third patient. These adverse events recovered after discontinuation of LY2334737. The last patient developed grade 3 hyponatremia and fatigue. Regarding the severe toxicity at the 50-mg dose level, it was concluded that this dose was nontolerable.

Subsequently, the following 3 patients were treated at the next lowest dose level of 40 mg daily, at which already 3 were treated without serious toxicity. However, in the expansion cohort, 2 patients developed DLTs. One patient developed grade 3 ALT elevations, which partly resolved after a week of rest and a dose reduction. The second patient, a 61-year-old male patient with pancreatic cancer, died suddenly, which was considered possibly related to LY2334737 intake. This patient was admitted to a hospital because of a partial obstruction of the duodenum due to pancreatic tumor, grade 3 hyponatremia, and elevated alkaline phosphatase and transaminase levels. Four days later, the patient complained of dyspnea, low blood pressure, and signs of hypoperfusion (hypovolemic shock). The following day the patient experienced abdominal pain and dyspnea and died suddenly. The events of dyspnea, hypovolemic shock, and sudden death were considered possibly related to study treatment.

Subsequently, 3 patients were treated with 30 mg LY2334737 daily. None of these patients experienced DLTs. On the basis of the incidence of DLTs, the MTD was estimated using the CRM to be 40 mg LY2334737. It was estimated that the median probability of having a DLT at this dose level was 25%.

The dose escalation arm of LY2334737 in combination with erlotinib (100 mg) started at the 20-mg LY2334737 dose level. DLTs were not observed in the 4 patients treated at this dose level. The following 6 patients were treated at the 40-mg dose level, of whom 2 experienced DLTs. These were grade 4 γ-glutamyl transpeptidase (GGT) and grade 3 elevated ALT levels. Both patients discontinued LY2334737 and erlotinib. The CRM resulted in an estimated probability of 31% having a DLT at the 40-mg LY2334737 and 100-mg erlotinib dose levels. Therefore, the MTD for the combination with erlotinib was set at 40 mg LY2334737.

A total of 56 of 65 patients experienced possibly drug-related treatment-emergent adverse events. The most frequent were fatigue (with an incidence between 50% and 70% for the different treatment arms), pyrexia and influenza-like illness, and gastrointestinal disorders, including nausea, vomiting, diarrhea, stomatitis, dysgeusia, and anorexia. In addition, liver toxicities reported included grade 4 GGT, ALT, and AST elevations. An overview of the observed adverse events is provided in Table 2. The severity of most toxicities was mild (grades 1–2). A dose-dependent increase in the occurrence of adverse events was observed, as patients who were treated with daily 20 mg LY2334737 or less, showed a substantially lower incidence of adverse events than patients treated with 30 mg LY2334737 or more. Furthermore, no grade 3 or 4 toxicities were observed in patients treated with 20 mg LY2334737 or less. The observed grade 3 and 4 toxicities at higher dose levels consisted mainly of fatigue and AST and ALT elevations.

The most commonly observed adverse event, fatigue, appeared early after start of LY2334737 and ceased quickly after drug discontinuation. Several patients presented with pyrexia up to grade 3. In general, the fever seemed to be well tolerated by patients and was resistant to treatment with acetaminophen or nonsteroidal anti-inflammatory drugs. Interruption of study medication resulted in recovery of symptoms. Besides fatigue, dose-dependent liver enzyme elevations were the most reported grade 3 and 4 toxicities. Mostly, these events occurred quickly after the first 14 days of LY2334737 treatment. Treatment discontinuation or a dose reduction to the next lower dose level mostly resulted in recovery to normal levels.

Hematologic toxicity was rarely observed. Grade 3 and 4 thrombocytopenia was observed in 2 patients treated with 50 mg LY2334737, but no neutropenia of any grade occurred. Severe, grade 3, hyponatremia was observed in 5 patients. This was not expected regarding the working mechanism or the toxicity profile of intravenous gemcitabine. A possible explanation could be reduced food intake by patients, as 3 of the patients with hyponatremia had also LY2334737-related anorexia.

The combination arm of LY2334737 and erlotinib showed a similar toxicity profile compared with LY2334737 monotherapy. However, a few adverse events were more frequently observed in the combination arm. These were diarrhea (60%), rash (50%), pyrexia (60%), and influenza-like illness (40%), which is consistent with the known toxicity profile of erlotinib.

In total 12 of the 65 patients discontinued the study because of adverse events. The most common adverse event causing discontinuation was fatigue.

A total of 49 patients provided blood samples for pharmacokinetic analyses after a single dose of LY2334737 (NRF), and a total of 28 patients provided data after 14 days of consecutive LY2334737 (NRF) administration.

After drug administration, LY2334737 was absorbed rapidly with maximum concentrations reaching 2.0 hours (10th and 90th percentiles: 1.0–6.0 hours) after drug intake. Conversion to gemcitabine occurred rapidly, and C_max of gemcitabine was reached at approximately the same time as LY2334737 at a median of 2.0 hours (10th–90th percentiles: 1.0–4.9 hours) after dose. The mean concentration–time curves after 40 mg LY2334737 are provided in Fig. 1A, and the pharmacokinetic parameters are presented in Table 3. A dose proportional increase in C_max and AUC of LY2334737 and gemcitabine was observed, and the median elimination half-life of LY2334737 and gemcitabine, determined using all pharmacokinetic data, were 1.77 and 1.83 hours, respectively. The variability in LY2334737 and gemcitabine exposure after administration of 40 mg on day 1 was 37% and 39%, respectively. The variability in C_max at this dose level was 62% and 72%. There was no statistically significant accumulation of LY2334737 and gemcitabine, although Fig. 2 shows a trend of increasing LY2334737 and gemcitabine exposure after 14 consecutive doses. Valproic acid could not be determined, as all samples fell below the lower limit of quantification (10 μg/mL).

Ten patients who participated in arm B were treated with LY2334737 in combination with erlotinib. The pharmacokinetics of LY2334737 in this combination arm was comparable with LY233437 monotherapy. The geometric mean exposure (AUC_{0–24 h}) to erlotinib was 8.2 μg h/mL (CV: 77%) at day 1 and 18 μg h/mL (CV: 120%) at steady state.

In study arm C, a novel LY2334737 drug formulation (RF) was compared with the old drug formulation applied in arms A and B (NRF). Twenty-three patients were included in this part of the study. No differences in systemic exposure to LY2334737 or gemcitabine were observed (AUC ratio LY2334737 RF:NRF of 1.02 with 90% CI of 0.91–1.14; AUC ratio gemcitabine RF:NRF of 0.91 with 90% CI of 0.80–1.02). However, the RF showed a significantly lower C_max than NRF, 32.1 versus 49.8 ng/mL LY2334737 and 3.12 versus 5.24 ng/mL gemcitabine (geometric mean values, 90% CI of ratio of least squares: 0.55–0.79 for LY2334737 and 0.49–0.71 for gemcitabine).

Plasma levels of dFdU, the major gemcitabine metabolite, were determined in all patients. dFdU has a long terminal half-life of 88 hours (CV: 32%) measured in patients treated in arm A. This means that at day 14, following 14 daily doses, steady state was almost reached (approximately 4 times the terminal half-life). At day 14, the geometric mean t_max was 4.0 hours (2.0–24 hours); C_max was 1.30 μg/mL (CV: 19%), and AUC_{0–24 h} was 28.3 μg h/mL (CV: 18%) dFdU, respectively. The pharmacokinetic ratios (day 14 vs. day 1) were 3.6 (CV: 32%) for C_max and 4.3 (CV: 20%) for AUC_{0–24 h}. The mean concentration–time curves of dFdU at day 14 are provided in Fig. 1B.

The active metabolite of gemcitabine dFdC-TP was measured in isolated PBMCs. The geometric mean AUC_{0–24 h} at days 1 and 14 after 40 mg LY2334737 was 80 ng h/mg PBMC (CV: 69%) and 79 ng h/mg PBMC (CV: 47%), respectively. This corresponds to approximately 27 h pmol/10⁶ cells (19). The geometric mean concentration–time curves observed after the different dose levels are provided in Fig. 2, bottom left panel.

In study arm C, the incorporation of gemcitabine in genomic DNA was measured in PBMCs. The results of this assay are given in Fig. 1D. The level of incorporated gemcitabine increased while patients were on treatment, and for most patients, a plateau level was not yet observed within the investigated time period of 14 days.

No statistically significant decrease or increase in CEC or VEGF levels during treatment, nor between arms A and B (with or without erlotinib), was observed, although patients treated in arm B showed a trend toward decreasing CECs and VEGF upon treatment for 21 days (data not shown).

A total of 51 patients were evaluable for at least one response assessment (Table 4). The best response within 22 patients was stable disease. Remarkable were 2 patients with mesothelioma who had progressive disease before start but showed stable disease for fourteen and seventeen 3-weekly cycles, respectively. One patient with prostate carcinoma had proven benefit of the treatment. This patient was assigned to the 40-mg dose level in combination with daily erlotinib 100 mg. The LY2334737 dose was, due to grade 3 ALT elevations, reduced in cycles 4 and 8 to 30 and 20 mg, respectively. Because there were no measurable lesions, this patient was evaluated for prostate-specific antigen (PSA), which was 89.4 μg/L at study entry and decreased to 32 μg/L (end of cycle 2). A complete response of PSA was observed after cycle 4 (PSA: 0.6 μg/L) and stabilized for 4 cycles. By the end of cycle 8, progression of PSA was observed (PSA: 20.6 μg/L).

This report describes the first in-human study of the oral gemcitabine prodrug LY2334737. The MTD was determined, for both monotherapy and in combination with 100 mg erlotinib, to be 40 mg daily administrations of LY2334737 for 14 days followed by 1 week of rest. Furthermore, the LY2334737 RF was considered bioequivalent to the NRF.

The most reported adverse events were fatigue, elevated liver enzyme levels, gastrointestinal toxicity, and flu-like illness (including pyrexia), which are adverse events also frequently observed after intravenous administration of gemcitabine. Hematologic toxicity, apart from grade 3/4 thrombocytopenia in 2 patients treated at the 50-mg dose level, was not observed. This is surprising, as hematologic toxicity is the predominant DLT of gemcitabine intravenously. A possible explanation for this is the lower systemic exposure to gemcitabine over time. At the MTD patients received a cumulative dose of 560 mg LY2334737 (equivalent to 304 mg gemcitabine) in a 21-day cycle, which is significantly lower than standard intravenous gemcitabine regimens (e.g. 1,250 mg/m² on days 1 and 8 of a 21-day cycle). However, high incidences of typical gemcitabine-related nonhematologic toxicities were observed at these relatively low doses of LY2334737. These discrepancies in toxicity profile probably originate in the oral route of administration and the short dosing intervals resulting in continuous exposure to LY2334737 and metabolites. Twice-weekly schedules were previously investigated for intravenous gemcitabine. The weekly schedules were better tolerated, but the most striking difference in toxicity between the weekly and twice-weekly schedule was the incidence of influenza-like illness, that is, 63% in the twice-weekly group versus 20% for the weekly schedule (22). Pyrexia and influenza-like illness were among the most observed toxicities in this study.

The pharmacokinetics of gemcitabine after intravenous administration is characterized by a high C_max and rapid decline in gemcitabine plasma levels after the end of the infusion, resulting in a short-term exposure to high levels of gemcitabine and metabolites. The continuous exposure to low levels of gemcitabine, achieved after daily LY2334737 intake, may contribute to cumulative nonhematologic toxicities, whereas the short-term high exposure after intravenous administration of gemcitabine results in hematologic toxicities.

AST and ALT elevations, after fatigue, were the most reported grade 3/4 toxicities. Studies in mice revealed that after multiple oral doses of gemcitabine, accumulation of phosphorylated gemcitabine (dFdC-TP) and phosphorylated dFdU (dFdU-TP) occurred. Especially, the accumulation of dFdU-TP in mouse liver was more pronounced following oral administration of gemcitabine than following intravenous administration (23). High levels of dFdU-TP were also observed in PBMCs of patients treated with oral gemcitabine. Therefore, the hepatotoxicity after daily intake of LY2334737 may be related to accumulation of gemcitabine metabolites in the liver.

LY2334737 was designed to overcome the extensive presystemic deamination of gemcitabine by cytidine deaminase to dFdU. This was required, as systemic exposure to gemcitabine after oral administration was very low (12).

Despite the reduced presystemic deamination due to the prodrug design, the total exposure to dFdU was still high and accumulation was observed after daily dosing of LY2334737. However, it is important to note that the accumulation ratio (day 14/day 1) of dFdU exposure was 4.45- or 0.75-fold lower than that observed following 2 weeks of daily dosing of oral gemcitabine (12). This indicates that the first-pass metabolism of gemcitabine into dFdU is lower following oral LY2334737 than following oral gemcitabine.

Both deamination of dFdC-monophosphate (MP) to dFdU-MP and cellular uptake of dFdU followed by phosphorylation result in the formation of dFdU-TP, which can be incorporated into DNA and RNA (24). Because of the long t_1/2, dFdU accumulates during the first 2 weeks of treatment and its plasma concentration is reduced only by about 75% (approximately twice the t_1/2) during the week of rest. The continuous exposure to dFdU may have contributed to the toxicity profile of daily LY2334737 treatment. Although there was a clear increase in toxicity with dose (see Table 2), it was not possible to assess more precisely the relationship between exposure of LY2334737 or its metabolite and toxicity because of both variability in systemic exposure and lack of knowledge of LY2334737 and metabolites concentration before systemic exposure at the level of the gut and liver.

The metabolite dFdC-TP and dFdC incorporated into DNA could be detected in peripheral PBMCs, although with high inter- and intrapatient variability. This indicates that the pharmacologically active form of gemcitabine can accumulate into these cells. Previously, it has been shown that this is a marker for the antitumor activity of gemcitabine (25, 26).

Hints of antitumor activity or stable disease were observed in 22 of 51 evaluable patients. Furthermore, a confirmed complete response of PSA was shown in a patient with metastatic prostate carcinoma.

Part C of this study compared two LY2334737 drug formulations: NRF versus RF. The RF showed slightly lower C_max values. This discrepancy most likely originates in the loading of the carrier beads (an excipient of the capsule formulation), which is much higher for the RF. Dissolution experiments showed that higher loading percentages of LY2334737 onto the carrier beads resulted in lower dissolution rates, possibly resulting in lower C_max values (data on file at Lilly). However, these discrepancies did not result in differences in exposure to LY2334737 and gemcitabine. Therefore, the NRF and RF were considered bioequivalent and future studies will be conducted with the RF, as this formulation resulted in smaller capsules and was easier to manufacture and store.

In conclusion, this study shows that LY2334737 can be safely administered to patients with solid tumors up to doses of 40 mg/d during 14 days followed by 1 week of rest with or without daily 100 mg erlotinib. The most frequently observed toxicities were fatigue, gastrointestinal toxicities, elevated liver enzyme levels, and influenza-like illness. The pharmacologically active forms of gemcitabine could be detected in isolated mononuclear cells and signs of antitumor activity were observed. An additional study is ongoing to explore different schedules of administration of LY2334737.

I. Garcia-Ribas, S. Callies, K.A. Benhadji, and C.A. Slapak are associated with Eli Lilly and Company.

The authors thank Susanne K. Rhoades and Roberta L. Balgobin for managing LY2334737 program, Takeshi Makiuchi for his study support, Enaksha R. Wickremsinhe for the establishment and validation of the dFdC assay, and Joanna Burke and Tess Lam for their statistical support and analysis.

This study was financially supported by Eli Lilly and Company.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.