Abstract
4′-Thio-β-
Fatigue (or weakness) is often a complication of chemotherapy. Effects can be mild and flu-like or severe and debilitating requiring bed rest. The fatigue may be due to drug-related anemia, thyroid or adrenal gland impairment, or centrally mediated via the hypothalamic-pituitary-adrenal (HPA) axis. Often in cancer patients, the cause is unknown (Bower et al. 2002). In the clinical investigation of a new oncology agent, 4′-thio-β-
4′-Thio-β-
Mechanisms associated with differences between antitumor activity of ara-C and OSI-7836 are not completely understood. The later compound, however, is retained longer in cancer cells. These compounds, as well as gemcitabine, require activation to triphosphate (TP) nucleotide metabolites (Johnson 2000; Parker et al. 2000), and are believed to act as analogs of deoxycytidine, competing with dCTP for incorporation into growing DNA strands, resulting in DNA chain termination, and ultimately triggering cell death (Secrist et al. 1995; Tiwari et al. 2000). At an equally toxic dose, OSI-7836TP accumulates more slowly, with approximately 100-fold less incorporation into DNA compared to ara-CTP. However, OSI-7836TP demonstrates a twofold longer half-life, slower metabolism, and greater cellular retention, resulting in higher overall exposure to the active TP metabolite over time. After 72 h, 1% of the initial concentration of the OSI-7836TP remains, compared to 0.01% of the initial concentration of ara-CTP (Parker et al. 2000).
To support the clinical development of OSI-7836, the overall drug toxicity was assessed in beagle dogs. In order to evaluate the potential differences in the toxicity due to frequency of administration, two different dose regimens were studied, each consisting of two 21-day cycles. In the first regimen, drug was administered on a day 1 and 8 schedule (drug administered on days 1, 8, 22, and 29), whereas in the second regimen drug was administered on a day 1 to 3 schedule (drug administered on days 1 to 3 and 22 to 24). For the treatment groups, the same total amount of drug was administered for each dosing regimen. Each dose was delivered as a 30-min intravenous (IV) infusion. The pharmacokinetic (PK) profile of each dosing regimen was also assessed.
Because patients experienced fatigue as a common dose-limiting toxicity, additional studies were performed in beagle dogs to better understand this effect. Hypotheses generated from the literature for fatigue included direct suppression of thyroid gland and/or adrenal gland function or an effect on the HPA axis. For these studies, dogs received a single 12 mg/kg IV dose of OSI-7836. Serum thyroxine (T4), adrenocorticotropic hormone (ACTH) and cortisol levels were evaluated pre- and post dose. In addition, a standard ACTH stimulation test to assess maximal adrenal function in the dog using synthetic ACTH, cosyntropin (Cortrosyn; Organon Pharmaceuticals, West Orange, NJ 07052), was conducted pre- and post OSI-7836 administration (Feldman 1995).
MATERIALS AND METHODS
Materials
OSI-7836 (C9H13N3O4S, molecular weight [MW] = 259.28) was supplied as a white-to-tan lyophilized cake or powder in a single use 20 ml type I glass vial (Gilead Sciences, San Dimas, CA).
Animals
All animal work was conducted in accordance with the Guiding Principles in the Use of Animals in Toxicology. Young adult beagle dogs (5 months) were obtained from Covance Research Products (Kalamazoo, MI). Dogs were in the weight range of 5.5 to 9.5 kg and ∼5 months of age. Lab Diet Certified Canine Diet no. 5007 (PMI Nutrition International, St. Louis, MO) and filtered tap water were provided ad libitum, except during designated fasting periods. Environmental controls were set to maintain temperatures of 64°F to 73°F, with a relative humidity of 34% to 70%. A 12:12-h light:dark cycle was maintained in the animal room. Animals were acclimated for a minimum of 19 days prior to study start. Animals were individually housed in dog runs with access to a minimum of 30 min of exercise three times per week.
Study Design
For the routine toxicity assessments, male and female dogs (3/sex) were used and the initial day of dosing was designated as study day 1. Dosages (3, 12, or 25 mg/kg/cycle) or saline vehicle in the control group were administered intravenously via the cephalic vein. Two schedules were assessed: one day 1 and 8 (1.5, 6, or 12.5 mg/kg/day) and one day 1 to 3 (1, 4, or 8.33 mg/kg/day) for dose administrations over two 21-day cycles.
Additional studies were conducted to assess the effects on lymphocyte subsets and on the thyroid and the HPA axes; dogs (2/sex) received a single 12 mg/kg IV bolus dose of OSI-7836. Blood samples were collected for lymphocyte subset evaluation by flow cytometry. Cells were stained for CD4 (T-helper), CD8 (T-suppressor), or CD21 (B–cell) surface markers. Serum T4, cortisol, and ACTH were measured predose and over 144 hours post dose.
Following a 14-day washout, the ACTH stimulation test was conducted before and after a second administration of 12 mg/kg IV bolus dose of OSI-7836. Blood samples were first collected for cortisol measurements prior to and 4 h after an IV bolus injection of OSI-7836 at a dosage of 12 mg/kg. Immediately after collection of the second blood sample, the ACTH stimulation test using an IV injection of 250 μg cosyntropin was conducted and blood samples for cortisol measurements were collected from the jugular vein 30 and 60 min post ACTH injection. The ACTH stimulation test was repeated 72 h post OSI-7836 administration. Additional serum samples for cortisol measurements were collected on days 3, 6, 10, and 14 post ACTH administration and frozen at ∼−20°C until analyzed. Serum cortisol was then measured in all samples during the same assay run.
Toxicological Assessment
Body weight data, food consumption data, physical examinations, ophthalmoscopic examinations, body temperature data, electrocardiographic examinations, clinical pathology samples, and plasma samples for drug concentrations were collected. Samples for hematology and clinical chemistry were collected predose and 3, 7, and 14 days following the end of the first drug cycle (day 8 for schedule 1 and day 3 for schedule 2).
At study termination (study day 50 for dose schedule 1 and day 45 for dose schedule 2) organ weights were measured for the liver, spleen, thymus, kidneys, adrenal glands, and brain. From the complete tissue list recommended by the Society of Toxicologic Pathology, tissues were collected, fixed in formalin, processed, and examined microscopically (Bregman et al. 2003).
Evaluation of Thyroid and Adrenal Gland Function
Individual samples were collected in glass tubes without anticoagulant, allowed to clot for approximately 20 min, the serum rapidly harvested, transferred to plastic tubes and frozen at −20°C. All hormones were measured by an Immunolite Chemiluminescent System (DPC, Los Angeles, CA). Retrospectively, we realized that the optimal sample management conditions for ACTH utilizes glass EDTA tubes and plasma for analysis; however, the baseline detection of measurable circulating levels of ACTH within the reference range for dogs supports the use of the data.
Initially, the T4, ACTH, and cortisol were measured prior to and 6 days after the IV bolus injection of 12 mg/kg OSI-7836. Pre- and post–OSI-7836 administration hormone values were compared to determine an effect. Blood was also collected for cortisol measurements prior to and 4 h after an IV bolus injection of normal saline (2 ml/kg) to assess potential procedure-related effects. Synthetic ACTH was used to stimulate adrenal gland secretion of cortisol before and following the administration of OSI-7836 as previously described. Blood samples for cortisol measurements were collected from the jugular vein prior to and 30 and 60 min post ACTH stimulation. Additional serum samples for cortisol measurements were collected 3, 6, 10, and 14 days post ACTH administration. All samples were stored frozen until assayed for cortisol. All blood samples for cortisol measurements were collected and all ACTH stimulation testing was conducted at approximately the same time of day for all phases of the study.
Plasma Protein Binding
OSI-7836 protein binding was measured in pooled dog plasma (Harlan Bioproducts, Indianapolis, IN) in the presence and absence of 0.4 mM tetrahydrouridine (THU) (Calbiochem Corporation, La Jolla, CA) to inhibit deamination. Plasma aliquots (0.5 ml) were fortified with 28000 DPM (12.6 nCi) of 14C-OSI-7836 (38 Ci/mMol; Moravek, Brea, CA) and binding was allowed to occur at 37°C for 30 min at final OSI-7836 concentrations of 100, 10, 1 or 0.1 μg/ml. A Centrifree Micropartition System (Amicon, Danvers, MA) was used to separate unbound drug. Binding at each concentration was analyzed in triplicate by liquid scintillation counting. The fraction unbound was calculated as DPM/ml ultrafiltrate divided by DPM/ml fortified plasma.
Determination of Drug Concentrations in Plasma
Venous blood samples were collected from all animals prior to each dose and immediately at the end of each infusion. Blood samples were collected at 0.25, 0.75, 2, 4, 7, and 10 h following the end of infusion on each day drug was administered in cycle 1, with the exception of day 2 on the day 1 to 3 regimen. All samples were collected in tubes containing EDTA and 0.4 mM THU. Samples were centrifuged, plasma harvested, and kept at −20°C until analyzed. The OSI-7836 and 4′-thio-ara-U concentrations were determined by a validated high-performance liquid chromatographic (HPLC) mass spectrometric (MS) method (Cedra Corporation, Austin, Texas). The range of the assay was 1.00 to 500 ng/ml for OSI-7836 and 5.00 to 2500 ng/ml for 4′-thio-ara-U. THU treated dog plasma was fortified with internal standard, 3′-azido-2′, 3′-dideoxyuridine (3-ADU), and samples were extracted with C18 solid phase extraction cartridges. Following a wash step, the analytes were eluted from the cartridge with methanol and evaporated to dryness under nitrogen. The dried extract was reconstituted and an aliquot was injected onto a SCIEX API 3000 LC-MS-MS equipped with a silica column. Peak areas of the m/z 260 to 112 OSI-7836 product ion and the m/z 261 to 113 4′-thio-ara-U product ion were measured against the m/z 254 to 113 3-ADU product ion of the internal standard. Quantification was performed using separate weighted (1/x 2) linear least squares regression analyses generated from fortified dog plasma combined calibration standards prepared immediately prior to each run.
Pharmacokinetic Analyses
The OSI-7836 and 4′-thio-ara-U pharmacokinetic parameters were assessed by noncompartmental analysis using WinNonlin version 3.1 (Pharsight, Mountain View, CA). Model 202 with the log/linear trapezoidal rule was used. However, for the day 3 dose of the day 1 to 3 schedule, NCA model 202 at steady state was used with the log/linear trapezoidal rule. For OSI-7836, the last three measurable time points were utilized for the estimation of the plasma elimination half-life. For 4′-thio-ara-U, time points from 4.5 h from the start of infusion until the last concentration measured (C last) were utilized to estimate the half-life. Pharmacokinetic parameters were determined for each animal in the study. From values obtained, the mean and standard deviation of each parameter were determined.
Statistical Analyses
Statistical Analysis of Pharmacokinetic Parameters
Statistical analysis was performed to test for differences (on the logarithmic scale) in dose and gender for dose-normalized area under the curve (AUC) and maximal plasma concentration (C max). This analysis used linear regression to model dose-normalized AUC and C max as a function of dose, and included a Dose × Sex interaction term. Repeated measures analyses were performed to test for differences between days, with both gender and dose effects included in the model.
Statistical Analysis of Other Parameters
Statistical analyses were performed on the clinical pathology (clinical chemistry, coagulation, hematology, urine specific gravity) parameters for each gender to determine differences due to schedule (day 1 and 8 dosing versus day 1 to 3 dosing), cycle (cycle 1 versus cycle 2), and dose group (vehicle, 1.5, 6, and 12.5 mg/kg/dose for the day 1 and 8 dosing, vehicle, 1, 4, and 8.33 mg/kg/dose for the day 1 to 3 dosing). The statistical analyses tested for pairwise comparisons as well as for overall effects.
Repeated-measures analyses of variance (rmANOVA) were performed for each clinical pathology parameter and gender to test for effects due to cycle, schedule, dose group, and study day. The models included effects for cycle, schedule, dose group, day within cycle, and animal. Rank-transformed values were analyzed when deviations from Normality were observed. The Dunnett’s tests and pairwise contrasts were performed to test for differences between the treatment and control groups (Dunnett’s) on the same day, to compare cycles on the same day of cycle (pairwise contrasts), to compare between schedules on the same day (pairwise comparisons), or to compare between the baseline or other time points within a group for each schedule and cycle (Dunnett’s).
ANOVA with Dunnett’s comparisons were performed to test for differences among treatment groups for each schedule and cycle on each day for body weights. Similar analyses were performed to test for differences in organ weights on three metrics: absolute organ weight, organ weight as a percentage of body weight, and organ weight as a percentage of brain weight. All comparisons were considered to be significant if p ≤.05.
RESULTS
In both dosing regimens, dogs had mild clinical toxicity of a similar magnitude at 12 or 25 mg/kg/cycle. There was a 5% decrease in body weight in the day 1 to 3 schedule in high-dose males for both cycles, but no effect was seen in animals on the day 1 and 8 dosing regimen. The body weights did recover after the test article was removed. Two animals exhibited emesis on days 1 and 2 following test article administration in the day 1 to 3 schedule. There were no significant differences (p >0.05) in the mean weight of adrenal glands, the thyroid glands, kidneys, or livers of treated dogs when compared to the control animals in either schedule (data not shown). There were no effects on electrocardiograms (ECGs), food consumption, ophthalmologic examinations, or body temperature.
Myelosuppression was observed in both regimens. The 25 mg/kg/cycle groups had lower total white blood cell (WBC) counts, absolute neutrophil counts, and absolute lymphocyte counts compared to the control animals for both cycles (Table 1). All WBC counts returned to normal values 7 days after the last dose of OSI-7836. Compared to control values, platelet counts were decreased in the high–dose groups for both schedules during both cycles (Table 1). There was an enhanced effect in the day 1 to 3 dosed animals, with maximal reductions 3 days after the completion of each drug cycle, but this effect resolved within 7 to 10 days. There was a mild increase in fibrinogen in animals on the day 1 and 8 schedule and a moderate increase in fibrinogen in the day 1 to 3 schedule for both cycles in the 25 mg/kg/cycle groups that partially resolved 7 days after dosing. Dogs on the day 1 to 3 dosing schedule also had a dose-dependent reduction in reticulocyte counts compared to control dogs during both cycles in all dose groups 3 to 7 days post dosing. The maximum reductions were 35%, 48%, and 67% for 3, 12, and 25 mg/kg/cycle, respectively. All other clinical pathology end points were unaffected. These counts had partially recovered by 19 days post dosing.
Treatment-related microscopic changes were limited to dose-dependent mild to moderate segmental atrophy of the seminiferous tubules at the mid and high doses of OSI-7836 in the day 1 and 8 regimen, one of three dogs receiving 12 mg/kg/cycle and two of three dogs receiving 25 mg/kg/cycle. Dogs on the day 1 to 3 schedule (two of three dogs receiving 12 mg/kg/cycle and all dogs receiving 25 mg/kg/cycle) demonstrated changes. The effect was dose related. The atrophy was characterized by an apparent absence of the various stages of spermatogenesis within segments of the seminiferous tubules. The affected tubules were lined by tall sustentacular cells and were likely to include residual stem cells capable of reestablishing spermatogenesis. These microscopic changes correlated with testicular weight reductions of 28% to 55% compared to control animals. The maximally tolerated dose (MTD) was 25 mg/kg/cycle for both dose regimens. The no-observed-adverse-effect level (NOAEL) was 3 mg/kg per cycle for both regimens based upon the histological and weight changes in the testes.
Lymphocyte Evaluation
The effect of OSI-7836 on subpopulations of lymphocytes was evaluated in an ensuing study in which dogs received a single 12 mg/kg dose by an IV infusion. The WBC counts again declined in this experiment with a nadir 24-h post drug administration, including of a 72% reduction in the absolute neutrophil count and a 78% reduction in the absolute lymphocyte count. Cell surface markers for B- and T-cell subpopulations indicated that the CD4, CD8, and CD21 populations were reduced (Table 2). These reductions were still apparent 72 h post dosing, although some recovery was evident. All three subpopulations appeared to be equally decreased by treatment with OSI-7836.
Evaluation of Thyroid and Adrenal Gland Effects
Following the single IV administration of OSI-7836 in dogs, the T4 concentrations approximated baseline values (Table 3). However, the cortisol levels in all animals were markedly reduced and remained reduced at the 144-h timepoint at levels that were below the limit of quantification of the assay (1 μg/dl). The ACTH levels were normal prior to OSI-7836 administration, but they were reduced following administration and remained reduced to the 144-h time point where all values were below the limit of quantification of the assay.
Following a 14-day washout, the ACTH stimulation test was conducted on these same animals prior to and after a second IV dose at 12 mg/kg OSI-7836. There was an ACTH stimulated increase of ∼5-fold in serum cortisol levels to 10.33 μg/dl before OSI-7836 administration. The ACTH stimulation test conducted immediately after or 3 days following the administration of OSI-7836 also demonstrated normal adrenal cortisol release (Figure 2).
Plasma Pharmacokinetics
OSI-7836 exhibited negligible binding to plasma proteins either in the presence or absence of tetrahydrouridine. Across all concentrations evaluated, the mean fraction unbound was 0.983 with a 95% confidence interval of 0.963–1.003 (n = 8). Thus, total plasma concentrations reported here are equivalent to free drug concentrations.
Two 21-day dose cycles of 3.0, 12.0, or 25 mg/kg/cycle were delivered by two different dosing schedules: either on study days 1 and 8, 22 and 29 administered at 1.5, 6.0, or 12.5 mg/kg/day or on study days 1 to 3 and 22–24 administered at 1.0, 4.0, or 8.33 mg/kg/day. Full pharmacokinetic sampling was performed only during cycle 1 (Figure 3), whereas trough and end of infusion concentrations were performed during cycle 2. Peak OSI- 7836 C max occurred near the end of infusion with a mean T max of 0.51 hours from the start of the 0.50-h infusion. Following C max, plasma levels declined with a mean (SD) elimination half-life of 2.2 (0.79) h.
Following the first OSI-7836 dose administration on study day 1, both plasma C max and plasma AUC0 → ∞ were dose proportional (Figure 4). No significant differences in plasma AUC0 → ∞ between genders were observed (p =.5684). The mean (SD) plasma clearance on study day 1 across all dose groups was 352 (74.4) ml/hr·kg (n =36), whereas the mean (SD) volume of distribution at steady state (V ss) was 896 (178) ml/kg (n = 36).
No significant differences in OSI-7836 plasma C max, AUC0 → τ or AUC0 → ∞ between the day 1 and day 8 or between the day 1 and day 3 doses were observed (p ≥.1738). However, for the 25 mg/kg/cycle group in the day 1 to 3 schedule, measurable OSI-7836 trough concentrations (5.05 to 13.9 ng/ml) of OSI-7836 were observed in both dose cycles. Plasma C max values in dose cycle 2 were similar to those observed in dose cycle 1(data not shown).
Across all dose groups, plasma concentrations of the inactive deaminated metabolite 4′-thio-ara-U peaked approximately 2.5 h following the start of the study day 1 OSI-7836 infusion and declined with a mean (SD) terminal half-life of 4.0 (0.76) h. Plasma 4′-thio-ara-U AUC0 → ∞ was also dose proportional (p =.1039) and no substantive differences in 4′-thio-ara-U plasma AUC between the day 1 and 8 or between the day 1 to 3 dose schedules were observed (data not shown). Across all dose groups on day 1 the median (range) plasma AUC0 → ∞ of the 4′-thio-ara-U was 17.1% (8.9% to 28.1%) of the parent drug (n =36).
DISCUSSION
OSI-7836 was evaluated in beagle dogs for toxicity, PK pro-filing, and as a potential species to assess human clinical fatigue. The major adverse effect demonstrated in male dogs was testicular atrophy with histopathological changes. The compound demonstrated an overall spectrum of adverse effects similar to other nucleoside analogues primarily including emesis and myelosuppression (Donehower, Karp, and Burke 1986; Toroutoglou et al. 1998). The nadir for myelotoxicity occurred 3 to 7 days after the final OSI-7836 dose for each schedule and returned to normal within 10 days. All clinical pathology changes returned to normal ranges within 19 days following OSI-7836 administration, with the exception of fibrinogen, absolute reticulocyte, and lymphocyte counts, which only partially recovered. Similar to other members in this class of drugs (e.g., gemcitabine), toxicity was influenced by the schedule of administration (Donehower, Karp, and Burke 1986; Herzig et al. 1987, Toroutoglou et al. 1998). Daily drug administration for 3 days was more toxic than the same total dose administered as a single weekly infusion. A second cycle of OSI-7836 neither exacerbated myelotoxicity nor caused cumulative toxicity under the evaluated conditions.
After a brief distribution phase, OSI-7836 was eliminated with a half-life of approximately 2.2 h. OSI-7836 displayed dose-linear pharmacokinetics across the dose range of 1.0 to 12.5 mg/kg following the day 1 dose and no changes in clearance were observed between the day 1 and 8 or between the day 1 and 3 doses within cycle 1. No changes in peak and trough plasma concentrations were observed in cycle 2 (data not presented).
Although in vitro studies have shown that pyrimidine nucleoside deaminase activity is lower in dogs than in most other species tested (Camiener and Smith 1965; Ho et al. 1975), the deamination of OSI-7836 still represented a significant pathway of elimination. The plasma clearance rate of OSI-7836 [352 ml/(hr·kg)] was slower than the reported value for cytarabine (ara-C) of 681 ml(hr·kg) in dogs (Scott-Moncrieff et al. 1991). This is consistent with earlier in vitro work indicating that OSI-7836 is a much less efficient substrate than cytarabine for both cytidine deaminase and deoxycytidine 5′-monophosphate deaminase in human leukemic CEM cell lines (Parker et al. 2000).
The clearance of OSI-7836 was similar to the plasma clearance observed for gemcitabine in dogs following a 3 mg/kg IV dose [Dose/AUC = 372 ml/(hr·kg)] (Shipley et al. 1992). Interestingly, the AUC of the deaminated metabolite, 4′-thio-ara-U, was approximately three-fold greater than that of the parent drug (Shipley et al. 1992). This result contrasts sharply with OSI-7836, where the AUC of the uracil metabolite was only 17.1% of the AUC of the parent drug. These data suggest that the clearance rates of these two metabolites differ appreciably from one another.
Our results indicate that the beagle dog was a useful species for the assessment of one possible cause of fatigue in cancer patients undergoing chemotherapy: centrally-mediated impairment of the HPA axis. We show that OSI-7836 did not adversely affect thyroid function in the dog, but, it did cause a marked reduction in the circulating cortisol and ACTH concentrations. The reduction in the circulating cortisol concentration was attributed to a deficiency of ACTH synthesis and/or release from the pituitary. The ACTH stimulation test prior to OSI-7836 administration demonstrated normal adrenal function with an ∼ five-fold increase in serum cortisol levels further supporting a centrally mediated effect.
Neurotransmitters from the central nervous system modulate the release of the hypophysiotropic hormones, corticotrophin-releasing hormone (CRH), and arginine vasopressin (AVP) from the hypothalamus. CRH and AVP are the predominant stimulating neurohormones for ACTH secretion from the anterior pituitary. Following its release from the pituitary into the circulation, ACTH stimulates the synthesis and secretion of cortisol by the adrenal cortex. Cortisol inhibits ACTH release; a negative feedback effect. The structure of CRH is identical in humans and dogs and canine ACTH differs from human ACTH by only one amino acid residue (Mol et al. 1994). One interpretation of our results is that OSI-7836 or a metabolite exerts a negative feedback for ACTH synthesis and/or secretion.
The findings in our study are consistent with clinical findings of low serum cortisol levels in breast cancer patients with fatigue versus those not experiencing fatigue (Bower et al. 2002). Findings from our phase 1 studies with OSI-7836 indicated fatigue was a dose-limiting toxicity in some patients (Siu et al. 2003; de Jonge et al. 2003). Studies with gemcitabine have also demonstrated fatigue, often mild and flulike, but infrequently debilitating (Brand, Capadano, and Tempero 1997; Catimel et al. 1994).
In conclusion, the OSI-7836 toxicities manifested as myelosuppression, testicular atrophy, and gastrointestinal effects may be expected effects of a potent cytotoxic compound. Hematologic findings showed complete or near complete recovery. We showed that thyroid gland function and the ACTH stimulation test for adrenal cortisol reserve were normal following the administration of OSI-7836. However, the circulating cortisol and ACTH concentrations were markedly decreased following the administration of OSI-7836 suggesting drug-related impaired function of the HPA axis. These findings further suggest that the dog may be a useful model for assessing one possible cause of fatigue in human patients undergoing chemotherapy.
Footnotes
Figures and Tables
Acknowledgements
The authors kindly thank Dr. Ray Bendele and Dr. Don Maul for insightful discussions related to this research.