Brendan M. MCGuire, Igor A. Zupanets, Mark E. Lowe, Xunjun Xiao, Vasyliy A. Syplyviy, Jon Monteleone, Sharron Gargosky, Klara Dickinson, Antonia Martinez, Masoud Mokhtarani,5 and Bruce F. Scharschmidt5
Glycerol phenylbutyrate (GPB) [or glyceryl tri- (4-phenylbutyrate), also referred to as HPN- 100] is an oral investigational agent under development for hepatic encephalopathy (HE) and urea cycle disorders (UCDs). It is a pro-drug of phenylbutyric acid (PBA), currently marketed as sodium phenylbutyrate (NaPBA), for the treatment of UCDs. It consists of glycerol with three molecules of PBA linked as esters. GPB is a pale yellow, nearly odorless and tasteless oil, whereas NaPBA has palatability issues, high sodium content, and high pill burden. The maximum approved daily dose of NaPBA (20 g) corresponds to 40 tablets containing ≈2,400 mg of sodium, which exceeds the daily allowance of 2,300 mg/day recommended in the US Department of Health and Human Services Dietary Guidelines for Americans, 2005 for the general population and 1,500 mg/day for individuals with hypertension or sodium retaining states.1 The corresponding dose of GPB is 17.4 mL, which contains no sodium.
NaPBA mediates excretion of waste nitrogen as shown in Fig. 1. PBA is absorbed from the intestine and converted by way of b-oxidation to the active moiety, phenylacetic acid (PAA). PAA is conjugated with glutamine in the liver and kidney by way of Nacyl coenzyme A-L-glutamine N-acyltransferase to form phenylacetylglutamine (PAGN).2 Like urea, PAGN incorporates two waste nitrogens and is excreted in the urine.
Because GPB contains no sodium and may be better tolerated than NaPBA, its safety and pharmacology were studied in healthy adults and adults with cirrhosis, as was the handling of GPB by human pancreatic lipases. Monte Carlo simulations were performed to assess metabolites blood levels and therefore clinical safety at doses approximating the highest approved dose of NaPBA for treatment of UCDs.
added at time zero. PLRP2 activity was determined with 10 lg of PLRP2 6 10 lg of colipase added at time zero. CEL activity was determined with 10 lg of CEL in the absence of colipase. Each reaction was monitored for 5 minutes. The reaction rate was determined from the slope of the linear curve. The rate of 100 mM NaOH titration during the assay was set to maintain a constant pH of 8.0 for PTL and PLRP2 and 50 mM NaOH for CEL. The activity of PTL and PLRP2 against tributyrin and triolein in 1 mM TrisHCl (pH 8.0), 2 mM CaCl2, 150 mM NaCl, and 4 mM sodium taurodeoxycholate and of CEL against tributyrin and triolein in the same buffer with 10 mM sodium cholate and no taurodeoxycholate was determined using the same methodology.
UP 1204-001. This was a phase 1, randomized, crossover, open-label study designed to assess safety, tolerability, pharmacokinetic (PK) equivalence, and bioequivalence in healthy adult subjects. Intravenous AMMONUL (a 10%/10% solution of sodium phenylacetate and sodium benzoate) and a formulated oral preparation of GPB were administered in addition to GPB (unformulated) and NaPBA, but only the results for NaPBA and unformulated GPB are reported in this study. Subjects received a single dose of either NaPBA or GPB on separate dosing days, at least 7 days apart. NaPBA and GPB were administered at a dose equivalent to 3 g/m2 of PBA. PK samples were taken predose and 0.25, 0.5, 1, 1.5, 2, 3, 4, 6, 8, 12, 24, and 48 hours postdose. Urine was collected from 0-4, 4-8, 8-12, and 12-24 hours postdose. PK variables were calculated for PBA, PAA, phenylacetylglycine (PAG), PAGN, phenylbutyrylglycine (PBG), and phenylbutyrylglutamine (PBGN). A test for intact GPB was also conducted in subjects receiving GPB.
Bioequivalence was assessed by calculating 90% confidence intervals for the ratio of geometric means between test and reference treatments. The ratios and confidence intervals were calculated in an analysis of variance model for log-transformed pharmacokinetic variables including treatment, period, and the treatment by period interaction as fixed effects and subject as a random effect.
UP 1204-002. This was an open-label study of the safety and PK equivalence of GPB in subjects with cirrhosis (Child-Pugh score A, B, or C [n ¼ 8 in each group]) compared with age- and sex-matched healthy subjects with normal hepatic function (n ¼ 8). Subjects received a single oral GPB dose (100 mg/kg/day) on day 1, two doses per day (12 hours apart) on days 8-14 (200 mg/kg/d), and a single dose on day 15 (100 mg/kg/d). The single oral dose on day 1 was a fasting dose, whereas the first dose on day 8 was given with a meal. The last GPB dose was administered on the morning of day 15 and was followed by 48 hours of plasma PK sampling and urine collection. PK blood samples were drawn at 0.25, 0.5, 1, 1.5, 2, 3, 4, 6, 8, 12, and 24 hours postdose on days 1, 8, and 15, and at 48 hours after dosing on days 1 and 15. Urine was collected from 0-4, 4-8, 8-12, and 12-24 hours postdose on days 1, 8, and 15 and at 24-48 hours postdose on days 1 and 15. PK samples were drawn fasting prior to the morning dose (trough) and 2 hours postdose on days 9-14. A 12-lead electrocardiogram was performed at screening on days 0 and 7, 2 hours postdose on days 1 and 15 (between 9:00 AM and 10:00 AM), and at follow-up (7 days after day 15).
Plasma and urine PK parameters were calculated for all subjects and summarized with descriptive statistics (number of patients, mean, standard deviation, median, minimum, and maximum). PK parameters were calculated using time concentration profiles for each subject, including area under the concentration versus time curve from time 0 (predose) to 24 hours (AUC0- 24), calculated using the linear trapezoidal rule; maximum plasma concentration at steady state (Cmax); and the time of maximum plasma concentration at steady state. The amount of PAGN excreted in urine over 24 hours was calculated from urinary concentration (by multiplying the urinary volume with urinary concentrations).
Monte Carlo simulations were performed to predict the average and uncertainty (5% and 95% prediction intervals) for simulated plasma PBA, PAA, and PAGN concentrations in a hypothetical clinical trial with 5,000 cirrhotic subjects dosed with GPB at 9 mL (≈9.9 g) twice daily. A concentration time profile was developed for each analyte corresponding to the mean as well as the 5% of patients with the highest and lowest levels.
The population PK model and corresponding PK parameter estimates used for the Monte Carlo simulations were developed using Nonmem VI (NONMEM; ICON Development Solutions, Ellicott City, MD) and PK data from protocols UP 1204-001 and UP 1204-002 and a phase 2 study in UCD patients (protocol UP 1204-003).9 Simulations were preformed using Trial Simulator software (TS2; Pharsight Corporation Inc., Mountain View, CA), assuming dosing at 8:00 AM and 6:00 PM (to coincide with breakfast and dinner), 7 days of dosing to ensure steady state concentrations were achieved, and frequent sampling (daily samples at 0, 0.25, 0.5, 0.75, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, and 24 hours). Because body surface area was a significant demographic covariate for clearance and volume of distribution parameters in the PK model, simulation was used to generate this demographic variable for each of the 5,000 hypothetical patients.
In Vitro Hydrolysis of GPB by Pancreatic
PTL, PLRP2, and CEL all hydrolyzed GPB (Table1). The specific activity (lmole fatty acid released/min/mg protein or U/mg) of PTL (≈600 U/mg) was ≈27-fold higher than that of PLRP2 (≈22 U/mg) when both were assayed in the presence of colipase and 4 mM sodium taurodeoxycholate and ≈2.4-fold higher than that of CEL (≈250 U/mg). For comparison, under the same assay conditions, the activity against tributyrin was 4,600 6 30 U/mg for PTL, 200 6 9.0 U/mg for PLRP2, and 260 6 12.0 U/mg for CEL and against triolein was 1,600 6 153 U/mg for PTL, 120 6 40 for PLRP2, and 30.3 6 6.0 for CEL.
Patient Demographics and Disposition
Twenty-four healthy adults were enrolled in protocol UP 1204-001, of whom 22 received each of the study drugs and completed the study according to the protocol.
Thirty-two subjects met the entry criteria and enrolled in protocol UP 1204-002 (Supporting Table 1). All subjects completed the study and were included in the analyses. Subject groups were generally well matched. There were more men than women in Child-Pugh groups A and B and equal numbers of men and women in Child-Pugh C and the healthy volunteer groups. None of the subjects in Child-Pugh A or the healthy volunteer group had HE or ascites. All subjects in Child-Pugh group B had mild ascites and stage I HE, and all subjects in Child-Pugh group C had mild or moderate ascites and stage I or II HE. Serum albumin, creatinine, and international normalized ratio were similar in all subject groups. Serum bilirubin increased with the degree of hepatic impairment, ranging from 0.74 mg/dL in the healthy volunteers to 3.46 mg/dL in Child-Pugh group C. Mean (standard deviation) Child-Pugh and model for end-stage liver disease (MELD) scores, respectively, increased commensurate with Child-Pugh grade (A ¼ 5.8 [0.5] and 7.3 [1.3]; B ¼ 8.3 [0.5] and 8.6 [2.1]; C ¼ 10.6 [0.5] and 12.6 [2.8]) among the cirrhotic subjects, and all 32 subjects had negative drug screens and alcohol breath test results at all assessments.
Safety and Tolerability (Supporting Table 2)Protocol UP 1204-001. Twenty-one adverse events (AEs) were reported by 10 subjects while receiving NaPBA compared with six AEs by two subjects while receiving GPB. The most frequently reported AEs with NaPBA were dizziness (n ¼ 5), headache (n ¼ 4), and nausea (n ¼ 3). One patient reported epigastric discomfort and one patient reported vomiting (n ¼ 2) while taking GPB.
Protocol UP 1204-002. There were no SAEs or AEs leading to withdrawal during the study. Overall, AEs were reported in 26 of 32 subjects. Among healthy volunteers, five of eight reported AEs, whereas seven of eight subjects in each of the Child-Pugh groups reported AEs. The most common system organ class was investigations (18 subjects); increased body temperature was reported by 10 subjects with cirrhosis and decreased platelet count was recorded for four subjects in Child-Pugh group A and one subject in healthy volunteer group D. Other common classes of AEs included gastrointestinal complaints (n ¼ 11) and nervous system disorders (n ¼ 8), particularly headache (n ¼ 7). Most AEs were considered not related (n ¼ 9) or possibly related (n ¼ 20) to the study medication, and no AEs were considered definitely related. Analysis of vital signs including oral temperature did not reveal clinically or statistically significant changes from baseline. The highest mean temperature recorded in any treatment group at any time was 37.2°C, and the highest temperature recorded in any individual subject at any time point was 38.2°C. Abnormal laboratory safety findings were common in subjects with hepatic impairment. There was no consistently observed pattern among hematology, coagulation, or chemistry (including liver enzymes), and changes after 7 days of dosing with GPB were clinically insignificant. Clinically significant changes in electrocardiogram were not observed with GPB dosing, nor were changes observed in the QTc intervals.
Pharmacokinetic AnalysesProtocol UP 1204-001. NaPBA resulted in higher plasma levels (both Cmax and AUC) of PBA, PAA, and PAGN than GPB; the 90% confidence intervals for the ratio of geometric means of each metabolite following GPB compared with NaPBA extended below the commonly used lower bioequivalence level of 0.8 (Fig. 2 and Table 2). The mean plasma half-lives of PBA, PAA, and PAGN were 0.7 (6 0.1) hours, 1.2 (6 0.2) hours, and 1.7 (6 0.5) hours, respectively, after NaPBA administration. The mean plasma half-life of PBA after GPB administration was 1.9 (6 1.7) hours and ranged from 0.8 to 7.4 hours. The plasma halflives for PAA and PAGN after GPB administration ranged from 1.0 to 1.8 hours and 1.9 to 16.9 hours, respectively.
Urinary excretion of PAGN was higher following NaPBA than following GPB (Table 3). However, urinary collection of PAGN was incomplete at 24 hours following GPB dosing, as PAGN was still detectable in plasma at 24 hours. In contrast, PAGN following NaPBA dosing was undetectable in the plasma by 24 hours and PAGN elimination considered complete by 24 hours. Taking into account the pattern of urinary excretion and plasma levels following the single dose arm of study UP 1204-002, the 0-48 hours urine collection was split into 0-24 and 24-48 hours to calculate the percentage of urinary PAGN that occurred after 24 hours (Table 4). It is estimated that 15% of urinary PAGN excretion in patients on GPB occurred beyond 24 hours, the time the collection was terminated for study UP 1204-001. When corrected for the estimated 15% excreted after 24 hours, PAGN excretion was essentially the same following NaPBA and GPB administration (Table 3).
Protocol UP 1204-002. Intact GPB was not detected in the systemic circulation; nor were the minor metabolites PAG, PBG, and PBGN. AUC0-t and C max for PBA and PAA tended to be higher in Child-Pugh groups B and C than in Child-Pugh group A or in the healthy subjects group, but these changes were not statistically significant (Table 2). Plasma PAGN levels did not differ among the study groups.
No consistent differences between cirrhotic subjects and healthy subjects were observed in the plasma PK variables examined on days 1 (single dose and fasting) or 15 (after multiple doses and at steady state). There were also no statistically significant differences in the PK characteristics of GPB when given after fasting (day 1) or with a meal (day 8).
Plasma PBA concentrations returned to near predose levels between doses during multiple dosing days 8-15 and did not reach steady state. By contrast, PAA and PAGN predose concentrations increased during the first 2 to 4 days of multiple dosing but did not increase consistently thereafter, indicating that steady state had been reached (Fig. 3). After dosing on day 15, the extent of exposure to PAA, but not PBA, significantly correlated with hepatic impairment, increasing with worsening MELD score. During multiple dosing, PAA accumulation in Child-Pugh C cirrhotic subjects exceeded that in other groups. However, this trend was attributable to a single Child-Pugh C subject that showed unusually high levels of PAA assessed as Cmax and AUC (208.8 lg/mL and 2,245.51 [(lg/mL)/ hour], respectively) after GPB administration, compared with all other subjects. This subject, who received GPB at a dose of 7 mL twice daily, exhibited a clinical profile similar to the other Child-Pugh C subjects. Reanalysis omitting this subject’s data resulted in mean PAA C max and AUC levels for the Child-Pugh C group similar to other subject groups. During repeated dosing, similar but less profound patterns of increased PAA levels compared with their group mean were noted in one healthy subject (AUC 420.32 [(lg/mL)/ hour], Cmax 61.31 lg/mL) and one Child-Pugh B subject (AUC 938.85 [(lg/mL)/hour], Cmax 65.40 lg/mL).
Urinary PAGN. PAGN was the major metabolite excreted: 42%-49% of the GPB dose administered was excreted as PAGN on day 1, 25%-45% on day 8, and 58%-85% on day 15 (Tables 3 and 4). Very low amounts of PBA and PAA were excreted in the urine (≈ 0.05% of the total GPB dose). There were no statistically significant differences in the amount of PAGN excreted between any of the Child-Pugh groups and the healthy subjects. Urinary PAGN excretion was significantly greater in all groups after multiple dosing compared with single dosing, a result consistent with the larger daily GPB doses and higher plasma PAA and plasma PAGN observed during the first 2 to 4 days of multiple dosing, after which steady state appeared to have been reached.
Simulations of 9 mL bid dosing were consistent with the pharmacokinetic findings observed in protocol UP 1204-002. PBA levels did not accumulate with repeated dosing, and PBA trough concentrations were predicted to be at or near baseline levels. PAA and PAGN, by contrast, did exhibit accumulation with repeated dosing and achieved steady state within 4 days. Simulations projected the median PAA concentration to be severalfold lower than the no observed adverse event level in primates (331 lg/mL [results not shown]) or the levels reported to be associated with neurological symptoms in human studies10,11 (Fig. 4).
GPB was hydrolyzed by all pancreatic enzymes tested, including PTL, CEL, and PLRP2 in order of specific activity. Compared with GPB, the specific activity of PTL is ≈8-fold higher against tributyrin, which has the same fatty acid chain length as GPB but lacks the phenyl group, and is ≈2.5-fold higher against triolein, a major dietary fat and physiological substrate. These findings suggest that the intestinal handling of GPB is likely similar to dietary triglycerides. Because PLRP2 and CEL, unlike PTL, are expressed at birth, the findings further suggest that GPB may be digested by newborns, which is of particular relevance to UCDs that can present shortly after birth.12-20 This requires confirmation in clinical studies.
The safety and tolerability of GPB with short-term dosing was generally satisfactory and comparable to NaPBA. There were no SAEs or deaths and no clinically significant changes in laboratory parameters.
Metabolite peak blood levels were lower after singledose administration to healthy adults of GPB compared with NaPBA, and urinary excretion of PAGN was prolonged. These findings are consistent with the gradual release of PBA by pancreatic lipases when delivered in the form of GPB compared with NaPBA, and the absence of intact GPB in blood or urine suggests that intestinal breakdown of GPB is complete. Although the lower PBA blood levels following GPB compared with NaPBA might be interpreted as consistent with lower bioavailability, urinary PAGN excretion was similar between the two drugs, suggesting that PBA may be converted to PAA and PAGN before reaching the systemic circulation.
GPB and NaPBA were metabolized similarly by healthy adults and cirrhotic subjects to PBA and PAGN, and steady state blood levels of PBA, PAA, and PAGN were achieved within 4 days of GPB dosing for both healthy adults and cirrhotic subjects. Three subjects, including one healthy adult and two with cirrhosis exhibited higher steady state PAA blood levels than their respective group means, suggesting that there may be individual differences in the rates of PAA formation and/or disposition independent of the presence or absence of liver disease. Blood metabolite levels were generally similar between healthy subjects and cirrhotic subjects, except that PAA levels tended to be higher in Child-Pugh C subjects and correlated positively with MELD score. However, this difference was largely attributable to a single Child-Pugh C subject and should therefore be interpreted cautiously.
PAA is of particular interest in that neurological toxicity has been reported in cancer subjects administered high doses of PBA or PAA intravenously and were associated with PAA blood levels ranging from ≈499- 1,285 lg/mL.10,11 Although the blood levels observed in cirrhotic subjects were well below the levels reported by Thibault and colleagues to be associated with neurological symptoms, Monte Carlo simulation analyses were conducted to further assess the anticipated PAA plasma concentrations in a hypothetical clinical trial involving 5,000 cirrhotic subjects administered GPB in a daily dose of 9 mL bid, which is slightly above the equivalent maximum approved dose of NaPBA for UCD patients (20 g/day of NaPBA is equivalent to 17.4 mL of GPB). As depicted in Fig. 4, Monte Carlo simulation indicated that even the highest 5% of predicted PAA concentrations were well below those associated with toxicity in phase 1 oncology studies.
Urinary PAGN is also of particular interest because it is stoichiometrically related to nitrogen scavenging. Cirrhotic subjects excreted essentially the same amount of PAGN as healthy adults, indicating that GPB should enhance excretion of waste nitrogen and lower ammonia in patients with cirrhosis. Assuming that dietary protein is ≈16% nitrogen by weight, that approximately 47% of dietary nitrogen is excreted as waste nitrogen, and that ≈60% conversion of PBA delivered as GPB is converted to urinary PAGN, then a GPB dose of 9 mL twice daily—which is essentially the same as the maximum approved dose in UCD patients— would be expected to mediate excretion of waste nitrogen associated with ≈26 g dietary protein, or ≈0.4 g/kg body weight for a 70-kg adult male.16 This would represent a substantial contribution to waste nitrogen excretion in cirrhotic subjects, in whom urea synthetic capacity is known to be impaired, and would likely compare favorably with the presumed reduction in ammonia production with agents commonly used for HE such as nonabsorbable disaccharides or antibiotics.21,22 The effect of nonabsorbable disaccharides or antibiotics on nitrogen metabolism is difficult to measure, and there is not a clinically useful biomarker for these agents such as urinary PAGN for GPB.22 These considerations, in conjunction with the results of a phase 2 study that suggest that GPB is at least as effective as NaPBA in lowering blood ammonia in UCD patients,9 suggest that GPB has the potential to lower blood ammonia in patients with cirrhosis and warrants further exploration for the treatment of HE.
HEPATOLOGY, Vol. 51, No. 6, 2010