The Role of Metabolites in Bioequivalency Assessment. II. Drugs with Linear Pharmacokinetics and First-Pass Effect

Simulations were conducted to address the question of whether metabolite data are required for bioequivalence evaluation of immediate release formulations with drugs exhibiting linear pharmacokinetics and first-pass effect. Plasma level-time profiles were generated for parent drug and metabolite using relevant rate constants obtained from a bivariate normal distribution and designated random error. Simulation results showed that the need for metabolite data (C_max) in the assessment of bioequivalence depends on the relative variability between the absorption process of the drug and first-pass route for metabolite(s). The importance of metabolite C_max data in the evaluation of rate of availability is clearly demonstrated for drugs with a high degree of intra-subject variation in the first-pass metabolism compared to the absorption process of the drug. Under such conditions, a wider confidence interval was found for the metabolite rather than parent drug. Opposite results were obtained when the intra-subject variance was high for drug absorption relative to first-pass effect. Discrepancies were observed for the scenarios in which the elimination pathway of the metabolite is more variable than the absorption process of the drug. The simulation results were in agreement with real bioequivalence data. It is thus recommended that, in the absence of the information on the relative variability of absorption and first-pass process, both parent drug and metabolite data be included for documentation of bioequivalence, should the metabolite(s) play an important role in the determination of efficacy and safety of the drug.     

The Role of Metabolites in Bioequivalency Assessment. I. Linear Pharmacokinetics without First-Pass Effect

The estimation of bioequivalency using metabolite data was investigated for immediate release formulations with drugs exhibiting linear pharmacokinetics and no first-pass effect. This was accomplished by generating parent drug and metabolite plasma level profiles assuming formation and excretion rate-limited pharmacokinetic models with absorption rate constants obtained from bivariate normal distributions and designated random errors. Simulation results indicated that bioequivalence determination using C _maxof parent drug and metabolite was independent of the metabolite models as evaluated by confidence interval approach. However, a clear difference with respect to the outcome of bioequivalence evaluation arises depending upon the utilization of C _max values for the parent drug and metabolite. The major reason for this disparity was attributed to the minimal effect of the absorption process for the parent drug on the formation of the metabolite. This phenomenon results in an apparent lower intrasubject variability for C _max of the metabolite and, in turn, a tighter confidence interval for C _max of the metabolite in comparison with the parent drug. The simulated results have been found to be in agreement with the bioequivalency data for acetohexamide, allopurinol, procainamide, and sulindac. In all cases, the interval of the 90% confidence limit for C _max of the metabolite is always smaller than that of the parent drug, regardless of the drug pharmacokinetics and the level of error contained in the data.     

The Role of Metabolites in Bioequivalence

The role of metabolites in bioequivalence studies has been a contentious issue for many years. Many papers have published recommendations for the use of metabolite data based on anecdotal evidence from the results of bioequivalence studies. Such anecdotal evidence has validity, but the arguments lack weight because the "correct" answers are always unknown. A more promising area of exploration is recommendations based on simulated bioequivalence studies for which the "correct" answers are known, given the assumptions. A review of the literature, however, reveals scant evidence of attempts to apply to real data the pharmacokinetic principles on which the recommendations from simulated studies relied. We therefore applied those principles (based on estimates of intrinsic clearance after oral administration of the parent drug) to four bioequivalence studies from our archives, in which the parent drug and at least one metabolite were monitored. In each case, the outcome is discussed in the context of the complexity of the metabolic processes that impact on the parent drug and the metabolite(s) during the first passage from the intestinal lumen to the systemic circulation. Our observation is that no simple generalization can be made such that each drug/metabolite combination must be examined individually. Our recommendation, however, is that in the interests of safety, bioequivalence decision-making should be based on the parent drug whenever possible.     

Bioequivalence Approaches for Highly Variable Drugs and Drug Products

Over the past decade, concerns have been expressed increasingly regarding the difficulty for highly variable drugs and drug products (%CV greater than 30) to meet the standard bioequivalence (BE) criteria using a reasonable number of study subjects. The topic has been discussed on numerous occasions at national and international meetings. Despite the lack of a universally accepted solution for the issue, regulatory agencies generally agree that an adjustment of the traditional BE limits for these drugs or products may be warranted to alleviate the resource burden of studying relatively large numbers of subjects in bioequivalence trials. This report summarizes a careful examination of all the statistical methods available and extensive simulations for BE assessment of highly variable drugs/products. Herein, the authors present an approach of scaling an average BE criterion to the within-subject variability of the reference product in a crossover BE study, together with a point-estimate constraint imposed on the geometric mean ratio between the test and reference products. The use of a reference-scaling approach involves the determination of variability of the reference product, which requires replication of the reference treatment in each individual. A partial replicated-treatment design with this new data analysis methodology will thus provide a more efficient design for BE studies with highly variable drugs and drug products. The opinions expressed in this report by the authors do not necessarily reflect the views or policies of the Food and Drug Administration (FDA).     

Limits for the Scaled Average Bioequivalence of Highly Variable Drugs and Drug Products

Purpose. To provide a rational procedure for establishing regulatory bioequivalence (BE) limits that can be applied in determinations of scaled average BE for highly-variable (HV) drugs and drug products. Methods. Two-period crossover BE investigations with either 24 or 36 subjects were simulated with assumptions of a coefficient of variation of 10, 20, 30, or 40%. The decline in the fraction of accepted studies was recorded as the ratio of geometric means (GMR) for the two formulations was raised from 1.00 to 1.45. Acceptance of BE was evaluated by scaled average BE, assuming various BE limits, and, for comparison, by unscaled average BE. A procedure for calculating exact confidence limits in two-period studies is presented, and an approximate method, based on the linearization of the regulatory model, is applied. Results. A mixed model is proposed for average BE. Accordingly, at low variabilities, the BE limit is constant, ±BEL_o, generally log(1.25). Beyond a logarithmic, limiting, switching variability (_o), in the region of HV drugs, the approach of scaled average BE is applied with limits of ±(BEL_o /_o). It is demonstrated that the performance of the mixed model corresponds to these expectations. The effect of _o and of the resulting BE limits is also demonstrated. Scaled average BE, with all reasonable limits for HV drugs, requires fewer subjects than an unscaled average BE. In two-period studies, the exact and approximate methods calculating confidence limits yield very comparable inferences. Conclusions. Scaled average BE can be effectively applied, with the recommended limits, for determining the BE of HV drugs and drug products. The limiting, switching variability (_o) will have to be established by regulatory authorities.     

An Approach for Widening the Bioequivalence Acceptance Limits in the Case of Highly Variable Drugs

Purpose. Highly variable drugs pose a problem in bioequivalence assessment because they often fail to meet current regulatory acceptance criteria for average bioequivalence (80–125%). This paper examines alternative approaches to establishing bioequivalence. Methods. Suggested solutions have included alternate study designs, e.g., replicate and multiple dose studies, reducing the level of the confidence interval, and widening the acceptance limits. We focus on the latter approach. Results. A rationale is presented for defining wider acceptance limits for highly variable drugs. Two previously described methods are evaluated, and a new method having more desirable properties is proposed. Conclusions. We challenge the one size fits all current definition of bioequivalence acceptance limits for highly variable drugs, proposing alternative limits or goal posts which vary in accordance with the intrasubject variability of the reference product.     

Evaluation of Bioequivalence of Highly Variable Drugs Using Monte Carlo Simulations. I. Estimation of Rate of Absorption for Single and Multiple Dose Trials Using Cmax

Purpose. A Monte Carlo simulation study was done to investigate the effects of high intrasubject variation in clearance (CL), and volume of distribution (V) on the calculation of the 90% confidence interval (CI) for Cmax for single dose and multiple dose studies. Methods. Simulations were done for both immediate release and sustained release scenarios. The simulated data were compared with clinical data from bioequivalence studies performed on indomethacin and verapamil. Results. Previous reviews and simulations have shown that the probability of failure for the Cmax for single dose studies was always greater than that for multiple dose studies. However, the results for the simulated scenarios currently investigated indicate that if intrasubject (period-to-period) variation in CL and V is high (% CV's above 25%, and 12%, respectively), multiple dose studies can exhibit a higher probability of failure for Cmax than do single dose studies. Furthermore, Cmax values from studies performed with a sustained release scenario are more sensitive to changes in Ka, CL, and V than are results of studies on immediate release products. As an example, the probability of failure for immediate release products in simulated single dose studies is about 11% and 21% when the mean difference in Ka is 10% and 20%, respectively; while, the probability of failure for multiple dose studies is about 36% regardless of the difference in Ka. The corresponding values for the probability of failure for sustained release products were 25%, 53% for single dose studies and 39% for multiple dose studies. The simulations also indicate that changes in the fraction absorbed have a greater effect on the estimation of Cmax in multiple dose regimens than in single dose studies. Conclusions. The results from these investigations indicate that multiple dose studies do not necessarily always reduce variability in Cmax.     

Bioequivalence of a Highly Variable Drug: An Experience with Nadolol

Purpose. To assess the bioequivalence of nadolol 40mg and 160mg tablets (Zenith-Goldline Pharmaceuticals) using Corgard^® 40mg and 160mg tablets (Bristol-Meyers Squibb) as reference products, to estimate the effect of food in the gastrointestinal tract on nadolol bioavailability, and to evaluate the effectiveness of standard pharmacokinetic metrics AUC_t, AUC, and C_max in bioequivalence determinations. Methods. Four bioequivalence studies were conducted as described in the FDA Guidance. Four additional studies of varying designs were conducted to establish bioequivalence of the 40mg tablet in terms of C_max. Results. Fasted and food-effect studies of the 160mg tablet clearly established bioequivalence and revealed an unexpected reduction in nadolol bioavailability from test and reference products in the presence of food. The food-effect study of the 40mg tablet (80mg dose) revealed a similar reduction in bioavailability from each product. Fasted studies of the 40mg tablet (80mg dose) established bioequivalence in terms of AUC_t and AUC. However, C_max criteria proved extremely difficult to meet in the initial 40mg fasted study because of the large variability, leading to additional studies and ultimately requiring an unreasonable number of subjects. Conclusions. Final results clearly established bioequivalence of both strengths and characterized an unexpected food effect which did not appear to be formulation-related. However, the C_max of nadolol is only slightly sensitive to absorption rate and the relatively large variability of C_max reduces its effectiveness as a bioequivalence metric. Findings suggest that bioequivalence criteria for highly variable drugs should be reconsidered.     

Mean Residence Time of Oral Drugs Undergoing First-Pass and Linear Reversible Metabolism

Equations for the mean residence times in the body (MRT) and AUMC/AUC of a drug and its metabolite have been derived for an oral drug undergoing first-pass and linear reversible metabolism. The mean residence times of the drug or interconversion metabolite in the body after oral drug are described by equations which include the mean absorption time (MAT), the mean residence times of the drug or metabolite in the body after intravenous administration of the drug, the fractions of the dose entering the systemic circulation as the parent drug and metabolite, and the systemically available fractions of the drug (F _p ^p) or metabolite (F _m ^p). Similarly, the AUMC/AUC of the drug and metabolite after oral drug can be related to the MAT, ratios of the fraction of the dose entering the systemic circulation to the systemically available fraction, the first-time fractional conversion of each compound, and AUMC/AUC ratios after separate intravenous administration of each compound. The F _p ^p and F _m ^p values, in turn, are related to the first-pass availabilities of both drug and metabolite and the first-time fractional conversion fractions. The application of these equations to a dual reversible two-compartment model is illustrated by computer simulations.     

Predictions of the In Vivo Clearance of Drugs from Rate of Loss Using Human Liver Microsomes for Phase I and Phase II Biotransformations

Purpose The utility of in vitro metabolism to accurately predict the clearance of hepatically metabolized drugs was evaluated. Three major goals were: (1) to optimize substrate concentration for the accurate prediction of clearance by comparing to K_m value, (2) to prove that clearance of drugs by both oxidation and glucuronidation may be predicted by this method, and (3) to determine the effects of nonspecific microsomal binding and plasma protein binding. Methods The apparent K_m values for five compounds along with scaled intrinsic clearances and predicted hepatic clearances for eight compounds were determined using a substrate loss method. Nonspecific binding to both plasma and microsomal matrices were also examined in the clearance calculations. Results The K_m values were well within the 2-fold variability expected for between laboratory comparisons. Using both phase I and/or phase II glucuronidation incubation conditions, the predictions of in vivo clearance using the substrate loss method were shown to correlate with published human clearance values. Of particular interest, for highly bound drugs (>95% plasma protein bound), the addition of a plasma protein binding term increased the accuracy of the prediction of in vivo clearance. Conclusions The substrate loss method may be used to accurately predict hepatic clearance of drugs.     

头孢克肟分散片在健康志愿者体内生物等效性

目的 :比较头孢克肟分散片与头孢克肟颗粒剂在健康志愿者体内的药动学及生物等效性。方法 :采用开放、随机、自身交叉对照试验法 ,用RP HPLC法测定血清中头孢克肟的浓度 ,并拟合药动学参数 ,以双单侧t检验法比较两种头孢克肟制剂AUC、Cmax和Tmax间的差异。结果 :头孢克肟分散片与头孢克肟颗粒剂在健康志愿者体内的药时曲线均符合一级吸收的单室开放模型 ,主要药动学参数 :Tmax分别为 ( 3 .5± 0 .5 )和 ( 3 .2± 0 .4)h ,Cmax(实测值 )为 ( 3 .0± 0 .5 )和 ( 3 .1± 0 .4)mg·L- 1 ,AUC( 0～ 2 4h) 分别为 ( 2 6.8± 4.7)和 ( 2 8.5± 3 .1)mg·h·L- 1 。两种头孢克肟制剂各主要药动学参数间均无统计学差异 (P >0 .0 5 ) ,头孢克肟分散片对头孢克肟颗粒剂的相对生物利用度为 94.0 %。结论 :两种头孢克肟制剂具有相同的生物效应。     

Evaluation of Bioequivalence of Highly Variable Drugs Using Clinical Trial Simulations. II: Comparison of Single and Multiple-Dose Trials Using AUC and Cmax

Purpose. Evaluating of the effects of high intrasubject variability in clearance (CL) and volume of distribution (V), on 90% confidence intervals (CIs) for AUC (Area Under the concentration Curve) in single and multiple-dose bioequivalence studies. The main methodology was Monte Carlo simulation, and we also used deterministic simulation, and examination of clinical trials. The results are compared with those previously observed for Cmax (maximum concentration.) Methods. The time course of drug concentration in plasma was simulated using a one-compartment model with log-normal statistical distributions of intersubject and intrasubject variabilities in the pharmacokinetic parameters. Both immediate-release and prolonged-release products were simulated using several levels of intrasubject variability in single-dose and multiple-dose studies. Simulations of 2000 clinical bioequivalence trials per condition (138 conditions) with 30 subjects in each crossover trial were carried out. Simulated data were compared with data from actual bioequivalence trials. Results. The current simulations for AUC show similar probabilities of failure for single-dose and multiple-dose bioequivalence studies, even with differences in the rate of absorption or fraction absorbed. AUC values from prolonged-release scenario studies are more sensitive to changes in the first order absorption rate constant ka, and to variability in CL and V than AUC from studies of immediate-release studies. Conclusions. We showed that multiple-dose designs for highly variable drugs do not always reduce intrasubject variability in either AUC or Cmax, although the behavior of AUC differs from Cmax. Single dose AUC to the last quantifiable concentration was more reliable than either single dose AUC extrapolated to infinity, or multiple dose AUC during a steady-state interval. Multiple-dose designs may not be the best solution for assessing bioequivalence of highly variable drugs.     

Evaluation of a limited sampling method used to determine the bioequivalence of highly variable drugs with long half-lives.

The usefulness of a limited sampling method (LSM) to determine the bioequivalence of highly variable drugs with long half-lives was investigated. The LSM uses multiple linear regression of observed drug plasma concentrations versus area under the curve (AUC) or C(max) (peak plasma concentration) to obtain a best set of coefficients, concentration times and intercept based upon the regression coefficient, R(2), to predict the selected pharmacokinetic parameter (i.e. AUC or C(max)). The LSM, used successfully in clinical settings, has also been suggested for data analysis of in vivo bioequivalence studies. Because the method has not yet been thoroughly tested under many conditions likely to be encountered in bioequivalence studies, a further investigation of the method's applicability to bioequivalence determination was undertaken. In the present study, training and test data sets incorporating various levels of intrasubject variability in clearance (CL) with different ratios for fraction absorbed (Fa) of test and reference drug formulations were used to further evaluate the applicability of the limited sampling method (LSM) to the evaluation of bioequivalence for drugs with long half-lives of elimination. Both simulated (a one-compartment pharmacokinetic (PK) model with first-order elimination) and experimental data were used in the study. The results indicated that the determination of bioequivalence using the LSM was significantly influenced by the ratio of Fa(test)/Fa(reference) and by the level of intrasubject error (variability) in CL. Therefore, use of the LSM to determine bioequivalence of drugs with long half-lives and highly variable in CL seems suitable only for formulations that have point estimates of Fa(test)/Fa(reference) within the range of 0.9-1.10. Because the Fa ratio range would have to be verified through use of observed data obtained from a pilot study, the practical utility of the LSM in the determination of bioequivalence would be severely limited.     

Membrane Transport in Hepatic Clearance of Drugs I: Extended Hepatic Clearance Models Incorporating Concentration-Dependent Transport and Elimination Processes

Purpose. The objective of the present study was to develop hepatic clearance models which incorporate a unidirectional carrier-mediated uptake and bidirectional diffusional transport processes for drug transport in the sinusoidal membrane of hepatocytes as well as nonlinear intrinsic elimination. Methods. Two models were derived which view the liver as two separate compartments, i.e., sinusoid and hepatocyte. Model I assumes the instantaneous complete mixing of drugs within each compartment (similar to that of the 'well-stirred' model), while model II assumes that the drug concentrations in both compartments decrease progressively in the direction of the hepatic blood flow path (similar to that of the 'parallel-tube' model). Computer simulations were performed using a range of steady-state infusion rates for a substrate, while varying theV _max (capacity) and K _m (Michaelis-Menten constant) for the carrier-mediated uptake process, the diffusional clearance, the V _max and K _m for the intrinsic elimination process, blood flow and protein binding. Results. Simulations in which V _max and K _m for the sinusoidal membrane transporter and the diffusional clearance were varied, demonstrated that these membrane transport processes could affect the clearance of drugs to a significant extent in both models. The estimates for clearance of substrates with the same pharmacokinetic parameters are always lower in model I than in model II, although the quantitative differences in parameter estimates between models varied, depending on the steady state infusion rates. Conclusions. These more general hepatic clearance models will be most useful for describing the hepatic clearance of hydrophilic compounds, such as organic anions or cations, which exhibit facilitated uptake and limited membrane diffusion in hepatocytes.     

Differentiation of Gut and Hepatic First-Pass Effect of Drugs: 1. Studies of Verapamil in Ported Dogs

Purpose. To investigate the relative contributions of the gut and liver to the first-pass loss of verapamil (VL) using anin vivo intestinal-vascular access port (IVAP) dog model. Methods. Basic pharmacokinetics of VL were determined after intravenous (IV: 0.5 mg/kg), portal venous (PV: 2 mg/kg), and duodenal (ID: 2 mg/kg) administration in IVAP dogs. Serial blood samples were collected for 8 h after dosing, and plasma was analyzed for unchanged drug by a high-performance liquid chromatography-fluorescence method. Extraction ratios in the liver and intestinal tract were determined from the area under the concentration-time curves for ID, PV, and IV administration. The functional role of CYP450 or secretory transporters such as P-gp on the gut and liver first-pass loss of VL was further studied using ritonavir, a known substrate or inhibitor of these processes. Results. The liver had a high intrinsic capacity for clearing VL because the absolute bioavailability (BA) of VL was 21.7% after PV administration. The BA of VL after ID administration was 23.5%; therefore, intestinal absorption was complete and intestinal extraction was negligible (ER_GI 0). The BA of VL increased from 23.5% to 66.2% in the presence of ritonavir primarily due to a reduction in hepatic extraction. Conclusions. Although the liver had a high intrinsic capacity for extracting VL, the contribution of gut to the first-pass loss of VL was negligible. Because of the additive effects of intestinal CYP3A-mediated metabolism and secretory transport, a significant gut first-pass effect was expected, but not observed in dogs. These studies demonstrate the utility of the in vivo IVAP dog model for evaluating the relative contribution of the gut and liver to the first-pass loss of drugs and for characterizing the functional role that CYP450 metabolism and/or secretory transporters play in drug-drug interactions and reduced oral bioavailability.     

Use of Parent Drug and Metabolite Data in Bioavailability Assessment of a Novel Diltiazem HC1 Once-Daily Product

Pharmaceutical Research -     

Use of Truncated Areas to Measure Extent of Drug Absorption in Bioequivalence Studies: Effects of Drug Absorption Rate and Elimination Rate Variability on this Metric

Purpose. To compare the applicability and accuracy of truncated area (AUCt; where t represents truncated time) versus area to the last quantifiable time point [AUC(O-T)] for assessing bioequivalence. Drugs with either very low or very high intra-subject variability in clearance (CL) were selected for study. Clearance variability was defined by the number of subjects with a quantifiable plasma value (Cp) at each collection time from 24 hrs to last collection time (T). Methods. Data for amiodarone and danazol, drugs with different distributions of subject CL were examined. For amiodarone, the number of subject samples observed (test + reference) at the time of the last quantifiable concentrations was 60 at 240 hrs(T), 16 at 144 hrs and 4 at 96 hrs; while danazol had 4 at 96 hr(T), 3 at 72 hrs, 16 at 60 hrs, 7 at 48 hrs, 14 at 36 hrs, 11 at 24 hrs, 13 and 2 at 16 and 12 hrs, respectively. Simulations (Scenarios A and B) were performed to obtain populations (N = 24) with CL patterns similar to those of amiodarone and danazol. For scenario A (CL pattern similar to amiodarone), log-normally distributed CL values (28.8 L/HR) with intra-subject coefficient of variation (CV) of 25%, 40% and 60% gave the desired CL pattern. Scenario B (CL pattern similar to danazol) required that a subpopulation with an increase in CL of 40% from baseline (i.e., 40.32 L/HR) in 5%, 10% and 20% of the population represent the desired distribution. Power was evaluated by the percentage of times the simulated trials were declared bioequivalent (i.e., the number of times the test vs. reference 90% CI was within 80–125%), while accuracy was determined when the true difference in fraction absorbed (i.e., 1.25) was within the CI. Each simulation was repeated 300 times. Results. The simulation results for Scenario A indicated that the statistical results using truncated area (AUCt) had power and accuracy equivalent to that obtained using the AUC(O-T) metric. However, results for Scenario B indicated that AUCt had less power and accuracy than that obtained using AUC(O-T). The confidence interval (CI) for amiodarone was the same whether AUC (0-T) or AUCt was used as the metric for extent, while for danazol, the AUC(O-T) and AUCt differed in the lower limit by 7%. Conclusions. The truncated area, AUCt, has the greatest power and accuracy when the population clearance is such that most subjects have measurable plasma concentrations at last collection time(T), resulting in a proportional loss of data from each subject.     

罗布麻叶对大鼠尿液内源性物质代谢的影响

目的：采用代谢组学方法考察罗布麻叶对正常大鼠尿液内源性物质代谢的影响。方法：罗布麻叶用10倍量70%乙醇回流提取,将大鼠分为空白对照组、罗布麻叶高剂量组及低剂量3组,连续灌胃给药29 d,运用SIMCA-P软件中的PCA法与PLS-DA法对比分析。结果：罗布麻叶给药组与空白对照组得分点达到很好的区分效果,连续给药22 d时,尿液内源性物质代谢改变最为明显。相比于罗布麻叶低剂量组,高剂量组与空白对照组之间差距更大。结论：罗布麻叶提取物对正常大鼠机体内源性物质代谢产生了影响,为进一步阐释罗布麻叶的药性研究工作提供了依据。     

The Pharmacokinetics of Antipyrine and Three of Its Metabolites in the Rabbit: Intravenous Administration of Pure Metabolites

Antipyrine (AP) is a commonly used probe of oxidative metabolism. Indirect evidence demonstrates formation rate limited disposition of its metabolites. Kinetic studies using antipyrine and its major metabolites 3-hydroxymethylantipyrine (HMA), norantipyrine (NORA), and 4-hydroxyantipyrine (OHA) were completed to investigate the metabolic fate of preformed antipyrine metabolite and to demonstrate directly formation rate-limited metabolite disposition in vivo. Bolus injections of antipyrine and preformed metabolites (40-50 mg/kg) were administered to male, New Zealand white rabbits. Plasma and urine were analyzed using HPLC. These studies demonstrate that HMA, NORA, and OHA are formation rate limited in the rabbit. NORA appears to undergo further extensive oxidative and conjugative metabolism. Unknown additional peaks were detected in urine after NORA dosing but not after HMA or OHA administration. Mass spectroscopy of the unknown HPLC eluents identified potential structures of these NORA metabolites.