Computational Opioid Prescribing: A Novel Application of Clinical Pharmacokinetics

ABSTRACT

We implemented a pharmacokinetics-based mathematical modeling technique using algebra to assist prescribers with point-of-care opioid dosing. We call this technique computational opioid prescribing (COP). Because population pharmacokinetic parameter values are needed to estimate drug dosing regimen designs for individual patients using COP, and those values are not readily available to prescribers because they exist scattered in the vast pharmacology literature, we estimated the population pharmacokinetic parameter values for 12 commonly prescribed opioids from various sources using the bootstrap resampling technique. Our results show that opioid dosing regimen design, evaluation, and modification is feasible using COP. We conclude that COP is a new technique for the quantitative assessment of opioid dosing regimen design evaluation and adjustment, which may help prescribers to manage acute and chronic pain at the point-of-care. Potential benefits include opioid dose optimization and minimization of adverse opioid drug events, leading to potential improvement in patient treatment outcomes and safety.     

Time-scheduled vs. pain-contingent opioid dosing in chronic opioid therapy

Some expert guidelines recommend time-scheduled opioid dosing over pain-contingent dosing for patients receiving chronic opioid therapy (COT). The premise is that time-scheduled dosing results in more stable opioid blood levels and better pain relief, fewer adverse effects, less reinforcement of pain behaviors, and lower addiction risk. We report results of a survey of 1781 patients receiving COT for chronic noncancer pain, in which 967 reported time-scheduled opioid dosing only and 325 reported pain-contingent opioid dosing only. Opioid-related problems and concerns were assessed with the Prescribed Opioids Difficulties Scale. We hypothesized that respondents using time-scheduled opioid dosing would report significantly fewer problems and concerns than those using pain-contingent dosing. Patients receiving time-scheduled dosing received substantially higher average daily opioid doses than those using pain-contingent dosing (97.2 vs. 37.2 mg average daily dose morphine equivalents, P < .0001). Contrary to expectation, time-scheduled opioid dosing was associated with higher levels of patient opioid control concerns than pain-contingent dosing (6.2 vs. 4.8, P = .008), after adjusting for patient and drug regimen differences. Opioid-related psychosocial problems were somewhat greater among patients using time-scheduled dosing, but this difference was nonsignificant after controlling for patient and drug regimen differences (5.9 vs. 5.0, P = .14). Time-scheduled dosing typically involved higher dosage levels and was associated with higher levels of patient concerns about opioid use. Controlled comparative effectiveness research is needed to assess benefits and risks of time-scheduled opioid dosing relative to pain-contingent opioid dosing among COT patients in ambulatory care.     

Development of breakpoints of cephems for intraabdominal infections based on pharmacokinetics and pharmacodynamics in the peritoneal fluid of patients

We developed breakpoints for cephem antibacterial agents for intraabdominal infections based on pharmacokinetics (PK) and pharmacodynamics (PD) at the target site. Cefepime (CFPM), cefotiam (CTM), cefozopran (CZOP), and flomoxef (FMOX) were each administered to 8–10 patients before abdominal surgery, and venous blood and peritoneal fluid (PF) samples were obtained. The drug concentrations in plasma and PF were determined and analyzed using population pharmacokinetic modeling. Using the pharmacokinetic model parameters, a Monte Carlo simulation was conducted to estimate the probabilities of attaining the bacteriostatic and bactericidal targets (40% and 70% of the time above the minimum inhibitory concentration (T > MIC), respectively) in PF. The bacteriostatic and bactericidal breakpoints were determined as the highest MIC values at which the bacteriostatic and bactericidal probabilities in PF were ≥80%, which values varied with drug and dosing regimen. Site-specific PK-PD-based breakpoints for CFPM, CTM, CZOP, and FMOX are proposed, and should help us to select appropriate cephems and design their dosing regimens for intraabdominal infections.     

The importance of software for the planning and adjusting of dosage. Application with aminoglycosides

We are presenting a Mac Intosh--Apple computer program (Excel*) for the aminoglycosides (AG) drug monitoring useful in the following situations: --initiation of an AG's dosing regimen depending on the desired peak and through concentration (AG's pharmacokinetic parameters have been calculated according to the AG's literature values). --an adequate dosing regimen after revision of the patient's parameters (calculated with the observed concentrations). The technique has been validated comparing desired (des. C) and measured concentrations (mes. C). --either at the beginning of the treatment, --or after revision of the individual parameters. In the first case, the correlation coefficient obtained between 14 des C and 14 mes. C is equal to 0.91 and is improved when individual parameters are calculated (0.94).     

Phenobarbital in intensive care unit pediatric population: predictive performances of population pharmacokinetic model

An external evaluation of phenobarbital population pharmacokinetic model described by Marsot et al. was performed in pediatric intensive care unit. Model evaluation is an important issue for dose adjustment. This external evaluation should allow confirming the proposed dosage adaptation and extending these recommendations to the entire intensive care pediatric population. External evaluation of phenobarbital published population pharmacokinetic model of Marsot et al. was realized in a new retrospective dataset of 35 patients hospitalized in a pediatric intensive care unit. The published population pharmacokinetic model was implemented in nonmem 7.3. Predictive performance was assessed by quantifying bias and inaccuracy of model prediction. Normalized prediction distribution errors (NPDE) and visual predictive check (VPC) were also evaluated. A total of 35 infants were studied with a mean age of 33.5 weeks (range: 12 days–16 years) and a mean weight of 12.6 kg (range: 2.7–70.0 kg). The model predicted the observed phenobarbital concentrations with a reasonable bias and inaccuracy. The median prediction error was 3.03% (95% CI: ?8.52 to 58.12%), and the median absolute prediction error was 26.20% (95% CI: 13.07–75.59%). No trends in NPDE and VPC were observed. The model previously proposed by Marsot et al. in neonates hospitalized in intensive care unit was externally validated for IV infusion administration. The model‐based dosing regimen was extended in all pediatric intensive care unit to optimize treatment. Due to inter‐ and intravariability in pharmacokinetic model, this dosing regimen should be combined with therapeutic drug monitoring.     

Optimal sampling schedule design for populations of patients

Generation of pharmacodynamic relationships in the clinical arena requires estimation of pharmacokinetic parameter values for individual patients. When the target population is severely ill, the ability to obtain traditional intensive blood sampling schedules is curtailed. Population modeling guided by optimal sampling theory has provided robust estimates of individual patient pharmacokinetic parameter values. Because of the wide range of parameter values seen in this circumstance, it is important to know how the range of parameter values in the population affects the timing of the optimal samples. We describe a new, simple technique to obtain optimal samples for a population of patients. This technique uses the nonparametric distribution associated with a nonparametric adaptive grid population pharmacokinetic analysis. We used the distribution from an analysis of 58 patients receiving levofloxacin for nosocomial pneumonia at a dose of 750 mg. The collection of parameter vectors and their associated probabilities were entered into a D-optimal design evaluation by using ADAPT II. The sampling times, weighted for their probabilities, were displayed in a frequency histogram (an expression of how system information varies with time for the population). Such an explicit expression of the time distribution of information allows rational sampling design that is robust not only for the population mean vector, as in traditional D-optimal design theory, but also for large portions of the total population. For levofloxacin, one reasonable six-sample design would be 1.5, 2, 2.25, 4, 4.75, and 24 h after starting a 90-min infusion. Such sampling designs allow informative population pharmacokinetic analysis with precise and unbiased estimates after the maximal a posteriori probability Bayesian step. This allows the highest probability of delineating a pharmacodynamic relationship.     

Efficacy and safety of scheduled dosing of opioid analgesics: a quality improvement study.

Scheduled dosing of opioids is believed to provide more effective analgesia when compared to as needed (PRN) administration of the drug; however, few studies have evaluated the value of this approach. Therefore, a quality improvement study was conducted to determine the efficacy and safety of scheduled dosing of opioid analgesics, using a 2-group parallel design. One medical unit in a large urban academic medical center employed scheduled dosing, whereas a comparable unit used PRN dosing. The primary outcome indicators included pain intensity ratings and opioid doses, along with adverse events. Scheduled dosing was associated with decreased pain intensity ratings. There were no statistically significant differences in the amount of opioid ordered, or the amount administered when comparing scheduled vs. PRN dosing. However, when the amount of opioid given was expressed as a percentage of the amount ordered, the difference between scheduled (70.8%) and PRN (38%) dosing was statistically significant (P = .0001). There was no difference in adverse events between the 2 groups. These findings suggest that scheduled dosing of opioids in an inpatient medical population provides improved analgesia with no increased risk of adverse events. PERSPECTIVE: Scheduled dosing of opioids in an inpatient medical population improves analgesia, theoretically by overcoming barriers to drug administration, as well as providing more stable plasma levels of the opioid.     

The application of cassette dosing for pharmacokinetic screening in small-molecule cancer drug discovery

Pharmacokinetic evaluation is an essential component of drug discovery and should be conducted early in the process so that those compounds with the best chance of success are prioritized and progressed. However, pharmacokinetic analysis has become a serious bottleneck during the 'hit-to-lead' and lead optimization phases due to the availability of new targets and the large numbers of compounds resulting from advances in synthesis and screening technologies. Cassette dosing, which involves the simultaneous administration of several compounds to a single animal followed by rapid sample analysis by liquid chromatography/tandem mass spectrometry, was developed to increase the throughput of in vivo pharmacokinetic screening. Although cassette dosing is advantageous in terms of resources and throughput, there are possible complications associated with this approach, such as the potential for compound interactions. Following an overview of the cassette dosing literature, this article focuses on the application of the technique in anticancer drug discovery. Specific examples are discussed, including the evaluation of cassette dosing to assess pharmacokinetic properties in the development of cyclin-dependent kinase and heat shock protein 90 inhibitors. Subject to critical analysis and validation in each case, the use of cassette dosing is recommended in appropriate chemical series to enhance the efficiency of drug discovery and reduce animal usage.     

Plasma and cerebrospinal fluid population pharmacokinetic modeling and simulation of meropenem after intravenous and intrathecal administration in postoperative neurosurgical patients

Combined intravenous and local intrathecal administration of meropenem in patients after craniotomy is widely used to treat intracranial infections. However, the optimal dosing regimen of meropenem has not been investigated, posing a risk to treatment efficacy. We aimed to identify significant factors associated with inter-individual variability in cerebrospinal fluid (CSF) pharmacokinetics of meropenem and to evaluate potential intravenous and intrathecal meropenem dosing regimens for the treatment of patients with intracranial infections. After the diagnosis of intracranial infection, 15 patients with an indwelling drain tube received intravenous and intrathecal administration of meropenem. Blood and cerebrospinal fluid (CSF) samples were obtained at the scheduled time to measure meropenem concentration. Plasma and CSF concentration-time data were fit simultaneously using a nonlinear mixed-effects modeling approach. A 3-compartmental model was selected to characterize the in vivo behavior of meropenem. Through population modeling, multiple covariates were tested about their impact on the meropenem pharmacokinetics. Considering CSF outflow via drain tube leading to a drug loss, the drug clearance in CSF (CL_CSF) was added to describe this drug loss. The covariate selection indicated that the drainage volume (mL/d) had a significant positive correlation with CL_CSF. Bootstrap and visual predictive check suggested a robust and reliable pharmacokinetic model was structured. The established final population model was useful to apply with simulation to identify meropenem dosing regimens for the treatment of patients with intracranial infections. With the goal of CSF concentrations exceeding the minimum inhibitory concentration during the therapy, we created a simple to use dosage regimen table to guide clinicians with drug dosing.     

Using disease progression models as a tool to detect drug effect

Generally, information required for approval of new drugs is dichotomous in that the drug is either efficacious and safe or not. Consequently, the purpose of most confirmatory clinical trials is to test the null hypothesis. The primary reasons for designing hypothesis testing trials are to provide the information required for approval using analyses techniques that are relatively straightforward and free of apparent assumptions. However, the information required for approval is very different from that used by prescribers for decision making. In the clinic, decisions must be made about dose adjustment for individual patients in the presence of additional therapies and co-morbidities. Choice of drug and dosing regimen is therefore a classical risk to benefit decision that is often poorly informed from the results of confirmatory trials. Therefore, providing answers to the more difficult question of how to use the drug in a clinical setting is essential.     

Population pharmacokinetics of ertapenem in juvenile and old rats

Ertapenem is a parenteral broad‐spectrum carbapenem active against Gram‐negative pathogens, which has been approved for treatment of different infectious situations in adults and children. Favourable pharmacokinetics and pharmacodynamics have been established in young adults. In the elderly, dosing regimen adaptations are not recommended. Nevertheless, pharmacokinetic studies in paediatrics have not been published yet. The aim of this study was to document whether age influenced ertapenem disposition by comparing its pharmacokinetics in three groups of rats. Rats were separated into three groups: very young rats 21‐day‐old, 10‐week‐old and 7‐month‐old rats. A population pharmacokinetic model was built and evaluated, using the NONMEM software. Pharmacokinetic parameters, interindividual variability and residual variability were estimated. The final model was evaluated by a bootstrap procedure and visual predictive check. The ertapenem concentration–time data were best described by a one‐compartment model with zero‐order input and first‐order elimination. Effect of very young and old ages was estimated on central volume and clearance. Model evaluation indicated that the model was robust and parameter estimates were accurate. Central volume was found to be dependent on age and increase with age. Although the dosing regimen was weight adjusted, clearance was found to depend not only on age but also on weight. This study clearly documents changes in ertapenem pharmacokinetics according to group of age. These results suggest that paediatric dosing regimen cannot be directly extrapolated from a pharmacokinetic model in young adults unless it took into account age‐induced modifications.     

A Bayesian adaptive phase 1 design to determine the optimal dose and schedule of an adoptive T-cell therapy in a mixed patient population

We present a novel Bayesian adaptive phase 1 design to determine the optimal dosing regimen for an adoptive T-cell therapy in a mixed patient population. Our design is motivated by a B-cell Non-Hodgkin Lymphoma trial evaluating multiple dosing regimens within multiple disease subtypes. A utility score is calculated from both safety and efficacy utility functions and used to guide dose-escalation decisions. We pool safety data across disease subtypes and use a single dose-toxicity model while sharing efficacy information between disease subtypes using a hierarchical dose-response model. In addition, an adaptive randomization approach is applied to dynamically assign patients to a regimen when more than one regimen is open for enrollment. We illustrate this study design through a simulated trial example, and we investigate the operating characteristics using simulation studies.     

A dosing regimen for immediate N-acetylcysteine treatment for acute paracetamol overdose

Context. Current treatment of paracetamol (acetaminophen) poisoning involves initiating a 3-phase N-acetylcysteine (NAC) infusion after comparing a plasma concentration, taken ≥4 h post-overdose, to a nomogram. This may result in dosing errors, a delay in treatment, or possibly more adverse effects – due to the use of a high dose rate for the first infusion when treatment is initiated. Objective. Our aim was to investigate a novel dosing regimen for the immediate administration of NAC on admission at a lower infusion rate. Methods. We used a published population pharmacokinetic model of NAC to simulate a scenario where a patient presents to the hospital 2 h post-overdose. The conventional regimen is commenced 6 h post-overdose when the 4-h plasma paracetamol concentration is available. We investigated an NAC infusion using a lower dosing rate initiated immediately on presentation. We determined a dosing rate that gave an area under the curve (AUC) of the concentration-time curve that was the same or greater than that from the conventional regimen on 90% of occasions. Results. Lower dosing rates of NAC initiated immediately resulted in a similar exposure to NAC. An infusion of 110 mg/kg over the first 5 h (22 mg/kg/h) followed by the last two phases of the conventional regimen, or 200 mg/kg over 9 h (22.6 mg/kg/h) followed by the last phase of the conventional regimen could be used. Conclusion. The novel dosing regimen allowed immediate treatment of a patient using a lower dosing rate. This greatly simplifies the current dosing regimen and may reduce NAC adverse effects while ensuring the same amount of NAC is delivered.     

Population pharmacokinetics of conventional and intermittent dosing of liposomal amphotericin B in adults: a first critical step for rational design of innovative regimens

There is increased interest in intermittent regimen of liposomal amphotericin B, which may facilitate use in ambulatory settings. Little is known, however, about the most appropriate dosage and schedule of administration. Plasma pharmacokinetic data were acquired from 30 patients receiving liposomal amphotericin B for empirical treatment of suspected invasive fungal infection. Two cohorts were studied. The first cohort received 3 mg of liposomal amphotericin B/kg of body weight/day; the second cohort received 10 mg of liposomal amphotericin B/kg at time zero, followed by 5 mg/kg at 48 and 120 h. The levels of liposomal amphotericin B were measured by high-pressure liquid chromatography (HPLC). The pharmacokinetics were estimated by using a population methodology. Monte Carlo simulations were performed. D-optimal design was used to identify maximally informative sampling times for both conventional and intermittent regimens for future studies. A three-compartment pharmacokinetic model best described the data. The pharmacokinetics for both conventional and intermittent dosing were linear. The estimates for the mean (standard deviation) for clearance and the volume of the central compartment were 1.60 (0.85) liter/h and 20.61 (15.27) liters, respectively. Monte Carlo simulations demonstrated considerable variability in drug exposure. Bayesian estimates for clearance and volume increased in a linear manner with weight, but only the former was statistically significant (P = 0.039). D-optimal design provided maximally informative sampling times for future pharmacokinetic studies. The pharmacokinetics of a conventional and an intermittently administered high-dose regimen liposomal amphotericin B are linear. Further pharmacokinetic-pharmacodynamic preclinical and clinical studies are required to identify safe and effective intermittent regimens.     

Population pharmacokinetics of voriconazole in adults

Voriconazole is a first-line agent for the treatment of invasive fungal infections. The pharmacology of voriconazole is characterized by extensive interindividual variability and nonlinear pharmacokinetics. The population pharmacokinetics of voriconazole in 64 adults is described. The patient population consisted of 21 healthy volunteers, who received a range of intravenous (i.v.) and oral voriconazole regimens, and 43 patients with proven or probable invasive aspergillosis, who received the currently licensed dosage. Voriconazole concentrations were measured using high-performance liquid chromatography (HPLC). The pharmacokinetic data were modeled using a nonparametric methodology and with a nonlinear pharmacokinetic structural model. The extent and consequences of pharmacokinetic variability were explored using Monte Carlo simulation. The relationship between drug exposure and clinical response was explored using logistic regression. Optimal sampling times were identified using D-optimal design. The fit of the nonlinear model was acceptable. Data from the healthy volunteers provided robust estimates for K(m) and the maximum rate of enzyme activity (V(max)). The Bayesian parameter estimates were more variable and statistically different in patients than in volunteers. There was a linear relationship between the trough concentration and area under the concentration-time curve (AUC(0-12)). There was no relationship between the AUC(0-12) and clinical response. The original parameter values were readily recapitulated using Monte Carlo simulation. Initial i.v. dosing resulted in higher AUC(0-12) and trough concentrations compared with oral dosing. Sample collection times of 1, 2, 3, 4, 8, and 12 h after an i.v. infusion are maximally informative times for future pharmacokinetic studies.     

Population Pharmacokinetics and Dosing Optimization of Vancomycin in Children with Malignant Hematological Disease

An increase in vancomycin dose has been proposed in adults with malignant hematological disease. As pediatric data are limited, our aim was to evaluate the population pharmacokinetics of vancomycin in order to define the appropriate dosing regimen in children with malignant hematological disease. Vancomycin concentrations were collected prospectively during therapeutic monitoring. Population pharmacokinetic analysis was performed using NONMEM software. Seventy children (age range, 0.3 to 17.7 years) were included. With the current recommended dosing regimen of 40 to 60 mg/kg/day, 53 children (76%) had subtherapeutic steady-state trough concentrations (C_ss/min of <10 mg/liter). A one-compartment model with first-order elimination was developed. Systematic covariate analysis identified that weight significantly influenced clearance (CL) and volume of distribution (V) with power functions of 0.677 for CL and 0.838 for V. Vancomycin CL also significantly increased with increases in creatinine clearance and seemed to be higher in children with malignant hematological disease than in the general pediatric population. The model was validated internally. Its predictive performance was further confirmed in an external validation by Bayesian estimation. A patient-tailored dosing regimen was developed based on the final pharmacokinetic model and showed that a higher proportion of patients reached the target C_ss/min than with the traditional mg/kg-basis dose (60% versus 49%) and that the risks associated with underdosing or overdosing were reduced. This is the first population pharmacokinetic study of vancomycin in children with malignant hematological disease. An optimized dosing regimen, taking into account patient weight, creatinine clearance, and susceptibility of the pathogens involved, could routinely be used to individualize vancomycin therapy in this vulnerable population.     

CYP2D6 Phenotype-Specific Codeine Population Pharmacokinetics

Codeine's metabolic fate in the body is complex, and detailed quantitative knowledge of it, and that of its metabolites is lacking among prescribers. We aimed to develop a codeine pharmacokinetic pathway model for codeine and its metabolites that incorporates the effects of genetic polymorphisms. We studied the phenotype-specific time courses of plasma codeine, codeine-6-glucoronide, morphine, morphine-3-glucoronide, and morphine-6-glucoronide. A codeine pharmacokinetic pathway model accurately fit the time courses of plasma codeine and its metabolites. We used this model to build a population pharmacokinetic codeine pathway model. The population model indicated that about 10% of a codeine dose was converted to morphine in poor-metabolizer phenotype subjects. The model also showed that about 40% of a codeine dose was converted to morphine in EM subjects, and about 51% was converted to morphine in ultrarapid-metabolizers. The population model further indicated that only about 4% of MO formed from codeine was converted to morphine-6-glucoronide in poor-metabolizer phenotype subjects. The model also showed that about 39% of the MO formed from codeine was converted to morphine-6-glucoronide in extensive-metabolizer phenotypes, and about 58% was converted in ultrarapid-metabolizers. We conclude, a population pharmacokinetic codeine pathway model can be useful because beyond helping to achieve a quantitative understanding the codeine and MO pathways, the model can be used for simulation to answer questions about codeine's pharmacogenetic-based disposition in the body. Our study suggests that pharmacogenetics for personalized dosing might be most effectively advanced by studying the interplay between pharmacogenetics, population pharmacokinetics, and clinical pharmacokinetics.     

Pharmacokinetic/pharmacodynamic evaluation of anti-infective agents

Pharmacokinetic/pharmacodynamic modeling has become an extremely important tool in evaluating and optimizing anti-infective therapy. By systematically linking the pharmacokinetic and pharmacodynamic properties of the anti-infective agent, it is possible to make educated decisions about the correct drug to be used, correct dosing regimen and to estimate the probability of success with the selected dose regimen. This article gives an overview of the current pharmacokinetic/pharmacodynamic approaches for anti-infective agents and discusses their use in optimizing drug therapy.     

Pharmacokinetic/ pharmacodynamic evaluation of anti-infective agents

Pharmacokinetic/pharmacodynamic modeling has become an extremely important tool in evaluating and optimizing anti-infective therapy. By systematically linking the pharmacokinetic and pharmacodynamic properties of the anti-infective agent, it is possible to make educated decisions about the correct drug to be used, correct dosing regimen and to estimate the probability of success with the selected dose regimen. This article gives an overview of the current pharmacokinetic/pharmacodynamic approaches for anti-infective agents and discusses their use in optimizing drug therapy.