PK in Late Phase Trials

Pharmacokinetic (PK) data (plasma drug concentrations) are routinely collected in early clinical studies. Several regulatory guidances describe the role of PK data in regulatory decisions.^{1, 2} From an industry perspective, delineating PK characteristics in early phase clinical trials aids in the selection of the promising molecules and helps to make a go/no-go decision.^{3, 4} However, the value of collecting the PK data, which typically entails sparse blood sampling (few samples collected per patient), in late phase clinical trials is not well appreciated or documented. In spite of regulatory guidances (population PK, exposure-response, dose-response) recommending collection of PK information in late phase (Phase II-III) clinical trials, whether to collect PK data in late phase clinical trials is still routinely debated. Industry scientists often are asked to justify the value of collecting PK information in late phase clinical trials. Practicality and costs are often presented as challenges that limit collection of PK.

There is no systematic documentation of the value of collecting PK data in late phase clinical trials either from a drug development or a regulatory perspective. Late phase clinical trials are mainly composed of early patient, dose-finding, pivotal, or registration trials. In the context of this article, both pivotal or registration trials, and those trials that have similar endpoints as the pivotal trials, such as Phase II dose-ranging trials, are considered late phase clinical trials.

The objective of this article is to systematically evaluate the role of PK collected in late phase trials in key regulatory decisions such as drug approval and labeling.

Materials and methods

Overall we utilized three approaches to evaluate the value of collecting PK in late phase clinical trials: internal experience, external survey, and selected case studies.

Internal experience. We documented the final regulatory action for each of the new drug applications (NDA) and biological license applications (BLA) submitted between 2002 and 2008 which required pharmacometrics review. Almost all submissions for a pediatric indication include sparse PK sampling in late phase trials, which critically drive the dosing decisions. Hence, pediatric submissions were excluded in the current survey. The survey included submissions for new molecular entities or new indication claims. Specifically, the following information was noted: availability of PK data with respect to the amount (number or percent of patients contributing to PK information) and type (rich/sparse/mix), role of PK in approval related decisions (pivotal/supportive), role of PK in labeling decisions (pivotal/supportive), and the role of PK in alleviating the need of additional trials (yes/no).

Population PK and exposure-response modeling were the predominant pharmacometric approaches used. Approval related decisions imply approval, approvable, not approvable, or complete response. Prior to August 11, 2008, approval related decisions by FDA's Center for Drug Evaluation and Research were categorized as "approval," "approvable," or "not approvable." Since then, instead of an "approvable" or "not approvable" letter, a "complete response" letter is issued to the applicant stating that the review period for the application is complete and it cannot be approved in its present form. The complete response letter describes specific deficiencies and when possible provides recommendations that the applicant might take to get the application ready for approval. The question on labeling was only applicable if the NDA or BLA was approved. The ranking choices included "pivotal," "supportive," "none," and "not applicable." If the availability of PK was critical in the regulatory decision, it was ranked "pivotal." That is, without the analysis involving PK information the regulatory decision would have been different. On the other hand, if the PK information was worthwhile in confirming the regulatory decision, it was ranked as "supportive." It should be noted that often supportive evidence is also required by the FDA to make decisions related to approval and labeling. Sparse PK information implies that few samples are collected from each patient in the trial, while rich PK information implies that complete concentration-time profiles were determined. The choice "mix" implies that some patients contributed towards rich, while some contributed towards sparse PK sampling.

External survey. An external survey was conducted in order to gain insights into the importance of PK data collection mainly from the pharmaceutical industry perspective. Clinical pharmacology scientists at pharmaceutical industries, academia, and contract research organizations (CROs) were asked to rank their choices on the following two questions as described in Table 1. The survey was posted on the Internet via two clinical pharmacology related online user forums (PharmPK and NMusers).

Table 1. Questions included in the external survey for reasons in favor of collecting PK (A) and against collecting PK (B).

Pharmaceutical scientists were asked to rank each of the five responses under (A) and four under (B) from least important (rank=1) to most important (rank=5). The results are tabulated as percent of responders selecting the highest rank to each of the choices for questions A and B.

Case studies. Case studies were selected that illustrated the role of collecting PK in late phase clinical trials. In general, the impact of PK collection is presented for the following areas: dose optimization via exposure-response (effectiveness and/or safety), approval related decisions, and the need for additional trials. Technical details of the analysis are intentionally not provided.

Results

Internal experience. There were 79 applications (66 NDAs and 13 BLAs) submitted between 2002 and 2008 that required pharmacometrics review. Of the 79 applications, 62 involved new molecular entities while 17 involved drugs that are already approved but were submitted to the FDA to seek a new indication. The impact of the PK information collected in late phase clinical trials on approval and labeling decisions is presented in Figure 1. Of the 79 submissions, 70 had PK data collected in the late phase clinical trials. A majority of the applications involved collection of sparse PK (61%), 35% had a combination of sparse and rich PK sampling, and approximately 4% involved rich PK sampling, in the late phase trials. Of the 39 reviews that had an impact on approval related decisions, 11 (28%) alleviated the need for additional trials. Of the 41 reviews in which PK was pivotal or supportive in labeling decisions, 13 (32%) contributed to statements in the dosage and administration section; 12 (29%) to the specific population; 4 (10%) to contraindications, warning, and precautions; and 34 (83%) to the clinical ­pharmacology section. Several reviews contributed to multiple labeling sections.

Figure 1. The impact of PK information collected in late phase clinical trials in approval and labeling decisions.

External survey. A total of 182 responses were received of which 137 (75%) were obtained from pharmaceutical industries while 20, 18, 4, and 3 were obtained from academia, CROs, regulatory agencies, and others, respectively. The results are summarized in Table 2. The most prominent reasons cited in favor of PK data collection were to conduct exposure-efficacy analysis as a source of evidence of effectiveness and to evaluate the effect of intrinsic and extrinsic factors on the drug's pharmacokinetics. On the other hand, the most important argument against collecting PK in late phase clinical trials was practicality or feasibility concerns.

Table 2. Summary of responses obtained via external survey highlighting the reasons in favor or against collecting PK data in late phase clinical trials.

Case studies. Case Study 1. The sponsor sought approval of etravirine, a non-nucleoside reverse transcriptase inhibitor (NNRTI) for the use in treatment experienced HIV-1 infected individuals in combination with other antiretroviral agents. Other approved NNRTIs include nevirapine, efavirenz, and delavirdine. Etravirine retained activity against many HIV isolates that contain the K103N mutation. Hence, etravirine addressed an unmet medical need in that it promised to provide a treatment option for treatment experienced patients with resistance to other NNRTIs in the setting of clinical failure.⁵ Registration trials conducted to support approval of etravirine enrolled patients who received HIV protease inhibitors, darunavir/ritonavir, as part of their optimized background regimen. A Phase I drug interaction study of etravirine with darunavir/ritonavir indicated that etravirine AUC (area under the plasma concentration vs. time curve), on average, decreased by 37%. However, if subjects take etravirine without darunavir/ritonavir or with other background antiretroviral regimens, it was expected that the exposures would be higher than that observed in the registration trials. For instance, there was data to believe that lopinavir/ritonavir in the background regimen could yield 60% higher exposures. In other words, etravirine exposures were expected to be 1.85 fold (85% higher) in the presence of lopinavir/ritonavir compared to (attenuated) exposures observed in registration trials.

Regulatory question: Is there adequate safety experience to approve etravirine with antiretroviral agents that might increase exposures of etravirine?

In this analysis, the lopinavir/ritonavir regimen was chosen as the prototype but the results can be applied to other background regimens that may increase etravirine exposures. The analysis was focused on the patients in the highest exposure quartile of the registration trials with two ­specific aims:⁶

• To derive the expected distribution of etravirine concentration across a population that might receive etravirine in combination with lopinavir/ritonavir, and compare it to exposures observed in the registration trials.

• To compare the adverse event profiles in the highest exposure quartile (AUC 6500-64000 ng*h/mL) to the patients in the lower three quartiles. Safety analyses consisted of comparison of common adverse events and evaluation of select laboratory abnormalities.

Pharmacokinetic data were available for 96% (576/599) of the etravirine treated patients. The expected distribution of etravirine concentration was achieved by multiplying the highest etravirine AUC value observed for each subject with pharmacokinetic data in Phase III by a factor of 1.85. Table 3 illustrates a comparison of observed and expected etravirine exposure by different AUC cut-offs. In the observed data, AUC greater than 30,000 was observed in <2% of patients when etravirine was administered with darunavir/ritonavir. However, ~6% of patients were expected to have AUC greater than 30,000 if etravirine is administered with lopinavir/ritonavir. On the other hand, almost 50% of subjects who would receive etravirine with lopinavir/ritonavir may have etravirine AUC between 10,000 to 30,000, while AUCs in this range were observed for 17% of subjects in Phase III. Important to note, this analyses assumed exposure increase in all subjects, however, only a certain fraction of the population will receive lopinavir/ritonavir. Therefore, numbers presented here are the worst case estimates and it is likely to observe exposures between 10,000 to 30,000 in 17-50% of the subjects to be treated with etravirine. In conclusion, the highest quartile (AUC range: 6,500 to 64,000 ng*h/mL) of etravirine exposure in the registration trials to some extent covers expected exposure if etravirine were to be co-administered with drugs that increase etravirine exposure.

Table 3. Comparison of expected etravirine exposure in a population that might receive etravirine (200 mg bid) with lopinavir/ritonavir and exposure observed in registration trials.

For the latter part of the analysis, adverse events (AE) and laboratory results were compared between 145 subjects in the highest quartile and 431 subjects in the lower three quartiles. Rash was the most frequent AE reported in the highest quartile group and a modest increase in rash (any type) in subjects in the highest quartile (21.4%) was noted as compared to subjects in lower quartiles (15.8%) and the placebo arm (13.5%). Additionally, there were differences in hypersensitivity (1.4% vs. 0%), hypertriglyceridemia (4.1% vs. 2.8%), peripheral neuropathy (5.5% vs. 2.6%), and anxiety (4.8% vs. 1.9%). Most of the AEs were resolved on treatment discontinuation and were not related to death or permanent discontinuations. In conclusion, there were no major safety concerns in subjects with high exposure. Thus, although limited, some safety experience was available at expected exposure if etravirine is co-administered with drugs that can increase etravirine exposure.

Etravirine was granted accelerated approval⁷ and was labeled for use in combination with lopinavir/ritonavir with caution for the treatment of HIV-1 infection in treatment-experienced adult patients.⁸

Given a continued public health need to have additional therapeutic options to treat HIV infections, approval of etravirine has allowed effective treatment regimens without limiting the antiretroviral combinations. The availability of PK data in combination with safety analysis was pivotal in providing assurance that etravirine could be approved for use with other antiretroviral agents that may increase etravirine exposure.

Case study 2. Oral paricalcitol (ZEMPLAR^®) capsule is approved for the prevention and treatment of secondary hyperparathyroidism associated with chronic kidney disease (CKD), stages 3 and 4. The sponsor conducted three clinical trials in stage 5 CKD patients to seek approval for oral paricalcitol. Paricalcitol dose was based on baseline intact parathyroid hormone (iPTH) level and the dose was selected to be iPTH/60 μg. The indication was not approved due to high-observed hypercalcemia rates (at that time defined as two consecutive elevations of Ca >11 mg/dL). The sponsor developed models to characterize the relationship between plasma paricalcitol and the measures of efficacy and safety such as iPTH, serum calcium (Ca) and serum phosphorous. The sponsor later conducted a new trial with lower dose (initial paricalcitol dose=iPTH/80 μg), which was projected using the exposure-response model developed for the three failed trials, to seek the same indication in patients on hemodialysis (HD) or continuous peritoneal dialysis (CPD). The proposed and tested paricalcitol dosing regimen (dose of iPTH/80 μg for HD and CPD patients with serum Ca <10.5 mg/dL) was effective in decreasing iPTH levels while the rate of hypercalcemia (now defined as two consecutive elevations of Ca>10.5 mg/dL) remained high and was driven by CPD patients. The agency issued an approvable letter and recommended the sponsor pursue a new iPTH based dosing regimen to obtain efficacy and safety information from a new clinical trial.

Regulatory question: Can an optimum dosing regimen be derived for CKD stage 5 patients on HD or CPD that balances efficacy and safety, alleviating the need of a new clinical trial?

An exposure-response model was developed for efficacy (two consecutive serum iPTH measurements decreased by 30% from baseline) and safety (two consecutive serum calcium measurements over 10.5 mg/dL; hypercalcemia) using pharmacokinetic and clinical response information from the three earlier conducted clinical trials in stage 5 CKD patients with iPTH/60 dosing regimen. The model was validated using an external dataset from the current clinical trial (lower dose: iPTH/80) and was able to predict the efficacy outcome (82.6% predicted versus 87.9% observed) and the hypercalcemia rates (2.0% predicted versus 1.6% observed) that were observed in the trial. However when the results were stratified by HD and CPD patient type, it was evident that hypercalcemia was higher in CPD patients compared to HD patients (0% in HD vs. 21.1% in CPD patients). Lowering of dose maintained acceptable efficacy and resulted in lowering of hypercalcemia. However, the predicted hypercalcemia rates in CPD patients still remained unacceptably high (31.4%). Simulations with switch–dosing schemes, starting with an initial low dose (iPTH/80 μg) for up to eight weeks followed by a higher dose (iPTH/60 μg) for up to 48 weeks did not provide either significant lowering of iPTH or hypercalcemia rates. Further exploration of the exposure-response model revealed that the higher baseline serum Ca levels as well as a higher potency for calcium synthesis drove the differences between CPD and HD. Hence, the lowering of paricalcitol dose alone did not result in lowered hypercalcemia rates in CPD population. Modifying the inclusion criteria to restrict the therapy to patients with lower screening calcium (Ca <9.5 mg/dL) was found to further decrease the hypercalcemia rates in both HD and CPD patient population without compromising efficacy (Table 4).

Table 4. Efficacy and safety assessments with different screening criteria.

Zemplar capsules were approved for use in stage 5 CKD patients on HD or CPD without additional clinical trials. The dosing regimen information derived via exposure-response modeling and simulation was included in the label.⁹

The pharmacokinetic information collected in the late phase clinical trials was vital in deriving the dosing regimen and alleviating the need of any additional clinical trials.

Case study 3. The sponsor sought approval for an orally administered prodrug, prasugrel, an inhibitor of platelet activation and aggregation for the reduction of atherothrombotic events and the reduction of stent thrombosis in acute coronary syndromes. Bleeding is the major risk component evaluated while considering the risk/benefit profile of these anticoagulants. The sponsor proposed a 60 mg QD loading dose followed by a 10 mg QD maintenance dose. A reduced maintenance dose of 5 mg QD for patients with body weight (BW) <60 kg or age ≥75 years was proposed to mitigate the increased risk of bleeding. The sponsor conducted one single Phase III trial to support the indication. Pharmacokinetic data was collected for the active metabolite of prasugrel from some late phase clinical trials (Phase II dose selection trials) but was not collected in the Phase III trial.

Regulatory question: Is the proposed dosing regimen for specific populations (BW <60 kg or age ≥75 years) adequate?

Exposure-response pooled-analysis conducted for bleeding using the data from six clinical pharmacology studies provided by the sponsor indicated a trend of increased bleeding with higher exposures of the active metabolite of prasugrel (Figure 2, left panel). Population pharmacokinetic analysis using the data from late phase clinical trials demonstrated weight as the major covariate affecting clearance such that patients with lower body weight had lower clearance and hence higher exposures. Age did not affect exposures after accounting for differences in weight. Specifically, patients with BW <60 kg were at increased risk of bleeding based on the AUC cutoff established in the exposure-bleeding relationship (Figure 2, right panel). Data from the Phase III trial showed that the risk for Thrombolysis in Myocardial Infarction major bleeding with prasugrel was higher in lower body weight (BW <60 kg) group (Hazard Ratio: 3.05 (2.01–4.62), p <0.0001) compared to patients with higher body weight probably due to higher exposures. It was also seen that efficacy was similar across different weight groups.

Figure 2. AUC-bleeding relationship (left) and AUC-body weight relationship (right) with the AUC cutoff of 88 ng*h/mL.

Pharmacokinetic simulations were conducted to show that the dose adjustment to 5 mg QD in patients weighing <60 kg will shift the exposures in the majority of these patients from the upper quartile to lower quartile of those seen in patients with body weight >60 kg (Figure 3). Thus, reduction of maintenance dose from 10 mg QD to 5 mg QD in patients with body weight <60 kg is acceptable.¹⁰

Figure 3. Simulation (N=2000) of the proposed dose of 5 mg in patients with body weight 60 kg.

Age ≥75 years was an independent predictor for increased risk of primary composite efficacy endpoint (cardio vascular disease/non-fatal myocardial infarction/non-fatal stroke) and major bleeding. In patients age >75 years, the efficacy of prasugrel was numerically better than clopidogrel with a similar risk for bleeding. Furthermore, after adjusting for body weight, the exposure of active metabolite of prasugrel did not increase with age. Hence, dose reduction in elderly patients due to age alone is not justified.

The dose recommendations requiring lower maintenance dose in patients less than 60 kg were included in the label. No dose reduction in elderly patients (≥75 years) was recommended.¹¹

Lighter patients were at higher risk of bleeding. The availability of PK information in the late phase clinical trials helped establish the body weight-exposure and exposure-bleeding relationship suggesting that lowering the dose in lower body weight patients (BW <60 kg) will reduce exposures thus reducing the chances of bleeding. At the same time, there will be no loss in efficacy as exposures in these lighter patients (<60 kg) receiving 5 mg would be similar to higher body weight patients (>60 kg) receiving the 10 mg dose. In summary, the PK and safety information in the late phase clinical trials along with efficacy information from the Phase III trial provided evidence to recommend an optimum dosing regimen in a patient population at higher risk of bleeding.

Case Study 4. The sponsor sought the marketing approval of lacosamide for the indication of the treatment of epilepsy as adjunctive therapy in patients with partial onset seizures aged 16 years and older. To support the efficacy claim, the sponsor conducted three pivotal trials to evaluate dose ranging from 200 mg to 600 mg. Due to large variability in PK, the dose response relationship was not apparent. Therefore, exposure-response analysis was performed.

Regulatory question: What is the optimum dose of lacosamide in treating patients with partial seizure?

The exposure-response relationship for lacosamide in treating patients with partial seizure was established in the subset of patients in three pivotal clinical trials where PK samples were available and baseline seizure was fully characterized. Our analyses focused on observations at two critical time points (i.e., by the end of titration phase and by the end of maintenance phase). The exposure was defined as AUC over a dosing interval of 12 hours at a steady state. The response was defined as a change from baseline of the average daily number of partial seizure.

An E_max model described the exposure-effectiveness relationship, as shown in Figure 4. The response curve started to flatten out beyond the median exposure of the 400 mg dose. Therefore, the 600 mg dose does not appear to provide additional benefit compared to the 400 mg dose.¹²

Figure 4. Exposure-response relationship for lacosamide in treating patients with partial seizure by the end of titration phase (A) and by the end of maintenance phase (B).

Based on the exposure-response analysis, the dosage and administration section of lacosamide package insert was updated. The following language was included:¹³

VIMPAT can be increased at weekly intervals by 100 mg/day given as two divided doses up to the recommended maintenance dose of 200 to 400 mg/day, based on individual patient response and tolerability. In clinical trials, the 600 mg daily dose was not more effective than the 400 mg daily dose, and was associated with a substantially higher rate of adverse reactions.

This conclusion could not have been arrived at based on the effect at each dose alone, owing to the overlapping PK across the doses. The dose-response curve appeared flatter than the exposure-response curve.

Discussion

The internal survey of NDA/BLAs submitted between 2002 and 2008 highlights the value of collecting PK in late phase clinical trials in approval and labeling related decisions. In most cases, sparse sampling with or without rich PK sampling in some patients was conducted. Adequate design and informative sampling for PK data can serve as a useful surrogate for an individual's exposure. Sparse PK samples were used mainly for population analysis to identify important covariates in the late 1990s. The current survey indicates that FDA and industry are employing PK in late phase clinical trials to support evidence of ­effectiveness, derive dosing regimen not directly studied in trials, and identify at-risk patients. About 56% (39/70) of the submissions with PK data in late phase clinical trials contributed towards approval related decisions. Most importantly, among these 39 submissions, exposure-response modeling alleviated the need for additional trials for 11, leading to reduced drug development cost and time. Our internal experience suggests that almost half of the submissions utilized the PK data to conduct exposure-efficacy analysis. This is a significant return on investment given the amount of time, money, and resources spent by a pharmaceutical company required to bring a drug candidate to market compared to the cost of collecting PK.

The etravirine case study represents a scenario where the PK data was useful in assessing the risk profile in a specific target population and allowed for the approval of etravirine with other anti-retroviral agents without any additional trials. In the case of paricalcitol, the dosing regimen was modified based on the exposure-response relationship and thus alleviating the need for additional trials. Oxcarbazepine was approved for treating partial seizures, as a single agent, in pediatrics four years and older based on pharmacometric bridging of exposure-seizure reduction relationships in adults and pediatrics receiving combination treatment.¹⁴ If the sponsor had not collected PK data in the combination use trials, such an approval would not have been possible. Other anti-epileptics are following a similar path. The primary basis for the approval of argatroban dosing in pediatrics was derived based on the exposure-anti-coagulation relationship.¹⁵ Again, without the PK data a similar regulatory decision might have not been reached. Further, PK data identified patients with compromised hepatic function as at-risk population requiring a different start dosing. The collection of PK data, for both cyclosporine and everolimus, in the first everolimus trial in heart-transplant patients allowed derivation of a dosing regimen with reduced risk of nephrotoxicity. The sponsor tested that simulation-derived regimen in renal transplant patients and found the risk of nephrotoxicity was as predicted. Everolimus is approved for prophylactic treatment of renal transplant patients. Transplant trials are challenging to conduct, and without the PK data, exploration of alternative dosing schemes for the next trial would not have been possible.¹⁶

Labeling decisions were affected for 87% (41/47) of the submissions. For prasugrel, PK information was vital in figuring out that increased bleeding in lower weight (<60 kg) patients was due to an increase in exposure. Hence, a lower dose was approved for these patients. More importantly, the lower dose would not result in loss of efficacy since PK would be within the effective exposure-range. Sponsor use population analyses based on sparse PK information in late phase clinical trials that provides most of the PK characterization for biologics. Eli Lilly reported that gemcitibine's PK evaluation was based on sparse PK sampling in late phase clinical trials.¹⁷ Pfizer noted that exposure-response analysis using the PK data from Phase III trials allowed alleviating the need for additional efficacy trials for gabapentin. Its labeling states that pharmacokinetic/pharmacodynamic modeling provided confirmatory evidence. Hoffman La Roche evaluated the integration of pharmacokinetic and pharmacodynamic principles in clinical development by looking at 18 projects in their development portfolio. Pharmacokinetic and pharmacodynamic principles were applied in every project independent of development phase and therapeutic area with selection of dosage for the clinical studies being their most important application. Furthermore, use of these principles resulted in significant time savings up to several months in many projects.¹⁸

For nine NDAs, PK data were not collected in late phase trials. For several of these, PK data might not have added much value. For example, WelChol (colesevelam hydrochloride) is a resin for local GI action that is not systemically absorbed. Another NDA was for an investigational drug with very short half-life. PK data were not collected in the registration trials for tetrabenazine. As tetrabenazine's major metabolites (a-dihydrotetrabenazine and - dihydrotetrabenazine) are further metabolized via CYP2D6 and poor metabolizers have higher exposures (3- and 9-fold, respectively), it might have been valuable to collect PK data. However, one of the trials involved dose titration using doses of 12.5 mg to 100 mg in each patient providing a rich individual dose-response. This dose-response provided the confirmatory evidence for effectiveness and alleviated the need for additional efficacy trials.¹⁹ For six NDAs, PK information was collected but did not directly affect labeling decisions. One of these NDAs was for a post-marketing safety study for ranolazine. It is challenging to speculate the impact if PK data were collected.

The results of the external survey suggest that one of the main drivers to collect PK data in late phase clinical trials is to provide evidence of effectiveness. Practicality or feasibility was the top reason given for not collecting PK data in late phase trials. However, based on our internal survey, 70 submissions that include well over 120 trials across various therapeutic areas had PK data collected. In fact, some of the trials involved collecting sparse PK data from thousands of subjects (e.g., Prasugrel). It's likely that the sponsors of these trials realized the value of collecting PK and/or implemented efficient clinical trial practices that allow them to collect PK data.

Spending time, money, and resources to collect PK information in late phase clinical trials is a significant return on investment for the pharmaceutical industry. It provides valuable information that can help in high-level approval related regulatory decisions and can also be useful in ­providing valuable labeling instructions, which otherwise would not be possible. We present results from two surveys (internal and external) that support the argument of collecting PK data in late phase clinical trials. We hope that the regulatory perspective presented here provides evidence to support and encourage collection of PK data in late phase clinical trials. This can reduce drug development cost and time, helping safe and effective drugs reach the target population in a timely manner and also provide the most appropriate labeling instructions for practicing physicians.

Nitin Mehrotra* e-mail: nitin.mehrotra@fda.hhs.gov, Venkatesh A. Bhattaram, Justin C. Earp, Jeffry Florian, Kevin Krudys, Joo-Yeon Lee, Fang Li, Jiang Liu, Anshu Marathe, and Hao Zhu are all Reviewers, Division of Pharmacometrics, Office of Clinical Pharmacology, at the US Food and Drug Administration. Yaning Wang is Associate Director of Science, Division of Pharmacometrics, Christine Garnett is Team Leader, Division of Pharmacometrics, Pravin R. Jadhav is Team Leader, Division of Pharmacometrics, and Rajnikanth Madabushi is Team leader, Division of Clinical Pharmacology 1 all at the US Food and Drug Administration. Christoffer W. Tornoe is Director, Quantitative Clinical Pharmacology at Novo Nordisk A/S. Jogarao V. Gobburu is Professor, Schools of Pharmacy, Medicine at the University of Maryland, Baltimore.

*To whom all correspodence should be addressed.

References

1. Food and Drug Administration, Guidance for Industry: Guidance for Industry Population Pharmacokinetics (FDA, Rockville, MD, 1999).

2. Food and Drug Administration, Guidance for Industry, Exposure-Response Relationships—Study Design, Data Analysis, and Regulatory Applications (FDA, Rockville, MD, 2003).

3. J.H. Lin and A.Y.H. Lu, "Role of Pharmacokinetics and Metabolism in Drug Discovery and Development," Pharmacological Reviews, 49 (4) 403-449 (1997).

4. I. Kola and J. Landis, "Can the Pharmaceutical Industry Reduce Attrition Rates," Nature Reviews, Drug Discovery, 3, 711-715 (2004).

5. Center for Drug Evaluation and Research, US Food and Drug Administration, "Etravirine Application, Office Director Memo," http://www.accessdata.fda.gov/drugsatfda_docs/nda/2008/022187s000_ODMemo.pdf.

6. Center for Drug Evaluation and Research, US Food and Drug Administration, "Etravirine, Clinical Pharmacology and Biopharmaceutics Review(s)," http://www.accessdata.fda.gov/drugsatfda_docs/nda/2008/022187s000_ClinPharmR_P1.pdf

7. Food and Drug Administration, Guidance for Industry: Fast Track Drug Development Programs—Designation, Development, and Application Review, (FDA, Rockville, MD, 2006).

8. Center for Drug Evaluation and Research, US Food and Drug Administration, Etravirine label, http://www.accessdata.fda.gov/drugsatfda_docs/label/2008/022187lbl.pdf.

9. Center for Drug Evaluation and Research, US Food and Drug Administration, Paricalcitol label, http://www.accessdata.fda.gov/drugsatfda_docs/label/2010/021606s007lbl.pdf.

10. Center for Drug Evaluation and Research, US Food and Drug Administration, Prasugrel Clinical Pharmacology and Biopharmaceutics Review(s), http://www.accessdata.fda.gov/drugsatfda_docs/nda/2009/022307s000_ClinPharmR_p54.pdf.

11. Center for Drug Evaluation and Research, US Food and Drug Administration, Prasugrel label, http://www.accessdata.fda.gov/drugsatfda_docs/label/2010/022307s002lbl.pdf.

12. Lacosamide, Center for Drug Evaluation and Research, US Food and Drug Administration, Lacosamide, Clinical Pharmacology and Biopharmaceutics Review(s), http://www.accessdata.fda.gov/drugsatfda_docsnda/2008/022253s000_022254s000_ClinPharm_P5.pdf.

13. Center for Drug Evaluation and Research, US Food and Drug Administration, Lacosamide label, http://www.accessdata.fda.gov/drugsatfda_docs/label/2009/022253s004,022254s001lbl.pdf.

14. V. A. Bhattaram, B. P. Booth, R. P. Ramchandani, B. N. ­Beasley, Y. Wang, V. Tandon, J. Z. Duan, R. K. Baweja, P. J. Marroum, R. S. Uppoor, N. A. Rahman, C. G. Sahajwalla, J. R. Powell, M. U. Mehta, and J. V. S. Gobburu, "Impact of Pharmacometrics on Drug Approval and Labeling Decisions: A Survey of 42 New Drug Applications," The AAPS Journal, 7 (3) E503-512 (2005).

15. R. Madabushi, D. S. Cox, M. Hossain, D. A. Boyle, B. R. Patel, G. Young, Y. M. Choi and J. V. S. Gobburu, "Pharmacokinetic and Pharmacodynamic Basis for Effective Argatroban Dosing in Pediatrics," Journal of Clinical Pharmacology, 51, 19-28 (2011).

16. Cardiovascular and Renal Drugs Advisory Committee, Center for Drug Evaluation and Research, US Food and Drug Administration, Everolimus briefing information, http://www.fda.gov/ohrms/dockets/ac/05/briefing/2005-4183-index%20with%20disclaimer.htm.

17. L. Zhang, V. Sinha, S. T. Forgue, S. Callies, L. Ni, R. Peck, and S. R. B. Allerheiligen, "Model-Based Drug Development: The Road to Quantitative Pharmacology," Journal of Pharmacokinetics and Pharmacodynamics, 33 (3) 369-393 (2006).

18. B. G. Reigner, P. E. Williams, I. H. Patel, J. L. Steimer, C. Peck and B. P. Van, "An Evaluation of the Integration of Pharmacokinetic and Pharmacodynamic Principles in Clinical Drug Development: Experience Within Hoffmann La Roche," Clinical Pharmacokinetics, 33, 142-152 (1997).

19. Center for Drug Evaluation and Research, US Food and Drug Administration, Tetrabenazine, Clinical Pharmacology and Biopharmaceutics Review(s), http://www.accessdata.fda.gov/drugsatfda_docs/nda/2008/021894s000_ClinPharmR_P1.pdf.