Share this article on:

Cerebrospinal Fluid and Plasma Pharmacokinetics of the Cyclooxygenase 2 Inhibitor Rofecoxib in Humans: Single and Multiple Oral Drug Administration

Section Editor(s): Bovill, James G.Buvanendran, Asokumar MD*; Kroin, Jeffrey S. PhD*; Tuman, Kenneth J. MD*; Lubenow, Timothy R. MD*; Elmofty, Dalia MD*; Luk, Pauline BS

doi: 10.1213/01.ANE.0000150597.94682.85
Anesthetic Pharmacology: Research Report

Cerebrospinal fluid (CSF) pharmacokinetics of orally administered cyclooxygenase 2 inhibitors, with single or multiple dosing, is of clinical relevance because it may relate to the analgesic efficacy of these drugs. We enrolled 9 subjects with implanted intrathecal catheters in the study. After 50-mg oral rofecoxib administration, the CSF drug concentration lagged slightly behind the plasma drug concentration. The ratio of the 24-h area under the drug-concentration curve (AUC) in CSF to plasma was 0.142. After daily dosing of rofecoxib 50 mg/d for 9 days, rofecoxib concentrations in plasma and CSF were larger on Day 9 than on Day 1, with the 24-h AUC on Day 9 more than twice the Day 1 AUC for both plasma and CSF. After nine consecutive daily doses of rofecoxib, the AUCCSF/AUCplasma ratio was 0.159. The important findings of this study are that CSF rofecoxib levels are approximately 15% of plasma levels and that repeated daily dosing more than doubles the AUC in CSF.

IMPLICATIONS: When the selective cyclooxygenase 2 inhibitor rofecoxib is given orally at 50 mg, the cerebrospinal fluid concentration of the drug is 15% of the plasma concentration. Daily dosing to reach steady-state more than doubles the amount of rofecoxib in the cerebrospinal fluid.

*Department of Anesthesiology, Rush Medical College at Rush University Medical Center, Chicago, Illinois; and †Merck Frosst Canada, Kirkland, Quebec, Canada

Funded by University Anesthesiologists S.C. and a Medical School Grant (MSGP#16380) from Merck & Co. to the Department of Anesthesiology, Rush Medical College, Chicago, IL.

Accepted for publication October 29, 2004.

Address correspondence and reprint requests to Asokumar Buvanendran, MD, Department of Anesthesiology, 1653 W. Congress Pkwy., Rush University Medical Center, Chicago, IL 60612. Address e-mail to

Animal studies demonstrate that surgical trauma induces cyclooxygenase 2 (COX-2) upregulation that locally sensitizes peripheral nociceptors and mediates central sensitization (1). Nonsteroidal antiinflammatory drugs (NSAIDs) decrease the inflammatory response associated with surgery (2), but their use is often limited in surgical patients because of concerns about increased bleeding (3), which is generally not observed with selective COX-2 inhibitors (4). Preoperative administration of rofecoxib has been shown to reduce postoperative pain (5,6), and we have demonstrated (7) that perioperative administration of a COX-2 inhibitor leads to improved postoperative outcome after knee arthroplasty.

The analgesic efficacy of oral COX-2 inhibitors may be related to blood-brain barrier penetration (8). Animal models of pain have demonstrated that COX-2 is induced in the spinal cord after peripheral inflammation (9) and that intrathecal COX-2 inhibitors reduce hypersensitivity (10,11). Previous studies from our laboratory (12) have shown upregulation of spinal COX-2 protein after surgical incision and enhancement of antinociception by an intrathecally administered COX-2 inhibitor in a surgically induced animal pain model (13).

Human data are consistent with findings in animals. Dental surgery studies (14,15) using microdialysis catheters at the molar extraction site showed that the administration of IM ketorolac (30 mg) after molar extraction decreased pain scores and the peripheral prostaglandin E2 (PGE2) response, whereas 1 mg of ketorolac IM produced only analgesia, with no change in peripheral PGE2, thus suggesting NSAID action at a nonperipheral site such as the spinal cord (14). In a randomized study in patients undergoing thoracotomy (n = 30), the oral administration of 100 mg of nimesulide (a COX-2 inhibitor) before incision and after surgery inhibited the increase in the cerebrospinal fluid (CSF) concentration of the PGI2 metabolite 6-keto-PGF compared with the control and ibuprofen groups, in which there were large increases in CSF 6-keto-PGF on the first two postoperative days (16). The patients randomized to the COX-2 inhibitor group also had decreased pain scores and postoperative opioid consumption, thus suggesting the clinical importance of suppression of CSF PG synthesis by a COX-2 inhibitor. These findings, along with other data, provide increasing evidence for the benefit of preoperative administration of oral COX-2 inhibitors for surgery (5–7,17).

Although we have demonstrated an inverse relationship between the plasma level of the selective COX-2 inhibitor rofecoxib and postoperative epidural drug consumption after knee arthroplasty (7) and although previous studies have assessed the pharmacokinetics of rofecoxib in plasma with either single (18–20) or multiple (18) dosing, no study has determined the amount of this compound that enters the CSF. If spinal COX-2 activity modulates pain perception, it is important to determine the extent and temporal pattern of distribution into the central nervous system of orally administered COX-2 inhibitors. Such knowledge is necessary to guide the appropriate timing of oral administration of COX-2 inhibitor before surgical incision. In addition, the time required to achieve peak CSF concentration in humans is important for assessing the maximum therapeutic effect of central COX-2 inhibition. The purpose of this study was to investigate, under well controlled conditions, the pharmacokinetic profile and CSF bioavailability of rofecoxib when administered orally with single or consecutive daily (steady-state) dosing.

Back to Top | Article Outline


Before subjects were enrolled, this study was approved by the IRB of Rush University Medical Center. Written informed consent was obtained from the subjects. Nine subjects of either sex who currently had an implanted SynchroMed® programmable drug pump with a sampling side port and intrathecal catheter were enrolled in the study. All of these individuals were receiving intrathecal opioids for the control of chronic pain. Inclusion criteria were age 30–75 yr, body weight 60–120 kg, and absence of any NSAID use for at least 2 wk before the study. Preenrollment verification of the patency of CSF sampling from the pump side port was performed. Subjects with previous allergic reactions to rofecoxib; known gastrointestinal disease, including prior upper gastrointestinal tract surgery; or significant renal or hepatic diseases were excluded. No subjects were receiving rifampicin or cimetidine.

Back to Top | Article Outline

Single-Dose Study

On Day 1, each subject arrived at the pain clinic between 6:00 and 7:00 am. The drug pump was programmed to the “no flow” position, and the subject was given a 20-mg dose of an oral slow-release morphine preparation (Kadian®) for pain control during the initial 1-day single-dose pharmacokinetic study. A peripheral IV catheter was placed for venous blood sampling. A baseline blood sample (3 mL) was collected, and CSF was sampled from the pump side port (the initial 1 mL was discarded, and the next 0.5 mL was collected for analysis). Each subject was then given a single oral 50-mg tablet of rofecoxib (Vioxx®; Merck & Co). Blood and CSF were then simultaneously sampled at 1, 2, 4, 6, 8, and 10 h after rofecoxib administration. During this time, the subjects remained in the pain clinic. After the 10-h sampling, the IV catheter was removed, and the subject was sent home. The programmable intrathecal pump was reprogrammed to the subject's previous opioid delivery rate after a single dose to account for the volume of the catheter length. The subjects returned to the pain clinic on the next day (Day 2), and venous blood and CSF were sampled 24 h after rofecoxib administration. Each subject's drug pump was then reprogrammed to its normal infusion mode.

Back to Top | Article Outline

Multiple-Dosing Study

A previous study of rofecoxib pharmacokinetics with daily oral dosing demonstrated that the trough (i.e., predose) concentration of rofecoxib in plasma was similar on Days 7 and 9 (18). We chose a 9-day protocol for determining the pharmacokinetics with multiple dosing. By following the dosing and sampling protocol of Day 1, each of the nine subjects took 50 mg of rofecoxib orally around 8:00 am on Days 2–8. Subjects received a telephone call each day to remind them of this dosing, which was verified by each subject's charting in a daily diary. On Day 9, each subject arrived at the pain clinic between 6:00 and 7:00 am. Drug administration and blood and CSF sampling were then performed as in the single-dose study.

Back to Top | Article Outline

Plasma and CSF Assay

Blood was collected in plasma tubes with heparin additive. The tubes were centrifuged within 2 h of collection, and 1 mL of plasma removed, pipetted into polypropylene tubes, and frozen at −80°C. CSF samples (0.5 mL) were directly transferred into polypropylene tubes and frozen at −80°C. Frozen samples were sent to Merck Frosst Canada for high-pressure liquid chromatography (HPLC) analysis of rofecoxib concentration.

For HPLC analysis, CSF was injected directly into the HPLC column. For plasma, the sample was mixed with an equal volume of acetonitrile and centrifuged, and the supernatant was used for injection. The lower limit of quantification (the smallest concentration on the standard curve that was measured within a 10% coefficient of variation) is 0.01 and 0.02 μg/mL for CSF and plasma, respectively. The interday coefficient of variation of the calibration standards was <4.0% and 4.3% for CSF and plasma, respectively, and the slopes and intercepts were within 10%. The intraday coefficient of variation was <3% for both CSF and plasma, and the slopes and intercepts were within 10%. The accuracy of the method is within 10% of the theoretical value for both CSF and plasma, and before each daily run, five standard samples are injected that must be accurate to within 2%. All calibration curves have a linear correlation of >0.999. There was no interference of the rofecoxib peak with other peaks from control plasma and CSF. Although we did not control for all of the subjects' medications that could potentially interfere with the HPLC assay, the Day 1 prerofecoxib baseline levels did not contain any peaks at the rofecoxib retention time. Samples were analyzed in three batches over 6 mo. The method used was an internal method used by Merck that is similar to the published method (21).

Back to Top | Article Outline

Pharmacokinetic Analyses

Single-Dose Study.

After the first rofecoxib dose on Day 1, pharmacokinetic variables were determined for both plasma and CSF in each subject (18–20). Area under the curve (AUC) was calculated for both CSF and plasma data by using the linear trapezoidal method from 0 to 24 h. The fraction of plasma rofecoxib in the CSF was calculated from the ratio of AUCCSF to AUCplasma.

Back to Top | Article Outline

Multiple-Dosing Study.

After the last rofecoxib dose on Day 9, for both plasma and CSF, pharmacokinetic variables were also determined for multidosed rofecoxib: 24-h AUC and the AUCCSF/AUCplasma ratio.

Back to Top | Article Outline


The mean age of the subjects was 51.9 yr (range, 41–75 yr), and the mean weight was 86.2 kg (range, 61.3–118.0 kg). There were five female and four male subjects. The 24-h concentration-time curve of rofecoxib in both plasma and CSF after 50-mg oral rofecoxib administration is shown in Figure 1. The CSF drug concentration lagged slightly behind the plasma drug concentration; e.g., at 1 h, the rofecoxib concentration in plasma was 30% of the maximum plasma concentration, whereas the CSF concentration was 14% of the maximum CSF concentration (it should be noted that some of the values at 1 h were below the lower limit of quantification of the assay, so this is only a rough estimate). In addition, the CSF/plasma ratio was only 0.067 at 1 h, versus 0.091 and 0.144 at 2 and 4 h, respectively. The maximum CSF/plasma ratio (0.159) occurred at 6 h. The AUCplasma was 5.40 ± 1.59 μg · h/mL (mean ± sd), and the AUCCSF was 0.72 ± 0.20 μg · h/mL. The time-averaged AUCCSF/AUCplasma ratio was 0.142 ± 0.046.

Figure 1

Figure 1

Figure 2 shows the 24-h concentration-time curve of rofecoxib in both plasma and CSF after the ninth daily 50-mg oral rofecoxib dose. The AUCplasma was 10.89 ± 3.68 μg · h/mL, the AUCCSF was 1.71 ± 0.62 μg · h/mL, and the AUCCSF/AUCplasma ratio was 0.159 ± 0.032. The rofecoxib concentrations in plasma and CSF were larger on Day 9 than on Day 1: the 24-h AUC on Day 9 was more than twice the Day 1 AUC. However, the CSF/plasma ratios on the 2 days were very similar (approximately 0.15).

Figure 2

Figure 2

Because of the presence of a secondary peak in the plasma concentration-time data of some subjects (most apparent in Fig. 1) and also because of insufficient sampling at later times, it was not possible to derive a meaningful estimate of the elimination half-life of rofecoxib. In addition, the lack of a clear peak in the plasma concentration-time data precluded accurate measurement of the time to maximum rofecoxib concentration and the maximum rofecoxib concentration.

Back to Top | Article Outline


This is the first human study to investigate the pharmacokinetics of a COX-2 inhibitor in the CSF and plasma simultaneously. The important finding of this study is that plasma rofecoxib enters the CSF with a CSF/plasma concentration ratio of approximately 15%. CSF rofecoxib at four hours was 0.04 ± 0.02 μg/mL. From our previous study (preoperative rofecoxib dosing), we obtained a CSF rofecoxib level of 0.03 ± 0.01 μg/mL at the time of spinal anesthesia (7); this was associated with immediately decreased postoperative epidural consumption. The time to achieve a 0.03 μg/mL concentration of CSF rofecoxib in this study after a single 50-mg dose of rofecoxib was approximately three hours. These data suggest that the preoperative administration of a single dose of 50 mg of rofecoxib will have the greatest likelihood of achieving a pharmacological effect at the spinal level at least three hours after oral dosing. At steady-state (after nine daily doses of oral rofecoxib 50 mg), the CSF rofecoxib concentration was more than 0.05 μg/mL at all times, so a more prolonged pharmacological effect can be anticipated.

Our plasma pharmacokinetic data after a single oral dose of rofecoxib are consistent with previous published human studies (18–20). We are in agreement with other investigators who have noted that the presence of secondary peaks in the plasma concentration-time curve “precludes estimation” (20) of the elimination half-life and makes it “difficult to measure properly” (18).

The main purpose of our study was to determine how much rofecoxib enters the central nervous system. A comparison of the CSF and plasma concentration-time curves shows that there was a minor delay in rofecoxib entering the CSF from the plasma. The CSF/plasma ratio was small (0.067) at one hour, and this is consistent with our previous study in which the CSF/plasma ratio was 0.073 when an oral 50-mg dose of rofecoxib was administered 24 and 2 hours before total knee arthroplasty and CSF was collected at the time of spinal anesthesia (7). In this study, by 4 hours the CSF/plasma ratio was 0.144, and even at 24 hours the ratio was 0.124, thus indicating that the relatively low early CSF rofecoxib level was due to a delay in the drug migrating from the plasma to the CSF. Other studies with oral drug administration and CSF sampling have shown similar results. For example, oral ondansetron, which has a CSF/plasma ratio similar to that of oral rofecoxib, had the smallest CSF/plasma ratios when drug distribution was assessed at <1.5 hours (21). The time delay in achieving maximum CSF rofecoxib concentration is not identical to the delay in blood-brain barrier transport, because the free drug must permeate the central nervous system capillaries and then migrate through the parenchyma to reach the CSF. Although there is also a blood-CSF barrier, the total surface area of the choroid plexus and other circumventricular organs is estimated to be 5000 times less than the surface area of the blood-brain barrier; therefore, the overall contribution to CSF drug levels is usually small (22).

With multiple dosing, the 24-hour CSF/plasma concentration ratio was 0.159. In an experiment performed in our laboratory, dogs were given 4 daily doses of oral rofecoxib over the range of 25–150 mg/d, and CSF and plasma concentrations were determined six hours after the last dose. In comparison to the data from our human subjects, the CSF/plasma ratio of rofecoxib for these animals remained between 0.25 and 0.40 over the entire dose range (23). In addition, rats ingesting rofecoxib 1.8–4.2 mg · kg−1 · d−1 in their feed obtained steady-state rofecoxib CSF/plasma concentration ratios between 0.19 and 0.35 (24). In a similar animal study using another COX-2 inhibitor, oral valdecoxib (100 mg/kg), the CSF/plasma ratio was 0.03–0.04 after a single dose (25).

Although the fraction of drug entering into the CSF with multiple dosing was not substantially different compared with single dosing (ratio of 0.159 vs 0.142), there are important advantages to the multiple dosing protocols. Once steady-state is achieved (estimated to be after four daily doses) (18), there is a basal level of rofecoxib in the plasma and CSF. At the start of Day 9, the zero-hour trough CSF rofecoxib level was 0.053 μg/mL, which is more than the peak level of 0.044 μg/mL achieved with a single 50-mg dose. This is reflected in the 24-hour rofecoxib AUC, which for multiple dosing is more than twice that achieved with a single dose. Hence, once steady-state is achieved with multiple dosing of rofecoxib, there is a much greater exposure of the spinal cord to the compound than with a single dose, and this may result in better-sustained therapeutic effects.

One limitation of this study is that there was insufficient sampling at later time periods. This may have led to some inaccuracy in the AUC calculation, because the drug concentration between 10 and 24 hours is represented by only two data points from each patient. Other variables, such as age, race, and possibly sex, may influence rofecoxib pharmacokinetics. In a plasma pharmacokinetics study (19) with eight male patients and four female patients, there was no difference in the AUC between sexes. However, as in our study, there was insufficient statistical power to compare pharmacokinetic differences according to sex. Future studies need to examine these factors and the entry of other oral COX-2 inhibitors into CSF, as well as the relationship between therapeutic efficacy and CSF concentrations of COX-2 inhibitor.

Back to Top | Article Outline


1. Woolf CJ, Chong MS. Preemptive analgesia: treating postoperative pain by preventing establishment of central sensitization. Anesth Analg 1993;77:362–79.
2. Walker JS. NSAID: an update on their analgesic effects. Clin Exp Pharmacol Physiol 1995;22:855–60.
3. Robinson CM, Christie J, Malcolm-Smith N. Nonsteroidal anti-inflammatory drugs, perioperative blood loss, transfusion requirements in elective hip arthroplasty. J Arthroplasty 1993;8:607–10.
4. Gilron I, Milne B, Hong M. Cyclooxygenase-2 inhibitors in postoperative pain management: current evidence and future directions. Anesthesiology 2003;99:1198–208.
5. Reuben SS, Connelly NR. Postoperative analgesic effects of celecoxib or rofecoxib after spinal fusion surgery. Anesth Analg 2000;91:1221–5.
6. Sinatra RS, Shen QJ, Halaszynski T, et al. Preoperative rofecoxib oral suspension as an analgesic adjunct for lower abdominal surgery: the effects on effort-dependent pain and pulmonary function. Anesth Analg 2004;98:135–40.
7. Buvanendran A, Kroin JS, Tuman KJ, et al. Effects of perioperative administration of a selective cyclooxygenase-2 inhibitor on pain management and recovery of function after knee replacement. JAMA 2003;290:2411–8.
8. Bartfai T. Telling the brain about pain. Nature 2001;410:425–6.
9. Ebersberger A, Grubb B, Willingale HL, et al. The intraspinal release of prostaglandin E2 in a model of acute arthritis is accompanied by an up-regulation of cyclooxygenase-2 in the spinal cord. Neuroscience 1999;93:775–81.
10. Samad TA, Moore KA, Sapirstein A, et al. Interleukin-1-beta-mediated induction of COX-2 in the CNS contributes to inflammatory pain hypersensitivity. Nature 2001;410:471–5.
11. Yaksh TL, Dirig DM, Conway CM, et al. The acute antihyperalgesic action of nonsteroidal, anti-inflammatory drugs and release of spinal prostaglandin E2 is mediated by the inhibition of constitutive spinal cyclooxygenase-2 (COX-2) but not COX-1. J Neurosci 2001;21:5847–53.
12. Kroin JS, Ling ZD, Buvanendran A, Tuman KJ. Upregulation of spinal cyclooxygenase-2 in rats after surgical incision. Anesthesiology 2004;100:364–9.
13. Kroin JS, Buvanendran A, McCarthy RJ, et al. Cyclooxygenase-2 inhibition potentiates morphine antinociception at the spinal level in a postoperative pain model. Reg Anesth Pain Med 2002;27:451–5.
14. Gordon SM, Brahim JS, Rowan J, et al. Peripheral prostanoid levels and nonsteroidal anti-inflammatory drug analgesia: replicate clinical trials in a tissue injury model. Clin Pharmacol Ther 2002;72:175–83.
15. Khan AA, Brahim JS, Rowan JS, Dionne RA. In vivo selectivity of a selective cyclooxygenase 2 inhibitor in the oral surgery model. Clin Pharmacol Ther 2002;72:44–9.
16. McCrory C, Fitzgerald D. Spinal prostaglandin formation and pain perception following thoracotomy: the role for cyclooxygenase-2. Chest 2004;125:1321–7.
17. Joshi W, Connelly NR, Reuben SS, et al. An evaluation of the safety and efficacy of administering rofecoxib for postoperative pain management. Anesth Analg 2003;97:35–8.
18. Depre M, Ehrich E, Hecken AV, et al. Pharmacokinetics, COX-2 specificity and tolerability of supratherapeutic doses of rofecoxib in humans. Eur J Clin Pharmacol 2000;56:167–74.
19. Werner U, Werner D, Mundkowski R, et al. Selective and rapid liquid chromatography-mass spectrometry method for the quantification of rofecoxib in pharmacokinetics with humans. J Chromatogr B Biomed Sci Appl 2001;760:83–90.
20. Halpin RA, Porras AG, Geer LA, et al. The disposition and metabolism of rofecoxib, a potent and selective cyclooxygenase-2 inhibitor, in human subjects. Drug Metab Dispos 2002;30:684–93.
21. Simpson KH, Murphy P, Colthup PV, Whelan P. Concentration of ondansetron in cerebrospinal fluid following oral dosing in volunteers. Psychopharmacology 1992;109:497–8.
22. Pardridge WM. Peptide delivery to the brain. New York: Raven Press, 1991.
23. Buvanendran A, Kroin JS, Luk P, et al. Blood brain barrier penetration of oral rofecoxib, a selective cyclooxygenase-2 inhibitor in dogs [abstract]. Anesthesiology 2002;97:A465.
24. Buvanendran A, Luk P, Kroin JS, et al. Central nervous system penetration of oral rofecoxib, a selective cyclooxygenase-2 inhibitor in rats [abstract]. Anesthesiology 2001;95:A868.
25. Zhang Y, Isakson P, Rathmell J, et al. Distribution of the COX-2 specific inhibitor valdecoxib into cerebrospinal fluid following oral administration. Am Pain Soc Abstract Database 2003;P765.
© 2005 International Anesthesia Research Society