Renal blood flow (RBF) and glomerular filtration rate (GFR) are reduced by 40–50% during strenuous exercise (14). Renal prostaglandins are essential in modulating renal perfusion under stress conditions. Prostaglandin formation is regulated by cyclooxygenase (COX) activity and there are two isoforms of cyclooxygenase, COX-1 and COX-2. COX-1 is constitutively expressed in most tissue and is associated with maintaining cellular integrity, vascular tone, and hemostasis. COX-2 is the inducible form of cyclooxygenase and is predominantly associated with acute inflammation; however, COX-2 is constitutively expressed in the human kidney (21). COX-1 appears to regulate PGE2 synthesis predominantly in the inner cortex and outer medullary regions to modulate tubular sodium and chloride reabsorption, and attenuate vasopressin-induced reabsorption of water, whereas COX-2 is associated with media smooth muscle and vascular pericytes of all segments of the pre- and postglomerular vasculature and regulates arteriolar tone to maintain RBF and GFR (21). COX-2 is thought to be induced in the human macula densa in conditions of low physiologic or compromised transepithelial NaCl transport (3,8,11,12). Therland and colleagues (21) propose that in sodium-replete individuals at rest, COX-2 protein in the macula densa is below current detection limits, but it can be rapidly induced and generate substrate for prostaglandin synthesis in states of low sodium intake, diuretic treatment, or altered renal perfusion. The role of COX-2 in exercise-induced changes in renal function is unknown.
The nonsteroidal antiinflammatory drugs (NSAID) inhibit cyclooxygenase, which results in decreased conversion of arachidonic acid to the endoperoxides, thereby decreasing prostaglandin synthesis. Recently, COX-2 selective inhibitors have been introduced to clinical practice, with the aim of inhibiting the formation of inflammatory prostaglandins without producing the unwanted side effects caused by the removal of cytoprotective effects of prostaglandins (2,5).
The use of NSAID to manage musculoskeletal injuries is an integral component of the management of sports injuries. The use of these agents may reduce the inflammation and discomfort associated with the injury, improving the athlete’s recovery time, and allow a quicker return to their physical activity. However, the use of NSAID also may have adverse consequences, including alterations in renal hemodynamics (17,23,24). In a previous study, we demonstrated that indomethacin (a nonselective cyclooxygenase inhibitor) significantly increased the exercise-induced reduction in RBF after 30 min of strenuous exercise. Changes in GFR paralleled the changes in RBF. Free water clearance postexercise was also significantly reduced by indomethacin (23). The role of selective COX-2 inhibition on exercise-induced changes in renal function is unknown.
This study was undertaken to investigate the role of selective COX-2 inhibition (celecoxib), compared with nonselective cyclooxygenase inhibition (indomethacin) or placebo on renal function following strenuous exercise.
We undertook a randomized double-blind cross-over study comparing indomethacin, celecoxib, and placebo on the exercise-induced changes in renal function. Twelve healthy male participants undertaking regular exercise were enrolled in the study, having given written informed consent. The study was approved by the Otago ethics committee. Sample size was based on our previous study (23) that detected a 31% reduction in RBF in eight subjects taking indomethacin; therefore, using the standard difference of the mean for RBF of 188 mL·min−1 and a difference of 115 mL·min−1, nine subjects were estimated to be required for an 80% power to detect a 30% difference in RBF at an alpha level of 0.05.
The participants were physically fit, undertaking regular exercise, nonsmokers, and taking no medications. The mean age of our participants was 24.3 yr (range 19–45 yr), with an average body mass index (BMI) of 24.6 (range 20.1–30.2). All experimental sessions took place in the Environmental Chamber at the Otago University School of Physical Education, at a dry bulb temperature of 25°C and a relative humidity of 50%.
Participants’ maximal aerobic power (V̇O2max) was measured directly while treadmill running. Respiratory gases were measured online (Sensor Medics 2900, SensorMedics Co., California) during a stepped run-to-exhaustion protocol, in which the treadmill speed began at 7 km·h−1 and incremented 1 km·h−1 each minute until volitional exhaustion. The V̇O2max (mL·min−1·kg−1 body mass) was estimated to be the highest 30-s mean (Table 1). The speed required to elicit a steady-state 80% of V̇O2max value was then calculated and verified using a second run with online gas analysis. The participant was required to run for a full 30 min at the established speed, during which ratings of perceived exertion were taken to confirm that the participant could run at this speed in the experimental sessions. This speed was maintained for all three experimental sessions.
Participants commenced their allocated medication 36 h before each experiment to achieve a steady state for each drug, and took the last dose at 0700 on the morning of an experimental session. The order of drugs was randomized, with the researchers and participants blinded to which drug was used in each trial. The drugs were labeled only as drug A, B, or C, and were dispensed in a plain container enclosed in a sealed envelope, which was labeled as drug A, B, or C. The packaging and randomization, using a random number generator, was undertaken by the department’s pharmacist. The randomization code was not released until all data were analyzed. Participants had at least 1 wk between each trial to allow an adequate washout period. All experimental sessions were carried out at the same time of day to reduce any influence of normal diurnal variation.
Participants had a light breakfast the morning of the session, having abstained from alcohol the previous night. The breakfast was the same for each of the three experimental sessions. Participants took the last of the drug with this breakfast and reported to the environmental chamber at 0800. Nude body mass, blood pressure, and a baseline urine sample were taken. Intravenous cannulae were placed in both forearms and a baseline blood sample was taken. An infusion of 1 L of 0.9% sodium chloride was given over 30 min. Additional water was then given orally to complete a 20 mL·kg−1 water load to maximize urine flow rate. Once water loading was complete, a priming dose of PAH (10 mg·kg−1) was given, followed by a continuous infusion of PAH 20% in saline solution delivering 12 mg of PAH per minute. After a 45-min equilibrium period to allow plasma levels of PAH to reach steady state, baseline measures of PAH clearance, creatinine clearance, and free water clearance were taken by timed urine and blood samples over a 1-h period with participants sitting comfortably at rest. Urine volumes were accurately measured and timed, with equivalent replacement volumes of water given to the participant to drink to maintain maximal hydration. After the baseline period, participants then undertook 30 min of treadmill exercise at the predetermined 80% V̇O2max. Blood and urine samples were taken immediately after exercise to establish the exercise clearances. During the following 2 h of recovery, urine was collected from the participant every 30 min, with blood samples taken at the midpoint between urine samples (23).
Aliquots of urine were stored at 4°C and later analyzed for PAH, creatinine, and osmolality. Plasma was obtained by centrifuging samples at 500 × g at 4°C for 10 min. Aliquots were taken for PAH, creatinine, sodium, and osmolality. Plasma creatinine and sodium was measured on a Roche modular analyzer (Roche Diagnostics), and urine creatinine on a Hitachi 911 analyzer (Hitachi), using the kinetic colorimetric technique. Plasma and urine osmolality were measured in an Advanced Instruments Osmometer model 3D3 using the freezing-point depression. Plasma and urine samples for PAH were stored at −20°C and analyzed in a single batch by Endolab Canterbury Health (Christchurch, NZ), using a standard colorimetric assay. The hematocrit was measured using the Sysmex XE-2100 analyzer by the hydrodynamic DC detection method.
GFR was measured by creatinine clearance and RBF derived from renal plasma flow measured by paraaminohippurate clearance, and the hematocrit. Free water clearance was derived from urine flow rate and osmolar clearances.
At low doses, PAH is almost completely removed at the first pass through the renal vasculature. This allows the assumption that the PAH concentration in the renal vein is “0,” so the amount of PAH leaving the kidneys in the urine will be equal to the amount of PAH entering the kidneys through the renal arteries. With a steady state, peripheral venous concentrations of PAH will be the same as renal arterial concentrations of PAH. RBF was then calculated from the renal plasma flow, as the RBF is the renal plasma flow divided by 1 − hematocrit (for formulae, see Appendix).
Endogenous creatinine clearance was used to measure GFR. With normal renal function, it is assumed that creatinine is completely filtered at the glomeruli; thus, creatinine clearance is equivalent to GFR (see Appendix).
The free water clearance is the urine flow rate minus the clearance of solutes measured by solute osmolality in the blood and urine. The osmolality clearance (COSM) was calculated for each given time point and deducted from the total volume of urine for that time period to get the free water clearance (V − COSM) (see Appendix).
Mixed models were used to analyze the separate and interactive effects of time and treatment (celecoxib, indomethacin, and placebo) on RBF, GFR, and free water clearance. The repeated measures nature of the study design was taken into account by using a random coefficients model, as implemented in the SAS procedure Mixed (SAS/STAT V8 Software User’s Guide, SAS Institute, Cary, NC). This allows for subject-to-subject variation in the profiles of outcome versus time, as well as the fact that measurements within a subject are generally more highly correlated than between subjects. The cross-over design was taken into account by testing period and order effects. Interest was primarily in testing the postexercise difference between the treatments and placebo; thus, the time by treatment interaction was the major focus in hypothesis testing.
Twelve healthy male volunteers were recruited for this study, and their demographics are displayed in Table 1. All participants completed the study. There were no detectable side effects related to the medications. RBF (Fig. 1) fell by 57–60%, and GFR (Fig. 2) fell by 34–37% of baseline values at the completion of the 30-min exercise period, but there was no significant difference between placebo, indomethacin, or celecoxib. Renal hemodynamics had returned to their baseline values by 2 h. Free water clearance was reduced at the end of the exercise period in all groups, with free water clearance remaining significantly reduced during the 2-h recovery period in both drug treatment groups when compared with placebo (indomethacin −2.43 ± 0.95 mL·min−1, P < 0.007; celecoxib −3.88 ± 0.94 mL·min−1, P < 0.0001) (Fig. 3). At baseline and immediately on completion of the 30 min of exercise, the free water clearance in the celecoxib and indomethacin groups was lower than the placebo group, but this just failed to reach significance (P = 0.06) (Fig. 3). There were no differences in plasma sodium concentrations or plasma arginine vasopressin concentrations between the groups at any time point (data not shown). Body weight demonstrated a small upward trend in the drug treatment groups at the completion of the study, but it did not reach significance (P = 0.07).
This study examined the impact of a nonselective cyclooxygenase inhibitor (indomethacin) or a selective cyclooxygenase inhibitor (celecoxib) compared with placebo on renal function during strenuous exercise. RBF fell by 57–60%, and GFR fell by 34–37% at the completion of 30 min of strenuous exercise and had returned to baseline values by the completion of the 2-h recovery period (Figs. 1 and 2). There was no significant impact of indomethacin, celecoxib, or placebo on the reduction of RBF or GFR. Free water clearance was reduced at the end of the exercise period in all groups, but it remained lower during recovery in both treatment groups than in placebo (indomethacin 5.97 ± 0.95 mL·min−1, P < 0.007; celecoxib 4.52 ± 0.94 mL·min−1, P < 0.0001; placebo 8.40 ± 0.95 mL·min−1) (Fig. 3). Free water clearance also tended to be lower in the celecoxib and indomethacin groups both before and at the end of the exercise (Fig. 3).
These results demonstrate that exercise produces a significant decrease in renal hemodynamics and free water handling as has previously been reported (23). Farquhar and colleagues (6) demonstrated that ibuprofen, under stressed conditions of salt restriction, dehydration, and heat, produced a significantly greater decrease in GFR after 45 min of exercise at 65% V̇O2max, when compared with placebo or acetaminophen. Similar reductions in GFR were evident in our study, but these did not reach significance. This may be related to the hydration protocol used in our study to maintain a maximal urine flow rate.
Previously, we have demonstrated that indomethacin significantly impaired free water clearances postexercise compared with controls (23), which was again demonstrated in this study but was also evident to a similar extent with selective inhibition of the cyclooxygenase 2 isoenzyme by celecoxib. The selective cyclooxygenase inhibitors have been in introduced into clinical practice to ostensibly reduce potential side effects related to nonselective cyclooxygenase inhibition (2,5).
The cyclooxygenase 2 isoenzyme is predominantly induced under inflammatory conditions; however, cyclooxygenase 2 is constitutively expressed in vascular tissue including the kidney (21). COX-1 appears to regulate PGE2 synthesis predominantly in the inner cortex and outer medullary regions to modulate tubular sodium and chloride reabsorption, and attenuate vasopressin-induced reabsorption of water, whereas COX-2 is associated with media smooth muscle and vascular pericytes of all segments of the pre- and postglomerular vasculature and regulates arteriolar tone to maintain RBF and GFR (21). However, COX-2 is thought to be induced in the human macula densa in conditions of low physiologic or compromised transepithelial NaCl transport (3,8,11,12). Therland and colleagues (21) propose that in sodium-replete individuals at rest, COX-2 protein in the macula densa is below the detection limit, but it can be rapidly induced and generate substrate for prostaglandin synthesis in states of low sodium intake, diuretic treatment, or altered renal perfusion. The results obtained from this study indicate that the action of indomethacin and celecoxib in inhibiting free water clearance postexercise is being mediated via inhibition of PGE2 formation via predominately the actions of cyclooxygenase 2.
Clinical studies have reported variable results of selective cyclooxygenase inhibition on renal function. Under conditions of mildly impaired renal function (especially in the elderly) or after a low sodium diet, cyclooxygenase inhibitors can affect renal function (15,17,18,20,25). However, it is important to remember that these studies examined patients under basal resting conditions only. The impact of selective cyclooxygenase inhibition on physiologically induced changes in renal function related to exercise in normal individuals has not previously been reported.
The potential clinical risk of impaired free water clearance after exercise is the development of hyponatremia and more rarely acute cerebral edema. There have been a number of observational studies reporting the occurrence of hyponatremia in athletes participating in endurance events. Speedy and colleagues (19) reported that up to 18% of participants in an “iron man” event were found to have serum sodium concentrations below 135 mmol·L−1 and 3.3% had levels below 130 mmol·L−1.
Exercise tolerance in the heat has been shown to be linearly related to serum sodium concentration, even above levels regarded as hyponatremia (22). Similarly, a number of authors have reported their experiences of marathon runners reporting to the medical services upon completion of the marathon, a number of whom had potentially life-threatening hyponatremia (1,4,9,10,13). Of these, Ayus and colleagues (1) reported seven cases of life-threatening noncardiogenic pulmonary edema and acute cerebral edema; one of whom died as a result of the acute cerebral edema. In these series, the identifiable causal factors were excessive water intake (both before and during the event), slower times (suggesting less well-conditioned athletes), female athletes, and the use of NSAID (1,4,9,10). The results of our study provide the physiological evidence implicating both NSAID as well as selective COX2 inhibitors as a major contributing factor to the hyponatremia through the inhibition of free water clearance by the kidneys.
In this study, all subjects received a 20 mL·kg−1 fluid load to ensure maximal diuresis, which is essential for the clearance techniques used to determine RBF and GFR. This potentially could have exaggerated the impaired free water clearances evident in this study but emphasizes the potential role of water loading before an event, as a contributing factor in the development of exercise-induced hyponatremia. In this study, two thirds of the fluid load was administered as a saline load to try and minimize the risk of hyponatremia; as in our first study (23), one participant retained 3 L of fluid and became hyponatremic after completion of the experimental session with indomethacin. In this study, there was otherwise no significant fall in plasma sodium concentration or excessive weight gain, despite the significant fall in free water clearances.
Unlike our previous study (23) with indomethacin alone, which demonstrated a significant reduction in RBF postexercise, we did not demonstrate a significant reduction in GFR or RBF with either indomethacin or celecoxib. In the first study, only recreational athletes took part, whereas in this study, fitness was higher with three of the participants actively participating in endurance events. Their degree of adaptation to exercise stress was greater, with less of a reduction in their RBF or GFR in response to exercise compared with the rest of the participants and is consistent with previous observations that aerobic fitness is associated with better maintenance of regional perfusions during exercise and/or heat stress (7,16). Although the numbers were small and not sufficient for statistical validation, nevertheless, these data would support the findings that the degree of physical training has an important impact on renal function during exercise.
A relatively small number of participants may have influenced the results, but the number who recruited and completed the study exceeded the power calculations based on our previous study (23). Likewise, Farquhar and colleagues (6) had a similar number of participants in their study, which demonstrated a significant change in GFR, in the ibuprofen group, after exercise. Familiarization with the study protocol could have influenced the results, but the randomization process and minimum 7-d spacing between sessions would have made this unlikely. Likewise, incomplete bladder emptying and hence an inaccurate urine collection is a potential source of error. However, all participants were aware of the need for complete emptying of their bladder each time they voided, and the baseline clearances remained consistent through out the study. The maintenance of a high urine flow rate with fluid loading would have minimized the possibility of incomplete bladder emptying due to low volumes. It would have been of interest to demonstrate similar findings with female athletes, given the observation that they may be at higher risk of exercise-induced hyponatremia as discussed above (1,9,13). For statistical validity, we would have needed to recruit an equal number of female athletes as well as controlling for timing with respect to menstrual phase. This is a proposed study yet to be undertaken.
In summary, strenuous exercise is associated with a reduction in RBF, glomerular filtration, and free water clearance, which can be modified in part by the degree of fitness of the athlete even at normalized work intensity. The alterations in renal hemodynamics and free water handling induced by exercise are, in part, modulated by renal prostaglandins. The use of drugs that inhibit prostaglandin formation can therefore significantly enhance the exercise-induced changes in renal function. In this study, both indomethacin, a nonselective cyclooxygenase inhibitor, and celecoxib, a selective cyclooxygenase-2 inhibitor, significantly reduced free water clearance in the recovery phase after 30 min of strenuous exercise. These findings provide the physiological basis for the observed cases of exercise-induced hyponatremia after endurance events (1,4,9,10,13). Antiinflammatory drugs play a major role in treating sports injuries, and athletes will frequently continue to take these agents as part of an early return to their sporting activity. It is therefore important that physicians involved in treating athletes are aware of these risks and advise their patients appropriately.
1. Ayus, J. C., J. Varon, and A. I. Arieff. Hyponatremia, cerebral edema, and noncardiogenic pulmonary edema in marathon runners. Ann. Intern. Med
. 132:711–714, 2000.
2. Bombardier, C., L. Laine, A. Reicin, et al. VIGOR study group: comparison of upper gastrointestinal toxicity of rofecoxib and naproxen in patients with rheumatoid arthritis. N. Engl. J. Med
. 343:1520–1528, 2000.
3. Castrop, H., F. Schweda, K. Schumacher, K. Wolf, and A. Kurtz. Role of renocortical cyclooxygenase 2 for renal vascular resistance and macula densa control of renin secretion. J. Am. Soc. Nephrol
. 12:867–874, 2001.
4. Davis, D. P., J. S. Videen, A. Marino, et al. Exercise
-associated hyponatremia in marathon runners: a two-year experience. J. Emerg. Med
. 21:47–57, 2001.
5. Davies, N. M., A. J. McLachlan, R. O. Day, and K. M. Williams. Clinical pharmacokinetics and pharmacodynamics of celecoxib: a selective cyclo-oxygenase-2 inhibitor. Clin. Pharmacokinet.
6. Farquhar, W. B., A. L. Morgan, E. J. Zambraski, and W. L. Kenney. Effects of acetaminophen and ibuprofen on renal function in the stressed kidney
. J. Appl. Physiol.
7. Fritzsche, R. G., and E. F. Coyle. Cutaneous blood flow during exercise
is higher in endurance-trained humans. J. Appl. Physiol.
8. Harris, R. C. Cyclooxygenase-2 in the kidney
. J. Am. Soc. Nephrol
. 11:2387–2394, 2000.
9. Hew, T. D., J. N. Chorley, J. C. Cianca, and J. G. Divine. The incidence, risk factors, and clinical manifestations of hyponatremia in marathon runners. Clin. J. Sport Med
. 13:41–47, 2003.
10. Hsieh, M., R. Roth, D. L. Davis, H. Larrabee, and C. W. Callaway. Hyponatremia in runners requiring on-site medical treatment at a single marathon. Med. Sci. Sports Exerc.
11. Kammerl, M. C., R. M. Nusing, H. W. Seyberth, G. A. Riegger, A. Kurtz, and B. K. Kramer. Inhibition of cyclooxygenase-2 attenuates urinary prostanoid excretion without affecting renal renin expression. Pflugers Arch.
12. Khan, K. N., S. K. Paulson, K. M. Verburg, J. B. Lefkowith, and T. J. Maziasz. Pharmacology of cyclooxygenase-2 inhibition in the kidney
. Kidney Int
. 61:1210–1219, 2002.
13. Noakes, T. Hyponatremia in distance runners: fluid and sodium balance during exercise
. Curr. Sports Med. Rep
. 1:197–207, 2002.
14. Poortmans, J. R. Exercise
and renal function. Sports Med
. 1:125–157, 1984.
15. Rossat, J., M. Maillard, J. Nussberger, H. R. Brunner, and M. Burnier. Renal effects of selective cyclooxygenase-2 inhibition in normotensive salt-depleted subjects. Clin. Pharmacol. Ther.
16. Sakurada, S., and J. R. Hales. A role for gastrointestinal endotoxins in enhancement of heat tolerance by physical fitness. J. Appl. Physiol
. 84:207–214, 1998.
17. Sanchez, P. L., L. M. Salgado, N. R. Ferreri, and B. Escalante. Effect of cyclooxygenase-2 inhibition on renal function after renal ablation. Hypertension
18. Schwartz, J. I., K. Vandormael, M. P. Malice, et al. Comparison of rofecoxib, celecoxib, and naproxen on renal function in elderly subjects receiving a normal-salt diet. Clin. Pharmacol. Ther
. 72:50–61, 2002.
19. Speedy, D. B., I. R. Rogers, T. D. Noakes, et al. Exercise
-induced hyponatremia in ultradistance triathletes is caused by inappropriate fluid retention. Clin. J. Sport Med.
20. Swan, S. K., D. W. Rudy, K. C. Lasseter, et al. Effect of cyclooxygenase-2 inhibition on renal function in elderly persons receiving a low-salt diet: a randomized, controlled trial. Ann. Intern. Med.
21. Therland, K. L., J. Stubbe, H. C. Thiesson, et al. Cycloxygenase-2 is expressed in vasculature of normal and ischemic adult human kidney
and is co-localised with vascular prostaglandin E2 EP4 receptors. J. Am. Soc. Nephrol.
22. Vrijens, D. M., and N. J. Rehrer. Sodium-free fluid ingestion decreases plasma sodium during exercise
in the heat. J. Appl. Physiol
. 86:1847–1851, 1999.
23. Walker, R., P. Fawcett, E. Flannery, and D. Gerrard. Indomethacin potentiates exercise
-induced reduction in renal hemodynamics in athletes. Med. Sci. Sports Exerc
. 26:1302–1306, 1994.
24. Whelton, A. Nephrotoxicity of nonsteroidal anti-inflammatory drugs: physiologic foundations and clinical implications. Am. J. Med
. 34:13S–24S, 1999.
25. Whelton, A., W. B. White, A. E. Bello, J. A. Puma, and J. G. Fort. Effects of celecoxib and rofecoxib on blood pressure and edema in patients > 65 years of age with systemic hypertension and osteoarthritis. Am. J. Cardiol
. 90:959–963, 2002.
Renal plasma flow (RPF) measured as PAH clearance (mL·min−1).
RPF = (urinary PAH)·(urinary volume)/(plasma PAH)·time
Renal blood flow (RBF) measured as renal plasma flow divided by 1 − hematocrit (mL·min−1).
RBF = RPF/1 − Hct.
Glomerular filtration rate (GFR) measured as creatinine clearance (mL·min−1).
GFR = (urinary creatinine)·(urinary volume)/(plasma creatinine)·Time
Solute clearance measured as osmololar clearance (Cosm) mL·min−1
Cosm = (urinary osmolality)·(urinary volume)/(plasma osmolality)·time
Free water clearance measured as urine flow rate (V) minus solute clearance (mL·min−1)
Free water clearance = V − Csom