Twelve physically active adults (six males, six females) between the ages of 18 and 45, who self-reported engaging in greater than 150 minutes per week moderate-intensity physical activity in the past six months, were recruited for this study. All testing and data collection were conducted at the Exercise Physiology Laboratory at the University of New Mexico during the summer months (June-August). All subjects resided in Albuquerque, New Mexico, and reported regularly exercising outdoors in the heat. Participants were excluded if they had a history of heat illness, stomach ulcers, were taking blood thinners, were pregnant (assessed prior to each visit using a urine pregnancy test), or had known allergies to NSAIDs. Participants came for a total of three visits including baseline testing and two experimental trials: one while consuming 600 mg of ibuprofen 12- and one-hour prior to exercise, and one while consuming an equivalent amount of corn starch (placebo) at the same time points. The protocol and procedures were approved by the local university institutional review board (protocol #2210021198), were registered as a Clinical Trial (Clinical Trials ID: NCT06247462), and were conducted in accordance with the Declaration of Helsinki.

During their initial visit, participants read and signed an informed consent document and completed a PAR-Q to be screened for health risks. Resting blood pressure was measured manually by the same researcher and subjects were excluded if their systolic and diastolic blood pressure were greater than 140 or 90 mmHg, respectively. If participants met the criteria to be in the study, height was measured with no shoes on using a stadiometer (SECA, seca 216, Chino, Calif., USA). While wearing a t-shirt and shorts, body weight and body composition were measured using a bioelectrical impedance scale (InBody770, InBody USA, Cerritos, CA). After a 10-minute running warm-up at a self-selected intensity, participants completed a baseline, graded exercise test on a treadmill to measure peak oxygen consumption (V̇O₂peak). The testing protocol was individualized, with either speed or grade increasing every minute, with a goal for the test to last between eight and 12 minutes (36). The graded exercise test was completed when the participant reached volitional fatigue. No other procedures were used to verify true V̇O₂max, thus the term V̇O₂peak is used throughout. Participants then underwent a 10-minute recovery period where they walked at a self-selected intensity. Finally, participants were instructed to run on the treadmill for a series of three-minute stages to determine the exercise intensity corresponding to 60% of their V̇O₂peak. This intensity was used for the exercise during the experimental trials. Oxygen consumption measured throughout the maximal exercise test and subsequent identification of 60% of V̇O₂peak was measured using eight-breath averages from the same metabolic cart (TrueOne 2400, ParvoMedics, Sandy, UT), which had been calibrated in accordance with manufacturer guidelines prior to each trial.

In a double-blind, randomized, crossover design, participants consumed either 600 mg of ibuprofen or placebo (corn starch) at both 12 hours and one hour prior to their exercise trials. A member of the research team not involved in the data collection or analysis was responsible for blinding and randomization using a random number generator. No confirmation of blinding was assessed. Participants came to the laboratory at least 48 hours after their baseline testing for these trials and were asked to refrain from using NSAID’s for one week prior to experimental trials. Participants were instructed to consume capsules, which appeared identical between conditions, 12 hours prior to their morning exercise trial (which began between 7:00 and 9:00 AM), and to adequately hydrate. Participants consumed a standardized breakfast provided by the researchers, containing between 400 and 600 calories (85–92 g carbohydrate, 5–12 g protein, 5–9 g fat), two hours prior to their exercise trial. Breakfasts were chosen by participants from a list of options which included selecting one carbohydrate/grain (oatmeal or granola bars), one fruit (banana or three clementines), and one juice (apple or orange). The breakfast chosen by each participant was matched during the second and third visit. Upon arriving at the laboratory one hour prior to exercise testing, participants ingested 600 mg of ibuprofen or placebo in unmarked capsules, then voided their bladder, and urine was assessed for hydration status by determining if the urine specific gravity (USG) was ≤ 1.020 (37). If subjects did not meet this criterion, trials were rescheduled. Next, participants waited for one hour without consuming any fluids and urine was again collected and the volume was measured to determine urine flow rate (UFR). This urine was also measured for USG and stored for later analysis.

Participants sat for 10 minutes before a blood draw was taken to collect serum and plasma. Next, nude body weight was measured, and participants were equipped with a heart rate monitor (Polar H10, Kempele, Finland) around their chest. Participants inserted a rectal thermistor 10 cm past the anal sphincter (Level 1 esophageal/rectal temperature probe, Smiths Medical, Minneapolis, MN, USA), which was connected to a receiver (4600 precision rectal thermometer, YSI Temperature, Dayton, OH). The participants then began their exercise trials, consisting of 60 minutes of treadmill running at 60% V̇O₂peak in a temperature-controlled environment set at 35 °C (Placebo: 35.1 ± 0.9 °C, Ibuprofen: 35.2 ± 0.9 °C, p = 0.37) and 45% relative humidity (Placebo: 44.0% ± 14.0%, Ibuprofen: 45.8% ± 11.9%, p = 0.22). The treadmill speed and grade were maintained constant during and between the trials, respectively. During the exercise trials, measures of core temperature, heart rate, rating of perceived exertion (38), thermal strain and comfort (39), dry bulb temperature, and humidity (QUESTemp QT-44, TSI QUESTemp, Shoreview, MN) were recorded at baseline and every ten minutes. At 30 minutes, participants ingested three milliliters of cool tap water (~12 °C) per kilogram of body weight. The rationale of the water temperature was to simulate what athletes (i.e., runners) usually experience during races at the water stations. As a safety measure to reduce the risk of exertional heat illness, exercise trials were immediately terminated if participants reached a core temperature of greater than 40.0 °C. GI symptoms were recorded via a visual analog scale, adapted from Gaskell et al., (2019), at pre-morning ingestion, pre-exercise, 30-minutes, post-exercise, and one-hour post-exercise. The scale included 18 GI symptoms classified as upper (belching, heartburn, upper abdominal bloating, stomach pain, urge to regurgitate), lower (flatulence, lower abdominal bloating, urge to defecate, abnormal stool), or other (nausea, dizziness, stitch) GI symptoms, on a scale from 0–10 where 0 indicates no symptoms; 1–3 indicates very mild symptoms; 4–6 indicates severe symptoms; and 7–10 indicates very severe symptoms. Scores from all time points were combined to assess incidence of GI symptoms.

After the exercise trials, participants were removed from the heat chamber and after 10 minutes in a seated position, a venipuncture to collect blood was performed. Urine was then collected, followed by nude body weight measurement. For the following hour, participants ingested a volume of water equivalent to the change in body weight from pre- to post-exercise. After this hour, another blood draw was conducted, and urine was collected to measure USG and UFR. The same experimental protocols with the alternate treatment were repeated for the final visit at least seven but no more than 28 days later (average of 16 ± 7 days between trials).

Urine was collected pre-, post-, and one-hour post-exercise and was immediately analyzed to measure volume for UFR and to manually assess USG using a refractometer (Cole-Parmer, RSA-BR90A, Vernon Hills, IL). After these initial analyses, urine was transferred to two mL tubes and stored at –80 °C until later analyses. Blood was collected into serum separating and EDTA vacutainer tubes through venipuncture from the antecubital vein immediately pre-, post- and one-hour post-exercise. To measure hematocrit, whole blood was transferred in triplicate to microcentrifuge tubes and centrifuged for five minutes. They were then analyzed via visual inspection and recorded as the nearest whole number using a micro-hematocrit reader (Micro-Capillary Reader No. 2201, International Equipment Company, Needham Heights, MA). Whole blood for serum samples was left to clot for approximately 30 minutes before centrifugation. Blood samples were centrifuged at 2200 rpm for 15 minutes at 22 °C to separate plasma and serum, respectively. They were then pipetted into two mL tubes respectively and stored at –80 °C until later analyses.

Concentrations of intestinal fatty acid binding protein (I-FABP) (Hycult Biotech, Uden, The Netherlands), lipopolysaccharide binding protein (LBP) (Hycult Biotech, Uden, The Netherlands), and soluble cluster of differentiation 14 (sCD14) (Hycult Biotech, Uden, The Netherlands) were measured at pre- and post-exercise timepoints in the plasma using enzyme-linked immunosorbent assay (ELISA) kits to assess intestinal damage (I-FABP) and endotoxemia (LBP, sCD14). To assess AKI, ELISA kits were used to measure concentrations of insulin-like growth factor-binding protein 7 (IGFBP7) (RayBiotech Life, Peachtree Corners, GA), tissue inhibitor of metalloproteinase 2 (TIMP-2) (RayBiotech Life, Peachtree Corners, GA), and neutrophil gelatinase-associated lipocalin (NGAL) (BioPorto, Gentofte, Denmark) in urine, while serum cystatin C (RayBiotech Life, Peachtree Corners, GA) was also analyzed to assess renal function at pre-, post-, and one-hour post-exercise timepoints. Urine samples were centrifuged and separated prior to analysis. Samples were diluted prior to analyses for I-FABP (1:2), LBP (1:1000), sCD14 (1:120), IGFBP7 (1:50), TIMP-2 (1:60), NGAL (1:250), and cystatin C (1:5000) to ensure the obtained values were on the standard curve for each ELISA. The product of IGFBP7 and TIMP-2 was calculated as this is a valid AKI marker approved by the US Food and Drugs Administration (NephroCheck^®), where values greater than 0.3 ng/mL²/1000 are considered an indication of increased risk for experiencing AKI in the following 12 hours (41) (42) (43). While the Kidney Disease Improving Global Outcomes guidelines define clinical AKI as a urine output of < 0.5 mL/kg/hour for at least six hours or an increase of serum creatine ≥ 0.3 mg/dL within 48 hours or 1.5-fold from baseline in seven days, these are normal physiological responses to exertional heat stress and may overestimate AKI (15). Thus, the aforementioned markers were used as they are more specific to renal damage rather than physiological changes in response to exertional heat stress and may be more accurate in assessing subclinical AKI (15). All ELISA analyses were completed according to manufacturer guidelines. Serum samples for pre- and post-exercise were sent to a commercial laboratory (Eve Technologies, Calgary, Canada) for analysis of a 15-plex assay of cytokines (Assay HDF15). Serum samples used to determine cytokine concentrations were not diluted prior to analyses. All assays were analyzed in duplicate, and intra-assay coefficient of variations for all markers were <10%.

To determine the sample size necessary to detect differences in the urinary NGAL response between placebo and ibuprofen conditions, an a priori power analyses was conducted using the effect size (ηp² = 0.33), an α-level of 0.05, power of 0.80 (1 − β), and assuming a correlation among repeated measures of 0.50. The effect size was used from a previous study by Semen et al. (2020), who investigated if runners that consumed NSAIDs prior to a race had greater increases in urinary NGAL. The analysis revealed a sample size of eight was required to detect differences in the NGAL response. A required sample size of eight subjects was also determined to detect differences using the same parameters and an effect size of ηp² = 0.30 for I-FABP based off of a study by Van Wijck et al. (2012) investigating the impact of ibuprofen on GI injury during exercise in thermoneutral conditions. Power analyses were completed using G-Power version 3.1.9.6 software.

Two-tailed paired sample t-tests were conducted to assess differences in peak core temperature and change in body weight between ibuprofen and placebo conditions. Changes in cytokines and markers of intestinal damage and endotoxemia were tested using two-way repeated measures analyses of variance (ANOVA) tests for main effects of condition (ibuprofen, placebo) and time (pre-, post-exercise), as well as an interaction between time and condition. The same testing was done to analyze changes in kidney injury and function for [IGFBP7•TIMP2], NGAL, and cystatin C, but peak post-exercise values were used instead of post-exercise. Peak post-exercise values were defined as the highest value from either the immediately post- or one-hour post-exercise timepoints. Peak values were analyzed rather than both post- and one-hour post-exercise for markers of AKI, as the aim of the study was to determine peak kidney injury rather than the time-course response. Mixed-effects models were used for the aforementioned variables in the case of missing data points. Markers with missing data points included LBP (1 data point: pre-exercise placebo), IL-1β (two data points: pre-, post-exercise ibuprofen, same subject), and IL-6 (five timepoints: one subject x all timepoints, one pre-exercise ibuprofen), which were missing due to falling outside the range of the standard curve.

Two-way repeated measures ANOVAs were used to assess changes in USG, UFR and hematocrit at pre-, immediately post-, and one-hour post-exercise timepoints. Changes in core temperature throughout the 60-minute exercise trials were compared using a mixed-effects analysis testing for the main effects of time (recorded at 10-minute intervals), condition (ibuprofen, placebo), and interaction between time and condition. This mixed-effects model was used to account for missing core temperature values, which occurred due to rectal probe malfunction during periods of testing. There were three times, from two trials (both placebo), when core temperature was not recorded. Any significant effects found during ANOVA or mixed effects model testing were further explored by comparing differences between different condition and time points using Tukey’s multiple comparisons post hoc test.

Due to non-normally distributed data assessed via visual inspection of Q-Q plot, analyses for IL-1Ra, IL-4, IL-5, IL12p40, IL-13, interferon-gamma (IFN-γ), [IGFBP7•TIMP-2], and NGAL were completed after log₁₀ transformation. The totaled GI symptoms were compared between conditions using a non-parametric Wilcoxon signed-rank test to handle non-normally distributed data. Incidence of GI symptoms was defined as a score of one or greater and is reported as a percentage. All statistical analyses listed above were completed using GraphPad Prism 9.4 (GraphPad Software Inc., La Jolla, CA, USA). Using RStudio (version 2021.09.2, R Development Core Team, Vienna, Austria), repeated measures correlations were conducted to assess the relationships between peak core temperature and the change (Δ) in LBP, sCD14, TNFα, IL-8, log₁₀[IGFBP7•TIMP2], and cystatin C from pre- to post- or peak post-exercise values. Effect size from pre-to post-exercise was calculated using Cohen’s d for all intestinal, immune, and AKI markers. An alpha level of 0.05 was set to determine significance for all tests.

Twelve participants (six females) were screened for this study. One male subject was removed due to reporting to trials dehydrated, as initial assessment of USG upon arrival to the laboratory was under the threshold of 1.020, however the pre-exercise USG assessed just prior to exercise was over this threshold. One participant, during their placebo trial, reached a core temperature of 40.0 °C prior to completing the 60-minute exercise protocol, and was removed from the heat chamber at minute 55 of running. Participant characteristics can be found in Table 1.

There were no differences in peak core temperature between the ibuprofen (38.9 0.6 °C) and placebo (39.0 0.5 °C) conditions (p = 0.73) (Figure 1B). Similarly, there was no effect of condition (p = 0.62) or interaction of condition and time (p = 0.78) on core temperature during the 60-minute exercise intervention; however, there was an effect of time (p < 0.001), with core temperature significantly increasing at each 10-minute interval (p < 0.05) (Figure 1A). There was a similar loss in body weight between the placebo (1.76% 0.72%) and ibuprofen (1.82% 0.59%) conditions (p = 0.27). There was no effect of time (p = 0.07), condition (p = 0.98), or interaction (p = 0.40) for hematocrit. Baseline hematocrit was 45% ± 4% in the placebo condition and 45% ± 4% in the ibuprofen condition. For the placebo condition, post-exercise hematocrit values were 46% ± 6% and 46% ± 5% for immediately post- and one-hour post-exercise, respectively, while values for the ibuprofen condition were 46% ± 3% and 47% ± 5%.

There was a significant time effect for UFR (p < 0.001), with decreases from pre- (473 ± 185 mL/hour) to post-exercise (77 ± 65 mL/hour), but there was no significant interaction effect (p = 0.63) or main effect of condition (p = 0.62). Similarly, there was a significant time effect of USG (p < 0.001), with increases from pre- (1.005 ± 0.004) to post- (1.012 ± 0.008), to one-hour post-exercise (1.017 ± 0.007), but there was no significant interaction effect (p = 0.71) or main effect of condition (p = 0.60). Post-hoc analyses showed that USG was only different between pre- and one-hour post-exercise in the placebo condition (p < 0.001), whereas both the post- (p = 0.03) and one-hour post-exercise (p < 0.001) values were significantly greater than the pre-exercise timepoint in the ibuprofen condition. There was an increase of urinary log₁₀ [IGFBP7•TIMP-2], from pre- to peak post-exercise (Placebo: Δ 1.8 ± 0.8 log10(ng/ml)²/1000, Ibuprofen: Δ 1.8 ± 0.9 log₁₀(ng/ml)²/1000) with no difference between conditions (Figure 2, A and B). Exploring the nonlog transformed data, four of the 11 (36%) participants in the placebo condition reached a peak post-exercise concentration greater than the NephroCheck^® threshold of 0.3 ng/mL²/1000, whereas six (55%) participants surpassed this value in the ibuprofen condition. There was an increase of urinary log₁₀ NGAL from pre-exercise to peak post-exercise values (Placebo: Δ 0.7 ± 0.5 log₁₀ ng/ml, Ibuprofen: Δ 0.9 ± 0.4 log₁₀ ng/ml), with no difference between conditions, and no interaction effect (Figure 2, C and D). Using the non-transformed data, there was only one trial, in the placebo condition, where a participant’s urinary NGAL surpassed the clinically used threshold of 150 ng/mL, indicative of increased risk of AKI (45). There was a significant main effect of time for serum cystatin C (Placebo: Δ 0.11 ± 0.14 mg/l, Ibuprofen: Δ 0.21 ± 0.28 mg/l), but there was not an interaction effect, or main effect of condition (Figure 2, E and F).

Results from the GI symptoms questionnaire can be found in Table 2. While not significantly different, the ibuprofen condition consistently had greater levels of incidence for upper, lower, other, and overall GI symptoms. Severity of symptoms throughout testing were low, as the highest reported symptom across all conditions and timepoints was a four out of 10, corresponding to “severe symptoms,” while all other reported symptoms were between one and three, corresponding to “very mild symptoms.” Running in the heat resulted in increased in intestinal cell injury indicated by an effect of time for I-FABP (Placebo: Δ 631 ± 446 pg/ml, Ibuprofen: Δ 576 ± 455 pg/ml) (Figure 3, A and B). Three of the 11 (27%) participants in the placebo condition had a change in I-FABP greater than 1000 pg/mL which has been proposed be a clinically significant threshold for increased risk of microbial translocation and endotoxemia (46). Only one participant (9%) surpassed this value in the ibuprofen condition. There was no significant effect of time, condition, or interaction for either marker of endotoxemia (LBP [Placebo: Δ −1.2 ± 3.2 μg/ml, Ibuprofen: Δ 1.0 ± 1.6 μg/ml], sCD14 [Placebo: Δ 58 ± 172 ng/ml, Ibuprofen: Δ 119 ± 216 ng/ml]) (Figure 3, C-F). Results for circulating cytokines can be found in Table 3. One hour of running in the heat increased concentration of several circulating cytokines including IL-1Ra, IL-6, IL-8, IL-10, MCP-1, and TNFα, indicated by a significant effect of time (p < 0.05). Three cytokines were found to have a significant effect of condition, as ibuprofen increased circulating IL-8 and decreased IL-4 and IL-13. There was also a significant interaction effect for IL-13, and a post hoc analysis showed that the post-exercise timepoint was significantly lower in the ibuprofen condition compared to the placebo condition (p = 0.002). There was a significant interaction effect for IL-5, but post hoc testing revealed no differences between timepoints.

Of the seven variables assessed to investigate the relationship between peak core temperature and gut-cytokine-kidney crosstalk, there were four significant correlates found (Figure 4). A positive relationship was found between peak core temperature and change in log₁₀IGFBP7•TIMP2. Change in both AKI markers (log₁₀IGFBP7•TIMP2 and cystatin C) were positively associated with changes in TNFα. Lastly, there was a positive relationship between changes in sCD14 and IL-8.