Before study commencement, an effect size was calculated based on recent data from our laboratory demonstrating that increases in a marker of renal oxidative stress (urinary liver type fatty acid binding protein) was modified by the magnitude of dehydration and elevations in core temperature (21)(f=0.29). Using this effect size and the observed correlation coefficient among repeated measures (r=0.75), a sample size of 12 participants was found to provide at least 80% power to detect a significant trial x time interaction using a two-sided test and an alpha=0.05 (G*Power 3.1.9.4). With this background, thirteen healthy adults (5 women) participated in this study to try to ensure that at least twelve data sets were obtained for most outcome variables. Participant characteristics are reported in Table 1. All participants were nonsmokers and reported to be free from any known cardiovascular, renal, metabolic, neurologic, or gastrointestinal diseases. Additionally, participants self-reported to be regularly engaged in regular moderate physical activity 3–5 days per week (22, 23). Women self-reported to be normally menstruating and were not pregnant at any point during the study, which was confirmed via a urine pregnancy test before each trial. Women were not on any form of birth control and were tested in the first 5 days of self-identified menstruation a time in which estrogen and progesterone are at their lowest levels, which is similar to other heat stress focused studies (24) The study was conducted throughout the calendar year in Bloomington, IN, USA.

Participants visited the laboratory on three separate occasions. The first visit included screening, familiarization, and a graded submaximal exercise test, the latter of which was used to determine exercise intensity during the subsequent experimental trials. The following two visits were experimental trials (see Experimental protocol) and were separated by at least 3 days. The mean duration between experimental trials was 38 ± 70 days, the variability is due to the greater length of time between trials to control for the menstrual cycle in women (73 ± 102 days) compared to men (17 ± 32 days). This study was approved by the Institutional Review Board at Indiana University (IRB#: 15402) and conformed to the Declaration of Helsinki, except for registration in a database. Informed written consent was obtained from all participants before participating in this study.

This study employed a block randomized, crossover design. An overview of the experimental protocol is presented in Figure 1. Participants arrived at the laboratory for the two experimental visits, having refrained from strenuous exercise, alcohol, and caffeine for 12 hours and from food for 2 hours. Upon arrival, participants voided their bladder and euhydration was confirmed via a urine specific gravity of ≤1.020 (75), and a negative pregnancy test was confirmed for women. Participants were weighed nude in a private room and then provided 250 mL of room temperature water to promote urine production over the proceeding 60 minutes. Participants then assumed a supine position, and a venous blood sample was taken after 20 minutes to account for shifts in plasma volume due to changes in posture (76). A pre-exposure urine sample was collected by participants completely voiding their bladder ~60 minutes after the previous urine void. A nude body measurement was then obtained. Following pre-simulated occupational heat stress measurements, participants entered the environmental chamber maintained at 38.6 ± 0.4°C and 54 ± 1% relative humidity. These conditions result in a wet bulb globe temperature of 33.1 ± 0.2°C and were chosen because outdoor workers in the U.S. (e.g., construction, agriculture, etc.) are often exposed to this level of environmental heat (77).

Upon entering the environmental chamber, participants assumed the seated position for pre- simulated occupational heat stress renal ultrasound measurements, which was immediately followed by a CPT. Participants then engaged in two hours of simulated occupational heat stress consisting of 10 minutes of walking on a treadmill and 5 minutes of rowing exercise every 15 minutes throughout the 2 hours (8 rounds total). For both treadmill walking and rowing, the external workload was set to an intensity sufficient to elicit an absolute oxygen consumption of 1.2 L/min. The external workload was established during the screening visit, and the external workload was maintained constant throughout both experimental trials. A target of 1.2 L/min was chosen because it represents the absolute oxygen consumption agricultural and construction workers experience when performing moderate work (78, 79). A novel aspect of this study is the use of rowing ergometry to simulate the upper body work that is common in manual labor. Specifically, rowing was used because it is a whole body, continuous exercise utilizing large muscle groups in the legs, back, shoulders, and arms, and better mimics manual tasks like hoeing, shoveling, or manual harvesting compared to treadmill walking alone (80). Thus, the combined use of treadmill walking and rowing better mimicked tasks workers experience. For example, workers often must walk to different areas on a job site to perform the necessary upper body work.

In one trial, participants were provided 237 mL of a non-caloric sports drink (Gatorade Zero) every 15 minutes during exercise to encourage ad libitum drinking (Drink trial). This approach is consistent with current occupational heat stress and hydration recommendations (81), and we have shown previously results in minimal dehydration (9, 82). In the other trial, the participants were not given fluids to drink (No Drink trial).

Following the first hour of exercise, participants were quickly (~1 minute) moved to the seated position for mid-exercise measurement of segmental artery blood velocity. At the end of the second hour of exercise, participants were again seated and underwent segmental artery blood velocity measurements, which was immediately followed by the post- simulated occupational heat stress CPT. Participants exited the environmental chamber after these measurements were conducted.

Immediately upon leaving the environmental chamber, a post- simulated occupational heat stress urine sample was collected. Participants then toweled off prior to the measurement of nude body weight. Participants assumed the supine position and a post- simulated occupational heat stress venous blood sample was obtained following 20 minutes of supine rest. Participants were then given 250 mL of room-temperature water to promote urine production, as we have done previously (7). Approximately 60 minutes after post-exposure urine collection, the participants provided a recovery urine sample. Participants then entered the environmental chamber maintained at 26.9 ± 0.8°C and 49 ± 4% relative humidity for recovery segmental artery blood velocity measurements and a recovery CPT. Immediately following the CPT, participants left the environmental chamber and assumed a supine position to collect a venous blood sample prior to the start of the carbon monoxide rebreathing test. Lastly, participants provided a recovery nude weight before they were permitted to leave the laboratory.

During the Drink trial, twelve participants completed the 2-hour simulated occupational heat stress, while one subject stopped at 90 minutes due to lightheadedness. During the No Drink trial, nine participants completed the 2-hour simulation. Three participants stopped at 90 minutes due to core temperature reaching 39.5°C (the ethical cutoff), and one participant stopped at 60 minutes due to lightheadedness. When the simulated occupational heat stress was terminated early, post- and recovery measurements were obtained in the same manner as if the entire 2-hour period were completed. One participant was excluded from the CPT analysis due to the inability to tolerate the CPT that rendered ultrasound-based images inaccurate. Two participants could not provide a urine sample at the recovery timepoint in the No Drink trial. Therefore, the number of samples for all parameters from urine samples reported at pre-exposure and post-exposure in both trials is n=13, while recovery samples were n=13 in the Drink trial and n=11 in the No Drink trial.

Prior to statistical analysis, all urinary markers were normalized to urine specific gravity to account for differences in urine concentration (83) and serum biomarkers were corrected to percent changes in plasma volume to account for differences in plasma volume between conditions. Data collected over time during simulated occupational heat stress were analyzed using two-way linear mixed models with repeated measures of trial and time. After visual inspection of predicted and actual residuals (i.e., QQ plots), it was identified that most urinary markers were not normally distributed. Therefore, statistical inference of these variables was completed after Log₁₀ transformation before statistical analyses, which normally distributed the data in all instances. Markers of AKI risk, inflammation, oxidative stress, and gastrointestinal permeability are presented as peak changes from pre-exposure. A two-tailed paired t-test or Wilcoxon test for non-normally distributed data were used to compare the peak changes in these variables between trials. Spearman’s rank correlations, to account for non-parametric data, were used to examine relations between the peak change in [IGFBP7•TIMP-2] and peak changes in urinary and serum markers of oxidative stress, inflammation, and gastrointestinal permeability in both trials. Responses to the CPTs were assessed as the change from minute zero of the CPT and these data were analyzed using independent two-way linear mixed models with repeated measures of trial and CPT time at pre-, post- and recovery timepoints. The peak elevations (heart rate, mean arterial pressure, segmental artery vascular resistance) and nadir (segmental artery blood velocity) responses relative to minute zero during each CPT were calculated and comparisons between trials and over time were made using two-way linear mixed models with repeated measures of trial and time. Two-tailed paired t-tests were used to examine changes in the percent change in nude body weight and sweat rate between trials. In all instances, the Geisser–Greenhouse correction was applied if sphericity could not be assumed. As determined a priori, if a significant interaction or main effect of trial was found, post hoc analyses were completed with Tukey’s multiple comparisons tests, which adjusts for multiple comparisons. All data were analyzed with GraphPad Prism software (version 8; La Jolla, CA). Data are presented as mean ± SD and individual values where possible. For pairwise comparisons, Cohen’s d_z effect size is provided along with the mean difference and 95% confidence intervals (CI). Statistical significance was set a priori at p ≤ 0.05, and actual p-values are reported where possible.

Core and mean skin temperature data are presented in Figure 2. In summary, core temperature increased during simulated occupational heat stress in both trials (Main effect of Time: p < 0.001). The increase in core temperature changed differently over time between trials (Time x Trial: p < 0.001) and was higher in the No Drink trial at 90 min (38.2 ± 0.3°C vs. 38.5 ± 0.4°C; p = 0.042) and 120 min (38.5 ± 0.4°C vs. 38.7 ± 0.4°C; p = 0.023). The rate of rise in core temperature in the No Drink trial (1.0 ± 0.3°C/h) was greater than the Drink trial (0.8 ± 0.2°C/h; mean diff. (upper, lower CI): 0.2 (0.4, 0.0)°C/h; Cohen’s dz: 0.739; p = 0.021). Despite the lack of difference between peak core temperature in the Drink (38.5 ± 0.4°C) and No Drink trial (38.6 ± 0.4°C; mean diff. (upper, lower CI): 0.1 (0.4, −0.1)°C; Cohen’s dz: 0.272; p = 0.346), the change in core temperature matched for the total time between trials was greater in the No Drink trial (1.7 ± 0.4°C vs. 1.5 ± 0.4°C; mean diff. (upper, lower CI): 0.3 (0.4, 0.1)°C; Cohen’s dz: 0.795; p = 0.014). Mean skin temperature increased over time (Main effect of Time: p < 0.001) in both trials and differed between trials (Main effect of Trial: p = 0.037), with pairwise comparisons indicating that mean skin temperature was higher in No Drink throughout the second hour of simulated occupational heat stress (p ≤ 0.020).

Heart rate data during simulated occupational heat stress are presented in Figure 2. To summarize, heart rate increased over time in both trials (Main effect of Time: p < 0.001) and the rise in heart rate was greater in the No Drink trial during most of the second hour (p ≤ 0.049). Similarly, heart rate represented as a percentage of age predicted maximal heart rate increased in both trials over time (Main effect of Time: p < 0.001) and was higher in the No Drink trial during most of the second hour (p ≤ 0.032).

Hydration-related measurements are presented in Table 2. In summary, participants in the No Drink trial lost 1.5 ± 0.5% more body weight compared to the Drink trial (mean diff. (upper, lower CI): −1.5 (−1.2, −1.7)%; Cohen’s d_z: 3.001; p < 0.001) (Table 2). Whole body sweat rate was not different between trials (p = 0.623). During the Drink trial, ad libitum fluid consumption was 1394 ± 316 mL.

In the present study, attenuating the development of dehydration during simulated occupational heat stress did not mitigate AKI risk, defined as elevations in urinary [IGFBP7·TIMP2] (Figure 3). While the change in core temperature was greater in the No Drink trial, participants achieved similar peak core temperatures (Figure 2) and sweat rates (Table 2), which allowed for a focus on fluid intake (i.e., hydration status) and its role in the risk of developing AKI, alongside renal oxidative stress and inflammation (discussed below). Fluid restriction elicited greater dehydration in the No Drink trial as evidenced by a greater loss in nude body mass (Table 2), increases in serum osmolality, and reductions in plasma volume (Table 2). However, we would be remiss not to acknowledge that ad libitum drinking did not completely ameliorate body fluid losses given that participants lost ~1% of their body weight even in the drinking trial (Table 2). Thus, it is possible that had we employed a complete fluid replacement trial we may have observed an attenuated rise in AKI risk during simulated occupational heat stress. It should be noted, however, that providing readily available cool liquids, and not prescribed drinking, is recommended during occupational heat stress (18). Moreover, the observed rate of fluid intake was in a range similar to that observed in agriculture workers during heat stress (0.7 L/h vs. 0.8 L/h (84, 85). Therefore, the present study indicates that while ad libitum fluid intake attenuates the magnitude of dehydration, this was insufficient to attenuate elevations in urinary [IGFBP7·TIMP2]. Such findings suggest that at a given level of hyperthermia, there may be a dehydration threshold after which AKI risk will be magnified, and that this threshold likely exceeds ~2.7% body weight loss (i.e., the mean body weight loss observed in the No Drink trial (Table 2).

Seemingly in contrast to our findings, Juett et al. (86) found that during high intensity intermittent running in a temperate environment, biomarkers of AKI risk were augmented with dehydration, but unfortunately, the authors did not measure core temperature. This is important because Chapman et al. (87) found that independently attenuating the magnitude of hyperthermia (via body cooling) and dehydration (via prescribed drinking) mitigated the rise in AKI biomarkers, but body cooling trial or prescribed drinking resulted in peak core temperatures that were similar between trials (~ +1.2°C). Thus, based on the findings of the present study, it may be that core temperature, rather than hydration status, plays a comparatively more important role in the development of AKI risk. This may be particularly true when the magnitude of hyperthermia exceeds some threshold value, which is likely greater than 38.0°C (9), but has not yet been defined. Moreover, it is likely that the interactive effects of hyperthermia and dehydration on AKI risk differ depending on the magnitude of changes in each of these factors.

Although urinary [IGFBP7·TIMP2], the a priori established biomarker of AKI risk, did not differ between the trials (Figure 3), it is interesting to note that the peak increase in urinary TIMP-2, which is preferentially secreted in the distal tubules (88), was greater in the No Drink trial (Figure 3). This finding is in contrast with previous work that identified no effect of prescribed drinking on the rise in urinary TIMP-2 concentrations during simulated occupational heat stress (7). These seemingly conflicting findings may be explained, at least in part, by the higher average relative work intensity in the present study (75% vs. ~65% (89) of estimated maximum heart rate) and the inclusion of upper body exercise to better simulate the physical demands of occupational tasks. These are notable differences because the cardiovascular response and likely the redistribution of blood flow away from the kidneys is influenced by the relative work intensity (11, 90) and the muscle mass utilized, with upper body work known to provoke a greater cardiovascular response (91, 92). We speculate that with this greater cardiovascular response oxygen delivery to the renal medulla, which anatomically houses portions of the distal tubule, may have been compromised to a greater extent in the No Drink trial due to augmented sympathetic outflow. This is supported by evidence that sympathetic activation induced by lower body negative pressure reduces renal medullary oxygenation in healthy adults (93). While we do not have regional renal blood flow measurements, in the current study we observed no effect of physical work or drinking on segmental artery vascular resistance (Figure 7), a marker of renal vasomotor tone (94–96), nor the segmental artery vascular responsiveness to sympathetic stimulation (i.e., CPT) (Figure 8), a marker of renal vascular control (65, 94, 95). Moreover, while indicators of kidney and tubular function (e.g., creatinine clearance, fractional excretion of sodium, etc.) exhibited the expected changes with simulated occupational heat stress (87, 97, 98), these indicators did not differ between trials (Table 3) and are unlikely to explain the greater increase in urinary TIMP-2 in the No Drink trial. Thus, from the current study, the mechanisms underlying the observed greater increase in urinary TIMP-2 in the No Drink trial remain unclear. Collectively, these findings highlight the need for future studies to clarify the role of hydration across a range of core temperatures and over longer work durations, to better understand its impact on AKI risk during occupational heat stress.

Contrary to our hypothesis, ad libitum fluid intake did not reduce markers of systemic (serum) or renal (urinary) inflammation (Figure 4) or oxidative stress (Figure 5) during simulated occupational heat stress. Serum MCP-1, our primary biomarker of systemic inflammation, was increased following simulated occupational heat stress but was unaffected by fluid intake. This finding is corroborated by previous work indicating that the systemic inflammatory response to heat stress is not modified by hydration in healthy young men (99). Similarly, urinary MCP-1, our primary biomarker of renal inflammation, was elevated following simulated occupational heat stress irrespective of hydration status. This is a novel finding. For example, previous work has demonstrated that the urinary inflammatory marker IL-18 is increased with prolonged passive heat exposure that rendered participants both hyperthermic and dehydrated (8) and that urinary MCP-1 is increased with simulated occupational heat stress when ad libitum drinking was allowed (17). The current study extends these findings to indicate that this increase in renal inflammation is not worsened when fluid intake is restricted. Similar to our study (Figure 7), this previous work also indicated that passive heat stress (8) and simulated occupational heat stress (17) resulted in an increase in urinary TRX-1, but not serum TRX-1, suggesting increases in oxidative stress are localized to the kidneys. Notably, independent of hydration status, peak changes in [IGFBP7•TIMP-2] and urinary MCP-1 and TRX-1 were significantly correlated (Figures 4 & 5), whereas correlations between peak changes in [IGFBP7•TIMP-2] and serum MCP-1 and TRX-1 were not statistically significant. This observation supports previous work during prolonged passive heat stress (8) which suggests a relation between renal inflammation and renal oxidative stress and the development of AKI risk. However, it should be noted that these relationships are only associative and should not be considered causal. As a result, future research is required to identify causal relationships between renal and/or systemic inflammation and oxidative stress and AKI risk in the context of occupational heat stress.

It has been proposed that heat-induced gastrointestinal permeability and the resultant systemic pro-inflammatory state occurring secondary to immune system activation caused by bacterial translocation is a contributor to AKI risk during occupational heat stress (62, 100). In the current study, biomarkers of gut permeability indicating enterocyte damage (I-FABP), but not bacterial translocation (LBP), were found to increase during simulated occupational heat stress but were not modulated by ad libitum fluid intake (Figure 6). This observation is in contrast to previous work demonstrating that minimizing dehydration mitigates the rise in markers of gastrointestinal permeability during exercise in the heat (101). That said, although in this previous study markers of gastrointestinal permeability were attenuated by fluid replacement, the fluid restricted trial resulted in a body mass loss of ~1.5%, which was similar to the body mass lost in the Drink trial in the present study (~1.3%), highlighting the need for further research on the dose-response effects of fluid replacement on markers of gastrointestinal permeability. Notably, while we observed seemingly parallel rises in I-FABP and serum MCP-1, our marker of systemic inflammation, none of these variables significantly correlated with the urinary [IGFBP7•TIMP-2], our primary marker of AKI risk (Figures 4 & 6). Thus, while markers indicative of gastrointestinal permeability, systemic inflammation, and AKI risk all increased independent of fluid intake, there is little evidence to link gastrointestinal permeability and the resulting systemic inflammation to AKI risk. This may be because the magnitude of hyperthermia was relatively homogenous and did not regularly exceed 39.0°C, which has been shown to consistently be associated with increased markers of gastrointestinal permeability (102). That said, further research is required to better discern the potential causal effects of immune system activation during occupational heat stress and its impact on kidney function and injury.

There are a few considerations that warrant attention. First, we would like to note that the actual number of participants with complete data sets (n=11) was less than the a priori power calculation (n=12). This is because two participants were unable to provide a urine sample at the recovery timepoint in the No Drink trial. Thus, it is possible that we were underpowered to observe an effect for the measured urinary markers. As a result, we have provided effect size metrics to aid in data interpretation. While we acknowledge this limitation, given the small effect size for the effect of trial on our primary outcome measurement of AKI risk (urinary [IGFBP7·TIMP2]) we think it is unlikely that one additional observation would have meaningfully changed the conclusions of this study. Second, a strength of this study is that it was a well-controlled study. Limitations of this approach, however, are that the obtained findings may not accurately reflect real-world exposures in which work shifts are 8–12 hours in duration, are repeated for several days, and co-exposure to other environmental factors (e.g., dust, noise, etc.). To highlight, this study involved a single two-hour occupational heat stress simulation. Therefore, it is possible that longer durations or repeated days of physical work with no fluid replacement would have unveiled greater elevations in renal and systemic oxidative stress and inflammation, and AKI risk. However, it is worth noting that previous work utilizing two hours of simulated occupational heat stress showed that prescribed fluid intake was able to avert elevations in AKI risk (7), suggesting that work duration is unlikely the primary explanation of our findings. Rather, our data are likely explained by the magnitude of hyperthermia experienced and the statistically significant, but relatively small, differences in hydration status. This conclusion is supported by recently published field work, in which the magnitude of hyperthermia, not hydration status, was identified as an independent predictor of cross-shift increases in AKI risk (urinary [IGFBP7·TIMP2]) in construction workers (103). Second, it is important to note that the biomarkers of AKI risk measured in this study are subclinical in nature and reflect an increased risk of AKI but do not indicate a clinical diagnosis. To this end, there is no established threshold for a meaningful increase in urinary [IGFBP7·TIMP2] in the context of occupational heat stress. Interestingly, however, all observations in the current study, independent of the experimental trial displayed an increase in urinary [IGFBP7·TIMP2] greater than 0.3 (ng/mL)²/1000 (Figure 3), the threshold used clinically (104). Thus, future research is required to improve the understanding and interpretation of AKI risk biomarkers in the context of occupational heat stress. Moreover, how biomarkers of AKI risk across a single or repeated episodes of occupational heat stress translate to the risk of developing CKDnt remains poorly understood (105). Third, NIOSH recommends that workers arrive at the worksite hydrated and that the worksite have cool fluids readily available (18). These recommendations guided our study design. However, we acknowledge that this may not always be what is encountered in the workplace, where, for example, workers regularly arrive at the worksite dehydrated (18, 79, 106). Lastly, renal blood flow was not directly measured. Instead, Doppler ultrasound was used to measure segmental artery blood velocity, which was interpreted to reflect changes in renal blood flow, as done previously (97). To increase the reliability of the operator-dependent ultrasound measurements and to ensure transparency the intra-operator coefficient of variation of our sonographer was measured and reported to aid in the interpretation of measurements obtained from Doppler ultrasound (95). Nevertheless, future studies should consider incorporating more direct measures of renal blood flow and/or renal plasma flow, particular focus should be on methods (e.g., MRI) that can quantify the regional distribution of blood flow within the kidneys.

Workers frequently experiencing occupational heat stress are at risk for heat-related illnesses (1), including heat-related AKI (1, 5) and CKDnt (1), the latter of which has been hypothesized to arise from a single clinically diagnosed AKI event or repeated subclinical AKI episodes (107). Experimental studies demonstrate that hyperthermia and dehydration are potential modulators of the development of heat-related AKI (7), while workplace based interventions that mitigate the development of hyperthermia and dehydration reduce the prevalence of clinical AKI (7, 106, 108, 109). Interestingly, our study found that attenuating dehydration via ad libitum fluid intake during simulated occupational heat stress, which followed occupational hydration recommendations (81, 110, 111), did not reduce AKI risk. We interpret such findings to demonstrate the relative importance of hyperthermia such that relatively small differences in hydration status have minimal effect of AKI risk during occupational heat stress when a given, not well defined, level of hyperthermia is present. Notably, we have also identified that simulated occupational heat stress increased markers of renal oxidative stress and inflammation, but this effect was not modified by ad libitum drinking. Moreover, while we observed increases in markers of systemic and renal inflammation and oxidative stress, only markers of renal inflammation and oxidative stress were correlated with increases in AKI risk. Such findings suggest that endogenous processes in the kidneys may contribute to the increased risk of AKI during occupational heat stress. Thus, future work should examine the potential causal role of renal inflammation / oxidative stress in the increased AKI risk during occupational heat stress. This information will be vital towards the development of novel strategies to protect the renal health of outdoor workers.

This study demonstrates that preventing dehydration by adhering to current occupational hydration guidelines did not attenuate increases in renal oxidative stress, renal and systemic inflammation, or AKI risk following two hours of simulated occupational heat stress. Additionally, the data show that elevations in AKI risk are associated with markers of renal inflammation and oxidative stress, regardless of fluid intake. To better protect the kidney health of outdoor workers, future research should focus on understanding the causal mechanisms by which occupational heat stress triggers renal oxidative stress and/or inflammation, thereby increasing AKI risk.