The present investigation was conducted in 2023, a year in which 379 runners (296 male/83 female, ~22% female) began the race, of which 328 (260 male/68 female, ~21% female) finished the race (86.5% finish rate) [33]. Temperature ranged from 10 to 27°C and relative humidity ranged from 30-75%. All athletes were provided written informed consent before participation and all procedures involving human research participants were approved by the Institutional Review Board at Georgia Southern University (H23253).

Participants were recruited up to six months prior to the 2023 WSER, all of which completed virtual questionnaires (training, nutrition, race preparation details) before any testing. Data included in this investigation was collected up to 72 hours before the beginning of the event (pre), and within 1 hour (30 ± 14 minutes, max = 59 minutes) of individuals finishing the race (post). At each time point, renal blood velocity was assessed as an applied measure of kidney perfusion, and spot urine and capillary blood samples were collected.

Participants were provided sealed, sterilized urine collection cups at the pre- and post-race data collections. From the provided urine samples, we measured USG (ATC 3-in-1 Scale Clinical Refractometer) at our field laboratory and stored aliquots of urine in cryogenic tubes for future analyses on dry ice prior to longer term storage at −80°C [38, 39]. While freezing can influence stability of certain biomarkers [38, 39], frozen samples were thawed simultaneously to minimize variability and subsequently analyzed at Auburn University School of Kinesiology for electrolyte concentrations (sodium, potassium, chloride) using a Smartlyte Easylyte Electrolyte Analyzer and osmolality using an Advanced Instruments Advanced 3D3 Osmometer model 3250 [40-43]. Notably, USG and urine osmolality are well accepted biomarkers of hydration status for athletes [35, 44]. Changes in urine electrolyte concentrations and ratios (e.g., sodium: potassium) have also been reported after endurance exercise events [45, 46]. As we have previously described [40-43], we measured all analytes in triplicate, and using these methods has resulted in a <1% intra-sample coefficient for urine electrolytes and urine osmolality [41].

Capillary blood was collected using Touch-Activated Phlebotomy device (TAP Micro, YourBio Health, Medford, MA). Blood was centrifuged at 735 x g for 10 minutes to separate serum, which was aliquoted and stored on dry ice prior to storage at −80°C. Serum CK was measured with a Roche Integra 400+ analyzer at the Indiana University School of Medicine Center for Diabetes and Medical Disease’s Translational Core Laboratory. We also measured serum creatinine as a proxy for glomerular function via the University of Alabama at Birmingham Biochemical Genetics Laboratory. Serum and urine creatinine samples were measured using liquid chromatography-mass spectrometry (LC-MS). Briefly, LC-MS was performed using tandem mass spectrometry using multiple reaction monitoring and quantified via stable isotope dilution [47]. For LC-MS experiments, 20 μL of the sample was deproteinated and diluted with a stable isotope in a single step. The diluted samples were subjected to isocratic HPLC with 10 mM ammonium acetate in 65% acetonitrile. Creatinine and d₃-creatinine were detected by electrospray ionization tandem mass spectrometry. For general kidney function we also measured pre- and post-race urine cystatin C as measures of glomerular filtration [43, 48].

Urine samples were also shipped to the Indiana University School of Medicine Center for Diabetes and Medical Disease’s Translational Core Laboratory for multiplex analysis to assess kidney function. Multiplex assays were performed in duplicate using Meso Scale Discovery assay kits (Meso Scale Diagnostics, Rockville, MD) according to the manufacturer's instructions. The assays were selected based on the specific analytes of interest, ensuring compatibility with the matrix being tested. The assay plates were pre-coated with capture antibodies specific to the target analytes. Regarding our selection of specific analytes, nephrin is a protein expressed in the glomerulus and is responsible for maintaining the glomerular barrier, and the excretion of nephrin being associated with glomerular damage [49]. As for biomarkers related to the renal tubules, neutrophil gelatinase-associated lipocalin (NGAL) is a protein expressed in the thick ascending loop of Henle and the collecting ducts and is detectable up to three hours after toxic kidney injury [50]. Insulin like growth factor binding protein 7 (IGFBP7) and tissue inhibitor of metalloproteinase 2 (TIMP2) induce cell cycle arrest [51]. While it remains unclear precisely where IGFBP7 and TIMP2 are secreted and what their specific function is surrounding kidney injury, the product of the two urinary biomarkers (IGFBP7●TIMP2) is a sensitive biomarker for the prediction of AKI [27]. Indeed, urinary IGFBP7●TIMP2 is the only United States Food and Drug Administration-approved biomarker for the clinical screening of AKI risk (proprietary name NephroCheck, Astute Medical, San Diego, CA) [27]. We chose this array of analytes based on our experience [40, 43, 50-54], the Indiana University Translation Core’s experience measuring them, their relatively high expression along the nephron such as the proximal tubule [28, 55] and ascending loop of Henle [56], as well as their documented effectiveness at representing sub-clinical kidney injury [28, 55-57].

Similar to our prior work, we indexed all urinary biomarkers of AKI to urine creatinine concentration [40, 43], which was an important methodological consideration since we collected spot urine samples. We indexed the AKI biomarkers to account for fluctuations in urine concentration and ensure more accurate comparisons across samples [58-61]. In clinical populations elevated spot urine creatinine alone has utility in identifying kidney injury-related risk of all-cause mortality [62] and hospitalization [63]. We anticipated indexing samples to creatinine would attenuate the magnitude of pre- to post-race changes [14]. However, we suspected elevations would persist, a finding supported by recent dehydration and endurance race studies [18, 27]. Nonetheless, urinary creatinine may be affected by prolonged exercise where it could be a function of changes in glomerular filtration rate or muscle breakdown. Therefore, we also indexed AKI biomarkers to urine osmolality and urinary cystatin C.

Renal and segmental artery blood velocity (RBV) were collected using doppler ultrasound (GE Logiq e) with a 3-5 MHz phased array transducer [65, 66]. Participants were positioned in the decubitus position on a massage table during each collection (pre- and post-race) and familiarized with a brief 10-second breath holding protocol to minimize the movement of the kidneys [65, 66]. The coronal approach was used to visualize both renal and segmental arteries. Color flow was used to confirm the internal structure of the renal vasculature. Once the renal vasculature was confirmed, each measure lasted between three and five cardiac cycles. Measures were collected in triplicate for both renal and segmental arteries for both timepoints. Semi-automated waveform tracing (GE software) within the ultrasound was used to analyze individual mean RBV and we report the average of replicate measures here.

Regarding a technical limitation, the diameter of the vessel walls of the renal vasculature cannot be accurately measured due to the resolution of Doppler imaging with the relatively deep kidney tissue depth. Using measures such as MRI for a field study would not be feasible therefore, we used Doppler ultrasound to quantify blood velocity [65]. The use of Doppler ultrasound to quantify hemodynamics in the renal and segmental arteries operates under the assumption that vessel diameter does not change during physiologically induced changes in blood flow. This assumption is based on evidence that, during infusion of a vasoconstrictor (adenosine) and vasodilators (isosorbide dinitrate, papaverine, dopamine, and fenoldopam), changes in renal blood flow are dependent on changes in blood velocity rather than changes in the diameter of the renal artery [67, 68]. Thus, changes in Doppler ultrasound-derived RBV are often interpreted as changes in renal blood flow [27, 66, 69, 70].

Additional hemodynamic measures were obtained from this study, which we have reported on previously [33, 71]. We used estimated aortic blood pressure to calculate renal vascular conductance (velocity ÷ pressure), as this more closely represents the pressure within the renal arteries, as opposed to brachial pressures. During pre- and post-race, aortic blood pressure was estimated via pulse wave analysis (SphygmoCor XCEL, AtCor Medical, Naperville, USA). Briefly, brachial artery pressure waveforms were transformed using a validated generalized transfer function to estimate an aortic pressure waveform and derive aortic blood pressure [72, 73].

All data were assessed for normality using a Shapiro-Wilks test. If data were normally distributed, paired samples t tests were used to determine differences between pre- and post- race measures; If data were not normally distributed, Wilcoxon ranked test was used. We used Cohen’s d to assess the effect size for normally distributed data [74] and interpreted effects as small (d=0.2-0.49), medium (d=0.5-0.79), and large (d > 0.8). We used the rank-biserial correlation (r_rb) to assess effect size for non-normally distributed data [75], interpreted as small (0-0.19), medium (0.2-0.29), large (0.3-0.39), and very large (0.4-1.0). Creatinine-indexed AKI biomarker change scores were calculated to correlate the relative change in AKI biomarkers, RBV and performance measures as [%Δ=((post−pre)∕pre)∗100] and were only calculated in participants with complete AKI and RBV data. Spearman’s rho was used for all correlations as neither pace, nor any biomarkers were normally distributed. Finally, participants were split based on hydration status at the end of the race (Euhydrated: USG ≤ 1.020, Hypohydrated: USG > 1.020) [35, 44] to assess whether finishing euhydrated attenuated the effect of the race on elevating urinary AKI biomarker concentrations. Since residuals for AKI biomarkers were not normally distributed, data were natural log-transformed (l_n) before running a mixed model ANOVA (Hydration status * Timepoint), however we visually depict the raw data. We used Fisher’s exact test to compare proportions between groups. Since this was a field study we decided to not screen for outliers unless we observed clear visual outliers (~3SD) outside of the range of the data distribution. Specifically, given the exploratory, field-based nature of the study, we did not conduct a formal power analysis [76], and due to the small sample size we used a deliberately lenient outlier criterion to preserve physiological variability and avoid introducing bias through arbitrary data exclusion, particularly female athletes who are historically underrepresented in this research area [77]. There were two data points that were removed with this criterion (one from NGAL/creatinine-pre and one from IGFBP7●TIMP2/creatinine-pre; Supplemental Figure 1). Post hoc pairwise comparisons with Bonferroni adjustment were completed where appropriate. Statistics were run in jamovi (v2.6.19) and figures were generated in Graphpad Prism (v10.4.1). Significance was set a priori to p ≤ 0.05. All data are presented as mean ± standard deviation for normally distributed data or median (interquartile range) for non-normally distributed data in the tables or as individual data points superimposed on box and whisker plots in Figures. Interquartile ranges are reported as single numerical values (Q3 – Q1) to reflect variability from pre to post race.

A schematic of participant recruitment and enrollment is presented as Figure 1. Thirty-six participants were included in the analyses investigating the effects of completing the WSER on biomarkers of AKI. Twenty-nine (29) of those 36 participants completed RBV for at least one artery and were included in investigating the effects of completing the WSER on RBV. Thirteen participants of the 29 who completed RBV for at least one artery, had complete renal (Paired; n=12) or complete segmental (Paired; n=12) artery velocity data and were included in exploratory correlations between RBV and biomarkers of AKI as both pre and post measures are needed to calculate change scores. Participant demographics for both subsets of participants can be found in Table 1.

Due the small number of female participants and unequal sample sizes (F=7, M=29), we did not separate participants by biological sex but we depict them with different symbols in figures. Additionally, we did complete a sex comparison for pre-race hydration given a prior study demonstrating sex differences in the change in hydration during an endurance cycling event [44]. Nonetheless, we observed no baseline sex differences in USG (male participants = 1.010 ± 0.006, female participants = 1.011 ± 0.008, p = 0.741) or urine osmolality (male participants = 460 ± 245, female participants = 421 ± 268 mOsm/kg, p = 0.667). Furthermore, we also observed a decrease in aortic mean blood pressure (p = 0.016, d = 0.423, pre: 93 ± 8 mmHg, post: 89 ± 9 mmHg) and an increase in serum CK concentration (p < 0.001, d = 1.582, pre: 132 ± 85, post: 13,373 ± 8414 U/L) in all the runners, similar to our previous findings [33].

Upon completion of the 2023 WSER, USG (Figure 2A; p < 0.001) and osmolality (Figure 2B p < 0.001) were increased. Using USG to classify hypohydration status (USG > 1.020), only 11% of participants (4/35) were hypohydrated at their pre-race collections whereas 77% were hypohydrated (27/35) upon completion Figure 2C (Fishers exact test p < 0.001). Race pace was positively correlated with changes in both USG (Figure 2D ρ = 0.34, p = 0.048) and urine osmolality (Figure 2E; ρ = 0.53, p = 0.002). The larger range in percent change for urine osmolality compared to USG reflects differences in scale. For example, an increase in urine specific gravity from 1.020 to 1.030 represents only a ~1% change, whereas an increase in urine osmolality from 200 to 800 mOsm/kg reflects a 300% change. A more extreme shift from 150 to 950 mOsm/kg represents a 533% increase. These differences explain the broader y-axis range for osmolality in Figures 2D & 2E. Of note, one participant had missing USG data that was not realized until all urine had been used for other measures, hence n = 35 as opposed to n = 36.

Comparisons between pre- and post-race serum creatinine, urine electrolyte concentrations, and raw urinary AKI biomarker concentrations are reported in Table 2. We observed pre- to post-race increases in urine potassium concentration (p < 0.001, r_rb = 0.84) and all raw urinary AKI biomarker concentrations (ps < 0.001); and decreases in urine sodium (p < 0.001, r_rb = 0.75) and urine chloride concentration (p < 0.001, r_rb = 0.97). As presented in Figure 3A-F, we also observed significant increases in urine creatinine concentration and all urine AKI biomarkers when indexed to urine creatinine (ps < 0.008), except for creatinine-indexed NGAL (p = 0.334) and TIMP2 (p = 0.287). Supplemental Figure 1 depicts the data with outliers included. The only change in interpretation is the interaction term for our mixed model ANOVA for time * hydration status on IGFBP7●TIMP2 (before outlier removal p = 0.104; after outlier is removed p = 0.003). Respective qq-plots have been included in Supplemental Figure 1 to demonstrate why these outliers were removed with the data points highlighted by a crossed-out box. In addition to the creatinine-indexed data, we indexed all AKI biomarkers to osmolality and cystatin C as presented in Table 4. These findings were generally consistent with the data presented in Figure 3. When indexed to osmolality, all AKI biomarkers were elevated from pre- to post-race except NGAL. When indexed to cystatin C, all AKI biomarkers were elevated from pre- to post-race except IGFBP7.

Comparisons between euhydrated (n = 8) and hypohydrated (n = 27) finishers are presented in Figure 4. Urinary creatinine (Figure 4A; p = 0.006) and creatinine-indexed IGFBP7•TIMP2 (Figure 4F; p = 0.003), exhibited significant time * hydration status interaction terms, but not creatinine-indexed nephrin, NGAL, IGFBP7, or TIMP2 (ps > 0.136). Regarding the creatinine and IGFBP7●TIMP2/creatinine (biomarkers demonstrating time x hydration status interactions), we also observed pre- vs. post-race pairwise differences in within the hypohydrated finishers’ group only (Figures 4A and F). Additionally, hypohydrated finishers exhibited higher post-race urinary creatinine and IGFBP7●TIMP2/creatinine than euhydrated finishers (Figures 4A and F). We observed no baseline differences between euhydrated and hypohydrated finishers in any AKI biomarker. Interestingly, we noted that all euhydrated finishers were males (0/7 females euhydrated vs. 8/28 males; Fishers exact test p = 0.166).

As presented in Figure 5, we did not detect any pre- vs. post-race race differences in renal artery blood velocity (Figure 5A; p = 0.186, d = 0.41), renal artery conductance (Figure 5B; p = 0.339, r_rb = 0.33), segmental artery blood velocity (Figure 5D; p = 0.483, d = 0.21), nor segmental artery conductance (Figure 5E; p = 0.622, r_rb = 0.18). Representative images of the renal ultrasound assessment of RBV are presented in Figure 5C and 5F.

We present simple bivariate correlations in our variables of interest, primarily correlations between race pace (km/hr) and pre- vs. post-race race differences in urinary AKI biomarkers, race pace and pre- vs. post-race race differences in hydration, and pre- vs. post-race race differences in hydration and urinary AKI biomarkers, in Table 4. We observed correlations between pace and changes in serum CK (ρ = −0.596, p = 0.028), urine sodium (ρ = 0.543, p = 0.004), urine potassium (ρ = 0.341, p = 0.049), urine chloride (ρ = 0.128, p = 0.468), and urine creatinine concentration (ρ = 0.433, p = 0.010). We observed no correlations between pace and changes in renal artery blood velocity (ρ = −0.109, p = 0.755) or conductance (ρ = −0.027, p = 0.946) or changes in segmental artery blood velocity (ρ = 0.373, p = 0.261) or conductance (ρ = 0.382, p = 0.248). There were no correlations between changes in any AKI biomarker and changes in renal artery velocity (ps > 0.297), renal artery conductance (ps > 0.366), segmental artery velocity (ps > 0.067), or segmental artery conductance (ps > 0.075). Additionally, changes in serum CK were negatively associated with changes in urine osmolality (ρ = −0.571, p = 0.045), but there were no correlations between serum CK and any urinary AKI biomarker (ps > 0.202). We were unable to correlate serum CK and measures of RBV due to insufficient complete data across both measures.

We would like to acknowledge the limitations of the present study. First, these data were collected in the field. Specifically, pre (~1890m) and post (~385m) measures were taken at different altitudes due to the spaces provided to collect data near the start and finish lines. Similarly, WSER 2023 was a relatively temperate year and resulted in the highest completion rate on record (per WSER website). Therefore, we hesitate to use our data to compare to other years (e.g., years where temperatures may have been substantially more elevated for the duration of the event) or other ultramarathons where environmental demands can vary widely. Additionally, most participants dispersed very shortly after the race, hence we could not collect recovery samples to determine the timeline of resolution for the increase in AKI biomarkers. In future iterations of the WSER, researchers should attempt to have participants provide a post-race sample perhaps 24 or 48 hours after the race and arrange for shipping to complete these measures as presentation of emergency symptoms may range between 17 and 32 hours post-race [37]. Lastly, it is possible that the timing of the last meal prior to the urine sampling could have influenced urine concentration of certain biomarkers such as osmolality and creatinine which are commonly used to assess urine concentration, potentially creating an artifact suggestive of hypohydration despite overall euhydration [83, 84]. We asked participants to refrain from eating for at least 3 hours before pre-race data collection; however, this was not strictly enforced, and adherence likely varied. Additionally, we did not control meal timing for the post-race measures.

The field study nature of the investigation also affected our RBV measures. We collected these data as close to immediately after the race as we could. Some participants finished the race in the evening and in the middle of the night, which we consequently collected RBV data at potentially near opposite times of day resulting in inconsistent light exposure, temperature, control of external stimuli, etc. between collections. Additionally, participant fatigue and shivering made obtaining post-race recordings of the renal and segmental artery difficult, hence the smaller paired sample size for these measures. Finally, to more comprehensively understand the physiological impact of these events, it is important to examine organ crosstalk. For instance, exercise is associated with both acute [85-87] and chronic [88] changes to the gut microbiome, which produces metabolites that may contribute to kidney injury (e.g., indoxyl sulfate, p-Cresyl sulfate). Future studies should examine these interactions to provide a more holistic understanding of the effects of ultramarathon on inter-related organ systems.