Impact of Nutrient Intake on Hydration Biomarkers Following Exercise and Rehydration Using a Clustering-Based Approach

Abstract

We investigated the impact of nutrient intake on hydration biomarkers in cyclists before and after a 161 km ride, including one hour after a 650 mL water bolus consumed post-ride. To control for multicollinearity, we chose a clustering-based, machine learning statistical approach. Five hydration biomarkers (urine color, urine specific gravity, plasma osmolality, plasma copeptin, and body mass change) were configured as raw- and percent change. Linear regressions were used to test for associations between hydration markers and eight predictor terms derived from 19 nutrients merged into a reduced-dimensionality dataset through serial k-means clustering. Most predictor groups showed significant association with at least one hydration biomarker: (1) Glycemic Load + Carbohydrates + Sodium, (2) Protein + Fat + Zinc, (3) Magnesium + Calcium, (4) Pinitol, (5) Caffeine, (6) Fiber + Betaine, and (7) Water; potassium + three polyols, and mannitol + sorbitol showed no significant associations with any hydration biomarker. All five hydration biomarkers were associated with at least one nutrient predictor in at least one configuration. We conclude that in a real-life scenario, some nutrients may serve as mediators of body water, and urine-specific hydration biomarkers may be more responsive to nutrient intake than measures derived from plasma or body mass.

Influence of Nutrient Intake on 24 Hour Urinary Hydration Biomarkers Using a Clustering-Based Approach

“…Because we are unaware of other any other studies reporting nutrient clustering and the association to hydration biomarkers, we are able to compare our findings only with those reported by Muñoz et al [ 38 ]. Comparing clustering results is made more complex by differences between the study designs.…”

“…To our knowledge, only one other study [ 38 ] has applied a clustering approach to nutrient consumption to examine relationships with hydration biomarkers. However, this other study was performed on trained athletes as they participated in a prolonged (161 km) cycling event, and its results were narrow in scope, and left open questions about study generalizability.…”

“…Recently, a hierarchical clustering model applied to 19 nutrients yielded 9 modules as follows: (1) glycemic load + carbohydrates + sodium, (2) protein + fat + zinc, (3) magnesium + calcium, (4) pinitol, (5) caffeine, (6) fiber + betaine, (7) water, (8) potassium + xylitol + erythritol + inositol, and (9) mannitol + sorbitol [ 38 ]. When determining the aforementioned clusters’ associations with both urinary and hematologic hydration biomarkers following a 161 km road cycling race, Muñoz et al [ 38 ] observed that urinary hydration biomarkers immediately following and 1 h post-race were most impacted by the nutrients consumed by participants leading up to, during, and immediately following the completion of the race. The resulting nutrient clusters and the associations with hematologic and urinary hydration markers emanating from this study is unsurprising given the impetus within the sport nutrition industry to optimize the composition of sport-related nutrition products for health and performance.…”

“…By conducting a cluster-based analysis, Muñoz et al, were able to better account for the collinearity between various nutrients that are often consumed concomitantly. By consolidating the nutrients into their representative clusters, the regression models derived by Muñoz et al [ 38 ], were properly weighted to reflect consumption patterns. Without this clustering, nutrients consumed commensurately would draw greater share of the total model variance, leading to disproportionate “leverage” within the model.…”

“…While the Muñoz et al [ 38 ] study provides a novel approach in beginning to develop an understanding how various nutrients cluster together and the resulting clusters’ association with hematologic and urinary hydration biomarkers, there are remaining gaps in the literature. First, the aforementioned study only assessed participants competing in an ultra-endurance road cycling event in hot environmental conditions, thus, the findings are limited in generalizability when compared to a broader sample population.…”