Post-exercise glycaemic rebound characterises exploratory glucose-lactate response patterns during incremental rowing in elite male rowers: a cross-sectional study

BACKGROUND: Blood lactate profiling is widely used to monitor metabolic stress in rowing, but lactate alone may not describe how circulating glucose behaves during and after severe exercise. This exploratory cross-sectional study examined whether paired capillary glucose and lactate kinetics describe distinct response patterns in elite male rowers. METHODS: Twenty-six national-level male open-weight rowers completed a discontinuous incremental rowing-ergometer test in a thermostat-controlled laboratory. The first five stages were 4-min submaximal bouts separated by 1-min passive recovery for blood sampling, followed by a free-paced all-out final stage. Capillary glucose and lactate were sampled at rest, during each inter-stage recovery interval, and at 3 and 5 min of recovery. Glucose slope during stages 4-6 (Kglu) and post-exercise glycaemic rebound (recovery 3-min glucose minus end-exercise glucose) were used for z-standardised k-means clustering. Cluster stability was examined using leave-one-out refitting and bootstrap resampling. Linear mixed-effects models tested group-by-stage interactions for glucose and lactate. Lactate-derived intensity proxies were analysed to assess whether the glucose patterns overlapped with traditional lactate markers. RESULTS: Two clusters were obtained (silhouette = 0.42): a Coupled response pattern (n = 17) and a Rebound-dominant response pattern (n = 9). The Rebound-dominant cluster was characterised descriptively by lower Kglu and larger early recovery rebound; these two variables were clustering inputs and are therefore not presented as confirmatory between-group tests in the abstract. Leave-one-out refitting reproduced the original assignment for all 26 rowers, and bootstrap resampling showed median label agreement of 96.2% (mean 91.9%) with the original solution. Because the glucose slope defined the clusters, the mixed-effects model showed the expected group-by-stage glucose interaction (-0.335 mmol/L/stage, P < 0.001), indicating a temporal rather than a baseline difference; the comparisons independent of the clustering - lactate slope, peak lactate, and lactate-derived proxies - did not differ between groups. The Rebound-dominant group had a higher body fat percentage than the Coupled group (14.8 +/- 2.7% vs. 11.8 +/- 3.0%, P = 0.021). CONCLUSIONS: High-intensity glucose slope and early recovery glycaemic rebound described exploratory glucose-response patterns that were not captured by lactate kinetics alone. The practical significance of these patterns remains unknown because individual workload-normalised outcomes, stage-6 duration, heart-rate time courses, gas-exchange data, and direct rowing-performance outcomes were not available in the metabolite-analysis dataset. Mechanistic and applied interpretations should therefore remain hypothesis-generating.