The Journal of Physiology | 2019
Reply from Chiel Poffé, Monique Ramaekers, Ruud Van Thienen and Peter Hespel
Abstract
In our recent article (Poffé et al. 2019) published in this journal, we demonstrated that ketone ester (KE) intake during a 3-week endurance training overload period blunts the development of overreaching symptoms as well as enhances endurance exercise performance. The criticisms by Korevaar et al. (2019) mainly relate to (i) the media coverage of the data, (ii) the presumed absence of predefined primary endpoints, and (iii) excess focus on positive outcomes, termed ‘cherry picking’. We fully agree with Korevaar et al. that the lay public would benefit from a more balanced presentation of the results in the media. However, the arguments with regard to the lack of predefined primary outcomes and ‘cherry picking’ are fully speculative. The major contribution of public funding to scientific research (Van Dalen et al. 2014) implies that media coverage is important in translating scientific output to the lay public. Some media channels provided a very well-balanced and nuanced discussion of our findings (Hutchinson, 2019; Van Teeffelen, 2019), but others did not. This resulted in invalid statements like ‘miracle drink’ or ‘superfuel’, as indicated by Korevaar et al. However, we are not responsible for this wording and obviously do not support such inappropriate terminology. Therefore, in this reply we confine our response to scientific facts and figures. Korevaar et al. correctly state that an a priori definition of primary output variables, with corresponding power calculation, is essential to avoid selective reporting in clinical trials. Good point, indeed. However, their criticism that both aspects were absent in our trial is incorrect. The primary outcomes and a power calculation were included in the study file which was approved by the Ethics Committee Research UZ/KU Leuven (registration number: B322201733747). Moreover, the Ethics Committee Research UZ/KU Leuven procedures comply with national and international clinical trials regulations, including the Declaration of Helsinki (for detailed information: https://www.uzleuven.be/nl/ethische-com missie/onderzoek/wet-en-regelgeving-voor -studies). Therefore, the suggestion by Korevaar et al. that we falsely reported to comply with the Declaration of Helsinki is irrelevant. But we agree that registration of the trial in an open access database would facilitate such verification, indeed. However, registration in a public database actually is only routinely done for clinical trials involving medicinal products (https://www. clinicaltrialsregister.eu/; https://databank klinischeproeven.be/en). Furthermore, the definition of primary endpoints in ‘classical’ clinical trials is often easy and straightforward. For instance, blood pressure is the logical primary output variable in studies addressing antihypertensive agents. However, in trials exploring novel fields, in which limited a priori consensus exists about clear-cut endpoints, such as in overreaching/overtraining (Urhausen & Kindermann, 2002), predefinition of primary endpoints is more complex. Nevertheless, we defined training load to be the primary endpoint of our study. We aimed to assess whether stimulation of anabolic signalling and muscle glycogen repletion by KE ingestion post-exercise (Holdsworth et al. 2017; Vandoorne et al. 2017) could attenuate training overload-induced fatigue, and thereby facilitate toleration of a higher training volume. Therefore, we imposed an expectedly ‘supramaximal’ training load to induce failure in adhering to the prescribed training programme. This resulted in higher training loads in the subjects on ketone ester supplements during the days before the performance tests (e.g. 30 min simulated time-trial (TT30min) and 90 s all-out sprint (90S)). Nonetheless, compared with baseline, subjects on KE, but not subjects on placebo, performed better in TT30min. Our data showed that training load in the final week of the training program was 15% higher in KE than in placebo. This effect has been excessively highlighted in the media. But clearly, this effect in healthy volunteers cannot be extrapolated to well-trained athletes, because this would imply that in some endurance events, female elite athletes on KE could probably outperform male athletes without KE, which makes no sense. But one should also recognize that in elite athletes even an ergogenic effect of <1% may be crucial (Paton & Hopkins, 2006). Furthermore, training performance was the pre-defined primary endpoint of our study. But post hoc, the effects of KE on the physiological parameters reflecting overreaching, such as maximal heart rate, urinary catecholamine excretion, hormonal food intake regulation and training-induced increase in growth differentiation factor 15 (GDF15), are far more interesting and conceivably will trigger further research in this promising field. Korevaar et al. also claim that we created ‘spin’ by adopting a ‘cherry picking’ strategy in the data presentation. They substantiate such a conclusion by indicating that confidence intervals around the 15% difference in training load and 120 min endurance performance test (EPT120min) ‘would show the uncertainty of this finding’. This statement is bizarre, not to say inappropriate, given that 95% confidence intervals (CI) and effect sizes (ES) are clearly reported in the results section (training load: CI = +237 to +2372 kJ, ES = 0.80; EPT120min: CI = +5 to +52 W, ES = 0.77) and in fact further underscore the relevance of the findings. Similarly, Korevaar et al. criticize the absence of adjustment for multiple statistical comparisons. But the statistics section in the manuscript clearly mentions that adjustment for multiple comparisons was done, indeed. Finally, the accusation of ‘cherry picking’ is entirely false. Positive outcomes were presented most transparently by showing individual data points as well as effect sizes and confidence intervals. In addition, also all negative observations were reported. Even in the abstract the authors report seven parameters that were unaffected by KE. From our perspective, Korevaar et al. highlight some valuable hallmarks in the context of good clinical practice. However, their criticisms regarding the output of our clinical trial are merely based on speculation and can be entirely rejected by facts and figures. The data shown in the article justify the valid conclusions of the study, and will hopefully stimulate further research to elucidate the role of ketone bodies in physiological regulation at rest and during exercise.