Since the formal back and forth in journals requires more patience than I often have, I decided to put a bit of commentary together here. The following includes the response from Jacob Wilson to our letter published in Nutrition & Metabolism. The response from Jacob Wilson is written in plain text and my responses are accented with a gray background (you know, because they are important ;). If you have already read the response, feel free to skip on to my comments. Enjoy!

Response to discrepancies in publications related to HMB-FA and ATP supplementation

by Jacob M. Wilson, Nutrition & Metabolism 2017 14:42. https://doi.org/10.1186/s12986-017-0201-7

Recently, Gentles and Phillips [1] wrote a communication requesting further explanations regarding our recent publications on ATP [2], HMB-FA [3] and co-ingestion of ATP and HMB-FA [4] on training adaptations, resulting from the same study with the clinical trial identifier: NCT01508338. We acknowledge that the authors have invested a great deal of time following our work on multiple platforms, and our reexamination only serves to further our understanding of the significance of these studies. In their effort to better understand our research, Gentles and Phillips [1] have submitted questions pertaining to the homogeneity of subjects’ characteristics between the three different published papers, and why there appears to be differences in the number of subjects in placebo groups across studies [2, 3, 4]. Our response addresses each of these issues and demonstrates that there are no discrepancies between papers but rather a misunderstanding of the papers previously published.

Gentles and Phillips [1] question the homogeneity between papers, providing a table showing that subject characteristics for Wilson et al. [2] and Wilson et al. [3] are even identical. The table by Gentles and Phillips is incorrect. The correct participant characteristics in Wilson et al. [2] are as follows: age 23.4 ± 0.7 with 1-RM of 1.71 ± 0.04, 1.34 ± 0.03 and 2.05 ± 0.04 times body weight for squat, bench press, and deadlift, respectively. We are uncertain as to where the numbers presented by Gentles and Phillips came from in their Table 1. We have provided the corrected Table 1 below. The values in the corrected table add a degree of heterogeneity which may reasonably be expected, but are still quite homogenous. This was accomplished by the randomization procedure. We arranged the subjects by strength, and then performed the randomization.

Table 1: Sample means and standard deviations

Wilson et al. according to [1] Entire group (n = 21)

Actual Wilson et al. [2] Entire group (n = 21)

Wilson et al. [3] Entire group (n = 20)

Lowery et al. [4] Entire group (n = 17)

Age (years)

21.6 ± 0.5

23.4 ± 0.7

21.6 ± 0.5

21.7 ± 0.4

1RM squata

1.7 ± 0.04

1.71 ± 0.04

1.7 ± 0.04

1.7 ± 0.07

1RM bench pressa

1.3 ± 0.04

1.34 ± 0.03

1.3 ± 0.04

1.3 ± 0.05

1RM deadlifta

2.0 ± 0.05

2.05 ± 0.04

2.0 ± 0.05

2.0 ± 0.06

aMean 1RM values are expressed relative to body mass

w

Wilson correctly points out the data in the first column of our table is incorrect. That was probably a copy and paste error on my part. However, that mistake does NOT invalidate our concerns about the homogeneous nature of the data between three papers. Authors “strive” to match subjects as best they can, but to be honest, it would seem quite difficult to recruite a group of subjects with this level of homogeneity; and they were able to do it three times! To illustrate this point, I have used the reported means and standard deviations for allometrically scaled 1RM squat, bench press and deadlift presented in column 2 above. In order to achieve an allometrically scaled 1RM squat, bench press and deadlift that corresponds to the means and standard deviations presented above, subject data would have to look something like the following:

Scaled 1RM squat (1.71 ± 0.04): 1.67, 1.76, 1.72, 1.67, 1.71, 1.73, 1.65, 1.70, 1.68, 1.78, 1.77, 1.77, 1.69, 1.72, 1.76, 1.71, 1.71, 1.67, 1.69, 1.66, 1.75

Scaled 1RM bench press (1.34 ± 0.03): 1.30, 1.36, 1.35, 1.41, 1.34, 1.25, 1.36, 1.41, 1.31, 1.32, 1.31, 1.35, 1.37, 1.36, 1.30, 1.35, 1.36, 1.34, 1.37, 1.35, 1.35.

Scaled 1RM deadlift (2.05 ± 0.04): 1.98, 2.02, 2.02, 2.04, 2.06, 2.08, 1.97, 2.04, 2.08, 2.06, 2.00, 2.04, 2.05, 2.06, 2.05, 2.06, 2.06, 2.02, 2.05, 2.07, 2.03

Pretty impressive! Even more impressive when you consider that while these subjects were “resistance trained”, nothing is known about the type of training they have been engaged in. Over the years I have worked with untrained and trained individuals as well as teams and athletes at a variety levels. I have never seen such a homogeneous group in terms of strength in my life; not even in weight class sports such as weightlifting and wrestling where I have known the training history and current training program of the athletes. But again, they were not able to find this level of homogeneity in just a single paper, but in all three papers. Who knows, maybe Wilson and company had a cloning lab and they just “forgot” to mention in it in their methods.

 

Gentles and Phillips also question the supplement protocol for the placebo group, allowing for the use of the same control group for all three publications from the same study. Subjects received three times per day a gel pack consisting of 1 g of HMB-FA, or a matching placebo gel pack; and once per day a capsule either containing 400 mg per day of Disodium ATP or a matching placebo. Subjects consumed a gel pack and the capsule 30 min before exercise as well as 1 gel pack before lunch and 1 before dinner. On the non-training days, participants consumed 1 gel pack and the capsule 30 min before breakfast and 1 additional gel pack with lunch and dinner. The daily supplementation protocols, 3 gel packs and 1 capsule, were identical in each group. The HMB-FA + ATP publication [4] describes the full supplementation protocol. For clarity, the ATP [2] and HMB-FA [3] publications describe the administered supplements relevant for the data being presented. The information for the ATP group provided in the clinical trial registration is updated and reflects the correct supplement protocol.

w

In other words, oops, our placebo groups don’t line up so let’s see how we can retrospectively redesign the study so the pieces appear to fit together. More on this later.

 

Lastly, Gentles and Phillips would like further clarification regarding the differences in the number of subjects in placebo groups across studies, 12 recruited, 3 drop-outs (2 due to injury, 1 due to time constraints), 9 completed for HMB-FA [3] and HMB-FA + ATP [4], and the 10 placebo-supplemented subjects completing the ATP study [2]. The ATP trial was completed in a later semester and thus, two additional subjects were recruited to the placebo group to blind the treatments. One of the additional placebo subjects was lost due to injury. Due to the cost and high intensity nature of the protocol, we felt it ethically correct to utilize placebo subjects from the original HMB-FA and HMB-FA + ATP cohort in addition to the ATP-supplemented subjects and the placebo subject added. The total number of subjects recruited for the ATP study was twenty-six, 14 in the placebo (4 total drop-outs, 3 due to injury, 1 due to time constraints) and 12 in the ATP group (1 drop-out due to injury).

w

This section is going to take a bit to unpack but initially I will focus on two sentences that are really responsible for the “Discrepancies about Discrepancies” title. In Wilson’s response he states the following.

“The ATP trial was completed in a later semester and thus, two additional subjects were recruited to the placebo group to blind the treatments. One of the additional placebo subjects was lost due to injury.”

There are several major problems with this statement.

First, consider what the above quote implies. This means the control group from the first study, which compared control vs. HMB vs. HMB + ATP cohorts, was used again in the study which occurred during a later semester that compared control vs. ATP cohorts. To reiterate, they used the same control group twice. If that does not sound odd enough, this issue is more troubling when you consider the fact that all three papers describe these studies as double-blinded.

So Jacob Wilson, I have two questions for you.

  1. If these studies were blinded, how would you have known which subjects were in the control group from the first study to reenlist them as controls during the second study performed in a later semester?
  2. If blinded, how would you know to add two more control group participants to replace the two that dropped out? Thanks to Stuart Phillips and Douglas Kalman for raising this point.

If these studies were double-blind as suggested in all three papers, there would be no way reenlist the same subjects as part of control vs. ATP study that was performed in a later semester.

Second, if you look at the study timeline on ClinicalTrials.gov, the start date was January 2012 with a completion date of October 2012. The statement by Wilson above suggests that the control group had to complete this study twice but did so during separate semesters. With this timeline the control subjects would have had to complete this 12 week (actually 14 weeks) protocol during the Spring and Summer semesters in order for Wilson’s statement above to be true. Did this control group lose all the strength and lean tissue they had just gained during the first time completing this protocol that could have only been a month or two apart? How could the control group be matched for strength and other factors if they had just made gains from completing the previous 12-week protocol?

Third, the authors have edited the information on ClinicalTrials.gov to match what he/they have said in their response to us.

Last but not least, I have email confirmation from Jacob Wilson himself stating these three papers were part of a single study that occurred simultaneously. You can download and view these emails here. Here is the statement from Jacob Wilson in his email to me which you can find on page 2 of the email linked above.

“In short yes we did all these studies at once. It was one large study divided into 3 papers which we mention in the HMB-ATP paper. So we had a control, HMB and HMB+ ATP group. For all 3 papers half of the subjects would be identical.”

So, this “change” in methods was not communicated in any of the studies, it was not mentioned in the ClinicalTrials.gov description of study methods (until they recently edited them), and Wilson said the exact opposite to me via email. What version of the methods are we to believe? The different versions presented in the 3 papers, the version originally presented on ClinicalTrials.gov, the version currently listed in this response, or the version presented in Wilson’s email to me?

 

We hope this serves to clear up any misinterpretation in our studies.

Sincerely,
Jacob Wilson, PhD
Corresponding Author

w

No Mr. Wilson, this has not served to clear up any misinterpretations in your study or studies (who knows what is correct). But I am sure you are well aware of that. You have simply dug a deeper hole for yourself and I am happy to watch you keep on digging.

 

References

  1. Gentles JA, Phillips SM. Discrepancies in publications related to HMB-FA and ATP supplementation. Nutr Metab. 2017;14:42. View Article Google Scholar
  2. Wilson JM, Joy JM, Lowery RP, Roberts MD, Lockwood CM, Manninen AH, et al. Effects of oral adenosine-5′-triphosphate supplementation on athletic performance, skeletal muscle hypertrophy and recovery in resistance-trained men. Nutr Metab. 2013;10:57. View Article Google Scholar
  3. Wilson JM, Lowery RP, Joy JM, Andersen JC, Wilson SMC, Stout JR, et al. The effects of 12 weeks of beta-hydroxy-beta-methylbutyrate free acid supplementation on muscle mass, strength, and power in resistance-trained individuals: a randomized, double-blind, placebo-controlled study. Eur J Appl Physiol. 2014;114:1217–27. View Article PubMed PubMed Central Google Scholar
  4. Lowery RP, Joy JM, Rathmacher JA, Baier SM, Fuller JC Jr, Shelley MC 2nd, et al. Interaction of beta-hydroxy-beta-methylbutyrate free acid and adenosine triphosphate on muscle mass, strength, and power in resistance trained individuals. J Strength Cond Res. 2016;30:1843–54. View Article PubMed Google Scholar