*7.1. General Principles*

The two original validation studies of the MuscleSound® technique [4,5] used protocols that allowed the biopsy to be taken from the identical site on which the image was captured. Meanwhile, as previously identified (Table 1), the outlying studies ([9] and the Bone study presented here) took care to standardise the sites from which both ultrasound images and biopsy samples were collected, but used different sites from the same muscle between and within treatments to accommodate best practice associated with the collection of sequential muscle biopsies. A key premise of the MuscleSound® protocol, at least in research scenarios, was that the location of the image for sequential assessments or comparison with biopsy assessments must be identical. However, much of the extended commentary about the protocol [10], and the specific criticisms of studies which found it did not provide a valid assessment of muscle glycogen [12], misunderstood or misrepresented the larger literature on the assessment of muscle glycogen. Specifically, comments about the variability of glycogen within muscle [10,12] demonstrated a failure to understand the capability of various assessments techniques.

This review has identified that all studies of muscle glycogen in humans utilised indirect and sampling techniques. The basis of such sampling, which occurred in both of the original MuscleSound® validation studies, is that a small piece of muscle collected in a biopsy needle or captured in an ultrasound image includes sufficient muscle fibres to represent the aggregated features of individual fibres. Indeed, both the MuscleSound® score and the chemical analysis of a biopsy sample represent the characteristics of "mixed muscle". Enhanced techniques of analysing biopsy samples include histochemical staining techniques to identify differences in the storage and utilisation of glycogen between different types of muscle fibres [50], and, more recently, electron microscopy of single fibres identified different sub-cellular locations of glycogen particles [21,22,51,52]. Such techniques have helped to understand exercise metabolism and mechanisms of fatigue during exercise. However, ultrasound techniques, just like chemical assessments of homogenates or mixed muscle samples, cannot achieve such a granular assessment. Rather, they provide an overarching, ye<sup>t</sup> still valuable, perspective of muscle fuel stores. In the case of muscle biopsies, there is a specific reason to require and validate the use of different sites for sequential biopsy samples; subsequent biopsies need to be taken from muscle that has not had its glycogen storage capacity impaired by trauma from the first biopsy [19]. However, it was shown that differences in the glycogen content of mixed muscle samples,

representing the average of a large number of individual muscle fibres from a number of individual sites across a muscle, are minor [23]. We acknowledge that the collection of biopsy samples from adjacent, but non-identical sites, or sites from contra-lateral limbs in scenarios involving symmetrical exercise protocols, may contribute to the technical error of measurement involved with chemical determination of muscle glycogen stores. Nevertheless, it is the basis of a robust literature involving many hundreds of studies, which have determined resting muscle glycogen concentrations in different populations [36], glycogen utilisation during exercise [1,53], and glycogen synthesis in response to diet [19,54,55].

Although the size of a biopsy sample can be measured (typically, 20–200 mg), the size and location of the actual site captured in the ultrasound image is uncertain. Since the analysis of the scan is undertaken via proprietary cloud-based software, the precise size and location of the sample and site, and its ability to represent the total muscle, ultimately lies with the company software, rather than the scan technician. Nevertheless, even if there are concerns about the validity of MuscleSound® glycogen assessments, there is some evidence of its reliability in estimating glycogen content across a muscle site. Indeed, the first study of the technique noted a significant correlation between the MuscleSound® glycogen content and its changes due to an exercise bout between two *separate* muscles. Here the correlation between glycogen stores of the rectus femoris and vastus lateralis were *r* = 0.93 (*p* < 0.0001), *r* = 0.91 (*p* < 0.0001), and *r* = 0.76 (*p* < 0.0001) for the pre-exercise, post-exercise, and exercise change scores, respectively [4]. Furthermore, in the Bone study reported in this review, we found a significant correlation between MuscleSound® scores at a single site and a shifting site within the same muscle, across a range of treatments (Figure 3). Therefore, in theory and in practice, there is evidence that in the absence of muscle damage, changes in glycogen in response to diet and exercise are similarly expressed across the gross aspect of a muscle.

#### *7.2. Specific Criticism of Studies That Fail to Support the MuscleSound® Technique*

The major rebuttal of data sets that have found that the MuscleSound® technique was unable to provide a valid measurement of muscle glycogen [10,12] is that "since glycogen is stored in different pools within a same muscle and therefore not uniformly stored, technically it is not possible to correlate the glycogen content from a very small portion of a muscle (1–2 cm2) with the glycogen content of an entire muscle" [10]. We have identified that this criticism of mixed muscle samples was confused with findings of sub-pools of glycogen within a single muscle fibre or between fibre types, and does not provide a legitimate understanding of broader muscle glycogen assessment. It is not necessary to make further comment on this issue. Nevertheless, if differences in the glycogen content of different sites within the same muscle do exist, this might provide an explanation for the lack of correlation between the biopsy site and ultrasound site in the dissenting studies discussed within this review. However, it fails to explain the failure of the MuscleSound® protocol to detect changes in its glycogen assessment metrics when across ultrasound scans taken at the same site on subsequent occasions. That such differences were both logical, based on knowledge of supervised diet and exercise treatments, and easily detected from chemical analysis of biopsy samples creates legitimate concern about the validity of the MuscleSound® technique.
