**1. Introduction**

The determination of muscle glycogen content is of key interest in sports nutrition due to its roles as a fuel source in athletic performance and a regulator of muscle metabolism and adaptation [1–3]. A technique that could achieve reliable and valid measurements, while being inexpensive, portable, and non-invasive, would have an enormous potential for increasing knowledge and enhancing practice, particularly in work involving elite athletes and field conditions. Therefore, there is understandable excitement around the commercialisation of ultrasound technology aimed at measuring muscle glycogen in both laboratory and field situations [4,5]. MuscleSound® (Glendale, CO) uses high-frequency ultrasound technology and image analysis from patented cloud-based software to estimate muscle glycogen content from the echogenicity of the image; this feature is based largely on muscle water content, which is presumed to act as a proxy for glycogen [4–6]. It was used in research activities to describe changes in muscle glycogen in response to diet and exercise, as well as being a commercially available tool to guide the preparation and recovery of athletes. Extended applications of MuscleSound® were more recently proposed, including the monitoring of the glycogen status of critically ill patients in hospital intensive care units as a measure of metabolic health [7]. Furthermore, ultrasound techniques are

**Citation:** Bone, J.L.; Ross, M.L.; Tomcik, K.A.; Jeacocke, N.A.; McKay, A.K.A.; Burke, L.M. The Validity of Ultrasound Technology in Providing an Indirect Estimate of Muscle Glycogen Concentrations Is Equivocal. *Nutrients* **2021**, *13*, 2371. https://doi.org/10.3390/nu13072371

Academic Editor: Kazushige Goto

Received: 12 May 2021 Accepted: 7 July 2021 Published: 11 July 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

being developed as reliable and valid protocols to monitor subcutaneous fat in athletic populations [8]. Nevertheless, this review will focus on the use of MuscleSound® to assess muscle glycogen within sporting populations and scenarios.

Although validation studies have been published [4,5], incongruous experiences with the use of this tool [9] led us to express concerns about a report on glycogen utilisation during a high-level football match using MuscleSound® technology [10]. We questioned the interpretation of these findings, including acknowledgement of data from our own groups, which refute the validity of the MuscleSound® technique [11]. Rebuttal from the study authors [12], which include co-founders of the technique and commercial company, dismissed our concerns based on assertions that the dissenting studies failed to understand the basis of the use of their tool. Furthermore, they noted that only one contradictory dataset is available in the peer-reviewed literature, while the other study remained an unpublished conference presentation [12].

Given the many uses of muscle glycogen measurements in sports nutrition research (e.g., investigations of strategies to enhance glycogen storage and enhanced understanding of strategies to enhance post-exercise adaptation) and practice (e.g., guiding individual athletes to optimally fuel for competition or achieve different levels of carbohydrate ((CHO) availability for training goals), it is important to discern whether the MuscleSound® technique provides valuable and actionable information or whether it might contribute to misleading research outputs and unsupported training and dietary practices. The aim of this review is to examine the available literature on the use of this ultrasound technology to measure muscle glycogen concentrations in athletic populations. To allow a complete account, we will include the results of the unpublished project, while providing full transparency over these data and the cause of their absence from the literature. Such data are valuable given the small number of studies of muscle glycogen involving MuscleSound® in general, and the independent nature of our investigation. We note, in particular, that our study provides the only comparison between biopsy-derived glycogen measurements and the newer estimated fuel level score provided by the MuscleSound® technology. The estimated fuel level is described in a company "position stand" as the metric by which athletes can be given actionable feedback about the suitability of their diet and exercise practices [13]. Although the original technique, producing an estimate of muscle glycogen content in arbitrary units, was described in two validation studies [4,5], we are unaware of any published work that confirms the reliability and validity of the updated technique which, as described in company literature, is the information provided for real-life uses.

#### **2. Brief Overview of Methods to Assess Muscle Glycogen**

Muscle glycogen is an important fuel store for exercise, as well as a key regulator of metabolism within the muscle cell [2]. Indeed, the development of methods to measure its presence and location within the muscle cell provided the first major advances in the science and practice of sports nutrition [14]. The first measurements of muscle glycogen were made possible by the introduction of the percutaneous biopsy technique to sports science in the late 1960s [15–17]. Subsequent modification of the technique included the addition of suction to increase the size of the sample collected [18] and movement of the location of the muscle site by 2 cm for subsequent biopsies to avoid the artefact of damage from the first [19]. This protocol is still used today, and is considered the "gold standard" for assessment of muscle glycogen stores, while acknowledging the invasiveness of the procedure and its downstream limitations on the subjects and environments in which it might be safely and logistically performed. A muscle sample collected by biopsy (typically 20–200 mg according to the size and type of needle) can be treated with several enzymic, histochemical, or electron microscopy procedures to determine the average glycogen content of mixed muscle, the specific glycogen content of fibre types, or the sub cellular location of glycogen, respectively [20–22]. In particular, validation studies showed that glycogen measurements from a homogenated muscle sample, freeze-dried to remove variables such as the presence of connective tissue and fluctuating cell water

content, provides a reliable measure of the glycogen content of the whole muscle under study [23].

Notwithstanding the utility of the biopsy technique in allowing an in-depth analysis of glycogen location within and between different muscle fibres, as well as enabling measurement of a vast host of other muscle metabolites, signalling molecules, and other "omic" interests, there is recognition that other techniques are needed to address the practical and ethical concerns associated with its use. The addition of magnetic resonance spectroscopy (MRS) to magnetic resonance imaging, and the increase in the power of the magnetic fields generated by such equipment, has allowed MRS to become an additional tool to indirectly assess muscle metabolites and fuel stores. This technique uses magnetic fields and radio waves to measure tissue glycogen by monitoring either 13C natural abundance levels, or 13C atoms incorporated into glycogen following the administration of a 13C substrate [20]. MRS techniques were shown to provide a reliable assessment of muscle glycogen in healthy and clinical populations [24] and have been applied to assessments of glycogen utilisation or storage in athletic populations [25,26]. Although MRS offers the advantages of providing rapid, non-invasive, and potentially repeated measurements of glycogen in various tissues, its disadvantages include the expense and the need to access specialised facilities and expert technicians.

## **3. MuscleSound® Measurements of Muscle Glycogen**

MuscleSound® is a commercially available tool which uses high-frequency ultrasound and patented software to derive an assessment of muscle glycogen. This technique, utilising a point of care device and cloud-based software, offers features that address the disadvantages of both biopsy and MRS-derived assessments of muscle glycogen; namely, it is a protocol that is non-invasive, relatively cheap, rapid, and portable. The original development of MuscleSound® was undertaken by researcher-practitioners who noted apparent correlations between ultrasound images and muscle glycogen content. The image greyscale produced on an ultrasound scan is based on the intensity of the ultrasound "echogenicity", or reflection of an ultrasound beam, with the beam being both produced and detected by the transducer forming the ultrasound image [27]. Ultrasound beams are reflected at the boundary between two materials with different acoustic impedances, with strong reflections showing as white on the ultrasound image and weaker echoes being grey [27]. Proprietary information within the MuscleSound® software aligns a darker image with greater glycogen stores using the principal that greater glycogen, and its associated water content in the muscle fibre, should reflect the lower echo intensity between soft tissue and water [5,27]. Conversely, in instances when glycogen is low and there is less fluid, the echo intensity is greater due to the increased visibility of other tissue boundaries, thus resulting in a brighter image [5,27].

The MuscleSound® technique, validated in two studies against biopsy-derived measures of muscle glycogen [4,5], produces a glycogen score in arbitrary units (hereafter identified as a.u.), where values are provided in bands of 5, between 0 and 100, with an unknown typical error. However, the assumed relationship between muscle glycogen and water was noted as a technical issue requiring further investigation in one of these foundation studies [5]. Indeed, according to an undated company position stand on the science and application of MuscleSound® located on the company website [13], further developments of the technique recognised scenarios in which muscle glycogen and water deviated from this relationship, and they provided recommendations for situations in which the use of MuscleSound® is considered optimal and those that are considered to be sub-optimal (see Section 6.1). According to this position stand, the current output from the MuscleSound® proprietary software provides a muscle energy status, representing the mean of an estimated fuel level and a muscle fuel rating. The company material describes muscle fuel as predominantly glycogen, with contributions from carnitine, creatine, and protein. The estimated fuel level is determined by "placing an image in context of the maximum (100) and minimum (0) points of glycogen obtained from a bank of images

captured for a specific participant", with the muscle fuel rating providing a separate rating compared to a large databank of images from many athletes [13]. We are unaware of any published validation studies of these new metrics, although the company literature promotes these values for field use in providing athletes with feedback about changes in muscle glycogen resulting from their diet and exercise strategies [13].

## **4. The Bone Study of MuscleSound®**

In 2014, we became aware of the newly launched MuscleSound® tool and realised both its potential to enhance our work as sports nutrition practitioners and applied researchers, and the opportunity to test its reliability and validity as an additional arm within a pre-existing project investigating interactions between manipulations of muscle glycogen and creatine content [28]. Although the main aim of the study was to investigate the effect of creatine and glycogen loading on cycling performance, we embedded a research arm to investigate artefacts in the measurement of lean mass by dual energy X-ray absorptiometry due to changes in muscle water content associated with changes in muscle creatine, glycogen, and water content [29]. We invited the MuscleSound® group to use this opportunity to further test their technology in scenarios that are very common in sports, but outside the conditions under which their own validation studies were conducted [4,5]. They were not involved in the study design, funding, or conduct; rather, we funded their visit to Australia to train us in the use of their technique. Although we had intended to use our own ultrasound equipment to capture images, IT security requirements at our workplace prevented us from uploading images to a cloud-based server. Therefore, MuscleSound® loaned us an ultrasound machine and provided complimentary results of muscle glycogen estimates derived from their proprietary software, using the original algorithm, for the duration of the study. A contract was signed to oversee return of the equipment and, on completion of the study, the contribution of our data towards further development of the proprietary algorithm. Following data collection, the equipment was returned, and a poster presentation was prepared for the 2016 annual meeting of the American College of Sports Medicine (ACSM).

In the days prior to the ACSM meeting, we received a directive from a legal firm engaged by MuscleSound® to withdraw the poster from the conference. Although we complied, an electronic version of the abstract was included in a review of methods to assess muscle glycogen in sports nutrition activities without our involvement [20]. To find a path to our ethical obligation to be transparent with all research outcomes via peer-reviewed publication, we subsequently agreed to allow the company to re-analyse the scans from our study using an updated algorithm and interpretation framework. However, recognising the commercial sensitivity of the two datasets, we decided to delay publication until other studies of the validity of the technique, undertaken by independent research groups, but partially funded by MuscleSound® [9], were released. Although the main results of the current study, that MuscleSound® failed to provide valid estimates of muscle glycogen, are no longer original, the additional range of scenarios that we have studied (supervised CHO loading, prolonged exercise depletion, and the addition of creatine loading) present novel outcomes. Furthermore, they include the only direct comparison of biopsy-derived glycogen assessments and the MuscleSound® estimated fuel level score, which is promoted in company material [13] as a commercially available use of this tool to provide actionable information to athletes about the suitability of their diet and exercise activities. Once published, we will provide the original data to the MuscleSound® company.
