Muscle Volume and Intramuscular Fat of the Tongue Evaluated With MRI Predict Malnutrition in People Living With Dementia: A 5-Year Follow-up Study

Abstract

Malnutrition is highly prevalent in older persons with dementia. Therefore, strong predictors of malnutrition in this population are crucial to initiating early interventions. This study evaluates the association between the probability of having malnutrition with the muscle volume and intramuscular fat (iMAT) of the masseter and the tongue in magnetic resonance imaging (MRI) of community-dwelling older persons diagnosed with mild dementia followed up for 5 years. This is a longitudinal study conducted in the western part of Norway. Muscle volume and iMAT of the tongue and masseter were computed from structural head MRI obtained from 65 participants of the Dementia Study of Western Norway using Slice-O-Matic software for segmentation. Malnutrition was assessed using the Global Leadership Initiative on Malnutrition Index. Linear mixed models were conducted. Having malnutrition at baseline was associated with lower muscle volume (odds ratio [OR] 0.60, standard error [SE] 0.20; p = .010) and higher iMAT (OR 3.31, SE 0.46; p = .010) in the tongue. At 5 years follow-up, those with lower muscle volume (OR 0.55, SE 0.20; p = .002) and higher iMAT (OR 2.52, SE 0.40; p = .022) in the tongue had a higher probability of presenting malnutrition. The masseter iMAT and volume were not associated with malnutrition in any of the adjusted models. In people diagnosed with mild dementia, tongue muscle volume and iMAT were associated with baseline malnutrition and the probability of developing malnutrition in a 5-year trajectory. In the masseter, there were no significant associations after adjustments.

The number of people with dementia is growing worldwide. Alzheimer’s disease (AD) and Dementia with Lewy Bodies (DLB) are the most common neurodegenerative dementias. These conditions are associated with dependency, poor quality of life, comorbidity, mortality, and high societal and family burden (1,2). Malnutrition is a prevalent condition in people living with dementia, which frequently associates with functional decline and mortality (3). There is a relevant link between nutritional status, loss of muscle mass, intramuscular fat (iMAT), inflammation, and dementia (4–6). Therefore, loss of muscle mass and higher iMAT could have an important role as early markers of malnutrition.

Much like visceral adipose tissue, iMAT secretes inflammatory cytokines leading to systemic inflammation (7). Proinflammatory status promotes muscular wasting and malnutrition (8). A recent study revealed that older adults with increased iMAT in the quadriceps had a higher risk of malnutrition (5). iMAT also constitutes a marker of muscle and mobility dysfunction in older adults (6,9,10). In addition, low muscle mass and sarcopenia have been identified as risk factors for falls, cardiovascular disease, loss of mobility and autonomy, frailty, and cognitive impairment (11,12). Therefore, muscle mass and iMAT are potential predictive markers of health outcomes in people living with dementia.

Muscle volume can be estimated using several techniques, such as total body or appendicular skeletal muscle mass, or by measuring specific previously validated anatomical sites. For muscle mass and structure, magnetic resonance imaging (MRI) and computerized tomography (CT) are considered the gold standard techniques. However, they involve high cost and lack portability limiting their use in clinical practice (13). Regarding specific anatomical sites, quantification of the lumbar muscle cross-sectional area by CT or MRI and the quadriceps assessment by ultrasound are good valid techniques for evaluating muscle loss and thus confirm the diagnosis of sarcopenia (14). However, in particular contexts, other anatomical areas such as cervical paravertebral and sternocleidomastoid muscle areas have demonstrated high reliability to identify sarcopenia in patients with neck cancer (15). In addition, in abdominal cancer, the abdominal skeletal muscle area has also been used to identify muscle wasting (16).

In the context of neurodegenerative disorders, as part of the standard clinical approach, people suspected of having dementia usually undergo head MRI or CT scans. Hence, masseter and tongue muscles could be opportunistically used as a surrogate measure of the general body muscle volume (17,18). Some studies have already reported associations between the masseter with abdominal muscle volume in CT and weight (19), and between the tongue muscle volume, measured using ultrasonography, and nutritional status in older persons (20). In addition, tongue and muscle are directly related to food intake, making them even stronger markers when referring specifically to malnutrition.

Overall, measuring masseter and tongue muscle and their iMAT in head MRI already performed in patients with neurodegenerative conditions is relatively inexpensive, free of radiation, accessible, and easily applicable in clinical practice. To date, little is known about the association of these structures with malnutrition. This study aimed to evaluate the muscle volumes and iMAT of the masseter and the tongue in head MRI scans from community-dwelling older persons diagnosed with mild dementia. We investigated the association of these structures with malnutrition and their capacity to predict further nutritional deterioration in a 5-year trajectory in people living with AD and DLB.

Materials and Methods

Setting and Participants

The “Dementia Study of Western Norway” (DemVest) is a longitudinal cohort study with yearly assessments up to 12 years. DemVest recruitment period was conducted between 2005 and 2013, and follow-ups are still ongoing. Participants consisted of referrals from dementia clinics, outpatient clinics in geriatric medicine, neurology, and old age psychiatry in the counties of Rogaland and Hordaland in Western Norway after being diagnosed with mild dementia by a physician. After agreeing to participate, the patient and caregiver were first seen by the study clinician, who performed a structured clinical interview of demographics, previous diseases, and drug history. The comprehensive assessment procedure included a clinical examination composed of physical, neurological, psychiatric, neuropsychological examinations, and routine blood tests.

Inclusion criteria were people with mild dementia defined as a Mini-Mental Status Examination (MMSE) score 20 or more or a Clinical Dementia Rating global score 1 and diagnosed with dementia. The inclusion period took place between 2005 and 2013. Exclusion criteria were absence of previously diagnosed moderate or severe dementia, delirium, previous bipolar or psychotic disorder, terminal illness, or recently diagnosed significant somatic disease. All participants at baseline were living at home, and 57% had a partner living with them. Participants were followed up yearly with comprehensive annual assessments. Details of the design, recruitment, clinical, and biomarker procedures are described elsewhere (21).

From an initial sample of 222 participants with different types of dementia, people diagnosed with mild AD and DLB followed up for 5 years were selected for this analysis. The final sample corresponded to those participants with complete assessments, from the same center (Stavanger), using the same MRI acquisition technique, and with available good-quality head MRI images at baseline (AD = 45, DLB = 20). Sample selection, dropouts due to death, and loss to follow-up are depicted in Supplementary Appendix 1. The regional ethics committee approved this study (approval code: 2010/633) and the Norwegian authorities for the collection of medical data. All data were handled and kept following national health and data privacy protocols. All participants signed an informed consent form before inclusion in the study.

Assessments

Dementia diagnosis

Diagnosis of early dementia was made according to the Diagnostic and Statistical Manual of Mental Disorders (14). Specific types of dementia were diagnosed according to the corresponding validated instruments (22,23). Pathological diagnosis was made on 56 participants of the DemVest cohort, with an accuracy above 80% compared to the clinical criteria (21).

Nutritional status

The Global Leadership Initiative on Malnutrition (GLIM) index was used to determine the nutritional status (24). Following inference rules, body mass index (BMI) and age were used to categorize individuals into 3 possible nutritional status groups: severe malnutrition, moderate malnutrition, or adequate nutritional status. Individuals whose BMI was less than 18.5 and were younger than 70 years and BMI less than 20 and 70 years or more were considered to be severely malnourished. Participants with BMI less than 20 and younger than 70 years of age and BMI less than 22 and older than 70 were considered to have moderate malnutrition. Participants who did not fit into the established categories were considered as not having malnutrition. As a required etiologic criterion, all studied participants had a diagnosis of dementia. For the primary set of analyses, moderate and severe malnutrition were analyzed together in a single group. We analyzed nutritional status at baseline and yearly during the 5 years of follow-up.

MRI calculations of iMAT and muscle volume

Because this is a new method, we had to calibrate and determine the tissue thresholds for each machine. Thus, to reduce variability data from the center, only the images from Stavanger were used, and images from other sites due to different scanners and acquisition techniques were excluded. Baseline MRIs were acquired from a 1.5-T Philips Intera scanner using this acquisition protocol for 3D T1-weighted sequence: repetition time/echo time 10.0/4.6 ms, flip angle 30.0°, 2-mm slice thickness with 1-mm spacing between the slices (1-mm slices with no gap), number of excitations 2, matrix 256 × 256, and field of view 26 cm. We conducted a visual quality check procedure, discarding those with movement artifacts and inadequate image quality. A standardized preprocessing method for harmonizing multiple collections of MRIs was applied, which consisted of movement correction and intensity normalization.

Using the thresholds for tissues of interest, the volumes and volume ratios of muscle, iMAT, and subcutaneous fat (SAT) were manually tagged and quantified at the regions of interest using Slice-O-Matic (Montreal, CA) software (Figure 1A) as previously described (25). The volumes of tissues and their signal intensity (eg, in muscle, a possible indicator of intracellular fat infiltration) were quantified.

Confounding variables

Demographic factors included in the analysis were sex and age. Comorbidities were assessed using the Charlson Index and were registered based on participants and informant reports (26). Cognition was evaluated using the MMSE in its validated version in Norwegian (27).

Statistical Analysis

A descriptive analysis was performed by estimating percentages for categorical variables and means and standard deviations for quantitative variables. We also evaluated the differences between groups using Pearson’s chi-squared test for categorical variables and the Kruskal–Wallis test for quantitative variables. The baseline variables considered potential confounders were age, sex, the MMSE score, and the Charlson Index score for comorbidities. The normality distribution assumption for continuous variables was analyzed using the Shapiro–Wilk test.

To analyze the association between baseline nutritional status as an outcome and tongue, left, and right masseter muscle measures as covariates, we conducted a logistic regression adjusting by the cofounders listed above and the type of dementia (AD/DLB) due to the possibility that any diagnosis, especially DLB, could have a more significant influence on the outcome, based on its reported poorer prognosis (28). We performed the same analysis using the longitudinal nutrition status using a logistic mixed model. To determine whether these MRI analyses could predict future malnutrition, we completed the analyses excluding participants without malnutrition at baseline. We fixed the significant probability at .05 to evaluate the covariates’ influence in the models using R version 3.6.0.

Results

Descriptive variables of the sample are given in Table 1. As shown, 27 (41.54%) were men, and 38 (58.46%) were women, and the mean age was 76.27 ± 6.70. There was an inverse relationship between muscle volume and iMAT; when muscle volume decreases, iMAT increases (Figure 1B). The prevalence of malnutrition at baseline was 23.7%.

At baseline, people with malnutrition had lower muscle volume (odds ratio [OR] 0.60, standard error [SE] 0.20; p = .010) and higher iMAT (OR 3.31, SE 0.46; p = .010) in the tongue. During the 5-year follow-up, those with lower muscle volume (OR 0.55, SE 0.20; p = .002) and higher iMAT (OR 2.52, SE 0.40, p = .022) in the tongue had a higher probability of malnutrition (Figure 2).

Subsequently, we performed an analysis excluding those with malnutrition at baseline. We found that the tongue muscle volume was significant as a predictor of malnutrition development during the follow-up (OR 0.63, SE 0.21; p = .027; Table 2).

Finally, the masseter iMAT and muscle volume were not associated with malnutrition in any of the adjusted models (Table 2).

Discussion

Tongue muscle volume and iMAT were independently associated with a higher probability of developing malnutrition in people with mild dementia. This finding was especially relevant when evaluating people without malnutrition at baseline, where the muscle volume of the tongue significantly predicted malnutrition. Following our results, iMAT and particularly muscle volume of the tongue offer a unique opportunity for early prediction of malnutrition in dementia, thus providing the possibility of new highly needed biomarkers that help predict prognosis (eg, higher risk of malnutrition), and promote early actions in older persons with dementia.

To the best of our knowledge, this is the first study analyzing this association in people living with dementia. However, previous studies have reported similar results in older adults, although focusing on muscle but not fat volumes. A Norwegian study detected lingual atrophy in more than 1/3 of the hospitalized older adults using morphological measures. Here, tongue atrophy was associated with anthropometric variables related to nutrition, serum concentrations of ascorbic acid, cholesterol, calcidiol, and general malnutrition (29). In addition, using ultrasonography, Tamura et al. (20) reported an association between tongue thickness and nutritional status.

By contrast, we did not find associations between malnutrition and iMAT or muscle volume in the masseter muscle. Hwang et al. (19) showed a significant correlation of the masseter muscle area analyzed via CT anthropometry with the abdominal muscle area, weight, and age. Our studies differ in many aspects, such as the imaging technique, segmentation, and assessment of the study variables. Hashida et al. (30) recently showed that geniohyoid muscle, but not masseter muscle, was associated with swallowing function after salvage surgery and radiotherapy in head and neck cancer, indicating that the tongue muscle can be more critical than masseter muscle for swallowing. It is then worth considering that masseter muscle volume can be altered by conditions not evaluated in the current study, such as bruxism, hypertonia in DLB, dental prosthesis, defective or missing teeth, the habit of chewing gum, temporomandibular joint disorder, or congenital and functional hypertrophies; common conditions in people living with dementia (31–33).

Like other muscles in the body, masseter and tongue muscles decline in volume and strength with aging (34,35). This becomes even more relevant in dementia when dysphagia is common, with a prevalence of up to 57% (36,37). The tongue plays an essential role in the deglutition process; therefore, adequate muscle volume in the tongue is required. On the other hand, malnutrition leads to loss of lean mass and strength in the whole body, including in the masticatory muscles (38). Therefore, a 2-way relationship exists between muscle loss and malnutrition when one condition reinforces the other (39,40).

As confirmed by our observations, fat infiltration follows muscle loss, further affecting mass and function. Due to the large volume of muscular tissue in the body, even slight infiltration of ectopic fat into muscles can lead to the local and systemic secretion of adipokines and fatty acids, which are known to be lipotoxic (ie, produces abundant toxic fatty acids and inflammatory cytokines) (41). Lipotoxicity could cause severe muscle atrophy and can also lead to increased circulating inflammatory markers and a systemic inflammatory reaction. Both muscular decline (42) and systemic inflammation (43,44) are also associated with poor prognosis and malnutrition.

In addition to this being one of the first studies to investigate the association of the iMAT and muscle volume of tongue and masseter in dementia, this study has several other strengths. DemVest includes long follow-up time and annual assessments from diagnosis to death, using structured validated instruments. The latter allowed us to determine malnutrition from mild to severe dementia. Also, diagnostic procedures were rigorous and accurate; the neuropathological diagnosis was available in a subgroup, demonstrating that the clinical diagnosis was accurate (21).

This study has some limitations. As we included primary care referrals, selection bias could happen when recruiting a higher number of people with more complicated health status or dementia. However, general practitioners were encouraged to refer any person with suspected dementia. Enrolled participants were treated according to updated guidelines for clinical pharmacological and nonpharmacological management which may influence the course of the nutritional status. As mentioned, studied muscle volumes can be altered by conditions that were not evaluated, such as missing teeth, bruxism, hypertonia in DLB, or dental prosthesis. The automatic segmentation method had some limitations, such as determining thresholds and the operator’s experience. However, Slice-O-Matic measurements have been previously validated against gold standards, demonstrating a high inter- and intrarater reliability and high reliability and validity (45). Furthermore, medications such as acetylcholinesterase inhibitors have been associated with gastrointestinal complaints, anorexia, and weight loss (46–48). However, the concomitant treatment and diet of the participants were not accounted for in the analysis. In addition, although living alone or with a partner can affect nutrition (49), the exact number who lived with a partner or alone during the study period was not recorded.

Participants’ anthropomorphic information varied between years of measurement. This accounts for the seemingly greater number of AD participants from Year 2 to Year 3, even when considering deaths and dropouts. The only anthropomorphic measurements recorded in the DemVest study were weight and height; therefore, it was impossible to do a more detailed nutritional evaluation. Nevertheless, in this study, we assessed malnutrition using the GLIM index. This tool has been shown as a predictive marker of prognostic, including incident mortality in several settings, including community-dwelling older adults and patients with proinflammatory diseases (24,50,51). To date, studies using the GLIM index have not provided specific guidance about including dementia as an etiological criterion. However, strong evidence links dementia with active inflammation, thus making it a valid etiology indicator (52,53).

Variables selected for the models’ adjustments, such as cognition or comorbidities, were chosen given their influence on nutritional status (54). Model adjustments by BMI were not directly performed because the outcome (malnutrition) contains this variable (24). An important issue is that the final sample from the DemVest included in the analysis was relatively small, decreasing the power of this study. In addition, because of mortality rates, the sample was significantly reduced to 5 years of follow-up. Particularly in DLB, a high mortality rate was observed after Year 3, which could have biased the estimates in the longitudinal mixed-effects models on malnutrition (55).

In summary, in participants diagnosed with mild dementia, tongue muscle volume and iMAT were associated with baseline malnutrition and the probability of developing malnutrition in a 5-year trajectory. Noninvasive quantification of these tissue volumes has the potential to become inexpensive and accessible prognostic markers in the field; however, more longitudinal studies with more specific and sensible methods are necessary.

Acknowledgments

We want to thank the participants, researchers, and technical staff who made the DemVest study possible, as well as staff and facilities provided by Centre for Age-Related Medicine (SESAM) at Stavanger, Norway.

Funding

This work was supported by the Norwegian government, through hospital owner Helse Vest (Western Norway Regional Health Authority) number 911973. It is also funded by the National Institute for Health Research (NIHR) Biomedical Research Centre at South London and Maudsley NHS Foundation Trust, and King’s College London. E.B.H. received a seed grant from the Australian Institute for Musculoskeletal Science and a Medical Research Future Fund fellowship under Rapid Applied Research Translation program in conjunction with the Melbourne Academic Centre for Health.

Conflict of Interest

The authors have no conflicts of interest to declare. The funders had no role in the design of the study, in the collection, analyses, or interpretation of data, in the writing of the manuscript, or in the decision to publish the results. The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR, or the Department of Health and Social Care.

References