A Novel Mri-Method Allows a Rapid and Robust Assessment of Muscle Quantity and Quality in Hemodialysis Patients.

Sarcopenia is a prevalent condition in patients on maintenance hemodialysis and associates with mortality. Research using magnetic resonance imaging (MRI) has demonstrated the importance of a proper body size-adjustment in the assessment of muscle mass, and that a muscle composition assessment including measurements of both muscle volume and fat inltration improves the prediction of comorbidity and survival related to sarcopenia. Such combined muscle composition assessment has not previously been performed in hemodialysis patients. Eleven hemodialysis patients were scanned using whole-body fat and water separated MRI and followed regarding survival and comorbidity for ve years. Muscle composition was assessed using AMRA® Researcher. Using data from 9615 UK Biobank participants, sex and BMI-matched muscle volume z-scores and sex-adjusted muscle fat inltration values were calculated for each patient. These measurements were then used for the calculation of a combined muscle score. Resulting three muscle measurements were associated with survival and comorbidity index.

The combined assessment including both body size-adjusted muscle volume and muscle fat in ltration can be used to analyze muscle composition in hemodialysis patients. MRI based muscle composition assessment re ected comorbidity and predicted survival in hemodialysis patients.

Background
Patients with chronical kidney diseases (CKD) has a degenerative muscle process associated with both protein wasting (with higher muscle degradation) and decrease of muscle synthesis (1). This leads to a loss of muscle mass, strength and function and make them more predisposed to sarcopenia. Although sarcopenia is a slow and age-related process, a chronic disease state such as CKD may lead to more rapid decline in muscle mass and function (2).
The relationship between increased prevalence of sarcopenia and progression of CKD is well described and is even more evident in patients with end stage kidney disease, especially when on dialysis treatment (3,4). Furthermore, sarcopenia in CKD patients is associated with a poor quality of life, increased morbidity, hospitalization and mortality (5,6).
The prevalence of sarcopenia in dialysis patients varies between 9.5 to 37.1% (7). This variation could be explained by differences between methods and cut-offs criteria used for sarcopenia assessment. A recent study using MRI has demonstrated that proper body size-adjustment is pivotal for identifying sarcopenia using muscle mass (8).
Muscle mass can be estimated by a variety of techniques, but MRI is considered a gold standard method (9). A standardized 6-minute MRI examination enables assessment of muscle volume and muscle fat in ltration (MFI) (10). MFI provides information of muscle quality and associates with low function, adverse outcomes, and mortality (2,11). The combined observation of low muscle volume and high MFI, adverse muscle composition (adverse MC), has been shown to improve functional correlations and predict hospitalization (8). Recently, adverse MC was linked to poor function and metabolic comorbidities within subjects with non-alcoholic fatty liver disease and all-cause mortality in general population (12,13).
In this study, hemodialysis (HD) patients were scanned using MRI and their muscle composition was assessed to investigate its association with comorbidity and mortality.

STUDY SUBJECTS AND CLINICAL DATA
In this prospective observational study, 11 adult patients undergoing HD at the University Hospital of Linköping were recruited between March-April 2014. The exclusion criteria were age less than 18 years and presence of any metallic implant. The Regional Ethical Review Board in Linköping approved (2013/475-31) the study protocol, and written informed consent was obtained from all patients.
Electronic health care records were used to calculate the nCIand to retrieve the ve-years survival data.

MEASUREMENTS OF BODY COMPOSITION BY MRI
Patients were scanned using rapid whole-body fat and water separated MRI. Fat-free muscle volume(FFMV) in the thighs and muscle fat in ltration (MFI) in anterior thighs were assessed using AMRA® Researcher (AMRA Medical, Linköping Sweden) (10). For the calculations, normative data from Page 4/15 FFMV: Fat-free muscle volume, the 'viable muscle tissue' (volume of all voxels with fat fraction <50%) in the thighs.
FFMVz-score:Fat-free muscle volume z-score. For each patient, a matched virtual control group (VCG) was strati ed among the participants from UK Biobank with complete muscle composition data. Each patient´s VCG included at least 150 individuals with the same sex and similar BMI. Based on each VCG, a personalized muscle volume z-score was calculated measuring how many standard deviations each patient was from the mean thigh FFMV/height 2 of their VCG. This variable is sex-, weight-, and height invariant and has been associated with poor function and increased hospitalization (8).
MFI: Muscle fat in ltration, the mean fat fraction within the 'viable muscle tissue' (FFMV) of the right and left anterior thighs. Due to the difference in magnitude between females (higher) and males (lower), a sex-adjusted MFI , MFI adj was calculated by subtracting the sex-speci c population median in the UK Biobank dataset from MFI.
Muscle comb : Combined muscle score, was estimated projecting MFI adj and FFMV z-score on the linear regression line describing the normal population relationship between MFI adj and FFMV z-score in the UK Biobank dataset.

STATISTICAL ANALYSES
Data was statistically analyzed using R.The Spearman rank correlation test was used to analyze the associations between muscle composition variables and comorbidity. To analyze the association between muscle composition variables and mortality, the Wilcoxon rank sum test and ROC analysis were used.

Results
A total of 11 patients were enrolled in the study. The majority were male with a mean age of 60.3 ± 12.3 years (Table 1). Seven patients exhibited lower muscle volume and eight patients higher MFI as compared with their matched controls (Fig 2). According to population wide cut-points for adverse MC, six patients presented with low muscle volume (<25 th percentile for FFMV z-score) and six patients with high MFI (>75 th percentile for MFI). A total of ve patients presented with adverse MC (low muscle volume coupled with high MFI).
The muscle volume z-score (FFMV z-score) and muscle fat measurement (MFI adj ) did not signi cantly correlate with comorbidity. However, when combining them into the Muscle comb score the correlation was statistically signi cant (Table 2). During the follow-up period of 68 months, ve of the 11 patients enrolled in the study died (two of cardiovascular disease, one of cerebrovascular disease, two of infection), four patients were kidney transplanted, and two patients remained on HD. Patients in the non-survivor group, when comparing with the survivor group, had signi cantly lower muscle volume z-score and higher MFI ( Table 1).
The three muscle composition variables, FFMV z-score, MFI adj and Muscle comb , exhibited a signi cant positive correlation with mortality (Table 3). Furthermore, the numerically strongest association was found for the combined variable, Muscle comb . A 2D-plot visualization of muscle volume (FFMV z-score) and sex-adjusted MFI (MFI adj ) showed that muscle composition was strongly associated with both comorbidity score and death (Fig 1).

Discussion
The present study demonstrates the feasibility of this MRI technique for assessment of muscle composition in HD patients and that combined assessment of muscle volume and muscle fat seems to predict mortality. Furthermore, the combination of these biomarkers, the combined variable Muscle comb , provides a more complete and muscle speci c description.
The current consensus for evaluating sarcopenia recommends assessing muscle quantity as well as muscle function (strength or performance) (2). Due to the high correlation between muscle quantity and body size, the patient's body size needs to be considered when assessing if they have a proper amount of muscle quantity or no. This is commonly done by creating a skeletal muscle index -muscle quantity divided by height 2 , weight or BMI. However, the most appropriate method of indexing muscle quantity remains still uncertain(8, 16). Recently, Linge et al. used MRI in a study of sarcopenic obesity and reported that common ways to adjust muscle quantity for body size (division by height 2 , weight or BMI) does not effectively normalize this relationship and confounds studies on sarcopenia. They suggested a body size adjustment for muscle volume using personalized virtual control groups, that ultimately enabled BMI-independent sarcopenia assessment(8). The same method was applied in this study.
The etiology of sarcopenia in CKD patients is still unclear but multiple factors are involved. Conditions that are directly related to CKD pathogenesis, such as cardiovascular diseases and diabetes mellitus (DM), are also independently associated with sarcopenia (17)(18)(19). Therefore, we analyzed the relationship between muscle composition and comorbidity index (nCI) in our patient cohort. We observed that the combined muscle score achieved by muscle volume z-score and MFI was signi cantly associated with the comorbidity index, but the association with the separate measurements did not reach signi cant level.
Our results suggest that the comorbidity score may identify patients who are more likely to have sarcopenia. Other studies have also found an association between nCI and sarcopenia supporting the idea that nCI could be used as a predictor to identify sarcopenia (20, 21).
As mentioned above, sarcopenia often co-exists with DM in HD patients and there is evidence that DM is an independent risk factor for the development of sarcopenia (17,22). Interestingly, in our study, all patients with DM belonged to the non-survivor group where adverse MC with low muscle volume z-score and high MFI were dominant.
Sarcopenia in CKD is associated with important clinical outcomes such as poor quality of life, cognitive impairment, depression, functional decline, frailty and hospitalization (5). Several studies have also reported that sarcopenia is an independent predictor of mortality in CKD patients (23). A new study on 9 000 individuals with CKD on UK Biobank reported that the presence of sarcopenia in CKD increased the mortality risk by 33% (24). In our study, we found that both muscle volume and MFI were signi cantly and independently associated with mortality and this association was even stronger when the combined score was assessed.
In line with our results, a study with patients undergoing hemodialysis reported that both low muscle mass and reduced muscle strength was associated with mortality (6). Interestingly, analysis of muscle composition data showed that adverse MC (low muscle volume z-score coupled to high MFI) seems to be associated with mortality in our patient group ( g 1). Recently published preliminary data on 40 000 participants in UK Biobank further showed that adverse MC, identi ed using the same method as in this study, is associated with four times higher mortality in general population (13) .These results, in combination with the results on the CKD patients in the present study, highlight that identi cation of adverse MC can be used to understand mortality risk related to sarcopenia across disease stages, not only in end stage disease ( g 2).
Hypoalbuminemia is a well-known predictor of mortality in patients with CKD (25) and it has been reported that there is an association between sarcopenia and hypoalbuminemia (24,26). In our study, we observed that patients in the non-survivor group (with a dominant adverse MC) had signi cantly lower serum albumin levels.
The strengths of this study include the detailed assessment of muscle composition (muscle volume and fat in ltration) with MRI and the use of quantitative muscle biomarkers with strong associations to functional performance and outcomes. The use of a highly standardized image acquisition protocol and systematic quality control procedures for muscle composition quanti cation enabled comparison between the study participants and virtual controls assessed in UK Biobank thus increasing the study size and enabling body size-independent sarcopenia assessment. The study has some limitations, including relatively low sample size and gender bias (only 2/11 of participants were women). A larger-scale study is needed to con rm our results.

Conclusions
This study demonstrates that this novel MRI method, combining body size-adjusted muscle volume and muscle fat in ltration, can be used to analyze muscle composition in CKD patients. Furthermore, MRI based muscle composition assessment did re ect comorbidity and could effectively predict survival in HD patients. The combined assessment including both muscle volume and muscle fat in ltration strengthened the predictive performance.
Our ndings suggest that MRI based muscle composition assessment can be used for identi cation of sarcopenia in CKD patients where early therapeutic interventions may prevent or delay the progression of sarcopenia. The results presented in this article have not been published previously in whole or part, except in abstract form.

FUNDING
Funding for image analysis of the UK Biobank data was gratefully received from P zer Inc.

Figure 1
Distribution of FFMV z-score and MFI adj among the patients in the dialysis dataset.
The number next to each observation is the nCI score. The ve observations with red circles around correspond to deaths. The blue dotted lines show the 1st, 5th, 25th, 75th, 95th, and 99th percentile in the UK Biobank dataset for each variable. FFMV z-score :virtual control group-based fat free muscle volume. MFI adj : muscle fat in ltration adjusted for sex. SD: standard deviation. VCG: virtual control group.

Figure 2
Coronal slice MRI scan and transversal slice with thigh muscle segmentations in 3 different patients Patient A with normal muscle composition. Patients B and C with adverse muscle composition, most notable in the patient C. FFMV z-score: virtual control group-based fat free muscle volume. SD: standard deviation. MFI: muscle fat in ltration.