Various myosteatosis selection criteria and their value in the assessment of short- and long-term outcomes following liver transplantation

Body composition and myosteatosis affect clinical outcomes in orthotopic liver transplantation (OLT). Here we aimed to compare the value and limitations of various selection criteria to define pre-transplant myosteatosis in the assessment of short- and long-term outcomes following OLT. We retrospectively analyzed the data of 264 consecutive recipients who underwent deceased donor OLT at a German university medical centre. Myosteatosis was evaluated by preoperative computed-tomography-based segmentation. Patients were stratified using muscle radiation attenuation of the whole muscle area (L3Muslce-RA), psoas RA (L3Psoas-RA) and intramuscular adipose tissue content (IMAC) values. L3Muslce-RA, L3Psoas-RA and IMAC performed well without major differences and identified patients at risk for inferior outcomes in the group analysis. Quartile-based analyses, receiver operating characteristic curve and correlation analyses showed a superior association of L3Muslce-RA with perioperative outcomes when compared to L3Psoas-RA and L3IMAC. Long-term outcome did not show any major differences between the used selection criteria. This study confirms the prognostic role of myosteatosis in OLT with a particularly strong value in the perioperative phase. Although, based on our data, L3Muscle-RA might be the most suitable and recommended selection criterion to assess CT-based myosteatosis when compared to L3Psoas-RA and L3IMAC, further studies are warranted to validate these findings.

While body composition (BC) may strongly vary among individuals, the generalized loss of muscle mass, function and strength defined as sarcopenia is frequently observed in critically ill patients 1 . A progressive sarcopenia is an underappreciated and frequent complication in patients with end stage liver disease (ESLD) and can be present in 40 to 60% of the patient undergoing orthotopic liver transplantation (OLT) 2 . Over the past ten years, an increasing number of reports demonstrated that the status of the skeletal muscle compartment has a significant prognostic value in various oncological and chronic diseases. Previous studies have shown the association of sarcopenia with inferior waiting list and post-transplant outcomes [3][4][5][6] . Excessive pathological intramuscular fat disposition called "myosteatosis" has recently been independently correlated with an increased risk of inferior outcomes in cancer and in end-stage liver disease 1,7,8 . Recent studies by our group have identified not only a high prevalence of myosteatosis but a strong association with adverse perioperative outcomes in patients undergoing OLT 4 .
Although, various techniques (e.g. dual-energy X-ray absorptiometry, magnetic resonance imaging, bioimpedance analysis) are used to assess patient BC in the clinical setting, cross-section imaging studies, such as computed tomography (CT), are recognized by the recent Clinical Practice Guidelines of the European Association for the Study of the Liver (EASL) as the gold standard for the quantification of clinically significant structural alteration in the skeletal muscle compartment 9 . Muscle mass (sarcopenia) and quality (myosteatosis) are usually estimated by segmentation of the cross-sectional area at the level of third lumbar vertebra (L3). Even though, myosteatosis is typically defined by low muscle radiation attenuation (RA) values in Hounsfield units (HU), there are multiple selection criteria introduced by different groups to characterize myosteatosis and identify patient at risk, without a clear international consensus 4,5,10 . Frequently utilized are the absolute values of muscle attenuation using sex-specific cutoffs of the whole skeletal muscle area (including psoas major, erector spinae, quadratus lumborum, transversus abdominis, external and internal obliques, and rectus abdominis) versus the bilateral psoas muscle area 4,10 . A novel selection criterion for the assessment of skeletal muscle quality and myosteatosis has been first described by Kitajima in non-alcoholic fatty liver disease and further explored by Hamaguchi et al. in the setting of living donor liver transplantation 5,11,12 . Intramuscular adipose tissue content or IMAC is defined as the lumbar multifidus muscle / subcutaneous fat tissue attenuation ratio 5,12 . Despite the fact that all of the above described muscle attenuation or myosteatosis selection criteria have been used in various patient cohorts, there is no clinical data directly comparing their value in the prediction of post-transplant outcomes in the setting of deceased donor OLT.
In this study we aimed to comprehensively assess the performance of three frequently used selection criteria for myosteatosis (L3Muscle-RA; L3Psoas-RA; L3IMAC) in predicting post-transplant outcomes in a large European single-center cohort of adult patients undergoing deceased donor OLT. Image analysis. Computed tomography imaging and CT segmentation were carried out as described before 4 . Briefly, image data of the most recent preoperative CT-scan were analyzed by the same investigator who was blinded for the remaining clinical data and outcomes of the patients. A single cross-sectional image/patient has been analyzed at the level of the third lumbar vertebra using the 3D Slicer software platform version 4.1 and BC module (https:// www. slicer. org/) as described before 4,13,14 . Table 1 shows the definitions and attenuation cutoff values used in the segmentation analysis. Sex-and cohort-specific cutoff values have been defined by the first and last quartiles of the corresponding body composition parameters (Q1 for intramuscular adipose tissue content-IMAC; Q4 for L3Psoas-Radiation Attenuation (RA) and for L3Muscle-RA; see Table 1 and Fig. 1 Clinical data collection and patient follow up. Clinical data were recovered from a prospective institutional database, medical charts and analyzed in a retrospective fashion. Liver allocation followed German national and international Eurotransplant regulations. The liver transplantation procedure was performed using a standardized approach of total cava replacement as previously described [16][17][18] . Perioperative treatment and immunosuppression were performed in a standardized fashion as described before 16,17 . The RWTH Aachen transplantation outpatient clinic and the responsible community-based hepatologists provided the follow-up data used in this study.  www.nature.com/scientificreports/ All definitions, scores and classifications used in this manuscript have been described by our group and by others in previous reports (including OLT risk scores [19][20][21][22] , definitions of extended criteria donor allografts-ECD and early allograft dysfunction-EAD [23][24][25] , indications for OLT listing 26 , Clavien-Dindo classification-CD and the Comprehensive Complication Index-CCI 27,28 , calculations of the length of ICU and hospital stay 29 , procedural costs 30 , peri-and postoperative transfusions 4 ).

Statistical analysis.
The primary endpoint of the present study was the incidence of 90-day post-OLT major morbidity (defined by CD ≥ 3b) 27 . Overall perioperative outcome, length of ICU-and hospital stay, mortality, EAD, procedural costs, long-term graft-and recipient survival were analyzed and reported as secondary endpoints.
Normal distribution was tested using the Kolmogorov-Smirnov-test for continuous variables. Data was reported as mean and standard deviation for normally distributed, and median (interquartile range-IQR) was displayed for non-normally distributed data. Absolute and relative frequencies were reported in case of categorical variables. For the statistical comparison of continuous variables, the Student t test, the Mann-Whitney U test, and the Kruskal-Wallis H test were used where appropriate. The Chi-square test and the Fisher's exact test were used, for the analysis of categorical data. To determine the ability of myosteatosis to predict perioperative outcome, uni-and multivariable logistic regression analyses were performed. Spearman correlation coefficient was used to further analyze the association of various clinical outcomes and myosteatosis. The further discriminative ability of the various myosteatosis selection criteria for the prediction of outcomes was compared using the receiver operating characteristic (ROC) analysis calculating the area under the receiver operating characteristic curve (AUROC). The Hosmer-Lemeshow Chi 2 goodness-of-fit test was applied to test model suitability. The level of statistical significance was defined as p < 0.05 and the statistical analysis has been performed using SPSS Statistics v24 (IBM Corp., Armonk, NY, USA).
Next, OLT recipients have been divided into quartiles, based on the distribution of L3Muscle-RA, L3Psoas-RA, L3IMAC values over the patient cohort (Fig. 4). This analysis, led to the observation that while L3Muscle-RA resulted in a satisfactory stratification of our patients even in individuals with superior muscle quality (Q1-Q3), L3Psoas and L3IMAC were not able differentiate in terms of outcomes between the patient quartiles with higher   Fig. 4) when L3Muscle-RA was used. However, this gradual or step-wise pattern was not observed when L3Psoas-RA or L3IMAC were used to stratify our cohort (see e.g. Fig. 4 B3 or B2). Based on this, despite their relatively good performance in the identification of high-risk individuals in the group analyses, L3Psoas-RA and L3IMAC were not able to differentiate between patients with better muscle quality and less advanced myosteatosis (Fig. 4). Analyzing the AUROCs, the best results in terms of the discriminative ability of the three tested parameters were obtained using 90-day mortality as outcome (Table 4). Here L3Muscle-RA and L3Psoas-RA showed satisfactory high AUROC values (> 0.7) with significant results and satisfactory model fit (L3muscle-RA: 0.762 p < 0.001; L3Psoas-RA: 0.751 p < 0.001; L3IMAC: 0.703 p = 0.077; Table 4). In this analysis, L3IMAC showed inferior performance compared to the other two parameters with either non-significant AUROC values due to broader confidence intervals or an insufficient model fit (Table 4).

Discussion
This study provides insights into the performance of various frequently adopted selection criteria of muscle radiation attenuation and myosteatosis in predicting short-and long-term outcomes following deceased donor liver transplantation. Although, all three parameters showed an overall satisfactory performance in predicting perioperative morbidity and mortality, L3Muscle-RA was superior in the quartile based, correlation, and AUROC analyses. Neither of the used myosteatosis selection criteria was able to identify patients at risk for inferior longterm graft and patient outcomes, which is in line with previous findings showing that the strong prognostic value of myosteatosis seems to be particularly important in the early postoperative period 4,5 .  www.nature.com/scientificreports/ This study builds on the limited but continuously accumulating body of published evidence that BC and especially sarcopenia and myosteatosis are associated with worse clinical outcomes in patients with ESLD 32 . While previous reports provide ample evidence on the association between sarcopenia and outcomes 2,33 , only a handful of recent studies have suggested a potential value of myosteatosis in the setting of liver transplantation 4,12 . Even in case of these sporadic publications, there is a large heterogeneity concerning patient cohorts and the used selection criteria to define myosteatosis 12 . The lack of an international consensus on methodical definitions complicates the interpretation of these findings and results in inconclusive systematic reviews and metaanalyses 33 . This may ultimately impede the translation of BC assessment into clinical practice guidelines and international recommendations.
Malnutrition and consequential alteration in BC can be assessed with a broad variety of screening tools which have been validated in the past in various patient cohorts 7,34,35 . Although, CT-based image analysis and quantification of muscle mass (morphological aspect of sarcopenia) and muscle quality (myosteatosis) are considered to be the gold standard in patients with liver disease, a number of research groups have introduced various selection criteria and cutoff values to assess BC and identify patients with clinically relevant BC alterations 9,12,32 . Not only the mean attenuation values of the entire lumbar skeletal muscle area (L3Muscle-RA in our present study) but also the total psoas density (L3Psoas-RA in our present study) are frequently used by various authors to characterize myosteatosis in patients with liver disease 10,36 . In a recent study by Kalafateli et al., they recommended the bilateral psoas attenuation to characterize myosteatosis 37 . Based on these, the central and deep location of the psoas muscle, the more simple and precise identification of its exact borders would facilitate a precise image analysis and segmentation. Furthermore, the density and form of the psoas muscle are presumably less influenced by abdominal distension and disease-related water retention compared to other abdominal muscle components (e.g. ventral abdominal musculature) 37,38 .
Besides L3Muscle-RA and L3Psoas-RA, the lumbar multifidus muscle / subcutaneous fat tissue attenuation ratio, known as IMAC (L3IMAC in our present study), was used in multiple Japanese studies to determine  www.nature.com/scientificreports/ myosteatosis 5,12,39 . In contrast to the absolute RA values, this novel parameter holds promise to reduce the variation between individual CT scans and patients, leading to an improved identification of clinically significant alterations 5,12 . A higher IMAC indicates an increased muscular adipose tissue content, thus a lower muscle quality 5,12 . In our present report, the presence of myosteatosis, defined by the sex-specific quartile-based cutoff values for L3Muscle-RA, L3Psoas-RA, and L3IMAC, was significantly associated with inferior perioperative outcomes. Patients with myosteatosis presented with significantly increased morbidity and mortality (increased major complication rates ≥ CD3b and cumulative CCI) over the first 90 days following OLT and showed higher intraoperative transfusion needs and longer stay on the ICU and in hospital. This inferior perioperative outcome was manifested in increased costs over the first 3 months. Although, there were no major differences in the performance of the three analysed selection criteria for recipient myosteatosis in terms of perioperative outcomes in our group analysis, L3Psoas-RA and IMAC seemed to be slightly inferior compared to L3Muscle-RA in the identification of patients at risk for EAD in our group analysis (Table 3). In our quartile-based, correlation and AUROC analyses, however, L3Muscle-RA showed a superior discriminative and diagnostic ability.
While several Japanese studies have explored the association of IMAC with the severity of non-alcoholic steatohepatitis (NASH) and the outcomes following LDLT 5,11,12,40 , and our group and others have extensively investigated L3Muscle-RA and L3Psoas-RA [1][2][3][4]33,37,41 , none of these previous reports attempted to compare various selection criteria for muscle RA and myosteatosis in a liver transplantation cohort. IMAC was first described by Kitajima et al. showing a relationship between an increasing IMAC and disease severity in NASH patients 11,40 . As the values of IMAC have improved over time following therapeutic intervention such as dietary changes and exercise, the authors proposed IMAC as a potentially valuable marker to non-invasively monitor therapeutic success in patients with chronic liver disease. However, IMAC has later also been adopted for the "non-NASH" setting and the Kyoto group has investigated its role following LDLT. In their pioneering report by Hamaguchi et al., they have found a strong association (p < 0.01) between high IMAC values and post-transplant survival using living donors 5,12 . In our present study the probability of graft-and patient survival did not differ significantly over the follow-up period below and above the L3Muscle-RA, L3Psoas-RA, L3IMAC cutoffs. However, likewise Table 5. Uni-and multivariable logistic regression analysis for 90-days major morbidity (Clavien-Dindo ≥ 3b). Values were given as numbers and (per cent). Results of the logistic regression were given as oddsratios with 95% confidence interval. *Factors showing a p value < 0.1 in the univariable analysis were included in the multivariable logistic regression model. Only significant results are shown. # To avoid a multicollinearity effect due to the inclusion of L3Muscle-RA, L3Psoas-RA, L3IMAC, the multivariable analyses were repeated for each of the three variables. a Based on the German Medical Chamber Guidelines 31 . Abbreviations used: BMI: body mass index, ECD: extended criteria donor allografts, MELD: model for end-stage liver disease, ICU: intensive care unit, L3Muscle-RA: lumbar 3 muscle radiation attenuation L3Psoas-RA: lumbar 3 Psoas radiation attenuation, L3IMAC: lumbar 3 intramuscular adipose tissue content.

Major complications (CD ≥ 3b) 1 n = 136
No-/ minor complications (CD1-3a) 1  www.nature.com/scientificreports/ in our findings, in the above-mentioned Japanese cohort a large number of the registered death events-thus the major difference in survival-occurred during the early post-LDLT phase with 90% of patients dying within the first year after LDLT 5,12 . Therefore, the lack of survival difference in our study may be attributed to our different statistical approach. To avoid the potentially interfering effects of early mortality we have excluded patients who died within the first 3 months after OLT (n = 20) from the analysis of long-term graft-and patient outcomes.
The findings of this study should be interpreted in the light of potential limitations. First, due to the retrospective nature of our analysis, the present study omitted any functional analysis of patient fitness and muscle strength which should be mentioned as an important limitation. Second, it is also necessary to consider whether the used L3Muscle-RA, L3Psoas-RA, and L3IMAC cutoffs used in our group analysis were adequate to identify patients at risk for inferior outcomes. Here we chose to use sex-specific cutoff values to identify patients belonging to the lower 25% in our cohort in terms of muscle quality according to the 3 different myosteatosis selection criteria. However, to reduce potential bias associated with this approach, we have also used further sophisticated and comprehensive statistical methods to analyze and report the diagnostic value and limitations of these three parameters from various angles (AUROC analysis, quartile-based distribution and correlation analysis). Third, our analyzed patient cohort shows the general characteristics of a heterogeneous European OLT patient cohort which carries the risk of a certain selection bias and may led to the underrepresentation of various indications and patient subgroups (e.g. high-MELD patients or patients with NASH). Fourth, CT scans used for segmentation analysis were obtained preoperatively as part of the clinical routine at heterogeneous time points and analysed in a retrospective and uncontrolled fashion.
Notwithstanding these limitations, this report is one of the first comprehensive studies assessing and comparing the value and limitations of three different but frequently reported radiation attenuation-based selection criteria for myosteatosis, demonstrating a comparable performance and similar shortcomings for all three parameters in predicting short-and long-term outcomes following deceased donor OLT. L3Muscle-RA has performed slightly superior compared to L3Psoas-RA and L3IMAC (depicted e.g. by the prediction of EAD as well as in a better linear correlation with ICU and hospital stay, CCI and costs or by its superior performance in the quartilebased or AUROC analyses). Based on these promising results, an international consensus and standardization of selection criteria would be highly desirable to improve comparability and reproducibility of findings and facilitate rapid translation of BC-based and malnutritional scores into clinical risk-assessment and outcome prediction in the setting of OLT. Further studies are warranted not only just to provide an external validation for these findings but to test the prognostic robustness of myosteatosis in various highly selected patient cohorts using a multi-center setting with a sufficient sample size and statistical power. It might be of particular clinical interest to investigate the prognostic role of myosteatosis in severe morbidity and mortality using a larger set of selected high-MELD patients which was not possible in a statistically meaningful way in the present single-center study.

Data availability
All relevant data were reported within the manuscript and the supplementary files. Further supporting data will be provided upon written request addressed to the corresponding author.