Almost nothing is known about the influence of the endurance burden on the specific changes in body composition regarding distribution of adipose and lean tissues in somatic and visceral compartments and in the body segments. Field studies on this topic mostly use methods which only allow indirect measurements and approximated calculations or simple estimations of total or local adipose or lean tissue proportions [11–14]. For TAT and subcutaneous adipose tissue (SCAT = SAST without intermuscular adipose tissue (IMAT) ), some of these indirect methods show more or less correlation to MRI findings . These methods are not able to predict the amount of visceral (VAT) or somatic adipose tissue (SAT) in the body [16, 32]. Being the first investigation in endurance field studies using the gold standard method  whole body MRI for such analyses, our results provide new data on the volume changes of fat and lean tissue in these different parts of the athlete’s body.
Age and gender related differences
Bale et al.  found a lower percentage of body fat in female elite marathon runners. In obese patients (BMI >27 kg/m2) Machann et al.  found that the amount and distribution of adipose tissue correlated with age (VAT increasing with age) and with gender (%SAT female > male, %VAT male > female). They found no consistent differences in TAT profiles between the selected age groups for both females (n = 40, mean age 45 years, SD 12 yrs., range 23 to 64 yrs.) and males (n = 40, mean age 45years SD 12 yrs., range 24 to 65 yrs.) in their group. Naturally, our group of ultra-runners with a comparable age distribution (n = 22, mean age 49 years, SD 12 years, range 27 to 69 years) showed a very low absolute mean volume of VAT at the start of TEFR09 (females: 0.5 L, males 1.8 L) compared to obese patients (females 1.5 to 4 L, males 4 to 6.8 L) . Statistical analysis of gender-related differences was not possible (only two females) in our group, but even these data indicate that a difference in VAT between men and women is not only visible in obese people, but is also visible in thin ultra-endurance athletes. Analysis based on age showed no correlation to fat distribution at the start (TAT, SAST, VAT) or to volume changes of lean and adipose tissue during TEFR09.
Changes in body composition
Different effects of endurance performance on body composition are described in the literature. Beyond dispute is the fact that endurance performance leads to a decrease in body mass, mainly body fat. Body fat is the main energy-rich substrate for endurance performance [34–37]. Therefore, endurance exercise leads to a reduction of subcutaneous tissue as demonstrated in several field studies [34, 36, 38].
The specific influence on energy turnover seems to depend on the type of endurance burden [1, 39]. In general, non-stop ultra-endurance races over hours, days or weeks without a break result in a decrease in body mass [1, 36, 40, 41] in which body fat as well as skeletal muscle seems to decrease [1, 36, 40–42]. In ultra-endurance performances with defined breaks, body mass may remain stable [43–45] or even increase  and body fat is reduced [34, 46, 47], whereas skeletal muscle mass seems to be spared [35, 43, 47] or may even increase . Our whole body MRI results show comparable results for an ultra-long MSUM over 64 days without any day rest; every subject decreased in BM(I), TV, TSV and TVV due to massive loss of TAT, SAT and VAT, respectively. Not every runner lost TLT and SLT during the TEFR09. Some of them showed increases, some decreases. Knechtle at al. found the same individual differences for lean tissue in ultra runners during a 1,200 km MSUM across Germany . If there are not sufficiently long breaks in ultra-endurance races, some participants might not find enough time for regeneration and restoration of their energy depots before the next stage. As the race progresses this leads to the utilization of muscle tissue for energy provision.
Raschka and Plat observed a mean loss of 1.75 kg body mass in an ultra-endurance run over 1,000 km within 20 days . In their investigation, there was a statistically significant decrease in body mass after day 8 until day 11, which then remained stable until the finish. In another investigation of 10 ultra runners (BIA), the mean loss of BM after a 1,200 km footrace was also not significant, but the loss of 3.9 kg fat mass was . Unfortunately, the authors gave no information about the relative changes of fat and lean body mass. Our results determined that a transcontinental ultra-long MSUM of 64 stages leads to a significant three times higher loss of body volume (9.5%) than published for body mass loss in deca-triathlons or 20 stage MSUMs [34, 47].
The relation of water and lipid to the density of human adipose tissue ranges from 0.925 to 0.97 kg/L . Assuming the middle value (0.948 g/L), in our investigation the ultra-athletes lost a total fat mass (TAT) of 4.8 kg in the mean (SAST 4.0 kg, VAT 0.8 kg), resembling the main part (91.8%) of body mass loss of 5.2 kg. The lean tissue of the human body has a higher density than adipose tissue and muscle tissue (range 1.05 to 1.06 g/L) and varies with age [49, 50], ranging between 1.10 and 1.11 g/L [51, 52]. With these data and knowing the mean relative reduction of TLT (1.2%), the mean loss of lean body mass can be calculated as about −0.67 kg at the end of TEFR09 in our subject group.
Visceral adipose tissue
Mediastino-abdominal lipomatosis is described as being associated with exertional dyspnea , non–insulin-dependent diabetes, type IV hyperlipidemia, and hyperuricemia. The abdominal VAT is an important independent risk factor for metabolic diseases in the older patient  and there is evidence that mainly abdominal VAT, which is morphologically and functionally different from abdominal SAST, is associated with the metabolic syndrome (insulin resistance, dyslipidemia, hypertension, obesity) and hyperinsulinemia [55–60], as well as linked inflammatory diseases . The real mean loss of relative IAAT while running a MSUM of nearly 4,500 km, was more than two thirds compared to the start in our group (Figure 13). We showed that endurance running also has an direct influence on intrathoracic fat, especially MAT, which decreased up to more than 40% in the mean (Figure 13). MAT is associated with hypertension, obesity, and iatrogenic Cushing syndrome [57, 62–65].
Until now, a specific treatment for the selective reduction of VAT is not known  and as our MR analyses showed that VAT decreased much more rapidly and vigorously than SAST (Figure 8), a very good and effective way to reduce the risk of metabolic disease is endurance running. As VAT decreases much faster and more than SAST, our investigation indicates that three-compartment measurement methods, such as SF-analyses and BIA, cannot give accurate assumptions or calculations for IAAT and MAT. Even the results of the four-compartment method cadaver study are false, when post mortem findings are transferred to physiological effects which occur from the impact of long lasting running on fat and lean tissue in vivo .
Finishers versus non-finishers
55% (n = 12) of the 22 ultra-runners treated with mobile whole body MRI for this study reached the last measurement interval; 10 dropped out earlier. In contrast, the dropout rate for all starters at TEFR09 and all subjects taking part in the TEFR project was 31% . Reasons for dropping out of this transcontinental MSUM race were overuse reactions of the musculoskeletal system of the lower extremities (80%, Figure 15), mainly concerning the myotendinous fascial system.
In a 17-day MSUM (1,200 km) Knechtle et al. found no differences between finishers and non-finishers regarding the anthropometric parameters, BMI, SF, CF, estimated skeletal muscle mass (estimated from SF and CF) and percent body fat (BIA) . With whole body MRI for differentiated body composition analysis, however, we found significant differences between finishers and non-finishers between both somatic and visceral volumes and between adipose and lean tissue volumes at the start and early beginning of the 4,500 km MSUM TEFR09 (Figure 16). Out results indicate that the risk of dropping out of such an ultra-long transcontinental footrace is significantly higher when the total body fat percentage is more than 21% to 25% at the start, in which the visceral fat percentage (VAT) shows a higher difference between finishers and non-finishers (71.5% in the mean) than the somatic fat compartment (SAST, 28.0%). Because VAT is affected by the endurance running burden most quickly and most deeply compared to somatic fat and other lean tissue (Figure 9) and is highly correlated with prerace performance regarding training volume and intensity and specific ultramarathon race-performance (50 km-race), our results indicate,that VAT is the most sensible predictor for the risk of non-finishing a transcontinental MSUM, such as the TEFR09. In ultra-runners there is not a high SAST or TAT, if the VAT is low.
Although training a distance of 4,500 to 5,000 km is not possible, participants of such MSUMs should acquire specific characteristics and levels regarding body composition and performance skills even before the race if they want to have a good chance to finish: VAT near 20% to 21%, training volumes of more than 100 km/week one year before the race and the performance intensity of 7.5 km/hour at a minimum allowing specific ultra-race records of less than 5 hours in 50 km-races or more than 178 km in 24-hour races. In other words, if these levels of prerace performance are reached for at least 15 months before the transcontinental race, the VAT (and SAST, TAT) as the sensible marker for specific body composition adaptation is also in an optimal range for low risk of non-finishing, because these parameters correlate in a mostly high level.
Because the subjects mainly fall out of the race due to overuse injuries in the myotendinous fascial system of the lower extremities, we tend to assume that the mentioned interdependent parameters of body composition and prerace ultra-running performance, lead to overuse injuries in the main stressed musculoskeletal organs, if they are not highly adapted as mentioned above; too little specific ultra endurance adaption and too much of VAT (and SAST) results in a high risk of severe soft tissue overuse in the legs and mostly happens in the early phase (Figure 15) of a transcontinental foot race.
Nearly every starter of TEFR09 showed, more or less often, overuse soft tissue problems of the myotendinous fascial structures of the legs during the race, but the feet are not a region for problems for experienced endurance runners in a MSUM . So the immense amount of mechanical stress on the musculoskeletal system when running nearly two marathons daily over a period of nine weeks can lead to these overuse syndromes without the obligatory necessity of prevalent (intrinsic) factors, such as ‘overweight’ (high VAT), suboptimal ultra-endurance prerace performance or mal-alignment of the legs (which was only seen in one female subject suffering from a bunion). The majority of the participants was able to ‘overrun’ more or less severe overuse soft tissue syndromes in the legs and reached the finish line . This indicates, that, despite the mentioned somatic parameters, other mentally based factors, such as pain resistance and personality traits, are also relevant for finishing or non-finishing a transcontinental footrace . One subject (male, 61 years old) had to stop the race after stage 38 (2,601 km run) due to a high tibial stress fracture which was detected in a specific MRI on at this day (Figure 15). The astonishing thing is not the stress fracture, because this can happen to every ultra runner when starting a transcontinental race, but the fact that the major pain and massive performance (running velocity) loss had already started at stage 36. This subject ran 228 km (three stages) with a complete high tibial fracture before stopping the race, because he interpreted the pain as a soft tissue injury due to overuse and tried to ‘overrun’ it before he asked for MRI control. Another participant (female, 46 years old) showed the same behavior when running 208 km (stage 46 to 48) with a ventral pelvic ring stress fracture before diagnosis could be done with mobile MRI . These examples and our prerace test on pain tolerance demonstrate that the resilience of the ultra athletes regarding pain is significantly higher than in a normal control group .
Body composition and performance
In specific treadmill investigations under laboratory settings, Millet et al. showed that a good single ultra-marathon performance needs specific running economy depending on the ability of maximal oxygen uptake being highly correlated with citrate synthase activity and capillary network . These physiological factors have not been investigated directly under race conditions in ultra-endurance events until now. Concerning this matter, only indirect parameters, such as anthropometric characteristics, are examined.
Several anthropometric factors are reported to affect performance in runners, but the presented data are inconsistent and often contradictory. Such differences are also present in the specific literature regarding anthropometrical predictors of performance outcome in ultra-marathons. Several factors are responsible for this. The numbers of volunteers are different, and in most reports they are limited and differ in gender and ethnic origin. Furthermore, the investigations are based on manifoldly different types of UM races. They can differ in the distance of running and number of stages, but also in altitude and/or external conditions.
Anthropometric parameters related to good performance are different in marathons and middle distance (half-marathon, 10 km) events . Knechtle et al. reported that anthropometry is not associated with performance in single mono-stage UM races (24 hours ).
In MSUM Knechtle et al. found no correlation between BM or body fat (BIA) and race performance in a 17-stage MSUM (‘Deutschlandlauf 2007’, 1,200 km) . In a cohort of 392 athletes, Hoffman found a significant relationship of BMI to finishing times in mono-stage UM running (161 km UM) . In single marathon runners abdominal and front thigh SF are correlated . The sum of eight SF-locations correlated significantly to 100 km race-time in a survey of three races in Knechtle et al. .
According to our results with a group of 22 subjects and using gold standard whole body MRI, in athletes taking part in a 64-day MSUM there are no relevant correlations between total volume, percentage fat and lean volumes of different compartments at the start and total race performance of subjects participating in TEFR09. For SAST, a significant correlation between percentage volume at the start and cumulative performance is seen at the beginning of TEFR09 (stages 1 to 8), but only at a medium to low effect size. Correlation of percentage fat and lean volumes to performance at the individual stages could only be shown in a few stages at a medium to low effect size. Looking at percentage volume distribution, the participants already started with a low percentage of body fat. Therefore, our results might confirm earlier findings of a negative relationship between the amount of subcutaneous fat tissue (thickness or volume), being the main fat tissue compartment of the body, and performance in single or multiday ultramarathon races. However, in a multistage ultramarathon over thousands of kilometers we found no relationship between body fat percentage or BM or BV and race performance using specific whole body MRI, as Knechtle et al. did with BIA . The majority of transcontinental MSUM participants ran not for winning but for finishing the race; therefore, running velocity was a priority only for a few of them. For single UM races, the race time and, therefore, the performance plays a more important role for the ultra-athletes, and body composition and fat distribution have a more significant influence, respectively.
Similar interpretation has to be done, when looking at segmental (somatic) tissue changes in the arms, legs and trunk during TEFR09. As for adipose and lean total somatic and visceral volumes (Figure 19), we also did an analysis of the relationship between segmental tissue volume changes and race performance (results not demonstrated graphically) and detected only a small to low medium effect size for correlations between SAST of all segments (UE, TR, LE) with cumulative race performance in the first eight stages of TEFR09. So, in our investigation, all segments show a significant relationship to race performance that is similar to that of SAST over all (Figure 19) without any exceptional segment findings, which explains the inconstant finding in the literature. Knechtle et al.  found an association between triceps SF thickness and performance in female 100 km ultra-runners. Tanaka and Matsuura mentioned this for CF of the thigh in the early eighties .
Some ultra athletes show adaption to the intense running burden of TEFR09 with muscle (SLT) increase in the legs, although they are already specialized in ultra running. These findings were not significant in the mean. For the trunk, a mean increase of SLT could also be detected in the first third of the race. This is explained by the gluteal and psoas muscles, which are part of the active motor system of the lower extremities but anatomically are placed in the trunk in our segmentation. All lean tissue segments showed a decrease in their volumes towards the end of TEFR09, indicating the high negative energy burden of transcontinental running.
After the first thousand kilometers the mean loss of TV per km, mainly caused by the SAST and VAT decrease, declined constantly up to more than half until the end of race (Figure 11). Despite lack of documentation of the nutrition and caloric intake but knowing that the subjects tried to ensure an optimum of energy intake, the decrease of fat volume loss can be explained by two factors: relevant metabolic changes regarding energy balancing  and improvement and optimization of running style during progression of the race. Not in multistage but in single stage ultra-running conditions such economical adaptations have already been shown by Millet et al. [80–82]. They could show significant changes of running mechanics and spring-mass behavior towards a higher mean step frequency (+4.9%) with shorter ground-feet contact time (−4.5%) and lower ground reaction force (−4.4%) due to functional leg length decrease (−13%) and increase of leg (+9.9%) and vertical stiffness (+8.6%) during the support phase of running between the early phase and the end of a 24-hour treadmill run . Millet et al. speculated that these changes in running mechanics contributed to the overall limitation of the potentially harmful consequences of such a long-duration run on the subjects’ musculoskeletal system. Transferred to MSUM conditions, such changes in running mechanics may also contribute to the necessity of the organism to optimize the running economy to a high-end level (as low an energy consumption as possible) due to the massive negative energy burden a transcontinental race requires. The changes Millet et al.  and other researchers had measured [83, 84] describe a running technique which requires only a low muscle power, because forceful eccentric load and step length are reduced. Besides the reduction of overuse risk for the musculoskeletal system this reduces the energy demand of the organism as well , even if the underlying mechanisms of the relation between energy cost of running and step variability remains unclear until now. If running economy could not be sacrificed in ultramarathons [86, 87] and the amount of change in running mechanics depends on the duration of running and distance towards a fatigue state, respectively [81, 85], it is even mandatory in transcontinental MSUM. Every subject in the TEFR-project showed a significant loss of BM and TV throughout the race, independent of the prerace overall status of body composition and performance or nutrition behavior during the race. The massive negative energy burden of a 4,500 km MSUM is also indicated by the significant loss of the grey matter in the brain . The analysis of specific laboratory markers of the required blood and urine samples may give more data about the metabolic changes during TEFR09 in the near future.
There was no general or individual nutrition plan offered or generated for the participants of TEFR09 or subjects of the TEFR-project, respectively. The athletes had a breakfast and a dinner served in different locations at the stage destinations, but these meals were organized and oriented at the local level at the last minute. The food supply points during the stages also offered products that changed every day and the athletes took additional individual food on their own throughout the race . Therefore, documentation and measurement of nutrition and caloric intake was not possible and a stringent documentation of nutrition by the subjects implied the risk of compliance problems.
Whole body mobile MRI protocols did not measure ectopic fat such as intracellular fat of organs (for example liver) and muscles (intramyocellular lipids: IMCL). For IMCL measurement, specific protocols for mobile 1H-MR-spectroscopy of the muscles of the lower legs were implemented in the TEFR-project . However, due to the dependence of this MR-method on a stable external magnetic field around the magnetom, the analysis of mobile1H-MR-spectroscopy during TEFR09 did not lead to valid data and needed further development and implementation of post-imaging proof algorithms.