In this MRI investigation of lower extremity muscles in FSHD patients, we quantitively assessed muscular fat infiltration in whole muscles from the distal to the proximal end, instead of only the central part of the muscle, as is common in quantitative MRI studies of diseased skeletal muscles. In this way, we demonstrate that, with a few exceptions, fat levels were highest at the distal end and lowest at the proximal end of the affected muscles. In particular, in muscles with an overall fat content between about 10 and 60%, the pattern of fat infiltration occurred as a (shifted) reversed sigmoid curve for which the slope (fat-infiltrating front) is more shallow at the higher fat content. Our longitudinal study revealed that these fat-infiltrating fronts moved to the proximal end of the muscle over the 3.5-year follow-up period. The fastest progression in the overall fat fraction was observed for muscles with mild-to-moderate fat infiltration, reaching a maximum rate of 4.9% per year in muscles with a 30–40% baseline fat fraction. Considering that muscular fat replacement is the ultimate result of cell death due to the expression of DUX4, we conclude that the disease process in FSHD starts at the distal end of leg muscles and proceeds at variable speed towards the proximal end. This conclusion is corroborated by our finding that TIRM-positive lesions, which are associated with disease activity10,17,21,24, preferentially occur at the distal end of affected muscles in the early stage of disease development in lower extremities of FSHD patients.
An important consequence of the results of our whole-muscle analysis is that the common practice in quantitative muscle MRI of assessing only a few slices at the centre of muscles may lead to large under- or overestimations of the fat fraction and its progression. In FSHD, this under- or overestimation varied from −11.8 to +23.8% for the baseline fraction and reached up to 3.4% for the fat fraction progression rate. This error can be minimised by analysing multiple transversal slices (5 or more) evenly distributed over the whole length of the muscles. In our study, semi-automatic muscle segmentation was essential for an efficient analysis of the whole muscles20.
To the best of our knowledge, this study is the first to quantitatively follow muscular fat fraction in FSHD patients along the whole muscle length, from the distal to the proximal end. The results of the cross-sectional part of this study, showing that affected muscles in the lower extremities of FSHD patients had higher fat fractions in the distal than proximal direction, are in line with earlier studies using quantitative T2 MRI for fat fraction assessment10,16 and reports using DIXON MRI for this assessment13,17. However, all these studies did not assess whole muscles end-to-end, so it remained unclear whether the proximal end of the muscle also had a relatively high-fat content and whether high-fat levels extended to the far distal end in the diseased state. Our data conclusively demonstrates that the fat fraction at the proximal muscle end is not higher than in more distal muscle parts. The findings strongly suggest that fat infiltration appears with a profile resembling a (shifted) reversed sigmoid curve and proceeds in a wave-like manner with variable speed from distal-to-proximal, with a slope (fat front) becoming more shallow during disease progression. These findings contrast previous suggestions that fat infiltration occurs via a quasi-linear gradient along the length of the muscle, a conclusion which can now be ascribed to the limited muscle coverage evaluated in these studies10,16. The occurrence of high-fat fractions starting distally and extending proximally during disease progression is typical for FSHD and differs from Duchenne muscular dystrophy, for which the highest fat fractions in the lower extremity muscles are found near both the distal and proximal ends, and the lowest at the muscle belly25.
It is remarkable that the majority of affected lower extremity muscles evaluated in the present study showed the typical pattern of fat infiltration, as described above. This implies that the far distal end of leg muscles in FSHD patients is a prime location for disease initiation, which is further supported by the preferential distal occurrence of TIRM lesions early in the disease development of a muscle. One might wonder why the distal end of these muscles is more susceptible to disease initiation than more proximal parts despite the wide variation in anatomy and function between muscles. The progression of fat infiltration also raises questions about processes underlying the spread of the disease towards the proximal end of the muscle.
Recent studies indicated that, after initial bursts of DUX4 expression in a sentinel myonucleus, the DUX4 proteins diffuse into neighbouring myonuclei and participate in the activation of DUX4 target genes7,18,19. In turn, the proteins of these genes activate further transcriptional factors leading to a transcriptional cascade, which is ultimately cytotoxic for myotubes. This gradually progresses along the whole muscle, consistent with a moving fat front. However, what initiates DUX4 bursts is still unknown, although studies on myoblasts from FSHD patients suggest that these bursts occur stochastically26,27. Our results indicate that specific conditions at the distal end of leg muscles are involved and, thus, that initiation is not purely stochastic. In Duchenne muscular dystrophy, it has been suggested that mechanical strain on the tendons contributes to disease initiation25, but in boys with this dystrophy, fat infiltrates muscles at both ends, in contrast to what we observed in muscles of FSHD patients.
The physiology of muscles may vary substantially along their length. Recently, the tibialis anterior was shown to have a relatively low oxidative capacity and post-exercise O2 supply in the distal part of the muscle28,29. This distally lower post-exercise O2 supply also occurs in the quadriceps and gastrocnemius30,31,32,33. Interestingly, oxidative stress in FSHD-affected muscle cells is implicated in disease onset by increasing DUX4 expression34. Moreover, DUX4 activates oxidative stress responses, leading to cell death35,36,37, and affected muscle cells are vulnerable to oxidative stress38,39. Therefore, it is of interest to know if (transient) hypoxia at the distal end of FSHD-affected leg muscles plays a role in disease initiation. Whatever the molecular conditions facilitating disease onset, our results provide a framework for further investigation into the nature of this process.
The overall change in the fat fraction of 1% per year indicates that fat progression in the lower extremities of FSHD patients at the individual level is slow. This percentage change is relatively low compared to those observed in other FSHD studies, which ranged from 0.86% per year to 5.4% per year10,14,15,24,40. This large variability among reported progression rates in a given patient cohort is not surprising, given that the rates are affected by multiple factors such as muscle type, the investigated location along the proximo-distal axis, the disease severity at baseline, and the applied MRI technique. For example, our study and that of others showed that fat infiltration in upper leg muscles progresses faster as compared to lower leg muscles14,15. Moreover, previous studies have assessed mainly central muscle areas, which can lead to over- or underestimation of fat progression (vide infra), which also depends on baseline fat fraction. Furthermore, more rapid fat infiltration rates have been reported in the muscles of FSHD patients involved in active disease processes, i.e., those showing TIRM-positive foci or those in a state of intermediate fat infiltration24. Finally, methodological differences, such as different MRI acquisition methods (e.g., 2 pt Dixon, multi-echo spin-echo) and data processing tools, may have slightly different sensitivities in the quantification of fat fractions. For example, the 2pt-Dixon study used in our study is likely to be less accurate than a multi-point Dixon, however, it does not affect precision and this precision is what really matters in longitudinal studies.
Despite that the leg muscle fat infiltration rate at the participant level was slow, fat infiltration can progress quickly in individual muscles. The fat replacement rate rapidly accelerates up to an average of 4.9% fat per year following fat infiltration onset until a muscle’s fat fraction becomes about 40%. At this stage, the fat replacement rate decelerates until muscles are completely fat-infiltrated. This typical pattern is in line with other longitudinal studies conducted in FSHD patients10,14,15. The pattern also fits the clinical picture reported in FSHD patients, with long stable periods alternating with short periods of rapid deterioration in single muscles or muscle groups41. In other muscular dystrophies, the relationship between the baseline fat fraction and the change in fat fraction has been studied in less detail. In late-onset Pompe disease, myotonic dystrophy type 1, and spinal muscular dystrophy, the change in fat fraction in muscles with elevated baseline fat fraction was larger than in muscles with fat fractions in the range of healthy controls42,43,44.
The presence of a distal-to-proximal fat-infiltrating front in diseased muscles has major implications for the application of quantitative MRI as a biomarker to assess treatments in clinical trials6,45,46. We demonstrated that evaluating fat fractions only in the central muscle parts, a commonly applied approach, instead of considering the whole muscle length, led to an average absolute error of 3.7% in baseline fat fraction and 0.6% per year in fat replacement rate in lower extremity muscles. The latter estimated absolute error is very large, representing 50% of the total change in fat fraction over 1 year commonly reported15,24. For instance, even if clinical trials intend to monitor disease progression in intermediately fat-infiltrated leg muscles—which are known to be most sensitive to changes—the corresponding evaluation of baseline fat fraction and fat replacement rate may both be largely biased if measurements are performed for slices located away from the fat progression front. Such an observation was made in a study of muscles in Duchenne muscular dystrophy, where a slight shift of 1.5 cm along the proximo-distal muscle axis led to a significant difference in the reported fat fraction25, especially in intermediately fat-infiltrated muscles. The absolute errors in baseline fat fraction and change in fat fraction can be substantially reduced when the five analysed slices are distributed along the muscle length. These findings demonstrate that whole-muscle imaging is essential to properly diagnose disease severity and to assess disease progression and response to therapy, such as in clinical trials. However, given that manual segmentation in quantitative assessments of whole muscle is very time-consuming, automatic or semi-automatic muscle segmentation approaches are critical in clinical assessments47. The analysis of an adequate number of slices properly divided over the length of muscles may represent a valid alternative.
In conclusion, fatty replacement in affected lower extremity muscles of FSHD patients commonly starts at the distal end of the muscles. It progresses relatively rapidly, in a wave-like manner, in the proximal direction in the mid-phase of disease development, after which it proceeds at a slower pace towards the proximal muscle end. In the early phase of disease development, TIRM-positive lesions dominantly occur at the distal end of affected muscles. These observations identify the distal end of lower extremity muscles as a prime location for disease initiation and are consistent with proposed mechanisms for disease progression in muscles. Factors involved in the activation of DUX4 expression to initiate disease have yet to be discovered, but oxidative stress at the distal end of leg muscles may play a role. Our analysis of whole muscles in FSHD demonstrates that the common practice in muscle MRI studies of assessing only a few slices in more central parts of muscles may lead to large under- or overestimations of muscle fat fractions and their change over time.