CT analysis of anatomical distribution of melorheostosis challenges the sclerotome hypothesis

Abstract

Melorheostosis (MEL) is a rare disease of high bone mass with patchy skeletal distribution affecting the long bones. We recently reported somatic mosaic mutations in MAP2K1 in 8 of 15 patients with the disease. The unique anatomic distribution of melorheostosis is of great interest. The disease remains limited to medial or lateral side of the extremity with proximo-distal progression. This pattern of distribution has historically been attributed to sclerotomes (area of bone which is innervated by a single spinal nerve level). In a further analysis of our study on MEL, 30 recruited patients underwent whole body CT scans to characterize the anatomic distribution of the disease. Two radiologists independently reviewed these scans and compared it to the proposed map of sclerotomes. We found that the disease distribution conformed to the distribution of a single sclerotome in only 5 patients (17%). In another 12 patients, the lesions spanned parts of contiguous sclerotomes but did not involve the entire extent of the sclerotomes. Our findings raise concerns about the sclerotomal hypothesis being the definitive explanation for the pattern of anatomic distribution in MEL. We believe that the disease distribution can be explained by clonal proliferation of a mutated skeletal progenitor cell along the limb axis. Studies in mice models on clonal proliferation in limb buds mimic the patterns seen in melorheostosis. We also support this hypothesis by the dorso-ventral confinement of melorheostotic lesion in a patient with low allele frequency of MAP2K1-positive osteoblasts and low skeletal burden of the disease. This suggests that the mutation occurred after the formation of dorso-ventral plane. Further studies on limb development are needed to better understand the etiology, pathophysiology and pattern of disease distribution in all patients with MEL.

1. Introduction

Melorheostosis (MEL; OMIM #155950) is a rare hyperostotic disease with a prevalence of 0.9 per million [1]. It has been described in children and adults. Patients can present with pain, range of motion limitations, limb swelling and limb length discrepancy, in growing children [2] (Jha et al. [20], JBMR). MEL typically affects long bones and is commonly unilateral. There is no reported geographic preponderance in disease distribution, as also suggested by a US-based patient registry (Fig. 1). MEL is not heritable with no reports of vertical transmission. In 2018, we found somatic activating mutations in MAP2K1 in 8 of 15 patients with MEL [3]. Whyte et al. have described a post-zygotic mutation in KRAS in the skin overlying the melorheostotic bone in a patient with familial osteopoikilosis and MEL in the setting of germline mutation in LEMD3 [4]. Germline mutations in LEMD3 have been implicated in MEL associated with osteopoikilosis or Buschke-Ollendorff syndrome but not sporadic MEL [5,6].

Fig. 1.

Distribution of melorheostosis across the US suggestive of random distribution of the disease.

Population data based on US 2010 Census, by ZCTA (Zip Code Tabulation Areas). Each dot represents 1000 people, and dots are randomly placed within each ZCTA according to the population to create a dot density map. Locations of incident melorheostosis cases (triangles) were obtained from the Melorheostosis Association website (http://www.melorheostosis.com/defaultfiles/Page439.htm and links thereof) for 2001–2015, and geocoded using the Nomination tool (http://wiki.openstreetmap.org/wiki/Nominatim) for reverse geocoding in the OpenStreetMaps API, implemented in the geopy Python package version 1.11.0 (https://pypi.python.org/pypi/geopy). The geocoding was done using Python version 2.7, the data for the dot density plot was generated using R (24) version 3.2 and the maptools package version 0.8–36 (25). The final map was generated using QGIS (26) version 2.10. Finding demonstrates that the prevalence of melorheostosis parallels the population density without any geographic predisposition. (Inspired by a presentation by Michael Whyte & Deborah Wenkert at the Melorheostosis Patient Conference, 2012 http://www.melorheostosis.com/default_files/Page2700.htm.)

MEL has a unique pattern of anatomic distribution which has intrigued scientists for decades - it is typically limited to one side of the limb (medial or lateral) but can progress proximo-distally to create the characteristic “dripping candlewax” appearance on radiographs. The affected bone is radiologically distinct from the unaffected bone. This pattern of disease distribution has thus far been largely attributed to the presence of sclerotomes. The word “sclerotome” in this sense refers to an area of bone which is innervated by a single spinal nerve level (as opposed to the embryological sclerotomes which develop from somites and give rise to the spinal column and paraspinal musculature). Because MEL is a rare disease with limited data, the sclerotome hypothesis has since been widely used to explain the pattern of anatomic distribution in the disease. Murray and McCredie were the first to propose that sclerotomes explain the unique anatomic distribution of MEL [7]. They reviewed the radiographs of 30 cases and found that in 19 of these cases, the distribution corresponded to a single sclerotome while the remaining 11 cases appeared to involve multiple sclerotomes. Publications from as recent as 2015 reference sclerotomes as the basis for the anatomic distribution of melorheostosis [8,9]. However, the very existence of sclerotomes is questionable with little scientific evidence to support their existence and largely presumed based on patterns of dermatomal innervation [10]. Anatomic studies have not demonstrated that sclerotomes exist in humans. The original mapping study which proposed the existence of sclerotomes in 1944 entailed stimulating individual areas of bone with needles (direct periosteal stimulation) in 26 volunteers and asking the subjects to describe points of radiation [11]. Maps were then drawn based on described location of pain. The original paper states “…the precise segmental innervation of the sclerotomal areas… cannot be given with any great precision” [11]. Several concerns have since surfaced about the methodology employed in proposing the sclerotomal map: a) accurate localization of referred pain can be difficult due to the dull and diffuse nature of pain b) precise number of skeletal points that were stimulated remains unclear c) unclear how these stimuli could be applied to whole skeleton in the presence of ligaments, tendons and muscles that cover most of the skeleton d) stimuli was applied to the periosteum, but segmental innervation of bone remains unknown and may be different from the periosteum and e) extent of pain reported by subjects in the study was likely influenced by both spinal and supra-spinal circuits [10]. Ivanusic evaluated the clinical studies on sclerotomes, development of skeletal innervation and the contributions of anatomical and physiological investigations and concluded that there is “little direct evidence” for the existence of sclerotomes [10]. Finally, no other disease has been demonstrated to conform to the sclerotomal distribution. Although sclerotomes were initially proposed to explain thalidomide-induced limb malformations by the same authors who proposed that the distribution of MEL follows sclerotomes [12], current research is increasingly suggestive of thalidomide’s action on cereblon protein and angiogenesis as the mechanism of dysmelia [13]. Hence, MEL is often cited as evidence that sclerotomes exist, and sclerotomes cited as a basis for MEL. This circular logic is unable to withstand scrutiny. Furthermore, this study has never been replicated.

We propose that the anatomic distribution of MEL can be explained by clonal proliferation of a mutated skeletal progenitor cell along the limb axis. The evidence supporting this hypothesis can be derived from studies on mice fetuses which have demonstrated the formation of restricted dorso-ventral (D-V) compartments in early stages of development of limb mesenchyme without any lineage compartmentalization along the proximodistal (P-D) or anteroposterior (A-P) axes [14–16]. Owing to the proximo-distal progression of limb proliferation/expansion, linear clones extending the P-D length of the limb occur commonly. Vertebrate limbs develop in a proximal-to-distal order. Undifferentiated cells underlie the apical ectodermal ridge (AER). As the limb grows, undifferentiated mesenchymal cells exit this region, proliferate and differentiate to create the limb skeletal elements progressing from proximal to distal (e.g. humerus to digits). Arques et al. studied the clonal expansion of single cells in the limb buds of the mouse [14]. They found that clones of a single cell have wide P-D expansion and can contribute to all three limb segments. Additionally, their work shows that expansions of single clones in the limb bud maintain relatively straight A-P distributions along the P-D axis with no lineage compartmentalization along the P-D or A-P axes. In contrast, an early D-V lineage compartmentalization was identified in the limb mesenchyme. This work shows that a single affected mesenchymal cell with a post-zygotic somatic mutation could create the unique patterns of affected tissue seen in melorheostosis (Fig. 2). To support our hypothesis, we evaluated the D-V compartment of affected bone in patients with confirmed somatic mutation in osteoblasts of affected bone [3]. We did not investigate the D-V compartment in patients whose mutation-status remains unknown as we wanted to study patients with confirmed underlying somatic mosaicism in MEL bone. The diagnosis of MEL is currently established by radiographic appearance of affected bone in the right clinical scenario. However, this makes the diagnostic criteria highly subjective and we suspect that what we know today as MEL, may be a mix of several diseases or disease-patterns.

Fig. 2.

Clonal proliferation and melorheostosis distribution.

A. On the left, a hindlimb of a 14-day mouse embryo genetically modified to produce LacZ in clones of developing cells. On the right, radiographs show anatomic distribution of melorheostosis in two different patients with upper extremity disease. Note the similar pattern.

B. On the left, a hindlimb of a 14-day developing mouse embryo genetically modified to produce LacZ in clones of developing cells. On the right, radiographs show anatomic distribution of melorheostosis in two different patients with lower extremity disease. Again, note the similar pattern.

2. Materials and methods

The original study was approved by the Institutional Review Board of the National Institute of Arthritis and Musculoskeletal and Skin Diseases (Clinicaltrial.gov NCT02504879). The eligibility criteria for study enrollment were – a) adult suspected or diagnosed to have melorheostosis and b) radiographic appearance consistent with MEL [17]. We confirmed the diagnosis of melorheostosis by the anatomic correlation of radiographic abnormality with increased uptake on ¹⁸F-NaF (¹⁸F-sodium fluoride) PET/CT studies. Thirty patients were hence recruited for the study of which fifteen underwent genetic testing. The non-contrast CT scan performed in conjunction with ¹⁸F-NaF PET/CT was reviewed by two radiologists (LK and AM) and compared to the proposed map of sclerotomes by Inman [7,11]. In cases where there was a discrepancy between their independent interpretations, a side-by-side consensus opinion was obtained. Of the 15 patients who underwent genetic testing, eight were confirmed to have somatic mutations in MAP2K1 with an allele frequency of 2.7–34% [3]. Evaluation of D-V confinement of the MEL lesion was performed in these eight patients with confirmed somatic mosaicism in osteoblasts of affected bone.

3. Results

Of the 30 patients whose CT scans were reviewed by the radiologists, we found that the distribution of MEL conformed to the distribution of a single sclerotome in five (17%) patients. In another 12 patients, melorheostotic lesions approximated a combination of contiguous sclerotomes with lesions spanning into contiguous sclerotomes but leaving large areas of involved scelrotomes free of the disease. It seemed more likely that these lesions violate the sclerotomal maps rather than MEL affecting contiguous sclerotomes. For example, patient Melo-21 (Fig. 3A) has melorheostosis from the humeral head, affecting the ulna, and the 3rd digit. The affected areas correspond to the C7 and C8 sclerotomes, however the radius and 5th digit is completely unaffected which are both covered by the C7 and C8 sclerotome regions. Melo-10 (Fig. 3B) has melorheostosis of the right lower extremity from the femoral head, extending down to the tibia, and to the great toe, corresponding to sclerotomes L3, L4, and L5. However, the greater trochanter and fibula are covered by L5, but are clearly uninvolved. This violation of the scelrotomal map raises questions about the pattern of disease distribution being attributed to sclerotomes.

Fig. 3.

Comparison of melorheostosis distribution to sclerotomal map.

A: Comparison of the distribution of melorheostosis in a patient with affected right arm (Melo-21; age 69 years) to the proposed sclerotomal map. With kind permission from Springer Science + Business Media: Skeletal Radiology, Melorheostosis and the Sclerotomes: A Radiological Correlation, Volume 4, 1979, 57–71, Murray and McCredie Figs. 1 and 2.

B: Comparison of the distribution of melorheostosis in a patient with affected right leg (Melo- 10; age 49 years) to the proposed sclerotomal map. The anatomic distribution of melorheostosis does not conform to the scelrotomal map in either of the two patients.

CT analysis for D-V restriction of bone affected with MEL showed that only one of eight patients (Melo-16) with confirmed somatic mutation associated MEL (MAP2K1) showed D-V confinement of the MEL lesion. This may be a consequence of low skeletal burden of MEL and a very low mutant allele frequency at 2.7% (Fig. 4).

Fig. 4.

CT scan showing compartmentalization of melorheostotic lesion in the dorso-ventral axis of the intermediate and lateral cuneiform bones of the left foot, in Melo – 16 (MAP2K1-positive, allele frequency − 2.7%; age 28 years). The short arrow points at the dorsal aspect and the long arrow at the ventral (plantar) aspect of the lesions.

4. Discussion

We found that the distribution of melorheostosis conformed to the distribution of a single sclerotome in only 17% of our patients. This is not consistent with findings reported by Murray et al. [7] who reported that in 19 of 30 patients, MEL conformed to a single sclerotome and in remaining 11 patients, disease involved multiple sclerotomes. Although several publications over the decades have repeatedly attributed the pattern of disease distribution in MEL to sclerotomes as proposed by Murray et al., this has not been specifically investigated by other scientists and the study has not been replicated. We propose that the anatomic distribution of MEL can be explained by the clonal proliferation of a mutation-positive skeletal progenitor cell along the limb axis. This is supported by experiments in mice fetuses where clones of cells which have been genetically modified to produce Lac-Z in clonal cells proliferate along longitudinal axis creating patterns of distribution like MEL (Fig. 2). D-V confinement of melorheostotic lesion in a patient (Melo-16) with low allele frequency and relatively small skeletal burden of the disease suggests that the mutation occurred after the formation of D-V plane in the embryo.

We acknowledge that the disease distribution in about 40% of the patients could be explained by a diseases process following contiguous sclerotomes. However, there are several arguments that make involvement of contiguous sclerotomes a biologically unlikely explanation. First the disease process (be it a mutation, deficit in the nerve, or change in paracrine factor) would have to occur after segmentation and subsequently drift into two segments. Second, we rarely see two entire sclerotomes involved. We more often see the melorheostotic bone bleeding into an anatomic area that is assigned to another sclerotome. In other words, we don’t see two adjacent branches of a tree affected, we see an affected branch and the distal portion of a near branch affected—difficult to explain. Finally, we often see a large portion of a sclerotomal region involved, yet a sizeable part of the sclerotome remains unaffected from the disease—also hard to support.

The pathophysiology of melorheostosis continues to evade researchers. Several possible hypotheses have been proposed to explain the cause of the disease including aberrant interaction between regional nerves and bones [18]. The asymmetrical random distribution of melorheostosis raised the possibility that melorheostosis may represent a form of post-natal neuropathy, affecting segmental spinal sensory nerves like herpes zoster, with resultant scarring of bone instead of skin [7]. In 2018, we found somatic mosaic mutations in MAP2K1 in affected but not unaffected bone in 8 of 15 patients with melorheostosis [3]. The precise identity of the mutated cell remains unclear but it seems likely that the mutation arises in a skeletal progenitor cell. It thus seems biologically appealing to propose that the unique distribution of the disease may be explained by the somatic proliferation of a mutation-positive skeletal progenitor cell along the longitudinal limb axis. This hypothesis also allows us to link the unique anatomic distribution of the disease with post-zygotic somatic mosaicism. While other skeletal disorders associated with somatic mosaic mutations like Maffucci syndrome (IDH1/IDH2) and Ollier disease (IDH1/IDH2), McCune Albright syndrome (GNAS) which affects multiple tissues like skin, bone and soft tissue and Klippel-Trelauney syndrome (PIK3CA) with cutaneous hemangiomas and bone and soft tissue hypertrophy have been described, none of these disorders appear to mimic the pattern of disease distribution seen in MEL. We suspect that this is likely related to the timing of mutation occurrence in the progenitor cell.

The D-V restriction of the pathological bone in patients with low skeletal burden of MEL and low proportion of mutation-positive cells further supports the clonal proliferation hypothesis. We suspect that in most patients with MEL, the mutation occurs in a very early limb bud mesenchymal skeletal progenitor cell such that the occurrence of the mutation precedes the formation of the D-V clonal restriction plane. In mouse fetuses, full D-V lineage restriction appears to take place after 10 days of embryo formation, equivalent to 31–32 days in human embryos [15,19]. It is thus conceivable that somatic mutations which respect D-V restriction will have a lower skeletal burden of the disease. Although, our findings are limited to patients with MAP2K1-positive melorheostosis, we anticipate that this finding can be extrapolated to all patients with somatic mosaicism in affected bone.

Strengths of the study include its relatively large sample size for a rare hyperostotic disease with a prevalence of “one in a million”, availability of whole body CT scans to evaluate the anatomic distribution of the disease and the value of multiple radiologists reviewing each CT scan. In contrast to the study by Murray et al., which used focal x-rays to ascertain sclerotomal distribution, our study uses whole body CT to define the extent of MEL in the entire skeleton. The major limitation of the study is its reliance on studies in mouse fetuses to support the clonal proliferation hypothesis. It is likely that D-V planar compartment has some variations in adult human bones. We are also limited by the small sample size of patients with MEL with confirmed somatic mosaicism in osteoblasts from affected bone.

5. Conclusion

The traditional sclerotomal hypothesis for the pathogenesis of melorheostosis is largely speculative, based on presumed sclerotomal innervation patterns. Here, we demonstrate that CT analysis of thirty patients with MEL challenges the sclerotome hypothesis. Recent work on clonal expansion of single cells in the limb bud in mice fetuses shows that it is possible for the distribution of a single clone to create patterns in limbs that are like the patterns of affected bone seen in MEL. Thus, a single post-zygotic somatic mutation in a mesenchymal cell of a limb bud could result in a pattern of clonal expansion like the patterns of anatomic distribution we see in MEL. This suggests that the clonal proliferation hypothesis can explain the patterns of affected bone that makes this disease so unique and perplexing. While we offer clonal proliferation as an alternate hypothesize, the main objective of this paper is to challenge the sclerotome hypothesis and encourage physicians to assess the pattern of disease distribution in patients.