Melorheostotic Bone Lesions Caused by Somatic Mutations in MAP2K1 Have Deteriorated Microarchitecture and Periosteal Reaction

Abstract

Melorheostosis is a rare non-hereditary condition characterized by dense hyperostotic lesions with radiographic “dripping candle wax” appearance. Somatic activating mutations in MAP2K1 have recently been identified as a cause of melorheostosis. However, little is known about the development, composition, structure, and mechanical properties of the bone lesions. We performed a multi-method phenotype characterization of material properties in affected and unaffected bone biopsy samples from six melorheostosis patients with MAP2K1 mutations. On standard histology, lesions show a zone with intensively remodeled osteonal-like structure and prominent osteoid accumulation, covered by a shell formed through bone apposition, consisting of compact multi-layered lamellae oriented parallel to the periosteal surface and devoid of osteoid. Compared with unaffected bone, melorheostotic bone has lower average mineralization density measured by quantitative backscattered electron imaging (CaMean: –4.5%, p = 0.04). The lamellar portion of the lesion is even less mineralized, possibly because the newly deposited material has younger tissue age. Affected bone has higher porosity by micro-CT, due to increased tissue vascularity and elevated 2D-microporosity (osteocyte lacunar porosity: +39%, p = 0.01) determined on quantitative backscattered electron images. Furthermore, nano-indentation modulus characterizing material hardness and stiffness was strictly dependent on tissue mineralization (correlation with typical calcium concentration, CaPeak: r = 0.8984, p = 0.0150, and r = 0.9788, p = 0.0007, respectively) in both affected and unaffected bone, indicating that the surgical hardness of melorheostotic lesions results from their lamellar structure. The results suggest a model for pathophysiology of melorheostosis caused by somatic activating mutations in MAP2K1, in which the genetically induced gradual deterioration of bone microarchitecture triggers a periosteal reaction, similar to the process found to occur after bone infection or local trauma, and leads to an overall cortical outgrowth. The micromechanical properties of the lesions reflect their structural heterogeneity and correlate with local variations in mineral content, tissue age, and remodeling rates, in the same way as normal bone.

Introduction

Melorheostosis (OMIM %155950) is an extremely rare sporadic sclerotic bone dysostosis with an estimated prevalence of 1 case per million.⁽¹⁾ First described in 1922 by Léri and Joanny,⁽²⁾ the disease is characterized by dense hyperostotic lesions with radiographic appearance of “dripping candle wax” affecting mostly the long bones of the lower and upper extremities and less frequently the axial or craniofacial skeleton.^(3–5) Melorheostosis typically manifests during childhood or adolescence and patients present first with symptoms of pain and joint stiffness, swelling, and deformity of the limb.⁽³⁾ One or several adjacent bones might be affected with or without cutaneous alterations overlying the bone lesion. More than 400 publications have described melorheostosis to date, with the diagnosis based exclusively on the presence of the characteristic dense and eccentric hyperostosis or cortical thickening of the affected bone without any features of bony destruction. Therapeutic management is limited to symptomatic management of skeletal pain because there is no cure.^(4,6,7) A few reports of clinical improvement after zoledronate treatment suggest that increased osteoclastic resorption and/or angiogenesis might be involved in the pathogenesis of the disease.^(3,7,8) Beyond this, the etiology of the exuberant melorheostotic bone overgrowth is yet to be elucidated. Becasue the abnormal tissue has not been well characterized, it is not yet clear which pathways could be targeted to improve the clinical condition of affected patients.

Two postzygotic mosaicisms causing melorheostosis were delineated within the last year. Whyte and colleagues reported a heterozygous somatic KRAS mutation in the scleroderma-like dermatosis overlying melorheostotic bone lesions in a 14-year old boy.⁽⁹⁾ The clinical background of this patient was, however, rather complex, because he was found to suffer additionally from osteopoikilosis, a hereditary bone dysplasia caused by germline inactivating mutation in LEMD3, a protein antagonizing TGF-ß/BMP signaling.^(9,10) Subsequently, Kang and colleagues identified somatic activating mutations in MAP2K1 in bone biopsy samples of eight melorheostotic patients diagnosed by the typical radiological “dripping candle wax” appearance of the lesion.⁽¹¹⁾ Histology of bone lesions revealed intense bone remodeling and—unexpectedly for a sclerosing bone disorder—accumulation of excessive unmineralized osteoid that was not observed in either the contralateral bone biopsy or in melorheostotic bone tissue of patients lacking a MAP2K1 mutation.^(11,12) These findings were consistent with in vitro studies showing that the expression of marker genes for osteoblast differentiation was markedly reduced and mineralization was inhibited in osteoblasts cultured from affected bone tissue.⁽¹¹⁾ Moreover, the ratio of RANKL/OPG transcripts was markedly increased, suggesting that cells with the MAP2K1 mutation stimulate osteoclast differentiation.

However, a phenotype of highly remodeled osteonal-like bone with high osteoid formation does not per se explain the classic radiological presentation of melorheostosis of cortical hyperostosis.⁽³⁾ To obtain further insights into growth and development of the melorheostotic lesions, we now performed a multi-method phenotype characterization of bone tissue properties by comparing affected and unaffected bone biopsy samples from six patients with confirmed MAP2K1 somatic activating mutations. In addition to standard histology, we used polarized light microscopy to investigate the local bone matrix organization and collagen fibril orientation, micro-computed tomography (micro-CT) to evaluate bone tissue porosity indicative for vascularity, quantitative backscattered electron imaging to evaluate the degree of bone matrix mineralization and the osteocyte lacunar density, and, finally, nanoindentation to assess the micromechanical properties of the bone tissue. The results suggest a model for pathophysiology of melorheostosis caused by somatic activating mutations in MAP2K1, in which the actual genetically induced lesion is covered by new bone formed through periosteal reaction similar to the process found to occur after bone infection or local trauma.⁽¹³⁾

Materials and Methods

Patient cohort

This study involved bone biopsy samples from 6 unrelated melorheostosis patients with somatic activating mutations in MAP2K1 presented recently.⁽¹¹⁾ Extensive clinical data were reported elsewhere and are compiled in Table 1.⁽¹²⁾ Briefly, the cohort consisted of 4 females and 2 males. Their average age at biopsy was 46.5 (±6.7) years; the average disease duration was 31.3 (±12.1) years.

Table 1.

Patient Cohort, Site of Biopsy, and Genetic Characteristics

Patient	Sex	Age at biopsy (years)	Age at onset (years)	Site of unaffected bone biopsy	Site of affected bone biopsy	Mutant allele in affected tissue (%)
Melo 02	Male	42	32	Left tibia	Right tibia	6.9
Melo 04	Female	42	10	Left tibia	Right ankle	10
Melo 09	Female	59	12	Left ulna	Right humerus	20
Melo 10	Female	49	8	Left iliac crest	Right tibia	12
Melo 18	Female	45	10	Right iliac crest	Left clavicle	25
Melo 19	Male	42	33	Right iliac crest	Right radius	31

Further details are published elsewhere.^(11,12)

The study protocol was approved by the Institutional Review Board of the National Institute of Arthritis and Musculoskeletal and Skin Diseases (Clinicaltrial.gov NCT02504879).

Bone biopsy samples

The patients underwent open surgical biopsy of the melorheostotic lesion. To minimize risks, only superficial parts of the affected bone tissue were surgically removed. The amount of bone tissue obtained was dependent on the extent and site of the lesion and thus varied considerably among the patients. From patients Melo 04 and Melo 18, two biopsy samples from affected bone were available (those from Melo 04 were obtained within a several-week interval). A contralateral, generally smaller sample of unaffected bone tissue was obtained as a control at the same surgery. Iliac crest was the preferred site. Three patients requested a different site for this elective procedure and were accommodated. Affected and unaffected bone samples were divided and immediately immersed either in culture media for DNA extraction for whole-exome sequencing and cell culture or fixed in 70% ethanol and embedded in polymethylmethacrylate for bone tissue analysis using standard procedures.⁽¹⁴⁾

Micro-CT

Scans were performed on two paired biopsy samples from Melo 04 (both samples from tibias) and Melo 09 (right humerus, affected bone and left ulna, unaffected bone). We used a micro-CT50 (Scanco Medical AG, Brüttisellen, Switzerland) operated at 90 kV with a 0.5 mm Al filter, and 750 ms exposure time per projection. A total of 1500 projections were captured over 180 degrees. Images with 4-μm resolution were obtained after 3D data reconstruction. The images were processed with the ImageJ software (https://imagej.nih.gov/ij/).⁽¹⁵⁾ High- and low-frequency noise was reduced by applying successively a 3D median filter with radius 2 pixel and a 3D gauss filter of size 3. The segmentation was performed using a local Phansalkar threshold with radius 50 and k value 0.15. After segmentation, the images were inverted in order to obtain the pores. The porosity (Po.V/TV, %), pore diameters (Po.Dm, μm), pore spacings (Po.Sp, μm), pore number (Po.N, 1/mm), and degree of anisotropy (DA, unitless) were evaluated using BoneJ.⁽¹⁶⁾ The 3D rendering of the pores was performed using Amira software (v6.5, FEI Visualization Sciences Group, Bordeaux, France). Pore diameters were arranged in histograms corresponding to the contribution of pores with a given diameter to the overall pore volume within the tissue volume. These histograms were further converted into cumulative distributions indicating the total volume of all pores smaller than a given diameter within the tissue. Pore diameters in the range of osteocyte lacunae were not resolved by the applied micro-CT resolution.

Bone histology and histomorphometry

For histological examinations, 4-μm-thick sections were cut from all tissue blocks with a hard tissue microtome (Leica SM2500, Nussloch, Germany), deplasticized with 2-methoxyethyl-acetate and stained with modified Goldner’s Trichrome.⁽¹⁴⁾ A light microscope (Axiophot, Zeiss, Oberkochen, Germany) equipped with a digital camera (AxioCam HRc, Zeiss) was used to obtain digital images of the sections that were analyzed using the ImageJ software.⁽¹⁵⁾ Additionally, the bone matrix organization was observed under polarized light. Using custom-made macro in ImageJ software,⁽¹⁵⁾ bone histomorphometry analyses were performed on four randomly chosen areas throughout each bone section of unaffected bone and the osteonal region of affected bone.

Quantitative backscattered electron imaging (qBEI)

Subsequently, the residual blocks were prepared for qBEI by grinding and polishing to obtain plane parallel surfaces and by carbon coating.⁽¹⁷⁾ The entire cross-sectioned bone sample area was imaged with a spatial resolution of 1.8 μm per pixel using a field emission scanning electron microscope (FESEM) (Zeiss Supra 40, Oberkochen, Germany) equipped with a four-quadrant semiconductor backscatter electron detector. The FESEM was operated with an electron energy of 20 keV. The gray levels of the obtained images reflecting the mineral/calcium content were calibrated by the material contrast of pure carbon and aluminum. Five variables were used to describe the bone mineralization density distribution (BMDD):⁽¹⁸⁾ CaMean: the average calcium concentration (weighted mean); CaPeak: the most frequently occurring calcium concentration (the peak position of the BMDD) in the sample; CaWidth: the width of the BMDD distribution (full width at half maximum) reflecting the heterogeneity in matrix mineralization; CaLow: the percentage of low mineralized bone area, which is mineralized below 17.68 weight% calcium, reflecting bone areas undergoing primary mineralization; and CaHigh: the percentage of highly mineralized bone matrix, having the calcium content above 25.30 weight% calcium.

Osteocyte lacunae sections (OLS) characteristics

Osteocyte lacunae were evaluated in all samples using qBEI analysis and thus investigated parameters refer to 2D sections through the 3D lacunae. The OLS size and shape were analyzed with a custom-made macro in ImageJ as previously described.⁽¹⁹⁾ Six to 12 qBEI scans with 130× nominal magnification (0.88 μm/pixel) were performed throughout the section to characterize the OLS by five parameters: OLS-density, the number of OLS per mineralized bone matrix area; OLS-porosity, OLS total area/(mineralized bone matrix area + OLS total area); OLS-area, mean value of the OLS areas per sample; OLS-perimeter, mean value of the OLS perimeters per sample; and OLS-AR, mean value of the OLS aspect ratio (AR) per sample. AR = 1 indicates a perfect circle and increasing values indicate increasingly elongated shape of the OLS. AR values >10 were excluded. The OLS were discriminated from the surrounding mineralized bone matrix and from the osteonal channels by setting a fixed gray level threshold at 55 (Ca wt% 5.2) and a size range between 5 μm² and 80 μm².

Nanoindentation

Nanoindentation was performed on bone samples taken from affected and non-affected sites from 3 patients (Melo 02, 04, 10) with a Berkovich-type diamond indenter of a TriboScope nanohardness tester (Hysitron Inc., Minneapolis, MN, USA). The sample was mounted on a motorized table that allows for movement in the plane normal to the axial motion of the tip. Rectangular regions of interest were selected predominantly in remodeled osteonal bone, according to previously taken backscattered electron images. These areas were scanned by nanoindentation with a step size ranging from 20 μm to 200 μm, depending on bone area size. The number of indents performed per sample varied from 145 to 546. This was done to obtain average values of hardness (H) and indentation modulus E_r. A load function with the following characteristics was used: 5000 μN maximum load, loading at a rate of 1000 μN/s, holding at maximum force for 60 seconds and unloading to 1000 μN at a rate of − 400 μN/s, a second holding time of 10 seconds, and finally unloading to 0 mN at a rate of −200 μN/s. The indentation modulus was calculated using the method of Oliver and Pharr as reported previously.^(20,21) It is calculated from the slope of the unloading curve in the region between 20% and 95% of the maximum load. The hardness H and indentation modulus E_r were then derived from the unloading contact stiffness S, and the indenter contact area Ac. Data points with E_r < 8 correspond to PMMA and were therefore excluded from the analysis.

$$H=\frac{P_{\max }}{A_c}{E}_R=\frac{\sqrt{\pi }}{2}\frac{S}{A_c}$$

Statistical evaluations

Statistical evaluation was performed with GraphPad Prism 6.0h (GraphPad Software, Inc, LaJolla, CA, USA). Comparison of histomorphometric parameters, BMDD, and osteocyte lacunae indices were based on paired t test (affected versus unaffected tissue in each patient). Comparison of hardness and indentation modulus between unaffected and affected tissue obtained from the same patient was evaluated by nonparametric Mann-Whitney test. Statistical significance was considered as p ≤ 0.05. Correlation analysis of BMDD with histomorphometric parameters or nanoindentation indices were done by Pearson analyses.

Results

All results were obtained from bone biopsy samples of melorheostosis patients with MAP2K1 mutations, comparing affected and contralateral unaffected bone from the same patient. MAP2K1 mutation allele frequency in affected bone tissue ranged from 6.9% to 31% (mean 17.5% (±9.4%) versus 0% in unaffected bone tissue, as determined by droplet digital PCR.⁽¹¹⁾

Histology of affected bone shows periosteal reaction followed by intense remodeling

Results of bone histomorphometry have been reported previously and are here summarized briefly.⁽¹¹⁾ Compared with unaffected bone samples, affected bone has markedly elevated number of osteoblasts and osteoclasts. Moreover, there is a more than sixfold increase in eroded surface and osteoid thickness, while osteoid surface/bone surface was found to be more than 50 times higher in affected bone.

Fig. 1 gives an overview of the typical microstructural differences between unaffected cortical bone (Fig. 1A–C) and melorheostotic lesions (Fig. 1D–H), both from tibias.

Fig. 1.

Overview of principal bone-tissue features of affected and unaffected bone. The cartoon in the upper part of the figure depicts a long bone with a classical “dripping candle wax” appearance and the development of the melorheostotic lesion. The images show typical features observed in unaffected (A–C) and affected bone (D–H). Unaffected bone consists of osteonal remodeled bone (OR), whereas in affected bone a zone of parallel lamellar bone formed through periosteal apposition (PA) and a zone of osteonal remodeled (OR) bone can be distinguished. Backscattered electron images from biopsy samples from unaffected (A) and affected (D) bone (Melo 04 and Melo 02, respectively, both tibia). Higher mineralized bone area appears brighter, whereas lower mineralized bone is rather dark gray. Both samples show typical features of osteonal remodeled (OR) bone. Affected bone is covered on the periosteal side by a band of nonporous bone (PA). (B–H) Histological images obtained from the above samples, Goldner staining: (B, E, G) OR and PA bone viewed in polarized light: unaffected (B) and affected (G) bone show concentric lamellae surrounding Haversian channels. PA is only viewed in affected bone and consists of compact multilayered lamellae oriented parallel to the periosteal surface over a large distance (E). (C, F, H) OR and PA bone viewed in transmission light: Red shows unmineralized osteoid, green mineralized bone. Unaffected OR (C) shows no osteoid, whereas osteoid and bone tissue porosity are highly increased in affected bone (F). In contrast, PA bone is rather compact, although single small channels with osteoid are already visible.

In unaffected bone, the backscattered electron image (Fig. 1A) shows dense cortical tissue with pores of rather constant size corresponding to Haversian canals surrounded by typical concentric lamellae as shown by polarized light microscopy (Fig. 1B). Transmission light microscopy of this sample reveals neither osteoblasts nor osteoclasts and absence of thick osteoid formation (Fig. 1C).^(11,12)

In contrast to this homogenous structure in unaffected bone, two strikingly different regions can be distinguished in the melorheostotic lesion. First, an outer shell with little porosity, and second, within the deeper region, a zone consisting of more porous bone (Fig. 1D). In polarized light, this outer shell appeared to be formed by long, compact multilayered lamellae, orientated strictly parallel to the bone surface (Fig. 1E). The parallel lamellar structure corresponds to bone formed through periosteal apposition (PA), whereas the deeper area shows concentric bone lamellae, a typical feature of osteonal remodeling (OR) (Fig. 1E, G).^(11,12) In sharp contrast to the unaffected bone, the deeper region of the melorheostotic lesion is characterized by high osteoblast and osteoclast activity and prominent osteoid formation (Fig. 1F, H), which is never observed in unaffected samples. The overall picture of events is that melorheostotic lesions are expanding laterally through robust primary bone deposition by the periosteum; this new compact bone tissue subsequently becomes perforated, vascularized, and remodeled (Fig. 2A, B). Newly formed channels can also be “closed” by filling the cavity with disorganized woven bone, leading to areas of secondary compact bone (Fig. 2C, D).

Fig. 2.

Histological images from affected bone at the transition of PA to OR. (A, B) Detail from affected bone from Melo 18. Identical region viewed in transmission light (A) and in polarized light (B): large vascular channels (asterisks) with prominent osteoid (red) perforate a zone of parallel lamellar bone, clearly visible in polarized light (white thick arrows). (C, D) Detail from affected bone from Melo 02. Identical region viewed in transmission light (C) and in polarized light (D): Closed vascular channel throughout an area of parallel ordered lamellae. The “footprint” of the vascular channel, delineated with broken lines, is recognizable by leftover of unmineralized matrix that appears now as “buried” osteoid (red, asterisk) (C) or as highly disorganized woven (D). Note in A and C that parallel lamellar bone is devoid of osteoid.

Bone tissue vascularity is significantly increased in affected bone

To compare porosity due to bone vascularity in affected and unaffected bone, we evaluated the region of osteonal remodeled bone (OR) of the melorheostotic lesion by micro-CT with a nominal resolution of 4 μm, which is too large to resolve osteocyte lacunae.

The 3D rendering of vascular channels in unaffected and affected bone from Melo 04 and Melo 09 (Fig. 3A) clearly shows that the porosity was elevated in the affected bone of both individuals compared with the unaffected bone. This observation was confirmed quantitatively. The number of pores (1/mm) was highly increased in affected bone (3.66 and 4.55) versus unaffected bone (2.75 and 2.16), while the mean pore spacing was reduced: affected bone (191.9 ± 50.3 and 137.0 ± 35.8 μm) versus unaffected bone (311.9 ± 84.7 and 318.5 ± 106.9 μm). The distribution of pore diameters was maximum at about 50 μm in all samples, with the exception of one particularly large channel in the unaffected bone tissue of Melo 09 that resulted in an extra peak at larger pore volume/per tissue volume in the latter sample (Fig. 3B). The cumulative histogram of the contribution of each pore size to the total porosity volume (Fig. 3C) shows a larger overall pore volume per tissue volume for the affected tissue. Only the pore volume of Melo 09 is also large due to the wide vessel in this specimen (arrows in Fig. 3). The degree of anisotropy of the pores in the unaffected samples was close to 1, which denotes a strong preferential orientation of the pores as observed in osteonal bone with the osteons oriented in the length axis of the bone. In contrast, the orientation of the pores in the affected samples was more heterogeneous with a degree of anisotropy of 0.6 to 0.7.

Fig. 3.

Micro-CT evaluations of bone tissue vascularity. (A) 3D renderings of cortical canals in unaffected and affected bone from Melo 04 and Melo 09 in a cube of edge length 1.5 mm arbitrarily chosen after micro-CT scans. The calcified tissue was segmented and then the images were inverted for the 3D visualization of the pores. Tissue porosity appears clearly increased in affected versus unaffected bone. Note the large vessel present in the scanned volume of the unaffected bone sample from Melo 09 (arrow). The nominal scan resolution is 4 μm. (B) Histogram of the contribution of pore size to the total pore volume per tissue volume. (Arrow: large vessel in the scanned volume from the unaffected bone tissue sample from Melo 09). (C) Cumulative histogram of the contribution of pores smaller than a certain size to the total pore volume per tissue volume. (Arrow: large vessel in the scanned volume from the unaffected bone tissue sample from Melo 09.) The colors of the labels in A (unaffected: Melo 04, Melo 09; affected: Melo 04, Melo 09) correspond to the colors of the lines in B and C.

Bone matrix mineralization is lower in affected than in unaffected bone

Quantitative backscattered electron imaging results obtained from the total surface of the sectioned bone area of all samples (paired biopsy samples from all 6 patients) revealed that affected bone tissue has lower average mineralization density (CaMean: −4.57%, p = 0.035) and nearly twice the portion of matrix undergoing primary mineralization (CaLow: +87%, p = 0.009) as unaffected bone (Table 2). The effect of MAP2K1 mutation on bone matrix mineralization is illustrated in Fig. 4A and B showing the results from Melo 04 and Melo 09: the backscattered electron microscope images of affected bone is very similar in both patients. Comparison of affected versus unaffected bone showed that the BMDD curve shifted toward a lower degree of mineralization in the melorheostotic lesions.

Table 2.

Comparison of Bone Tissue Characteristics (Total Bone Area) From Melorheostotic Lesions Versus Unaffected Bone

Bone sample	Unaffected (n = 6)	Affected (n = 8)a	p Value
BMDD parameters
CaMean (weight % calcium)	23.23 (1.20)	22.17 (0.49)	0.035*
CaPeak (weight % calcium)	24.02 (1.23)	23.70 (0.39)	0.43 (ns)
CaWidth (Δ weight % calcium)	4.31 (0.66)	4.87 (0.68)	0.15 (ns)
CaLow (% bone area)	4.87 (2.07)	9.11 (2.52)	0.009**
CaHigh (% bone area)	26.03 (19.37)	15.55 (4.02)	0.11 (ns)
OLS parameters
OLS-density (number OLS/mm2)	216.3 (70.34)	277.9 (35.73)	0.050
OLS-porosity (%)	0.572 (0.15)	0.797 (0.06)	0.013*
OLS-area (μm2)	27.04 (1.75)	29.07 (1.74)	0.03*
OLS-perimeter (μm)	17.76 (1.08)	17.78 (1.82)	0.93
OLS-aspect ratio	2.62 (0.41)	2.55 (0.27)	0.71

BMDD = bone mineralization density distribution; OLS = osteocyte lacunae sections.

Values are given as mean (±SD).

^aFor Melo 4 and Melo 18, two affected samples were available and evaluated separately.

^*p ≤ 0.05

^**p < 0.01

^****p < 0.0001 (results from paired t test).

Fig. 4.

Comparison of BMDD between affected and unaffected bone. (A) Backscattered electron microscope image of affected bone biopsies obtained from Melo 04 (left) and Melo 09 (right). Note that despite the differences in size and shape of the sample, the bone tissue structure seems very similar. Scale bar = 1000μm. (B) Corresponding histograms show that the BMDD curve from affected bone is shifted to the left, ie, toward lower mineral content compared with unaffected bone in both individuals. (C) Correlation of osteoid indices obtained by bone histomorphometry and CaLow obtained by qBEI: The percentage of bone area with low calcium concentration (CaLow), representing the areas undergoing primary mineralization is highly correlated with osteoid thickness and surface. Note that affected and unaffected bone separate in two groups. Unaffected bone (that has very low bone formation^(11,12)) has little osteoid formation and reduced CaLow. Biopsy samples were obtained from 6 patients. From Melo 04 and Melo 18 two samples of affected bone were available.

A correlation analysis between the BMDD parameter CaLow and osteoid indices evaluated by bone histomorphometry of all bone samples (pooling affected and unaffected biopsies) revealed a strong positive correlation between CaLow and osteoid surface, as well as osteoid thickness (r = 0.67, p = 0.008; 0.68, p = 0.009, respectively) (Fig. 4C).

Mineral content is reduced in periosteal versus remodeled bone in melorheostotic bone

Because affected bones from Melo 02, Melo 04, and Melo 18 contained sufficient intact periosteal surface, we evaluated BMDD separately in the parallel lamellar bone (PA) and in the osteonal remodeled (OR) bone. Within a melorheostotic lesion, CaMean in PA was reduced up to 12% compared with the adjacent OR bone (Table 3). Consistently, the BMDD curve from PA shifted toward a lower degree of mineralization compared with OR (Fig. 5). The lower mineral content in PA is mirrored by lower CaPeak and CaHigh values, while there is no clear trend for CaLow in OR (Table 3).

Table 3.

BMDD Parameter Findings in Bone Formed Through Periosteal Apposition (PA) and Osteonal-like Bone (OR)

	Melo 02-affected		Melo 10-affected		Melo 18-affected
BMDD parameter	PA	OR	PA	OR	PA	OR
CaMean (weight % calcium)	19.41	22.17	21.95	22.73	20.88	22.77
CaPeak (weight % calcium)	20.97	23.57	22.70	23.92	22.18	24.09
CaWidth (Δ weight % calcium)	5.72	5.37	4.68	3.81	5.03	4.51
CaLow (% bone area)	23.19	9.40	5.05	5.34	11.92	6.68
CaHigh (% bone area)	1.76	16.49	5.21	12.75	3.73	18.52

BMDD = bone mineralization density distribution

Fig. 5.

Comparison of BMDD between periosteal and osteonal remodeled bone. (A, B) Backscattered electron microscope images from affected bone from Melo 05. Same sample as in Fig. 1D at larger magnification. Higher mineralized bone area appears brighter, whereas lower mineralized bone is rather dark gray. (A) Outer part of the lesion, bone formed through periosteal apposition (PA). The bone tissue appears less porous and darker (less bright) than the osteonal remodeled (OR) bone in the deeper region of the melorheostotic lesion (B). Note the large vascular channels in OR (asterisks) are often surrounded by a dark seam representing newly formed, lowly mineralized bone formed through osteonal remodeling. Scale bars = 250 μm. (C) BMDD curves from bone formed through periosteal apposition (PA) and through osteonal remodeling (OR). Note that the BMDD peak from PA is shifted toward lower mineral content in comparison to OR.

Osteocyte lacunar porosity is increased in melorheostotic bone

2D-microporosity showed a significant increase in the osteocyte lacunar porosity (OLS-porosity) in affected bones (+39%; p = 0.01), due to a combined increase of OLS-density (+28%; p = 0.05) and OLS-area (+8%; p = 0.03). In contrast, OLS-perimeter and OLS-AR were similar in affected and unaffected bones (Table 2).

Micromechanical properties of melorheostotic bone correlate with its mineral content

The relationship between mechanical properties and mineral content is plotted for each sample in Fig. 6: median values of hardness (Fig. 6A) and indentation modulus (Fig. 6B) against the typical calcium concentration (CaPeak). We found a strongly significant positive correlation with r = 0.8984, p = 0.0150 and r = 0.9788, p = 0.0007, respectively. This indicates that micromechanical properties are primarily determined by the mineral content of the bone material.

Fig. 6.

Bone material properties assessed by nanoindentation. (A, B) Correlation between the mode of the BMDD peak (CaPeak, most frequently occurring calcium concentration) and hardness (A) and indentation modulus (B). Note that the micromechanical properties of bone tissue are highly dependent on mineral content of the bone, independently of the mutation. (C, D) Example of histograms of hardness (C) and indentation modulus (D) values showing their distribution in the measured area of Melo 04. Both histograms are clearly shifted to smaller values in the affected areas.

In Melo 02 and Melo 04, hardness and indentation modulus were significantly decreased in melorheostotic versus unaffected bone (p < 0.0001). The histograms of these mechanical parameters clearly shifted toward lower values at the affected sites (shown for Melo 04 in Fig. 6C and D). In Melo 10, however, the hardness and indentation modulus as well as mineral content are found to be slightly but significantly increased at affected sites compared with unaffected bone. Note that the unaffected bone sample from Melo 10 was obtained from iliac crest and not from the cortex of limb bone (Table 1).

Discussion

Melorheostosis is a rare non-inherited dysostosis that is often described in the literature as a “sclerosing” bone dysplasia.^{(3,5,22–25)} Radiographs of melorheostosis show dense, overgrown, and expansive cortical bone. We previously described the distinct histology of melorheostotic lesions with “dripping candle wax” appearance caused by somatic mutations in MAP2K1.^(11,12) We showed that the hyperostotic tissue consists of two morphologically distinctive regions: a shell formed through periosteal bone apposition consisting of multilayered lamellae oriented parallel to the outer surface, which covers a distinctive zone of osteonal-like bone. Furthermore, melorheostotic bone appeared more porous than unaffected bone. Here, we expanded our analysis of the bone lesion by quantifying vascular and osteocyte lacunar porosity, the degree of bone matrix mineralization, and micromechanical properties.

Our evaluations indicate that melorheostotic lesions caused by activating mutations in MAP2K1 are the product of sustained processes of bone modeling and remodeling resulting in a heterogeneous microstructure in terms of collagen fibril orientation, porosity, and mineral content. The classic radiological feature of melorheostosis,⁽³⁾ a thickening or eccentric hyperostosis of the cortex, is shown here to result from robust focal de novo bone formation by the periosteum, identified in polarized light microscopy as multiple layers of mineralized collagen fibrils oriented parallel to the periosteal surface. This primary lamellar bone is compact and devoid of osteoid. With increasing distance from the periosteal surface, this dense structure becomes perforated by numerous vascular canals and, conversely, there is a gradual decrease of parallel lamellar bone, which becomes highly remodeled into an osteonal-like structure with typical concentric lamellae surrounding newly formed channels. In sharp contrast to unaffected cortical bone, affected osteonal-like bone exhibits abundant osteoclasts, osteoblasts, excessive osteoid formation, and high porosity. Consistently, the number of vascular channels in the lesions was shown by micro-CT to be markedly elevated, suggesting that angiogenesis might also play an important role in the pathophysiology of melorheostosis, as it does in tumors caused by hyperactivation of the MEK1-ERK1/2 pathway or in extracranial arteriovenous malformation.^(26,27) Although the median pore diameter of 50 μm, corresponding to the typical size of Haversian channels, was similar in affected and unaffected bone, enlarged vascular spaces with multiple blood vessels were also often viewed in melorheostotic bone.⁽¹²⁾ However, there is a notable difference in the orientation of the channels: in unaffected bone, we observed a predominant alignment of the Haversian channels and concomitantly osteons along the longitudinal axis of the bone. In contrast, in affected bone, the channels are much less oriented, mirroring a rather isotropic outgrowth of the lesion. Interestingly, we also captured refilled, and thus closed, channels that were presented in histological sections as islands of disordered woven bone with buried osteoid throughout areas of parallel-oriented bone lamellae.

The osteocyte lacunar density, evaluated by 2D analysis of qBEI, was increased by nearly 40%, which is possibly related to enhanced cell proliferation and/or altered cell differentiation, as observed in cell cultures from affected bone.⁽¹¹⁾ It has been speculated indeed, that the quantity of osteocytes generated during bone formation depends, on the one hand, on the number of preexisting osteoblasts and, on the other hand, on their matrix synthesis activity.^(28,29) Thus, in melorheostotic bone, more osteocytes might be encased in the matrix because a larger osteoblastic pool is present and/or because less matrix is synthesized at the single-cell level, which would be in line with a reduced COL1A1 expression observed in affected osteoblasts.⁽¹¹⁾ Of note, pre-osteoblasts and osteocytes both produce RANKL, the key protein of osteoclastogenesis; an increased ratio of RANKL/OPG transcripts was previously demonstrated in osteoblast cultures from affected bone.^(11,30,31) Taken together, our bone tissue and cell culture evaluations show very consistently that activating mutations in MAP2K1 lead to metabolically overactive bone, in terms of increased turnover and remodeling, in line with increased tracer uptake on bone scintigraphy.^{(3,11,12,32,33)}

An important consequence of combined intense bone formation and remodeling activity is that affected bone has an obviously much younger average tissue age than unaffected bone, which has low turnover as shown by bone histomorphometry.^(11,12) Because younger bone has lower mineral content (due to the time course of mineral accumulation in newly formed bone),⁽³¹⁾ this would also explain the distinctly reduced bone matrix mineralization in affected bone, as measured by qBEI (CaMean: −4.5%, p = 0.04). Moreover, CaLow, a BMDD parameter that represents the percentage of bone area of lowest mineral content, was nearly doubled in the melorheostotic lesions. Such an increase in CaLow is generally observed in situations of high bone formation and/or when the onset of mineralization is delayed and osteoid accumulates.^(18,34,35) We further explored the relationship between BMDD and bone histomorphometry parameters and found, all data pooled, a strong positive association between CaLow and osteoid thickness and surface. It is noteworthy that the values clustered as two groups: unaffected bone has low osteoid and low CaLow, whereas affected bone has high osteoid formation and a high proportion of lowly mineralized bone. The latter mirrors the extensive undermineralization of the matrix in the osteonal-like region of the affected bone formed through remodeling processes.

Material stiffness and hardness of bone are normally dependent on mineral content and on the properties of the organic matrix.⁽³⁶⁾ We evaluated the local mechanical parameters by nanoindentation of affected as well as unaffected bone from melorheostosis and found a nearly perfect positive correlation between indentation modulus and typical calcium content (CaPeak), independent of the skeletal site and MAP2K1 mutation (r = 0.978, p = 0.0007). This demonstrates that changes in the micromechanical properties are predominantly due to changes in the mineral content of the bone matrix as evaluated with qBEI and not due to specific alterations in the organic matrix of the melorheostotic lesion.

At first sight, the lower mineralization density combined with nanomechanical properties that depend entirely on mineral content seems counterintuitive, given the clinical description of melorheostosis. Melorheostosis is considered as a “sclerosing” bone disorder, often described as “more dense bone,” which would suggest more highly mineralized and harder bone as in some other skeletal dysplasia.^(23,37,38) This is also supported by clinical observation of an extreme surgical hardness of the bone of the lesion.⁽¹¹⁾ Our present work suggests that this extreme resistance of the melorheostotic material is not due to abnormal mineral accumulation in the lesion or to a modified and thereby stronger bone matrix, but rather to the periosteal reaction, which covers the lesion with parallel lamellar bone that is known to be resistant against fracturing and cutting and, therefore, likely also against surgical excision.^(39,40)

Pathological states often elucidate previously hidden biological pathways. Melorheostosis is generally considered as a disorder of excessive bone formation. However, our in vitro^(11,12) and bone tissue studies suggest that melorheostosis is also a disorder of excessive bone remodeling. In fact, the new bone formation leading to the (out)growth of the melorheostotic lesion is laid down by the periosteum and totally lacks the typical features associated with the activating mutation in MAP2K1. The primary lamellar bone is compact and shows no abnormal osteoid accumulation. Interestingly, our qBEI analysis revealed an even lower degree of mineralization in this compact bone than in the osteonal-like bone, which is a strong indication that the parallel lamellar bone is younger and thus more recently formed than the remodeled bone. A very similar situation of robust periosteal apposition of lowly mineralized parallel lamellar bone was recently reported in the skull of a patient with an activating LRP5 mutation.⁽³⁷⁾ It is well known that the inner layer of the periosteum, the cambium, contains numerous blood vessels and osteogenic cells that allow radial bone expansion during skeletal growth. Thus, the periosteal bone-forming potential remains throughout life and bone formation is reactivated in metabolically critical situations.^(13,41) In fact, periosteal cells can lay down either highly organized and lowly mineralized lamellar bone as observed in the melorheostotic lesions or highly disorganized and highly mineralized woven bone. The latter is formed when a bony scaffold has to be quickly formed during fetal development, postnatal growth spurt, in callus during fracture healing, or after osteotomy.^(19,42–45) In contrast, periosteal lamellar bone deposition occurs much more slowly, in agreement with a progression of melorheostotic lesions over years.^(44,46,47) Periosteal new bone formation is often viewed as a physiological response to focal insults or systemic condition such as infections, hemorrhages, inflammations (periostitis), osteonecrosis, tumor processes, and malign neoplasms such as Ewing’s sarcoma or Caffey’s disease.^(13,48–51) In fact, so-called “periosteal reactions” occur widely and represent a rather nonspecific mechanism to maintain bone strength, counteracting the effects of cortical lysis or endocortical bone loss by forming a bridge over damaged cortical area.⁽¹³⁾ Activating MAP2K1 mutations lead to highly proliferative osteoblasts, poor matrix mineralization, accelerated bone remodeling, and highly increased osteoclastogenesis.^(11,12) Such bone tissue would be rapidly resorbed with dramatic consequences for bone integrity. One could speculate that as a physiologic response, primary compact lamellar bone is laid down by periosteal cells that may not even harbor the mutation. However, due to the presence of MAP2K1-mutation positive cells, this newly formed bone becomes gradually remodeled and replaced by highly porous melorheostotic bone, which again generates a periosteal reaction in an irregular pattern according to the MAP2K1 somatic mosaicism, leading over time to the typical dysostotic bone with “dripping candle wax” appearance. We do not currently know the factors that drive the periosteal reaction in the melorheostosis lesion. It is, however, interesting that inflammation seems to act as a strong local stimulus.^(13,49) In that sense, it seems possible that inflammatory cytokines produced during overactive bone remodeling also trigger periosteal bone formation.

In conclusion, our investigations have shown that activating mutations in the MAP2K1 oncogene lead to metabolically overactive bone with a highly vascularized network and elevated microporosity. The outgrowth of the melorheostotic lesion apparently results from a periosteal reaction consisting of the deposition of newly formed compact bone as a physiological response to counteract the perforation and progressive loss of the underlying tissue. The micromechanical properties of the bone material correlate with the local variations in mineral content, depending on modeling and remodeling rates in the same way as normal bones, thus indicating that the intrinsic properties of the organic matrix are not altered. Further exploration of the role of angiogenesis in the pathophysiology of melorheostosis with activating mutations in MAP2K1 and the testing of MEK1 inhibitors as a therapeutic approach are warranted.