Discussion
Our analyses revealed that the mineral present in the calcinosis lesions of juvenile and adult myositis patients consists of carbonate apatite and not hydroxyapatite or fluoroapatite, as it was referred to in earlier case reports of surgically removed calcinosis specimens from patients with calcinosis cutis, JDM and DM, using XRD [
9‐
12]. Nielsen and colleagues [
10] stated that the accuracy of the selected area diffraction method used in their study was not sufficient to distinguish between hydroxyapatite and or fluoroapatite. The mineralized regions also contained small amounts of proteins that were significantly lower than those present in bone (Table
2), as determined in the FTIR-RM analyses. A higher mineral-to-protein ratio than that of bone has been previously reported in deposits from five JDM patients, using FTIR spectroscopy on powders from ground specimens and FTIR imaging [
8] that was performed on small 2 to 4 μm sections, while we mapped the whole specimens. The Ca/P molar ratios that were determined in two regions of Calc1 and one region of Calc2 with SEM/SDD-EDS (Table
2) were close to that of stoichiometric hydroxyapatite (1.67). Those results add important new information about the composition of the calcified nodules in this clinical disorder.
XRD performed on two of the five specimens that were analyzed with FTIR-RM (Calc7 and Calc8) showed that the mineral is highly crystallized apatite. The crystallographic results (XRD) corroborate those of the elemental analysis (SEM/SDD-EDS). Similar FTIR spectra were obtained from calcified nodules present in the five different specimens that were subjected to FTIR-RM and Ca/P ratios obtained with SEM/SDD-EDS from two of these five specimens were very similar between the two specimens and between the two large calcified regions in the same specimen (Calc1). These findings indicate that although the deposits differ in their sizes, internal distribution and microstructure, their chemical composition is the same.
The FTIR-RM and XRD data in this study, in confirming the carbonated apatitic nature of the mineral phase - which is the nature of the mineral in normal calcifying tissues (enamel, dentin, cementum, cartilage and bone) - indicate that it is safe to make direct assumptions about the mineral content from comparison of the qBSE signal levels from these different tissues. The findings that the crystallinity (by XRD) of the apatite was more than two times higher than that of bone and dentin and closer to that of enamel, that the Ca/P ratios (by SEM/SDD-EDS) are close to that of stoichiometric hydroxyapatite, and that the deposits contained significantly less organic matrix than bone and dentin (5% to 38% relative to that present in bone in representative spots in specimens obtained from various patients, by FTIR-RM), correlate with the results obtained from the qBSE imaging that the mineral density of the calcified areas was much higher than bone, often higher than dentin and the more highly mineralized regions within calcified cartilage, and closer to that of enamel (Table
2). This also correlates with the nanoindentation measurements, in that the stiffness values of the highly mineralized regions in the analyzed calcinosis samples (33.7 ± 6.9 and 29.0 ± 6.7 GPa) are greater than human dentin (20 GPa to 27 GPa) [
33] and normal human bone (16.8 ± 2.8 GPa) [
34] and below those of enamel (70 GPa to 100 GPa) [
35]. This also correlates with the histological findings of an absence of collagen (the main component of the organic matrix in bone and dentin) in highly mineralized areas.
Earlier microscopic examinations support our PLM findings of the scarce presence of collagen inside the calcified regions. Paegle [
9] proposed that the random orientation of the long axes of the crystals, their large size and lack of association with collagen and elastin, suggested that the crystallization occurred in the extracellular spaces, mitigating against the direct role of collagen in mineral deposition in calcinosis coutis. Nielsen and colleagues [
10] claimed that they demonstrated with electron microscopy (EM) that only elastic fibers calcified and not collagen in the case of a five-year-old girl suffering from JDM and universal calcinosis. Kawakami and colleagues [
11] reported that the calcified deposits formed in the site of fibrinoid degeneration and were located partly on collagen fibers, but mainly on cell debris in foci of degeneration. Globular and/or membranous structures (originated from the degenerated cells of the stroma) that were observed in these calcified regions, were hypothesized to be concerned with initial calcification in this case of calcinosis universalis associated with dermatomyositis. Landis [
36] showed that in an adult DM case, the long axes of the mineral rods followed the long axes of associated collagen, but without a definitive structural relation with the collagen period.
Our study demonstrates the advantages of FTIR-RM as a non-destructive method capable of mapping large thick specimens (as big as 120 × 25 × 14 mm). Full DM specimens (as large as 14 × 10 mm) were fully mapped at various planes and subsequently could be remounted and analyzed by XRD, μCT and SEM/SDD-EDS, while the only published results of FTIR imaging of a JDM specimen presented an FTIR image obtained from an area of 312 × 375 μm of a 2 to 4 μm thin section [
8]. In our study, the XRD analyses performed at the same areas on the same specimens corroborated the FTIR-RM findings, confirming that the only inorganic compound present in the DM deposits was indeed apatite and that it was much more crystallized than bone. The FTIR-RM measurements in our study were performed in a manner to resolve the spatial distribution of both the mineral and the organic matrix, as well as the distribution of the deposits within the specimens. In this way, we fully determined the composition, distribution and spatial orientation of the mineral in the cross sections of the whole specimens, as well as in various depths of the bulk. The μCT performed on three specimens confirmed that the distribution of the calcified deposits seen in the 2D FTIR-RM maps represented very well the 3D network of the deposits.
This is the first report of the use of SEM/SDD-EDS to perform detailed elemental analyses and mapping on DM deposits. The similarity of all the nodules (including the tiny particles) in the Ca, P, Na and Mg maps (Figures
1i,
2f and
2g) indicate that these elements precipitated together in the same spots, forming one homogenous mineral composed of carbonate apatite (FTIR-RM, XRD) containing Na, Mg and trace amounts of Cl, Al and S. Earlier X-ray micro analysis study detected Ca, P and small amounts of Cl and S in the deposits of a JDM and universal calcinosis case [
10]. No values for the amounts of these elements were reported. In addition, an electron probe microanalysis demonstrated that the calcified deposits in a case of calcinosis universalis associated with DM [
11] consisted mainly of Ca and P and that stromal matrices contained traces of Ca and P and small amount of S. The S peak existed mainly in the stromal tissues. No values for the amounts of these elements were reported either, while the two deposits obtained from JDM and JPM patients that we mapped with SEM/SDD-EDS, consisted mainly from Ca, P and O with Na and Mg as minor elements and trace amounts of Cl, Al and S, and their relative amounts were determined (Table
2). The finding of Mg in these DM calcinosis deposits is novel and of interest, as Mg is a known physiological mineralization inhibitor [
37].
Our XRD analyses were performed on particles that were scraped from the white chalky part of a native unprocessed specimen (Calc7) and on the highly mineralized regions (white parts) of the cross sections of embedded and polished whole native specimens (Calc7 and Calc8) and not on fully ground specimens that may include proteins and other organic materials from the surrounding tissue, that are not necessarily part of the calcified regions as was performed in earlier studies [
8,
11]. Our study shows patterns of apatite that is much more crystallized than bone and dentin, with highly resolved peaks in the 30 to 34° 2θ region. A published XRD pattern [
12] obtained from a 0.5 × 0.5 mm area of a JDM sample with synchrotron transmission XRD also showed more resolved peaks than in bone, while XRD patterns similar to bone with c-axis length in the same range as in bone were reported in fat and muscle samples from a JDM patient in another study [
8]. The quantitative XRD results obtained in our study (Table
2) cannot be compared directly to these of Stock and colleagues [
12], because the width of the 002 peak was not reported in that publication. Crystallite sizes of 220 Å and 240 Å were estimated for one JDM sample and for a trabecular bone sample (respectively) using the FWHM of the 002 peak in Scherrer's equation [
12].
This is the first report of the use of qBSE imaging to study the detailed morphology and mineral density of soft tissue calcification with such high resolution. A great advantage of the BSE block face approach to tissue analysis concerns its depth of resolution. The information depth is of the order of 0.5 μm [
38], and the lateral resolution can be better than 0.1 μm depending on the pixel size chosen. Thus, we have a resolution that is orders of magnitude better than X-ray microanalysis and X-ray imaging. We have about 0.5 μm electron optical section in an area limited only by the block dimensions. Further, the 'section' in a well embedded PMMA block face is intact, because it is the intact tissue in the residual block face, whereas any physical section will be incomplete and deformed, particularly if it contains tiny calcified elements. These exceptional specifications of the qBSE resulted in information about incremental growth lines and substructural details (Figure
6h). Another advantage is that the same surface of these blocks can be mapped with FTIR-RM without any additional treatment or alteration, as we did when Calc1 and Calc2 blocks were imaged first with qBSE, subsequently mapped with FTIR-RM and SEM/SDD-EDS, and then scanned with μCT that supplied the internal structure of the deposits inside the specimen (Calc2).
Comparing the FTIR-RM and SEM/SDD-EDS maps and the μCT and qBSE images with the clinical data (Table
1) reveals a correlation between the duration of calcinosis and the morphological distribution of the mineral in the specimens. The duration of the presence of calcinosis in samples Calc1 (FTIR-RM, qBSE, SEM/SDD-EDS), Calc2 (FTIR-RM, μCT, qBSE, SEM/SDD-EDS), Calc3 (qBSE), Calc6 (qBSE) and Calc10 (FTIR-RM), where islands of mineral are seen, is less than 12 months, and the patients' illness duration was six years or less. The illness duration for the patient of Calc4, that also exhibits the islands morphology, was 4.9 years while the duration of the calcinosis was at least three years. The patients with specimens of homogeneous and continuous mineral that filled the entire 'tissue sac' (Calc7 (FTIR-RM, μCT) and Calc8 (FTIR-RM, μCT)) had illness for 12 and 30 year (respectively) and calcinosis in these locations for at least 12 and 27 years. A relation between the duration of treatment and the mineral distribution was observed in a smaller study using μCT [
12], in which continuous uniform mineralization was observed in a specimen removed from a patient who received treatment for only two months (9.5 years after the diagnosis of JDM), while smaller variable mineral blocks were observed in two other specimens removed from patients treated for 6.5 years and 7.2 years (6.7 years and 7.5 years after onset of JDM, respectively). Stock and colleagues [
12] proposed possible origins for the inhomogeneous vs. uniform structure of JDM calcifications, including differences in duration of untreated inflammation, in the TNFα-308A polymorphism, and in physical constraints at the calcification site. That proposed relation between the morphology and location where the growing mass was confined or free to expand, is questionable from our larger study. Samples with the island morphology (Calc1 and Calc2) and other with continuous morphology (Calc7 and Calc8) were all from confined locations (elbow, toe, and shoulder).
The steep gradient pattern of mineral density observed in the background material in Figures
6d and
6e (seen as changing colours from blue to yellow-red in the top left of the images), and the solidly red regions in the right parts of the images, suggest that the mineralization process occurred in at least two stages, first with the formation of small mineralized nodules, followed by a wave of mineralization that incorporated such nodules into larger mineralized structures. The sizes of the nodules range from a few μm to about 100 μm (e.g., the largest particle in Figure
6d). Note that many islands of this size range were seen in the FTIR-RM mineral map of the same specimen (Calc2) when mapped with 40 × 40 μm spatial resolution (Figure
1f) and in the SEM/SDD-EDS P, Ca, Na and Mg maps (Figure
1i). Even highly mineralized regions displayed substructures (Figures
6h and
6i), and occasionally incremental growth lines, suggesting a resting period between rounds of mineralization, or variations in the rate of mineralization (arrows in Figure
6h). The gradient of background level mineralization seen in this study by FTIR-RM, μCT and qBSE is of considerable interest in understanding the biology of this disease. The morphology shows that widespread mineralization of large numbers of separate small zones or patches may occur, as if there were equal numbers of niduses at which the process commenced, perhaps within a short span of time. The small and highly mineralized patches may later become integrated via a general spread of mineralization (as seen also in the SEM/SDD-EDS elemental maps of the right region of Calc1; Figure
2g). Gradients in mineral concentration were observed in this secondary mineralization process, and incremental growth lines reminiscent of other normal and pathological calcifying tissues [
22‐
27,
33] were also visible. Although the exact time scale in DM is not always known, we observed in this study that the small island morphology is dominant in deposits that resided for relatively short times and continuous mineral morphology in deposits of very long duration.
Biological mineralization involves the replacement of tissue water space with mineral. The less 'protein' present per unit volume and the more water in the 'matrix', the more the matrix can be mineralized. Thus dental enamel, which loses its protein matrix during mineralization, is the most densely calcified tissue [
27,
28]. Cartilage matrix contains highly hydrated protein polysaccharide complexes and comparatively little collagen compared with bone and dentin, but when mineralized, it is more densely mineralized [
28]. Soft tissue cells contain about 10% to 15% protein solids, and soft tissue collagen is less densely aggregated than in bone and dentin matrix. This explains the present findings that calcified soft tissue areas are more or less intermediate in position in terms of the density of calcification between bone, dentin and enamel, as well as in tissue stiffness. There was no observable difference between the modulus versus mineral concentration behavior for the two samples tested (Calc3 and Calc4, Figure
8d). This implies that the mineral in these deposits has a similar ultrastructural arrangement. Mineral concentration is important in explaining stiffness of calcified tissues, and we confirm that there is a linear correlation (R
2 = 0.58) between mineralization and modulus in the densely mineralized regions (Figure
8d). The more water that is displaced and the more robust the ultimate mineral particles, the stiffer a tissue becomes. However, the continuity and/or contiguity of the individual mineral particles also directly influence this parameter [
39]. Thus an explanation for variations in stiffness at the same mineral concentration includes variations in mineral crystal contact [
28]. The steep rise in stiffness observed at a mineral density greater than 2.5 g/ml suggests a percolation threshold; that is, sharply increased continuity of the stiff phase by contact between mineral particles. The growth and impingement of adjacent crystals within dense deposits could produce such a change in stiffness and would be associated with the more well-defined X-ray diffraction patterns observed.
There are similarities and differences between the composition of DM deposits and other forms of pathological calcification. Although the mineral in the DM deposits is composed of homogenous and highly crystallized carbonate apatite (FTIR-RM, XRD, SEM/SDD-EDS) containing Na and Mg and traces of Cl, Al and S (SEM/SDD-EDS) with Ca/P ratios of 1.63 to 1.69, explanted bioprosthetic calcified heart valves were composed of carbonate apatites containing Mg (0.06 to 0.36 wt%) with Ca/P ratios ranging from 1.34 to 2.12 [
40], and with Ca/P ratios of 1.33 to 2.01 for bioprosthetic valves and of 1.62 to 2.13 for natural valves in another study [
41]. SEM-EDS and XRD in this report [
41] revealed that the mineral of these calcified valves was a mixture of dicalcium phosphate dihydrate, octacalcium phosphate and carbonate apatite. Carbonate apatite, β-tricalcium phosphate and octacalcium phosphate-like phase were found in different locations in the same dental calculus specimens with FTIR-RM and micro XRD [
42]. Only traces of proteins were found in these specimens. Carbonate apatite was reported in another study of dental calculus as a separated phase with collagen at the calculus-cementum interface and with lower concentrations of organic phase attributed to microorganisms away from the interface [
43]. Carbonate apatite was found occasionally in urinary stones as a mixture with calcium oxalate [
44,
45]. The concentrations of the main chemical elements in pulp stones appear to be similar to that of the DM deposits that we determined: average Ca/P molar ratios of 1.69 were reported for 10 samples of pulp stones along with 0.51 wt% Mg and 0.75 wt% Na [
46]. But, although we could not detect F (the SEM/SDD-EDS detection limit for F is 0.05 wt%), the pulp stones in the above study contained 0.88 ± 0.05 wt% F and K, Cl, Mn, Zn and Fe were present at trace concentrations. Al and S (found as trace elements in Calc1 and Calc2) were reported also in soft plaque and calcified plaque deposits from human coronary arteries [
47]. S was detected also in brain calcinosis along with Na, Mg and K [
48]. Osteopontin, which has previously been detected in JDM deposits [
8], was found in several other forms of pathological calcification: human pulp stones [
49], urinary stones [
50], human atherosclerotic lesions [
51], and dental calculus [
52]. Note that although dental calculus, pulp stones and urinary stones contain osteopontin (a known bone formation protein), they are very different from bone as well. Type I collagen was detected in the whole area of a pulp stone, while higher magnification reveled stronger staining along the growth lines of the stone [
49]. Comparison of the presence of collagen in other forms of pathological calcification is more complicated. It is reported with cardiovascular deposits that were pulverized and/or powdered, but it could originate from the surrounding tissue or the heart valves tissue [
40,
41]. With no detailed position resolved study available, we cannot compare it with our results of the absence of collagen within the central mineralized section of the DM specimens.
Blood is supersaturated with respect to the concentrations of calcium and phosphate ions necessary to spontaneously precipitate apatite [
53]. The physiological calcification inhibitors that are normally present might be missing or negated by other factors in the case of the pathological soft tissue mineralization associated with DM. Spontaneous vascular calcification, as a result of loss of mineralization inhibitors such as pyrophosphate and matrix gamma-carboxyglutamic acid (GLA) protein that are expressed normally in human blood vessels, was observed by Rutsch and colleagues [
54], and spontaneous calcification of arteries and cartilage occurred in mice lacking the matrix GLA protein [
55]. A recent
in vitro study [
56] showed that nanocrystals of carbonate apatite with composition and morphology analogous to atherosclerotic plaque formed
in vivo, precipitated from human serum-like solutions that did not contain inhibitors, supporting the mechanism of spontaneous precipitation of carbonate apatite in the absence of physiological inhibitors. The waves of mineralization that we observed in our qBSE, FTIR-RM and SEM/SDD-EDS maps, and the presence of only one inorganic compound (carbonate apatite containing Na and Mg and traces of Cl, S and Al) with small amounts of proteins, might support the mechanism of spontaneous precipitation in DM calcification as well. We did not determine the nature of other inhibitors and/or nucleators present in the deposits in this study, but we might learn more in future studies.
Our new findings might aid in designing better therapies for the dystrophic calcification associated with DM. Because the findings that the crystallinity of the apatite in the DM deposits is higher than that of bone (more crystallized apatite is less soluble than poorly crystallized apatite) and its Ca/P ratios are close to stoichiometric hydroxyapatite, the presence of Mg in the deposits, the significantly smaller amount of collagen than in bone, and the possibility that the calcium and phosphate, normally present in affected tissues, may have precipitated as carbonate apatite due to local loss of mineralization inhibitors, different therapeutic agents than those that are currently used for DM may be needed to prevent, slow the progression of and/or dissolve these calcified deposits. One potential class of agents that may be of interest therapeutically are the bisphosphonates (derivatives of the biological inhibitor pyrophosphate), which inhibit the conversion of calcium phosphate into crystalline hydroxyapatite and the nucleation of hydroxyapatite [
57]. This is supported by anecdotal cases of improvement in DM calcification following administration of bisphosphonate therapy [
58,
59]. In addition, supplemental Mg may help prevent and/or retard the progression of the calcinosis. Mg has been used to treat several other kinds of soft tissue calcifications, including myositis ossificans traumatica, calcific bursitis, traumatic osteoarthropathy, and others. Local application of magnesium sulfate into the lesion and peroral administration of magnesium lactate resulted in disappearance or reduction in size of the lesions [
60]. Previous approaches have been trying to treat underlying inflammation, and we suggest using combination of bisphosphonates and other mineralization inhibitors such as Mg at the phase of rapid deposition, presumably at the peak of these waves of mineralization that we saw in our qBSE, FTIR-RM and SEM/DDE-EDS maps.