Spontaneous cervical artery dissection (sCAD) is one of the commonest causes of ischaemic stroke among the young and middle aged.1 sCAD occurs commonly in healthy appearing individuals and is preceded by mild or no trauma. However, the occurrence of multiple sCAD, the association with infection and the seasonal pattern with a pronounced peak in autumn/winter strongly argue against a random occurrence of sCAD and in favour of a specific underlying pathomechanism.2 3 The following observations suggest that an underlying connective tissue disorder might represent a predisposition to sCAD: (1) Ehlers–Danlos syndrome type IV (EDSIV), Marfan syndrome and a number of other connective tissue disorders are rarely associated with sCAD4 and (2) Brandt et al found connective tissue abnormalities in skin biopsies of approximately 50–60% of patients with sCAD.5 6 Their most common and important findings were irregularly shaped so called “flower-like” or “composite” collagen fibrils which resemble collagen fibrils found in the skin of patients suffering from EDSIII and in a more severe form, classical EDS (EDSI or EDSII). Brandt et al used a pathoanatomical examination by a very experienced examiner and judged connective tissue aberrations on a 3 point scale but did not employ objective (eg, morphometric) measures. However, it should be noted that a recent investigation did not show clinical signs of a connective tissue disorder in patients with sCAD.7

After unblinded training using skin biopsies of patients with sCAD, control individuals and a patient with EDS, our experienced light and electron microscopist (WV) sought to confirm the results described in the previous paragraph in a blinded setting but was not able to reliably discriminate between cases and controls, which is a prerequisite for any further studies (eg, genetic). Therefore, we performed a detailed morphometric analysis of collagen fibrils in the dermis of 20 patients with sCAD and in 18 control subjects (table 1) to determine whether objective morphometric parameters can distinguish patients with sCAD from controls, either on an individual basis or as a group.

MATERIALS AND METHODS

The study protocol was approved by the ethics committee of the University of Muenster and written informed consent was obtained from the participants. Skin biopsies were obtained from the dorsal side of the upper arm close to the elbow by scalpel biopsy. Compression of the biopsy by tweezers was carefully avoided. The tissue was divided and processed for electron microscopy (see below). Control biopsies were obtained from healthy volunteers (14 samples) or from autopsies (control samples 5, 11, 12, 14) (table 1). Controls did not suffer from connective tissue disorders or dissections. The skin biopsy of a patient with EDSIII served as a positive control for composite fibrils. Diagnosis of sCAD required a suggestive history, typical clinical signs and the detection of an intramural haematoma on transversal MRI of the neck or proven tapering stenosis or string sign of the affected cervical artery on intra-arterial digital subtraction angiography.

Electron microscopy, morphometry and statistics

The skin specimens were fixed in glutaraldehyde fixative, stained with osmium tetroxide and embedded in epoxy resin. Transmission electron micrographs were made using digital imaging plates (Ditabis, Pforzheim, Germany). Digital photomicrographs were prepared for morphometric analysis using the program Adobe Photoshop 7.0 (Adobe Inc., San Jose, California, USA) and morphometric analysis was performed with the program analySIS (SIS, Münster, Germany) (fig 1). At least three representative fields in the deep dermis of cross sectioned collagen fibrils, free of cellular structures or elastic fibrils, were chosen for morphometry by the examiner and surrounded by a polygon. Areas in the vicinity of other structures, especially elastic material, were avoided because they contained abundant aberrant collagen fibrils, both in control individuals as well as in patients with sCAD. The sides of the polygon were drawn at a distance to the nearest collagen fibrils which equalled approximately half the distance between collagen fibrils in the field. The following basic parameters were measured: (1) total number of fibrils analysed (n), (2) fibril diameter in nanometres, (3) fibril density expressed as fibrils per square micrometres (n/μm²), (4) “relative fibril area” expressed as percentage of the area occupied by fibrils per total area analysed, expressed in per cent and (5) “form factor”: the parameter “form factor” is a generic parameter of the program analySIS which assesses the roundness of objects. It is especially sensitive to irregularly formed contours which are the hallmark of cross sectioned composite fibrils and less sensitive to ovoid deformation which can occur if collagen fibrils are not cut exactly perpendicularly. Form factor adopts values between 1.0 (perfect circle) and 0.5 (very irregularly shaped). Microscopic investigations and morphometric measurements were performed and tabulated, blinded for the disease status. All measured parameters except for the total number of fibrils analysed are continuous variables for which we calculated mean and SD as a measure of variation. Comparisons between groups were performed using the unpaired two tailed Student’s t test with a significance level of p⩽0.05. Since most representative fields of collagen fibrils chosen for morphometric analysis did not contain any composite fibrils (fig 2) we decidedly searched for these aberrant fibrils in the deep dermis of the skin biopsies by spending approximately 20 min on the electron microscopy searching for these fibrils in each sample. The number of composite fibrils identified is given as a semiquantitative score (− to ++++) in table 1.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Dermal collagen morphology and preparation of electron micrographs for morphometry. (A) Representative electron micrograph of dermal collagen of control sample C3. (B) Same electron micrograph as in (A) after digital preparation for morphometry. (C, D) Patient with spontaneous cervical artery dissection (sample P7).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 Morphology of aberrant collagen fibrils in a patient with spontaneous cervical artery dissection (sCAD) and in a control sample. (A) Collagen fibrils of control sample C12 demonstrating mild collagen fibril aberrations (arrow). (B) Mild collagen fibril abnormalities in sCAD patient sample P1 (arrow). (C) Frank composite fibrils in sCAD patient sample P1 (arrowhead). (D) Composite fibrils in the skin of the Ehlers–Danlos syndrome type III reference sample.

RESULTS

The demographic, clinical and morphometric characteristics of the patients with sCAD and controls are summarised in table 1. Mean age of the patients with sCAD was 41 years while the mean age of the controls was 37 years. None of the patients with sCAD showed any clinical signs of EDS (eg, hyperextensible skin or joints). Figure 1 shows representative electron micrographs of collagen fibril fields in the deep dermis of a control (C3) and an sCAD patient (P7) before and after digital preparation for morphometric analysis.

Pathoanatomical examination of skin biopsies revealed that flower-like or so-called composite fibrils in the deep dermis were, as expected, abundant in the reference biopsy obtained from a patient with EDSIII. Composite fibrils were also quite common in one patient with left internal carotid artery dissection (table 1, P1, +++) and somewhat less prevalent in another patient with dissection of both internal carotid arteries (table 1, P16, ++) but were absent in all other skin biopsies from patients with sCAD. However, composite fibrils were also found in two skin biopsies of control individuals (table 1, C12, ++; C13, +) (figs 1, 2).

Next, we performed a morphometric analysis of collagen fibrils focusing again on collagen fields in the deep dermis. We measured, on average, approximately 760 collagen fibrils in each skin biopsy (824 in patients with sCAD and 692 in controls) (table 1).

The mean diameter of collagen fibrils in the skin of patients with sCAD was significantly smaller than in the skin of controls (84 vs 92 nm, p = 0.027) (table 1, fig 3).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3 Diameter vs density of dermal collagen fibrils in skin biopsies. Diameter of collagen fibrils is measured as mean diameter per individual. (A) Control group. (B) Spontaneous cervical artery dissection (sCAD) group. Depicted are the means and (SD) of the diameter of each individual as well as the regression line (solid line) and the 95% confidence intervals (broken lines).

The variation in collagen fibril diameters expressed as SD was not significantly different between patients with sCAD and controls (p = 0.57, table 1). There was also no obvious correlation between collagen fibril diameter variation expressed as SD and the aberrant fibril score. Interestingly, sCAD patient skin sample P1 showed by far the largest diameter variation (SD 24 nm) which was only comparable with the reference sample of an EDSIII patient (SD 20 nm). However, patient P1 has repeatedly been examined by one of the authors (GK) and did not show any clinical features of a connective tissue disorder, as judged by the Beighton criteria.8 All other skin biopsies showing aberrant collagen fibrils did not exhibit larger fibril diameter variation than the samples without aberrant fibrils.

Average fibril density, expressed as fibril number/μm², was higher in the skin of patients with sCAD than in controls (95 vs 77 fibrils/μm²; p = 0.001) (table 1, fig 3). The higher fibril density in skin biopsies of patients with sCAD “compensates” for the smaller diameters of the individual fibrils leading to comparable values for the relative fibril area (area of fibrils/total area analysed consisting of the fibrils and the surrounding ground substance, expressed as per cent) in patients with sCAD and controls (51 vs. 48%, p = 0.08, table 1, fig 3). Figure 3 shows that an excellent inverse linear correlation between the fibril diameter and fibril density is present in patients with sCAD as well as in control individuals (R = −0.82 for control samples and R = −0.95 for sCAD samples) underlining the fact that the relative fibril area, which is proportional to the fibril diameter times the fibril density, is relatively constant between individuals as well as between groups.

Analysis of the roundness of the collagen fibrils using the parameter “form factor” did not reveal any differences in the shape of collagen fibrils in skin between patients with sCAD and controls.

Despite the differences in the comparison between groups, none of the measured parameters was able to discriminate between patients with sCAD from controls on an individual basis.

DISCUSSION

Aberrations of connective tissue in the deep dermis of patients with sCAD on electron and light microscopy were first described by Brandt et al in 1998.5 6 Skin biopsy findings with unrounded so-called “composite collagen fibrils” often resembled the findings in skin biopsies of patients with EDSIII. Similar, but even more severe, changes are observed in classical EDS (EDSI and EDSII). In addition, elastic tissue abnormalities, especially frayed elastic fibrils with a “moth eaten” appearance, were assessed by Brandt et al. Both collagen and elastic fibril abnormalities were summarised and graded together in one score (expressed as: 0, +, ++, +++). Summarising all publications using this grading system, more than half of patients, but no control subjects, showed connective tissue aberrations. The most recent publications by the same group reported on a considerably larger population of patients with sCAD than those in our study.9 They did not use the above mentioned grading system but differentiated between three different types of connective tissue pathology (resembling EDSIII with predominant aberrant collagen fibrils (43/126 skin sCAD skin biopsies), EDSIV-like (13/126 sCAD skin biopsies) and elastic fibre abnormalities only (16/126 sCAD skin biopsies)) and demonstrates an inheritance pattern suggesting autosomal dominant inheritance of dermal connective tissue abnormalities in three families in whom the index patient suffered from sCAD.9 Interestingly, suggestive genetic linkage with two loci on chromosomes 10 and 15 was later described in one of these families while the same loci were excluded in the two other families.10 However, analysis of dermal connective tissue in families with more than one member affected by sCAD surprisingly revealed no abnormalities in most of these patients.11 None of the reports stated that the pathologist was blinded to the disease status.

Even after prolonged training, our electron microscopist (WV) was unable to reliably discriminate between the dermal connective tissue of patients with sCAD and controls. The elastic tissue in particular was morphologically extremely variable in patients with sCAD as well as in controls, foreclosing a systematic and especially a morphometric analysis. Aberrant so-called composite collagen fibrils were found in 2/20 skin biopsies from patients with sCAD and, in contrast with previous results which did not find any aberrant collagen fibrils in controls, also in a comparable number of control biopsies (2/18).

The main reason why we performed the morphometric analysis presented here was to objectively asses the morphology of collagen fibrils in patients with sCAD and control individuals to determine whether morphometric parameters might distinguish between sCAD and control biopsies on an individual basis or as a group. While the morphometry as such is a rather objective method, the choice of the sections subjected to morphometry does necessarily introduce a subjective examiner dependent component. The morphometric parameters were chosen bearing the previously reported findings in mind. Fibril diameter, fibril diameter variation and relative fibril area aimed at detecting thin, loosely packed fibrils with a diameter variation found in various forms of EDS, while the parameter “form factor” was chosen to detect unrounding of fibrils typical of so-called flower-like and composite fibrils. As described in detail above, the only differences detected were slightly thinner but more densely packed fibrils in the reticular dermis in the sCAD patient group compared with the control group (table 1). Another interesting finding, albeit not related to sCAD, was the very high linear correlation between fibril diameter and fibril density, leading to very similar relative fibril areas between individuals in both groups. One could hypothesise that this mechanism may compensate for interindividual differences in collagen fibril diameters by forming fewer fibrils of large diameter in some individuals and more fibrils of thinner diameter in others as a way to generate similar mechanical strength of the skin.

How can the apparent differences between the results of previous publications and our own findings be reconciled? One explanation could be that the previously published results were a result of overinterpretation of slight random or environmentally caused dermal connective tissue alterations caused by lack of blinding and objective parameters. It is well known that various environmental influences may lead to the formation of composite collagen fibrils.12 Alternatively, the dermal connective tissue alterations in patients with sCAD are so subtle that they escaped our visual and morphometric analysis. The skin pathologist who performed all of the analyses in the investigations by Brandt (I Hausser) is a widely renowned expert in ultrastructural skin abnormalities related to connective tissue disorders and we cannot exclude the fact that very slight variations in different elements of the connective tissue might have escaped our pathological examination and morphometric measurement.5 6 9^–11 Our morphometric investigation of collagen fibrils in skin biopsies of patients with sCAD did not show a difference between patients with sCAD and controls on an individual basis. Therefore, this method is unable to provide a novel diagnostic marker for sCAD. However, we did find an association between thinner and more densely packed fibrils and sCAD.

In our opinion, a blinded examination of a large series of skin biopsies of patients with sCAD versus controls by different examiners and the development of objective measures of dermal connective tissue aberrations which can easily be reproduced would be a valuable step ahead in the assessment of dermal connective tissue aberrations in patients with sCAD.