Vascular Ehlers–Danlos syndrome (vEDS), a rare genetic disease with an autosomal-dominant trait, is caused by pathogenic variants within the collagen type III alpha-1 chain (COL3A1) gene. The resulting quantitative and qualitative alterations in type III collagen, typically lead to organ fragility, expressed clinically in young adults as arterial dissections and ruptures, bowel perforation, gravid uterine, and organ ruptures [1,2,3,4].

The premature death of patients with vEDS syndrome is frequently related to spontaneous rupture of arterial aneurysms and/or dissections. These arterial lesions are frequent and always looked for in the CT scan performed in case of complication or in case of screening during the monitoring of these patients. Arterial lesions are most often localized on the branches of the abdominal aorta and supra-aortic branches. Pulmonary vascular segment involvement is very rare [5, 6].

Furthermore, chest involvement, manifesting as hemo-/pneumothoraces and pulmonary hemorrhages, is not uncommon in vEDS patients, but has not been formally characterized in a large patient cohort. Indeed, pulmonary involvement and parenchymal disease were identified only in case reports and case-series describing cysts, intra-pulmonary cavitary lesions, and interstitial disease [7,8,9,10,11,12,13,14,15,16,17].

To further document pulmonary disease in vEDS, a study on clinical and CT imaging of lung involvement was undertaken in a large series of consecutive vEDS patients, with corresponding histology when available.

Methods

Population

All patients with molecularly confirmed vEDS referred, between 2004 and 2016, to the French National Referral Centre for Rare Vascular Diseases were included in this retrospective cohort study when CT scans had been obtained for either symptom (index cases) or screening after genetic diagnosis (relatives). All vEDS-related symptoms, clinical characteristics (smoker status, body mass index (BMI; kg/m2), and history of respiratory events) were collected in a dedicated database with systematic follow-up.

Respiratory events were assessed from the initial evaluation to the last follow-up visit to our institution. Respiratory events were defined as pneumothorax, hemothorax, or hemoptysis.

Genetic testing and the database were managed in compliance with the French legislation (French Bioethics Law no. 2004–800), and written informed consent was obtained from all patients. As reported previously [18], based on the results of COL3A1-mutation screening, distinct variants were divided into three groups as follows: glycine substitutions, splice-site and in-frame insertions-deletions/duplications (ins/ del/dup), and variants leading to haplo-insufficiency.

Chest CT scans

CT acquisitions were performed using either a 4-row or 64-row CT system (General Electrics) in the cephalocaudal direction from the lung apex to the ischial tuberosity to include fully the chest, abdomen, and pelvis. The patients were instructed to hold their breath during scanning. In all cases, image acquisition was done during the arterial phase after an intravenous bolus injection of 350 g/L of iodinated contrast material at a rate of 3.5 mL/s, in accordance with an institutionally defined protocol optimized to screen for vEDS arterial lesions. The volume of contrast medium to be injected was calculated based on the patient’s body weight, with the total dose ranging from 80 to 130 mL; it was followed by a 50-mL saline flush at 3.5 mL/s. The exposure parameters for the CT scans were as follows: 80 to 120 kVp, 50 to150 mA with dose modulation (angular and z). Axial chest images were reconstructed with 0.625- to 1.2-mm slice thickness, high resolution and soft tissue algorithms, and an optimized field of view of 25 to 30 cm.

All chest CT images were retrieved from the institution’s picture-archiving and communication system (PACS) to be analyzed on a dedicated workstation with both mediastinal (level and width = 50 Hounsfield units (HU) and 400 HU, respectively) and lung window settings (level and width = −600 HU and 1500 HU, respectively). Minimum- and maximum-intensity projection algorithms were applied to each set of CT scans with 5-mm slice thickness to facilitate detection of nodules and hypodense structures (width 500 HU; level − 950 HU).

Two chest radiologists (S.B. and P.A.G.) blinded to clinical data conducted a consensus review chest CT scans performed either on a systematic basis or because of respiratory events occurring during follow-up. Only the first HRCT examination systematically obtained for each vEDS patient at our institution or scans obtained because of respiratory events occurring during follow-up were analyzed herein. To clarify the anatomic distribution of abnormalities, the lungs were divided into axial upper and lower zones at the level of the tracheal carina. The presence and distribution of the following features were defined according to the last Fleischner Society statement [19]. Centrilobular emphysema (CLE) was considered to be small poorly or well-defined areas of low attenuation surrounded by normal lung without visible walls. Advanced destructive emphysema (ADE), representing advanced CLE, was defined as a generalized decrease of lung attenuation without focal hypoattenuation. Interlobular septa were often preserved and splayed, facilitating the identification of pulmonary lobular hyperexpansion. In addition, the central pulmonary vessels were often distorted, splayed, and narrowed with decreased branching (architectural distortion). Paraseptal emphysema (PSE) was defined as subpleural and peribronchovascular foci of low attenuation separated by intact interlobular septa thickened by associated mild fibrosis. For regions of low attenuation adjacent to visceral pleura, PSE may be further classified as peripheral, mediastinal, diaphragmatic, or fissural. Bullae were defined as avascular low-attenuation areas, > 1 cm in diameter, with a thin but visible wall. Pulmonary cyst was a round parenchymal lucency or a low-attenuation area delimited by a well-defined interface with the normal lung. Nodule corresponded to a poorly or well-defined rounded opacity, measuring up to 3 cm in diameter, with homogenous soft tissue attenuation. Cavitary lesions were either gas-filled spaces or low-attenuation areas associated with a > 4-mm-thick wall within pulmonary consolidation, within a mass or a nodule.

Vascular abnormalities were assessed like thoracic arterial aneurysm/dissections.

The Haller index was defined as the ratio of the transverse diameter (the horizontal distance of the inside of the ribcage) to the anteroposterior diameter (the shortest distance between the vertebrae and sternum) [20].

Histology

Because of respiratory events requiring surgery, five histologic samples from four different patients were available and could be analyzed using previously described criteria [21], and others, such as pleural thickening and tobacco-induced changes. The histologic findings of these samples were compared to those obtained in five control subjects, who had undergone lung surgery for recurrent pneumothoraces. For each patient, three or four representative 4-μm-thick tissue samples were examined after staining with hematoxylin, eosin and saffron, Sirius red, and iron staining. Two independent pathologists blinded to patient status conducted the histologic examinations using previously described criteria [21].

Statistical analyses

Quantitative variables are expressed as mean ± standard deviation or median (interquartile range). Quantitative variables were compared with unpaired t tests or Mann–Whitney tests and qualitative variables with chi-square with Yates correction or Fisher’s exact tests when appropriate. Variables associated with an abnormal CT scan at the 10% level (p ≤ 0.10) were then introduced into a multivariable logistic regression model. Statistical analyses were performed using R (https://www.R-project.org/) software. p ≤ 0.05 defined significance.

Results

Patients and clinical characteristics

Among 138 vEDS patients eligible to participate between April 2004 and February 2016, two were excluded because of breathing artifacts impacting the analysis of lung CT images. Clinical characteristics and genotypes of the 136 analyzed patients, including 86 index cases and 50 first-degree relatives, are reported in Table 1. Their mean age was 37 ± 13 years, 83/136 (61%) were women, and 50/136 (36.8%) were current (n = 30) or former (n = 20) smokers with an average exposure of 12 pack-years. Among identified pathogenic variants, glycine substitutions within the triple helix domain predominated, followed by splice-site variants and in-frame insertions-deletions, and variants leading to haplo-insufficiency.

Table 1 Demographics of 136 subjects with vEDS (n = 136)
Full size table

Seven out of 136 (5.1%) patients had an associated respiratory disease (asthma (n = 2) or sarcoidosis (n = 2), pulmonary embolism (n = 1), sleep-apnea syndrome (n = 1), and bronchial cancer (n = 1)). Before and during the study period, 24/136 (17.6%) patients developed respiratory events, mainly pneumothorax (n = 17), and several with hemothorax (n = 4) or hemoptysis (n = 3); 11 of them required thoracic surgery. For the prior and current respiratory event (n = 24), 20 were index cases and 4 were related cases.

Imaging findings

Chest CT scans were performed in 125 vEDS patients at the initial evaluation in our institution and in 11 patients because of clinical events occurring during follow-up. Six patients had CT images obtained for acute chest events, pneumothorax (n = 1), pulmonary hemorrhage (n = 1), hemopneumothorax (n = 2) (Fig. 1), or pleural effusions related to vascular dissection (n = 2). Five patients had acute clinical events related to sub-diaphragmatic vascular dissections without thoracic consequences. No vascular abnormalities were found.

Fig. 1
figure 1

Axial thin-section CT images (a, b) in the lung window of a 48-year-old male non-smoker with vEDS, showing diffuse round-glass opacities, suggestive of pulmonary hemorrhage, and some low-attenuation areas (asterisk)

Full size image

Comparisons of patients’ clinical characteristics according to the presence or absence of CT pulmonary abnormalities are reported in Table 2. An abnormal chest CT scan included all parenchymal and pleural abnormalities. Chest CT scans were abnormal for 78/136 (57.4%) patients. The two groups did not differ significantly for age, sex, weight, genotype, or smoking history. However, patients with pulmonary abnormalities were significantly taller. In multivariate analysis, the height remained significantly higher in abnormal CT groups (OR = 1.08, IC95% [1.02–1.15], p = 0.01). Fifty-nine patients who did not present respiratory symptoms or prior respiratory event had abnormal chest CT findings. Two patients with no parenchymal changes on CT had a pneumothorax. The incidence of pneumothorax was significantly higher in the group with parenchymal abnormalities. Among chest CT characteristics (Table 3), emphysema was the most frequent abnormality, observed in 44/78 (56.4%) patients, followed by linear opacities connected to pleura observed in 28/78 (35.8%), clusters of calcified micronodules in 9/78 (11.5%) (Fig. 2a), and cavitating nodules in 4/78 (5.1%) (Fig. 2b). One patient developed interstitial pneumonitis (peribronchovascular consolidation compatible with organizing pneumonia) and another had small nodules related to sarcoidosis. Three patients had a thick apical cap, two patients had ground-glass areas, and one patient had small nodular ground-glass opacities mainly in the upper lobes. Emphysema was centrilobular in all patients with emphysema, and paraseptal in 22/44 (62.2%) (Fig. 3). CLE was about equally frequent in the upper and lower lobes. PSE was seen, in decreasing frequency, along the peripheral, mediastinal, diaphragmatic, or fissural pleura. Peribronchovascular paraseptal emphysema was found, but less often. Four patients had ADE in the lower lung zones (Fig. 4). Linear opacities connected to pleura (Figs. 4 and 5) were mostly connected to the diaphragmatic, but less frequently to mediastinal or fissural pleura. The mean Haller index was 2.34 ± 0.52.

Table 2 Comparison of clinical and genetic characteristics according to CT scan abnormalities in vEDS
Full size table
Table 3 Chest CT findings of the 136 Subjects with vEDS
Full size table
Fig. 2
figure 2

a Multiple calcified micronodules in the left lower lobe (arrows) in a 38-year-old female non-smoker with vEDS (axial 5-mm maximum-intensity projection). b Cavitating nodule in the left lower lobe of a 28-year-old female non-smoker with vEDS

Full size image
Fig. 3
figure 3

Thin-section CT images of a 21-year-old male non-smoker with vEDS. Reformatted axial (a, b) and sagittal images of the right lung (c). Multiple paraseptal emphysematous spaces in the apex of the left upper lobe (a) and the lung bases of the right lung (particularly along the diaphragm) (b, c). Note the presence of several small nodular round-glass opacities in the upper lobes (a, c)

Full size image
Fig. 4
figure 4

Axial thin-section CT images in the upper (a) and the lower (b) lung zones, and reformatted right sagittal image (c) of a 38-year-old male non-smoker with vEDS. Note the presence of centrilobular emphysema in the upper lobes (a), and paraseptal emphysema along the fissures and the diaphragmatic pleura. Linear opacities are connected to the diaphragmatic pleura and fissures (arrows)

Full size image
Fig. 5
figure 5

a, b Axial thin-section CT images through the lung bases of a 40-year-old male non-smoker with vEDS. Note the areas of advanced destructive emphysema in the anterobasal segment of the lower lobes and linear opacities through the left lung bases (asterisk)

Full size image

CLE and PSE were identified in 21/50 (42%) smokers and 23/86 (6.7%) non-smokers. Emphysema distribution in the upper or lower lobes, linear opacities connected to pleura, clusters of calcified small nodules, cavitating nodules, and Haller index did not differ between smokers and non-smokers (Table 4). Subanalysis on lung abnormalities on CT did not show statistical difference between the different types of COL3A1 variants (Table 2).

Table 4 CT Findings according to smoker status in subjects with vEDS
Full size table

Histologic findings

Histologic examination findings are summarized in Table 5. Emphysema lesions were found in three patients with ruptured alveoli, accompanied by diffuse pleural scarring or localized pleural fibrosis. The peribronchovascular emphysema seen in histologic findings (Fig. 6a, b) in our series could correspond to the emphysema seen on CT scan. Fibrous nodules and calcifications/ossifications were not observed in the examined specimens (in our series). vEDS patients 2 and 4 had a very specific pattern of diffuse hemorrhage. Hemosiderin resorption was particularly notable in patient 4, accompanied by alveolar wall thickening, probably secondary to chronic intra-alveolar hemorrhage. Pulmonary arteries in vEDs patients were more irregular and thin-walled, with more diffuse hemorrhage and hematomas, than in controls (Fig. 6c, d). No intraluminal outgrowth was observed.

Table 5 Pleuropulmonary histology in 5 sample from 4 subjects with vEDS
Full size table
Fig. 6
figure 6

Examples of histology of lung sections in vEDS. a, b Hemalun Eosine Saffron (HES) stain demonstrating emphysematous changes in the first (a) and second (b) samples of case 1. Note the diffuse pleural thickening (black arrows, × 2 and × 4 magnification). c Hemorrhagic bullae. Sirius red–stained histologic sections on the second resection at 27 years old (d), demonstrating the arterial thickness variations (arrows) commonly seen in vEDS (× 20 magnification)

Full size image

Discussion

Herein, we described a large cohort of vEDS patients with long-term follow-up, their clinical presentations, and CT imaging of disease-related pulmonary involvement, with corresponding histologic findings for a few of them. This is the largest cohort of vEDS patients, whose frequencies of pulmonary lesions were routinely analyzed without selection bias. Lung CT abnormalities were not influenced by types of COL3A1 variants. Types of COL3A1 variants have been shown to have significant influence on the onset of arterial and digestive complications, as well as on overall mortality. Indeed, patients with splice-site variants seem to develop organ complications earlier and have a higher mortality rate at a given age than patients with glycine substitutions. The patients with the latest onset of complications are patients with variants leading to haplo-insufficiency, typically coming to medical attention a median 12 to 21 years later than patients with glycine substitutions and splice-site variants, respectively. In this study, lung CT abnormalities were not influenced by types of COL3A1 variants, and more interestingly, patients with haplo-insufficiency had similar prevalence of CT abnormalities than the more severe glycine and splice-site variants. In contrast to the literature [3], we did not find more pectus excavatum in this population of vEDS patients. The normal Haller index indicated the absence of a link between parenchymal abnormalities and morphology of the thoracic cavity. In histologic finding, pleural thickening is probably the consequence of the presence of hemorrhage that limits resorption [22]. Dar et al reported a vEDS case presenting with hemoptysis and hemopneumothorax and described increased fibrosis on the pleural surface with an increased vascularity on surgical lung biopsy [13].

We did not find any cysts, since the low-attenuation areas in the lung parenchyma without visible walls corresponded to emphysema, as confirmed by the histologic examinations of five samples. The notion of cysts in journals is based on a few reported cases [23, 24]. Therefore, by analyzing thin-section chest CT images, vEDS can be differentiated from cystic lung diseases, which can be responsible for familial spontaneous pneumothorax in several genetic diseases, including sporadic or tuberous sclerosis complex-associated lymphangioleiomyomatosis, α1-antitrypsin deficiency, Marfan syndrome, Birt-Hogg-Dubbé syndrome, and cystic fibrosis [23, 24].

Among pulmonary lesions described in detail in our series, there was no finding of underlying interstitial lung disease or fibrosis as well on CT scan as on histologic findings. Chest CT abnormalities concerned only emphysema, linear bands, cavitating nodules, and clusters of calcified micronodules.

The presence of emphysema in this disease was previously described in a case report [9] and eight of nine vEDS patients included in a histopathology study [21]. Although tobacco smoking is the cardinal risk factor for centrilobular emphysema, 64% of our patients had never smoked. Hence, tobacco was not statistically related to the presence of emphysema and distribution. Compared to smaller series of vEDS patients [1, 3], pneumothorax (12.5%) and hemothorax (3%) frequencies were very similar.

The mechanism of emphysema onset and its development in vEDS is still unclear. Dilacerations of the alveolar wall could be one of the underlying mechanisms. Kawabata et al argued that fragility of the alveolar wall, which contains type III collagen, can also be responsible for other pathologies associated with parenchymal and pleural damage [21]. Corrin et al also suggested that spontaneous tearing might cause pulmonary emphysema formation in vEDS patients [25], with spontaneous laceration of lung tissue being the key pathological feature. According to this mechanical hypothesis, it is not surprising to observe in our vEDS patients an alveolar rupture distribution which predominates at the level of alveolar attaches to subpleural and peribronchiolar conjonctive tissue, instead of being panlobular. Our findings also support the hypothesis that abnormalities might not be the cause, but rather the consequence of an event and might be the residual scar of a past respiratory event. Linear opacities connected to pleura were also found in 28 (20.6%) of our patients, mainly diaphragmatic, representing a previously undescribed feature of this disease. Linear opacities probably represent atelectasis or interlobular septa adjacent to the emphysematous spaces. Atelectasis may be secondary to pleural fibrosis, since it was also seen in histologic specimens. These lines may represent connected thickened interlobular septa made more conspicuous by extensive adjacent parenchyma destruction. [26,27,28,29]

The third abnormality found in our study was cavitating nodules with similar observations made in some case reports [30, 31]. These probably correspond to hematomas caused by spontaneous arteriolar ruptures. Because of blood-vessel fragility and abnormal connective tissue repair of hemorrhagic foci, hematomas shrink to ultimately evolve into fibrous nodules.

According to the largest pleuropulmonary histopathology series of vEDS cases reported to date, eight of nine patients had fibrous nodules, six of them with ossifications probably explained by osseous metaplasia found within these fibrous nodules [21]. The same mechanism probably explains the multiple calcified micronodules seen in nine of our patients.

We also compared CT and histologic findings. No cysts were seen. Smokers and non-smokers had emphysema. The results of the histologic analysis suggest that emphysema is an indicator of alveolar rupture and perforated nodules would indicate hematomas. The cause of these abnormalities is a matter of speculation but we advance that these abnormalities seem to resemble thoracic pulmonary blast. Pulmonary blast lesions are due to the impact of a pressure wave on the chest wall with hyperpressure in the airways followed by a secondary depression wave. The main anatomic injuries seen after a blast explosion are pneumothorax, pneumatocele, hemato-pneumatocele, and alveolar rupture [32], i.e., the same chest injuries seen in vEDS. A pathophysiological mechanism could perhaps be the greater sensitivity to air-pressure waves in vEDS.

This study has several limitations. First is its retrospective nature with no pulmonary function tests for comparison with CT abnormalities. Because we included asymptomatic cases, missing pulmonary function tests is unavoidable in routine practice. In addition, we only analyzed the patients’ first chest CT scans to determine whether or not they were normal. Although unlikely, lung parenchymal changes could be due to other concomitant lung diseases. Further studies are needed to analyze the evolution of the lung parenchyma under treatment in this context and its potential link with vascular disease. Emphysema is not specific; however, in young patients or non-smokers, it could be indicative. Emphysema has been reported in other genetic diseases. Several syndromes, such as cutis laxa, Niemann-Pick disease, and telomeropathies, are related to emphysema. Genome-wide association studies demonstrated that several genes are associated with emphysema on CT [33]. Emphysema associated with alpha-1-antitrypsin deficiency is panlobular, characterized by uniform destruction of the pulmonary lobule leading to widespread areas of abnormally low attenuation without visible walls, and in contrast to CLE, the destruction is generalized and more prominent in the lower lobes. Emphysema in non-smokers has also been reported in patients with connective tissue diseases [34] but always in association with lung fibrosis [35, 36].

In conclusion, based on a large cohort of patients with molecularly confirmed vEDS, we were able to identify different previously undescribed pulmonary lesions and report their frequencies. The most common CT feature, including in clinically asymptomatic patients, was emphysema, suggesting alveolar wall ruptures. Identification by radiologists of this unexpected pulmonary pattern in seemingly healthy young subjects without known vEDS should trigger diagnostic investigations. The prognostic value and evolution of these parenchymal abnormalities remain to be characterized. We also need to determine to what extent these CT findings are associated with respiratory or other arterial events and outcomes.