Involvement of the pulmonary arteries in patients with Takayasu arteritis: a prospective study from a single centre in China

Background Takayasu arteritis (TA) is a large vessel vasculitis that can involve pulmonary arteries (PAs). We studied multiple clinical characteristics related to pulmonary artery involvement (PAI) in TA patients. Methods We enrolled 216 patients with TA from a large prospective cohort. PAI was assessed in each patient based on data from magnetic resonance angiography/computed tomography angiography. Pulmonary hypertension, cardiac function, and pulmonary parenchymal lesions were evaluated further in patients with PAI based on echocardiography, the New York Heart Association Functional Classification, and pulmonary computed tomography, respectively. These abnormalities related to PAI were followed up to evaluate treatment effects. Results PAI was detected in 56/216 (25.93%) patients, which involved the pulmonary trunk, main PAs, and small vessels in the lungs. Among patients with PAI, 28 (50%) patients were accompanied by pulmonary hypertension, which was graded as ‘severe’ in 9 (16.07%), ‘moderate’ in 10 (17.86%), and mild in 9 (16.07%). Twenty-six (46.43%) patients showed advanced NYHA function (III, 20, 35.71%; IV, 6, 10.71%). Furthermore, 21 (37.50%) patients presented with abnormal pulmonary parenchymal lesions in the area corresponding to PAI (e.g. the mosaic sign, infarction, bronchiectasis). During follow-up, two patients died due to heart failure and pulmonary thrombosis. In the remaining patients, the abnormalities mentioned above improved partially after routine treatment. Conclusions PAI is common in TA patients. PAI can cause pulmonary hypertension, cardiac insufficiency, and pulmonary parenchymal lesions, which worsen patients’ prognosis.


Introduction
Takayasu arteritis (TA) is a chronic, granulomatous large vessel vasculitis. It involves the aorta and its main branches predominantly and leads to vascular thickness, stenosis, and occlusion [1]. In addition, TA occurs preferentially in young (20-40 years) women and has a relatively high prevalence in Asian countries compared with that in the USA and European countries [2].
Besides the aorta and its branches, the pulmonary arteries (PAs) are involved in TA. PAs have been reported to be involved in 6.9 to 80% of TA patients from different populations [3][4][5][6][7][8]. The stenosis, occlusion, or embolism of PAs can cause pulmonary hypertension (PH), perfusion defects, or even pulmonary infarction [9][10][11]. Recently, we also found multiple other pulmonary parenchymal lesions in the lung lobes with pulmonary artery involvement (PAI). In addition, right ventricular function is susceptible to damage in patients with PH due to increased afterload, which has a direct adverse impact on the prognosis [12]. Therefore, PAI in patients with TA is life-threatening because it damages the cardiopulmonary function.
Multiple imaging modalities can be employed to detect PA abnormalities. These include digital subtraction angiography (DSA), magnetic resonance angiography (MRA), computed tomography angiography (CTA), positron emission tomography-computed tomography (PET-ECT), and lung ventilation/perfusion (VQ) scan [10,13,14]. Among them, DSA is applied rarely due to its invasiveness although it is the 'gold standard' to assess PAI. By contrast, CTA and MRA are undertaken more frequently because they show the structure of and inflammation in the blood vessels, respectively [15]. PET-CT can also be employed to detect vascular inflammation [16]. Lung VQ scans can display the vascular emboli for segmental (or sub-segmental) PAs [10].
We investigated the clinical characteristics, pulmonary parenchymal lesions, and cardiac functions in TA patients with PAI by combining multiple imaging modalities (MRA, CTA, PET-CT, lung VQ scan, echocardiography, and high-resolution computed tomography (HRCT). Our aim was to elicit a better understanding of TA patients with PAI to aid rational treatment for these patients and improve their prognosis.

Methods and materials
Patients This was an observational study based on a prospective cohort named East China Takayasu Arteritis (clinical trial number, NCT03893136). This cohort was established from a single centre (Zhongshan Hospital, Fudan University) in China, and all the patients in this cohort were Han Chinese. In this cohort, all patients were classified as TA according to the classification criteria set by the American College of Rheumatology in 1990 [17]. The clinical data of this cohort were recorded using a standardised form and inputted into an electronic database. Patients diagnosed from 1 January 2013 to 31 May 2019 were enrolled in the present study. Patients with concurrent disorders that could involve the lungs (e.g. pulmonary infection, asthma, tumours) were excluded. Finally, 216 patients were included in this study (flowchart was shown as Supplementary Figure 1).

Disease assessment
At the initial visit, demographic data and clinical features (systemic and ischaemic symptoms, physical signs) were recorded. Laboratory workup (routine tests for blood and urine, erythrocyte sedimentation rate (ESR), C-reactive protein (CRP) level, liver function test, kidney function test, hepatitis-B/C test, interferongamma release assay, antibodies against Mycoplasma species, sputum pathogens) was undertaken.
Vascular involvements were evaluated by whole-body MRA. If MRA was contraindicated, CTA or PET-CT of the thoraco-abdominal region was done. The type of vascular involvement in TA patients was classified according to the imaging classification created by Hata et al. [18]. Echocardiography was also carried out to evaluate patients' cardiopulmonary conditions.
Pulmonary CT was undertaken for each patient. Additional pulmonary CTA or lung VQ scans were carried out for patients in whom a pulmonary embolism/thrombosis was suspected. CT-guided transthoracic lung biopsy was conducted for patients with confusing pulmonary lesions. Histopathology studies using staining (haematoxylin and eosin, immunohistochemical, acid-fast, silver) were done to evaluate vascular inflammation and detect possible pulmonary infections. Tissue culture was also carried out to clarify if the pulmonary lesions were caused by an infection.
The Kerr criteria [19] were used to assess disease activity: (i) systemic symptoms (infection, tumours, etc. were excluded), (ii) elevated ESR level, (iii) symptoms or signs of vascular ischaemia (weakened pulse or pulselessness, vascular bruits, asymmetric blood pressure), and (iv) positive imaging results. New onset or worsening of two or more criteria indicated 'active disease'.

Follow-up
Patients were followed up each month during the first 6 months and every 2-3 months after that as planned. During follow-up, MRA was repeated every 6 months, whereas other imaging examinations (e.g. pulmonary HRCT, echocardiography) were employed as clinically required.
For patients suspected of having pulmonary thrombosis, anticoagulant therapy (e.g. warfarin) was given. For patients with PH, bosentan, sildenafil, or Adempas was also prescribed according to clinical conditions.

Pulmonary arterial involvement
For newly diagnosed patients with TA, their whole-body MRA, echocardiology, and pulmonary CT were routinely performed. Combining patients' symptoms and imaging results, patients' PAI can be preliminarily evaluated, then patients with suspicious pulmonary lesions will be required to perform extra examinations such as pulmonary CTA or lung VQ scan to confirm it.
Vascular inflammation, as well as stenosis, dilation, or occlusion of the pulmonary trunk (PT), right/left main PAs, or branches of the right/left main PAs, were identified upon MRA or CTA. Thrombosis/embolism in sub-segmental PAs was confirmed by lung VQ scans combined with CTA. 'Vascular inflammation' was defined as an increased signal intensity of the vascular wall upon MRA compared with that of the back muscle situated beside the vertebral column of the same slice [15] or standard uptake value (SUV) of 18 F-fluorodeoxyglucose on PET-CT > 2.5 in the PA [19]. Vascular stenosis was assessed if the vascular diameter of the lesioned segment was less than that of the adjacent normal segment. Vascular dilation was diagnosed if the vascular diameter of the lesioned segment was < 50% greater than the vascular diameter of the upper or lower normal segments. A 'vascular aneurysm' was documented if the vascular diameter of the lesioned segment was > 50% of the vascular diameter of the upper or lower normal segments.
Imaging results were analysed independently by two experienced radiologists blinded to the clinical data. Discordant interpretations were settled by a discussion based on additional clinical information until consensus was reached.

PH and heart function
PH was considered if patients had typical clinical symptoms (dyspnoea, chest pain, haemoptysis) and satisfied at least one of the following criteria: (i) mean PA pressure ≥ 25 mm, pulmonary artery wedge pressure ≤ 15 mmHg, and pulmonary vascular resistance > 3 Wood units at rest as assessed by right heart catheterization (RHC) [20] and (ii) estimated pulmonary arterial systolic pressure (ePASP) ≥ 40 mmHg and peak systolic velocity of the tricuspid valve > 3.4 m/s upon echocardiography [21]. Echocardiography was applied in patients with RHC contraindications. PH classes were also analysed in TA patients according to different causes: left heart disease, lung disease and/or hypoxia, chronic thromboembolism, or unknown multifactorial mechanisms [22].
Impaired right heart function occurs frequently in patients with chronic pulmonary disorders (especially PH). Cardiac function was also assessed in patients with PAI according to the New York Heart Association (NYHA) Functional Classification [24].

Pulmonary parenchymal lesions
Pulmonary parenchymal lesions within lung segments with PAI were analysed by pulmonary CT or PET-CT. Typical pulmonary features due to PA abnormalities (the mosaic sign, pleural effusion, bronchiectasis, pulmonary infarction) were analysed.

Statistical analyses
Categorical variables are presented as frequencies and percentages. Continuous variables are presented as the mean ± SD. The significance of those parameters among the two groups was determined by the chi-squared test, Fisher's exact test, or unpaired Student's t test, as appropriate. P < 0.05 (two-sided) was deemed significant. SPSS v20.0 (IBM, Armonk, NY, USA) was used for statistical analyses.

PAI
Among 56 patients with PAI, 14 (25%) patients presented PT, right pulmonary artery (RPA), and left pulmonary artery (LPA) lesions; 10 (17.86%) patients had RPA and LPA involvements; 7 (12.5%) and 3 (5.36%) patients demonstrated PT and RPA or LPA lesions, respectively, while single PT, RPA, or LPA involvement was observed in 10 (17.86%), 8 (14.29%), and 5 (8.93%) patients, respectively. Their specific features were listed in Table 2, and the representative images were shown in Fig. 1. Among these patients, 105 PAs were involved,  Occlusion (n (%)) 0 (0) Lung biopsy was carried out in two of the 56 patients with PAI, which indicated haemorrhagic infarction of pulmonary lesions, accompanied by fibrinoid necrosis of the small vessels and infiltration of inflammatory cells (Fig. 3). Special stainings such as PAS staining and acid-fast staining were negative, which excluded common pulmonary infections (Fig. 3). These results confirmed vasculitis in the pulmonary parenchymal lesions of TA patients.  Ground-glass opacity (e) in the right upper lobe of a TA patient with right upper pulmonary arterial branches involvement (f). Cavitation (g) and mass-like consolidation (h) in the patient with severe stenosis of the right main PA (i) and balloon dilation for each); at the time point of 12th month, one (1/26, 3.85%) patient's pulmonary arterial lesions got worsened, while the remaining (25/26, 96.15%) patients' pulmonary arterial involvement kept stable; at the time point of 24th month, pulmonary arterial lesions were improved, stable, and more severe in 2 (11.11%), 14 (77.78%), and 2 (11.11%) patients, respectively. Representative pulmonary CTA and VQ scan images posttreatment were shown in Fig. 4a-f and Supplementary Figure 2B.

Follow-up of patients with PAI
Although patients' pulmonary arterial lesions did not change a lot after medical treatment, their overall conditions were improved. Cough/sputum in four cases after medications and haemoptysis in three cases (one after medication combined with endovascular treatment, two after medications) had resolved. In addition, ePASP upon ultrasonography tended to decrease during patients' follow-ups (baseline, 62.32 ± 28.76 mmHg, n = 27; 6th month, 58.73 ± 24.03 mmHg, n = 15; 12th month, 62.69 ± 29.75 mmHg, n = 13; 18th month, 49.75 ± 27.21 mmHg, n = 4; 24th month, 52.11 ± 34.01 mmHg, n = 9; 36th month, 39 ± 13.45 mmHg, n = 3; Fig. 4g). Eleven patients showed improvement in cardiac function (from IV to III in three patients, from III to II in four cases, from II to I in four patients), 20 patients showed no change, and cardiac function worsened in three cases (from II to III in two cases, from I to II in one patient, Fig. 4h).
Nine patients with lung parenchymal lesions at baseline had repeated pulmonary CT (previously, four cases had pulmonary infarction, two had pleural effusion, two had the mosaic sign, and one had pulmonary cavitation). After treatment, pulmonary infarction in 2/4 (50%) patients had dissipated, 2/2 (100%) patients with pleural effusion had a reduced effusion volume, and 1/1 (100%) patient with pulmonary cavitation has improved. However, no changes were observed regarding other pulmonary lesions upon repeat pulmonary CT. Representative pulmonary CT images were demonstrated in Fig. 4i-l.

Discussion
We discovered that one quarter of TA patients can have PAI, which presented in the PT, main PA, lobar artery, and segmental (or even sub-segmental) arteries. The prevalence of PH was~15% in patients with TA whereas, among TA patients with PAI, it was 50%, which contributed to the impairment of cardiac function. PAI could also cause pulmonary parenchymal lesions (37.5%). Therefore, PAI in patients with TA can affect multiple aspects of cardiopulmonary functions, which worsens the prognosis.
One quarter of patients had PAI, a figure similar to that reported in other countries (5 to 36.7%) [9,25]. Not all patients presented with symptoms such as cough or haemoptysis, which suggested that PAI in TA can be silent [26]. In addition, we did not find an association between PAI and disease activity. This finding was in accordance with a report from China [27] but contrary to a study from Korea that discovered PAI in a high proportion of patients with clinically active TA [5]. Given that TA is indolent, patients might be diagnosed at different stages of TA. Hence, the relationship between PAI and disease activity might differ in different studies.
Stenosis was the most common imaging presentation of the involved PAs, followed by dilation and emboli. Two studies have shown that stenosis or occlusion was the most common presentation of involved PAs in TA [25,28]. Additional lung VQ scans for patients with low perfusion upon MRA revealed pulmonary emboli to be very common in TA patients, which involved multiple segmental (or sub-segmental) PAs. Therefore, low PH is a direct result of PAI. In the present study, half of the patients with PAI suffered from PH. Among them, the ratio of left heart disease-related PH (group 2) was more than that of pulmonary embolism-related PH (group 4), a finding that is in accordance with a report from Wang and colleagues [28]. This phenomenon can be explained by the higher prevalence of aortic involvement than PAI in TA. Moreover, we observed that the PH and cardiac insufficiency were moderate to severe in patients with PAI, implying cardiopulmonary function was badly damaged in patients with PAI.
In TA, PAI can also induce pulmonary parenchymal lesions, which was consistent with the past reports [29,30]. Various pulmonary lesions (the mosaic sign, infarction, pleural effusion, bronchiectasis, cavitation) were observed in regions corresponding to PAI. The mosaic sign and pulmonary infarction are direct pulmonary manifestations on account of hypoperfusion [11,31,32]. Recurrence of pleural effusions has been reported to be associated with disease activity [33,34]. Bilateral effusions are probably induced by cardiac insufficiency, whereas a unilateral effusion might be related to pleural inflammation or small vessel vasculitis in peripheral pulmonary areas [35]. Bronchiectasis has been reported in one case study of TA [36], which we assumed was the result of the involvement of the bronchial arteries in TA. Poor blood perfusion can cause chronic destruction of bronchial tissue and further bronchiectasis. Moreover, cavitation has been reported in TA [37,38], which can be treated with glucocorticoids combined with immunosuppressants, further confirming that cavitation is mediated by abnormal autoimmune reactions.
Patients with PAI should be monitored closely and treated comprehensively. In the present study, pulmonary thrombosis developed in two patients with primary PAI, which was probably due to endothelial injury under the condition of vascular inflammation. Since they also had chronic heart failure due to long-term PH, it was assumed that their cardiopulmonary dysfunction contributed to their death to a great extent. Besides the usual treatment for TA, PH-targeting agents are necessary for patients with PH (especially group 4 PH). In some cases, surgery is Fig. 4 Pulmonary and cardiac conditions after treatment. a-f Pulmonary arterial changes before and after treatment: the right main pulmonary artery was shown upon CTA after balloon dilation intervention as well as immunosuppressive treatment in one patient (prednisone and azathioprine; a before treatment; b after treatment); blood supply to the right lung was also increased upon lung VQ scan after 1 year of treatment with prednisone and sirolimus in another patient (c, e before treatment; d, f after treatment). g The ePASP upon ultrasonography tended to decrease during patients' follow-ups (baseline, 62.32 ± 28.76 mmHg, n = 27; 6th month, 58.73 ± 24.03 mmHg, n = 15; 12th month, 62.69 ± 29.75 mmHg, n = 13; 18th month, 49.75 ± 27.21 mmHg, n = 4; 24th month, 52.11 ± 34.01 mmHg, n = 9; 36th month, 39 ± 13.45 mmHg, n = 3) despite of treatment regimens. h Patients' cardiac function was improved in most patients despite of treatment regimens. i, j Infarcted lesions dissipated after 12 months of treatment with prednisone and AZA as well as agents for PH in one patient. k, l The cavitation in the right lung apex became smaller after treatment with prednisone and sirolimus for 7 months in one patient also indicated. Our study demonstrated that symptoms, pulmonary pressure, and cardiac function were improved markedly after treatment in most patients, but some patients did not respond well. Hence, ascertaining the factors responsible for such different therapeutic responses must be undertaken. In addition, novel treatment strategies must be explored. These strategies will lay the foundation for future precise treatment for TA.
Some limitations should be addressed in the present study. Firstly, since RHC is invasive and would be contradicted in some patients with vascular stenosis, ePASP upon echocardiography was mainly applied in the present study to evaluate patients' PH, which may decrease the accuracy to some extent. Secondly, pulmonary arterial thrombosis might be low estimated since pulmonary CTA and VQ scan were not routinely performed for all the patients enrolled.