Patients

A total of 31 consecutive patients with AAV treated at a single tertiary center were enrolled in the study. The inclusion criteria comprised the diagnosis of AAV with the presence of antibodies against myeloperoxidase (anti‑MPO) or proteinase 3 (anti‑PR3), no disease activity, defined as 0 points in the Birmingham Vasculitis Activity Score, version 3 (BVAS v.3),¹⁹ and age between 18 and 70 years. We excluded the patients with comorbid conditions potentially affecting the nervous system, such as diabetes mellitus, uncontrolled thyroid disorders, or chronic kidney disease (stage 3 or higher), patients on medications, such as β-blockers, rate‑controlling calcium channel blockers, or tricyclic antidepressants, and those with contraindications to MRI of the head (eg, metallic implants, artificial heart valves, vascular clips, and other potentially ferromagnetic foreign bodies, claustrophobia, pacemakers, neurostimulators, and other biostimulators). One patient enrolled in the study did not finally undergo fMRI due to significant anxiety.

The control group consisted of 30 healthy volunteers who matched the study group in terms of age and sex. Each participant provided a written informed consent. The study was conducted in accordance with the Declaration of Helsinki. It was reviewed and approved by the Bioethics Committee of the Jagiellonian University (KBET/286/B/2012). Demographic and medical history data were collected from the available medical records.

Autonomic symptoms questionnaire

Each participant was asked to complete the COMPASS‑31 questionnaire, which consists of 31 questions grouped into 6 domains: orthostatic, vasomotor, secretomotor, gastrointestinal, urinary, and pupillomotor. The score for each domain was weighted according to the previously described methodology⁸; then, all the scores were added to give a total score ranging from 0 to 100 points.

Magnetic resonance imaging data acquisition

MRI was performed using a 3T scanner with a 32‑channel head coil (Magnetom Skyra, Siemens Healthcare, Erlangen, Germany).

The procedure was divided into 2 stages. In the first stage, anatomical scans were performed. Subsequently, fMRI was carried out during the Valsalva maneuver.

High‑resolution, whole‑brain anatomical images were acquired using TSE T2, FLAIR, and SWI3D sequences with respective transverse slices of 1.5, 3, and 5 mm, as well as with SPC T2 and T1 MPRAGE sequences in 0.9‑mm sagittal slices. For the T1 MPRAGE scan, a total of 176 sagittal slices were obtained (voxel size, 0.94 mm³; matrix size, 256 × 256; field‑of‑view [FOV], 240 mm × 240 mm; repetition time [TR], 2300 ms; echo time [TE], 2.29 ms; flip angle, 8 °). Then, the scans were transferred to and analyzed by an experienced radiologist, who was blinded to the participant data.

For functional data acquisition, the gradient‑echo EPI sequence was used. During the task sequence, 155 volumes, each consisting of 47 axial slices, were acquired with the following parameters: TR, 2500 ms; TE, 27 ms; 3‑mm isotropic voxel, flip angle, 90 °; FOV, 192 mm × 192 mm, GRAPPA acceleration factor 2. The task sequence was followed by a resting‑state sequence with the same scan parameters, lasting approximately 10 minutes (240 measurements). The first 4 volumes (dummy scans) were discarded from the analysis due to the magnetic saturation effect. A gradient fieldmap was obtained with the same geometric parameters.

Physiological data were collected using a scanner‑compatible electrocardiogram, a respiratory belt placed around the abdomen, and a pulse‑oximeter placed on the index finger of the left hand.

Valsalva maneuver

To apply the Valsalva maneuver in a comparable way in each patient, it was conducted according to the protocol based on the previous studies by He et al¹⁷ and Bohr et al.¹⁸ Each patient was asked to perform the Valsalva maneuver 6 times during the fMRI procedure. The task was explained in detail to each participant and a preprocedural trial was conducted. The visual aid was prepared and generated using E–Prime 2.0 software (Psychology Software Tools, Pittsburgh, Pennsylvania, United States). It was presented on a 32‑inch screen located behind the MRI scanner and approximately 100 cm from the head coil. The participants were able to see the screen using a single mirror placed on the head coil.

The single Valsalva maneuver lasted 1 minute and consisted in a 16‑second forceful exhalation against a single‑use plastic tube with an antimicrobial filter, connected to a custom‑made pressure sensor located outside the MRI scanner room (Ober‑Consulting, Poznań, Poland). Task performance feedback was displayed to the patient graphically in the form of a grey bar changing its length proportionally to the pressure load, and switching its color to green once a threshold of 35 mm Hg for women and 40 mm Hg for men was achieved. The participants were asked to achieve the threshold in the shortest possible time. Following termination of the forceful expiration, a 44‑second rest period ensued, and the instruction “Breathe peacefully through your nose” was displayed on the screen. The participants were asked not to remove the plastic tube from their mouth between the maneuver attempts and not to use their hands as support for the tube.

Functional magnetic resonance imaging analysis

Preprocessing of the fMRI data was performed using the Analysis of Functional NeuroImage (version 17.3.03)²⁰ and the FMRIB Software Library (FSL; version 5.0.9).²¹ Anatomical images were skull‑stripped (3dSkullStrip) and coregistered to the Montreal Neurological Institute (MNI) space using nonlinear transformation (3dQWarp). The cerebrospinal fluid (CSF) mask was created using a segmentation procedure (3dSeg). The functional data preprocessing started with despiking (3dDespike) and slice timing correction (3dTshift); subsequently, motion (3dvolreg) and image distortion (Fugue, FSL) corrections were performed. After coregistration to the skull‑stripped anatomical images, the functional images were coregistered to the MNI space using the transformation matrix from nonlinear anatomical normalization. Then, the images were rescaled to represent the percent signal change, and the time course of CSF was extracted. Finally, functional data were detrended and the CSF signal was regressed out.

Data from 3 participants were discarded from further analysis (due to technical issues—missing task performance data [n =1] or invalid task performance [n = 2]).

The generalized least‑squares time series analysis with the prewhitening option (3dREMLfit) was performed using the model that included 6 movement parameters and 3 task regressors: the Valsalva maneuver modeled as the 16‑second block and cue, and the rest onset modeled as the event. As a result, the map of activity during the task was generated. Subsequently, the comparison between the patients and controls was performed using the t test.

The region‑of‑interest (ROI) analysis was performed using a mask generated from the Harvard–Oxford subcortical atlas. The time course was extracted from preprocessed data and 30‑second epochs were averaged across correct trials for each participant, and then across participants for both groups.

Statistical analysis

Standard descriptive statistics were used. The normality of distribution of variables was checked with the Shapiro–Wilk test. To compare the 2 groups, the χ² test (with Yates correction, if needed) for dichotomous variables and the Mann–Whitney test for continuous variables were used. A P value below 0.05 was assumed as significant. Calculations were performed with StatSoft Statistica 13 software (StatSoft, Tulsa, Oklahoma, United States).

Patient characteristics

A total of 31 patients (12 women, 19 men) were included in the study group. Of those, 25 had anti‑PR3 antibodies and 6 had anti‑MPO antibodies. The most prevalent organs involved were the respiratory system, kidneys, and ENT. All patients were treated with steroids and most of them also with cyclophosphamide during the remission induction phase. The most frequent comorbidities were arterial hypertension, hypothyroidism, and secondary immune deficiency.

The control group consisted of 30 sex- and age‑matched individuals (12 women, 18 men). Six of them had a history of arterial hypertension.

The general characteristics of the AAV cohort and a comparison of selected parameters between the AAV patients and controls are presented in Table 1. Previous cerebral ischemic stroke and cerebral hemorrhage were reported by 1 AAV patient each. None of the controls reported a cerebral episode.

Autonomic dysfunction symptoms

The patients with AAV had a higher median (interquartile range [IQR]) COMPASS‑31 score than the controls (12.86 [4.45–23.91] vs 2.99 [0–7.12], respectively; P <⁠0.01). The patients with nervous system involvement did not show a higher COMPASS‑31 score than those without such symptoms (median [IQR], 15.73 [6.73–26.02] vs 10.96 [4.45–18.28], respectively; P = 0.71).

Structural magnetic resonance imaging

Structural MRI findings are presented in Table 2. The most frequently reported changes in the brain area were nonspecific white matter lesions (WMLs) in various locations. They were found in half of the study group and in half of the controls. In both groups, 20% (3/15) of the abnormalities were described as numerous or moderate. Arterial hypertension was present in 40% (6/15) of the AAV patients with WMLs, and 40% of the controls.

In the AAV group, the radiological signs of sinusitis were more common than in the controls (80% vs 50%, respectively; P = 0.03), with severe sinus involvement present only in the AAV group.

Valsalva maneuver

Overall, 2 out of 60 participants did not perform the task correctly, that is, they did not reach the threshold level of pressure. The rest of the participants performed the task with 82% accuracy—almost 5 of 6 Valsalva trials were correct. There was no difference in the task performance between the study patients and controls (P = 0.14).

Functional magnetic resonance imaging results

The generalized least‑squares analysis revealed several clusters activated during the Valsalva maneuver (P <⁠0.001; t >4.6): the right intraparietal sulcus, right middle temporal gyrus, right superior temporal gyrus, right hippocampus (partially including the amygdala), left thalamus, bilateral motor and sensory cortices, bilateral putamen, as well as the cerebellum and brainstem. Detailed information is provided in Table 3 and Figure 1.

The ROI analysis was performed for the bilateral amygdala, putamen, and pallidum clusters. The results are presented in Figures 2 and 3. There were no differences between the AAV group and the control group.

Conclusion

Patients with AAV experience symptoms related to the ANS dysfunction more often than healthy individuals; however, no differences in the functioning of the brain centers responsible for the ANS have been demonstrated based on fMRI performed during the Valsalva maneuver. Symptoms of ANS dysfunction in AAV are more likely to originate from peripheral nerve damage, because central autonomic control is preserved. Further studies investigating the nature of the autonomic symptoms are needed.