Impaired small fiber conduction in patients with Fabry disease: a neurophysiological case–control study

We included 76 consecutive Fabry patients in this mono-center case–control study (median age 43 years, range 16–73) in whom the diagnosis of FD was confirmed by measurement of α-GAL activity in leucocytes (http://​www.​metabolic-genetic-disease.​gmxhome.​de; Munich, Germany) and genetically ascertained. The patients were prospectively recruited (2009–2011) through the Wurzburg Fabry Center for Interdisciplinary Therapy (FAZIT), University of Wurzburg. FAZIT is a tertiary referral center where patients are seen from all over Germany to confirm the diagnosis and to initiate treatment. The cohort included 31 men (median age 39 years, 16–62) and 45 women (median age 43 years, 18–73). Thirty-four patients (24 men, 10 women) were on enzyme replacement treatment (ERT; biweekly infusions of agalsidase beta [1 mg/kg body weight] in 29 patients and of agalsidase alpha [0.2 mg/kg body weight] in 5 patients). The median time on ERT at study enrolment was 4.7 years (range 0.1-9.3 years). QST and skin punch biopsy data of 66 subjects from our patient cohort were part of a data set published earlier [5].

We compared our data with data of age- and gender-matched healthy controls as detailed below. Inclusion criteria for healthy controls were: ≥16 years, no FD, no neuropathy, no neuropathic pain or other sources of pain, normal sural nerve conduction.

The study was approved by the Wurzburg Medical School Ethics Committee and was conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all study participants.

All patients underwent thorough neurological examination and were assessed using pain questionnaires: the German version of the Neuropathic Pain Symptom Inventory (NPSI) [9, 10] and the Graded Chronic Pain Scale (GCPS), modified to a four-week recall [11]. The NPSI investigates pain intensity and characteristics resulting in a sum score between 0 (no pain at all) and 1 (maximum pain). The 24-hour recall version was used. From the GCPS we used the total score of the three pain intensity items as an indicator of pain severity, and the total score of the three items rating interference with social, occupational, and recreational activities as a disability score. To address depressive symptoms we used the German version of the Center for Epidemiologic Studies Depression Scale (“Allgemeine Depressionsskala”, ADS) [12]. The ADS ranges from 0 to 60; a score ≥ 16 is assumed clinically significant. The one week recall version was used. To make the diagnosis of small fiber neuropathy, clinical presentation, QST findings, and intraepidermal nerve fiber density (IENFD) were assessed according to Lacomis [13] (methods description see below). Depending on the number of pathological findings, definite, probable, and possible small fiber neuropathy was diagnosed. Laboratory tests included whole blood and differential cell counts; C-reactive protein; serum electrolytes; renal, liver, and thyroid function tests; erythrocyte sedimentation rate. Renal function was assessed by determining the glomerular filtration rate (GFR). Renal function, as one indicator of FD severity, was considered normal if GFR ≥ 60 ml/min/1.73 m² and reduced if GFR < 60 ml/min/1.73 m²[14‐16].

The right sural nerve was investigated in all study participants using surface electrodes and following a standard procedure [17] to exclude large fiber neuropathy. The results were compared with laboratory normal values: for antidromic sural nerve sensory nerve action potential (SNAP) amplitude ≥ 10 μV for age < 65 years, ≥ 5 μV for age > 65 years; for sural nerve conduction velocity (NCV) > 40 m/s for all ages.

QST was performed using a calibrated device (Somedic, Hörby, Sweden) and following a standardized procedure [18]. For individual analysis patients` data were compared with published reference values [19]. For group analysis patients` data were compared with values of age- and gender-matched healthy controls. All subjects were investigated at the left dorsal foot. Based on the log transformed raw values for each QST item a z-score sensory profile was calculated as follows: z-score = (value of the subject – mean value of controls)/standard deviation of controls. Negative z-scores indicate loss of sensation, positive z-scores indicate gain of sensation. We determined cold and heat detection thresholds (CDT, HDT) and the ability to detect temperature changes (thermal sensory limen, TSL) as small fiber functions. Paradoxical heat sensation (PHS) was recorded if the subject experienced cold as heat. We additionally determined the vibration detection threshold (VDT) as large fiber function. The control group for QST measurements consisted of 76 age- and gender matched healthy volunteers (45 female, 31 male; median age: 44 years, 16–73). The difference in age between patients and matched controls was three years at maximum.

PREP were recorded as previously described [20]. PREP were elicited by consecutive stimulation at the right and left side from face (above eyebrow), hands (medial phalanx second and third digit), and feet (dorsum) using superficial planar concentric electrodes (Inomed Medizintechnik GmbH, Lübeck, Germany) and a stimulator (Digitimer DS7A, Welwyn Garden City, UK). The potentials were recorded from Cz by a subcutaneously placed needle electrode referred to linked earlobes (A1 - A2) of the international 10–20 system using Signal Software (Version 2–16; Cambridge Electronic Design, Lt., UK). For stimulation 20 triple pulses with an intensity of two-fold of the individual pain threshold, duration of 0.5 ms, and random inter-stimulus interval of 15 to 17 seconds were applied. The potentials were recorded using the following setting: gain: x 5000, bandwidth: 1 Hz-1 kHz, sweep length: 400 ms, digitalization sampling rate: 2.5 kHz. The individual pain threshold was determined by stimulation of the area of interest twice with increasing and decreasing current intensities until the subject reported a pin-prick sensation. The average value was calculated as the individual pain threshold. Two sets of averaged curves (from n = 10 single sweeps each) were investigated for reproducible N1- (i.e. first negative peak), P1- (i.e. subsequent positive peak) latencies and peak-to-peak amplitudes (PPA) using MATLAB software (Version 7.7.0.471, The Math Works, Ismaning, Germany). All records were individually evaluated to exclude technical or biological artifacts by the same investigator who was blinded as for the diagnosis: data were assessed off-line using coded files. The control group for PREP recordings consisted of 65 healthy controls (41 female, 25 male; median age: 45 years, 21–75). Subjects with cardiac pacemakers or with seizures in their medical history were excluded.

For assessment of intraepidermal nerve fiber density (IENFD) 5-mm skin specimens were obtained (punch device by Stiefel, Offenbach, Germany). Biopsies were taken from the lower leg (10 cm above the lateral malleolus) and from the back (at th5 level). Skin samples were processed as described previously [21]. They were immunoreacted with antibodies to protein-gene product (PGP) 9.5 (Ultraclone, UK, 1:800; primary antibody) with goat anti-rabbit IgG labelled with cyanine 3.18 fluorescent probe (Amersham, USA, 1:100; Cy3, secondary antibody), and IENFD were quantified by an observer blinded to the identity of the specimen following published rules [22]. As reference value we took data of a cohort of normal samples collected in our laboratory: lower leg: n = 110 (63 females, 47 males), median age: 50 years (range 20–84), median IENFD 7 fibers/mm, range 1–15 fibers/mm; back: n = 42 (23 female, 19 males), median age: 50, range 20–81, median IENFD 22 fibers/mm, range 6–40 fibers/mm.

We used IBM SPSS Statistics Version 20.0 (IBM, Ehningen, Germany) for statistical analysis and creation of graphs. Non-normally distributed data were compared with the non-parametric Mann–Whitney test. Data with normal distribution were analyzed using one-way ANOVA. Data distribution was tested with the Kolmogorov-Smirnov-test and by observing data histograms. Results of non-normally distributed data are given as median and range and are illustrated as box plots. Results of normally distributed data are given as mean +/− standard deviation and are illustrated as bar graphs. For correlation analyses we used the bivariate Spearman correlation. P-values < 0.05 were considered significant.

Table 1 gives baseline data of the patient population. Neurological examination was normal in 44/76 cases (58%). The pathological findings mainly in the peripheral nervous system are summarized in Table 1. Neurophysiological assessment of the sural nerve revealed normal SNAP and NCV in 71/76 (93%) cases; in four patients a slight reduction in SNAP and NCV was found and in one case no potential could be obtained as a sign of sensory polyneuropathy. In none of the patients motor symptoms indicative of neuropathy were present. As for the presence of small fiber neuropathy: 16 patients had definite (11 men, 5 women), 35 patients had probable (17 men, 18 women), and 25 patients had possible small fiber neuropathy (3 men, 22 women). Assessment of pain using the NPSI did not show any intergroup differences, as almost all patients reported on “no pain” in the last 24 hours (data not shown). The GCPS (4-week recall) revealed higher pain intensities in FD patients (p < 0.001 for both genders) and higher disability scores due to pain (p < 0.05 for males, p < 0.001 for females) compared to healthy controls (Additional file 1: Figure S1a). Forty-three patients did not take analgesic medication; 19 patients used analgesic drugs on demand (n = 8: acetaminophen; n = 5: aspirin; n = 4: tramadol; n = 1 each: tilidine, ibuprofen, metamizol); twelve patients were on regular analgesic medication (n = 4: pregabalin; n = 2: gabapentin; n = 2: metamizol; n = 1 each: flupirtine, carbamazepine, tilidine, tramadol). The ADS showed higher scores for male and female FD patients compared to controls (p < 0.001 each, Additional file 1: Figure S1b). None of the patients was on antidepressant therapy. In 15/76 (20%) cases renal function was impaired; five male patients were on dialysis.

In line with previous reports [4, 5, 23] we found elevated CDT (p < 0.01) and WDT (p < 0.05) in male Fabry patients compared with male controls (Figure 1a). Female patients did not differ from female controls (Figure 1b). Male patients had higher thermal perception thresholds and impaired vibration sense compared to female FD patients (CDT: p < 0.05, WDT: p < 0.001, TSL: p < 0.01, VDT: p < 0.01). PHS was present in none of the male patients and in 9/45 (20%) female FD patients. As shown earlier male patients with impaired renal function (GFR < 60 ml/min/1.73 m²) had most severe impairment in thermal perception (Additional file 2: Figure S2). Women with reduced α-GAL activity did not differ in QST parameters from women with normal enzyme activity.

PREP data from both sides were pooled, since there was no difference between the right and left side for any subject group, investigated area, or PREP parameter. Stimulus intensities did not differ between groups for all three areas. PREP N1 and P1 latencies did not differ between FD patients and controls (Figure 2a-c). Peak-to-peak amplitudes (PPA), however, were lower in male Fabry patients compared to male controls upon stimulation at face (p < 0.01), hands (p < 0.05), and feet (p < 0.01, Figure 2d-f). Male Fabry patients also had lower PPA compared to female patients (hand: p < 0.05, foot: p < 0.01, Figure 2d-f). When comparing data of male Fabry patients with GFR < 60 and GFR > 60 with data of healthy male controls, only patients with impaired renal function showed reduced PPA after eliciting PREP from face (p = 0.012), hands (p = 0.007), and feet (p = 0.007, Figure 3a-c). Stimulus intensities did not differ between patients and controls at any site. Women with reduced α-GAL activity did not differ from women with normal enzyme activity as for PREP parameters. Additional file 3: Figure S3 shows an example of a PREP recording after stimulation at the face.

Twenty male patients (20/31, 65%) and 27/45 female patients (60%) agreed to skin punch biopsy. In line with previous findings [5] we found reduced distal and proximal IENFD in male patients compared to healthy controls. Fiber reduction was most prominent in male patients with impaired renal function: distal (p < 0.001) and proximal (p < 0.05) IENFD was lower compared to male healthy controls (Additional file 4: Figure S4). Similarly, female patients with reduced renal function had lower distal IENFD than female controls (p < 0.05, Additional file 4: Figure S4a), while skin innervation was not different at the back (Additional file 4: Figure S4b). Lower GFR correlated with lower distal IENFD (correlation coefficient: 0.508, p < 0.001) and with lower proximal IENFD (correlation coefficient: 0.369, p < 0.05). Women with reduced α-GAL activity did not differ from women with normal enzyme activity as for skin innervation.

A positive correlation was found between PREP PPA elicited at the feet and thermal perception thresholds (CDT, WDT) determined at the feet of Fabry patients (Figure 4a, b). In male patients a positive correlation was found for CDT (correlation coefficient: 0.564, p < 0.01) and for WDT (correlation coefficient: 0.390, p < 0.05). In female patients a positive correlation was found only for CDT (correlation coefficient: 0.340, p < 0.05). TSL and VDT did not correlate with PREP PPA in both genders (Figure 4c, d). Also, distal IENFD did not show a correlation with PREP parameters obtained after stimulation at the feet (illustrated for PPA in Figure 4e).