Twenty-four female patients with PBC (mean age: 60.4 ± 7.1 years; range: 45-73 years) from the Hepatological Outpatient Unit of Semmelweis University, Budapest and 20 age-matched healthy female controls (mean age: 59.3 ± 6.8 years; range: 44-72 years) were recruited for this cross sectional study. The diagnosis of PBC was based on characteristic clinical and laboratory data, AMA M2 positivity and liver biopsy. The severity of liver disease was assessed by histologic classification. Stage I: 2, stage II: 5, stage III: 12 and stage IV: 5 patients. Full medical history was taken, followed by thorough physical and neurologic examination in each patient and control. Patients were only included if they were normotensive, ie. no history of hypertension, and at the time of inclusion visiting office blood pressure < 140/90 mmHg calculated by the mean of three measurements using Korotkov’s technique and no evidence of disease known to affect autonomic function (e.g. other hepatic disease, cardiovascular, kidney, endocrinologic, neurologic and psychiatric disorders including alcoholism). None was taking any antihypertensive drugs or other medications, apart from ursodeoxycholic acid, vitamin D and calcium supplementation[18], and none had ascites. The healthy controls were recruited from the staff of our institution and their family members. Every participant was asked to refrain from consuming caffeine and alcoholic beverages, and tobacco products 12 h before autonomic testing.

The autonomic function was explored by the five standard cardiovascular reflex tests[19]. Heart rate tests (heart rate responses to deep breathing, the 30/15 ratio and the Valsalva ratio) mainly reflect parasympathetic function while blood pressure responses to sustained handgrip and standing primarily allow the assessment of sympathetic integrity. Patients with at least one abnormal or two borderline cardiovascular tests were considered to have autonomic neuropathy. The same research assistant using a computerized ECG-recording-analyzing system developed by Innomed Inc, Budapest, Hungary, performed all reflex tests.

Two channel 24-h ECG recordings were done by CardioTens equipment (Meditech, Budapest, Hungary). This device complies with the requirements of the British Hypertension Society and the Association for the Advancement of Medical Instrumentation protocols. Automatic filters were used to continuously restore baseline and filter background and muscle noise. Analysis of stored data was done by Medibase software. The recording was also edited using visual control and manual corrections were made to omit ectopic beats, arrhythmic events and noise effects and only normal-to-normal beats (NN intervals) were used for further analysis. Ratio of normal beats to total number of beats was > 95% in both groups.

To characterize 24-h HRV, time domain and frequency domain methods were used. Since there were several parameters cited in the literature to assess HRV, we selected a limited number of parameters according to the recommendations of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology[11]. Statistical time domain parameters could be calculated from either the direct measurements of NN intervals or from the differences between NN intervals. The following parameters were calculated from direct measurements of NN intervals: standard deviation of all NN (SDNN) intervals reflecting all the cyclic components responsible for variability and standard deviation of the averages of NN (SDANN) intervals in all 5-min segments of the entire recording, an estimate of the changes in heart rate due to cycles longer than 5 min.

Statistical time domain parameters deriving from NN interval differences are RMSSD and pNN50. RMSSD (the square root of the mean of the sum of the squares of differences between adjacent NN intervals) is an estimate of short-term components of HRV. PNN50 (the proportion of adjacent NN intervals differ by more than 50 ms) was considered to reflect the vagal tone of the heart.

A simple geometric time domain parameter, HRV triangular index (HRVTI) was also computed. HRVTI is the integral of the density distribution (the number of all NN intervals) divided by the maximum of the density distribution. HRVTI represents overall HRV.

In the frequency domain analysis (power spectral density analysis) of heart period oscillations low- (0.04-0.15 Hz) and high- (0.15-0.4 Hz) frequency bands of the power (i.e. variance) spectrum (power distribution as function of frequency) was performed. The following frequency domain measures were computed: TP (total power: variance of all NN intervals), LF (power in the low frequency range) which is under both sympathetic and parasympathetic influences, HF (power in the high frequency range) which is an acknowledged measure of the parasympathetic (i.e. vagal) modulations.

Peripheral sensory function was characterized by the evaluation of the current perception threshold (CPT) with a neuroselective diagnostic stimulator (Neurotron, Baltimore, MD, USA), which permits transcutaneous testing at three sinusoidal frequencies (2000 Hz, 250 Hz and 5 Hz). The intensity of the stimulating current was changed within the range from 0.01 to 9.99 mA. The neurometer is the first instrument designed for the overall assessment of all types of sensory fibres. As demonstrated by the results of comparative trials conducted earlier[20], CPT values measured during high frequency stimulation correlated best with tests of large fibre function and low frequency CPT values correlated with tests of small fibre function. Median and peroneal nerves (digital branches) were studied.

The Local Regional Committee of Science and Research Ethics approved the study. Written informed consent was obtained.

All analyses were performed using Statistica Software. Data are expressed as mean ± SD and were compared between groups by Student’s t-test. Correlations between variables were analysed by partial correlation coefficient calculation adjusted for age. P < 0.05 was regarded as statistically significant.

Using standard cardiovascular tests[18], 14 PBC patients (58%) had at least one abnormal autonomic function test. Among these patients parasympathetic neuropathy was found in 8 (57.1%) patients, sympathetic nerve dysfunction was observed in 2 patients, and 4 subjects had both parasympathetic and sympathetic damage.

As a novel finding we proved that both time domain and frequency domain parameters of HRV were significantly reduced in PBC patients compared to controls. Each patient had at least one abnormal parameter of HRV.

Results of cardiovascular reflex tests and HRV parameters are presented in Table 1. The heart rate response to deep breathing (P = 0.001), as well as to standing (P = 0.03) and Valsalva manoeuvre (P = 0.01) was significantly lower in PBC patients than in age matched control subjects. A more profound decrease of systolic blood pressure after standing (P = 0.03) was found in patients compared to healthy controls. By HRV analysis most of the time-domain indices were significantly lower in patients than in controls (PNN50: P = 0.0008; HRVTI: P = 0.004; RMSSD: P = 0.006 and SDNN: P = 0.015). PBC patients also showed significantly lower total power (P = 0.0001), power of LF band (P = 0.00007) and of HF band (P = 0.004).

At least one abnormal sensory parameter was detected in 13 patients (54%), of whom 12 had hyperaesthetic type, and only one had hypoaesthetic type sensory nerve dysfunction. Among patients with sensory neuropathy the lower extremities were affected in all 13 patients, while 3 patients had abnormal CPT values at upper extremities testing. Lower CPT values, indicating hyperaesthesia, were found in PBC patients compared with age matched controls at peroneal nerve testing at all three frequencies (P < 0.01) as well as at median nerve testing at 250 Hz (P = 0.03). The CPT values of patients and controls are shown in Table 2.

After adjustment for age, the longer duration of the disease was associated with less prominent increase of diastolic blood pressure during sustained handgrip test (r = -0.52, P = 0.01). Duration of the disease also correlated with reduced SDNN and SDANN (r = 0.47 and r = -0.45, P<0.05, for both). The severity of PBC (stage) was found to negatively correlate with lower HRVTI (r = -0.6, P = 0.01) and lower SDANN (r = -0.49, P = 0.04) as well.

Partial correlation analysis revealed that lower prothrombin activity was associated with lower heart rate response to standing (r = 0.79, P = 0.006) as well as to deep breathing (r = 0.63, P = 0.04). The serum albumin positively correlated with SDNN and HRVTI (r = 0.47 and 0.57, P < 0.05, for both). Serum AST and ALT levels negatively correlated with SDNN (r = -0.54, P = 0.01, for both). Positive correlations of SDNN (r = 0.62, P = 0.004) and Valsalva ratio (r = 0.51, P = 0.02) with serum trigliceride levels were found. These relationships remained significant after age adjustment.

Negative correlations of serum ALT with CPT values at median nerve testing at 250 HZ (r = -0.56, P = 0.005) as well as with CPT at peroneal (r = -0.48, P = 0.03) and median nerve (r = -0.45, P = 0.02) at 5 Hz were revealed. AST and ALP levels were inversely related to CPT values at peroneal nerve testing at all three frequencies, as well as to those at median nerve testing at 250 Hz and 5 Hz. An inverse relationship was also found between serum bilirubin levels and CPT values at median nerve testing at 5 Hz (r = -0.43, P = 0.04). None of these relationships was altered by adjustment for age.

Interestingly no correlation was found between peripheral sensory nerve function and duration and severity of PBC. Furthermore, no association was found between sensory nerve function and serum lipid levels, prothrombin activity and serum albumin levels.

Reduced total power was associated with lower CPT values testing median nerve at 250 Hz (r = 0.69, P = 0.0001), and at 5 Hz (r = 0.62, P = 0.002), as well as with those testing peroneal nerve at 2000 Hz (r = 0.53, P = 0.01). HF-power was positively related to CPT values at peroneal nerve testing at 2000 Hz (r = 0.51, P = 0.01), LF-power correlated positively with CPT values testing median nerve at 250 (r = 0.53, P = 0.01), as well as at 5 Hz (r = 0.47, P = 0.03). A significant positive correlation was observed between the SDNN and the CPT values testing the median nerve at 2000 Hz (r = 0.60, P = 0.003), as well as at 250 Hz (r = 0.53, P = 0.01) and at 5 Hz (r = 0.50, P = 0.02). Lower SDANN values were associated with lower CPT values at the median nerve at 2000 Hz (r = 0.52, P = 0.01). The PNN50 values correlated positively with CPT values testing the peroneal nerve at 2000 Hz (r = 0.50, P = 0.02) and the beat-to-beat variation was also positively related to CPT values at the peroneal nerve at 250 Hz (r = 0.47, P = 0.02), even after adjustment for age.