This study hypothesized that patients with liver cirrhosis are at increased risk of developing peripheral neuropathy, so we aimed to highlight the importance of good screening and early detection of subclinical polyneuropathy in patients with compensated cirrhosis and to explore any significant association between characteristics of liver cirrhosis and peripheral nerve dysfunction.
All patients in our study were post-hepatitis C being the most common cause of liver cirrhosis in Egypt. Moreover, most of them had mild liver cirrhosis based on Child–Pugh classification (92% were Child-A). We excluded patients with decompensated cirrhosis due to their frailty and disability caused by chronic malnutrition and muscle wasting. To assess the neuropathy symptoms, we used a subjective questionnaire (NSC) and it revealed that 42% of the patients were asymptomatic and 52% had mild to moderate pain and paresthesia.
We conducted extensive electrophysiological assessment (sensory and motor nerve conduction studies of four limbs and QMUP analysis of two distal muscles) and using the proposed defining criteria of HN (more than two abnormal values), the prevalence of HN in our patients was 82%. The pattern of peripheral nerve dysfunction was almost axonal degeneration (or mixed axonal and demyelinating) and involved both motor and sensory fibers and ranged from mild to moderate affection.
Most previous studies showed a mixed pattern of peripheral nerve involvement (axonal and demyelinating, sensorimotor affection) in agreement with our findings [4, 6, 14]; however, other studies showed histopathologic evidence of segmental demyelination and remyelination in sural nerves of patients with hepatic neuropathy [15, 16].
The inter-group comparison revealed that cirrhotic patients had significantly abnormal nerve conduction finings (most commonly ulnar and peroneal distal motor latency and conduction velocity, ulnar sensory latency, amplitude, and conduction velocity, in addition to absent lower limb sensory responses in a significant proportion of the patients).
In reviewing the literature, there is a large controversy regarding the prevalence of HN (ranging from 20 to 63% based on clinical findings and from 18 to 73% based on electrophysiological studies) which may be explained by the different approaches in defining the presence of neuropathy and the different spectrum of the selected cirrhotic patients due to different etiologies and disease severity [3,4,5,6].
The higher prevalence in our study could be attributed to the extensive approach in electrophysiological testing of sensory and motor nerves in all limbs together with the use of quantitative MUP analysis which offers a more objective way in assessment of peripheral nerve motor fibers and muscles compared to the qualitative and subjective approach of routine EMG test.
This quantitative approach is carried out through analysis of basic and derived metrics extracted from a group of selected MUPs from the tested muscles. To our knowledge, most previous studies which evaluated the hepatic-related polyneuropathy have only focused on routine nerve conduction or autonomic function testes and no one used quantitative MUPs analysis in cirrhotic patients.
In the current study, the prevalence of HN was higher when using the combined electrophysiological methods (NCS and QEMG) than using a single modality (58% and 48% respectively). This combined approach may serve for early detection of any subtle or subclinical peripheral nerve dysfunction in the compensated cirrhotic patients.
Moreover, the results of comparing different electrophysiological parameters using ROC curve analysis reported that the most sensitive parameters were (in order) ulnar sensory distal conduction velocity and peak latency, common peroneal conduction velocity and ulnar distal motor latency. These sensitive measures together with lower limb sensory responses could be used as a rapid screening tool to detect cirrhotic patients with polyneuropathy. In concordance with these results, the peroneal, ulnar, and sural nerve conduction responses were among the mostly discriminating nerves of HN in previous studies [4].
In our study, patient age and the severity of liver disease (assessed by both child and ALBI scores) were found to have a significant direct mild to moderate correlation with the severity of HN. Moreover, a significant positive correlation was found between smoking and incidence of HN.
Similarly, Jain et al found that PN was higher in cirrhotic patients older than 60 years compared to the younger age but odds ratio for age was not statistically significant [6]. Moreover, Santoro et al. found in their large multi-center study a significant correlation between the old age and prevalence of PN in HCV population [17]. In a large screening study by Beghi et al., the prevalence of PN among healthy elderly people with no recognized risk factors was 1.6% compared to 10.9% in age-matched patients with non-alcoholic liver cirrhosis [18].
It is well known that elderly are at more risk of peripheral nerve dysfunction due to processes of denervation that occur with normal aging at accelerated rate beyond 60 years of age, and so, it may be difficult to discriminate the effects of normal aging from pathological conditions [19,20,21].
Turning now to the positive correlation between smoking and HN, a similar relation was found in another study [6]. Moreover, Clair et al. concluded in their review a positive association between smoking and incidence of diabetic PN through its microvascular complications and its adverse effects such as systemic inflammation, oxidative stress, endothelial dysfunction, and the resulting hypoxemia [22].
If we moved to the relation between severity of liver cirrhosis and HN, previous studies showed contradictory results. While our finding of a positive interrelation accords with some studies [14, 23, 24], others showed no significant correlation [4,5,6].
It is interesting to find in this study a statistically highly significant inverse moderate correlation between the albumin level and the findings of most nerve conduction studies and FDI duration. The lower the albumin level, the more abnormal electrodiagnostic values. Few previous studies showed a similar relation between low serum albumin and autonomic neuropathy in cirrhotic patients [3, 25]. Moreover, other studies showed that low serum albumin level had adverse effects on nerve conduction outcomes and was significantly associated with the severity of polyneuropathy in patients with type 2 DM [26, 27].
The exact pathophysiology underlying HN is not well known; however, different mechanisms were suggested in the literature including the liver failure itself, primary causes of liver cirrhosis, or its common comorbidities.
The significant correlation explored in this study between severity of liver disease, low serum albumin level, and HN increases the likelihood that the metabolic dysfunction caused by the chronic liver disease itself rather than its etiology is the primary pathogenetic factor of HN.
Hepatocellular damage leads to disordered metabolism, reduction of cellular oxidative processes, and accumulation of toxic substances. This disruption results in functional and toxic inhibition of nerve axonal membrane and a possible damage to Schwann cells [16, 28, 29]. Moreover, alteration of nitrogen metabolism and disordered insulin metabolism were suggested to have a role in the pathogenesis of HN [28, 29].
Another early theory of portosystemic shunting was introduced by Chopra et al. [28]; however, this theory could be criticized by Kardel at al. who found no significant differences in cirrhotic patients who undergone portocaval shunts [30].
HCV infection is another important independent risk factor for developing HN especially when associated with cryoglobulinemia [31, 32]. All patients of our study were secondary to HCV infection, and so the confounding effect of HCV generating HN could not be explored or compared to other etiologies of CLD. Other studies had addressed this issue. For example, Chaudry et al. found that the prevalence of HN was not significantly different between cirrhotic patients caused by HCV infection and other etiologies [24]. On the other hand, Cocito et al. found that HN was more prevalent in HCV-related liver cirrhosis compared to other disease groups; however, they stated in their final conclusion that the primary cause of nerve damage could be the toxic and metabolic changes resulting from hepatic failure and that the association between neuropathy and HCV infection could be random [23].
HCV-related HN was mostly explained by virus-triggered immune-mediated mechanisms resulting in ischemic nerve damage through vascular and perivascular inflammation and axonal degeneration [33, 34]. Other studies hypothesized that the nerve dysfunction results from reduced vascular perfusion due to an imbalance between potent vasoconstrictors and vasodilators leading to nerve axonal hyperexcitability [35,36,37].
The significant relation of low serum albumin to HN highlights its role being the most abundant circulating protein in human plasma and has significant effect on peripheral nerve function through its important anti-inflammatory and antioxidative functions. Given that inflammation and oxidative stress are important pathophysiologic processes involved in microvascular and peripheral nerve dysfunction, this may explain the significant effect of hypoalbuminemia on development of HN [38,39,40,41].
Our study had several strengths. First, it included a homogenous group of compensated cirrhotic patients secondary to HCV infection. Second, we excluded many confounding factors regarding etiologies and comorbidities of CLD aiming to strengthen the cause-and-effect relationship between CLD and PN. Third, we compared the patient group to healthy age-matched control group to minimize the confounding effect of old age. Lastly, we conducted extensive electrophysiological assessment which included for the first time QMUP analysis in a such group of patients. On the contrary, our study lacked assessment of small fiber neuropathy using autonomic functions and quantitative sensory tests.
