Liver involvement in patients with Gaucher disease types I and III

Background & aims Gaucher disease (GD) is a multisystemic disease. Liver involvement in GD is not well characterised and ranges from hepatomegaly to cirrhosis and hepatocellular carcinoma. We aim to describe, and assess the effect of treatment, on the hepatic phenotype of a cohort of patients with GD types I and II. Methods Retrospective study based on the review of the medical files of the Gaucher Reference Centre of the Hospital de Clínicas de Porto Alegre, Brazil. Data from all GD types I and III patients seen at the centre since 2003 were analysed. Variables were compared as pre- (“baseline”) and post-treatment (“follow-up”). Results Forty-two patients (types I: 39, III: 3; female: 22; median age: 35 y; enzyme replacement therapy: 37; substrate reduction therapy: 2; non-treated: 3; median time on treatment-MTT: 124 months) were included. Liver enzyme abnormalities, hepatomegaly, and steatosis at baseline were seen in 19/28 (68%), 28/42 (67%), and 3/38 patients (8%), respectively; at follow-up, 21/38 (55%), 15/38 (39%) and 15/38 (39%). MRI iron quantification showed overload in 7/8 patients (treated: 7; MTT: 55 months), being severe in 2/7 (treated: 2/2; MTT: 44.5 months). Eight patients had liver biopsy (treated: 6; MTT: 58 months), with fibrosis in 3 (treated: 1; time on treatment: 108 months) and steatohepatitis in 2 (treated: 2; time on treatment: 69 and 185 months). One patient developed hepatocellular carcinoma. Conclusions GD is a heterogeneous disease that causes different patterns of liver damage even during treatment. Although treatment improves the hepatocellular damage, it is associated with an increased rate of steatosis. This study highlights the importance of a follow-up of liver integrity in these patients.


Methods
This is a retrospective study, based on the review of the medical records of the GD types I and III patients followed at the Gaucher Reference Centre of the Hospital de Clínicas de Porto Alegre, Brazil (GRC-HCPA) from 2003 to 2018. HCPA is a public, university hospital located in Southern Brazil. Inclusion criteria were: a) having biochemical or genetic diagnosis of GD; b) not having any other primary liver disease, as determined by clinical and laboratory features and serological screening for hepatitis B and C.
At the GRC-HCPA, patients have regular appointments every 3-4 months and most exams are made in an annual basis unless an acute event prompts a more frequent evaluation. The following exams were performed at baseline for most patients: complete blood count, chitotriosidase activity, aspartate-transaminase (AST), alanine-transaminase (ALT), and abdominal ultrasonography (US). The following exams are performed yearly: AST, ALT, γ-glutamyltransferase (γGT), direct bilirubin (DB), indirect bilirubin (IB), prothrombin time, alkaline phosphatase, total and fractional cholesterol, triglycerides, serum creatinine, blood urea, calcium, phosphorus, US, serum protein electrophoresis, serum immunoglobulins, transferrin saturation/ironbinding capacity, and serum iron. The following exams are performed every three months: complete blood count, serum ferritin, and chitotriosidase activity. All patients are tested for serological markers of viral hepatitis at initiation of treatment and again according to clinical indication. Alpha-foetoprotein (AFP) is not ordered for patients without cirrhosis due to its dubious efficacy as a screening test for hepatocellular carcinoma [22]. The presence of hepatomegaly was ascertained by US or by physical exam (when US was not available). The presence of steatosis was assessed by US. Elastography for fibrosis assessment is not routinely performed. Other exams are performed according to clinical indication [23]. All patients had genotyping of GBA and HFE by nextgeneration sequencing.
Immunological and iron metabolism findings of our cohort have already been described by Vairo et al. [24] and Koppe et al. [25], respectively.
Statistical analyses were performed using the SPSS software (IBM Inc., v.18); for comparison of frequencies of categorical variables, the χ 2 test was used. Patients were compared regarding the findings before the onset of treatment ("baseline" data points) and during treatment until last follow-up ("follow-up" data points). Findings were considered abnormal at baseline or at follow up if altered in at least two measurements for each datapoint, or one measurement when it was the only one available.

Ethics statement
This study was approved by the Institutional Review Board of HCPA (CEP/HCPA), Porto Alegre, RS, Brazil (projects #13-0537 and #15-0083). All studies were conducted according to the Declaration of Helsinki. Written informed consent was obtained from all subjects or, when <18 years-old, from their parents.

Subjects
Forty-two patients were included (n = 39, type I; n = 3, type III; female = 22; median time on treatment: 124 months). One patient with GD type I (pt 26D) was excluded from the follow-up data analysis due to diagnosis of active chronic hepatitis B. One patient with GD type I (pt 26A) had serological evidence of spontaneously cured hepatitis B. No other patients had signs of other liver diseases, such as drug-related liver injury, autoimmune hepatitis, or viral hepatitis.
No patient had a history of blood transfusions in the past. A total of 36 patients had measurements of serum transferrin saturation after treatment; of these, 6 had decreased values and 5 had increased values (Supplementary Table 1). Four patients had used ferrous sulphate supplements in the past, one of them only during pregnancy (Supplementary Table). No patient was homozygous or compound heterozygous for pathogenic variants in the HFE gene, ruling out the concomitant diagnosis of HFE-associated haemochromatosis (MIM: #235200).

Laboratory findings
Laboratory findings of all patients are shown in Table 1 Out of the 28 patients with liver enzymes (AST, ALT, or γGT) data at baseline, 19/28 (68%) had abnormal liver enzymes in at least two measurements. At follow-up, 21/38 (68%) had abnormalities in at least one liver enzyme in at least two measurements. History was positive for excessive alcohol intake in two patients (19B and 26B).
Serum transferrin saturation, immunoglobulins, and serum protein electrophoresis results during treatment can be found in the supplementary table. Immunoglobulin measurements and serum protein electrophoresis results were available for 36 patients during treatment; of these, 26 had an abnormal serum immunoglobulin measurement at least twice and 20 had increased γ-globulins in serum electrophoresis at least twice.

Table 1
Liver enzymes in patients with Gaucher disease.
Steatosis was present in 3/39 (8%) of patients at baseline and in 15/ 38 (39%) at follow-up. In 6 patients, there was regression of steatosis within 2 years of US detection. Of these, none had any significant change in body-mass index (BMI) but two had changes in the ERT regimen (for patient 15, there was an increase in the imiglucerase dosage from 45 IU/Kg to 60 IU/Kg; for patient 19C, there was a switch from taliglucerase alfa to imiglucerase). Twelve out of the 16 patients (75%) with steatosis were overweight or obese, with 4 patients (two whose steatosis regressed, one that maintains the finding, and one that denied treatment and further follow-up) having a normal BMI. A significant difference was found between the frequency of overweight/obesity in patients with and without persistent steatosis (77.8% vs 40%, p = .047, Pearson's χ 2 ). Blood lipid levels were available for 7 of the 9 patients (78%) with non-regressing steatosis during treatment. All 7 patients had dyslipidaemia (four with high triglycerides, three with high total cholesterol and LDL, and five with low HDL). Levels were available for 31 patients without non-regressing steatosis -of these, 28 (90.3%) had dyslipidaemia (10 with high triglycerides, 5 with high total cholesterol and LDL, and 25 with low HDL). No significant difference was found between patients with and without non-regressing steatosis and the presence of dyslipidaemia (p = .814, Pearson's χ 2 ).
Twelve patients in the cohort had cholelithiasis, and 7 of them underwent cholecystectomy (pts. 13, 18, 19A, 20, 23, 25A, and 25B). Patient 23 had cholecystectomy before initiation of treatment for GD. Eight out of the patients with cholelithiasis were overweight or obese, but no significant difference in the prevalence of overweight/obesity was found between the patients with and without cholelithiasis (66.7% vs 40.7%, p = .135, Pearson's χ 2 ).
Other US findings observed in the cohort were: cysts, haemangioma, solid nodule compatible with an adenoma or a haemangioma, portal hypertension that resolved with initiation of ERT, and cirrhosis with HCC. The two cysts of unknown diagnosis were present in a pair of brothers with GD type I who also had steatosis (pts 29A and 29B). The older brother passed away at the age of 65 due to multiple myeloma. The cyst in the younger brother, now aged 65, is 5 mm in diameter and is stable since it was diagnosed 2 years ago. The patient with cirrhosis

Magnetic resonance iron quantification
Liver iron quantification by magnetic resonance had been performed in 7 patients with GD type I on treatment with ERT (Table 4). Iron overload was observed in 6/7 (85%) patients, ranging from 50 to 280 μmol/g (reference value (RV): <36). All the patients with iron overload had high ferritin values, ranging from 244 to 3011 ng/mL. Two patients had a high level of iron overload (>79 μmol/g [27]) -one was a 55-year-old male patient whose MRI --------y = years-old; BMI = body mass index; I = imiglucerase; T = taliglucerase alfa; V= = velaglucerase alfa; E = eliglustat; A = alglucerase; M = miglustat. a Metabolic syndrome is defined as the presence of at least three of the following: obesity, high triglycerides level, increased blood pressure, and elevated fasting blood glucose (reduced HDL level was not considered as a criterium because it is a feature of GD). b Steatosis regressed within two years of US detection. c Steatosis at liver biopsy only.

Liver biopsy
Six patients with GD type I had a liver biopsy done ( Fig. 1; Table 5) when on-treatment. One patient was found to have Gaucher cells in the liver parenchyma; One patient had atypical Gaucher cells in a cirrhotic parenchyma with severe iron overload in hepatocytes and Kupffer cells, and, in a subsequent biopsy, a moderately differentiated HCC [26]. Two patients who have had mild to moderate steatosis on ultrasound had a biopsy confirming macrovesicular steatosis -one also with evidence of cholestasis and a few foci of inflammation, and the other with mild haemosiderosis. Two patients had steatohepatitis with mild activity: a 27-year-old female with a BMI of 28.8 Kg/m 2 who did not show any sign of steatosis in the ultrasound, had normal serum blood glucose and lipid profile except for a low HDL (which is expected in GD) and that was on SRT with eliglustat at the time of the biopsy; and a 58 year-old man had moderate-to-severe haemosiderosis of hepatocytes and Kupffer cells, elevated triglycerides and total and LDL cholesterol, and low HDL, albeit a normal blood glucose, and signs of steatosis in the liver ultrasound, and that was on ERT at the time of biopsy.
Two patients with GD type I underwent liver biopsy before treatment initiation. A 34-year-old female's biopsy showed bridging (stage 3) fibrosis and scattered Gaucher cells; in the other, a 20-year-old woman, peri-sinusoidal fibrosis was noted together with high serum AST, ALT, and γGT, and a normal liver ultrasound.

Discussion
For the past few decades, the liver involvement in GD has become subject of great importance in the patients' management. It is now recognised that hepatomegaly is only one of the manifestations of hepatic compromise in GD, and more attention is needed to all the possible comorbidities that may arise from it. In our cohort, we observed that a significant number of patients have mildly increased markers of hepatic and biliary damage before the treatment initiation and throughout the clinical follow-up, indicating that a low-grade process of liver damage is not fully corrected by the treatment. This finding resembles the study by James et al. from when effective treatment for GD was not available [28], in which most patients with GD had mild-to-moderate transaminase elevations. In more recent cohorts, these alterations have also been found in a lesser proportion of patients [20,21]. However, the impact of these alterations is still unclear. Nascimbeni et al. have shown that levels of liver enzymes are not correlated with liver fibrosis [20]. The contribution of chronic liver damage to the development of other complications such as iron deposition, since chronic hepatitis and liver disease are strongly associated with hemosiderosis [29], has not been fully explored to date. The high frequency of patients with elevations in γGT may also be related to the known biliary alterations caused by GD [30] such as changes in bile composition, increased incidence of cholelithiasis, or with the chronic inflammatory process that happens in the disease [3,4,24] causing biliary damage.
A significant proportion of patients had bilirubin elevations, both before and during treatment. Most elevated bilirubin values corresponded to direct bilirubin, which points toward a biliary cause rather than overproduction (e.g., haemolysis). It is difficult to establish a clinical significance of this finding, It is known that GlcCer and glucosylsphingosine (GlcSph) [31,32] interact with a series of transporters of the ABC (ATP-binding cassette) family, including ABCB1 [33]; It is also known that the bile of patients with GD is different than in the general showing thick bridging fibrosis characteristic of cirrhosis, as well as substitution of the local hepatic parenchyma by moderately differentiated hepatocellular carcinoma cells.

Table 5
Findings in the liver biopsy.
population, being composed of lower total lipid concentration and, in some patients, high relative concentration of sphingolipids [30]; and finally, that ABC transporters such as ABCB1 are capable of transporting GlcCer and GlcSph [34] across cell membranes, and are modulated by these complex lipids [33]. ABCB1 is localized at the canalicular membrane contributing to the bile formation and xenobiotic excretion [35] it is possible that, due to ABC-mediated efflux, the higher levels of GlcCer present in bile [36] lead to canalicular disturbances that may cause an impaired flow of bilirubin, leading to the slightly high levels of DB observed.
Iron homeostasis is being increasingly recognised as a key factor of GD's pathogenesis [25]. In a recent article by Lefèbvre et al. [37], it was reported that a local overstimulation of hepcidin related to the lower enzymatic activity of GCase causes iron to be sequestered within macrophages and other cell types, leading to a lower level of free iron, transferrin-bound iron and a higher production of ferritin by the liver. In our study, we observed that several patients with GD have high hepatic iron levels as measured by magnetic resonance, two of the tested patients with levels consistent with severe iron overload -whilst in one patient it may be caused by other risk factors such as alcoholism, steatohepatitis, and a pathogenic HFE variant, in the other patient the only obvious risk factor is obesity, and the low transferrin value with exceedingly high ferritin confirm the predictions by Lefèbvre et al.
Other studies have observed increase liver iron concentration in GD patients [38], with a positive correlation with serum ferritin concentration. On liver biopsy, positive iron staining has been described extensively [28,39] both in Kupffer cells and in hepatocytes, similar to what was observed in our cohort. Data on pre-(median = 19%, n = 8 patients) and post-treatment (median = 28%, n = 13 patients) values for serum transferrin saturation in this cohort have been described by Koppe et al [25], with no significant difference (p = .138).
The main ultrasound finding in our cohort was steatosis, with predominance in overweight/obese patients. Our findings differ from the Israeli cohort, which has a much lower prevalence of fatty liver and a higher prevalence of focal lesions [40]. In the Israeli study, 500 patients were evaluated by US, of which 39 had ultrasonographic evidence of hepatic disease -of these, two-thirds were on ERT and one-fourth was splenectomised. ERT is a potent inducer of weight gain due to slowing the increased basal metabolic rate of patients before treatment [19,41]; thus, it may be difficult to establish whether the high prevalence of steatosis is a manifestation of GD itself, a complication of its treatment, or a comorbidity. A significant proportion of our patients had dyslipidaemia, which indicates that metabolic syndrome may play a role as a confounder in the development of steatosis in these patients [42]. Remarkably, a young patient being treated with eliglustat that had a hepatic biopsy done during cholecystectomy was diagnosed with steatohepatitis, regardless of having no signs of steatosis. This case raises two questions: whether ultrasound can be relied upon as a mean of screening for liver disease in GD patients; and whether steatohepatitis may be a manifestation of GD, since the only known risk factor that the patient had for steatohepatitis (a BMI of 28.8 Kg/m 2 ) is hardly considered enough for a sole causal factor; and, as the blood glucose and lipid levels of this patient were normal except for a low HDL, which is a marker of GD, dyslipidaemia and metabolic syndrome are not strongly suspected. Another possible cause for the steatohepatitis in this patient could be what is becoming known as "lean fatty liver disease" -that is, non-alcoholic steatosis (NAFLD) or steatohepatitis (NASH) in patients with few or no risk factors for such [43]. Although in the classical definition of "lean NASH" the patient's BMI is normally <25 Kg/m 2 [43], despite some authors advocating for the use of a BMI of <30 Kg/m 2 in Western populations [44], it is expected that patients with "non-lean NASH" are male, of older age, and have hypertension, insulin resistance, or hypercholesterolaemia -none of which is present in this patient [43]. It is speculated that lean NASH arises from "metabolic obesity" in non-obese people, which is reflected by the higher distribution of fat to the visceral intraabdominal organs [44,45], along with classical risk factors such as insulin resistance and hypercholesterolaemia [44] -none of which were present in this patient -and genetic predisposition due to polymorphisms in genes associated with lipid metabolism [44,46]. Liver fibrosis is shown to be increased in a significant proportion of patients [20], especially in those who were splenectomised [38], and it is a major risk factor for HCC [39]. Liver fibrosis is correlated with increased severity of GD [20], although its correlation with biomarkers of disease activity is still controversial [20,38]. In the pre-ERT era, when no specific treatment for GD was available, liver fibrosis was a common finding [28], and often culminated in a massive central area of hypocellular fibrotic tissue [47,48] that led to portal hypertension and other clinical manifestations of cirrhosis [28].
Cholelithiasis is a frequent comorbid process of GD with about 30-45% [49,50] lifetime incidence in these patients. Although the causes for this increased incidence are not completely elucidated, some authors speculate that the excretion of GlcCer in the bile may increase its lithogenicity, predisposing to the formation of gallstones [30,36,49]. In our cohort, we have observed a similarly increased prevalence of cholelithiasis in GD patients compared to the general population, with 12 patients affected in a total of 41.

Conclusion
In this study, we presented a comprehensive summary of the hepatic manifestations in a well-characterised cohort of patients with GD, showing that several patients have lingering alterations that may indicate a smouldering process of liver damage which is not completely avoided by standard therapy. It is also noticeable that many patients have liver steatosis or steatohepatitis, with a noticeable increase in prevalence during treatment with ERT, but it is still unclear whether it reflects a consequence from the treatment, a feature of the disease, or a coincidental finding.