The Impact of Phlebotomy in Nonalcoholic Fatty Liver Disease: A Prospective, Randomized, Controlled Trial Leon A. Adams,1,2 Darrell H. Crawford,3,4 Katherine Stuart,4 Michael J. House,5 Timothy G. St. Pierre,5 Malcolm Webb,6 Helena L.I. Ching,2 Jenny Kava,7 Michael Bynevelt,8 Gerry C. MacQuillan,1,2 George Garas,1,2 Oyekoya T. Ayonrinde,7,9,10 Trevor A. Mori,1 Kevin D. Croft,1 Xianwa Niu,1 Gary P. Jeffrey,1,2 and John K. Olynyk7,9,10 Iron is implicated in the pathogenesis of liver injury and insulin resistance (IR) and thus phlebotomy has been proposed as a treatment for nonalcoholic fatty liver disease (NAFLD). We performed a prospective 6-month randomized, controlled trial examining the impact of phlebotomy on the background of lifestyle advice in patients with NAFLD. Primary endpoints were hepatic steatosis (HS; quantified by magnetic resonance imaging) and liver injury (determined by alanine aminotransaminase [ALT] and cytokeratin-18 [CK-18]). Secondary endpoints included insulin resistance measured by the insulin sensitivity index (ISI) and homeostasis model of assessment (HOMA), and systemic lipid peroxidation determined by plasma F2-isoprostane levels. A total of 74 subjects were randomized (33 phlebotomy and 41 control). The phlebotomy group underwent a median (range) of 7 (1-19) venesection sessions and had a significantly greater reduction in ferritin levels over 6 months, compared to controls (2148 6 114 vs. 238 6 89 ng/mL; P < 0.001). At 6 months, there was no difference between phlebotomy and control groups in HS (17.7% vs. 15.5%; P 5 0.4), serum ALT (36 vs. 46 IU/L; P 5 0.4), or CK-18 levels (175 vs. 196 U/L; P 5 0.9). Similarly, there was no difference in end-of-study ISI (2.5 vs. 2.7; P 5 0.9), HOMA (3.2 vs. 3.2; P 5 0.6), or F2isoprostane levels (1,332 vs. 1,190 pmmol/L; P 5 0.6) between phlebotomy and control groups. No differences in any endpoint were noted in patients with hyperferritinemia at baseline. Among patients undergoing phlebotomy, there was no correlation between number of phlebotomy sessions and change in HS, liver injury, or IR from baseline to end of study. Conclusion: Reduction in ferritin by phlebotomy does not improve liver enzymes, hepatic fat, or IR in subjects with NAFLD. (HEPATOLOGY 2015;61:1555-1564)
N
onalcoholic fatty liver disease (NAFLD) is present in 10%-30% of the world’s population. This burden of disease is translating to an increasing rate of end-stage liver disease requiring liver transplantation or resulting in mor-
bidity and death.1 Despite this, a relative minority of subjects with NAFLD develop liver injury, which, histologically, is characterized as nonalcoholic steatohepatitis (NASH). Understanding the factors leading to NASH and liver injury is key to developing
Abbreviations: ALT, alanine aminotransaminase; BMI, body mass index; CK-18, cytokeratin-18; ESS, Epworth Sleepiness Scale; HbA1c, glycated hemoglobin; HDL, high-density lipoprotein; HIC, hepatic iron concentration; HOMA-IR, homeostasis model of assessment of insulin resistance; HS, hepatic steatosis; IQR, interquartile range; IR, insulin resistance; ISI, insulin sensitivity index; LPO, lipid peroxidation; MRI, magnetic resonance imaging; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; NEFA, nonesterified free fatty acid; OGTT, oral glucose tolerance test; SD, standard deviation; SF-36, the 36-item Short Form Health Survey; TGs, triglycerides. From the 1School of Medicine & Pharmacology, University of Western Australia, Crawley, Australia; 2Department of Gastroenterology & Hepatology, Sir Charles Gairdner Hospital, Perth, Australia; 3School of Medicine, University of Queensland, Brisbane, Australia; 4Department of Gastroenterology, Greenslopes Private Hospital, Brisbane, Australia; 5School of Physics, University of Western Australia, Crawley, Australia; 6Department of Hematology, Fremantle Hospital, Fremantle, Australia; 7Department of Gastroenterology, Fremantle Hospital, Fremantle, Australia; 8Department of Radiology, Sir Charles Gairdner Hospital, Perth, Australia; 9 Faculty of Health Sciences, Curtin University, Perth, Australia; 10Institute for Immunology and Infectious Diseases, Murdoch University, Perth, Australia. Received September 7, 2014; accepted December 12, 2014. Additional Supporting Information may be found at onlinelibrary.wiley.com/doi/10.1002/hep.27662/suppinfo This study was supported by a Clinical Research Grant from the Raine Medical Research Foundation and the Western Australian Government Department of Health. Resonance Health Ltd. (Claremont, Australia) provided support for the magnetic resonance scans. 1555
1556
ADAMS ET AL.
treatment strategies to reduce NAFLD-related morbidity. Iron may potentially act as a source of oxidative stress and injury within the liver.2-4 In animal models of fatty liver, iron loading is associated with the development of hepatic necroinflammation and fibrosis.5,6 Among patients with NAFLD, increased serum ferritin levels have been associated with the presence of NASH and advanced fibrosis in some studies, but not others.7,8 Similarly, the relationship between hepatic iron and liver fibrosis is conflicting in the literature with positive and negative studies.2,4,9,10 Iron has also been implicated as a cofactor in the pathogenesis of insulin resistance (IR), which is universal among individuals with NAFLD and is implicated in the progression of liver injury.11 Cell-based and animal studies have demonstrated that iron induces IR in myocytes, adipose tissue, the pancreas, and liver.12-17 Among subjects with hereditary hemochromatosis and iron overload, insulin secretion and sensitivity is altered and changes following venesection.18 Notably, however, the impact of iron does not appear to be limited to subjects with pathological iron overload. For example, epidemiological data has demonstrated dietary heme iron intake to be a risk factor for the development of diabetes in the absence of iron overload.19 Pilot clinical studies have suggested that removal of iron by venesection in subjects with normal serum ferritin and transferrin values leads to improved IR and serum alanine aminotransaminase (ALT) levels.20-22 In particular, a case-control study of NAFLD subjects demonstrated reduction in IR with venesection among subjects with both normal and raised baseline serum ferritin levels.23 The progression of these observations linking iron with liver injury and IR has led to the postulation that iron represents a novel therapeutic target for subjects with NAFLD.3 Thus, we hypothesized that venesection in patients with NAFLD would reduce liver steatosis and liver injury, as well as improve IR and systemic levels of oxidative stress. Owing to a lack of high-level evidence to support this postulation, we aimed to test our hypothesis by performing a 6-month prospective, parallel, randomized, controlled trial comparing venesection and lifestyle
HEPATOLOGY, May 2015
advice against lifestyle advice alone in 74 adults with NAFLD.
Patients and Methods Subjects. Subjects were recruited between November 2010 and December 2012 from the hepatology clinics at Sir Charles Gairdner Hospital, Fremantle Hospital (both located in Perth, Western Australia, Australia) and Greenslopes Private Hospital (Brisbane, Queensland, Australia). Adults with hepatic steatosis (HS) confirmed on imaging (ultrasound or computed tomography scan) that were teetotallers or consumed 0.3 for both). (B) Median (95% confidence interval) serum ALT was not different between venesection and control groups throughout the trial (P > 0.3 for all time points).
Results Study Population Seventy-four individuals were randomized: 41 to the control arm and 33 to the venesection arm. In the control arm, 9 subjects withdrew before the first follow-up assessment (Fig. 1). In the venesection arm, 4 subjects withdrew before any intervention. One individual completed four venesections and the first follow-up assessment and subsequently withdrew. This subject was included in the analysis with their last observation brought forward. The baseline characteristics of the groups were evenly matched (Table 1). The cohort was predominantly middle aged with a slight preponderance of males. The majority were overweight or obese and approximately one third had a history of either diabetes or hypertension. No subject had clinical evidence of cirrhosis. Eight subjects were taking statins (5 in the control group and 3 in the venesection group), 3 were taking fish oil (2 in the control group and 1 in the venesection group), and 2 were taking fenofibrate (1 in each group). Doses remained unchanged throughout the trial. Waist circumference and BMI remained stable throughout the study, with no significant change over time or difference between groups (P > 0.1 for all comparisons).
ADAMS ET AL.
1559
Venesection The intervention group had a median of seven (range, 1-19) phlebotomy sessions. Although serum ferritin levels fell in both groups during the study, the reduction was significantly greater in the intervention group, compared to controls (2148 6 114 vs. 238 6 89 ng/mL; P < 0.001). Similarly, there was a reduction in transferrin saturation from baseline to end of study, which was greater in the phlebotomy group compared to the control group (25.8 6 9.2 vs. 21.5 6 8.8 %; P 5 0.07). Venesection was well tolerated. One patient felt faint for a short period following her first venesection, but recovered quickly and underwent further venesections without incident. No other side-effects occurred. Efficacy HS and ALT. HS determined by MRI was not different between groups at baseline or at month 6 (P > 0.3 for both comparisons), as illustrated in Fig. 2A. The change in HS from baseline to 6 months was not different between venesection and control groups (21.5 6 4.7 vs. 21.6 6 5.9 %; P 5 0.9). There were no within-group differences between baseline and month 6 HS (P > 0.1 for both groups, paired t test). The proportion of subjects who achieved >25% reduction in serum ALT was not different between venesection and control groups (30.3% vs. 20.0%; Table 2. End-of-Study (Month 6) Characteristics of Control and Venesection Groups Characteristic
Control Group (n 5 32)
Venesection Group (n 5 29)
BMI, kg/m2 30.2 (4.8) 32.0 (4.9) Waist circumference, cm 102 (11) 108 (12) ALT, IU/L 46 (39-74) 36 (25-95) AST, IU/L 39 (26-46) 33 (26-51) Bilirubin, mg/dL 0.8 (0.4) 0.8 (0.3) Alkaline phosphatase, IU/L 84 (23) 81 (76) Albumin, mg/dL 4.4 (0.3) 4.3 (0.3) CK-18, U/L 196 (157-268) 175 (136-374) Hepascore 0.33 (0.24) 0.39 (0.27) TG, mg/dL 133 (106-204) 120 (89-177) HDL-cholesterol, mg/dL 46 (10) 48 (12) Free fatty acids, mmol/L 0.35 (0.21) 0.39 (0.18) F2-isoprostanes, pmol/L 1,190 (1,008-1,871) 1,332 (1,062-1,832) Glucose, mg/dL 99 (86-108) 98 (84-116) Insulin, mU/L 13 (8-17) 13.5 (7-22) HbA1c, % 5.6 (5.47-6.2) 5.7 (5.4-6.5) HOMA 3.2 (1.8-4.3) 3.2 (1.7-5.7) ISI 2.7 (1.8-4.6) 2.5 (1.6-4.9) Hemoglobin, g/dL 148 (12) 145 (11) Platelet, 3109/L 221 (41) 218 (68) Ferritin, ng/mL 159 (166) 111 (78) Transferrin saturation, % 29.5 (9.1) 26.7 (11.2) Hepatic IC, mmol/kg 18 (12-27) 15.5 (14-21) HS, % 15.5 (10.8) 18.1 (10.4)
P Value
0.2 0.04 0.4 0.8 0.9 0.6 0.2 0.9 0.4 0.7 0.8 0.5 0.6 0.8 0.6 0.7 0.6 0.9 0.4 0.9 0.2 0.3 0.6 0.4
Abbreviations: AST, aspartate aminotransferase; IC, inhibitory concentration.
1560
ADAMS ET AL.
Fig. 3. IR. (A) Median (IQR) of the ISI was not different between venesection and control groups throughout the trial (P > 0.7 for all time points). (B) Median (IQR) of HOMA was not different between venesection and control groups throughout the trial (P > 0.5 for all time points).
p 5 0.4). Serum ALT was not different between groups at baseline or 3 or 6 months (P > 0.3 for all time points), as shown in Fig. 2B. After adjustment for baseline ALT or ferritin, there was no difference in end-of-treatment ALT between groups (P > 0.5 for both). Similarly, the change in mean ALT levels from baseline to end of treatment was the same between venesection and control groups (211 6 30 vs. 213 6 35 IU/L; P 5 0.8). In the treatment group, the change in ALT from baseline to end of treatment did not correlate with the change in serum ferritin from baseline to end of treatment (Spearman’s rho: 0.02; P 5 0.9). IR. Glucose homeostasis and IR were assessed using multiple methods, including fasting glucose, insulin, HbA1c, HOMA, and the Matsuda ISI. There was no significant difference between groups at any time point (baseline and month 3 or month 6) in any of these measures (Table 2; Figs. 3A,B). Similarly, there was no within group differences from baseline to 6 months in any of these measures (P > 0.3 for all, data not shown). Serum Lipids, LPO, and Hepatocyte Apoptosis. Serum lipids, including TGs, HDL-cholesterol, and NEFA levels, were not different between groups at
HEPATOLOGY, May 2015
baseline or 3 or 6 months (P > 0.1 for all). Similarly, plasma F2-isoprostane levels, as a marker of systemic LPO, were not different between groups at baseline or end of treatment (Tables 1 and 2). The change in F2isoprostane levels from baseline to end of treatment was not different between the two groups (Fig. 4A). Hepatic apoptosis was not altered by venesection treatment, with serum CK-18 fragment levels the same between groups at baseline and at 6 month (Tables 1 and 2). There was no difference in change in CK-18 fragment levels from baseline to end of treatment between venesection and control groups (217.5 [247, 12.2] vs. 29.0 [-44.5, 14.0]; p 5 0.7), as illustrated in Fig. 4B. Quality of Life. Quality of life was not different between control and venesection groups at any time point, with each domain of the SF-36 scoring similarly at baseline and 3 and 6 months Supporting Table 1 Supporting Figs. 1 and 2. Within the control and venesection groups, there was a significant increase in general health perception from baseline to 6 months (66.8-75.1, P 5 0.003 for controls; 59.4-65.3,
Fig. 4. Change in hepatocyte apoptosis and LPO. (A) Change in serum F2-isoprostane levels between end of treatment and baseline in venesection and control arms. P 5 0.7 (Mann-Whitney’s U test). (B) Change in serum CK-18 levels between end of treatment and baseline in venesection and control arms. P 5 0.7 (Mann-Whitney’s U test). Abbreviation: EOT, end of treatment.
HEPATOLOGY, Vol. 61, No. 5, 2015
ADAMS ET AL.
1561
Table 3. Impact of Venesection Compared to Controls in NAFLD Patients With Hyperferritinemia (n 5 36) Baseline Characteristic
ALT, IU/L AST, IU/L CK-18, U/L F2-isoprostanes, pmol/L Glucose, mg/dL Insulin, mU/L HbA1c, % HOMA ISI HS, %
Month 6
Control Group (n 5 15)
Venesection Group (n 5 21)
P Value
Control Group (n 5 15)
Venesection Group (n 5 21)
P Value
59 (32-90) 44 (31-73) 260 (184-525) 1,240 (993-1,721) 5.7 (5.2-6.3) 13 (10-18) 5.5 (5.3-6.3) 3.3 (2.5-4.4) 2.5 (1.5-3.7) 21.2 (12.9)
69 (36-111) 35 (27-48) 272 (154-552) 1,220 (1,046-1,538) 5.3 (5.0-6.8) 15 (9-20) 5.8 (5.4-6.2) 4.2 (2.0-5.4) 2.1 (1.7-3.9) 18.4 (9.1)
0.8 0.3 0.9 0.7 0.6 0.7 0.5 0.7 0.9 0.5
65 (45-88) 42 (33-48) 197 (188-531) 1,373 (1,024-1,909) 5.8 (5.4-6.0) 15 (8-18) 5.6 (5.2-6.2) 3.6 (2.5-4.7) 2.4 (1.6-4.6) 17.2 (11.2)
42 (28-118) 33 (26-58) 162 (136-442) 1,268 (912-1,411) 5.6 (5.0-6.4) 14.5 (9.2-23) 5.6 (5.4-6.4) 3.9 (2.3-5.6) 2.0 (1.5-3.6) 18.5 (9.3)
0.3 0.3 0.3 0.4 0.7 0.8 0.8 0.9 0.8 0.7
Abbreviation: AST, aspartate aminotransferase.
P 5 0.06 for the venesection group). There were no other within-group differences in the other domains of the SF-36 from baseline to 6 months. Somnolence scores, as determined by the ESS, were not different between groups at any time point Supporting Table 1. Similarly, there were no within-group differences from baseline to 6 months (P > 0.3 for both groups). Analyses by Ferritin Status and Number of Venesections. Subanalysis was performed in 36 patients who had hyperferritinemia at baseline. Measures of liver injury, HS, IR and sensitivity, and LPO were not different between the control and venesection groups at baseline or end of treatment (Table 3). Similarly, there was no difference in the change in these parameters from baseline to end of treatment between the control and venesection groups (P > 0.3 for all comparisons; Fig. 5A-F). Among subjects who underwent venesection (n 5 29), the reduction in serum ferritin from baseline to end of treatment was significantly correlated with the number of venesection sessions (Spearman’s rho: 20.63; P < 0.001), as shown in Table 4. However, in contrast, there was no significant correlation between measures of liver injury, HS, IR, or LPO with the number of venesection sessions.
Discussion In this prospective, randomized, controlled clinical trial conducted in patients with NAFLD, we have demonstrated that iron removal by way of venesection was not associated with any significant improvement in (1) HS, (2) liver injury, (3) IR or sensitivity, (4) measures of LPO, or (5) quality of life. The lack of benefit was observed across the range of serum ferritin levels, and no dose-response relationship was observed between the number of venesection sessions and any of the study endpoints.
A number of studies have demonstrated an association between intrahepatic iron and severity of liver injury in patients with NAFLD; however, these studies have had conflicting findings of the importance of hepatocellular versus reticuloendothelial iron.2,4,9 Furthermore, cross-sectional studies have not demonstrated whether iron deposition is causal or an epiphenomenon of liver injury in NAFLD. Our findings suggest that the levels of hepatic iron found in NAFLD patients do not have a significant causal role in the genesis of liver injury. A number of previous clinical trials in individuals with NAFLD have demonstrated an improvement in serum ALT and/or HOMA levels with phlebotomy.20,23,35-37 However, these have been single-arm uncontrolled or nonrandomized studies, which are prone to bias and “regression to the mean” phenomena, which may account for supposedly positive findings. It is pertinent to note that a large, randomized, controlled trial was required to demonstrate the lack of efficacy of ursodeoxycholic acid in the treatment of NAFLD after a number of small, uncontrolled trials had previously suggested a therapeutic benefit.38 The only previous randomized, controlled trial of venesection in NAFLD subjects demonstrated an improvement in HS with venesection, but not lobular inflammation or hepatocyte ballooning.39 In this small study of 36 patients, only 50% of subjects underwent an end-of-study liver biopsy, limiting the ability to draw definitive conclusions from this study. A limitation of our trial was the lack of histological assessment and thus ability to determine the impact of phlebotomy on liver fibrosis and direct measures of liver injury. A recent consensus meeting has recommended magnetic resonance–determined HS and serum ALT as suitable primary endpoints for shortterm phase II trials.40 We utilized MRI to determine HS, which is in close agreement with magnetic
1562
ADAMS ET AL.
HEPATOLOGY, May 2015
Fig. 5. Change in measures of liver injury (A and F), HS (B), IR (C and D), and LPO (E) from baseline to the end of treatment in control and venesection groups among patients with hyperferritinemia at baseline (n 5 36). There was no significant difference between groups (P > 0.3 for all comparisons). Abbreviation: EOT, end of treatment.
resonance spectroscopy–determined HS and histologically determined HS.41-43 However, we cannot exclude that minor changes in HS were undetected. A strength of our study is the multiple measures, including serum markers of hepatocyte apoptosis (CK18 levels), LPO (F2-isoprostane levels), multiple measures of insulin metabolism, and patient quality of life. We were unable to detect any impact of phlebotomy, despite having a cohort size similar to other clinical
studies where clinical benefits have been demonstrated.33,44,45 However, there is the possibility that our study was underpowered and improvements in the primary or secondary outcomes may not have been detected. Furthermore, we cannot exclude that venesection may be of benefit in certain subgroups of patients with NAFLD, such as those with NASH or increased hepatic iron concentration, or that a longer duration of treatment may be therapeutic.
HEPATOLOGY, Vol. 61, No. 5, 2015
ADAMS ET AL.
Table 4. Correlation Between Number of Venesection Sessions and Change in Outcomes From Baseline to End of Treatment (n 5 29) All Venesection Patients (n 5 29) Characteristic (Month 6 to Month 0 Value)
Spearman’s rho
P Value
Ferritin ALT, IU/L AST, IU/L CK-18 F2-isoprostanes, pmol/L Glucose, mg/dL Insulin, mU/L HbA1c, % HOMA ISI HS, %
20.69 0.02 0.03 20.01 0.23 0.13 20.002 0.005 0.08 20.04 0.32
N
onalcoholic fatty liver disease (NAFLD) is present in 10%-30% of the world’s population. This burden of disease is translating to an increasing rate of end-stage liver disease requiring liver transplantation or resulting in mor-
bidity and death.1 Despite this, a relative minority of subjects with NAFLD develop liver injury, which, histologically, is characterized as nonalcoholic steatohepatitis (NASH). Understanding the factors leading to NASH and liver injury is key to developing
Abbreviations: ALT, alanine aminotransaminase; BMI, body mass index; CK-18, cytokeratin-18; ESS, Epworth Sleepiness Scale; HbA1c, glycated hemoglobin; HDL, high-density lipoprotein; HIC, hepatic iron concentration; HOMA-IR, homeostasis model of assessment of insulin resistance; HS, hepatic steatosis; IQR, interquartile range; IR, insulin resistance; ISI, insulin sensitivity index; LPO, lipid peroxidation; MRI, magnetic resonance imaging; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; NEFA, nonesterified free fatty acid; OGTT, oral glucose tolerance test; SD, standard deviation; SF-36, the 36-item Short Form Health Survey; TGs, triglycerides. From the 1School of Medicine & Pharmacology, University of Western Australia, Crawley, Australia; 2Department of Gastroenterology & Hepatology, Sir Charles Gairdner Hospital, Perth, Australia; 3School of Medicine, University of Queensland, Brisbane, Australia; 4Department of Gastroenterology, Greenslopes Private Hospital, Brisbane, Australia; 5School of Physics, University of Western Australia, Crawley, Australia; 6Department of Hematology, Fremantle Hospital, Fremantle, Australia; 7Department of Gastroenterology, Fremantle Hospital, Fremantle, Australia; 8Department of Radiology, Sir Charles Gairdner Hospital, Perth, Australia; 9 Faculty of Health Sciences, Curtin University, Perth, Australia; 10Institute for Immunology and Infectious Diseases, Murdoch University, Perth, Australia. Received September 7, 2014; accepted December 12, 2014. Additional Supporting Information may be found at onlinelibrary.wiley.com/doi/10.1002/hep.27662/suppinfo This study was supported by a Clinical Research Grant from the Raine Medical Research Foundation and the Western Australian Government Department of Health. Resonance Health Ltd. (Claremont, Australia) provided support for the magnetic resonance scans. 1555
1556
ADAMS ET AL.
treatment strategies to reduce NAFLD-related morbidity. Iron may potentially act as a source of oxidative stress and injury within the liver.2-4 In animal models of fatty liver, iron loading is associated with the development of hepatic necroinflammation and fibrosis.5,6 Among patients with NAFLD, increased serum ferritin levels have been associated with the presence of NASH and advanced fibrosis in some studies, but not others.7,8 Similarly, the relationship between hepatic iron and liver fibrosis is conflicting in the literature with positive and negative studies.2,4,9,10 Iron has also been implicated as a cofactor in the pathogenesis of insulin resistance (IR), which is universal among individuals with NAFLD and is implicated in the progression of liver injury.11 Cell-based and animal studies have demonstrated that iron induces IR in myocytes, adipose tissue, the pancreas, and liver.12-17 Among subjects with hereditary hemochromatosis and iron overload, insulin secretion and sensitivity is altered and changes following venesection.18 Notably, however, the impact of iron does not appear to be limited to subjects with pathological iron overload. For example, epidemiological data has demonstrated dietary heme iron intake to be a risk factor for the development of diabetes in the absence of iron overload.19 Pilot clinical studies have suggested that removal of iron by venesection in subjects with normal serum ferritin and transferrin values leads to improved IR and serum alanine aminotransaminase (ALT) levels.20-22 In particular, a case-control study of NAFLD subjects demonstrated reduction in IR with venesection among subjects with both normal and raised baseline serum ferritin levels.23 The progression of these observations linking iron with liver injury and IR has led to the postulation that iron represents a novel therapeutic target for subjects with NAFLD.3 Thus, we hypothesized that venesection in patients with NAFLD would reduce liver steatosis and liver injury, as well as improve IR and systemic levels of oxidative stress. Owing to a lack of high-level evidence to support this postulation, we aimed to test our hypothesis by performing a 6-month prospective, parallel, randomized, controlled trial comparing venesection and lifestyle
HEPATOLOGY, May 2015
advice against lifestyle advice alone in 74 adults with NAFLD.
Patients and Methods Subjects. Subjects were recruited between November 2010 and December 2012 from the hepatology clinics at Sir Charles Gairdner Hospital, Fremantle Hospital (both located in Perth, Western Australia, Australia) and Greenslopes Private Hospital (Brisbane, Queensland, Australia). Adults with hepatic steatosis (HS) confirmed on imaging (ultrasound or computed tomography scan) that were teetotallers or consumed 0.3 for both). (B) Median (95% confidence interval) serum ALT was not different between venesection and control groups throughout the trial (P > 0.3 for all time points).
Results Study Population Seventy-four individuals were randomized: 41 to the control arm and 33 to the venesection arm. In the control arm, 9 subjects withdrew before the first follow-up assessment (Fig. 1). In the venesection arm, 4 subjects withdrew before any intervention. One individual completed four venesections and the first follow-up assessment and subsequently withdrew. This subject was included in the analysis with their last observation brought forward. The baseline characteristics of the groups were evenly matched (Table 1). The cohort was predominantly middle aged with a slight preponderance of males. The majority were overweight or obese and approximately one third had a history of either diabetes or hypertension. No subject had clinical evidence of cirrhosis. Eight subjects were taking statins (5 in the control group and 3 in the venesection group), 3 were taking fish oil (2 in the control group and 1 in the venesection group), and 2 were taking fenofibrate (1 in each group). Doses remained unchanged throughout the trial. Waist circumference and BMI remained stable throughout the study, with no significant change over time or difference between groups (P > 0.1 for all comparisons).
ADAMS ET AL.
1559
Venesection The intervention group had a median of seven (range, 1-19) phlebotomy sessions. Although serum ferritin levels fell in both groups during the study, the reduction was significantly greater in the intervention group, compared to controls (2148 6 114 vs. 238 6 89 ng/mL; P < 0.001). Similarly, there was a reduction in transferrin saturation from baseline to end of study, which was greater in the phlebotomy group compared to the control group (25.8 6 9.2 vs. 21.5 6 8.8 %; P 5 0.07). Venesection was well tolerated. One patient felt faint for a short period following her first venesection, but recovered quickly and underwent further venesections without incident. No other side-effects occurred. Efficacy HS and ALT. HS determined by MRI was not different between groups at baseline or at month 6 (P > 0.3 for both comparisons), as illustrated in Fig. 2A. The change in HS from baseline to 6 months was not different between venesection and control groups (21.5 6 4.7 vs. 21.6 6 5.9 %; P 5 0.9). There were no within-group differences between baseline and month 6 HS (P > 0.1 for both groups, paired t test). The proportion of subjects who achieved >25% reduction in serum ALT was not different between venesection and control groups (30.3% vs. 20.0%; Table 2. End-of-Study (Month 6) Characteristics of Control and Venesection Groups Characteristic
Control Group (n 5 32)
Venesection Group (n 5 29)
BMI, kg/m2 30.2 (4.8) 32.0 (4.9) Waist circumference, cm 102 (11) 108 (12) ALT, IU/L 46 (39-74) 36 (25-95) AST, IU/L 39 (26-46) 33 (26-51) Bilirubin, mg/dL 0.8 (0.4) 0.8 (0.3) Alkaline phosphatase, IU/L 84 (23) 81 (76) Albumin, mg/dL 4.4 (0.3) 4.3 (0.3) CK-18, U/L 196 (157-268) 175 (136-374) Hepascore 0.33 (0.24) 0.39 (0.27) TG, mg/dL 133 (106-204) 120 (89-177) HDL-cholesterol, mg/dL 46 (10) 48 (12) Free fatty acids, mmol/L 0.35 (0.21) 0.39 (0.18) F2-isoprostanes, pmol/L 1,190 (1,008-1,871) 1,332 (1,062-1,832) Glucose, mg/dL 99 (86-108) 98 (84-116) Insulin, mU/L 13 (8-17) 13.5 (7-22) HbA1c, % 5.6 (5.47-6.2) 5.7 (5.4-6.5) HOMA 3.2 (1.8-4.3) 3.2 (1.7-5.7) ISI 2.7 (1.8-4.6) 2.5 (1.6-4.9) Hemoglobin, g/dL 148 (12) 145 (11) Platelet, 3109/L 221 (41) 218 (68) Ferritin, ng/mL 159 (166) 111 (78) Transferrin saturation, % 29.5 (9.1) 26.7 (11.2) Hepatic IC, mmol/kg 18 (12-27) 15.5 (14-21) HS, % 15.5 (10.8) 18.1 (10.4)
P Value
0.2 0.04 0.4 0.8 0.9 0.6 0.2 0.9 0.4 0.7 0.8 0.5 0.6 0.8 0.6 0.7 0.6 0.9 0.4 0.9 0.2 0.3 0.6 0.4
Abbreviations: AST, aspartate aminotransferase; IC, inhibitory concentration.
1560
ADAMS ET AL.
Fig. 3. IR. (A) Median (IQR) of the ISI was not different between venesection and control groups throughout the trial (P > 0.7 for all time points). (B) Median (IQR) of HOMA was not different between venesection and control groups throughout the trial (P > 0.5 for all time points).
p 5 0.4). Serum ALT was not different between groups at baseline or 3 or 6 months (P > 0.3 for all time points), as shown in Fig. 2B. After adjustment for baseline ALT or ferritin, there was no difference in end-of-treatment ALT between groups (P > 0.5 for both). Similarly, the change in mean ALT levels from baseline to end of treatment was the same between venesection and control groups (211 6 30 vs. 213 6 35 IU/L; P 5 0.8). In the treatment group, the change in ALT from baseline to end of treatment did not correlate with the change in serum ferritin from baseline to end of treatment (Spearman’s rho: 0.02; P 5 0.9). IR. Glucose homeostasis and IR were assessed using multiple methods, including fasting glucose, insulin, HbA1c, HOMA, and the Matsuda ISI. There was no significant difference between groups at any time point (baseline and month 3 or month 6) in any of these measures (Table 2; Figs. 3A,B). Similarly, there was no within group differences from baseline to 6 months in any of these measures (P > 0.3 for all, data not shown). Serum Lipids, LPO, and Hepatocyte Apoptosis. Serum lipids, including TGs, HDL-cholesterol, and NEFA levels, were not different between groups at
HEPATOLOGY, May 2015
baseline or 3 or 6 months (P > 0.1 for all). Similarly, plasma F2-isoprostane levels, as a marker of systemic LPO, were not different between groups at baseline or end of treatment (Tables 1 and 2). The change in F2isoprostane levels from baseline to end of treatment was not different between the two groups (Fig. 4A). Hepatic apoptosis was not altered by venesection treatment, with serum CK-18 fragment levels the same between groups at baseline and at 6 month (Tables 1 and 2). There was no difference in change in CK-18 fragment levels from baseline to end of treatment between venesection and control groups (217.5 [247, 12.2] vs. 29.0 [-44.5, 14.0]; p 5 0.7), as illustrated in Fig. 4B. Quality of Life. Quality of life was not different between control and venesection groups at any time point, with each domain of the SF-36 scoring similarly at baseline and 3 and 6 months Supporting Table 1 Supporting Figs. 1 and 2. Within the control and venesection groups, there was a significant increase in general health perception from baseline to 6 months (66.8-75.1, P 5 0.003 for controls; 59.4-65.3,
Fig. 4. Change in hepatocyte apoptosis and LPO. (A) Change in serum F2-isoprostane levels between end of treatment and baseline in venesection and control arms. P 5 0.7 (Mann-Whitney’s U test). (B) Change in serum CK-18 levels between end of treatment and baseline in venesection and control arms. P 5 0.7 (Mann-Whitney’s U test). Abbreviation: EOT, end of treatment.
HEPATOLOGY, Vol. 61, No. 5, 2015
ADAMS ET AL.
1561
Table 3. Impact of Venesection Compared to Controls in NAFLD Patients With Hyperferritinemia (n 5 36) Baseline Characteristic
ALT, IU/L AST, IU/L CK-18, U/L F2-isoprostanes, pmol/L Glucose, mg/dL Insulin, mU/L HbA1c, % HOMA ISI HS, %
Month 6
Control Group (n 5 15)
Venesection Group (n 5 21)
P Value
Control Group (n 5 15)
Venesection Group (n 5 21)
P Value
59 (32-90) 44 (31-73) 260 (184-525) 1,240 (993-1,721) 5.7 (5.2-6.3) 13 (10-18) 5.5 (5.3-6.3) 3.3 (2.5-4.4) 2.5 (1.5-3.7) 21.2 (12.9)
69 (36-111) 35 (27-48) 272 (154-552) 1,220 (1,046-1,538) 5.3 (5.0-6.8) 15 (9-20) 5.8 (5.4-6.2) 4.2 (2.0-5.4) 2.1 (1.7-3.9) 18.4 (9.1)
0.8 0.3 0.9 0.7 0.6 0.7 0.5 0.7 0.9 0.5
65 (45-88) 42 (33-48) 197 (188-531) 1,373 (1,024-1,909) 5.8 (5.4-6.0) 15 (8-18) 5.6 (5.2-6.2) 3.6 (2.5-4.7) 2.4 (1.6-4.6) 17.2 (11.2)
42 (28-118) 33 (26-58) 162 (136-442) 1,268 (912-1,411) 5.6 (5.0-6.4) 14.5 (9.2-23) 5.6 (5.4-6.4) 3.9 (2.3-5.6) 2.0 (1.5-3.6) 18.5 (9.3)
0.3 0.3 0.3 0.4 0.7 0.8 0.8 0.9 0.8 0.7
Abbreviation: AST, aspartate aminotransferase.
P 5 0.06 for the venesection group). There were no other within-group differences in the other domains of the SF-36 from baseline to 6 months. Somnolence scores, as determined by the ESS, were not different between groups at any time point Supporting Table 1. Similarly, there were no within-group differences from baseline to 6 months (P > 0.3 for both groups). Analyses by Ferritin Status and Number of Venesections. Subanalysis was performed in 36 patients who had hyperferritinemia at baseline. Measures of liver injury, HS, IR and sensitivity, and LPO were not different between the control and venesection groups at baseline or end of treatment (Table 3). Similarly, there was no difference in the change in these parameters from baseline to end of treatment between the control and venesection groups (P > 0.3 for all comparisons; Fig. 5A-F). Among subjects who underwent venesection (n 5 29), the reduction in serum ferritin from baseline to end of treatment was significantly correlated with the number of venesection sessions (Spearman’s rho: 20.63; P < 0.001), as shown in Table 4. However, in contrast, there was no significant correlation between measures of liver injury, HS, IR, or LPO with the number of venesection sessions.
Discussion In this prospective, randomized, controlled clinical trial conducted in patients with NAFLD, we have demonstrated that iron removal by way of venesection was not associated with any significant improvement in (1) HS, (2) liver injury, (3) IR or sensitivity, (4) measures of LPO, or (5) quality of life. The lack of benefit was observed across the range of serum ferritin levels, and no dose-response relationship was observed between the number of venesection sessions and any of the study endpoints.
A number of studies have demonstrated an association between intrahepatic iron and severity of liver injury in patients with NAFLD; however, these studies have had conflicting findings of the importance of hepatocellular versus reticuloendothelial iron.2,4,9 Furthermore, cross-sectional studies have not demonstrated whether iron deposition is causal or an epiphenomenon of liver injury in NAFLD. Our findings suggest that the levels of hepatic iron found in NAFLD patients do not have a significant causal role in the genesis of liver injury. A number of previous clinical trials in individuals with NAFLD have demonstrated an improvement in serum ALT and/or HOMA levels with phlebotomy.20,23,35-37 However, these have been single-arm uncontrolled or nonrandomized studies, which are prone to bias and “regression to the mean” phenomena, which may account for supposedly positive findings. It is pertinent to note that a large, randomized, controlled trial was required to demonstrate the lack of efficacy of ursodeoxycholic acid in the treatment of NAFLD after a number of small, uncontrolled trials had previously suggested a therapeutic benefit.38 The only previous randomized, controlled trial of venesection in NAFLD subjects demonstrated an improvement in HS with venesection, but not lobular inflammation or hepatocyte ballooning.39 In this small study of 36 patients, only 50% of subjects underwent an end-of-study liver biopsy, limiting the ability to draw definitive conclusions from this study. A limitation of our trial was the lack of histological assessment and thus ability to determine the impact of phlebotomy on liver fibrosis and direct measures of liver injury. A recent consensus meeting has recommended magnetic resonance–determined HS and serum ALT as suitable primary endpoints for shortterm phase II trials.40 We utilized MRI to determine HS, which is in close agreement with magnetic
1562
ADAMS ET AL.
HEPATOLOGY, May 2015
Fig. 5. Change in measures of liver injury (A and F), HS (B), IR (C and D), and LPO (E) from baseline to the end of treatment in control and venesection groups among patients with hyperferritinemia at baseline (n 5 36). There was no significant difference between groups (P > 0.3 for all comparisons). Abbreviation: EOT, end of treatment.
resonance spectroscopy–determined HS and histologically determined HS.41-43 However, we cannot exclude that minor changes in HS were undetected. A strength of our study is the multiple measures, including serum markers of hepatocyte apoptosis (CK18 levels), LPO (F2-isoprostane levels), multiple measures of insulin metabolism, and patient quality of life. We were unable to detect any impact of phlebotomy, despite having a cohort size similar to other clinical
studies where clinical benefits have been demonstrated.33,44,45 However, there is the possibility that our study was underpowered and improvements in the primary or secondary outcomes may not have been detected. Furthermore, we cannot exclude that venesection may be of benefit in certain subgroups of patients with NAFLD, such as those with NASH or increased hepatic iron concentration, or that a longer duration of treatment may be therapeutic.
HEPATOLOGY, Vol. 61, No. 5, 2015
ADAMS ET AL.
Table 4. Correlation Between Number of Venesection Sessions and Change in Outcomes From Baseline to End of Treatment (n 5 29) All Venesection Patients (n 5 29) Characteristic (Month 6 to Month 0 Value)
Spearman’s rho
P Value
Ferritin ALT, IU/L AST, IU/L CK-18 F2-isoprostanes, pmol/L Glucose, mg/dL Insulin, mU/L HbA1c, % HOMA ISI HS, %
20.69 0.02 0.03 20.01 0.23 0.13 20.002 0.005 0.08 20.04 0.32
