Phase II clinical trial of phlebotomy for non-alcoholic fatty liver disease


Correspondence to:

Dr M. D. Beaton, Division of Gastroenterology, London Health Sciences Centre - University Hospital, Room ALL-132, 339 Windermere Road, London, ON N6A 5A5, Canada.




Elevated iron indices are described in non-alcoholic fatty liver disease and iron reduction has been suggested as a potential therapy.


To determine whether phlebotomy is an effective therapy for non-alcoholic fatty liver disease.


Patients with biopsy proven non-alcoholic fatty liver disease underwent baseline evaluation to determine severity of metabolic and liver disease. A Phase II trial of phlebotomy was carried out to achieve near-iron depletion (serum ferritin ≤50 μg/L or haemoglobin 100 g/L). Repeat liver biopsy, anthropometric and biochemical measurements were performed 6 months following the end of treatment. Primary outcome was improvement in liver histology, assessed using the non-alcoholic fatty liver disease activity score.


Thirty-one patients completed follow-up. Iron reduction resulted in a significant improvement in the non-alcoholic fatty liver disease activity score (−0.74 ± 1.83, P = 0.019). Reductions in individual histological features of lobular inflammation (−0.29 ± 1.07, P = 0.182), steatosis (−0.26 ± 0.82, P = 0.134), hepatocyte ballooning (−0.19 ± 0.70, P = 0.213) did not achieve significance nor did the score for fibrosis (−0.32 ± 0.94, P = 0.099).


This prospective Phase II study of phlebotomy with paired liver biopsies evaluating phlebotomy therapy in non-alcoholic fatty liver disease patients suggests that iron reduction may improve liver histology. However, the effect size of phlebotomy raises questions of whether treatment could have sufficient clinical significance to justify a definitive Phase III trial. This trial has been registered with the US National Institute of Health (, Identifier NCT 00641524).


Non-alcoholic fatty liver disease (NAFLD) is the most common cause of liver disease in the western world and its incidence, as well as that of its complications continues to rise. The impact of NAFLD on health resources is significant, with costs and utilisation reported to be 26% higher than healthy individuals.[1] As such, its treatment has become an area of much interest. Nonpharmacological lifestyle interventions positively impact NAFLD and the metabolic syndrome and are recommended first-line therapies.[2, 3] Unfortunately, these changes are difficult for patients to sustain long term and so other treatment options continue to be investigated. To date, however, there remains a lack of convincing data supporting the widespread use of pharmacological agents or bariatric surgery.[4]

Hyperferritinemia is common in the metabolic syndrome[5] and NAFLD.[6] Evidence is mounting that elevated iron is linked to increased risk of complications including cardiovascular outcomes, diabetes severity and progression of hepatic disease.[7, 8] Recently, one of the largest studies evaluating iron and liver histology in NAFLD demonstrated that elevated serum ferritin (SF) identified NAFLD patients with more advanced disease.[9] Previous studies have suggested a benefit of iron reduction therapy on NAFLD severity and insulin sensitivity.[10-12] To date, however, the impact of this therapy on liver histology, considered to be the gold standard for evaluating efficacy of liver disease treatment, remains uncertain.

The purpose of a Phase II trial is to obtain some evidence of treatment response and safety in small groups of patients.[13] While Phase II results are not definitive; they are helpful for justifying and designing larger Phase III trials.

The primary purpose of this study was to conduct a Phase II trial of the effects of phlebotomy in 31 individuals with biopsy proven NAFLD. A secondary aim was to use the estimated effect size to calculate the sample size needed for a definitive Phase III trial.



We recruited 31 consecutive adults (18–75 years) referred to London Health Sciences Centre (LHSC) for NAFLD between April 2009 and November 2010 through letters sent to family physicians and gastroenterologists. Informed consent was obtained from each patient included in the study. Participants required a diagnosis of NAFLD using American Association for the Study of Liver Diseases diagnostic criteria.[4] Patients were excluded if they had any other liver disease or decompensated cirrhosis defined by stage 4 fibrosis on biopsy with liver synthetic dysfunction clinically or biochemically. Patients were also excluded if they had a contraindication to obtaining a liver biopsy, were unable or unwilling to provide informed consent, or had a history of alcohol intake of greater than 10 g/day in women or 20 g/day in men. Female patients were excluded if they were pregnant or breastfeeding, or of child-bearing potential and not using a reliable method of contraception. Patients with type 2 diabetes (DM) were required to have stable doses of antihyperglycaemic medications for at least 6 months. All patients underwent testing for mutations of the two major mutations of the HFE gene (C282Y and H63D). Subjects homozygous for the C282Y HFE mutation were excluded.

Study design and organisation

Patients meeting diagnostic criteria for NAFLD were offered therapy with phlebotomy. This consisted of the removal of 500 mL of blood at each biweekly or monthly session, depending on patient tolerance and availability, until near-iron depletion (NID) was reached. NID was defined as either a SF of ≤50 μg/L or a decline in haemoglobin to ≤100 g/L. All patients received counselling regarding healthy eating and recommended lifestyle modifications, but individual diets were not prescribed or monitored.

Evaluation and monitoring of patients

All subjects underwent anthropometric and laboratory assessments prior to therapy and at the end of follow-up. Compliance and adverse effects were evaluated on each of these occasions. Anthropometric assessment included body mass index (BMI) and waist: hip ratio. Laboratory parameters included alanine aminotransferase (ALT), aspartate aminotransferase (AST), gamma-glutamyltransferase, albumin, bilirubin, international normalised ratio, haemoglobin, platelet count, serum ferritin (SF), transferrin saturation, erythrocyte sedimentation rate (ESR), c-reactive protein (CRP). Glycated haemoglobin, fasting lipids, glucose and insulin levels were measured and insulin resistance was determined using the homeostasis model assessment (HOMA-IR).[14] Total iron removed via phlebotomy for each subject was calculated by multiplying the number of phlebotomy sessions by the amount of iron in 500 mL of whole blood (250 mg/500 mL).[15]

Liver histology

Percutaneous liver biopsy was performed in all participants prior to initiation of therapy with a median of 29 days (range: 12–355 days) prior to first phlebotomy session, and at the end of follow-up (6 months following attainment of near-iron depletion). This follow-up time frame for the second biopsy was chosen to assess the longer term effect of phlebotomy on liver histology. All biopsies were scored according to the extent of steatosis (on a scale of 0–3), lobular inflammation (0–3) and hepatocellular ballooning (0–2) present using the NAFLD activity score (NAS)[16]; fibrosis (0–4) was scored separately. A diagnosis of NASH required the presence of steatohepatitis, active disease (NAS ≥ 3) plus a hepatocellular ballooning score ≥1.

Liver iron assessment was according to a 5-point semi-quantitative grading scale using a Prussian blue stain to grade the intensity of stainable iron deposits in hepatocytes from 0 (none) to 4+, where 0 is absent or barely discernible staining of iron granules at 400× magnification and 4+ is masses of iron visible at 10× magnification or with the naked eye.[17, 18] Following completion of the study, the adequacy of all biopsy specimens was assessed, aggregate length recorded and histology formally reviewed by a central, blinded pathologist with expertise in liver biopsy interpretation. As per study design, final analysis was based on the results of this formal review. Hepatic iron concentration (HIC) was determined using high resolution inductively coupled plasma – mass spectrometry using a Finnigan MAT Element High Resolution ICP-MS located in the Trace Elements Laboratory of LHSC.[19] A HIC >36 μmol/g indicates hepatic iron overload.[20]

Liver stiffness assessment

Liver stiffness measurements using transient elastography (TE) were carried out using the FibroScan (Echosens; Paris, France) XL probe. This is a non-invasive surrogate measure of liver fibrosis and has been validated in NAFLD.[21] However, its accuracy in assessing response to a therapeutic intervention is not as well known.

Experienced operators performed all examinations as per the manufacturer's recommendations. As an indicator of variability, the ratio of the interquartile range of liver stiffness to the median value (IQR/M) was calculated. Examinations with fewer than 10 valid measurements and/or an IQR/M >30% were considered unreliable.

The study protocol conformed to the principles of the Declaration of Helsinki as reflected in a priori approval by the Research Ethics Board at Western University on 5 May 2008. This trial has been registered with the US National Institute of Health (, Identifier NCT 00641524). The author and co-authors had access to the study data and reviewed and approved the final manuscript.

Primary and secondary outcomes

The primary outcome was histological severity of liver disease as assessed by the improvement in liver biopsy measured by the NAS. The proportion of subjects who met the criteria of no worsening of fibrosis, an improvement by 1 or more points in the total NAS score and no increase in the individual scores for steatosis, lobular inflammation or hepatocyte ballooning was also determined.

Secondary outcomes included improvements in individual scores for fibrosis, steatosis, lobular inflammation and hepatocyte ballooning as well as HIC. Additional outcomes included improvement in serum transaminases, SF, IR, lipid profile, TE and anthropometric measures. A comparison between histological responses to phlebotomy in patients with normal vs. increased baseline SF was also conducted. The upper limit of normal (ULN) was defined as <200 μg/L for female subjects and <300 μg/L for males, as per the Hemochromatosis and Iron Overload Screening Study (HEIRS).[22] Similar comparisons were made between those with baseline haemoglobin above and below the median level.

Statistical analysis

The primary outcome of improvement in liver histology was assessed using the Wilcoxon Signed rank test. Secondary outcomes were determined using paired t-test. Sample size was derived using the methods of Cohen,[23] with 80% power to detect a difference in fibrosis as small as 2/3 of a standard deviation, of −0.46 (2-tailed, alpha = 5%) with n = 36. An additional seven subjects were added to allow for dropouts. All analyses were performed using sas version 9.3 (2012 SAS Institute Inc., Cary, NC, USA). Nominal, two-sided p-values were used and were considered to be statistically significant if P < 0.05; no adjustments for multiple comparisons were made.


Patient characteristics

Subjects were recruited between January 2009 and November 2010. Forty-three subjects were enrolled, 31 completed study follow-up and were included in this analysis. Twelve subjects withdrew citing difficulties attending treatment or follow-up visits. The characteristics, baseline and end of treatment laboratory test results and biopsy features of these patients are shown in Table 1.

Table 1. Patient characteristics at baseline and end of follow-up
 BaselineEnd of follow-upP-valuea
  1. Data are presented as mean ± standard deviation or N (%).

  2. a

     Paired t-test was used to compare results before and after phlebotomy therapy.

  3. b

     The homeostasis model assessment for insulin resistance (HOMA-IR) was used to calculate insulin resistance according to the following formula: [glucose (mmol/L) × insulin (mU/L)]/22.5. Higher numbers indicate greater insulin resistance.

  4. c

     NASH was defined as: definite steatohepatitis according to the reading pathologist with a NAS of 3 or more and a score of at least 1 for hepatocellular ballooning.

Demographic factors
Age (years)49 ± 12  
Male19 (61)  
Hypertension19 (61)18 (58) 
Type 2 Diabetes9 (28)9 (28) 
Hyperlipidaemia7 (22)7 (22) 
Anthropometric measures
BMI (kg/m2)33.3 ± 5.033.6 ± 5.20.18
Waist-to-Hip Ratio0.96 ± 0.10.97 ± 0.10.27
Serum biochemistry
AST (U/L)47.0 ± 20.638.8 ± 14.20.03
ALT (U/L)63.5 ± 30.649.7 ± 18.20.004
Bilirubin (μmol/L)11.63 ± 6.2112.18 ± 8.50.76
Ferritin (μg/L)383.9 ± 331.187.9 ± 78.1<0.001
Transferrin Saturation (%)33.2 ± 13.130.7 ± 14.30.41
ESR (mm/h)16.7 ± 15.315.9 ± 15.40.67
CRP (mg/L)9.6 ± 16.06.0 ± 8.60.33
Metabolic factors
Fasting Serum Glucose (mmol/L)6.52 ± 1.86.18 ± 1.240.14
Glycated haemoglobin (%)6.6 ± 1.76.3 ± 1.10.38
Insulin resistanceb3.0 ± 1.82.5 ± 1.20.21
Triglycerides (mmol/L)1.7 ± 0.61.7 ± 0.90.76
Total cholesterol (mmol/L)4.7 ± 1.24.6 ± 1.20.22
HDL-Cholesterol (mmol/L)1.0 ± 0.31.0 ± 0.30.28
Hepatic tissue
NASHc23 (74)17 (55) 
Transient Elastography (kPa)11.6 ± 7.510.4 ± 6.40.41
Liver Biopsy Length (mm)22.8 ± 7.616.1 ± 4.7 
Hepatic Iron concentration (HIC) (μmol/g)16.5 ± 17.85.2 ± 4.7<0.001
HIC > 36 (0–36 μmol/g)2 (6.5)0 

Regarding HFE genotype, three subjects were C282Y heterozygotes and two were H63D heterozygotes. There was one H63D homozygote and no compound heterozygotes. Only one subject, a C282Y heterozygote, had excess liver iron (HIC 62.2 μmol/g).

Phlebotomy therapy

Median baseline haemoglobin was 147 g/dL (range: 116–174 g/dL) and SF 295 μg/L (range: 60.5–1305 μg/L). Among those with baseline SF above the median, a mean of 10 phlebotomy sessions was required to achieve NID, representing an average of 2339 mg of iron. For those below the median baseline ferritin, a mean of five sessions was required, representing an average of 1265 mg of iron. Although 46% more iron was removed from the group with baseline SF above the median, among both groups, the mean haemoglobin declined by approximately 15 g/L, with a posttreatment mean of 133 g/L.

Post-hoc subgroup analyses were performed correlating the total iron removed with change in NAS using Spearman's Rank Correlation rho (ρ). For the overall group (N = 31), ρ = 0.16, = 0.32. Among subjects with baseline SF less than or equal to the median, ρ = −0.27, P = 0.41 and for those with baseline SF above the median, ρ = 0.04, P = 0.80.

Histological outcomes

Phlebotomy resulted in a statistically significant improvement in NAS (P = 0.019), but no improvement in fibrosis nor the individual NAS components; lobular inflammation, steatosis or hepatocyte ballooning (Table 2). It also induced a significant reduction in hepatic iron as measured by HIC (P < 0.001). Representative pre- and posttreatment liver histology of one subject are shown in Figure 1.

Table 2. Primary outcome – improvement in histological features
CharacteristicMean change from baselineP-valuebNumber of patients with improvementa
  1. Data are presented as mean ± s.d. or N (%).

  2. a

     For NAS, this represents the proportion of subjects with a 1 or greater decrease in NAS and no increase in steatosis, lobular inflammation, hepatocyte ballooning or fibrosis. Improvement in steatosis, lobular inflammation, hepatocyte ballooning or fibrosis is defined as a decrease of at least 1 point.

  3. b

     P-values from Wilcoxon Signed rank test.

NAS0.74 ± 1.830.01912 (38.7)
Steatosis0.26 ± 0.820.13413 (41.9)
Lobular inflammation0.29 ± 1.070.18211 (35.5)
Hepatocyte ballooning0.19 ± 0.700.2139 (29.0)
Fibrosis0.32 ± 0.940.09912 (38.7)
Figure 1.

The pre- and posttreatment liver histology of a representative subject are shown. Trichrome Stain, 400× (original magnification). (a) Pre-treatment NAS 6 (0–8), fibrosis stage 2 (0–4), serum ferritin 436.3 μg/L, HIC 6.8 μmol/g, ALT 70 U/L. (b) Posttreatment NAS 2, fibrosis stage 1, serum ferritin 29.5 μg/L, HIC 2.3 μmol/g, ALT 24 U/L.

Thirty-nine per cent of patients (12 of 31) demonstrated a reduction of 1 or more points in NAS and no increase in fibrosis, steatosis, lobular inflammation or hepatocyte ballooning. (Table 2) Twenty-six per cent of patients (8 of 31) demonstrated a reduction of a minimum 2-point improvement in NAS with at least a 1-point improvement in more than one category and no worsening fibrosis (Table S1).[24] Six (26%) subjects with NASH at baseline demonstrated resolution of steatohepatitis, defined as a NAS of ≤2 or a decrease in the NAS of ≥3 points with no increase in fibrosis. Figure 2 shows the change in fibrosis score over time.

Figure 2.

The change in fibrosis stage over the duration of phlebotomy therapy and 6-month follow-up period prior to the end of treatment of liver biopsy is shown. There was an average of 448 ± 118 days between the pre- and posttreatment biopsies.

Subjects with baseline SF above the ULN had higher NAS, fibrosis and individual NAS component scores prior to treatment and demonstrated a significant reduction in the NAS compared with those with baseline SF below the ULN (Table S2). No significant reduction in fibrosis or the individual NAS component scores was seen in either group following treatment, nor was there a significant difference in the change in scores between groups following treatment. (Table 3) The relationship between the change in histological score and decrease in serum ferritin is shown in Figures 3a through 3d.

Table 3. Outcomes among subjects with baseline SF above and below ULN
CharacteristicSF ≥ ULN (= 17)SF < ULN (= 14)
Mean change from baseline P-valuea Mean change from baseline P-valuea
  1. ULN = Serum ferritin upper limit of normal; defined as 200 μg/L in females and 300 μg/L in males.

  2. Data represent change from baseline for each group, presented as mean ± s.d.

  3. a

     P-values from Wilcoxon Signed rank test.

NAS1.00 ± 1.460.0110.43 ± 2.210.544
Steatosis0.35 ± 0.930.2130.14 ± 0.660.688
Lobular Inflammation0.41 ± 1.120.2100.14 ± 1.030.813
Hepatocyte Ballooning0.24 ± 0.560.2190.14 ± 0.860.766
Fibrosis0.18 ± 1.070.6720.50 ± 0.760.063
Figure 3.

The change in liver histological scores in relation to reduction in serum ferritin. (a) Steatosis and decrease in serum ferritin. (b) Lobular inflammation and decrease in serum ferritin. (c) Hepatocyte ballooning and decrease in serum ferritin. (d) Fibrosis and decrease in serum ferritin.

Similar results were noted among those with baseline haemoglobin above the median. In that group, baseline NAS and individual component scores, but not fibrosis, were higher than those with baseline haemoglobin below the median. A significant reduction in the NAS was noted in those with baseline haemoglobin above the median, but neither group demonstrated a significant reduction in fibrosis or the individual NAS components scores, nor was there a significant difference in the change in scores between groups following treatment (Table 4).

Table 4. Outcomes among subjects with baseline haemoglobin above and below mediana
CharacteristicHaemoglobin ≥ 147 g/L (N = 17)Haemoglobin < 147 g/L (N = 14)
Mean change from baseline P-valueb Mean change from baseline P-valueb
  1. Data represent change from baseline for each group, presented as mean ± s.d.

  2. a

     Median baseline haemoglobin 147 g/dL (range: 116–174 g/dL).

  3. b

     P-values from Wilcoxon Signed rank test.

NAS1.12 ± 2.120.0380.29 ± 1.330.410
Steatosis0.29 ± 0.850.2670.21 ± 0.800.531
Lobular Inflammation0.47 ± 1.230.2360.07 ± 0.83>0.999
Hepatocyte Ballooning0.35 ± 0.700.1680.00 ± 0.680.168
Fibrosis0.24 ± 0.660.2890.43 ± 1.220.305

Prior to treatment, 14 subjects (45%) had detectable liver iron on biopsy. This was mild (+1) in all cases and the pattern of distribution was within hepatocytes in 6, reticuloendothelial cells (RES) in 4 and mixed in 4. In only two subjects did posttreatment biopsy demonstrate persistent mild iron staining; this was within hepatocytes in one and RES cells in the other. Figure 4 shows the change in NAS relative to the reduction in HIC with phlebotomy.

Figure 4.

Change in NAFLD Activity Score (NAS) relative to the change in Hepatic Iron Concentration (NAS) (μmol/g) following treatment. There was an average of 448 ± 118 days between the pre- and posttreatment biopsies.

Biochemical and anthropometric outcomes

Phlebotomy resulted in significant improvements in ALT and AST and, as anticipated, SF. There was no evidence to suggest a significant association between phlebotomy and changes in measures of IR, glucose or insulin, lipid profile or systemic inflammation as measured by ESR and CRP. There was no significant change in BMI or Waist-to-Hip Ratio. Although many metabolic and biochemical factors did not change by a statistically significant degree, it is noteworthy that changes were in a clinically desirable direction in 15 of 18 (83%) indicators.

Transient elastography

In keeping with the lack of reduction in fibrosis on liver histology, no significant change in liver stiffness as measured by TE was noted (P = 0.45).

Adverse events

The occurrence of adverse experiences was monitored throughout the study. One patient required an extension of the interval between treatments to every 6 weeks due to fatigue following phlebotomy sessions. No significant adverse events occurred and no patients discontinued therapy as a result of adverse events.


Phase II studies are appropriate for preliminary evaluations of treatments, particularly in conditions that lack a widely accepted efficacious, safe and efficient treatment. This small Phase II study showed statistically significant reductions in the NAS score, ALT, AST, SF and hepatic iron concentration. Although not statistically significant, improvements in steatosis, lobular inflammation, hepatocyte ballooning, and fibrosis were all in the predicted direction. The proportion of NASH patients who improved was 26% (95% CI: 11.3–46.6). No important adverse events were noted. Taken together, these results suggest that phlebotomy to near-iron depletion could improve liver histology and function in patients with NAFLD in a safe and well-tolerated fashion.

In this study, each patient served as his or her own control, and the measures used were objective. However, the absence of an untreated control group means that the findings must be viewed with caution. While not as strong as a concurrent control group, changes in the same outcome in untreated NAFLD patients from RCTs of other treatments can put the present findings in perspective. Two RCTs, with placebo groups of 83 and 29, reported mean changes in NAS scores of −0.5 and −0.1, respectively, for a weighted average of −0.4. When subtracted from the mean change in NAS, we observed of −0.74, the difference is −0.34 or 0.19 of one standard deviation. As a treatment effect size, this is considered small. To have 80% power to detect this difference or larger at the 5% level of significance would require two groups of about 450 patients.

Aside from recommendations regarding appropriate caloric intake and lifestyle changes, no other specific interventions were instituted and no reduction in weight or BMI was seen over the study period, which suggests that the effects noted on liver histology are due to the effect of phlebotomy. Although the majority of outcome measures changed in the predicted direction with treatment, the lack of improvement in weight and BMI reflects the difficulty this group of patients has in achieving these changes over the period of follow-up, despite being part of a clinical trial with regular medical follow-up. While the effect of phlebotomy on insulin resistance was not statistically significant, a reduction in HOMA-IR scores of approximately one quarter of a standard deviation was seen. It is noteworthy that the estimated effect size of 0.26 was consistent with that observed in the treatment group of trials of pentoxifylline (ES = 0.32) and pioglitazone (ES = 0.18).[25, 26] As shown above, a larger sample size and concurrent random controls are needed to better estimate the effect of phlebotomy on IR in these patients.

The results of this study are consistent with previously published work demonstrating that subjects with elevated baseline SF are likely to receive the greatest benefit from iron reduction.[12] Also in keeping with previous studies,[27-31] an important observation in this trial was that despite the majority of subjects having elevated SF, only two had hepatic iron overload (HIC > 36 μmol/g) at baseline. There was, however, a significant reduction in HIC in the overall study population, indicating that phlebotomy successfully removes tissue iron. As most subjects had normal HIC at baseline, we feel it is unlikely that the improvement in liver histology is solely due to reduction in liver iron. It is quite possible that iron reduction has anti-inflammatory effects independent from decreasing tissue iron.[32, 33] Iron is a known potent catalyst of oxidative stress[34, 35] and may play a role in inducing fibrogenesis.[36]

There is evidence to support haemoglobin as a risk factor for NAFLD, particularly in individuals without the metabolic syndrome.[37-39] Potential mechanisms of liver injury have been suggested to relate to higher iron accumulation as well as increased blood viscosity raising peripheral resistance and reducing hepatic blood flow.[37, 40] Although the majority of patients in this study met diagnostic criteria for the metabolic syndrome, it is interesting to note that subjects with higher baseline haemoglobin had more severe histological disease. As well, iron reduction therapy resulted in an improvement in the NAS. These findings support this relationship and warrant further evaluation of the role haemoglobin level may play in NAFLD pathogenesis.

The impact of HFE gene status on iron accumulation in NAFLD has been evaluated. A recent large multicentre study concluded that although hepatocyte iron accumulation was associated with a higher risk of moderate/severe fibrosis, no significant association between the presence of HFE mutations and liver disease severity was found. It was concluded that HFE genetic testing has limited value in predicting those patients with more severe disease.[27] Similar results were seen in our study where of the six subjects carrying HFE mutations, only one demonstrated hepatic iron accumulation. This is in keeping with a relatively weak association between HFE mutations and hepatic iron deposition, which has been suggested to be due to the low penetrance of HFE mutations.[41]

We recognise that the results of the present trial are limited by the lack of concurrent random controls.[25, 26] Additional limitations include the relatively mild overall degree of iron overload and liver disease at baseline (mean NAS = 3.8), combined with variability in the severity of individual NAS parameters and clinical characteristics, including degrees of hepatic iron deposition. To reliably assess changes in liver fibrosis, studies with a 12-month treatment duration are recommended.[24] In the case of phlebotomy for iron reduction, treatment duration cannot be standardised, as time to iron depletion varies in each individual. An attempt was made to control for this by obtaining the final liver biopsy 6 months following the end of treatment, thus demonstrating the longer term histological effect of this intervention. As the sample size was relatively small, it was not possible to perform multivariable analysis to control confounding and examine interactions.

NAFLD patients have been shown to have increased overall mortality compared with matched control populations.[42-44] NASH is presently the third most common reason for liver transplantation in the United States[45] and is projected to become the most common indication in coming years.[46] The identification of treatment modalities for NAFLD is therefore an important, albeit challenging area of research. Despite a significant amount of work to date, we still have little to offer these patients. It has been stated that the ideal treatment ‘should be one that decreases overall mortality, including liver-related and cardiovascular deaths, while remaining safe, widely available, and relatively inexpensive.’[24] Phlebotomy is appealing as it easily fulfils the final three criteria, something that cannot be said for some other therapies that have been evaluated for this condition. Although our results suggest that phlebotomy has a positive impact on liver disease in NAFLD as assessed by the NAS, no statistically significant improvement in fibrosis or steatohepatitis was seen. An additional important consideration is whether the estimated treatment effect is clinically significant enough to warrant a randomised trial with an untreated control group, particularly given the ethical and feasibility challenges posed by serial liver biopsies in several hundred patients.

This Phase II trial helps to address the important clinical question of whether iron depletion could have a beneficial effect on liver histology in NAFLD and was undertaken as a hypothesis-generating study to determine whether a larger, Phase III randomised trial would be prudent. On the basis of our results, we suggest that future studies of iron reduction therapy be focused on those patients with baseline SF of at least 300 μg/L and steatohepatitis on liver biopsy to more definitively ascertain the magnitude of benefit in this condition. registration number; NCT 00641524.


Guarantor of the article: Dr M. D. Beaton.

Author contributions: Melanie D Beaton MD contributed to the study concept and design, acquisition of data, analysis and interpretation of data, drafting of manuscript and obtained funding. Subrata Chakrabarti MD contributed to the analysis and interpretation of liver histology data. Mark Levstik MD contributed to the recruitment of study subjects. Mark Speechley PhD performed the statistical analysis and study design, critical revision of manuscript. Paul Marotta MD performed the recruitment of study subjects. Paul Adams MD contributed to the study concept and design, analysis and interpretation of data, critical revision of manuscript. All authors approved the final version of this manuscript.


Declaration of personal interests: Dr Beaton has served as a speaker for Janssen. Dr Adams has served as a consultant for Biomarin and has received research funding from Gilead.

Declaration of funding interests: This study was funded in full by a research grant from the Canadian Liver Foundation.