Digestive Diseases and Sciences

, Volume 53, Issue 3, pp 815–822

Phlebotomy Improves Therapeutic Response to Interferon in Patients with Chronic Hepatitis C: A Meta-Analysis of Six Prospective Randomized Controlled Trials

Authors

    • William Beaumont Hospital
  • Laith H. Jamil
    • William Beaumont Hospital
  • Mamtha Balasubramaniam
    • Research InstituteWilliam Beaumont Hospital
  • Raymond Koff
    • University of Connecticut Health Center
  • Herbert L. Bonkovsky
    • University of Connecticut Health Center
Original Paper

DOI: 10.1007/s10620-007-9945-7

Cite this article as:
Desai, T.K., Jamil, L.H., Balasubramaniam, M. et al. Dig Dis Sci (2008) 53: 815. doi:10.1007/s10620-007-9945-7

Abstract

Prospective randomized controlled trials (RCTs) comparing phlebotomy and interferon (IFN) treatment to IFN alone in patients with chronic hepatitis C (CHC) have suggested a benefit for the phlebotomy group. However, statistical significance was achieved in only one of these trials. We performed a meta-analysis of RCTs comparing phlebotomy and IFN to IFN alone for the treatment of CHC. The MEDLINE database and Cochrane registry of controlled trials were searched using the key words “phlebotomy” and “treatment of hepatitis C.” Reference lists of review articles discussing the interaction between iron and CHC, and prospective RCTs comparing phlebotomy plus IFN therapy to IFN alone were searched to identify additional RCTs that compared phlebotomy plus IFN to IFN alone. Peto odds ratios with their 95% confidence intervals and Forrest plots were generated for each variable to assess the relationships among the studies that had provided that information. Statistical analysis was performed using Comprehensive META-Analysis version 2.0. Six prospective RCTs were identified: all used sustained viral response (SVR) as an endpoint. The three largest RCTs excluded patients with cirrhosis. Two RCTs specifically included only patients with either high ferritin or high hepatic iron content. IFN treatment regimes varied. Length of treatment varied between 6 and 12 months. The phlebotomy plus IFN group and the IFN group did not differ with respect to the percentage of patients with cirrhosis or genotype 1. SVR was attained in 50/182 (27%) patients in the phlebotomy plus IFN group, compared to 22/185 (12%) patients in the IFN group. Peto odds ratio for SVR in phlebotomy plus IFN group was 2.7; 95% CI 1.6–4.5, < 0.0001. All five RCTs published in manuscript form showed a trend towards a benefit from the phlebotomy plus IFN in attaining SVR, and the results of the meta-analysis were not dependent on any single RCT, since excluding any single RCT did not change the results. Phlebotomy improves the SVR in response to IFN treatment in patients with CHC. Confirmation of this will require RCT with detailed pre-treatment iron studies and appropriately powered to demonstrate a statistically significant benefit.

Keywords

PhlebotomyChronic hepatitis CInterferonMeta-analysis

Introduction

Iron is a critical factor in the growth of a number of microbes [1], and sequestration of iron is recognized as a crucial host response to infection [2]. In patients infected by the hepatitis C virus (HCV), iron stores, as reflected by serum ferritin levels or by hepatic iron content, are often elevated [3]. The impact of reducing iron stores on the natural history and in the management of chronic hepatitis C (CHC) has not been fully defined. Therapeutic phlebotomy to reduce iron stores in patients with CHC results in a decline in serum aminotransferase levels [4], and a few studies have even reported histological improvement following phlebotomy [5, 6]. In one small series of patients with CHC and transfusion-related iron overload, therapeutic phlebotomy alone led to regression of cirrhosis [6]. High hepatic iron content in CHC has been reported to reduce the likelihood of a sustained viral response (SVR) to monotherapy with interferon (IFN) [713], but not to treatment with the combination of IFN and ribavirin [14, 15].

Randomized controlled trials (RCTs) comparing phlebotomy plus IFN to IFN alone in the treatment of CHC have suggested an advantage for the phlebotomy plus IFN group [1621], although this advantage reached statistical significance in only one of these trials [18], perhaps because of small numbers of subjects.

Therefore, we have conducted a meta-analysis of prospective RCTs that compared treatment with IFN after therapeutic phlebotomy to treatment with IFN alone in patients with CHC.

Methods

The MEDLINE and OVID databases were searched from 1992 to 2005 using the keywords “phlebotomy” and “hepatitis C” and “iron and treatment of hepatitis C.” The reference lists of review articles and RCTs regarding iron and hepatitis C were also searched. The COCHRANE database was searched for RCTs and review articles regarding “iron and hepatitis C,” and the reference lists of these review articles also were searched. The authors of RCTs that compared phlebotomy plus IFN to IFN alone were contacted to determine if they were aware of additional RCTs.

Trials were included only if they were prospective and reported randomization, if an intention-to-treat analysis was possible, and if SVR was included as an endpoint.

Three separate reviewers extracted data. Data extraction was replicative and reproducible among all three reviewers.

Trial quality

The methodological quality of trials included in the meta-analysis was scored using the Jadad composite scale [22, 23]. A maximum of 5 points can be attained, and a score ≤2 signifies a low quality study while a score of ≥3 signifies a high quality study. A trial involving phlebotomy could not conceivably be blinded, so it is hard to conceive of a trial involving phlebotomy attaining a score >3. Quality scores for the trials are provided in Table 1.
Table 1

Jadad quality scoresa of the trials included in this report

First author/country of origin/reference no.

Randomization

Method used to generate sequence

Double-blind

Method of double-blind

Description of withdrawals and dropouts

Total points

Van Thiel/USA/[20]

1

1

0

0

1

3

Fong/USA/[19]

1

0

0

0

1

2

Fontana/USA/[16]

1

1 (Stratified randomization scheme)

0

0

1

3

Fargion/Italy/[17]

1

1 (Stratified randomization scheme)

0

0

1

3

Carlo/Italy/[18]

1

0

0

0

1

2

Rossini/Italy/[21]

1

0

0

0

1

2

aJadad score calculation

Item

Score

Was the study described as randomized (this includes words such as randomly, random, and randomization)?

0/1

Was the method used to generate the sequence of randomization described and appropriate (table of random numbers, computer-generated, etc.)?

0/1

Was the study described as double-blind?

0/1

Was the method of double-blinding described and appropriate (identical placebo, active placebo, dummy, etc.)?

0/1

Was there a description of withdrawals and dropouts?

0/1

Deduct one point if the method used to generate the sequence of randomization was described and it was inappropriate (patients were allocated alternately, or according to date of birth, hospital number, etc.).

0/−1

Deduct one point if the study was described as double-blind but the method of blinding was inappropriate (e.g., comparison of tablet vs. injection with no double dummy).

0/−1

All trials reported “randomization,” and two [16, 17] of the trials reported a “stratified randomization” scheme. Randomization was based on opening a sealed envelope in a third trial [20], and the method of randomization was not provided in any of the remaining three trials [18, 19, 21]. Two trials [16, 17] reported an intention-to-treat analysis, while the other trials provided comprehensive follow-up and reported all dropouts, so that an intention-to-treat analysis was possible. None of the trials reported blinding of the investigators or the patients; this would seem virtually impossible in a protocol involving therapeutic phlebotomy.

Statistical analysis

Each of the variables studied, namely SVR, genotype distribution, and prevalence of cirrhosis was analyzed separately for the six studies, wherever applicable. Peto odds ratios with their 95% confidence intervals (C.I.) and Forrest plots were generated for each variable to assess the relationships among the studies that had provided that information. Between-study tests of heterogeneity were performed using the I2 statistic which has been shown to be more reliable and powerful than the χ2 test based on the Mantel-Haenszel statistic when small numbers of studies are available [23, 24]. By definition, it is the percentage of total variation across studies that is due to heterogeneity rather than chance [23]. Negative values of I2 are put equal to zero so that 0% ≤ I2 ≤ 100% [23]. A value of 0% indicates no observed heterogeneity, and larger values show increasing heterogeneity [23]. As a guide, I2 values of 25% may be considered low, 50% moderate, and 75% high [24]. As there were fewer than 10 studies under review in this study, the possibility of small study and publication bias was not investigated by using funnel plots and Egger-weighted regression [2].

For comparison purposes, between-study tests of heterogeneity were also performed with those based on Mantel-Haenszel for fixed effects reported (as all our studies were homogeneous) with significant P-values implying that the dispersion in treatment effects was more than would be expected by chance. P-values less than an alpha of 0.05 (probability of type I error) were considered statistically significant. Statistical analysis was performed using Comprehensive META-Analysis version 2.0, and the I2 statistic was computed by hand.

Results

Six prospective RCTs were identified that met the criteria for inclusion in the meta-analysis [1621]. Five were published as full manuscripts, and one was published as an abstract [21]. The five trials published in manuscript form used IFN thrice weekly or daily. The trial published in abstract form was the only one that used pegylated IFN plus ribavirin treatment in patients with CHC and concomitant porphyria cutanea tarda [21]. Although exact treatment regimens differed from trial to trial, control and study groups received the same treatment within each trial either with or without iron reduction by therapeutic phlebotomy.

Patient population

The patient population is described in Table 2. Two trials from Italy included only patients with high iron stores, as assessed by either high ferritin levels or high normal hepatic iron content, but both of these trials specifically excluded patients with cirrhosis [17, 18]. A multicenter study from the United States included patients with or without iron overload, but also excluded patients with cirrhosis [16]. These three trials were the three largest trials and included almost 75% of the patients in the meta-analysis.
Table 2

Selected demographics and clinical features of subjects studied

Trial, by author

M/F

Genotype 1/total (n, %)

Cirrhosis/total

AA/total

Control

Phleb

Control

Phleb

Control

Phleb

Control

Phleb

Van Thiel [20]

12/3

13/2

NR

NR

5/15

6/15

NR

NR

Fong [19]

17/4

14/3

15/21 (71%)

14/17 (53%)

4/21

1/17

12/21

8/17

Fontana [16]

26/16

23/17

32/42 (76%)

27/40 (67%)

0/42

0/40

7/42

1/40

Fargion [17]

38/19

48/9

1–4 39/57 (68%)

1–4 36/57 (63%)

0/57

0/57

NR

NR

Carlo [18]

38/2

40/3

34/40 (85%)

37/43 (86%)

0/40

0/43

NR

NR

Rossini [21]

10/0

9/1

8/10 (80%)

10/10 (100%)

NR

NR

NR

NR

Total

141/44

147/35

128/170 (75%)

124/167 (74%)

9/175

7/172

19/63

9/57

M Male, F female, AA African American, Phleb phlebotomy, NR not reported

Five of the six studies provided data on viral genotype, while one study did not [20]. Our meta-analysis both included and excluded the latter study.

Across all studies, the control group and the phlebotomy group did not differ with respect to the number of patients with cirrhosis or genotype 1 (Table 2).

African heritage was mentioned in two studies, and African Americans were more likely to be present in the control group [16, 19].

Treatment protocols for the clinical trials

The treatment regimens, history of prior treatment, pre-treatment iron status, and major outcomes are summarized in Table 3. Standard IFN was given thrice weekly in two trials for 6 months [16, 19], and IFN was given thrice weekly in higher doses for 1 year in the two Italian trials [17, 18]. In one trial, which was conducted in patients who previously had failed treatment, daily IFN was given for 6 months [20]. In the trial reported in abstract form only [21], patients were treated with pegylated IFN and daily ribavirin dosed according to body weight for an uncertain period, but because 90% of patients were genotype 1, we assumed a treatment period of 48 weeks.
Table 3

Measures of iron status at baseline, history of prior therapy for CHC, IFN treatment regimes, and SVR rates in the six trials

Trial, by author

IFN Rx regimes

H/O prior Rx for HCV

Iron status at baseline (mean ± SD)

IFN Rx SVR/total (n, %)

Phleb + IFN Rx SVR/total (n, %)

P-value

Van Thiel [20]

5MU IFN daily × 6 months

Prior failures

HIC (μg/g dry wt): C 9,466 ± 63, P 1,033 ± 82

2/15 (13)

9/15 (60)

NR

Fong [19]

3MU IFN-α2b tiw × 6 months

Previously untreated (from 1st author)

Ferritin (ng/ml): C 196 ± 61, P 200 ± 185

1/21 (4.8)

5/17 (29.4)

0.07

Fontana [16]

3MU IFN-α2b tiw × 6 months

Previously untreated

Ferritin (ng/ml): C 253 ± 55, P 235 ± 27

3/42 (7)

7/40 (17)

0.2

Fargion [17]

6MU IFN tiw × 4 months, then 3MU tiw × 8 months

Previously untreated

HIC 1,100  ± 640 μg/g dry wt

9/57 (15.8)

16/57 (28.1)

0.17

Carlo [18]

6MU QOD IFN (α2a or α2b) × 6 months then 3MU QOD × 6 months

Previously untreated

Ferritin increase (ng/ml): men >400, women >300

6/40 (15)

12/43 (28)

<0.05

Rossini [21]

PEG IFN and Rib weight-based Rx

NR

Ferritin (ng/ml): P 549 ± 375, C 690 ± 352

1/10 (10)

1/10 (10)

NR

Total

   

22/185 (12)

50/182 (27)

<0.0001

CHC Chronic hepatitis C, IFN interferon, SVR sustained viral response, HIC hepatic iron content, C control, P phlebotomy, Rib ribavirin, NR not reported

Peto odds ratio for SVR in phlebotomy plus IFN group 2.7; 95% CI 1.6–4.5 < 0.0001

Phlebotomy protocol

The phlebotomy protocol for all studies is outlined in Table 4. Iron depletion was induced in five studies. In the sixth trial [18], however, patients were phlebotomized very conservatively only to ferritin <100 ng/ml, which is not an iron-depletion state.
Table 4

Protocols used for iron depletion

Trial, by author

Phlebotomy protocol

Van Thiel [20]

250 ml phlebotomy weekly until Hgb <11 g/dl. Repeat when Hgb ≥11 g/dl

Fong [19]

500 ml phlebotomies every 2 weeks (mean 5.1 sessions, range 2–8) until 2 of 3 criteria were fulfilled: persistent Hgb value 3 g below the pre-phlebotomy level, iron-binding saturation <15%, serum iron <50 μg/dl. Repeated in 10 patients to maintain them in iron-deficient state during IFN treatment

Fontana [16]

400–500 ml every 1–2 weeks until mild iron deficiency anemia until 2 of 3 criteria were fulfilled: hematocrit <35%, ferritin <10 ng/ml, serum transferrin saturation <10%. Repeat during IFN therapy if hematocrit >35% and either ferritin >15 ng/ml or transferrin saturation >15%

Fargion [17]

400 ml weekly phlebotomy (mean of 9.1 ± 4.5 phlebotomies) until transferrin saturation <20% and serum ferritin <10 ng/ml in the presence of at least 2 g/dl of Hgb decrease

Carlo [18]

300 ml phlebotomy every 10–15 days (average 8 sessions) until serum ferritin value <100 ng/ml

Rossini [21]

450 ml phlebotomy every 2 weeks (mean 10.8 sessions, range 6–15) until 2 of 3 criteria were fulfilled: ferritin <10 ng/ml, transferrin saturation <20%, Hgb ≤11.5 g/dl

Virologic outcomes

An advantage for the phlebotomy group was reported in all five trials published as manuscripts: SVR rates were higher in the phlebotomy group than in the control group. However, this difference reached statistical significance in only one of these studies [18]. In the sixth study, published as an abstract, equal SVRs were observed in the phlebotomy and control groups [21]. After pooling the results of all the trials, the SVR rate was significantly higher in the phlebotomy group, 27% (50/182), than in the control group, 12% (22/185) (< 0.0001) (Table 3).

The Peto odds ratio for SVR in the phlebotomy group was 2.7 (95% C.I. = 1.6–4.5; < 0.0001) (Fig. 1). The results of the meta-analysis were not dependent on any single trial, because excluding any single trial did not change either the results or the conclusions.
https://static-content.springer.com/image/art%3A10.1007%2Fs10620-007-9945-7/MediaObjects/10620_2007_9945_Fig1_HTML.gif
Fig. 1

Forrest plot comparing SVR between phlebotomy plus IFN to IFN alone

Histology

Paired liver biopsy samples obtained before and after treatment were available in three of the studies. Unfortunately, paired liver biopsy samples before and after treatment were available in fewer than 45% of the patients in the two largest trials with the highest quality ratings [16, 17]. Paired liver biopsy samples were available in all of the patients in the third trial [20], but this was a smaller trial. All three trials found greater histological improvement in the phlebotomy group. Fontana et al. [16] found that the phlebotomy group demonstrated an improvement in the inflammation score after treatment, while the control group did not show an improvement in inflammation [16]. Fargion et al. [17] found that the phlebotomy group demonstrated an improvement in the hepatic activity index and fibrosis scores, whereas the control group demonstrated an improvement in the hepatic activity index but not the fibrosis score [17]. The histology data from these two studies must be interpreted with caution because paired liver biopsies were available in a minority of patients, and the virologic outcomes of the patients with paired liver biopsies were not reported. Thus, the apparent histological benefit may relate in part to the fact that paired liver biopsies were not uniformly obtained from the same proportions of responding and nonresponding patients. The only trial in which paired liver biopsies were obtained in all patients found significant histological improvement using the Knodell score in the phlebotomy group but not the control group [20].

Discussion

This meta-analysis of published RCTs is limited by the small number of published trials, the small number of patients included, and the heterogeneity of the trial designs, the patient populations, and the reported data. Nonetheless, this analysis provides support for the notion that phlebotomy to induce iron depletion enhances the efficacy of nonpegylated IFN monotherapy for CHC. The major limitation of the data we report here is that nonpegylated IFN monotherapy is no longer the standard of care therapy for CHC. The only trial that studied the influence of phlebotomy on response to the current standard of care, namely, pegylated IFN and ribavirin, was undertaken in patients with CHC and porphyria cutanea tarda, a disorder associated with excessive stores of iron. Limited data suggest that patients with both CHC and porphyria cutanea tarda are likely to be less responsive to IFN-based treatment than those without porphyria cutanea tarda [25]. Therefore, the enhanced response reported in the phlebotomy group is encouraging. On the basis of our findings, it may be reasonable to undertake a large, statistically powered RCT that evaluates the efficacy of phlebotomy to induce iron deficiency as an adjunct to treatment with pegylated IFN with or without combination ribavirin treatment. One concern in considering such a study is the fact that iron depletion by therapeutic phlebotomy can induce anemia and ribavirin-induced hemolysis also can cause anemia. This may make patient management difficult.

Two separate studies have shown that pre-treatment liver iron content does not correlate with the response to combination treatment with IFN and ribavirin [14, 15]. But both of these studies were performed in patients without high hepatic iron content (HIC). The mean HIC was only 523 μg/g dry weight in one study [14] and less than 700 μg/g in the other study [15]. Previous studies have shown an HIC of 1,100 μg/g dry weight to be a cutoff value separating responders from nonresponders [713].

Two other studies have reported that high iron stores may predict nonresponse to IFN and ribavirin treatment for CHC [26, 27]. Distante et al. [26] found that only 10% of patients with a ferritin >500 ng/ml responded to treatment with IFN and ribavirin.

The largest study to evaluate the relationship between hepatic iron stores and response to combination treatment with IFN and ribavirin is the secondary analysis of data from the HALT C trial.

Analyzing clinical and histological data from 1,145 patients, Bonkovsky et al. found that stainable iron in endothelial cells or in portal triads predicted a lower likelihood of SVR to treatment with IFN and ribavirin [27]. This relationship is particularly impressive for two reasons. First, it was found despite the fact that subjects with Scheuer iron grade 3 or 4+ on baseline biopsy were excluded from analysis. Second, the relationship between stainable iron in portal triads and endothelial cells and SVR was found in patients with advanced fibrosis (Ishak fibrosis score >2).

Thus, phlebotomy to induce iron depletion seems especially relevant to evaluate prospectively in patients with stainable iron in the portal triads or endothelial cells and/or baseline hepatic iron concentrations exceeding 1,100 μg/g dry weight before combination treatment with IFN and ribavirin.

Ribavirin-related hemolytic anemia may result in increased hepatic iron stores, and this phenomenon may impair the efficacy of combination treatment. As a result, it may be beneficial to induce iron depletion before beginning ribavirin treatment. Then too, subsets of patients with CHC are unable to tolerate treatment with ribavirin. This includes patients with end-stage renal disease who are on dialysis, those who are allergic to ribavirin, and those who suffer severe hemolysis in response to ribavirin. In these patients, the use of phlebotomy as an adjunct to treatment with pegylated IFN is particularly worthy of consideration and further study.

Even in the absence of pharmacologic therapy, there is evidence that phlebotomy to induce iron depletion may aid in disease management. Phlebotomy is known to reduce serum aminotransferase levels [4]. A number of studies have shown that phlebotomy to induce iron depletion may lead to regression of fibrosis. In the three trials in which paired liver biopsies before and after treatment with IFN or phlebotomy plus IFN were evaluated [16, 17, 20], all reported histological improvement in the phlebotomy group. Di Bisceglie et al. compared treatment with phlebotomy alone to phlebotomy plus IFN, and found that, after 1 year, the phlebotomy-only group demonstrated mild histological improvement that did not reach statistical significance [28]. Studies with limited numbers of patients have suggested that phlebotomy can induce regression of fibrosis in patients with CHC and regression of cirrhosis in CHC patients with thalassemia who also had transfusion-related iron overload.

Additionally, it has been reported that HIC is correlated with the risk of developing hepatocellular carcinoma [29, 30]. If these observations are correct, it is possible that therapeutic phlebotomy in patients with CHC may provide a benefit to patients even in the absence of an SVR to current therapy.

The safety of therapeutic phlebotomy is well established. Published studies by the American Red Cross have reported serious morbidity related to blood donation of 1 unit as being 1/186,000 [31]. “Serious morbidity” was defined as an event that required a visit to a hospital emergency room or admission to a hospital. Mortality after donation of 1 unit of blood was less than that predicted by actuarial table [31]. The safety of phlebotomy is in stark contrast to the use of one aspirin tablet a day, which is associated with a risk of cerebral hemorrhage in 1/3,000 patients per year of use [32].

Iron depletion may also be achieved by other means, such as by chelation therapy. Indeed, Bayraktar et al. reported that only a few doses of deferoxamine (Desferal) led to improved rate of response to IFN-α among patients with hepatitis B [33], suggesting that a labile, readily chelatable pool of iron may be of much importance. Alternatively, the use of orally active deferasirox (Exjade, Novartis), recently approved for use in the United States and Europe, may provide sufficient iron depletion. However, the safety and efficacy of these chelators in patients with CHC remains to be established in prospective randomized trials.

Even with the current advances in the treatment of CHC, at least half of the patients with genotype 1 who are treated with combination pegylated IFN-ribavirin regimens do not achieve an SVR [24, 25], and for those who cannot tolerate ribavirin, SVRs are even lower. For these patients, strategies such as therapeutic phlebotomy before beginning IFN-based therapy may be an appropriate consideration. However, in an era of evidence-based medicine, the need for an adequately powered prospective, randomized study of this question looms large.

Acknowledgements

Grant support: Supported by NIH contracts NO1 DK29236 and UO1 DK065193 and NIH grant RO1 DK38825 (to HLB).

Copyright information

© Springer Science+Business Media, LLC 2007