A

hBscdtt©

K

1

moaibnnsiicp

0d

Vaccine 25 (2007) 709–718

Granulocyte macrophage colony-stimulating factor as anadjuvant for hepatitis B vaccination: A meta-analysis

Mario Cruciani a,∗, Carlo Mengoli b, Giovanni Serpelloni a,Romualdo Mazzi a, Oliviero Bosco a, Marina Malena a

a Center of Preventive Medicine, HIV Outpatient Clinic, V. Germania, 20-37135 Verona, Italyb Department of Histology, Microbiology, and Medical Biotechnology, University of Padua, Padua, Italy

Received 26 April 2006; received in revised form 20 July 2006; accepted 11 August 2006Available online 22 August 2006

bstract

The efficacy of granulocyte macrophage colony-stimulating factor (GM-CSF) to enhance the immune response to hepatitis B virus vaccineas been object of several reports. We searched for randomized controlled clinical trials comparing GM-CSF given concomitantly to hepatitisvirus vaccine to vaccine given alone or with placebo. Data on rates of seroconversion (anti-HBs titers >10 IU/ml) from 13 studies (734

ubjects) produced combined estimates that favored GM-CSF as compared to controls: rate ratio after a single immunization was 1.54 [95%onfidence interval (CI), 1.04–2.27] and 1.20 (95% CI, 1.02–1.42) at the end of the vaccination cycle. Using a logistic approach a significant

ose/response effect of GM-CSF was seen. Moreover, in renal failure patients who have responded to the vaccine, GM-CSF increased anti-HBsiters. Our findings suggest that GM-CSF induced a significant effect in terms of response rate and achievement of an earlier seroconversiono the vaccine in the overall populations examined, in renal failure patients and in healthy individuals.

2006 Published by Elsevier Ltd.

vaccin

ibiid

crvohe

eywords: Granulocyte macrophage colony-stimulating factor; Hepatitis B

. Introduction

Immunization is the most effective way to prevent trans-ission of hepatitis B virus (HBV) and, hence, the devel-

pment of acute or chronic hepatitis B. Seroprotectionfter recommended schedule and routes of immunizations achieved in 90–99% of immunocompetent individuals,ut hyporesponsiveness to HBV vaccine is well recog-ized in immunocompromised people [1–3]. Predictors ofon-response include increasing age, male gender, obe-ity, tobacco smoking, alcoholism and immunocompromis-ng chronic disease [4–8]. Actually, the antibody responses lower in patients with diabetes mellitus, renal failure and

hronic liver disease, as well as in immunocompromisedatients, such as those infected with HIV.

∗ Corresponding author. Tel.: +39 045 8076266; fax: +39 045 8622239.E-mail address: crucianimario@virgilio.it (M. Cruciani).

ebjp

t

264-410X/$ – see front matter © 2006 Published by Elsevier Ltd.oi:10.1016/j.vaccine.2006.08.015

e; Meta-analysis

Strategies to improve the HBV vaccine response rate havencluded the use of higher vaccine dose or increasing num-er of doses, use of different route of administration (e.g.,ntradermal versus intramuscular administration), accelerat-ng dosing schedule and use of adjuvants such as antigenelivery systems and various immunomodulators [3,9–17].

There is growing evidence that granulocyte macrophageolony-stimulating factor (GM-CSF) enhances the immuneesponse to vaccines direct against both infectious agents andarious cancers [16,17]. GM-CSF has a variety of effectsn immune responses and coimmunization with GM-CSFas been shown to increase the antibody response and tonhance the proliferative response of T cells [18,19]. Thefficacy of GM-CSF as adjuvant to hepatitis B vaccine haseen object of several clinical trials conducted in healthy sub-

ects, patients with end-stage renal disease and HIV-infectedatients [20–44].

This study aims to systemically identify and summarizehe quality of the controlled trials available and the effects

7 accine

Gpb1

2

2

2DutlgvAct

2

iGHCilca

2

dhdsb

2

otwcfirdl

r

ma

qstm(om(a

vv(udpvcrws

2

ffae

C

3

tfetu[dapaa

3

10 M. Cruciani et al. / V

M-CSF as adjuvant to HBV vaccine (part of this study wasresented at the 45th Interscience Conference on Antimicro-ial Agents and Chemotherapy, Washington, DC, December6–19, 2005).

. Materials and methods

.1. Search strategies

The search was carried out on MEDLINE (1966–March005), EMBASE (1980–March 2005), The Cochraneatabase of Systematic Reviews (Issue 1, 2005). MeSH termssed were “Granulocyte macrophage colony-stimulating fac-or”, “vaccines”, “hepatitis B vaccine”, “adjuvants, immuno-ogic”. The syntax used for the MEDLINE searches was:ranulocyte macrophage colony-stimulating factor ANDaccines; granulocyte macrophage colony-stimulating factorND HBV vaccines; adjuvant, immunologic AND HBV vac-

ine. The computer search was supplemented by consultinghe bibliographies from the articles retrieved.

.2. Selection criteria and outcome measures

We included randomized controlled clinical trials compar-ng the efficacy of hepatitis B vaccine given with or withoutM-CSF as vaccine adjuvant. Data on rate of response toBV vaccine (anti-HBs >10 UI/l) in patients receiving GM-SF and controls were extracted. The meta-analytical pool-

ng was performed at two time-points: after the first and theast (third or fourth) vaccine dose during the immunizationycle. We also extracted data, where possible, on anti-HBsntibody titres, side effects and on hematological profile.

.3. Quality assessment

We assessed the methodology of each trial with a scaleeveloped by Jadad et al. that scores (from a low of 0 to aigh of 5) the randomization, double blinding and reports ofropouts and withdrawals [45]. Each trial was independentlycored by two of us and any areas of disagreement arbitratedy a third.

.4. Statistical analysis

A conventional meta-analysis was performed with usef the Mantel–Haenszel fixed-effects model and applyinghe DerSimonian and Laird random effects model in caseshere the heterogeneity test give a p value <0.1 [46,47]. We

alculated both the study-specific and the common, 95% con-dence intervals (CIs) by the method of Woolf [48]. We usedate ratio (RR) as measure of the effect size, and the proce-

ure to combine the 2 × 2 tables was the Mantel–Haenszelike method by Greenland and Robins [49,50].

Differences in anti-HBs antibody titers among GM-CSFecipients and controls were evaluated by the standardized

sc

25 (2007) 709–718

ean difference, which is the difference in means divided bystandard deviation [50,51].

Sensitivity analysis was performed for determining ifuantitative results differed with the exclusion of individualtudies. We used random effect meta-regression to explorehe influence of possible sources of heterogeneity on treat-

ent effect. This was done according to the study populationhealthy subjects versus dialysis patients) and the numberf GM-CSF doses (single dose versus multiple doses). Theeta-regression was done by regressing the study-level RR

dependent variable) against a study-level covariate treateds a dicotomic variable.

Finally, the influence of four explanatory (independent)ariables [GM-CSF dose, primary vaccination procedureersus booster in non-responders, HBV vaccine schedulenumber of administered doses), and renal failure] was eval-ated employing a maximum likelihood approach. GM-CSFosage and HBV vaccine schedule were quantitative, whereasrimary vaccination and renal failure were binary (yes/no)ariables. Since the outcome was a binomial variable (suc-essful immunization: yes/no), a random effects logisticegression analysis was done. The heterogeneity of the dataas addressed by stratifying for the different studies. The

oftware package Stata 9.1 was used for this task.

.5. Assessment of publication bias and heterogeneity

Graphical funnel plots were generated to visually inspector publication bias [52]. The statistical methods for detectingunnel plot asymmetry were the rank correlation tests of Beggnd Mazumdar and the regression asymmetry test of Eggert al. [52,53].

The heterogeneity of study results was assessed by theochran’s Q and by a test of inconsistency (I2) [54,55].

. Results

As shown in the flow diagram (Fig. 1), 31 poten-ially relevant clinical trials were identified and retrievedor more detailed evaluation. Of these, six studies werexcluded because reported data not relevant to study ques-ion. We also excluded six uncontrolled clinical trials eval-ating GM-CSF as an adjuvant for hepatitis B vaccination20–22,24,38,44], and five studies controlled but not ran-omized [25,27,32,35,37]. One report [34] was identifieds duplicate publications and was considered under theirrimary reference [42] Therefore, we included in the meta-nalysis data retrieved from 13 randomized clinical trials fortotal of 734 patients [23,25–31,33,39–43].

.1. Description of studies and quality assessment

Table 1 summarizes the main characteristics of includedtudies. Data were published as full-length paper or, in twoases, as abstract. Seven studies were conducted in patients

M.C

rucianietal./Vaccine25

(2007)709–718

711

Table 1Main characteristics of randomized clinical studies included in the analysis

Reference Study population (no. pts, sex); age inGM-CSF and controls

Vaccine schedule GM-CSF No. pts with HBsAb response/total (%)

After 1 dose End cycle

GM-CSF Control GM-CSF Control

Studies in healthy subjects[23] Healthy, primary vaccination (108 M); mean

age (overall) 22.7 yearsOne dose, 20 �g s.c. or i.m., or 4 �g i.d. 20–80 �g, s.c., i.m. or i.d. 11/81 (13.5) 0/27 – –

[28] Healthy, primary vaccination (108; 49 M, 59F); mean age 28.7 (40 �g)−28.4 (80 �g) and27.7

Three doses (0, 1, 6), 10 �g 80 or 250 �g i.m., at T0 and 1 mo 10/71 (14.0) 2/34 (5.8) 67/68 (98.5) 32/34 (94.1)

[36] Healthy elderly, primary vaccination (45; 20M, 25 F); mean age (overall) 74.0 years

Three doses (0, 1, 6 mo), 10 �g 80 or 250 �g i.m., at T0 and 1 mo – – 11/30 (36.6) 8/15 (53.3)

[43] Healthy, primary vaccination (54; 25 M, 29F); mean age 37.1 and 37.5 years

One dose, 40 �g 300 �g s.c. 23/27 (85.1) 11/27 (40.7) – –

[41] Healthy non-responders (60; 29 M, 31 F),mean age 42.0 and 44.4

Three doses (0,1,2 mo), 10 vs. 40 �g 125 �g i.m. at T0, 1 and 2 mo 11/30 (36.6) 13/30 (43.3) 19/30 (63.3) 16/30 (53.3)

Studies in hemodialysis and/or chronic renal failure[30] Hemodialysis, non-responders (30; 18 M, 12

F); mean age 51 (40 �g)−43 (80 �g) and 52years

One booster, 40 �g 40 or 80 �g 6/18 (33.3) 6/10 (60.0) – –

[31] Hemodialysis, primary vaccination (12; 10M, 2 F), 47.5 and 44.4 years

Three doses (0, 1, 6 mo), 40 �g s.c. 150–250 �g s.c., at T0 3/6 (50) 0/6 5/6 (83.3) 2/6 (33.3)

Hemodialysis, non-responders (16, 11 M, 5F), 38.8 and 41.4 years

One booster dose, 40 �g 150–250 �g s.c., T0 7/8 (87.5) 2/8 (25.0) – –

[33] Chronic Renal failure, primary vaccination(60; 33 M, 27 F); mean age 41 and 38 years.

Four doses (0, 1, 2, 6 mo), 20 �g 150 �g s.c. at T0, or To and 1 mo 13/24 (54) 7/20 (34) 23/24 (95.8) 18/20 (92)

[39] Chronic Renal failure or hemodialysis,primary vaccination (54; 38 M, 16 F)

Three doses (0, 1, 2 mo), 40 �g 150 �g s.c. at T0, 1 and 2 mo – – 24/26 (92) 18/28 (64)

[26] Hemodialysis, primary vaccination (15; 9 M,6 F) 38.7 and 30.5 years.

Four doses (0, 1, 2, 6 mo), 40 �g 3 �g/kg s.c., at T0 2/6 (33.3) 1/9 (11.1) 6/6 (100) 4/9 (44.4)

[42] Chronic Renal failure, primary vaccination(64; 40 M, 24 F); 45.6 and 45.8 years

Four doses (0, 1, 2, 6 mo), 40 �g 50–150 �g s.c. at T0 – – 25/28 (89.2) 20/27 (74.0)

[29] Hemodialysis (30; 19 M, 11 F); mean age42.4 and 43.2 years

Two doses (0, 1 mo), 40 �g 150 �g s.c. at T0 and 1 mo 10/15 (66.6) 5/15 (33.3) – –

Studies in HIV-1 infected patients[40] HIV-infected, primary vaccination (80, 42

M, 38 F); median age 25.9 and 28.1 yearsThree doses (0, 1, 6 mo), 40 �g 20 �g i.m. at T0 12/40 (30) 6/40 (15) 29/40 (72.5) 24/40 (60)

Abbreviations: M, male; F, female; i.m., intramuscular; s.c., subcutaneous; i.d., intradermal; T0, 24 h before first dose of vaccine or booster; mo, months.

712 M. Cruciani et al. / Vaccine

whvris

tds[aam

3

atomnd

r

i[iiatap

3r

y

ttd

iosTdsstbsi

3

s[r(a(Hr(

3

TR

GI

T

Fig. 1. Meta-analysis profile summarizing trial flow.

ith chronic renal failure and hemodialysis, five studies inealthy individuals, and one in HIV-infected patients. HBVaccine was given as primary vaccination or as booster in non-esponders. Dosages of GM-CSF ranged from 20 to 300 �g,n a single (eight studies) or multiple administration (fivetudies), given subcutaneously or intramuscularly (Table 1).

The median Jadad score for quality was 2.0 (range 1–5);he mean score was 2.2 ± 1.4. All the included studies wereescribed as randomized, but the method to generate theequence was described and appropriate in four studies only23,26,30,43]. Three studies were defined as double blind,nd one as single blind [23,28,30,36]. Only the study defineds a single blind trial clearly reported concealment of treat-ent allocation [23].

.2. Results of the meta-analysis

Due to heterogeneity, both clinical and statistical, meta-nalysis was performed with a random effects model. Theime course of the immune modulation exerted by GM-CSFn the immunization process was explored by repeating theeta-analytical pooling at two time-points during the immu-

ization cycle: after the first dose of vaccine and after the lastose (the third or fourth).

Fig. 2 shows RRs and 95% CI for individual studies foresponse rate after the first and last vaccine dose. Visual

ifi

able 2esults of the linear regression analysis

Regression coefficient Standard error

M-CSF dose 0.00483 0.001685ntercept 0.659359 0.415805

he overall significance of the model was p = 0.0042 (Wald χ2, 2 d.o.f. = 8.21).

25 (2007) 709–718

nspection of forest plots shows that, with few exceptions30,36,41], the treatment effect was consistent among stud-es. The studies produced combined estimates that favoredn a significant way GM-CSF as compared to controls. RRnd related 95% CI of successful immunization were: afterhe first vaccine dose, 1.54 (95% CI, 1.04–2.27; p = 0.03);t the end of vaccination cycle, 1.20 (95% CI, 1.02–1.41;= 0.023).

.3. Sensitivity analysis, meta-regression and logisticegression

In sensitivity analysis, the exclusion of any single studyielded only minimal change on the effect size.

In meta-regression analysis, we found that response rateso GM-CSF did not differ in hemodialysis patients comparedo healthy individuals, and in individuals receiving a singleose or multiple doses of GM-CSF.

The logistic regression demonstrated a monotonic increas-ng relationship between antibody response rate and dosagef GM-CSF, whereas vaccine schedule, primary vaccinationchedule, and renal failure did not show a significant impact.herefore, the final model had only a predictor, GM-CSFosage (Table 2). The logistic model including the differenttudies as strata confirmed the presence of heterogeneity attudy level. The between studies variance (rho) accountedo 28.2% of the total variance, the null hypothesis of rho = 0eing rejected (p < 0.0001). The predicted probability of auccessful immunization in function of the GM-CSF dosages shown in Fig. 3.

.4. Anti-HBs antibody titers

Anti-HBs antibody titers were provided in eleventudies, as mean, mean ± S.D., or geometric mean23,26,28,30,31,37,39–43]. Using the mean ± S.D. values aseported in 4 papers conducted in patients with renal failureFig. 4), the standardized mean difference of antibody titerst the end of vaccination cycle was statistically significant1.84; 95% CI, 0.97–2.71; p < 0.0001) [26,31,39,42]. Anti-Bs titers reported in these studies were higher in GM-CSF

ecipients (median 275, range 69–663) compared to controlsmedian 44, range 14–243).

.5. Other outcomes

Since there was inconsistency in measuring and in report-ng side effects related to study drugs and biochemical pro-les, these data are presented only descriptively.

z p 95% Confidece interval

2.87 0.004 0.001527 0.0081331.59 0.113 −0.1556 1.474322

M. Cruciani et al. / Vaccine 25 (2007) 709–718 713

F for ther ed at twd square

rsts[w

itl

ig. 2. Pooled rate ratio (RR) estimates and their 95% confidence intervalseceiving GM-CSF and controls. The meta-analytical pooling was performuring the immunization cycle. Studies are identified by first author. Size of

Adverse events assessed as being related, or possiblyelated to vaccine and/or GM-CSF were reported in alltudies published as full-length papers but one [27]. In

hree studies (one placebo controlled) the occurrence ofide effects was similar in GM-CSF recipients and controls30,31,41]. In the remaining eight studies adverse eventsere reported mostly with GM-CSF. The prevalence and

edba

outcomes rate of response to HBV vaccine (anti-HBs >10 UI/l) in patientso time-points: after the first (top) and after the last (bottom) vaccine dose

s is proportional to the weighted RR.

mportance of side effects varied with GM-CSF dosage,hough on average they were of mild to moderate nature, self-imiting and of short duration. No severe or life threatening

vents were reported and, in studies with multiple GM-CSFoses [28,29,36,41], no participants withdrew from the studyecause of adverse events. In the large majority of casesdverse events possibly related to GM-CSF consisted of mild

714 M. Cruciani et al. / Vaccine

Fd(i

sf

[

ottaGb

4

Gseci

omeiatGc

Ffo

ig. 3. Probability of positive response to vaccination at various GM-CSFoses. Y-axis: predicted success rate of vaccination. X-axis: GM-CSF dose�g). Broken line: success rate, with the corresponding 95% confidencenterval (C.I.) (grey area).

oreness at the injection site, fatigue, headache, nausea, andever.

Data on total leucocytes count were reported in six studies23,26,28,36,40,42]. Five studies provided mean value with-

arac

ig. 4. Meta-analytical pooling of anti-HBs antibody titers as standardized mean difrom the report of Anandh et al. [31] are stratified according to the vaccination cyclf Kher et al. [39] are stratified in three groups depending on creatinine values (fro

25 (2007) 709–718

ut standard deviation (S.D.), and one mean and S.D. Overall,he median value of total leucocytes (cells × 103 mm−3) inhese studies was 7.6 (range 5.1–15.6) in GM-CSF recipientsnd 6.7 (range 4.7–7.8) in controls, with a median increase inM-CSF recipients of 1.1 (range −0.1 to 7.8), which cannote regarded as clinically significant.

. Discussion

Early reports have found a very promising effect ofM-CSF as an adjuvant to HBV vaccine [20,22,23]. Sub-

equently, the efficacy of this approach has been widelyxplored, and a number of controlled and uncontrolled clini-al trials have evaluated the adjuvancy properties of GM-CSFn combination with HBV vaccine [24–44].

The vaccine adjuvant properties of GM-CSF are basedn a variety of effects on immune responses, which includeacrophage activation, increasing MHC class II antigen

xpression, enhancing cell maturation and migration, enhanc-ng memory cell generation via T and B cell activation,nd inducing localized inflammation at the site of injec-ion [16–19,56]. However, the exact mechanism by whichM-CSF may improve the immune response to HBV vac-

ination is unclear and deserves further research. Of note,

study investigating the effect of GM-CSF in primary non-

esponding hemodialysis patients has shown a decrease in thentigen presenting capacity of peripheral blood mononuclearells and in the number of circulating dendritic cells [44].

ference. Data from four reports in renal failure patients [26,31,34,39]. Datae (primary vaccination and boosts in non-responders). Data from the reportm 2–4 to 8 mg/dl).

accine

raed

GeC(

nhpootpsr(mbtttsumiifiy

fiAainauuraisrhitt

hiul

mhvveHPwatorc6sDau

vtlrwavw

eaqoutsbo

mlTimam

arL(dr

M. Cruciani et al. / V

In this meta-analysis, we have included data from 13andomized studies in 734 subjects evaluating GM-CSF asdjuvant to HBV vaccine. Since there was a substantial het-rogeneity among studies, both clinical and statistical, a ran-om effect model was applied.

The studies produced combined estimates that favoredM-CSF as compared to controls. The pooled rate ratio

ffect estimate after a single immunization was 1.54 (95%I, 1.04–2.27), and at the end of the vaccination cycle 1.20

95% CI, 1.02–1.41).The trials analyzed were conducted in hemodialysis or

on-hemodialysis patients with chronic renal failure, inealthy individuals and, in one study, in HIV-1 infectedatients. Since chronic renal failure patients have a high riskf contracting hepatitis B infection, HBV vaccination is rec-mmended for susceptible hemodialysis patients [3]. Unfor-unately, seroconversion rates are much lower in renal failureatients compared to healthy people. Early studies usingtandard doses and schedules of both plasma-derived andecombinant yeast vaccines showed poor antibody response∼50%) in hemodialysis patients [57]. Patients with ure-ia who were vaccinated before they required dialysis have

een shown to have higher seroconversion rates and antibodyiters [58]. In these patients the major objective of vaccina-ion is to obtain a rapid, adequate and sustained seroprotec-ive level of anti-HBs antibodies. Not surprisingly, severaltudies have evaluated GM-CSF as adjuvant in renal fail-re patients and seven of these studies were included in theeta-analysis [26,29–31,33,39,42]. Regardless to the admin-

stration of GM-CSF, rates of response to the vaccine variedn the included studies according to the degree of renal insuf-ciency and were lower, as expected [39,42,58], in hemodial-sis patients compared to non-dialysis patients.

In the present meta-analysis, we also included data fromve studies conducted in healthy adults [23,28,36,41,43].fter the pilot trial by Tarr et al. [23], the use GM-CSF

s adjuvant to HBV vaccine in healthy individuals stud-es was based on different motivations, which included: theeed to produce an earlier seroconversion rate in individu-ls at immediate risk of HBV infection (health care workers,nprotected sexual partners of seropositive persons) or in vol-ntary kidney donors; the attempt to induce seroprotectionesponse in healthy non-responders to primary vaccination;nd the attempt to reverse the age-related defect in primarymmune response in elderly individual. Of the five includedtudies conducted in healthy subjects, one provided responseate after a single vaccine dose [23], one was conducted inealthy elderly undergoing primary vaccination [36], and onen healthy non-responders [41]. Thus, it is not surprising thathe response rates in controls from these studies are lowerhat those expected in the general population.

As for renal failure patients, persons with HIV-1 infection

ave a high risk of acquiring HBV infection and show anmpaired response to HBV vaccine [3]. Moreover, individ-als coinfected with HIV-1 and HBV, especially those withow CD4 nadir counts, are at increased risk for liver-related

ali2

25 (2007) 709–718 715

ortality [59]. Therefore, the importance of prevention ofepatitis B in people with HIV infection is of absolute rele-ance, and strategies to produce rapid and increased serocon-ersion rate are highly desirable. Recently, Das Gracas Sasakit al. have evaluated the efficacy of GM-CSF as adjuvant forBV vaccine in a cohort of 80 HIV infected individuals [40].atients were randomized to receive GM-CSF or placeboith the first 40 �g dose of HBV vaccine, followed by two

dditional doses of vaccine at 1 and 6 months. The results ofhis study show that GM-CSF increases the immunogenicityf HBV vaccine in HIV infected individuals, both in terms ofate of response (62% versus 30% 1 month after the last vac-ine dose) and average of antibody titers (210 mIU/ml versus44 mIU/ml). Basing on these favorable results, a new studyponsored by the National Institute of Allergy and Infectiousiseases evaluating the safety and effectiveness of GM-CSF

s an adjuvant for HBV vaccination in HIV infected individ-als has been designed [60].

Regardless to the study population, response rate in indi-idual studies shows that, with few exceptions [30,36,41], thereatment effect was consistent among studies. Our cumu-ative data show that the rise in response rate in GM-CSFecipients was significant after the first dose of vaccine asell as at the end of vaccination cycle. Using meta-regression

nalysis we have shown that the effect of GM-CSF on HBVaccine response was significant in hemodialysis patients asell as in healthy individuals.Some limitations of this meta-analysis need to be acknowl-

dged. As with all meta-analyses, our conclusions can be asccurate as the trials upon which they are based. Based onuality scores, the methodological quality of the studies wasn average not ideal, though varied considerably in individ-al studies. Among the 13 randomized studies included inhis analysis, only 4 described methods of randomization; 3tudies were defined as double blind and one as single blind,ut concealment of treatment allocation was clearly reportednly in one study.

Publication bias is a significant threat to the validity ofeta-analysis. In the present meta-analysis, evidence of pub-

ication bias with statistical methods was not detectable.here was, however, evidence of heterogeneity among stud-

es. Because in the presence of heterogeneity a fixed-effectsodel yields artificially narrow confidence intervals, we usedrandom-effects model, which incorporates a moment esti-ator of the between-trial components of variance.There was also clinical heterogeneity in the trials

nalysed. Of note is that the studies employed differentecombinant human GM-CSF (Leucomax®, Lucomax®,eukine®, Sargramostim®) and HBV vaccine preparations

Recombivax®, Engerix B®, Enivac-HB®), at differentoses and number of administration, and by differentoutes. Moreover, the time intervals at which antibodies

ssessments were made varied among studies. There was aarge heterogeneity in the dosage of GM-CSF in the studiesncluded in the analysis. The GM-CSF dosages ranged from0 to 300 �g, and these differences have been described as

7 accine

aihGw

sGitmstps[1irfv

uacm

aCMtmn

ccGtptte

A

ton

R

[

[

[

[

[

[

[

[

[

[

[

[

16 M. Cruciani et al. / V

possible explanation for discrepancies of results obtainedn different studies. Actually, using logistic regression weave shown that the treatment effect was dependent onM-CSF dose. Predicted response rate strongly increasedith increase in GM-CSF dosage.Data from studies conducted in hemodialysis patients

how that in individuals who have responded to the vaccineM-CSF increased the anti-HBs titers. In fact, the standard-

zed mean difference of antibody titers at the end of vaccina-ion cycle was statistically significant Since the extent of the

aximal antibody response has been correlated with the per-istence of protective antibody over time, it may be supposedhat GM-CSF provides a lasting protection in renal failureatients responding to immunization [61,62]. This hypothe-is has been confirmed by Kher et al. in a follow-up study39]. In this study antibody titers in renal failure patients

year after the last vaccine dose were significantly highern GM-CSF recipients compared to controls. Anti-HBs titeranged from 71 to 203 mIU/ml in GM-CSF group androm 19 to 21 mIU/ml in controls, depending on creatininealues.

Within the short period of administration and the dosessed, GM-CSF was well tolerated. The low incidence ofdverse events in the analyzed studies is in agreement withlinical experience of low dose of GM-CSF in patients withyelodisplastic syndrome or aplastic anemia [63,64].Among the key goal of HBV vaccination is to obtain rapid

nd sustained anti-HBs titers. Our findings suggest that GM-SF hasten the immunization process in a significant way.oreover, in hemodialysis patients who have responded to

he vaccine, GM-CSF produced higher anti-HBs titers, whichay translate in a significant longer lasting protection and less

eed of boosts.From this systematic review and meta-analysis we con-

lude that the adjuvant effect of GM-CSF to hepatitis B vac-ine appears to be biologically confirmed. In addition, higherM-CSF dosage would increase the probability of response

o HBV vaccine. Further well-designed studies with an appro-riate follow-up are needed to establish the optimal dose andhe number of GM-CSF injections, the cost-effectiveness ofhis intervention, and the duration of protective antibody lev-ls over time with this adjuvant.

cknowledgments

We would like to thank Dr. Vijay Kher for providing addi-ional data from his study for our analysis. Potential conflictsf interest: All authors—no conflicts. Financial support:one.

eferences

[1] Egea E, Iglesias A, Salazar M, Morimoto C, Kruskall MS, Awdeh Z,et al. The cellular basis for lack of antibody response to hepatitis Bvaccine in humans. J Exp Med 1991;173(3):531–8.

[

25 (2007) 709–718

[2] Alper CA, Kruskall MS, Marcus-Bagley D, Craven DE, Katz AJ, BrinkSJ, et al. Genetic prediction of nonresponse to hepatitis B vaccine. NEngl J Med 1989;321(11):708–12.

[3] Recommendations of the Advisory Committee on Immunization Prac-tices (ACIP): use of vaccines and immune globulins for persons withaltered immunocompetence. MMWR Recomm Rep 1993 Apr 9; 42(RR-4):1–18.

[4] Fisman DN, Agrawal D, Leder K. The effect of age on immunologicresponse to recombinant hepatitis B vaccine: a meta-analysis. ClinInfect Dis 2002;35(11):1368–75.

[5] Esser MT, Marchese RD, Kierstead LS, Tussey LG, Wang F, ChirmuleN, et al. Memory T cells and vaccines. Vaccine 2003;21(5–6):419–30.

[6] Wiedmann M, Liebert UG, Oesen U, Porst H, Wiese M, Schroeder S,et al. Decreased immunogenicity of recombinant hepatitis B vaccine inchronic hepatitis C. Hepatology 2000;31(1):230–4.

[7] Huang CC. Hepatitis in patients with end-stage renal disease. J Gas-troenterol Hepatol 1997;12(9–10):S236–41.

[8] Arslanoglu I, Cetin B, Isguven P, Karavus M. Anti-HBs response tostandard hepatitis B vaccination in children and adolescents with dia-betes mellitus. J Pediatr Endocrinol Metab 2002;15(4):389–95.

[9] Mitwalli A. Responsiveness to hepatitis B vaccine in immuno-compromised patients by doubling the dose scheduling. Nephron1996;73(3):417–20.

10] Rey D, Krantz V, Partisani M, Schmitt MP, Meyer P, Libbrecht E, etal. Increasing the number of hepatitis B vaccine injections augmentsanti-HBs response rate in HIV-infected patients. Effects on HIV-1 viralload. Vaccine 2000;18(3):1161–5.

11] Rawer P, Willems WR, Breidenbach T, Guttmann W, Pabst W, Schut-terle G. Seroconversion rate, hepatitis B vaccination, hemodialysis, andzinc supplementation. Kidney Int Suppl 1987;22:149–52.

12] Quiroga JA, Castillo I, Porres JC, Casado S, Saez F, Gracia Martinez M,et al. Recombinant gamma-interferon as adjuvant to hepatitis B vaccinein hemodialysis patients. Hepatology 1990;12(4 Pt 1):661–3.

13] Donati D, Gastaldi L. Controlled trial of thymopentin in hemodialy-sis patients who fail to respond to hepatitis B vaccination. Nephron1988;50(2):133–6.

14] Jungers P, Devillier P, Salomon H, Cerisier JE, Courouce AM.Randomised placebo-controlled trial of recombinant interleukin-2 inchronic uraemic patients who are non-responders to hepatitis B vac-cine. Lancet 1994;344(8926):856–7.

15] Bryan JP, Sjogren MH, Perine PL, Legters LJ. Low-dose intrader-mal and intramuscular vaccination against hepatitis B. Clin Infect Dis1992;14(3):697–707.

16] Disis ML, Bernhard H, Shiota FM, Hand SL, Gralow JR, HusebyES, et al. Granulocyte-macrophage colony-stimulating factor: aneffective adjuvant for protein and peptide-based vaccines. Blood1996;88(1):202–10.

17] Taglietti M. Related Vaccine adjuvancy: a new potential area of devel-opment for GM-CSF. Adv Exp Med Biol 1995;378:565–9.

18] Morrissey PJ, Bressler L, Park LS, Alpert A, Gillis S. Granulocyte-macrophage colony-stimulating factor augments the primary antibodyresponse by enhancing the function of antigen-presenting cells. JImmunol 1987;139(4):1113–9.

19] Okamura S, Tanaka T, Yamaga S, Omori F, Niho Y. The effects ofrecombinant human granulocyte-macrophage colony-stimulating fac-tor on the induction of lymphokine-activated killer cells in vitro. Int JImmunopharmacol 1991;13(5):587–93.

20] Jikki PN, Anuradha G, Narayen, Joshi N. The efficacy of granulocytemacrophage-colony stimulating factor (GM-CSF) as adjuvant to hep-atitis B vaccine in patients of CRF. Ind J Nephrol 1998;8:134–5.

21] Martin J, Bosch O, Moraleda G, Bartolome J, Quiroga JA, CarrenoV. Pilot study of recombinant human granulocyte-macrophage colony-

stimulating factor in the treatment of chronic hepatitis B. Hepatology1993;18(4):775–80.

22] Hess G, Kreiter F, Kosters W, Deusch K. The effect of granulocyte-macrophage colony-stimulating factor (GM-CSF) on hepatitis B vac-cination in haemodialysis patients. J Viral Hepat 1996;3(3):149–53.

accine

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

M. Cruciani et al. / V

23] Tarr PE, Lin R, Mueller EA, Kovarik JM, Guillaume M, Jones TC.Evaluation of tolerability and antibody response after recombinanthuman granulocyte-macrophage colony-stimulating factor (rhGM-CSF) and a single dose of recombinant hepatitis B vaccine. Vaccine1996;14(13):1199–204.

24] Carlsson T, Struve J. Granulocyte-macrophage colony-stimulating fac-tor given as an adjuvant to persons not responding to hepatitis B vaccine.Infection 1997;25(2):129.

25] Anandh U, Dhanraj P, Nayyar V, Ballal HS. Granulocyte macrophagecolony stimulating factors(GM-CSF) as adjuvant to hepatitis-B vacci-nation in hemodialysis patients: a preliminary report. Ind J Nephrol1997;7(3):109–11.

26] Kapoor D, Aggarwal SR, Singh NP, Thakur V, Sarin SK. Granulocyte-macrophage colony-stimulating factor enhances the efficacy of hepatitisB virus vaccine in previously unvaccinated haemodialysis patients. JViral Hepat 1999;6(5):405–9.

27] Sudhagar K, Chandrasekar S, Rao MS, Ravichandran R. Effectof granulocyte macrophage colony stimulating factor on hepatitis-B vaccination in haemodialysis patients. J Assoc Physicians India1999;47(6):602–4.

28] Hasan MS, Agosti JM, Reynolds KK, Tanzman E, Treanor JJ, EvansTG. Granulocyte macrophage colony-stimulating factor as an adju-vant for hepatitis B vaccination of healthy adults. J Infect Dis1999;180(6):2023–6.

29] Malhotra KK, Saxena S, Mahajan M, Mishra P, Tewary G, Hazarika A.Role of granulocyte macrophage colony stimulating factors in hepatitis-B vaccination in haemodialysis patients. A prospective trial. Ind JNephrol 1999;9(2):38–40.

30] Evans TG, Schiff M, Graves B, Agosti J, Barritt ML, Garner D, etal. The safety and efficacy of GM-CSF as an adjuvant in hepatitis Bvaccination of chronic hemodialysis patients who have failed primaryvaccination. Clin Nephrol 2000;54(2):138–42.

31] Anandh U, Bastani B, Ballal S. Granulocyte-macrophage colony-stimulating factor as an adjuvant to hepatitis B vaccination in main-tenance hemodialysis patients. Am J Nephrol 2000;20(1):53–6.

32] Varma PP, Mahapatra CC. Granulocyte macrophage colony stimulat-ing factor (GMCSF) – an adjuvant to hepatitis B vaccine in ESRDpatients – how many doses. 32nd Annual Conference of Indian Soci-ety of Nephrology, September 2001. Ind J Nephrol 2000;10:101. Abst.CNO 10.

33] Agarwal SK, Irshad M, Dash SC, Pandey RM. Prospective randomizedtrial of GMCSF as adjuvant for hepatitis B vaccination in CRF. 32ndAnnual Conference of Indian Society of Nephrology, September 2001.Ind J Nephrol 2000;10:102. Abst. CNO 11.

34] Singh NP, Mandal SK, Thakur A, Anuradha S, Prakash A, AgarwalSK. Efficacy of GM-CSF (in varying doses) as adjuvant to hepatitis Bvaccination in CRF patients. 32nd Annual Conference of Indian Societyof Nephrology, September 2001. Ind J Nephrol 2000;10:103–4. Abst.CNO 12.

35] Prakash R, Gupta A, Sharma RK, Gulati S, Sharma AP. Role of GM-CSF as adjuvant to hepatitis B vaccine in CRF patients. 32nd AnnualConference of Indian Society of Nephrology, September 2001. Ind JNephrol 2000;10:118. Abst. CNO 15.

36] Looney RJ, Hasan MS, Coffin D, Campbell D, Falsey AR, Kolassa J,et al. Hepatitis B immunization of healthy elderly adults: relationshipbetween naive CD4+ T cells and primary immune response and evalu-ation of GM-CSF as an adjuvant. J Clin Immunol 2001;21(1):30–6.

37] Jha R, Lakhtakia S, Jaleel MA, Narayan G, Hemlatha K. Granulo-cyte macrophage colony stimulating factor (GM-CSF) induced sero-protection in end stage renal failure patients to hepatitis B in vaccinenon-responders. Ren Fail 2001;23(5):629–36.

38] Wang J, Zhu Q, Zhang T, Yu H. A pilot study on the combined therapy

of granulocyte-macrophage colony-stimulating factor and hepatitis Bvaccine on chronic hepatitis B virus carrier children. Chin Med J (Engl)2002;115(12):1824–8.

39] Kher V, Aggarwal M, Singhal MK. Potentiation of seroconversion ofhepatitis B vaccination with GM-CSF in CRF. A randomized controlled

[

25 (2007) 709–718 717

trial—follow up data. Abs. 33rd Annual Conference of Indian Societyof Nephrology, November 2002. Ind J Nephrol 2002;12(4):175–6.

40] Das Gracas Sasaki M, Foccacia R, de Messias-Reason IJ. Efficacyof granulocyte-macrophage colony-stimulating factor (GM-CSF) as avaccine adjuvant for hepatitis B virus in patients with HIV infection.Vaccine 2003;21(31):4545–9.

41] Kim MJ, Nafziger AN, Harro CD, Keyserling HL, Ramsey KM, Dru-sano GL, et al. Revaccination of healthy nonresponders with hep-atitis B vaccine and prediction of seroprotection response. Vaccine2003;21(11–12):1174–9.

42] Singh NP, Mandal SK, Thakur A, Kapoor D, Anuradha S, Prakash A,et al. Efficacy of GM-CSF as an adjuvant to hepatitis B vaccination inpatients with chronic renal failure—results of a prospective, random-ized trial. Ren Fail 2003;25(2):255–66.

43] Krishnamurthy K, John GT, Abraham P, Jacob CK. Granulocytemacrophage colony stimulating factor augmented hepatitis B vaccineprotocol for rapid seroprotection in voluntary kidney donors. Ind J MedRes 2004;119(4):162–4.

44] Verkade MA, van de Wetering J, Klepper M, Vaessen LM, WeimarW, Betjes MG. Peripheral blood dendritic cells and GM-CSF as anadjuvant for hepatitis B vaccination in hemodialysis patients. KidneyInt 2004;66(2):614–21.

45] Jadad AR, Moore RA, Carroll D, Jenkinson C, Reynolds DJ, GavaghanDJ, et al. Assessing the quality of reports of randomized clinical trials:is blinding necessary? Controlled Clin Trials 1996;17(1):1–12.

46] Mantel N, Haenszel W. Statistical aspects of the analysis of data fromretrospective studies of disease. J Natl Cancer Inst 1959;22(4):719–48.

47] DerSimonian R, Laird N. Meta-analysis in clinical trials. ControlledClin Trials 1986;7(3):177–88.

48] Woolf B. On estimating the relation between blood group and disease.Ann Human Genet 1955;19(4):251–3.

49] Greenland S, Robins JM. Estimation of a common effect parameterfrom sparse follow-up data. Biometrics 1985;41(1):55–68.

50] Deeks JJ, Altman DG, Bradburn MJ. Statistical methods for examiningheterogeneity and combining results from several studies in meta-analysis. In: Egger M, Davey Smith G, Altman DG, editors. Systematicreviews in health care: meta-analysis in context. 2nd ed. BMJ Books;2001. p. 285–312.

51] Rosenthal R. Parametric measures of the effect size. In: Cooper H,Hedges LV, editors. The handbook of research synthesis. New York:Russell Sage Foundation; 1994.

52] Egger M, Davey Smith G, Schneider M, Minder C. Biasin meta-analysis detected by a simple, graphical test. BMJ1997;315(7109):629–34.

53] Begg CB, Mazumdar M. Operating characteristics of a rank correlationtest for publication bias. Biometrics 1994;50(4):1088–111.

54] Cochran WG. The combination of estimates from different experi-ments. Biometrics 1954;10:101–29.

55] Higgins J, Thompson S, Deeks J, Altman D. Measuring inconsistencyin meta-analyses. BMJ 2003;327(7414):557–60.

56] Lin R, Tarr PE, Jones TC. Present status of the use of cytokines as adju-vants with vaccines to protect against infectious diseases. Clin InfectDis 1995;21(6):1439–49.

57] Stevens CE, Alter HJ, Taylor PE, Zang EA, Harley EJ, Szmuness W.Hepatitis B vaccine in patients receiving hemodialysis. Immunogenic-ity and efficacy. N Engl J Med 1984;311(8):496–501.

58] Seaworth B, Drucker J, Starling J, Drucker R, Stevens C, HamiltonJ. Hepatitis B vaccines in patients with chronic renal failure beforedialysis. J Infect Dis 1988;157(2):332–7.

59] Thio CL, Seaberg EC, Skolasky Jr R, Phair J, Visscher B, Munoz A, etal. Multicenter AIDS Cohort Study. HIV-1, hepatitis B virus, and risk ofliver-related mortality in the Multicenter Cohort Study (MACS). Lancet

2002;360(9349):1921–6.

60] Safety and effectiveness of an adjuvant in improving immune responseto hepatitis B virus vaccine in HIV infected individuals. Clini-cal trials.gov Identifier: NCT00272493. Available at: http://www.clinicaltrials.gov/ct/show/NCT00272493. Accessed 24 January 2006.

7 accine

[

[

[

18 M. Cruciani et al. / V

61] Hadler SC, Francis DP, Maynard JE, Thompson SE, Judson FN,Echenberg DF, et al. Long-term immunogenicity and efficacy of hep-

atitis B vaccine in homosexual men. N Engl J Med 1986;315(4):209–14.

62] Magnani G, Calzetti C, Campari M, Lehndorff H, Pizzaferri P, RossiE. Immune response to hepatitis B vaccine and duration of protectionin a dialysis unit. Acta Biomed Ateneo Parmense 1987;58(1–2):41–7.

[

25 (2007) 709–718

63] Estey EH, Kurzrock R, Talpaz M, McCredie KB, O’Brien S, KantarjianHM, et al. Effects of low doses of recombinant human granulocyte-

macrophage colony stimulating factor (GM-CSF) in patients withmyelodysplastic syndromes. Br J Haematol 1991;77(3):291–5.

64] Kurzrock R, Talpaz M, Gutterman JU. Very low doses of GM-CSFadministered alone or with erythropoietin in aplastic anemia. Am JMed 1992;93(1):41–8.