Sie sind auf Seite 1von 10

Hematol Oncol Stem Cell Ther (2017) 10, 116– 125

Available at


journal homepage:


Hydroxyurea for nontransfusion-dependent

b-thalassemia: A systematic review and
Ali H. Algiraigri a,b,c,*, Nicola A.M. Wright d, Elizabeth Oddone Paolucci c,
Aliya Kassam c

Department of Hematology, King Abdulaziz University Hospital, Jeddah, Saudi Arabia
King Faisal Special Hospital and Research Center, Jeddah, Saudi Arabia
Department of Community Health Science, University of Calgary, Calgary, Alberta, Canada
Department of Pediatrics, Alberta Children’s Hospital, Calgary, Alberta, Canada

Received 20 December 2016; accepted 15 February 2017

Available online 6 April 2017

Non-transfusion dependent Objective/background: Nontransfusion-dependent b-thalassemia (NTDbT) syndromes consist
of b-thalassemia intermedia and moderate hemoglobin E/b thalassemias. They are character-
b-thalassemia intermedia;
ized by varying degrees of chronic anemia and a wide spectrum of complications due to inef-
Meta-analysis; fective erythropoiesis and iron overload from chronic transfusions. Hydroxyurea (HU), an oral
Blood transfusion chemotherapeutic drug, is anticipated to decrease disease severity.
Methods: We performed a meta-analysis to evaluate the clinical efficacy and safety of HU in
NTDbT patients of any age. MEDLINE, EMBASE, Cochrane databases, and major conference pro-
ceedings for studies that assessed HU in NTDbT patients were searched. Qualities of eligible
studies were assessed by National Institutes of Health tools.
Results: Seventeen studies, collectively involving 709 patients, fulfilled the eligibility criteria.
HU was associated with a significant decrease in transfusion need in severe NTDbT with com-
plete and overall (50%) response rates of 42% and 79%, respectively. For mild NTDbT, HU
was effective in raising hemoglobin by 1 g/L in 64% of patients.
Conclusion: HU appears to be effective, well tolerated, and associated with mild and transient
adverse events. NTDbT patients may benefit from a trial of HU, although large randomized

* Corresponding author at: Wadi An Nakhl, AL Rawdah, Jeddah 23433, Saudi Arabia.
E-mail address: (A.H. Algiraigri).
1658-3876/Ó 2017 King Faisal Specialist Hospital & Research Centre. Published by Elsevier Ltd.
This is an open access article under the CC BY-NC-ND license (
Hydroxyurea for b-thalassemia intermedia 117

clinical trials assessing its efficacy should be conducted to confirm the findings of this meta-
analysis and to assess its long-term toxicity and response sustainability.
Ó 2017 King Faisal Specialist Hospital & Research Centre. Published by Elsevier Ltd. This is an
open access article under the CC BY-NC-ND license (

Introduction Methods

One of the most common inherited diseases worldwide is Data sources and searches
b-thalassemia, characterized by a reduced ability to pro-
duce hemoglobin (Hb) [1]. Although b-thalassemia is com- A comprehensive systematic search of the literature was
mon among people originating from the Mediterranean, conducted to evaluate the clinical efficacy and safety of
Middle East, Central Asia, India, and Southern China, it HU in patients with NTDbT. MEDLINE (1946–September
is no longer limited to these geographical areas due to 2016), EMBASE (1974–September 2016), and Cochrane Cen-
migration to different regions of the world [2]. Despite tral Register of Controlled Trials (September 2016) were
the complexity of b-thalassemia genotypes (more than searched using the following keywords: ‘‘hydroxyurea,”
200 different mutations of b-globin genes), two broad ‘‘hydroxycarbamide,” ‘‘hydrea,” or ‘‘droxia” and ‘‘tha-
phenotypes of clinically significant b-thalassemia are rec- lassemia.” Clinical trial registries for the last 5 years were
ognized based on the Hb steady state and the need for searched, including, World Health Organi-
blood transfusions: transfusion-dependent b-thalassemia zation International Clinical Trials Registry Platform, and
(TDbT) and nontransfusion-dependent b-thalassemia major conference proceedings (American Society of Hema-
(NTDbT). tology and European Hematology Association). Hand
NTDbT encompasses b-thalassemia intermedia and mod- searches were also conducted using reference lists from pri-
erate E/b-thalassemia [3]. Although NTDbT constitutes the mary studies. Searches were not restricted by language,
moderate form of b-thalassemia and its name may falsely publication date, or publication type, but for human partic-
reassure, it represents a wide spectrum of severity where ipants only.
at the one end there are patients suffering from severe Preferred Reporting Items for Systematic reviews and
symptoms, similar yet milder than TDbT patients; however, Meta-analyses and Meta-analysis of Observational Studies
at the other end, individuals are less affected but more sev- in Epidemiology guidelines for meta-analyses were used as
ere than the minor form [1]. While patients with NTDbT are the basis of our method [5,6]. The meta-analysis protocol
usually not dependent on blood transfusion for their entire for this study was registered with the Prospective Register
lives, they have a wide range of transfusion requirements of Systematic Reviews (registration number:
and suffer from a variety of complications [3]. Such compli- CRD42014010138).
cations are typically related to profound anemia, ineffec-
tive erythropoiesis, and transfusion-related implications Study selection
[3]. In addition, other specific complications have been
ascribed to NTDbT and include pulmonary hypertension One reviewer (A.A.) screened the citations, first by title
and thrombosis [3]. and abstract, and then by a review of the complete article
The common goal in NTDbT management is to correct as indicated. Randomized controlled trials (RCTs) and
the anemia, and allow for normal growth and development observational studies (sample size 10) that assessed the
[3]. To date, blood transfusion is the mainstay of care for clinical efficacy and/or safety of HU alone, for 3 months
patients with the severe form of NTDbT; however, transfu- or longer, in patients with NTDbT of any age were eligible
sions come with the risk of blood-borne infections, iron for inclusion. Exclusion criteria included b-thalassemia
overload, and adverse reactions that could become a major major, severe E/b-thalassemia, any combination therapy
source of morbidity and mortality. As a result, different with HU, and case reports, case series, or studies with a
drugs have been investigated in order to reduce transfusion sample size of <10 patients. If a study included NTDbT
requirements [4]. Hydroxyurea (HU), a fetal Hb inducer that and TDbT patients treated with HU, NTDbT patients were
is expected to improve chronic anemia and decrease the included but remaining patients were excluded. If a study
need for blood transfusions by reducing a/b-globin chain compared two or more treatment arms, the HU-only arm
imbalance, has shown potential [4]. While the current was included. In addition, if a study compared two or more
NTDbT clinical guidelines indorsed HU as a treatment fixed doses of HU in different arms, the most effective arm
option, it restricted the use of HU to certain clinical compli- was selected, as this mimics clinical practice where dose is
cations [3]. Furthermore, there have been no comprehen- titrated to reach the most effective maximum tolerated
sive systematic reviews or meta-analyses that address the dose.
role of HU in patients with NTDbT. Thus, an in-depth assess- While transfusion dependency (TD) may represent a
ment of the efficacy and safety of HU in NTDbT is common feature for both NTDbT and TDbT, this can lead
warranted. to ambiguity about the type of b-thalassemia. As such, we
118 A.H. Algiraigri et al.

further categorized TD b-thalassemia to either lifelong Heterogeneity and publication bias

TDbT or NTDbT using the primary authors’ definition/labels
of the disease. In cases with no clear distinction in the pri- Heterogeneity between studies was assessed by visual
mary study (i.e., only provided as TD b-thalassemia), we inspection of forest plots to detect overlapping 95% CIs,
asked the primary authors to clarify this issue. If that was by a formal statistical test of the significance of the hetero-
not possible, those patients/studies were excluded. geneity (chi-squared test; p < 0.05) and by estimation of the
To limit the expected clinical heterogeneity of NTDbT, percentage of heterogeneity between trials that could not
the disease was subcategorized into two groups: mild (no be attributed to sampling error (I2) [10]. The I2 is expressed
or <4 transfusions per year) and severe (4 transfusions as a percentage, with I2 values of 25% indicating low, 50%
per year). indicating moderate, and 75% indicating high heterogeneity
Data abstraction and quality assessment Publication bias was assessed graphically using funnel
plots and the Egger test, which quantified the asymmetry
Two reviewers (A.A. and A.K.) extracted the data indepen- of the plot [11,12]. The latter tests the null hypothesis that
dently, using data extraction forms for each type of NTDbT. small studies give the same results as large studies. An alpha
Key characteristics were extracted from eligible studies and level of p < 0.05 was deemed significant for the statistical
recorded. These characteristics included author, year and calculations in this meta-analysis. All data were analyzed
country of publication, study design, sample size, age of using Stata, version 13.0 [13].
sample, blood transfusion history, HU dose, response rate
(RR), types of adverse events (AEs), and follow-up duration.
Information was requested from primary authors when it
was not available in the published papers.
The quality of the included studies was assessed indepen-
Study selection
dently by two reviewers (A.A. and A.K.) using the National
Institutes of Health (NIH) quality assessment tool for con- The initial literature search yielded 943 references, after
trolled intervention studies as well as the NIH tool for duplicates were removed, and was updated to April 2015.
before–after (pre–post) studies with no control group [7]. One author (A.A.) then screened the titles and abstracts,
These tools are composed of multiple questions assessing and 882 were excluded for not meeting eligibility criteria.
the potential risk for selection bias, information bias, mea- The remaining 61 references were assessed on the basis of
surement bias, and confounding. Reviewers and then used their full text for inclusion or exclusion using the criteria
these questions to judge each study to be of ‘‘good,” indicated above. Of these, 36 studies were excluded, with
‘‘fair,” or ‘‘poor” quality. Disagreements between review- the most common reason being inadequately reported out-
ers were resolved by discussion until consensus was come data. Seventeen studies met our inclusion criteria
reached. for the meta-analysis; see Fig. 1 for Preferred Reporting
Items for Systematic reviews and Meta-analyses study flow
Data synthesis and analysis diagram. From the pool of 17 studies available for our
meta-analysis, nine included patients with severe NTDbT
and 10 included those with mild NTDbT.
The effect size of our meta-analysis was the proportion of
patients who responded to HU by either decreasing transfu-
sion needs or raising Hb in severe and mild types, respec- Study characteristics
tively. This proportion was called the RR. In severe NTDbT
patients who had a decrease in transfusion needs, complete Severe NTDbT
response rate (CRR) was defined as complete cessation of All nine studies were published over the last decade
blood transfusion post-HU therapy, and overall response (between 2005 and 2014) and conducted in seven countries:
rate (ORR) was defined as 50% reduction in transfusion three in Egypt; two in Iran; one each in India and Pakistan;
needs post-HU therapy. For mild NTDbT, the RR was defined and one multinational study (Canada, UK, and USA) [14–22].
as a 1 g/L increase in Hb. All studies were single arm (pre–post design) with no con-
Since most of the included studies were of single-arm trol group except one RCT [15]. In the single-arm studies,
designs with no control arms, pooled estimates of the treat- the prestudy arm used a historical control and the poststudy
ment effect for each outcome (CRR and ORR for severe arm was the treatment group where HU was given. For the
NTDbT; RR for mild NTDbT) across the studies were calcu- RCT, HU was compared with HU and erythropoietin, so the
lated as proportions (the responders overtreated sample HU-only arm was included [15]. All included studies were
size with HU), together with their 95% confidence intervals published in English and as full-text papers, except for
(95% CIs). These analyses were performed using a recently one where only the abstract was in English [20]. The studies
published ‘‘metaprop” Stata command developed for the collectively enrolled 416 patients. Sample size ranged from
use of proportions as effect sizes [8]. Owing to the 20 to 120. Study populations were a mixture of children and
expected heterogeneity within and between studies, we adults in all studies. The mean or median duration of the
used the random effect model, which takes into considera- studies ranged from 6 months to 50 months. Study charac-
tion between- and within-study variations and provides a teristics are summarized in Table 1.
more conservative analysis of the studies than the fixed All the studies enrolled b-thalassemia intermedia
effect model [9]. patients except one, which enrolled E/b-thalassemia
Hydroxyurea for b-thalassemia intermedia 119

Identified records from the databases ASH and EHA conference abstracts
and bibliographies of relevant articles and bibliographies of relevant articles
(n = 1,210) (n = 11)

Records after excluding duplicates (n = 943)

Records screened (n = 943)

Records excluded according to selection criteria (n = 882)

Full-text articles assessed for eligibility (n = 62)

Full-text articles excluded (n = 37)

Abstract published later as full-text article (n = 3)
Combination therapy (n = 3)
Editorials (n = 2)
Case series (n = 2)
Not enough data to calculate outcome (n =13)
Review (n = 2)
Sample size < 10 (n = 8)
Others (n = 4)

Studies included in qualitative synthesis (n = 25)

Studies included in meta-analysis (n = 17)

Fig. 1 PRISMA flow diagram of literature search. Note. ASH = American Society of Hematology; EHA = European Hematology
Association; PRISMA = Preferred Reporting Items for Systematic reviews and Meta-Analyses.

patients only [22]. All patients were requiring blood transfu- (pre–post) studies with no control group raised several
sion at least four times per year. Transfusion interval ranged potential methodological limitations. For instance, the lack
from monthly to four times per year. All but four studies of prespecified eligibility/selection criteria for the study
[17,19,20,22] stated their blood transfusion thresholds, population was a common issue in most of the included
although the trigger points were different, ranging from studies, especially for severe NTDbT. Although all the par-
Hb of 6 g/dL to that of 8 g/dL. Most studies based their ticipants in the studies were represented in results, it was
transfusion decisions on the Hb level and/or clinical indica- not clear whether all eligible participants were in fact
tions, such as symptomatic anemia, poor growth, and poor enrolled, raising concerns about a selection bias. A small
quality of life. sample size was a common issue for both types of NTDbT,
which potentially affects the generalizability of their find-
Mild NTDbT ings. All studies did not state whether people assessing
Study characteristics are summarized in Table 2. The stud- the outcomes were blinded to the participants’ interven-
ies were published between 2005 and 2014, and conducted tions. This could bias the results by affecting the actual out-
in seven countries: three in India; two each in Iran and Pak- comes of the participants in the studies. For more details,
istan; one each in Egypt and Italy; and one multinational see Table S1.
study (Canada, UK, and USA) [14,17,23–30]. All studies The two included RCTs [15,23], one for each type of
were single arm (pre–post design) with no control group NTDbT, were of ‘‘fair” quality as they satisfied nine of the
except one RCT, which compared two-dose regimens of 14 NIH criteria used to assess the quality of RCT. As such,
HU (10 mg/kg/day vs. 20 mg/kg/day). Only the superior they are at moderate–high risk of bias as they lack blinding
arm (10 mg/kg/day) was included as per our inclusion crite- and allocation concealment, and were of small size; see
ria. The studies collectively enrolled 373 patients. Sample Table S2.
size ranged from 15 to 106. Study populations were a mix-
ture of children and adults in all studies except one which Intervention and safety
included only adults [26]. The mean or median duration of
the studies ranged from 6 months to 156 months. In all studies, HU was used as a single intervention with no
All studies except one reported the mean Hb increment in other treatment apart from blood transfusions. HU was
the responders, which ranged between 1.5 g/L and 2.4 g/L. given as a single daily oral agent in all studies. The dose
Four studies showed a >2 g/L increment in the mean Hb of HU was similar across all the studies, ranging from
among responders [17,27,29,30], while five studies revealed 10 mg/kg/day to 25 mg/kg/day. The response to HU was
an increment between 1.5 g/L and 2 g/L [14,23,24,26,28]. measured in similar ways and entailed a decrease in the
need for transfusions or increase in Hb. The studies used
Study quality HU for 3–6 months before judging its efficacy.
All studies reported AEs related to HU, which were tran-
For the 15 observational studies, assessment of study qual- sient and improved with temporary cessation of the drug
ity using the NIH quality assessment tool for before–after and/or adjustment of the dose. These include transient
Table 1 Study characteristics for severe NTDbT.
Author Country Design Agea (range) Blood HUa N Overall Complete AEs F/Ua Quality
(y) transfusiona (mg/kg/d) response: n (%) response: n (%) (mo) (NIH tool)
El Beshlawy et al. [14] Egypt Pre/postb 12.7 (2–36) 4/y 20 25 23 (92) 11 (44) Transient 35.4 Fair
neutropenia and
Elalfy et al. [15] Egypt RCTc 9.1 (5–17.5) Q 5.8 wk 25 40 25 (62.5) 6 (15) Transient 12 Fair
Mokhtar et al. [16] Egypt Pre/post 13 (4–25) 6/yd 20.7 40 30 (75) 8 (20)d Transient 18 Fair
Ishaq et al. [17] Pakistan Pre/post 8.6 (1.8–29) 4/y 15 29 21 (72.4) 16 (55.2) Mild GI symptoms 12 Good
Hashemi et al. [18] Iran Pre/post 16.5 (3–40) 4/y 10–15 20 15 (75) 9 (45) Transient >6 Fair
and GI symptoms
Italia et al. [19] India Pre/post NR (5–40) 4/y 15–20 35 25 (71.4)d 19 (54.3)d Neutropenia 22 Fair
Yavarian et al. [20] Iran Pre/post 16.8 Regular 10–12 80 69 (83) 30 (37.5) Vomiting and 12 Poor
Karimi et al. [21] Iran Pre/post 13.5 (4–35) Monthly 8–12 120 106 (88.3) 83 (69.2) No significant AEs 50 Good
Singer et al. [22] Canada, UK, Pre/post 15.1 (1.8–35) Regular 18–20 27 NR 12 (44.4) Transient 24 ± 9 Good
and USA myelosuppression
Note. AEs = adverse events; d = day; F/U = follow-up; GI = gastrointestinal; HU = hydroxyurea; mo = month; N = sample size, NIH = National Institutes of Health; NR = not reported;
NTDbT = nontransfusion-dependent b-thalassemia; RCT = randomized clinical trial; wk = week; y = year.
Mean or median.
Only HU-alone arm was included.
Obtained from the primary author.

A.H. Algiraigri et al.

Hydroxyurea for b-thalassemia intermedia
Table 2 Study characteristics for mild NTDbT.
Author Country Design Agea (range) HUa N Responders: Mean Hba AEs F/Ua (mo) Quality
(y) (mg/kg/d) n (%) (g/dL) (NIH tool)
in responders
Bohara et al. [23] India RCTb 16.7 (NR) 10 32 18 (56.3) 1.7 Transient 6 Fair
myelosuppression and
mild GI symptoms
El Beshlawy et al. [14] Egypt Pre/postc 12.7 (2–36) 4/y 75 56 (74.7) 1.5 Transient neutropenia and 35.4 Fair
Karimi et al. [24] Iran Pre/postc 18.1 (4–45) 8–15 106 56 (52.8) 1.8 No significant AEs 96–156 Good
Ishaq et al. [17] Pakistan Pre/post 8.6 (1.8–29) 15 26 23 (88.5) 2.1 Mild GI symptoms 12 Good
Ansari et al. [25] Pakistan Pre/post 5.4 (3–13) 16 27 12 (44.4)d NR Transient 24 Good
myelosuppression and
mild GI symptoms
Rigano et al. [26] Italy Pre/post 37 (18–59) 14.6 24 17 (70.8) 1.7 No significant AEs 52 Good
Ehsani et al. [27] Iran Pre/post 10.7 (5–19) 20 16 11 (68.8) 2.1 No significant AEs 6 Poor
Singer et al. [28] Canada, Pre/post 13.7 (3–27) 18–20 15 8 (53.3) 1.5 No significant AEs 10.2 Fair
Panigrahi et al. [29] India Pre/post NR (5–32) 10–20 15 8 (53.3) 2.4 No significant AEs 12 Good
Dixit et al. [30] India Pre/post 10 (4–50) 10–20 37 26 (70.3) 2.1 Transient 12 Good
myelosuppression and
mild GI symptoms
Note. AEs = adverse events; d = day; F/U = follow-up; GI = gastrointestinal; Hb = hemoglobin; HU = hydroxyurea; mo = month; N = sample size; NIH = National Institutes of Health; NR = not
reported; NTDbT = nontransfusion-dependent b-thalassemia; RCT = randomized clinical trial; y – year.
Mean or median.
Only the low-dose arm was included.
Obtained from the primary author.

122 A.H. Algiraigri et al.

myelosuppression, transient transaminitis, and mild gas- RR of HU

trointestinal symptoms such as vomiting and diarrhea. No
long-term AEs such as cancer, infertility, or end-organ dam- Severe NTDbT
age were reported. For severe NTDbT patients, HU was associated with com-
plete cessation of regular blood transfusion with a CRR of
42% (95% CI: 29–56%); heterogeneity was considered high
with an I2 of 86.9%, p < 0.01. In addition, HU was associated
with a significant decrease (50%) in transfusion need
among severe NTDbT patients with an ORR of 79% (95% CI:
71–86%); heterogeneity was moderate with an I2 of 65.3%
and p < 0.01, as depicted in Fig. 2.

Mild NTDbT
HU was associated with an increase in Hb of >1 g/L from
baseline in mild NTDbT patients with an RR of 64% (95%
CI: 55–73%); heterogeneity was considered moderate with
an I2 of 63.9%, p < 0.01, as depicted in Fig. 2.

Analysis of heterogeneity and publication bias

< .001

Graphical inspection of forest plots for CRR, ORR, and RR in

Fig. 3 revealed nonoverlapping 95% CIs for CRR, but slightly
overlapping 95% CIs for ORR and RR, suggesting some
heterogeneity among studies. This was confirmed by the
statistically significant high to moderate I2 values.
Owing to an insufficient number of studies that provided
data, it was not possible to perform either subgroup analysis
or metaregression for the age of participants (pediatric vs.
adult), HU dose (10 mg/kg/day vs. 20 mg/kg/day), age at
first blood transfusion, transfusion threshold, and type of
the thalassemias (b-thalassemia intermedia vs. moderate
Egger test showed no evidence of publication bias of
studies on CRR (p = 0.24), ORR (p = 0.13), and RR
< .001
(p = 0.81). These were consistent with the symmetry of
the funnel plots; see Supplementary Fig. 1.


To our knowledge, this study is the first comprehensive

meta-analysis evaluating the clinical efficacy and safety of
HU in NTDbT. Seventeen studies met our inclusion criteria,
but most were observational studies. These studies enrolled
a total of 709 patients. Our meta-analysis concluded that
HU had good clinical efficacy in reducing transfusion
requirements and increasing Hgb in NTDbT, and it was well
tolerated with no major long-term adverse effects. In sev-
ere NTDbT (4 transfusions per year), HU was associated
with complete cessation of blood transfusion in 42% of study
participants. Moreover, there was a >50% reduction in trans-
fusion needs in nearly 80% of study participants. For mild
NTDbT (no or <4 transfusions per year), HU was effective
in raising Hb by at least 1 g/L in nearly two-thirds of study
participants. However, there was moderate heterogeneity
among studies for both types of NTDbT, with higher hetero-
Fig. 2 Forest plots of hydroxyurea responses in NTDbT geneity for severe NTDbT.
patients. (A) Complete response rate in severe NTDbT. (B) Heterogeneity was observed despite stringent inclusion/
Overall response rate in severe NTDbT. (C) Response rate in exclusion criteria and subcategorization of NTDbT. This was
mild NTDbT. Note. CI = confidence interval; NTDbT = nontrans- not unexpected given the poor correlation between the
fusion-dependent b thalassemia; ES = effect size. genotype and phenotype of b-thalassemia [31] and blood
Hydroxyurea for b-thalassemia intermedia 123

Fig. 3 Proposed management algorithm for NTDbT patients. Note. AEs = adverse events; CBC = complete blood count; d = day;
Hb = hemoglobin; HbF = fetal hemoglobin; HU = hydroxyurea; MCV = mean corpuscular volume; mo = month; NTDbT = nontransfu-
sion-dependent b-thalassemia; Retics = reticulocytes; wk = week.

transfusion practices that differ from one region to another mortality directly related to drug usage in the studies.
depending on institutional policies and blood availability. There were no documented long-term AEs such as leukemia,
However, when the outcome was assessed objectively, as any cancer type, or any chronic organ damage such as liver
in mild NTDbT by measuring Hb, the heterogeneity was or kidney dysfunctions; however, the follow-up duration
moderate as compared with substantial in severe NTDbT was not long enough to completely determine the inci-
where the decision to transfuse is influenced by many dence, if any, of long-term AEs.
factors. The comprehensive search of this meta-analysis identi-
Complete cessation or significant reduction (such as fied only three published cases of leukemia post-HU in
more than 50%) in chronic blood transfusions is a significant b-thalassemia [35–37], two of which were unlikely related
medical gain. Such reduction in blood transfusions will to HU due to a very short interval between the usage of
decrease in the immediate, and short- and long-term risks the drug and development of leukemia in one case and
and complications of blood transfusions. More specifically, the retrospective suspicion of coexistence of chronic leuke-
this leads to reduction in iron overload and its related mia in another case prior to the use of HU [35,37]. This left
end-organ complications and failure, less frequent clinic only one case of suspected association between HU use and
visits, and potentially less frequent monitoring of iron over- leukemia (chronic myelogenous leukemia post 5 years of HU
load. Although quality of life was not assessed in the in b-thalassemia intermedia) despite HU being used for
included studies, it was evident that a decrease or complete other hemoglobinopathies over 2 decades [36]. This reassur-
cessation of blood transfusion could be considered a sub- ing result is consistent with another meta-analysis that
stantial gain for patient quality of life. specifically addressed the long-term carcinogenicity of HU
Blood transfusions can be a significant financial burden, among the nonmalignant conditions, where no association
particularly in developing countries where most thalassemic between long-term use of HU and leukemia was found [38].
patients live. HU is an economical drug and costs approxi- The main limitations of this meta-analysis are the follow-
mately US$95 for a 1-month supply of an average 70 kg adult ing: (A) the conclusions are based on a limited number of
[32]. The cost of one unit of blood is US$316 [33] (minimum studies, with a relatively small sample size; (B) there was
of 2 units per month) and between US$1500 and US$3760 for an absence of control arms in the included studies; (C) there
a 1-month supply of an average 70 kg adult for iron chelators were relatively short follow-up periods for the HU treat-
deferoxamine and deferasirox, respectively [34]. More ment (an average of 1–2 years in most studies, which limits
widespread use of HU in thalassemia could potentially conclusions about the long-term efficacy); and (D) there was
decrease the direct and indirect costs of transfusions as well potential for selection bias across most studies due to a lack
as iron chelation therapies significantly. of prespecified eligibility/selection criteria in most of the
Apart from the transient AEs, most patients tolerated the included studies.
drug well. The most commonly reported AEs included tran- Despite the abovementioned limitations and acknowl-
sient bone marrow suppression, mild elevation of liver edging the significant need to conduct robust experimental
enzymes, nausea, and vomiting. There was no reported studies in this area, the results of this meta-analysis suggest
124 A.H. Algiraigri et al.

potential efficacy of HU therapy in patients with NTDbT. status, b-globin genotype, a-globin genotype, and molec-
Following discussions with individual patients and family ular determinants of increased fetal Hb production.
members along with a structured monitoring plan to ensure Although randomization according to specific predictors
efficacy and safety, we recommend a 3–6-month trial of HU might be challenging, subgroup analysis may reveal cer-
therapy in patients with NTDbT based on the results of this tain associations.
meta-analysis as well as the following implications: (A) seri-
ous consequences of chronic blood transfusions and their
complications; (B) growing evidence of the long-term safety Conflicts of interest
of HU; (C) ability to assess efficacy of HU in a short period
(within a few months); and (D) availability and affordability
All authors state that they do not have any conflict of inter-
of HU (especially for the developing countries).
ests to declare.


Implications for practice Acknowledgments

We wish to thank the authors of the studies included in our

The best available evidence used in this meta-analysis indi-
meta-analysis for providing additional information about
cates that HU is a potentially effective drug in the manage-
ment of NTDbT. It is well tolerated, and associated with their studies at our request.
A.A. envisioned and designed the study, wrote the protocol,
only mild and transient AEs. A risk/benefit analysis drives
searched the literature, extracted the data, evaluated the
the recommendation to proceed with using the drug in the
quality of eligible studies, analyzed and interpreted the
clinical setting, where the current therapy (chronic blood
data, and wrote the manuscript; A.K. edited the protocol,
transfusions) is associated with short- and long-term compli-
extracted the data, evaluated the quality of eligible studies,
cations that can increase disease morbidity and mortality.
analyzed and interpreted the data, and edited the manu-
Prevention or reduction in such complications, using HU,
script; and N.W. edited the protocol, interpreted the data,
outweighs the potentially unknown long-term adverse
and edited the manuscript.
effects of HU therapy.
We have suggested a clinical algorithm for the use of HU
in NTDbT, as depicted in Fig. 3. These recommendations
were based on the present meta-analysis as well as the pub- Appendix A. Supplementary material
lished guidelines for NTDbT and Sickle cell anemia (SCA)
Supplementary data associated with this article can be
Implications for research found, in the online version, at
Although HU appears to be potentially effective in the man-
agement of NTDbT across several studies, more evidence is References
needed to further consolidate the validity of this meta-
analysis. The ideal design would be that of a randomized, [1] Weatherall DJ, Clegg JB. Thalassaemia: classification, genetics
placebo-controlled trial. Longitudinal observational studies and relationship to other inherited disorders of haemoglobin.
that address the long-term sustainability of response and In: The thalassaemia syndromes. Oxford, UK: Blackwell
monitor long-term toxicity are also highly needed, since Science Ltd; 2008. p. 121–32.
such a drug can be used for several years. Additional [2] Vichinsky EP. Changing patterns of thalassemia worldwide. Ann
research is also required to address the many other ques- N Y Acad Sci 2005;1054:18–24.
[3] Taher A, Vichinsky E, Musallam K, Cappellini MD, Viprakasit V.
tions that were identified in this meta-analysis. Particular
Guidelines for the management of non transfusion dependent
issues for future trials include the following:
thalassaemia (NTDT). Nicosia, Cyprus: Thalassaemia Interna-
tional Federation; 2013.
1. The optimal HU dosage in b-thalassemia needs to be [4] Musallam KM, Taher AT, Cappellini MD, Sankaran VG. Clinical
determined, as different investigators used varying doses experience with fetal hemoglobin induction therapy in patients
of HU in their studies. However, most of the published with beta-thalassemia. Blood 2013;121:2199–212.
studies reported clinical responses using a dose range [5] Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie
of 10–20 mg/kg/day. Whether a fixed low-dose regimen D, et al. Meta-analysis of observational studies in epidemiol-
versus an escalated dose approach according to toxicity ogy: a proposal for reporting. JAMA 2000;283:2008–12.
(maximal tolerated dose) is better remains to be [6] Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting
items for systematic reviews and meta-analyses: the PRISMA
statement. Ann Intern Med 2009;151:264–9.
2. Cost effectiveness and patient quality of life should be
[7] NIH. Study quality assessment tools. Available at: <http://
assessed pre- and post-HU therapy. These could be
assessed as secondary objectives of an RCT or observa- vascular-risk-reduction/tools> [accessed March 10, 2015].
tional study. [8] Nyaga VN, Arbyn M, Aerts M. Metaprop: a Stata command to
3. There is great debate about the predictors of effective perform meta-analysis of binomial data. Arch Pub Health
response to HU in NTDbT patients such as splenectomy 2014;72:39.
Hydroxyurea for b-thalassemia intermedia 125

[9] DerSimonian R, Laird N. Meta-analysis in clinical trials. Control [25] Ansari SH, Shamsi TS, Ashraf M, Perveen K, Farzana T, Bohray
Clin Trials 1986;7:177–88. M, et al. Efficacy of hydroxyurea in providing transfusion
[10] Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring independence in beta-thalassemia. J Pediatr Hematol Oncol
inconsistency in meta-analyses. BMJ 2003;327:557–60. 2011;33:339–43.
[11] Egger M, Smith GD, Schneider M, Minder C. Bias in meta- [26] Rigano P, Pecoraro A, Calzolari R, Troia A, Acuto S, Renda D,
analysis detected by a simple, graphical test. BMJ et al. Desensitization to hydroxycarbamide following long-
1997;315:629–34. term treatment of thalassaemia intermedia as observed in vivo
[12] Begg CB, Mazumdar M. Operating characteristics of a rank and in primary erythroid cultures from treated patients. Br J
correlation test for publication bias. Biometrics Haematol 2010;151:509–15.
1994;50:1088–101. [27] Ehsani MA, Hedayati-Asl AA, Bagheri A, Zeinali S, Rashidi A.
[13] StataCorp. Stata statistical software: release 13 [computer Hydroxyurea-induced hematological response in transfusion-
program]. College Station, TX: StataCorp LP; 2013. independent beta-thalassemia intermedia: case series and
[14] El Beshlawy A, El-Ghamrawy M, Abou El-Ela M, Said F, Adolf S, review of literature. Pediatr Hematol Oncol 2009;26:560–5.
Abdel-Razek AR, et al. Hydroxyurea in the management of [28] Singer ST, Vichinsky EP, Larkin S, Olivieri N, Sweeters N,
pediatric B-thalassemia intermedia: eight years’ follow-up in Kuypers FA, et al. Hydroxycarbamide-induced changes in
Egypt. Haematologica 2014;99:737–8. E/beta thalassemia red blood cells. Am J Hematol
[15] Elalfy MS, Adly AA, Ismail EA, Elhenawy YI, Elghamry IR. 2008;83:842–5.
Therapeutic superiority and safety of combined hydroxyurea [29] Panigrahi I, Dixit A, Arora S, Kabra M, Mahapatra M, Choudhry
with recombinant human erythropoietin over hydroxyurea in VP, et al. Do alpha deletions influence hydroxyurea response in
young beta-thalassemia intermedia patients. Eur J Haematol thalassemia intermedia? Hematology 2005;10:61–3.
2013;91:522–33. [30] Dixit A, Chatterjee TC, Mishra P, Choudhry DR, Mahapatra M,
[16] Mokhtar GM, Tantawy AA, Adly AA, Ismail EA. Clinicopatho- Tyagi S, et al. Hydroxyurea in thalassemia intermedia—a
logical and radiological study of Egyptian beta-thalassemia promising therapy. Ann Hematol 2005;84:441–6.
intermedia and beta-thalassemia major patients: relation to [31] Cao A, Galanello R. Beta-thalassemia. Genet Med
complications and response to therapy. Hemoglobin 2010;12:61–76.
2011;35:382–405. [32] Moore RD, Charache S, Terrin ML, Barton FB, Ballas SK. Cost-
[17] Ishaq F, Mannan J, Seyal T, Abid H, Hassan S. Efficacy and side effectiveness of hydroxyurea in sickle cell anemia. Investiga-
effects of hydroxyurea in patient with thalassemia intermedia. tors of the multicenter study of hydroxyurea in sickle cell
Pak Paediatr J 2011;35:8–12. anemia. Am J Hematol 2000;64:26–31.
[18] Hashemi A, Abrishamkar M, Jenabzade AR, Eslami Z. Hydrox- [33] Kavanagh BD, Fischer 4th BA, Segreti EM, Wheelock JB,
yurea can reduce or eliminate transfusion requirements in Boardman C, Roseff SD, et al. Cost analysis of erythropoietin
children with major and intermediate thalassemia. Iran J Blood versus blood transfusions for cervical cancer patients receiving
Cancer 2009;1:147–50. chemoradiotherapy. Int J Radiat Oncol Biol Phys
[19] Italia KY, Jijina FJ, Merchant R, Panjwani S, Nadkarni AH, 2001;51:435–41.
Sawant PM, et al. Response to hydroxyurea in beta thalassemia [34] Delea TE, Sofrygin O, Thomas SK, Baladi J-F, Phatak PD, Coates
major and intermedia: experience in western India. Clin Chim TD. Cost effectiveness of once-daily oral chelation therapy
Acta 2009;407:10–5. with deferasirox versus infusional deferoxamine in transfusion-
[20] Yavarian M, Fayazi N, Rostami N, Shamsaei M, Karimi M. dependent thalassaemia patients. Pharmacoeconomics
Efficacy and side effects of hydroxyurea in patients with 2007;25:329–42.
thalassemia. Med J Hormozgan Univ 2007;11:109–14. [35] Bradai M, Pissard S, Abad MT, Dechartres A, Ribeil JA, Landais
[21] Karimi M, Darzi H, Yavarian M. Hematologic and clinical P, et al. Decreased transfusion needs associated with hydrox-
responses of thalassemia intermedia patients to hydroxyurea yurea therapy in Algerian patients with thalassemia major or
during 6 years of therapy in Iran. J Pediatr Hematol Oncol intermedia. Transfusion 2007;47:1830–6.
2005;27:380–5. [36] Alavi S, Safari A, Sadeghi E, Amiri S. Hematological malignan-
[22] Singer ST, Kuypers FA, Olivieri NF, Weatherall DJ, Mignacca R, cies complicating beta-thalassemia syndromes: a single center
Coates TD, et al. Fetal haemoglobin augmentation in E/beta(0) experience. Blood Res 2013;48:149–51.
thalassaemia: clinical and haematological outcome. Br J [37] Kosaryan M, Vahidshahi K, Karami H, Ehteshami S. Effect of
Haematol 2005;131:378–88. hydroxyurea on thalassemia major and thalassemia intermedia
[23] Bohara VV, Ray S, Chakrabarti P, Ray SS, Nath UK, Chaudhuri U. in Iranian patients. Pak J Med Sci 2009;25:74–8.
Optimizing the dose of hydroxyurea therapy for patients with [38] Algiraigri AH, Radwi M. Long-term safety of hydroxyurea in
beta-thalassemia intermedia (Hb E-beta-thalassemia): a single sickle cell anemia and other benign diseases: systematic
center study from eastern India. Hemoglobin 2014;38:44–8. review and meta-analysis. Blood 2014;124:560.
[24] Karimi M, Haghpanah S, Farhadi A, Yavarian M. Genotype– [39] Yawn BP, Buchanan GR, Afenyi-Annan AN, Ballas SK, Hassell
phenotype relationship of patients with beta-thalassemia KL, James AH, et al. Management of sickle cell disease:
taking hydroxyurea: a 13-year experience in Iran. Int J Hematol summary of the 2014 evidence-based report by expert panel
2012;95:51–6. members. JAMA 2014;312:1033–48.