This is the protocol for a review and there is no abstract. The objectives are as follows:
To compare the efficacy and safety of IDA versus other anthracyclines in induction therapy of newly diagnosed AML patients.
Description of the condition
Acute myeloid leukaemia (AML) is a heterogeneous group of clonal malignant myeloid disorders which have clinical similarities but distinct morphologic, immunophenotypic, cytogenetic and molecular features. It is the most common type of myeloid leukaemia with an overall incidence of 3.7 cases per 100,000 persons. The incidence of AML increases with age and the median year at presentation is approximately 65 years. AML represents 80% to 90% of acute leukaemia cases in adults but accounts for fewer than 15% of leukaemia cases in children younger than 10 years (Greer 2009a). AML is slightly more common among populations of European background and acute promyelocytic leukaemia (APL), a distinct subtype of AML, has a higher incidence in populations of Latino or Hispanic descent (Douer 1996; Estey 1997).
AML is characterised by an increased number of immature myeloid cells (blasts) in bone marrow, peripheral blood and other tissue, resulting in impaired haematopoiesis manifested by cytopenias (Lowenberg 1999). It results from genetic alterations in normal haematopoietic stem cells that induce differentiation arrest or excessive proliferation of the affected cells, or both (Jabbour 2006). Several factors have been implied in acquiring AML that include exposure to ionising radiation, benzene and cytotoxic chemotherapy (Estey 2006). The World Health Organization (WHO) classifies AML into five major categories: (a) AML with recurrent genetic abnormalities; (b) AML with multilineage dysplasia; (c) AML and myelodysplastic syndromes (MDS), therapy-related; (d) AML not otherwise categorised; and (e) acute leukaemia of ambiguous lineage (Greer 2009a). Typical clinical presentations of AML are fatigue and weakness, haemorrhage, or infections and fever due to decreases in red cells, platelets or white cells, respectively. Additionally, leukaemic infiltration of various tissues can produce a variety of corresponding symptoms such as hepatomegaly, splenomegaly, lymphadenopathy, leukaemia cutis (the outermost, nonvascular layer of the skin) and so on (Lowenberg 1999). Besides these common clinical presentations of AML, acute promyelocytic leukemia (APL) possesses additional characteristics. It was used to be considered as the most fatal subtype of AML because of its potential fatal haemorrhage due to consumptive coagulopathy. But now it is regarded as the most curable subtype for its highly sensitivity to all-trans retinoic acid (ATRA) and arsenic trioxide (ATO) (Wang 2008).
Studies on the pathogenesis and prognosis of AML have made revolutionary progress; however, the treatment for AML remains unsatisfactory. Only 40% to 45% of AML patients enjoy long-term disease-free survival (DFS) and most patients still die of their disease primarily due to persistent or relapsed AML (Burnett 2011). More progress in the treatment of AML is required.
Description of the intervention
Treatment of AML consists of two phases: remission-induction therapy phase and post-remission therapy phase. The former one aims to attain a complete remission (CR), while the latter aims to maintain the CR. Achieving CR by remission-induction therapy is essential to prolonging survival and obtaining a cure for AML patients. For several decades, a combination of an anthracycline and cytarabine (Ara-C) has been the standard for remission-induction therapy of AML (Dohner 2010). Therefore, selecting the most effective and tolerable anthracycline is key to maximising treatment outcomes.
Daunorubicin (DNR) is the most widely used anthracycline. The standard dose of DNR used in remission-induction therapy is 45 mg/m²/d for three days (Fernandez 2009). A combination of three days of DNR at a dose of 40 mg/m²/d to 60 mg/m²/d and seven days of Ara-C at a dose of 100 mg/m²/d to 200 mg/m²/d generally has been used for more than 40 years (Burnett 2011; Lowenberg 1999). With this regimen, approximately 60% to 80% of adults with AML achieve CR, whereas only 40% to 45% of patients enjoy long-term disease-free survival (DFS) (Burnett 2011; Lowenberg 1999; Tallman 2005; Zittoun 1995). Additionally, DNR tends to cause serious cumulative injury to the heart resulting in congestive cardiomyopathy and, ultimately, congestive heart failure, which is usually refractory to medical therapy. Other common side effects include myelosuppression, nausea, vomiting, diarrhoea, alopecia and mucositis (Greer 2009b).
To improve the efficacy and reduce the side effects of remission-induction therapy, various alternative anthracyclines were developed and introduced into clinics in the 1980s, among which idarubicin (IDA) is a most promising one (Johnson 1998). IDA, also called 4'-demethoxydaunorubicin (4-DMDR), is a DNR derivative synthesised by replacing the C-4 methoxyl group with a hydrogen atom (Arcamone 1976). With this minor structural alteration, IDA has several theoretical advantages over the parent compound: (1) IDA has a more effective antileukaemia activity (Casazza 1980); (2) IDA is active by both intravenous and oral routes of administration (Ganzina 1986); (3) IDA has an ability to overcome the multidrug resistant (MDR) phenotype and reduces the development of drug resistance (Berman 1992); (4) IDA is less cardiotoxic and is well tolerated (Cersosimo 1992). IDA was registered and approved by the Food and Drug Administration (FDA) of USA in 1990. The standard dose of IDA used in remission-induction therapy is 12 mg/m²/d for three days (Ohtake 2011). At present, IDA has been used as the first-line therapy at a dose of 10 mg/m²/d to 12 mg/m²/d for three days in younger adult patients (18 to 60 years) with newly diagnosed AML, or relapsed/refractory AML (Dohner 2010).
How the intervention might work
IDA is an anthracycline antineoplastic agent. It mediates control of AML by two molecular mechanisms. First, IDA inhibits DNA topoisomerase II, which is a nuclear enzyme that modulates DNA topology by passing a double-stranded DNA through a transient break in the DNA backbone. By poisoning the enzyme to prevent it from religating (i.e. binding back together) cleaved DNA, IDA converts topoisomerase II into a toxin, resulting in high levels of transient protein-associated breaks in the genome of treated cells. Second, IDA intercalates into base pairs of DNA and generates free radicals to break the DNA strand. Both eventually lead to the death of leukaemia cells (Greer 2009b). Experimental laboratory studies have indicated that IDA and DNR have equal affinity for DNA and comparable inhibitory effects on DNA topoisomerase II (Ganzina 1986). The higher antileukaemia activity of IDA may result from its metabolite idarubicinol, which is more active and has a longer half-life than the metabolite of DNR (Robert 1992). For the ability of overcoming the MDR phenotype, some studies suggest that IDA has a high lipophilic (having an affinity for, tending to combine with, or capable of dissolving in lipids) coefficient and is less of a substrate for P-glycoprotein (P-gp) than DNR, which acts as an active efflux pump, thereby allowing for greater intracellular drug accumulation (Berman 1992; Supino 1977).
A great number of phase I/II trials support the activity of IDA in AML. In phase I trials, IDA was demonstrated to be less cardiotoxic and the dose-limiting toxicity of the drug was myelosuppression (Berman 1983; Kaplan 1982). In later phase II trials, as a single agent, IDA induced CR in about 20% of adult patients with relapsed or refractory AML (Carella 1984; Hayat 1984). Combining IDA with Ara-C increased CR to a range of 24% to 70% in similar groups of heavily pretreated patients (Berman 1989; Harousseau 1987; Lambertenghi-Deliliers 1987). In previously untreated AML, more than 80% of patients achieved CR after being treated with a combination of IDA, Ara-C and etoposide (an anti-cancer agent which kills cancer cells by inhibiting their DNA synthesis) (Carella 1987). For newly diagnosed APL, IDA, when combined with ATRA, induced a CR rate higher than 80% either in adults or in elderly patients (Avvisati 1996; Latagliata 1997).
On the basis of these trials, numerous prospective, randomised controlled trials (RCTs) testing the superiority of IDA versus other anthracyclines including DNR have been conducted in previously untreated AML (Beksac 1998; Berman 1991; Creutzig 2001; Harousseau 1996; Indrak 2001; Mandelli 1991; Mandelli 2009; Morita 2010; Ohtake 2011; Pignon 1996; Reiffers 1996; Rowe 2004; Vogler 1992; Wiernik 1992). However, the outcomes of these RCTs are inconsistent. Three initial RCTs comparing standard dose IDA (12/13 mg/m²/d for three days) with standard dose DNR (45/50 mg/m²/d for three days) reported a superior CR rate for the IDA group (Berman 1991; Vogler 1992; Wiernik 1992). However, the long-term follow-up of the three RCTs revealed that the IDA group had a better overall survival (OS) than the DNR group in only one of the three RCTs (Berman 1997). A study published by Mandelli in 1991, which compared IDA (12 mg/m²/d for three days) with DNR (45 mg/m²/d for three days) in elderly AML patients (age greater than 55 years), failed to demonstrate any significant difference in CR rate, OS and relapse free survival (RFS) between the two arms (Mandelli 1991). In another study published by Mandelli in 2009, the use of IDA (10 mg/m²/d for three days) was superior to DNR (50 mg/m²/d for three days) in terms of disease-free survival (DFS), survival from CR and OS, but was similar to mitoxantrone (12 mg/m²/d for three days) (Mandelli 2009). Moreover, in two recent studies, doubling the dose of DNR from the standard dose (45 mg/m²/d for three days) to 90 mg/m²/d for three days significantly improved the CR rate and duration of OS (Fernandez 2009; Lowenberg 2009). In a more recent study conducted by Ohtake, DNR at a dose of 50 mg/m²/d for five days was found to be equivalent to IDA (12 mg/m²/d for three days) in CR rate, RFS and OS without increasing the risk of infection or cardiomyopathy (Ohtake 2011). Therefore, the superiority of IDA versus other anthracyclines remains a matter of debate.
Why it is important to do this review
Anthracyclines have been the core treatment for AML for several decades; thus, selecting the most effective and tolerable anthracycline is key to maximising treatment outcomes. In spite of the theoretical advantages of IDA, RCTs comparing induction therapy based on IDA with those based on other anthracyclines have conflicting results. There is no evidence that would definitively prove the superiority of IDA versus other anthracyclines with respect to CR rate, RFS, DFS and OS. We would like to assess which anthracycline is the most effective to be used for induction therapy. Although a meta-analysis for IDA is available (AML Collaborative Group 1998), it only included RCTs published before 1996 and many new RCTs have been published since then (Beksac 1998; Creutzig 2001; Indrak 2001; Mandelli 2009; Morita 2010; Ohtake 2011; Rowe 2004). It is important to update the information by including all new trials. Therefore, we are planning to undertake this systematic review to obtain definitive evidence on the role of IDA versus other anthracyclines in the treatment of AML. Our review will inform about the current status of clinical practice and provide some guidance for future clinical studies in this area.
To compare the efficacy and safety of IDA versus other anthracyclines in induction therapy of newly diagnosed AML patients.
Criteria for considering studies for this review
Types of studies
We will accept only randomised controlled trials (RCTs) in this review and we will include both full text and abstract publications. We will exclude quasi-randomised trials and cross-over trials due to the risk of bias.
Types of participants
Participants are patients with newly diagnosed AML according to French-American-British (FAB) or WHO diagnostic criteria, or both, irrespective of age, gender and ethnicity. For studies with mixed populations, we will include data from the AML subgroups. Alternatively, if subgroup data for AML patients are not available, we will include the whole study if at least 80% of the patients have AML.
Types of interventions
Induction therapy based on other anthracyclines, e.g. daunorubicin (DNR), doxorubicin (DOX), aclarubicin (ACR) and mitoxatrone (MIT): any form of application, any dose.
Types of outcome measures
Overall survival (OS): defined as the time from randomisation or entry into study to death from any cause or to last follow-up.
Disease-free survival (DFS): defined as the time from CR to first relapse, or death from any cause or the last follow-up.
Complete remission (CR) rate: CR is defined by bone marrow blasts < 5%, no blasts in peripheral blood, absence of extramedullary disease, absolute neutrophil count > 1.0 x 109/L, platelet count > 100 x 109/L and independence of red cell transfusions (Dohner 2010).
Death on induction therapy.
Relapse rate: relapse is defined by recurrence of bone marrow blasts > 5%, or reappearance of blasts in the blood, or development of extramedullary disease (Dohner 2010).
Adverse events (haematologic toxicity including mean duration with a neutrophil count < 1.0 x 109/L, mean duration with a platelet count < 100 x 109/L), and mean duration of hospitalisation days after induction treatment; non-haematologic toxicity including grade 3/4 toxicity of nausea/vomiting, alopecia, diarrhoea, mucositis, hepatic, renal or cardiac dysfunction, etc.).
Quality of life.
Search methods for identification of studies
We will adopt search strategies from those suggested in the Cochrane Handbook for Systematic Reviews of Interventions (Lefebvre 2011). We will apply no language restriction to reduce language bias.
We will search the following bibliographic databases:
Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library, latest issue, see Appendix 1 for search strategy).
MEDLINE (1969 to present, see Appendix 2 for search strategy).
EMBASE (1974 to present, see Appendix 3 for search strategy).
Chinese BioMedical Literature Database (CBM) (1978 to present, see Appendix 4 for search strategy).
We will search conference proceedings of the following societies for the years that are not included in CENTRAL (1990 to present, see Appendix 5 for search strategy):
We will search databases of ongoing trials:
Searching other resources
We will handsearch:
Conference Proceedings of the Chinese Society of Hematology (1980 to present).
References of all identified trials, relevant review articles and current treatment guidelines.
We will contact experts in this area and pharmaceutical companies for extra unpublished and ongoing studies.
Data collection and analysis
Selection of studies
Two review authors (XL, SX) will independently screen all the obtained titles and abstracts from the mentioned resources and reject studies that are obviously irrelevant. We will make our best efforts to obtain the full text of all potentially relevant studies. Then these full-text studies will be screened independently by the two review authors with the inclusion criteria stated in the section "Criteria for considering studies for this review". If necessary, we will contact the authors of studies for further information to make a decision about eligibility. The review authors will not be blind to the study authors' names, institutions and journal of publication. Any disagreement between the two review authors will be resolved through discussion. According to PRISMA, we will document the overall number of studies identified, included and excluded, and the reasons for exclusions at every stage of searching and screening of the literature in a flow diagram (Liberati 2009; Moher 2009).
Data extraction and management
Two review authors (XL, SX) will independently extract data from the included studies according to Chapter Seven of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a). We will resolve disagreements by consensus. We will record all the extracted data on paper data collection forms and enter them into Review Manager 5 (RevMan 2011). We will extract the following groups of data:
Quality assessment: sequence generation, allocation sequence concealment, blinding, incomplete outcome data, selective outcome reporting, and other concerns about bias.
Study characteristics: title, first author, contact address, publication data, publication status (published, published as abstract or unpublished), duplicate publications, country, language, trial design, aims, setting (inpatients or outpatients), data (defined as recruitment initiation year), centre (single centre or multicentre), trial sponsor, inclusion/exclusion criteria, reasons for exclusion, sample size, power calculation, comparability of groups, subgroup analysis, stopping rules described, duration of follow-up, results, and conclusion.
Participant characteristics: age, gender, ethnicity, ECOG status, total number recruited/randomised/analysed, FAB subtype (M0-M7, not assessed), cytogenetics (favourable, intermediate, adverse, not assessed), treatment history, additional diagnoses, lost to follow up numbers, and dropouts (percentage in each arm) with reasons;
Interventions: experimental and control interventions, time, dosage, regimen, cycles and route of interventions, compliance to interventions, additional interventions given, and any differences between interventions.
Results of outcomes: overall survival (hazard ratio (HR); 95% confidence interval (CI)/P value), disease-free survival (HR; 95% CI/P value), complete remission rate, death during induction phase, relapse rate, adverse events and quality of life.
Whenever possible, we will seek missing data from the authors of studies.
Assessment of risk of bias in included studies
Two review authors (XL, SX) will independently assess quality and risk of bias for each included study. According to the recommendations in Chapter Eight of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011b), we will use a questionnaire for the following criteria:
Blinding (participants, personnel, outcome assessors).
Incomplete outcome data.
Selective outcome reporting.
Other sources of bias.
Our judgement of the review will involve an answer for each criteria based on a three-point scale (low risk of bias, high risk of bias and unclear) and a summary description. We will resolve disagreements between the two review authors by consensus.
Measures of treatment effect
For dichotomous outcome data, attainment of CR will be counted as an event for 'CR rate'; death due to hypoplastic marrow or progressive disease, or death before marrow re-evaluation for 'death on induction phase'; relapse for 'relapse rate'; and development of adverse events for 'adverse events'.
For dichotomous data, we will calculate risk ratios (RR) as measures of treatment effect with 95% CI as measures of uncertainty. For continuous data, we will calculate mean difference (MD) with 95% CI or, if different scales are used, standardised mean difference (SMD) with 95% CI. For time-to-event data, we will calculate HR with 95% CI using the methods described by Parmar (Parmar 1998) and Tierney (Tierney 2007).
Unit of analysis issues
For parallel group designed RCTs in which participants were individually randomised to one of two intervention groups, and a single measurement for each outcome from each participant was collected and analysed, we will use individual participant as a unit of analysis.
For RCTs with three arms, we will just lump the two comparators together in a 1:2 comparison, rather than including each comparison IDA to control separately (to avoid double counting the IDA patients).
Dealing with missing data
We will deal with the missing data as suggested by the Cochrane Handbook for Systematic Review of Intervention (Higgins 2011c) and the National Research Council (NRC) report on missing data (Little 2012):
Document the reasons why data are missing as clearly as possible.
Decide on a primary assumption about the missing-data mechanism, including "missing at random (MAR)" and "missing not at random (MNAR)".
For data that are MAR, we will analyse the only available data (i.e. ignoring the missing data); For data that are MNAR, we will input the missing data with replacement values and treat these as if they were observed (e.g. last observation carried forward, imputing an assumed outcome such as assuming all were poor outcomes, imputing the mean, imputing based on predicted values from a regression analysis).
Perform sensitivity analyses of primary outcomes to assess how sensitive results are to reasonable changes in the assumptions that we made.
Address the potential impact of missing data on the findings of the review in the Discussion section.
Assessment of heterogeneity
We will detect heterogeneity of treatment effects across studies using the Chi2 test with a significance level at P < 0.1. We will also use the I² statistic for quantifying inconsistency (Deeks 2011). We will use the following rough guide to interpretation of I²:
0% to 40%: low heterogeneity .
30% to 60%: may represent moderate heterogeneity.
50% to 90%: may represent substantial heterogeneity.
75% to 100%: considerable heterogeneity.
We will explore potential causes of heterogeneity by sensitivity and subgroup analysis as defined below.
Assessment of reporting biases
We will make our best efforts to minimise the impact of reporting biases by searching comprehensively for studies that meet the eligibility criteria for the review, including unpublished studies and trial registries, contacting the original investigators and pharmaceutical companies of IDA, making no restriction on location or language, and carefully examining the author, institute and detailed information of studies. If at least ten studies are included in future review, the possibility of publication bias will be assessed using a funnel plot (Sterne 2011).
We will perform data analyses according to the recommendations of Chapter Nine of the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2011). We will use the Cochrane statistical package Review Manager 5 (RevMan 2011) to synthesise data. One review author (XL) will input data into the software and a second review author (YT) will check it for accuracy. We will firstly adopt a fixed-effect model for meta-analyses. If there is substantial heterogeneity, we will also carry out a random-effects analysis. We will choose the Mantel-Haenszel method for dichotomous data outcomes and the inverse variance method for both continuous data outcomes and time-to-event data outcomes.
Subgroup analysis and investigation of heterogeneity
Different dosages of IDA and different dosages of DNR (e.g. 8 mg/m²/d IDA versus 45 mg/m²/d DNR, 8 mg/m²/d IDA versus 60 mg/m²/d DNR, 8 mg/m²/d IDA versus 90 mg/m²/d DNR, 10 mg/m²/d IDA versus 45 mg/m²/d DNR, 10 mg/m²/d IDA versus 60 mg/m²/d DNR, 10 mg/m²/d IDA versus 90 mg/m²/d DNR, 12 mg/m²/d IDA versus 45 mg/m²/d DNR, 12 mg/m²/d IDA versus 60 mg/m²/d DNR, 12 mg/m²/d IDA versus 90 mg/m²/d DNR, or other doses used).
Cytogenetic risk stratification (favourable, intermediate, or adverse).
Anthracycline agent types of control intervention (e.g. DNR, DOX, ACR, or MIT)
Age (e.g. age < 60, age > 60, or otherwise defined for elderly AML).
AML subtypes (APL or other subtypes of AML)
We will perform sensitivity analysis based on:
Methodological quality of the studies, including all studies versus including only those studies with low risk of bias.
Fixed-effect versus random-effects model.
We are grateful to Ina Monsef (Trial Search Co-ordinator of the CHMG) for her assistance with the search strategy and Sabine Kluge (Managing Editor of the CHMG) for her assistance with this protocol.
Appendix 1. CENTRAL search strategy
Appendix 2. MEDLINE search strategy
OVID MEDLINE search strategy
|1||exp Leukemia, Myeloid, Acute/ |
|2||leukemia, myeloid/ |
|3||acute disease/ |
|4||2 and 3 |
|5||(acut$ or akut$ or agud$ or aigu$).tw,kf,ot. |
|6||((myelo$ or mielo$ or nonlympho$ or granulocytic$) and (leuk?em$ or leuc$)).tw,kf,ot. |
|7||5 and 6 |
|10||1 or 4 or 7 or 8 or 9 |
|16||(desmethoxydaunorubicin$ or demethoxydaunorubicin$).tw,kf,nm,ot. |
|17||(imi-30$ or imi30$).tw,kf,nm,ot. |
|18||(nsc-256439$ or nsc256439$).tw,kf,nm,ot. |
|23||10 and 22 |
|24||randomized controlled trial.pt. |
|25||controlled clinical trial.pt. |
|28||drug therapy.fs. |
|34||32 and 33 |
|35||10 and 22 and 34 |
Appendix 3. EMBASE search strategy
|1|| 'ACUTE GRANULOCYTIC LEUKEMIA':exp|
|4||#2 AND #3|
|5||acut*:ab,ti,tt OR akut*:ab,ti,tt OR agud*:ab,ti,tt OR aigu*:ab,ti,tt|
|6||((myelo*:ab,ti,tt OR mielo*:ab,ti,tt OR nonlympho*:ab,ti,tt OR granulocytic*:ab,ti,tt) AND (leuk*em*:ab,ti,tt OR leuc*:ab,ti,tt))|
|7||#5 AND #6|
|10||#1 or #4 or #7 or #8 or #9|
|16||desmethoxydaunorubicin*:ab,ti,tt OR demethoxydaunorubicin*:ab,ti,tt|
|17||imi-30*:ab,ti,tt OR imi30*:ab,ti,tt|
|18||nsc-256439*:ab,ti,tt OR nsc256439*:ab,ti,tt|
|22||#11 OR #12 OR #13 OR #14 OR #15 OR #16 OR #17 OR #18 OR #19 OR #20 OR #21|
|23||#10 AND #22|
|24||'randomized controlled trial'/exp OR 'single blind procedure'/exp OR 'double blind procedure'/exp OR 'crossover procedure'/exp|
|25||random*:ab,ti OR placebo*:ab,ti OR allocat*:ab,ti OR crossover*:ab,ti OR 'cross over':ab,ti OR trial:ti OR (doubl* NEXT/1 blind*):ab,ti|
|26||#24 OR #25|
|27||'animal'/de OR 'animal experiment'/de OR 'nonhuman'/de|
|29||#27 AND #28|
|30||#27 NOT #29|
|31||#26 NOT #30|
|32||#10 AND #22 AND #31|
Appendix 4. CBM search strategy
|1||myeloid leukemia AND idarubicin |
|2||myeloid leukemia AND demethoxydaunorubicin |
Appendix 5. Other search strategy
|1||myeloid leukemia AND idarubicin|
|2||myeloid leukemia AND idamycin|
|3||myeloid leukemia AND idaralem|
|4||myeloid leukemia AND zavedos|
|5||myeloid leukemia AND desmethoxydaunorubicin|
|6||myeloid leukemia AND demethoxydaunorubicin|
|7||myeloid leukaemia AND idarubicin|
|8||myeloid leukaemia AND idamycin|
|9||myeloid leukaemia AND idaralem|
|10||myeloid leukaemia AND zavedos|
|11||myeloid leukaemia AND desmethoxydaunorubicin|
|12||myeloid leukaemia AND demethoxydaunorubicin|
Contributions of authors
Xi Li: developing protocol, identification of studies, screening studies, extracting data, conducting analysis, drafting report.
ShuangNian Xu: developing protocol, screening studies, extracting data, conducting analysis, revising the protocol and the review.
Ya Tan: identification of studies, screening studies, conducting analysis.
Jieping Chen: developing protocol, identification of studies, screening studies, drafting report, revising the protocol and the review.