Oral administration of Chinese herbal medicine during gestation period for preventing hemolytic disease of the newborn due to ABO incompatibility: A systematic review of randomized controlled trials

Background About 85.3% of hemolytic disease of the newborn (HDN) is caused by maternal-fetal ABO blood group incompatibility. However, there is currently no recommended “best” therapy for ABO incompatibility during pregnancy. Objectives To systematically assess the safety and effectiveness of oral Chinese herbal medicine (CHM) for preventing HDN due to ABO incompatibility. Methods The protocol of this review was registered on the PROSPERO website (No. CRD42016038637).Six databases were searched from inception to April 2016. Randomized controlled trials (RCTs) of CHM for maternal-fetal ABO incompatibility were included. The primary outcome was incidence of HDN. The Cochrane risk of bias tool was used to assess the methodological quality of included trials. Risk ratios (RR) and mean differences with 95% confidence interval were used as effect measures. Meta-analyses using Revman 5.3 software were conducted if there were sufficient trials without obvious clinical or statistical heterogeneity available. Results Totally 28 RCTs involving3413 women were included in the review. The majority of the trials had unclear or high risk of bias. Our study found that the rate of HDN and the incidence of neonatal jaundice might be 70% lower in the herbal medicine group compared with the usual care group (RR from 0.25 to 0.30).After treatment with herbal medicine, women were twice as likely to have antibody titers lower than 1:64 compared with women who received usual care(RR from 2.15 to 3.14) and the umbilical cord blood bilirubin level in the herbal medicine group was 4umol/L lower than in those receiving usual care. There was no difference in Apgar scores or birthweights between the two groups. Conclusions This review found very low-quality evidence that CHM prevented HDN caused by maternal-fetal ABO incompatibility. No firm conclusions can be drawn regarding the effectiveness or safety of CHM for this condition.


Objectives
To systematically assess the safety and effectiveness of oral Chinese herbal medicine (CHM) for preventing HDN due to ABO incompatibility.

Methods
The protocol of this review was registered on the PROSPERO website (No. CRD42016038637). Six databases were searched from inception to April 2016. Randomized controlled trials (RCTs) of CHM for maternal-fetal ABO incompatibility were included. The primary outcome was incidence of HDN. The Cochrane risk of bias tool was used to assess the methodological quality of included trials. Risk ratios (RR) and mean differences with 95% confidence interval were used as effect measures. Meta-analyses using Revman 5.3 software were conducted if there were sufficient trials without obvious clinical or statistical heterogeneity available.

Results
Totally 28 RCTs involving3413 women were included in the review. The majority of the trials had unclear or high risk of bias. Our study found that the rate of HDN and the incidence of neonatal jaundice might be 70% lower in the herbal medicine group compared with the usual care group (RR from 0.25 to 0.30).After treatment with herbal medicine, women were twice as likely to have antibody titers lower than 1:64 compared with women who received PLOS

Introduction
Maternal-fetal incompatibility is one of the most common immunological causes for spontaneous abortions, with 15% being related to maternal-fetal ABO incompatibility [1]. About 85.3% of hemolytic disease of newborn (HDN) is also caused by maternal-fetal ABO incompatibility [2], which may result in neonatal jaundice. It is estimated that27% of newborns in China have ABO incompatibility compared with 15% worldwide [3]. During pregnancy, there are often no symptoms in the pregnant woman with ABO incompatibility, but 30% of babies born to these women have HDN, with 19% having moderately severe to severe hemolysis [4]. Intrauterine therapy such as maternal plasma exchange or intrauterine fetal blood transfusions are accepted therapies during pregnancy for maternal-fetal Rh incompatibility, as this condition has a higher risk of morbidity and mortality. Even though maternal plasma exchange is relatively safe, the procedure can be difficult and expensive [5]. On the other hand, intrauterine fetal blood transfusions have the potential risk for preterm pre-labor rupture of the membranes [6] (PPROM, 0.1%-2% risk), intra-amniotic infection (chorioamnionitis, 0.3%-1.2% risk), preterm labor, and other fetal complications (1.3%-2.5% risk).There is currently no recommended "best" therapy for preventing ABO incompatibility during pregnancy. In China intravenous injections of glucose and vitamin C in combination with oral Vitamin E are the commonly used for this condition.
ABO blood group incompatibility is recorded as Taihuang (fetal jaundice) or Tailou (threatened abortion) according to traditional Chinese medicine (TCM) diagnosis. In TCM theory, syndrome differentiation for this disease often referred to retained dampness-heat stagnation [7] and kidney-qi deficiency [8]. Consequently, clearing heat and draining dampness, warming the kidney and fortifying the spleen are the general treatment principles for this condition [9]. Previous studies published in China indicated that Herba Artemisiae Scopariae (Yinchen) and Rhizoma Rhei (Da Huang) have high contents of blood group A and B substances, which may resist antibodies from maternal-fetal incompatibility [10]. 6,7-dimethoxy coumarin, one of the ingredients of Herba Artemisiae Scopariae (Yinchen) is thought to protect the liver [11] and cause jaundice resolution [12]. A previous systematic review indicated a potential benefit for using combined therapy of Chinese and Western medicine to treat maternal-fetal ABO incompatibility [1]. However, due to the poor methodological quality of the included studies and obvious clinical heterogeneity between the trials, the study could not draw firm conclusions.

Objectives
To critically appraise the existing randomized controlled trials (RCTs) of Chinese herbal medicine (CHM) for treating ABO incompatibility during pregnancy to prevent hemolytic disease of the newborn, and provide evidence-based evaluation of the safety and effectiveness of the oral CHM.

Methods
The protocol of this review was registered on the PROSPERO website (No. CRD42016038637) and can be retrieved through https://www.crd.york.ac.uk/PROSPERO/display_record.asp? ID=CRD42016038637 (see S1 File).This review was conducted and reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement guidelines (see S2 File).

Criteria for considering studies for this review
Randomized controlled trials (RCTs) which compared oral administration of CHM (alone or in combination with other treatments) with other therapies (or no treatment)for treating maternalfetal ABO incompatibility were included in our review. We defined ABO incompatibility as the situation when the maternal blood type is "O", and the paternal blood type is A, B or AB resulting in an antibody titer (IgG anti-A or anti-B) that is higher than 1:64. The primary outcome was the incidence of HDN; secondary outcomes of interest were: the level of antibody titer after treatment, the incidence of neonatal jaundice, neonatal bilirubin levels (including neonatal umbilical cord blood bilirubin and total serum bilirubin), other measurements of health status of the newborn (such as Apgar scores and birthweight) and adverse events. To be included in our evaluation, at least one of the above outcomes need to be reported in the trial. The above strategies were adapted for each specific database, with the use of Chinese characters for relevant key words when searching Chinese databases.

Study selection
Two review authors (Cao H and Wu R) independently assessed the titles, abstracts and keywords of every record retrieved to determine relevance and whether the study was an RCT according to the inclusion criteria. Full articles were retrieved for further assessment, with reasons for exclusion being recorded. Where differences in opinions existed, they were resolved by discussion with the other authors until a consensus was reached.

Data extraction and management
Data concerning details of the included studies were extracted independently by two review authors (Cao H and Wu R) using a piloted data extraction form. The data extraction form included general information, study design, participant information, interventions and outcomes. Authors of relevant studies identified were contacted if needed to obtain information regarding additional references, unpublished trials, or data missing from the original publication. Disagreements were resolved by consensus.

Assessment of risk of bias in included studies
Two authors (Cao H and Han M) assess the methodological quality of the included trials independently. Selection bias (random sequence generation and allocation concealment), detection bias (blinding of outcome assessment), attrition bias (incomplete outcome data), reporting bias (selective reporting), and other biases (including sample size calculation, inclusion/exclusion criteria for patients' recruitment, comparability of baseline data, funding sources, and any other potential methodological flaws that may have influenced the overall results) were assessed according to the criteria from the Cochrane Handbook for Systematic Reviews of Intervention [13]. Since all of the outcomes of this review were measured objectively and were therefore unlikely to be influenced by the lack of blinding, and there was no ideal placebo control for Chinese herbal medicine (especially for the herbal decoction), we did not assess for performance bias (with blinding of participants and personnel). The level of bias was assessed for each trial (as low, high or unclear risk). A study was considered to have a low risk of bias if all seven risk of bias items met the criteria as "low risk", a study was considered to have high risk of bias if at least one of the seven items was assessed as "high risk".

Data analysis
Data were summarized using risk ratios (RR) with 95% confidence intervals (CI) for binary outcomes or mean difference (MD) with 95% CI for continuous outcomes. We used Revman 5.3 software from the Cochrane Collaboration for data analyses. Meta-analysis was used if there was acceptable homogeneity in the study design, participants, interventions, control, and outcome measures. Statistical heterogeneity was tested by examining I 2 [14], with an I 2 greater than 50% indicating a possibility of statistical heterogeneity. Both fixed effects models and random effects models were used if there was a possibility of statistical heterogeneity between trials. Pooling analysis were not performed if there was large statistical heterogeneity (I 2 >75%). Publication bias was explored by funnel plot analysis. Subgroup analyses were conducted for different baseline characteristics or different treatment duration if data were sufficient. Sensitivity analyses were used to determine whether the conclusions differed if eligibility was restricted to studies without high risk of bias, or if a fixed effect/random effect model had been applied.

Results of the search
A totally of 262 papers were identified through literature searching, and 58 full text papers were retrieved for further assessment. Finally, 32 publications of 28trials  met our predefined criteria and were included in this review. There were four duplicate studies [16&17, 31&32, 33&34, 35&36]. One paper [47] was excluded due to plagiarism. All of the included trials were conducted and published in China. Details of the searching and selection of included studies are shown in Fig 1. 1:64 at baseline, and the remaining trials included women with antibody titers higher than 1:128.
Twenty-seven of the 28 included trials assessed the use of oral CHM for preventing hemolytic disease of the newborn due to ABO incompatibility, with 22 trials [15, 16, 18-20, 22-26, 29-33, 37, 39, 41, 43-46] using herbal decoctions, four [21,27,28,40] using herbal liquid (patent) and one trial [36] using herbal granule. The remaining trial [42] assessed the use of oral administration and external washing of CHM. According to TCM theory, all of the included trials used clearing heat and draining dampness as their treatment principle. The top five most . Except Rhizoma Glycyrrhizae, which was used for coordinating the drug actions of a prescription, the other four herbs were all used for clearing heat. Herba Artemisiae Scopariae and Fructus Gardeniae also works on draining dampness.
Twenty-three of the included trials compared herbal decoction with usual care, four trials [16,21,24,28] assessed herbal decoction as an add-on treatment combined with usual care compared to usual care alone, and one trial [41] compared herbal decoction with no treatment. Usual care included vitamin supplements(in all trials: oral or intravenous vitamin C at 100-200mg three times daily and oral vitamin B6 at 20mg twice or three times daily), oxygen inhalation (in 18 trials, for 20-30minutes once daily) and oral phenobarbital (in 10 trials: at 20-30mg twice or three times daily for two to four weeks before delivery).
Details of the characteristics of the 28 included trials are shown in Table 1, and the components of the herbal prescriptions and adverse events are shown in Table 2.

Risk of bias of included studies
All of the included trials stated that the treatments were randomly assigned, however only five of them [16,18,19,26,30] were assessed as having low risk of selection bias because they reported that a random numbers table was used to generate random allocation. Five trials [21,23,31,36,43] had unequal numbers between the groups, with two of them [23,43] reporting they excluded participants after randomization if the blood type of the newborn was 'O'. These five trials were evaluated as having high risk of selection bias. The remaining seventeen trials did not provide sufficient information to judge their risk of selection bias. Furthermore, none of the included trials mentioned allocation concealment, thus all of the trials were evaluated to have unclear risk of bias regarding to this item.
Except for three trials [19,22,25] which employed a third party to measure treatment outcomes, the other twenty-five trials were assessed as having unclear risk of detection bias.
Two trials [30,36] were evaluated as having low risk of attrition bias because they clearly explained the reasons for drop outs, with equal numbers of drop outs (1 or 2 cases) in each group which is unlikely to impact on the final results. Three trials were evaluated as having high risk of attrition bias: one trial [37] reported a few cases of drop outs due to high and unchanged antibody titers during treatment and two other trials [23,43] excluded participants after randomization if the blood type of the new born were 'O'.
Since the protocols of the trials were unavailable, we determined the risk of reporting bias according to whether the trial reported all the key outcomes (HDN and antibody titer after   Only one trial [30] was assessed as having low risk of other bias, as this trial was well designed with an appropriate sample size calculation. Two trials [21,24] had only one author and another two trials [28,43] had poor methodology which may influence the results. These four trials were defined as having high risk of other bias.
In summary, twelve of the included trials [15,21,23,31,36,37,38,[43][44][45] had poor methodological quality due to the potential high risk of selection, attrition, reporting or other bias. Only five trials [16,18,19,26,30] were assessed as having low risk of bias because of adequate randomization and completeness of the report. The remaining trials all had unclear risk of bias. The risk of bias summary for each trial is showed in Fig 2.

Effects of interventions
Primary outcomes. Hemolytic Disease of the Newborn Data from twelve trials that reported the rate of HDN and compared CHM with usual care were pooled in a meta-analysis. The baseline antibody titer in two of the trials [20,29] were higher than 1:64, and in the other ten trials [19,22,23,26,30,31,33,36,42,46] were higher than 1:128, thus we performed subgroup meta-analyses on these 12 trials. With potential statistical heterogeneity between trials (50%< I 2 <75%), the subgroup analysis of participants with a baseline antibody titer that was higher than 1:64showed no difference between herbal medicine and usual care in decreasing the incidence of the HDN (RR 0.25, 95%CI 0.03 to 2.07, 215 participants, 2 trials, I 2 = 56%, random-effect model, p = 0.20), however, herbal medicine may superior to the usual care when the baseline antibody titer of the participants was higher than 1:128 (RR 0.29, 95%CI 0.17 to 0.51, 1331 participants, 10 trials, I 2 = 62%, random-effect model, p<0.0001). The overall meta-analysis (Fig 3) showed that CHM was better at reducing the incidence of HDN compared to usual care (RR 0.30, 95%CI 0.18 to 0.49, 1546 participants, 13 trials, I 2 = 58%, random-effect model, p<0.00001). The funnel plot for this comparison showed considerable asymmetry (Fig 4A), which indicates that publication bias cannot be ruled out for this outcome.
One trial [25] also compared CHM with usual care and reported that no case of HDN occurred. One other trial [41] found CHM was superior to no treatment in reducing the  number of HDN (RR 0.36, 95%CI 0.14 to 0.87, 170 participants). Another trial [16] reported no difference between herbal medicine plus usual care and usual care alone in reducing the incidence of the HDN when the baseline antibody titer of participants was higher than 1:128 (RR 0.42, 95%CI 0.17 to 1.04, 60 participants).
Another two trials [21,23] only reported the numbers of the newborn with Apgar scores of less than 8, with no difference between the CHM and usual care groups (RR 0.97, 95%CI 0.29 to 3.22, 177 participants) and between CHM plus usual care and usual care alone (RR 0.67, 95%CI 0.12 to 3.82, 98 participants).
Adverse events Twenty of the 28 included trials did not report whether there were any adverse events during the treatment period. Six [21,23,30,38,41,42] of the remaining eight trials found no adverse events in herbal medicine group. Only two trials [16,37] mentioned the adverse events which were present in both herbal medicine and control groups. One reported a case (3.33%) of abortion/stillbirth in herbal group and four cases (13.33%) of abortion/stillbirth in control group, and the other trial reported two cases (2.17%) of abortion/stillbirth in herbal group and ten cases (11.11%) of abortion/stillbirth in control group.
For the meta-analyses in which the fix-effects model and random effects model were both used as sensitive analysis, the results were consistent irrespective of potential heterogeneity between trials. Only the results of the meta-analysis using the random-effects model were reported in this review.

Summary of main results
This review included 28 randomized controlled trials which assessed the effects of herbal medicine for preventing hemolytic disease of the newborn in pregnant women with ABO incompatibility and antibody titers that were higher than 1:64. The majority of the trials had unclear or high risk of bias according to the mentioned assessment criteria. The main results found that the rate of HDN and the incidence of neonatal jaundice in herbal medicine group might be 70% lower compared than in the usual care group (25.3% in usual care group, RR from 0.25 to 0.30); twice as many women taking herbal medicine (RR from 2.15 to 3.14) had post-treatment antibody titers lower than 1:64 compared with those receiving usual care, and umbilical cord blood bilirubin in herbal medicine group was 4 umol/L less than in the usual care group. There was no difference in Apgar scores and birthweights between groups. However, considering the poor methodological quality of the included trials, there remains some uncertainty about whether Chinese herbal medicine prevents HDN due to the ABO incompatibility.
Only a few included trials mentioned adverse events, with six trials reporting no adverse event in the herbal medicine group, and another two trials reported lower rates of fetal loss in the herbal medicine group (3%) than in the control group (13%). Due to insufficient information regarding the safety of herbal medicine for pregnant women, more trials will be needed to confirm its safety.

Overall completeness and applicability of evidence
We searched two English databases and four Chinese databases without language restrictions. However, all of the 28 included trials from this review were conducted and published in China. In countries outside of China, there is currently no standard therapy for treating maternal-fetal ABO incompatibility during pregnancy, with therapeutic procedures only being provided if the newborn was diagnosed with HDN after birth. Thus, the results of this review may have significant implications for providing a preventative therapy for HDN caused by ABO incompatibility.

Quality of the evidence
As previously mentioned, only five trials were evaluated as having low risk of bias. The remaining 23 trials all have poor methodology and have high or unclear risk of selection, attrition, reporting and other bias.
We employed the Grading of Recommendations Assessment, Development and Evaluation (GRADE) criteria ratings to assess the quality of the evidence; factors that downgraded the quality include imprecision, inconsistency, indirectness, limitations and bias of the evidence. For the meta-analyses in this review, we downgraded the level of evidence mainly because of the potential for performance, attrition, and reporting bias and the inconsistency of the results between the included studies. Subsequently, we could find only very low-quality evidence that CHM may reduce the incidence of HDN and reduce antibody titers, and low-quality evidence that CHM may reduce the incidence of neonatal jaundice and umbilical cord blood bilirubin levels ( Table 3 Summary of finding table).

Agreements and disagreements with other studies or reviews
Two other reviews [1,48] were found on this topic, which were all dissertations for masters' degrees and written in Chinese, with one of them being published in 2015 [1]. One review included eleven trials and the other included 20 trials. Both of them used a total effective rate as the primary outcome measure, which is a composite outcome index combining the reduction of antibody titers, rate of HDN and adverse birth outcomes post treatment. Although this outcome index is commonly used in China, it is difficult to determine its clinical relevance since the definition of "effectiveness" is inconsistency between studies. These two reviews also conducted meta-analyses to investigate the effects of CHM in reducing the rate of HDN, with results consistent with our studies, showing that CHM was more likely to reduce the incidence of HDN compared to usual care (RR = 0.19 or 0.25, p<0.05). One of the reviews also reported there was no difference between CHM and usual care in Apgar scores and birthweight.
Our review included 28 trials and assessed the effects of CHM for reducing the incidence of HDN, the antibody-titer, the incidence of neonatal jaundice and umbilical cord blood bilirubin. Subgroup-meta analyses were conducted for the participants with different baseline antibody titer values. We found similar results to the previous reviews, showing that CHM is effective in reducing the incidence of HDN (RR = 0.30, p<0.05), with the overall quality of the evidence appraised according to the GRADE criteria.

Implications for practice
Maternal-fetal ABO incompatibility may be asymptomatic during pregnancy, and patients may not be aware they have this condition. As there are currently no guidelines or standard therapies to prevent HDN caused by ABO incompatibility during pregnancy, this study may have identified a possible therapy for preventing this condition.
According to the review, five herbs are frequently used to prevent HDN caused by maternal-fetal ABO incompatibility, including Herba Artemisiae Scopariae (Yin Chen), Rhizoma Glycyrrhizae (Gan Cao), Fructus Gardeniae (Zhi Zi), Radix Scutellariae (Huang Qin), and Radix et Rhizoma Rhei (Da Huang). Among these five herbs, Yin Chen has been commonly used for treating jaundice and HDN.A laboratory study [49] found that liver fibrosis induced by dimethylnitrosamine in rats was improved by the administration of Yinchenhao Decoction (in which Yin Chen was the primary active ingredient), with a reduction in hydroxyproline and a significant improvement in liver function and hepatic histology after 2 weeks of treatment. Da Huang was found to promote the secretion of bile, causing gallbladder and biliary sphincter relaxation which may relieve jaundice. The potential mechanism may be through increased expression of transport proteins associated with the metabolism of bile acids to reduce the accumulation of bile acids and other toxic compounds in the liver [50,51], which could improve the ultrastructure of liver cells, thus affecting the intracellular enzyme activity and the concentration of calcium ions in the cells [52]. Huang Qin is thought to be a heatclearing, phlegm-removing herb, traditionally used to cool heat, clear damp-heat, and calm the fetus [53]. It is also found to contain compounds with anti-inflammatory and antimicrobial

Weigh of newborn (kg)
The mean Birthweight in the control groups was3.10 The mean Birthweight in the intervention groups was0.06 higher(0.04 lower to 0.15 higher) 578 È⊝⊝⊝ (5 studies) very low 1,2,3 *The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: Confidence interval; RR: Risk ratio GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate. 1 Majority trials had unclear risk of selection, detection, attrition and reporting bias 2 The I square test showed a significant statistical heterogeneity among trials (I-square larger than 50%) 3 The asymmetric funnel plot indicated the possibility of publication bias. 4 The I square test showed a significant large statistical heterogeneity among trials (I square larger than 70%) https://doi.org/10.1371/journal.pone.0180746.t003 effects [54].Zhi Zi has effects on the contraction of gallbladder, containing iridoid and Saffron glucoside which may increase the secretion of bile [55].Even though there are some studies (cited as above) tried to investigate the potential mechanism of the specific herbs in treating HDN, we need to aware that the mechanisms of how the Chinese herbal medicine could eliminate maternal IgG antibodies were not evidenced. Regarding safety considerations, medications used during pregnancy require extra caution. According to Chinese herbal medicine traditions, herbs that work on clearing heat are usually forbidden during pregnancy, as it is believed that oral intakes of these herbs may cause abortions and other serious adverse birth outcomes. However, four of the top five most frequently used herbs in the included trials from this review are herbs that have the function of clearing heat [53]. The majority of trial participants took the herbal decoction for at least half of their pregnancy until labor, with no or less adverse events (such as fetal loss) reported in the treatment group compared with the usual care group. This finding suggests that it may be safe for pregnant women with ABO incompatibility to use CHM with herbs that have the function of clearing heat and draining dampness, under the guidance of herbal medicine practitioners. According to our study, the incidence of HDN from ABO incompatibility may decrease from 25.3% to 7.6% when CHM is used during pregnancy, compared to usual care. However, we have low confidence in certainty of the safety issue using Chinese herbal medicine to prevent or treat ABO HDN based on the current evidence.