There is a lack of consensus in the treatment of infertile women with unexplained positive autoantibody. For better understanding the effect of a commonly used drug, GCs, in this population, we conducted this systematic review and meta-analysis, and the major finding of our study is that GCs use has positive effect on improving clinical pregnancy rate and live birth rate in women with unexplained autoantibodies. Clinical pregnancy rate and live birth rate were improved only in ANA positive women without specific autoimmune disorder, but not in those with positive ACL. In addition, the effect on live birth rate is significant when GCs are administered before confirmation of pregnancy rather than after pregnancy.
4 RCTs and 2 cohort studies provided data on clinical pregnancy rate, in which 4 independent studies favored GCs use. Our systematic review and meta-analysis based on current studies confirmed the efficiency of GCs administration in improving clinical pregnancy rate in ANA but not APL positive women, which implies the different pathogenetic roles of diverse autoantibodies in establishing or maintening pregnancy. However, for subgroup with APL, we find data extracted from 2 studies are heterogenous and the sample size is small. Thus, examining whether GCs use can improve clinical pregnancy rate in APL-positive women needs further research.
The analysis based on 5 studies providing live birth rate shows the same results that GCs use has a positive effect in increasing live birth rate only in ANA positive patients, but not in APL. These results indicate that patients with different autoantibodies may also have different immunologic characteristics and clinical outcomes, thus showed different responses to GCs.
Additional subgroup analysis was added to help determine the best time to start GCs therapy, and showed that preconception use of GCs improved live birth rate, while post-conception GCs administration did not. A meta-analysis conducted by Dan et al. showed that women experiencing unexplained recurrent miscarriage benefit significantly from prednisolone treatment after confirmation of pregnancy in terms of an increased live birth rate compared with placebo [38]. However, the population in the studies included in their meta-analysis was different from ours. They used high uterine natural killer cell density (> 5%), a test that is not popular in daily practice, as inclusion criteria, and known cause for recurrent miscarriage as exclusion criteria, but no mention of autoantibodies [39, 40]. Therefore, difference in target population of the two studies may underlie the inconsistency in terms of the efficacy of GCs treatment. Furthermore, no effect in post-conception GCs treatment may also indicate risk in fetal exposure to exogenous GCs. A prospective controlled study collected and followed 311 pregnancies with systemic use of GCs in the first trimester. Higher rates of miscarriage (11.5% versus 7.0%, P = 0.013) and preterm birth (22.7% versus 10.8%, P < 0.001) were observed in GCs exposed group compared to the controls [41]. To date, prednisolone maintains a Category D rating with the Food and Drug Administration in the USA, indicating routine administration is not recommended. Therefore, further well-designed studies are needed to determine the best initial time and length of GCs therapy.
Miscarriage rate is not improved by GCs administration in neither 6 studies included nor our systematic review and meta-analysis. Same result is obtained from further subgroup analysis concerning the initiation time of GCs treatment. The miscarriage rate in subgroup “GCs administration before pregnancy” in our study is 9.4%, which is comparable to 13.5% reported by Anderson et al. [42]. Only 1 study is included in post-conception GCs administration subgroup, and the miscarriage rate is lower than control group without statistical significance. As discussed before, early fetal exposure to exogenous GCs may cause risk in miscarriage. Therefore, identifying the optimal indication for GCs use after confirmation of pregnancy warrant further well-designed study with large sample size.
Our study has several limitations that have to be considered when it comes to interpretation of the data. First, as mentioned before, our meta-analysis included only seven studies due to lack of related studies. However, because the question is so important and sample size is small in each study, and couldn’t give clinicians effective guidance, so it is necessary to have meta-analysis and systematic review to help us get more comprehensive information. Although several borderline effects needs to be confirmed by larger sample size, our study still provides preliminary information that is important for further studies. Second, the dosage, type and the therapeutic course of GCs among included trials were of a great discrepancy. Using placebo or not is also inconsistent between studies. Third, the type and dose of adjuvant therapy varies in studies included, and the effect of these therapies, such as oral aspirin, on changing pregnancy outcome are not clear[43–45]. The effect of GCs use in women with unexplained positive autoantibody needs further investigations of high quality to confirm. To gain better understanding of this therapy, well-designed prospective, randomized, controlled clinical trials must be proposed.
Based on our finding, the recommendations for future study design are listed below. First, the type of GCs should not be long acting GCs including dexamethasone or betamethasone, for they can pass through placenta and cause adverse effects such as fetal malformation. Second, the dose is not necessarily high, 10 mg/d prednisone/prednisolone or less is possibly enough to have positive effect. Third, the therapeutic course covering 1–3 months before confirmation of pregnancy or controlled ovarian hyperstimulation is recommended, although the effect of post-conception GCs therapy initiation needs further understanding. Adjuvant therapy such as anticoagulants can be added, but should be comparable in both control and GCs groups. Forth, patients should be divided into different subgroups according to their type of autoantibodies.