Arterial/venous thrombosis, fetal loss and stillbirth in pregnant women with systemic lupus erythematosus versus primary and secondary antiphospholipid syndrome: a systematic review and meta-analysis

EMBASE (www.​sciencedirect.​com), MEDLINE database of medical research articles, and Google Scholar were carefully searched for relevant publications comparing SLE with PAPS and/or SAPS in pregnant women.

In addition, official websites of major rheumatology and maternity journals were also reviewed for relevant articles.

The following words/terms/phrases were used during the search strategy:

This search which was carried out in accordance to the PRISMA guideline [8], was restricted to English publications.

Studies were included if:

Studies were excluded if:

In this analysis, pregnant women with SLE alone, pregnant women with PAPS and pregnant women with SAPS (most of the time it was associated with SLE) were included.

SLE is defined as an autoimmune disorder which affects mainly women of child-bearing age. There is no exact cause of SLE, however, genetic and environmental factors have shown to be among the causes. Painful swollen joints, malar rash, oral ulcers, photosensitivity, renal and cardiovascular symptoms and inflammation are among its manifestations.

APS also known as Hughes syndrome is defined as an autoimmune disorder with the presence of anti-phospholipid antibodies and anticardiolipin antibodies, manifesting as arterial and venous thrombosis and pregnancy related complications as the common symptoms. PAPS implies that the disorder is not due to and not co-existing with other disorders. However, SAPS implies that the disorder has been caused secondary to another disease.

Outcomes which were assessed through this analysis were:

Major outcomes were arterial/venous thrombosis, fetal loss and stillbirth, whereas the other outcomes were minor endpoints. The reported outcomes in patients with SLE versus PAPS and in patients with SLE versus SAPS were listed in Tables 1 and 2 respectively.

After a careful assessment of eligibility of the respective studies, the following information was extracted/collected by two independent reviewers (PKB, and MZSS):

These data were carefully cross-checked to ensure that no data was missing. Any disagreement which followed during this data collecting was resolved by the third author (FH).

Since all the eligible studies were observational studies, quality assessment was carried out by the Newcastle Ottawa Scale (NOS) [9] using a ‘star system’ method whereby stars were given based on certain assessment criteria. A maximum total number of nine stars were possible. Higher scores indicated better qualities of the studies.

Analytical software: RevMan version 5.3.

Statistical parameters: Risk ratios (RR) with 95% confidence intervals (CIs).

Interpretations: Heterogeneity is a major concern in meta-analyses [10]. To ensure consistency of the results, heterogeneity was assessed by the Q-statistic test whereby a P value less or equal to 0.05 would imply a statistically significant result. Heterogeneity was also assessed by the I² statistic test with a value less than 50% representing a low level of heterogeneity and a fixed effects model was used, whereas a value above 50% indicated a higher level of heterogeneity whereby a random effects model was used.

Sensitivity analysis was carried out by an ‘exclusion method’ whereby one study was excluded each time and the results which were obtained were observed for any significant deviation.

Publication bias which was another feature often encountered in a meta-analysis, was visually interpreted using funnel plots which were generated through the RevMan software.

This is a meta-analysis and ethical or board review approval was not required.

Following this search process, a total number of 1812 articles were obtained:

EMBASE database: 608;

MEDLINE database: 648;

Google Scholar: 527;

Official websites of specific journals which are related to rheumatology and obstetrics: 29.

Following an assessment of the titles and abstracts, which was an integral part of the eligibility criteria, 1747 articles were eliminated for irrelevancy.

Sixty-five (65) full-text articles were assessed for eligibility. However, further elimination was carried out based on the following conditions:

Finally, only 10 studies [11‐20] were selected for this meta-analysis as shown in Fig. 1.

The quality of the studies, which was assessed by the NOS has been shown in Table 3.

A total number of 941 participants were included in this analysis:

Type of studies: observational studies.

Patients’ enrollment period: 1986–2014 as shown in Table 4.

First of all, SLE was compared with APS (PAPS and SAPS). Results of this analysis showed APS to be associated with a significantly higher risk of fetal loss (RR: 4.49, 95% CI: 2.09–9.64; P = 0.0001) as shown in Fig. 2. In addition, stillbirth was also significantly increased with APS (RR: 6.65, 95% CI: 2.14–20.60; P = 0.001). Similarly, arterial/venous thrombosis was also significantly higher in the APS group (RR: 3.95, 95% CI: 1.28–12.16; P = 0.02) as shown in Fig. 3. However, infants who were low for gestational age, and preterm delivery were not significantly different (RR: 1.73, 95% CI: 0.77–3.87; P = 0.18), (RR: 1.07, 95% CI: 0.67–1.70; P = 0.79) and (RR: 0.53, 95% CI: 0.10–2.76; P = 0.45) respectively as shown in Fig. 2.

When patients with PAPS were compared with patients who suffered from SLE alone, fetal loss was still significantly higher with the former (RR: 4.89, 95% CI: 1.79–13.40; P = 0.002) as shown in Fig. 4. Arterial/venous thrombosis was also significantly higher in patients with PAPS (RR: 4.15, 95% CI: 1.00–17.16; P = 0.05) (Fig. 5).

When SAPS were compared with SLE, the current results showed arterial/venous thrombosis, and stillbirth to still be significantly higher with SAPS (RR: 7.73, 95% CI: 2.22–26.89; P = 0.001), and (RR: 8.07, 95% CI: 2.81–23.15; P = 0.0001) respectively (Fig. 6). However, infants who were low for gestational age were not significantly different between these two groups (RR: 0.98, 95% CI: 0.40–2.36; P = 0.96) (Fig. 6).

Fetal loss significantly favored SLE and was therefore significantly higher in patients with SAPS (RR: 5.92, 95% CI: 2.06–16.98; P = 0.0009) (Fig. 7). However, premature delivery was not significantly different (RR: 1.23, 95% CI: 0.54–2.80; P = 0.62) (Fig. 7).

When patients with PAPS were compared with patients who suffered from SAPS, no significant difference was observed in arterial/venous thrombosis (RR: 1.11, 95% CI: 0.86–1.43; P = 0.43) as shown in Fig. 8. In addition, fetal loss was also not significantly different between PAPS and SAPS (RR: 0.86, 95% CI: 0.34–2.21; P = 0.76) as shown in Fig. 9.

The overall results of this meta-analysis have been summarized in Table 5.

Sensitivity analysis which was carried out in all the sub-groups showed consistent results across the studies. Funnel plots which were graphically generated from the RevMan software, showed very little evidence of publication bias across all the studies that assessed clinical outcomes related especially to fetal loss, arterial/venous thrombosis and stillbirth as shown in Figs. 10 and 11.

The inclusion of observational data might have affected the results. However, several high-quality studies were obtained after an initial assessment of the methodological qualities, in addition to a low to moderate level of heterogeneity observed among the subgroups. Another limitation would be the restricted number of patients which might have affected the results. But it should be noted that, only a few research was carried out on this issue and therefore, only limited data were available to be used in this analysis. Another limitation was the fact that no data concerning medication use during the pregnancy stage was reported, and the influence of these medications on corresponding outcomes could not be assessed. The studies published by Tarr et al. and Munoz Rodriguez et al. showed a high number of patients with thrombosis. However, data about anti-coagulation treatments in these patients could not be extracted. A few studies showed the use of heparin and aspirin among the pregnant women. Use of these medications might also have affected the results of this analysis. In addition, a few cohorts such as the PROMISSE cohort, the French APS cohort could not be included because they involved indirect data that could not be used in this analysis. In addition, the study by Tarr et al. also included a minor number of patients who were not pregnant. This could be a limitation, however, the results were not affected due to a very small sample size of non-pregnant participants in that particular study. Also, it was possible that PAPS was a forerunner of SLE, and it remained un-noticed, and was considered as SAPS. This could have had an influence on the results too.