In this cohort, we included 156 patients with fetal death at ≥ 10 GW. Twelve patients did not complete initial aPL testing, and 18 patients did not complete repeat aPL testing due to loss to follow-up or distance constraints. One hundred and twenty-six patients were therefore included in the diagnostic analysis of APS. Maternal APS was diagnosed per the 2006 Sydney classification criteria in 10.3% (13/126) of the cohort. Similar proportion was noted in a case–control study that had positive tests for aPLs (aCL or aβ2GPI) in 9.6% of stillbirths [12]. However, that study included patients without LA testing and did not include those who were fetal death at 10–20 GW. In out cohort, the proportion of APS was slightly higher than the 8.8% (12/136) of FCA although the difference was not statistically significant. The diagnosis of FCA did not include those chromosomal deletion or duplication at < 5 Mb because there is no convincing evidence that those microdeletion / microduplication are associated with fetal death although the variants classified as pathogenic according to American College of Medical Genetics (ACMG) criteria.

aPLs are the crucial diagnostic marker of APS. In our study, aβ2GP-1 IgM exhibited high diagnostic performance for the diagnosis of APS in patients experienced fetal death (AUC, 0.918 [95% CI, 0.79–1.00]). This result is consistent with the result that isolated aPLs IgM positivity is associated with pregnancy morbidity[13]. One study has reported that aβ2GP-1 isotypes are associated with the lowest live birth rate and highest incidence of stillbirth, compared with the presence of aCL or LA alone[14]. Thus, aPL testing should be emphasized more and the positive aβ2GP-1 IgM should be given attention in women who experienced fetal death.

During follow-up, of 13 patients with APS, seven were persistently positive for aPLs, four exhibited fluctuation, and one had negative conversion. Some scholars have reported that the rate of aPL positivity decreases during follow-up in primary APS, estimating that seroconversion occurs in between 8.9 and 59% of patients over time, and hydroxychloroquine has been identified as the most effective pharmacological agent to reduce aPL titers [15]. Another study found that aPL titers decreased modestly during pregnancy among patients who were positive [16]. We could not conduct an analysis based on GW or interventions due to the small number of patients in these subgroups. But from real-world individual observations, we noticed that the fluctuations of aPLs (solid-phase assays) were independent of the gestational stage and clinical interventions (Table 2, column 8 and 9). The factors contributing to the fluctuation of aPLs were unknown. Notably, clinicians need to further monitor individuals who were suspicious for APS but have only one time negative aPL result.

Four patients with subsequent pregnancies received therapy according to guidelines and delivered at term. None of the enrolled patients experienced a thrombotic event. A retrospective study found that aPLs at low titers and fulfilling the 2006 Sydney criteria were associated with pregnancy morbidity and that treatment by LDA plus LMWH appeared to improve outcomes [17]. Furthermore, there is evidence that isolated IgM was rare in thrombotic APS, but more frequent in obstetric APS [13]. Our results, with a preponderance of positive aβ2GP-1 IgM and relatively infrequent positive LA, supported these conclusions.

Our study was initiated in 2021. In 2023, new classification criteria for APS, the ACR/EULAR criteria were published. When the ACR/EULAR criteria were used in our cohort, none of the patients fully met the classification conditions for APS. An international study reported that the percentage of obstetric APS dropped from 26.9% per Sydney criteria to 3.2% per ACR/EULAR criteria [18]. This observation coupled with our results indicate that the new criteria may have an extremely low sensitivity to obstetric APS. The ACR/EULAR criteria assign a low weight to fetal death in the absence of preeclampsia and/or placental insufficiency [19]. In our cohort, half of APS patients were observed before 20 weeks of gestation. Experience holds that preeclampsia and placental insufficiency cannot manifest before 20 weeks of pregnancy [5]. Clearly, although the new 2023 ACR/EULAR classification criteria for APS were established, they should not be adhere dogmatically in patients experienced fetal death.

The strengths of our study include the prospective design, relatively comprehensive etiological investigation, longitudinal follow-up data for aPL serostatus changes over time and pregnancy outcomes post-diagnosis, which can guide clinical expectations and inform future research. In this study, the GW was carefully assessed to be close to the time of occurrence of fetal death rather than that of recognition. Our findings from real clinical settings have provided insight into APS diagnosis and management in cases of fetal death, which can be useful for clinicians treating similar cases.

Our study has several limitations. First, it was the absence of a control cohort with viable pregnancies for comparison, which prevented us from establishing a stronger association between aPLs and fetal death. Second, LA detection in our study did not use two tests with different assaying principles, which may have decreased the sensitivity of LA, contributing to the inconsistency between our current finding and a previous report that LA is the best predictor of adverse pregnancy outcomes [20]. Additionally, the generalizability of our results may be limited due to the relatively homogeneous study population and a relatively small sample size.

In conclusion, maternal APS was found to be one of the important causes of fetal death, contributing 10.3% of cases of fetal death at ≥ 10 weeks of gestation, slight ahead of fetal chromosomal abnormalities. Follow-up indicated that the aPL serostatus may fluctuate significantly in some patients with APS.