In this study, we identified five factors associated with adverse pregnancy outcomes in pregnant women with adenomyosis through univariate and multivariate analysis of the training and the validation cohorts: whether natural conception, previous parity, type of adenomyosis, history of infertility or adverse pregnancy outcomes, history of uterine body surgery. Among the above factors, in addition to the type of adenomyosis, the other four factors can also reflect the severity of the effect of adenomyosis on fertility and pregnancy outcomes. For example, univariate and multivariate analysis showed that the previous parity (≥ 1), which had the largest weight in the model, was a protective factor (OR of the previous parity < 1). It was not difficult to understand, because if a patient had a previous history of delivery, it indicated that although the patient had adenomyosis, it may have not had a serious adverse impact on the patient's fertility yet, the intrauterine environment was still suitable for embryonic development until delivery at a larger gestational week. On the contrary, univariate and multivariate analysis showed the history of infertility or adverse pregnancy outcomes was a risk factor for major adverse pregnancy outcomes. Patients with a history of infertility or adverse pregnancy outcomes indicated that adenomyosis may have already seriously affected the fertility of the patient so that the patient cannot conceive normally, or even if the patient was able to conceive, the patient's intrauterine environment may no longer be suitable for the development of embryos, resulting in repeated pregnancy but also repeated abortion. The patient may have multiple pregnancies but cannot be pregnant to the larger gestational week (after 28 weeks) to give birth (that was, the gravidity increased, but the parity did not increase). In conclusion, our analysis showed that "multiple pregnancies but few births" or "multiple pregnancies with no birth" may be an important influencing factor for adverse pregnancy outcomes in pregnant women with adenomyosis, in which whether patients have given birth in the past (i.e., previous parity ≥ 1 vs 0) was a directly related factor, because it reflected whether the intrauterine environment of most patients was suitable for embryonic development. Some studies have reported that the invasion of endometrium and interstitium in patients with adenomyosis destroyed the structure of the myometrium and the continuity of the uterine junction zone, changed the function of the myometrium, and affected the remodeling of the uterine spiral artery, thus it affected the normal implantation of the embryo. Secondly, myometrial fibrosis in patients with adenomyosis affected the physiological expansion of the uterus as the gestational age increased, which increased the risk of miscarriage and premature birth in patients with adenomyosis.[16–18] These studies explained the pathophysiological mechanism of adenomyosis affecting the fertility of patients, and the previous parity (≥ 1 vs 0) was an important "phenotype" that reflected whether the intrauterine environment was suitable for embryonic development. Similarly, whether the patient is naturally pregnant or not, the history of previous uterine body surgery such as conservative adenomyosis lesion resection can also laterally reflect the severity of adenomyosis and the impact of adenomyosis on embryo implantation and development.[4, 8, 19]
Based on the above situation, we established a nomogram model including the above five factors. This model can provide a comprehensive and individualized prediction of the risk of adverse pregnancy outcomes for each pregnant woman with adenomyosis, which was undoubtedly very interesting and practical. Next, we stratified the patients according to the optimal risk threshold of the model. The probability of adverse pregnancy outcomes in patients in the high-risk group was undoubtedly much higher than that in the low-risk group. This suggested that in addition to routine obstetric care and monitoring should be applied for each pregnant woman at risk of adverse pregnancy outcomes, we also need to develop a personalized diagnosis and treatment plan for them. For patients with high risk of adverse pregnancy outcomes, management and monitoring should be strengthened in the first and second trimesters of pregnancy, and the number of obstetric examinations and the items of obstetric examination should be appropriately increased, especially after 12 weeks of gestation. Because we found that the time of miscarriage in this study was mostly concentrated after 12 weeks of gestation, which may be caused by the above-mentioned myometrial fibrosis in patients with adenomyosis, with the increase of gestational weeks gestational age, the mechanical physiological expansion of the uterus was affected. Particularly, we found that for patients in the high-risk group, whether in the training cohort or in the validation cohort, the probability of developing pregnancy related complications (hypertensive disorder complicating pregnancy, PPROM, GDM, postpartum hemorrhage, etc.) was also much higher than that in the low-risk group, which undoubtedly increased the risk of adverse pregnancy outcomes on another level, thus entered a "vicious circle".[7, 20] Therefore, it suggested that low-dose aspirin, heparin, and immunosuppressants may also should be used in early pregnancy to improve uterine blood perfusion for pregnant women in the high-risk group of adverse pregnancy outcomes. The duration and dose of administration should be adjusted according to the effect on pregnancy outcomes to reduce the probability of related complications such as preeclampsia, SGA.[17, 21] At the same time, obstetricians should closely monitor the patient's various indicators (including various biochemical indicators and imaging examinations), be wary of the occurrence of complications (such as gestational hypertension, PPROM, GDM, etc.), and strengthen prediction and prevention of adverse pregnancy outcomes. If complications have occurred, related complications should be actively treated according to relevant guidelines.[22] For the perinatal management of patients in the high-risk group, the choice of delivery timing and delivery method should consider multidisciplinary (obstetrics, gynecology, imaging, etc.) consultation and be carefully decided according to the specific conditions of the patients. It should be noted that pregnant women with a history of uterine body surgery (such as adenomyomectomy or adenomyosis lesion resection in the past) had a progressively higher risk of uterine rupture with increasing gestational age (3 patients with uterine rupture in the training and validation cohort had a history of adenomyosis or adenomyoma surgery in the past), and these patients should be treated as high-risk pregnancy management regardless of the risk predicted by the model. Such pregnant women should terminate their pregnancy by caesarean section (especially in patients conceived through ART).[5, 23, 24] For intrapartum and postpartum management of patients, it had been reported in the literature that pregnant women with adenomyosis were a high-risk group for postpartum hemorrhage. In our study, the proportion of postpartum hemorrhage in the high-risk group was much greater than that in the low-risk group. Some patients should be alert to the occurrence of postpartum hemorrhage or even severe and refractory postpartum hemorrhage and should be managed served as the high-risk group of postpartum hemorrhage during delivery to strengthen prevention. In summary, for patients in the high-risk group, we should pay more attention to the management of the whole process of pregnancy.[21, 25]
This study has significant advantages. First of all, the model can predict the risk of adverse pregnancy outcomes in all pregnant women with adenomyosis. Second, the nomogram model used only 5 readily available clinicopathological factors (previous parity, natural conception or not, type of adenomyosis, history of infertility or adverse pregnancy outcomes and history of uterine body surgery) to predict the risk of adverse pregnancy outcomes in pregnant women with adenomyosis accurately. Finally, our model was externally validated and showed good accuracy and stability. This model can help clinicians to improve pregnancy outcomes by early intervention in pregnant women with adenomyosis.
Of course, there were certain limitations in this study. This study was a retrospective study, and the exclusion of some cases in the access to clinical data may lead to selective bias, so we need more prospective studies to validate it further. In addition, the factors currently included in the model in this study were limited to clinical indicators, and there was a lack of objective pathophysiological, serological, and imaging indicators, and more predictive indicators should be included in the future to increase the performance of the model.