Adenomyosis and IVF — A Story Full of Holes

Adenomyosis is defined as the benign invasion of endometrial glands and stroma into the myometrium of the uterus (Bird et al., 1972). While first described in 1860 by von Rokitansky and named in 1925 by Frankl (Benagiano and Brosens, 2006), adenomyosis has remained under-represented in research.

Our understanding of this disease has been, in part, stymied by vague and inconsistent diagnostic criteria. Adenomyosis is diagnosed based on the gold standard of histology of the uterus following hysterectomy. However, histological diagnosis is not standardised. There is a lack of consensus regarding the diagnostic criteria, including the site and number of biopsies and the depth of gland/stromal invasion required (McCaughey et al., 2024). Calls for international standardisation of these diagnostic criteria have resulted in a Delphi Consensus, which is currently underway (McCaughey et al., 2024).

Surrogate diagnostic tests have aimed to provide identification of adenomyosis in the absence of histology. This has been particularly relevant where individuals desire current or future fertility. Imaging evidence of adenomyosis (IEOA) is defined using ultrasound and magnetic resonance imaging (MRI). Consensus regarding the diagnostic criteria for ultrasound-based diagnosis of adenomyosis was established by the Morphological Uterus Sonographic Assessment (MUSA) group (Van den Bosch et al., 2015). In 2022, these criteria were modified due to poor inter-rater agreement (Harmsen et al., 2022). MRI has comparable sensitivity and specificity to ultrasound (Alcázar et al., 2023). Nevertheless, both MRI and ultrasound lack sensitivity and specificity when compared to histology (Chapron et al., 2020).

Recent studies have attempted to demonstrate that adenomyosis is a predictor of poor IVF outcomes. Data pertaining to ‘adenomyosis’ and IVF outcomes have been produced based on a diagnosis of adenomyosis made using IEOA. IEOA is accessible, and, unlike obtaining histological evidence of adenomyosis through either hysterectomy or uterine biopsy, IEOA can be obtained without a negative impact on fertility.

However, caution should be used when drawing conclusions based on data regarding IEOA and IVF outcomes. The low sensitivity and specificity of IEOA when compared to histology suggest that the features that IEOA describes may not be unique to adenomyosis. This raises the possibility that IEOA may be seen in the presence of other gynaecological conditions, and, as such, the cause for a change in IVF outcomes becomes less clear.

Moreover, while many studies have explored IEOA and IVF outcomes, only six studies have a prospective design (Benaglia et al., 2013; Cozzolino et al., 2024; Higgins et al., 2021; Mavrelos et al., 2017; Neal et al., 2020; Salim et al., 2012). The distinction between correlation and causation is of relevance when reviewing the studies exploring the association between IEOA and IVF outcomes. Unless specifically designed to do so, retrospective studies are not able to infer causation. As such, prospectively collected data are paramount. The six prospective studies assess fertility outcomes in 654 women with IEOA and 1914 controls. Methodological limitations and heterogeneity predominate.

While diagnosis is based on ultrasound in all studies, IEOA is variably defined. Only two studies use the MUSA features of adenomyosis (Higgins et al., 2021; Neal et al., 2020). The minimum number of features of IEOA required for a positive diagnosis varies from 1 (Cozzolino et al., 2024; Higgins et al., 2021; Mavrelos et al., 2017; Neal et al., 2020) to 2 (Benaglia et al., 2013).

Potential confounders are managed inconsistently. While all six studies limit inclusion to the first IVF cycle, potential confounders include patient age at oocyte retrieval (oocyte age), presence of other gynaecological pathologies (endometriosis, fibroids, endometrial polyps and hydrosalpinges), embryo development to cleavage versus blastocyst (embryo age) and number of embryos transferred. One study prospectively matches controls based on oocyte age, number of embryos transferred, embryo age and study period (Benaglia et al., 2013), while three studies adjust for potential confounders using logistic regression modelling (Cozzolino et al., 2024; Higgins et al., 2021; Neal et al., 2020). All but one study (Higgins et al., 2021) exclude subjects with hydrosalpinges. However, there is greater variation in tolerance of coexistent fibroids and variation in oocyte age, number of embryos transferred and embryo age.

The primary outcome for four studies is live birth rate — there is a general agreement that live birth rate should be the primary outcome for IVF studies. However, it is less clear as to what this measure is applied: stimulation cycle commenced, cycle where embryo transfer occurred or other. All six studies list clinical pregnancy as an outcome. Forest plots displaying the clinical pregnancy and live birth outcomes for these studies are shown in Figs. 1 and 2 (prepared using STATA version 18). Of note, these are raw values. The lack of adjustment for potential confounders, as would be seen in randomised controlled trial data, is likely to affect the results. In particular, increasing oocyte age and coexistent fibroids would potentially reduce IVF success, contributing to a false impression of impact from adenomyosis.

It is on this basis, then, that the following becomes clear.

First, it should be required that a consistent definition is used when diagnosing IEOA. Regarding ultrasound, as the MUSA group consensus exists, this should be used until a better evidence-based definition is published. There is a need to clarify what sonographic signs (if any) are associated with poorer IVF outcomes.

Second, more prospective studies are required to assess the interaction between IEOA and fertility outcomes. Studies need to be designed with adequate power (using the above summary odds ratios as a basis for the calculation) and should adjust for the main potential confounders, including but not limited to oocyte age, presence of endometriosis, presence of fibroids, embryo age, embryo ploidy status, number of embryos transferred and fresh versus thawed status. Given the relatively small changes in outcome rates being tested, sample sizes are likely to be large, making multi-site studies necessary. There is no value in the publication of further retrospective data on this subject.

Third, while the question of whether IEOA has an impact on fertility outcomes remains, exploration of interventions to treat adenomyosis with a view to aiding IVF success is ethically questionable. The limited evidence for IEOA and for its association with altered IVF outcomes needs to be recognised rather than ignored.

CONFLICTS OF INTEREST

MH obtains income as a fertility specialist and is a shareholder in an IVF clinic. HM has no conflicts of interest.

ORCID

Martin Healey  https://orcid.org/0000-0003-2832-4747

Helen C. McNamara  https://orcid.org/0000-0003-1867-3142