108 Additionally, in the external cohort, patients with adenomyosis more often had a history of endometriosis compared to patients without adenomyosis, as diagnosed on MRI or laparoscopy (56.4% vs. 38.5%, p=0.014). This difference was not statistically significant in the original dataset. Several studies have shown an association between adenomyosis and endometriosis. There is ongoing debate regarding if they should be seen as two separate conditions or as one disease spectrum with regard to their pathophysiological mechanisms (72,209–211). These differences in endometriosis prevalence between the datasets may have implications for the model’s performance, given the clinical overlap in symptoms between endometriosis and adenomyosis, such as dysmenorrhoea. Furthermore, in terms of symptoms, dysmenorrhoea was found to be statistically significantly between groups (66.2% vs. 44.3%, p = 0.011). Despite the fact this symptom was not significantly different in the original dataset, it was still included in the model, since dysmenorrhoea is the most commonly reported complaint in adenomyosis. The difference between the datasets can be explained by the subjective nature of this symptom. The fact that more patients had a history of endometriosis in the adenomyosis group may also contribute to this. Since the prevalence of dysmenorrhoea in the original dataset was higher in the adenomyosis group (73.3% vs. 63.1%, p = 0.491), despite not being significantly different, this may not significantly affect the model’s performance however. As for MRI characteristics, both datasets showed statistically significant differences for the same variables. It is important to note that the reliability of the JZ thickness on MRI as a predictor of adenomyosis is debatable. Several studies have reported a high reliability of the JZ thickness on MRI as predictor of adenomyosis, especially the use of a JZ thickness of >12 mm (33,47). However, recent studies suggest that JZ thickness may not be a reliable predictor for adenomyosis on its own and that symptoms and other MRI features should be weigh more heavily in adenomyosis (29,45,198). This was taken into account in the initial study. As a result, several factors were included in the model. Moreover, as in initial study, the presence of HSI foci was the strongest predictive MRI feature in this study. This is in accordance with previous studies that describe these HSI foci as main indicator for
RkJQdWJsaXNoZXIy MTk4NDMw