We enrolled patients who visited Qilu Hospital, Shandong University, between September 2021 and October 2022. The inclusion criteria for the study group were as follows: 1) patients diagnosed with adenomyosis based on clinical symptoms and imaging findings such as ultrasound, 2) premenopausal patients aged 18–50 years, 3) patients with dysmenorrhea with or without menorrhagia, 4) patients who had not received hormone therapy in the last three months, and 5) patients without reproductive system malignancies or pelvic inflammatory disease. Premenopausal patients without myometrial lesions or a history of reproductive system malignancies or infection were included in the control group. Further, non-sexual or menopausal women, patients with a previous or current history of reproductive system malignancies, and pregnant patients or those with coinfection of the reproductive system were excluded from the study. Finally, we recruited 39 premenopausal women as the control group and 57 premenopausal women as the adenomyosis group.

Demographic data and detailed medical histories were recorded before TVUS. Demographic information, including age, body mass index (BMI; kg/m2), age at menarche, gravidity, parity, mode of delivery, frequency and duration of menstrual periods, amount of menses, degree of dysmenorrhea, medication, and surgical history, were collected and recorded using questionnaires. All patients’ medical records, including intraoperative and pathological findings, were also recorded. The degree of dysmenorrhea was quantified using a numerical rating scale (NRS) from 0–10, with 0 representing no pain and 10 representing maximum pain. NRS scores were grouped into three levels: 0–3, none or mild; 4–6, moderate; and 7–10, severe. The amount of menses was recorded as mild, moderate, or severe based on the subjective evaluation of the patients.

The eutopic endometrium and adenomyotic tissue samples were collected from 20 patients with adenomyosis who underwent surgery. For the control group, specimens of the eutopic endometrial and normal myometrial tissues were obtained from 15 patients who underwent hysterectomy.

This study was approved by the ethics committee of the Medical Integration and Practice Center, Shandong University (approval number SDULCLL2022-1–21). All patients included in the study signed informed consent forms.

TVUS and strain elastography were performed by a single-trained ultrasound specialist with several years of experience in gynaecological sonography, especially for adenomyosis. All enrolled patients underwent both conventional TVUS and strain elastography using the Nuewa R9 with a DE10-3WU transvaginal probe (Mindray, Shenzhen, China) less than 1 d preoperatively or during the outpatient examination. The sonographer was blinded to patients’ clinical information. Conventional TVUS was initially performed to diagnose adenomyosis according to the Morphological Uterus Sonographic Assessment criteria [13, 14]. The following information was recorded: volume of the uterus, thickness of the anterior and posterior walls, site and extent of typical adenomyotic lesions, and presence of ovarian endomyoma, endometrial polyps, and uterine fibroids. The volume of the uterus was recorded in the control group. The uterine and lesion volumes were calculated using the following formula for an ovoid: volume = D1 × D2 × D3 × 0.52, where D1, D2, and D3 represent the vertical, transverse, and anteroposterior diameters of the uterus or lesion, respectively.

Strain elastography was subsequently performed on all recruited patients. The elastograms of typical lesions were visualized in real time, followed by a B-mode image. To evaluate the strain ratio of the lesion, we applied external pressure using an ultrasound probe to produce a deformation. Three cycles of gentle compression and decompression were then performed. The elastogram images were colour-coded as red, yellow, green, and blue. The different colours represented the tissue stiffness relative to that of the endometrium or adjacent bowel. Blue represented the softest tissue, and red represented the hardest tissue. For the adenomyosis group, a region of interest (ROI) was set in the typical lesion area of the uterus, while another region of interest (Ref) was set in the adjacent normal myometrium. The stiffness of the lesion was semi-quantified using the ratio of Ref/ROI. The higher the ratio, the greater the stiffness of the lesion. For the control group, we set two regions of interest (ROI1 and ROI2) in the uterus, and the ratio of ROI1/ROI2 represented the stiffness of the normal myometrium.

To eliminate possible bias and maintain consistency, three strain ratios were measured for each patient, and the mean value was used as the stiffness of the lesion or normal myometrium.

Serial 4-mm sections were obtained from each block. Routine deparaffinization and rehydration were performed. The first resultant slide was stained with hematoxylin and eosin (H&E) to confirm the pathological diagnosis.

Then, adenomyotic lesions and the eutopic endometrium of patients with adenomyosis and the control group underwent IHC staining for TGF-β, α-SMA, and PGP9.5. For antigen retrieval, the EDTA buffer (pH = 8.0) was microwaved on high for 5 min to a boil, and then the sections were heated in EDTA buffer on low for 15 min and cooled to room temperature. Next, the sections were treated with 3% hydrogen peroxide to block the activity of endogenous peroxidase for 30 min at 37℃. After the samples were blocked with 5% bovine serum albumin (BSA; Boster, Wuhan, China) for 50 min at 37℃, they were incubated with the primary antibody against TGF-β (1:100; Abcam, Cambridge, England), α-SMA (1:100; Abcam), and PGP9.5 (1:250; Abcam) overnight at 4℃. The sections were rinsed three times with PBS buffer and incubated with the horseradish peroxidase (HRP)-labelled secondary anti-rabbit/mouse antibody for 30 min at 37℃. The sections were then washed with PBS and treated with glucose oxidase-diaminobenzidine for microscopic observation. Finally, they were incubated with hematoxylin for 30 s, differentiated, and stained with anti-blue. The expression level of the targeted substance was semi-quantitatively calculated based on the proportion of positive cell areas using ImageJ software (National Institutes of Health, USA). To reduce bias and ensure reliability, a series of 3–5 randomly selected images for every section were taken to obtain a mean value.

Masson’s trichrome staining was used to detect collagen fibres in the lesion tissue in adenomyosis samples and the normal myometrium in control samples; it was performed according to common protocols using a Masson’s trichrome staining kit (Solarbio, Beijing, China). After Masson’s trichrome staining, images were observed and captured using a microscope. Collagen fibres appeared blue, whereas smooth muscle fibres and red blood cells appeared red. The extent of fibrosis was represented by the relative content of collagen fibres, which was evaluated by the ratio of positive fibres to the total tissue area and calculated using Image-Pro Plus 6.0 (Media Cybernetics Inc., Bethesda, Massachusetts, USA).

Statistical analyses were conducted using SPSS Statistics version 26.0 (IBM Corp., Armonk, NY, USA) and GraphPad Prism 8.0.1 software. All continuous variables are expressed as mean ± standard deviation (SD) or median (interquartile [IQR]), depending on whether these data conform to a normal distribution, as assessed by the Shapiro–Wilk test. Categorical variables are presented as frequencies. Continuous variables conforming to a normal distribution between two groups were compared using the Student’s t-test, while continuous variables conforming to a normal distribution among three or more groups were compared using a one-way analysis of variance (ANOVA). For continuous variables that did not conform to a normal distribution, a nonparametric rank sum test was performed. Pearson’s X2 or Fisher’s exact test was used to compare differences in categorical variables. Correlations between tissue stiffness and dysmenorrhea and between the extent of fibrosis and the level of TGF-β, α-SMA, and PGP9.5 were assessed by Pearson’s coefficient. P-values < 0.05 were considered statistically significant.