The value of DCE-MRI in assessing histopathological and molecular biological features in induced rat epithelial ovarian carcinomas

This study was approved by Institutional Review Board of Jinshan Hospital of Fudan University and performed in strict accordance with the Guide for the Care and Use of Laboratory Animals of the National Science and Technology Committee of China. Fourty-six female SD rats (8 weeks old), Shanghai Laboratory Animal Research Center, Shanghai, China) were fed a standard diet and were housed in pathogen-free facilities at 25 ± 1 °C. The right ovary of each rat was surgically exposed, packed a cloth strip (5 mm × 5 mm) coated 2 mg 7,12-dimethylbenz[A] anthracene (DMBA, Sigma Chemical Company, St. Louis, MO, USA), and closed with the surround fatty substance. Benzylpenicillin injection of 10⁵ units was administered intraperitoneally to prevent infection before the abdominal wall was closed. The surgical procedures were referred to previous study [9]. Based on our preliminary experiment, rat EOCs in situ model were successfully established at 180 days after DMBA exposure.

One hundred and eighty days after DMBA induction, MRI was carried out using a 3.0 T scanner (Verio, Siemens Healthcare, Erlangen, Germany) and a coil specially designed for rats. First, routine axial T1- and T2-weighted imagings with and without fat faturation were performed to detect the lesions. Second, DCE-MRI using fast low angle shot-2D (FLASH 2D) T1WI with fat saturation (TR/TE, 7.92 ms/2.28 ms) was performed in axial, sagittal and coronal planes after the caudal intravenous administration of 0.2 mmol/kg gadopentetate dimeglumine (Gd-DTPA, Magnevist; Bayer Schering, Berlin, Germany) and at a rate of 0.3 ml/s. A total of 40 phases of images were sequentially acquired with a temporal of 6 s. The scanning parameters were as follows: slice thickness = 1.0 mm, flip angle = 15°, gap = 1 mm, matrix size = 224 × 370, field of view = 80 mm × 62.5 mm and excitation = 4. The total scan duration was 4 min.

Using Tissue 4D software (Siemens, Erlangen, Germany) and two-compartment (Tofts) modeling, the DCE-MRI images were evaluated by two blinded radiologists each with 6 years of experience in pelvic MRI [10]. The values of the image signal intensity were measured in the slice with a maximum diameter of ovary lesions. By avoiding vessels and flow artifacts as much as possible, three circular regions of interest (ROI) (20–50 mm²) were selected in the solid areas of every lesion, achieved time-signal intensity curves (TICs) and generating the following quantitative parameters automatically: 1) volume transfer constant, K^trans; 2) reverse volume transfer constant, K_ep; and 3) extravascular extracellular space (EES) volume per unit volume of tissue, Ve (the relationship among these three parameters was as follows: K_ep = K^trans/V_e) [11]; and 4) initial areas under the time-intensity curve, IAUC. According to the study of Thomassin-Naggara [12], TIC was divided into three types:type I was defined as a gradual increase without a well-defined shoulder; type II was defined as a moderate initial increase relative to that of myometrium followed by a plateau; type III was defined as an initial increase steeper followed by a plateau.

At target time point, rats were immunized by intraperitoneal injection of 10% chloral hydrate, and blood was then taken from the heart. After centrifugation, the supernatants were collected and analyzed for the evaluation of cancer antigen 125 (CA125) in murine serum by ELISA (R&D Systems, Minneapolis, MN) according to the manufacturer’s instructions.

After MRI scanning, all rats were sacrificed, dissected and examined for gross abnormalities. Macroscopical specimen of entire reproductive tracts were harvested from the rats and fixed in 10% formalin. The specimen was dissected into sections at 1-3 mm interval by L.W., with 10 years of experience in human and murine gynecological pathology. The tumors were analyzed for morphology, and then the tissues were embedded in paraffin and cut into 3-μm-thick slices. Hematoxylin-eosin (HE) staining was performed to observe the histological type. Immunohistochemical (IHC) (Primary antibodies of CD31, VEGF and Ki67 purchased from Cell Signaling Technology, CA, USA) was applied to investigate the MVD, expression of VEGF and Ki67 in positive cells in EOCs. MVD was analyzed in CD31-stained vascular endothelial cells and determined as the mean microvessel at 3 microscopic fields of vision (HPF, × 200). VEGF expression displayed as brownish yellow granules in the cytoplasm and intercellular spaces, and determined as the mean integrated optical density (IOD) measured by Image-Pro Plus image analysis software at 3 HPFs (×200). Ki67-positive expression indicated the appearance of yellow or brown granules in the nucleus and determined as the percentage of positive tumor cells in 1000 tumor cells assessed at 10 HPFs (×200) with 1000 random cells.

Data were analyzed using SPSS 22.0, and values were represented as the means ± standard deviation. Analysis of variance was used for data analysis between multiple groups, and the values between every two groups were analyzed by independent sample T test. The ordinal date was analyzed by rank sum test. Pearson’s correlation was used to analyze the relationship between the DCE-MRI parameters and VEGF, MVD, Ki67 and CA125 expression as follows: a correlation coefficient between 0.75 and 1.00 was considered highly relevant, that between 0.50 and 0.74 was considered moderately relevant, that between 0.25 and 0.49 was considered weakly relevant and that ≤0.249 was not considered relevant. [13]. For T test and rank sum test, p < 0.05 was considered significant; for correlation analysis, p < 0.01 was considered significant.

Based on previous study and our preliminary experiment, there was best result of induced EOCs and survival rates at 180 days after DMBA implantation. Fourty-four of 46 rats survived and 42 of 44 developed ovarian lesions. All cases were confirmed by postoperative pathology: 1 cyst, 1 cystadenoma, 35 EOCs (19 serous adenocarcinomas, 9 mucous, 4 mixed, 2 endometrioid, 1 clear cell carcinoma) and 5 other non-epitholial malignancies, Thirty-five EOCs accounted for 87.5% of all malignant ovarian tumors. Thirty-five EOCs accounted for 87.5% of all ovarian malignancies, The maximum diameter of tumors ranged from 15.2 ~ 31.0 mm with mean diameter of 25.5 ± 0.76 mm. Tumors appeared as cystic in 6 cases, solid-cystic in 25 cases and solid in 4 cases on conventional MRI and were proven by pathology.

DCE-MRI showed none of type I(0%) TIC, 6 type II (17.14%), and 29 type III (82.86%), with statistical significance (p<0.001). Compared with tumors with curve type II TIC, tumors with type III showed significantly higher histological grades (p<0.01, Table 1).

The mean K^trans value was 0.22 ± 0.06, mean K_ep value was 0.48 ± 0.13, mean V_e value was 0.51 ± 0.12 and mean IAUC value was 27.62 ± 8.15 in 35 rat EOCs. DCE-MRI parameters in different histological grades are summarized in Table 2. The mean K^trans, K_ep, and IAUC values significantly increased along with tumor grades increasing. There were significant differences in the overall comparison and in pairwise comparisons (all p < 0.05) except for IAUC in grade 2 vs. grade 3. There was no significant difference in the mean V_e values among the three grades (p > 0.05).

As shown in Table 3, the MVD, expressions of VEGF and Ki67-positive tumor cells and CA 125 level between tumors with type II and type III TIC showed statistical significance (p<0.01) (Figs. 1, 2, 3 and 4).

As shown in Table 4, the mean K^trans was highly positively correlated with MVD (23.46 ± 3.602) and VEGF ((32.13 ± 17.68) × 10³) and was moderately positively correlated with the Ki67-positive cell proportion (34.91 ± 12.70) % and CA125 level (33.13 ± 82.17) U/mL (all p<0.01); K_ep was moderately positively correlated with MVD, VEGF, Ki67 and weakly positively correlated with CA125 (all p<0.01); no significant correlation was observed between V_e and MVD, VEGF, Ki67 or CA125 (all p>0.05); IAUC was highly positively correlated with VEGF, moderately positively correlated with MVD and Ki67, and weakly positively correlated with CA125 (all p<0.01).