Minimal apparent diffusion coefficient value of the solid component to differentiate borderline and malignant ovarian epithelial tumours: a preliminary report

Introduction

Ovarian tumours are the second most common gynaecological cancer after cervical tumours. The prevalence increases with age, from 15-16 per 100,000 population in women aged 40-44 years to 57 per 100,000 population in women aged 70-74 years [1]. Epithelial tumours, which account for about 90% of all ovarian tumours, are divided into three types based on histopathology: benign, borderline, and malignant. Differentiating borderline from malignant tumours is crucial in determining the appropriate surgical intervention, as well as the need for adjuvant therapy. This distinction is also important to accommodate the reproductive capability of patients [1-3].
Studies commonly distinguished the benign and malignant ovarian tumours on magnetic resonance (MR) characteristics. However, results were varied [3,4]. Diffusion-weighted imaging (DWI) has been shown as a promising approach to determine whether ovarian tumours are benign or malignant. Li et al. [5] showed the difference using the apparent diffusion coefficient (ADC) on DWI, and it was 90.1% sensitive and 89.9% specific. Meanwhile, another study by Fuji et al. [6] suggested that the ADC was not significant in distinguishing benign from malignant ovarian tumours. Both studies were focused on benign-malignant differentiation [5,6].
Overall, it is still difficult to differentiate borderline from malignant tumours solely based on MR imaging [5-7]. Hence, this study focuses on the use of the minimal ADC (mADC) value to differentiate the borderline and malignant ovarian tumours.

Material and methods

Patient selection

This retrospective study was using secondary data from the Department of Anatomical Pathology. A total of 21 subjects from January 2016 until 2018, who were diagnosed with the borderline and malignant tumours based on histopathology and also underwent MRI using DWI-ADC technique, were included in this study.

Imaging technique and analysis

Abdominal and pelvic MR examination was performed using 1.5 T Avanto (Siemens Healthcare, Erlangen Germany) or Optima (General Electric, Boston, Massachusetts, USA). Standard T1-weighted images (repetition time [TR], 500s; echo time [TE], 9.5 ms) and T2-weighted images (TR, 4000 ms; TE, 95-250 ms) were performed. Gadobenate dimeglumine (MultiHance, Bracco, Milan, Italy) contrast was injected (0.1 mmol/kg of contrast material) via antecubital vein using a power injector at a rate of 2-3 ml/s. DWI was performed with b-values of 0, 800, and 1000 s/mm².
Minimum ADC values were acquired using free-hand range of interest (ROI) on the ADC map with images from other sequences to define the solid area. ROI included only the solid component of the tumour, performed in area of at least 10 mm².
Free-hand technique on the solid component of all 21 subjects was performed in consensus by two radiologists with 4 and 15 years of experience, who were unaware of the histopathological results. Tumours with pelvic tissue or other organ invasion, peritoneal metastases, or extra pelvic or distant metastases were excluded from the analysis.

Data analysis

All statistical analysis was performed using commercial software (SPSS version 24; IBM corporation, Armonk NY, USA). The intraclass correlation coefficient (ICC) was used to determine the reliability between the examinations. Mann-Whitney U test (abnormal distribution shown by Shapiro-Wilk test) was used to test the difference of mADC between borderline and malignant. Cut-off points were then obtained using the receiver operating characteristic (ROC) analysis.

Results

Figure 1 shows the solid component of ovarian tumour measurement by using free-hand ROI on the ADC map performed in an area of 10 mm². The results from two different radiologists were then analysed with intraclass correlation coefficient (ICC), and showed they excellent reliability in both borderline and malignant tumours (0.961 and 0.947, respectively) (Table 1).
There was a significant different (p = 0.001) in mADC value of borderline tumours and malignant tumours. ROC curve was then performed to obtain the area under the curve (AUC) and determine the optimal cut-off point (Figure 2).
Area under curve showed mADC as an excellent (0.92 ± 0.06, p = 0.001) variable to differentiate borderline from malignant tumours. Upon ROC analysis, the optimum mADC value was identified at 0.628 × 10–3 mm²/s (Figure 3) with sensitivity of 100%, specificity of 80%, positive predictive value (PPV) of 84.6%, and negative predictive value (NPV) of 100% (Table 2).

Discussion

This study showed a significant difference of mADC value between the borderline and malignant ovarian tumours. In previous studies, a lower ADC value correlated with a higher tumour cellular density [7-9]. Compared to benign tumours, malignant epithelial tumours showed an increased proliferation rate, which contributes to higher cellular density [9-11]. Thus, it is suggested that the solid component of the malignant tumour demonstrates a higher restricted diffusion and appears as a low signal intensity on ADC.
The solid component of both borderline and malignant tumours was then measured by two different radiologists using a free-hand technique on the ADC map. We used the intraclass correlation coefficient (ICC) to measure the reliability and agreement between the two examiners. The ICC value in our study was 0.961 and 0.947 for the borderline and malignant tumours, respectively. This means that the reliability is excellent [12]. This measurement was also used by Li et al. [13].
Several studies used the ROI selection approach to differentiate borderline and malignant ovarian tumours. Mimura et al. [14] used the semi-automatic ROI selection to gain the ADC value and distinguish the borderline from malignant ovarian tumours. This study showed a cut-off ADC value of 0.9 × 10–3 mm²/s with sensitivity of 61.9% and 93.8% [14]. The approach is different to the one used in our study because they used semi-automatic ROI, while we used a free-hand technique involving two different examiners.
The involvement of radiologists for the differentiation of the borderline and malignant ovarian tumours was also found in other studies. Denewar et al. [15] involved two radiologists as examiners to analyse the ROI of solid components on the ADC map with the area of ROI was 10 mm². There were 60 subjects included in this study, and the cut-off point between the borderline and malignant tumours was 1.53 × 10–3 mm²/s and showed 69% sensitivity and 81% specificity [15]. Likewise, Li et al. [13] used a similar method to evaluate the ROI of solid components. The total sample used in this study was 52 subjects, and the cut-off point was 1.36 × 10–3 mm²/s, with 88.2% sensitivity and 88.6% specificity [13]. The most recent study was conducted by Kim [16] in 2019. This study involved 70 subjects who underwent MRI examination. Of these, 63 subjects underwent DW-MRI examinations. Two radiologists were assigned to analyse the ROI of the solid component of tumours and determine the ADC value. The cut-off point between the borderline ovarian and malignant ovarian tumours was 1.05 × 10-3 mm²/s, with 74% sensitivity and 80% specificity [16].
With the exception of Mimura et al., who used a semi-automatic approach, the method of previous studies was similar to our study, i.e. two independent radiologists were assigned to evaluate the ADC value of solid components.
Our study showed the lowest result in the cut-off point of ADC value (0.628 × 10–3 mm²/s), which was statistically significant (p = 0.001). Moreover, this study showed the highest sensitivity and specificity among other previous studies. We suggest that a free-hand technique on the solid component of the ovarian tumours, the reliability between the two independent radiologists, and the histopathological data play an important role in differentiating borderline and malignant ovarian tumours.
This study has several limitations. This preliminary report used small size samples, which may affect the cut-off point, sensitivity, and specificity of this study. Another limitation is that we only measured the ADC value of the solid component of the ovarian tumours, while other studies also assessed other aspects such as the vessel permeability and the cell density of the solid components [15]. Lastly, although the ICC value showed excellent reliability and agreement between the two independent radiologists, a free-hand technique still requires further validation.

Conclusion

The minimum ADC value of the solid components can be valuable. This is a highly applicable method that can assist clinicians in considering the treatment approach. As of a preliminary report, we suggest further study using a larger sample size and similar method to determine the significance of the minimal ADC value in differentiating borderline and malignant ovarian tumours.

Conflict of interest

The authors report no conflict of interest.

References

1. Lalwani N, Shanbhoque AKP, Vikram R, et al. Current Update on Borderline Ovarian Neoplasms. Am J Roentgenol 2010; 194: 330-336.
2. Takeuchi M, Mayumi K, Nishitani H. Diffusion-weighted magnetic resonance imaging of ovarian tumors: differentiation of benign and malignant solid components of ovarian masses. J Comput Assist Tomogr 2010; 34: 173-176.
3. Hellyanti T, Tjahjadi H. Accuracy of frozen biopsy on ovarian tumor epithelial type. Indonesian Pathology Journal 2012; 21: 37-43.
4. Wakefield JC, Downey K, Kyriazi S, deSouza NM. New MR techniques in gynecologic cancer. Am J Roentgenol 2013; 200: 249-260.
5. Li W, Chu C, Cui Y, et al. Diffusion-weighted MRI: a useful technique to discriminate benign versus malignant ovarian surface epithelial tumors with solid and cystic components. Abdom Radiol 2012; 37: 897-903.
6. Fujii S, Kakite S, Nishihara K, et al. Diagnostic accuracy of diffusion-weighted imaging in differentiating benign from malignant ovarian lesions. J Magn Reson Imaging 2008; 28: 1149-1156.
7. Nasr E, Hamed I, Abbas I, Khalifa NM. Dynamic contrast enhanced MRI in correlation with diffusion weighted (DWI) MR for characterization of ovarian masses. Egypt J Radiol Nucl Med 2014; 45: 975-985.
8. Mansour S, Wessam R, Raafat M. Diffusion-weighted magnetic resonance imaging in the assessment of ovarian masses with suspicious features: strengths and challenges. Egypt J Radiol Nucl Med 2015; 46: 1279-1289.
9. Fischerova D, Zikan M, Dundr P, Cibula D. Diagnosis, treatment, and follow-up of borderline ovarian tumors. Oncologist 2012; 17: 1515-1533.
10. McCluggage WG. The pathology of and controversial aspects of ovarian borderline tumors. Curr Opin Oncol 2010; 22: 462-472.
11. Hauptmann S, Friedrich K, Redline R, Avril S. Ovarian borderline tumors in the 2014 WHO classification: evolving concepts and diagnostic criteria. Virchows Arch 2017; 470: 125-142.
12. Koo TK, Li MY. A guideline of selecting and reporting intraclass correlation coefficients for reliability research. J Chiropr Med 2016; 15: 155-163.
13. Li HM, Zhao SH, Qiang JW, et al. Diffusion kurtosis imaging for differentiating borderline from malignant epithelial ovarian tumors: a correlation with Ki-67 expression. J Magn Reson Imaging 2017; 46: 1499-1506.
14. Mimura R, Kato F, Tha KK, et al. Comparison between borderline ovarian tumors and carcinomas using semi-automated histogram analysis of diffusion-weighted imaging: focusing on solid components. Jpn J Radiol 2016; 34: 229-237.
15. Denewar FA, Takeuchi M, Urano M, et al. Multiparametric MRI for differentiation of borderline ovarian tumors from stage I malignant epithelial ovarian tumors using multivariate logistic regression analysis. Eur J Radiol 2017; 91: 116-123.
16. Kim SH. Assessment of solid components of borderline ovarian tumor and stage I carcinoma: added value of combined diffusion- and perfusion-weighted magnetic resonance imaging. Yeungnam Univ J Med 2019; 36: 231-240.