Serous borderline ovarian tumours: an extensive review on MR imaging features

Abstract

Serous borderline ovarian tumours (SBOTs) are an intermediate group of neoplasms, which have features between benign and malignant ovarian tumours and for which, fertility-sparing surgery can be offered. MRI in imaging of SBOTs is, therefore, crucial in raising the possibility of the diagnosis, in order to present the patient with the most appropriate treatment options. There are characteristic MRI features that SBOTs demonstrate. In addition, recent advanced techniques, and further classification into subtypes within the borderline group have been developed. The aim of this article is to review the MRI features of SBOT and provide the reporter with an awareness of the imaging tips and tricks in the differential diagnosis of SBOT.

Introduction

Borderline ovarian tumours (BOT) account for 15–20% of non-benign ovarian neoplasms.¹ These tumours show low malignant potential with intermediate mitotic activity, nuclear atypia and lack of obvious invasion of the ovarian stroma.² Pathologically, they are classified according to the type of surface epithelial cell.³ The serous and mucinous BOTs are the most common subtypes, which account for approximately 50–55% and 35–45% of all BOTs, respectively.^4,5

The genetic profile of ovarian tumours underpins the dual pathway of carcinogenesis.⁶ Serous borderline tumours (SBOT) demonstrate similar genetic and molecular alterations as low-grade serous carcinomas (LGSC) with sequence mutations in KRAS and BRAF oncogenes and very low number of DNA copy number changes.^4,5,7 This supports the hypothesis that LGSC develops from SBOT with specific mutations of KRAS and BRAF with activation of the mitogen-activated protein kinase (MAPK) signal transduction pathway.^6,7 On the contrary, most of the advanced stage, high-grade serous carcinomas have mutation of p53 with high level of DNA copy number changes.^4,6,7 In this second pathway, high-grade serous carcinomas are thought to develop de novo from surface epithelium of the ovary or from surface inclusion cysts.⁷

About 9–15% of all serous neoplasms are known to be the borderline type.² The SBOTs are usually seen in a younger age group compared to invasive malignancy, with patients typically presenting between 40 and 50 years old.³ In addition, patients also tend to present at an early stage with a better prognosis when compared with malignant ovarian tumours.⁸ In approximately one-third of SBOTs, peritoneal implants accompany the adnexal mass.⁹ Extra ovarian invasive implants have been shown to demonstrate similar biological behaviour and disease progression compared to low-grade serous carcinoma (LGSC).^9,10 According to the World Health Organisation (WHO) classification, the invasive implants are designated as LGSC and all other implants that are non-invasive are defined as “implants”.^4,11 This histologic classification of implants is an important prognostic marker since invasive peritoneal disease (LGSC) is associated with shorter overall survival.^9,10

Radiological diagnosis of SBOT is of paramount importance since those tumours have better clinical prognosis with higher patient survival rates than their malignant counterparts, even with the intraperitoneal dissemination or lymph node metastasis.^3,12 However, the imaging features of BOTs may overlap with both benign and malignant tumours, which cause challenges in the preoperative diagnosis based on imaging.¹³

The aim of this review is to illustrate the MR imaging features of borderline tumours with emphasis on SBOT. The imaging pitfalls and challenges of diagnosing SBOT will also be highlighted.

Nomenclature

Borderline tumours were first described historically as “semi-malignant” tumours of the ovary by Taylor et al¹⁴ in 1929. According to the current World Health Organisation (WHO) classification of female genital tumours (2020), the term “borderline tumour” is the accepted term, whereas the related terminology including non-invasive low-grade serous carcinoma, atypical proliferative serous tumour and serous tumour of low malignant potential are no longer recommended terminology.¹¹ The subtypes of SBOT are further defined as typical SBOT and micropapillary/cribriform variant.¹¹

Patient characteristics, signs and symptoms

Patients with SBOT tend to be younger than their malignant counterparts with an average age of 42 years at the time of diagnosis.^15,16 About 45% of these patients are under the age of 40.¹⁷ This is particularly important since fertility-preserving surgery is a viable treatment option to be considered.

The clinical signs and symptoms of SBOT are typically non-specific including vague abdominal pain or bloating, but less often, the disease may present with gynaecological (menstrual disorder, vaginal bleeding), gastrointestinal (nausea, dyspepsia, gas retention) or urinary (urinary urgency or frequency) symptoms.^8,17 In addition, it has been reported that a significant proportion (23%) may have no symptoms at all with imaging providing an incidental diagnosis.¹³

Serum CA 125 is not often a useful marker for the diagnosis of SBOT due to the low sensitivity and specificity in early diagnosis.¹⁸ Serum CA 125 level has been shown to be raised only in 50% of patients with stage-I disease, whereas it was significantly higher in more than 80–85% of patients with stage III or IV disease.¹⁸ Also, the level of CA 125 may overlap between BOT and stage I ovarian cancers and is not sufficient alone to distinguish these entities.¹⁷

Pathology of serous borderline ovarian tumours

BOTs are characterised by higher epithelial proliferation than their benign counterparts and variable nuclear atypia but without destructive stromal invasion when compared to invasive carcinomas.³ The histologic subtypes of BOTs are distinguished depending on the epithelial cell type, comprising serous and mucinous the most common and endometrioid, clear cell, Brenner and seromucinous types the least common.⁴

The WHO classification of SBOT is dependent upon the cystic serous tumour containing more than 10% borderline architecture on histopathological examination.¹⁹ Macroscopically SBOTs have one or more cysts which are lined by polypoid excrescences and closely packed papillae in an endophytic growth pattern.³ However, in almost half of SBOTs, the papillary vegetations cover the outer surface of the ovary, presenting an exophytic growth.³ Histologically, SBOTs display two distinctly different morphological phenotypes.⁷ Typical SBOTs, which behave in a benign fashion, are characterised by paucicellular, oedematous or hyalinised fibrous stroma and multiple fibrous papillae with extensive and complex hierarchical branching covered by epithelial cells which form multilayered cellular tufts.^2,5 The other morphologic variant is characterised by nonhierarchical, micropapillary pattern, so-called serous borderline tumour-micropapillary variant (non-invasive low-grade serous carcinoma), in which micropapillae arise directly from large, often fibrotic papillae.^4,11 When compared to typical SBOT, micropapillary variant is reported to be more often bilateral, have a higher frequency of exophytic growth and present mostly at an advanced stage with relatively poor outcome.^2,7

Radiological approach and MRI features

Ultrasound (US) is routinely used as the first-line modality for assessment of females with adnexal lesions. The gray-scale and colour Doppler US characteristics could help to distinguish BOTs from invasive ovarian tumours.²⁰ Furthermore, dynamic transvaginal pelvic ultrasound can be helpful in the assessment of the relationship between the ovary and the mass and to confirm the ovarian origin.¹³ Several studies have described typical sonographic characteristics of BOTs, which include cystic ovarian lesions with papillary projections in SBOTs and large cystic ovarian mass with internal septa in mucinous borderline ovarian tumours (MBOT).^20,21 As a strong sign of malignancy, papillary projections are more common in BOTs than in invasive cancer but may also be seen in benign tumours such as adenofibromas.²² In addition, “microcystic pattern” of papillary projections, solid component(s) and/or septa was reported as a novel sonographic marker of BOTs recently.²³ Doppler ultrasound may show minimal to moderate increased vascularity particularly within the vegetation pedicle, in contrast to the higher grade tumours with marked vascularity.^24,25 According to a recent meta-analysis, US including Doppler modality is an accurate preoperative method in discriminating BOT from benign/malignant ovarian masses with a sensitivity of 0.66 and specificity of 0.85.²⁶ However, the specific diagnosis of BOT is still challenging by solely use of US.

Magnetic resonance imaging (MRI) is used to characterise adnexal masses confidently as well as serving as a problem-solving method, given its capability of multiplanar imaging and morphological and functional assessment.^27,28 It is also used in staging of ovarian cancer in some institutions due to high sensitivity in detection of peritoneal implants.^29,30

In reference to European Society of Urogenital Radiology (ESUR) recommendations for MR imaging of the sonographically indeterminate adnexal masses, the basic imaging technique should include a sagittal T₂-weighted imaging (T2WI) sequence of the pelvis and a pair of T1WI and T2WI sequences covering the adnexal mass in the same orthogonal (axial or coronal or oblique) plane with identical slice thickness.³¹ Nevertheless, high-resolution T2WI sequences are of importance to demonstrate the architecture of papillary projections on MRI.¹³ DWI also helps in the differentiation of borderline and malignant epithelial ovarian tumours and in the detection of peritoneal implants.^32,33

Serous BOT is a non-invasive, low-grade, proliferative serous epithelial neoplasm, which is a different entity clinically and radiologically compared to other borderline tumours of the ovary.¹¹ Macroscopic morphology of these tumours might help to differentiate them from other malignant tumours of the ovary. Naqvi et al⁸ classified the pure epithelial SBOTs as “surface papillary” (i.e. solid mass with papillary projections arising from the outer wall of the ovary, with a branching pattern) and “cystic papillary” subtype (i.e. cystic lesion with papillary formations, either endophytic or exophytic) (Figure 1).

Figure 1.

Schematic presentations of serous (a, b, d, e) and mucinous (c, f) borderline ovarian tumours. Endophytic (a,d) and exophytic (b,e) growth patterns of papillary projections are illustrated separately. Normal ovarian parenchyma is illustrated at the edge of the lesions with follicles in it. In cystic papillary subtype of serous borderline tumours (i.e. cystic ovarian lesion with papillary projections, either endophytic or exophytic), there may be single (a) or multiple (d) papillary projections. In surface papillary subtype of serous borderline tumours (i.e. solid mass with papillary projections arising from the outer wall of the ovary, with a branching pattern), ovarian parenchyma may have a normal morphology (b), or it may be distorted often with a triangular shape (e). Mucinous borderline ovarian tumour is a large multilocular cystic mass often with irregular or thickened septations (c, f). Different viscosities of mucin in the locules cause “stained glass appearance”.

Presence of bilaterality is one of the features that is associated with SBOT. About one-third of SBOTs are bilateral.⁵ Bilateral involvement was also described as prognostic factor of disease recurrence and is also risk factor for subsequent serous carcinoma.^15,34

A predominantly cystic mass with papillary projections is the classical morphological appearance of SBOT. A spherical T2 hyperintense cyst is seen with internal low-to-intermediate signal papillary projections arising from the inner/outer wall or septation (i.e. endophytic or exophytic) or mural nodules on T2WI (Figures 2 and 3).⁸ Tanaka et al¹⁶ reported that one of six patients had papillary architecture and internal branching with cysts in their case series while the others had that growth pattern without cyst. Another study on 76 cases with SBOT showed that 49% of the patients had entirely intracystic tumours (i.e. cystic lesion with endophytic papillary formations).³⁵ In the study of Kawaguchi et al³⁶ comparing LGSC and SBOTs, 23 of 25 SBOTs were predominantly cystic with 18 including papillary growth pattern with internal branching.

Figure 2.

22-year-old with left serous borderline ovarian tumour. Sagittal (a), coronal (b) and axial (c) T₂-weighted MR images demonstrate intermediate signal intensity endophytic papillary projections (red arrows) in a cystic lesion (yellow arrows). Post-contrast axial T1W image (d) shows slight enhancement of papillary projections (red arrow) with slight irregular enhancement of the cyst wall (yellow arrow).

Figure 3.

32-year-old with serous borderline ovarian tumour. Sagittal (a) and axial (b) T2W MR images demonstrate high signal intensity exophytic papillary projections (red arrows) arising from the outer wall of a cystic right adnexal lesion (yellow arrows) with a branching architecture. Normal ovarian parenchyma (white arrow) is seen on top of the lesion. U, uterus.

The papillary projections are defined as enhancing solid components arising from the inner/outer wall or septation of the lesion with a branching architecture according to the recently published O-RADS MRI lexicon.³⁷ Small and few papillary projections associated with a thin-walled cystic ovarian structure suggest benignity, whereas papillary projections accompanying a larger solid portion suggests malignancy (Figure 4).^8,38 Similarly, Thomassin-Naggara et al³⁹ reported that papillary projections were more frequent in BOTs, whereas a larger solid portion and septa were more frequent in malignant ovarian tumours which supports that the presence of papillary projections, either endophytic or exophytic, is important in distinction of those tumours.

Figure 4.

44-year-old with high-grade serous carcinoma of the left ovary. Sagittal (a), coronal (b) and axial (c) T2W MR images show that numerous endophytic and exophytic papillary projections (yellow arrows) accompany irregular solid component (red arrows) with T2 intermediate signal intensity which was confirmed to be invasive malignancy pathologically.

Exophytic growth pattern (i.e. T2 hyperintense papillary projections arising from the surface of the ovary) are occasionally observed in SBOTs with frequencies between 15 and 30% (Figure 5).^40,41 This pattern was not observed in other borderline tumours such as mucinous or seromucinous subtypes in studies or case series.^40–42 However, in the study of Kawaguchi et al³⁶, 2 of 7 (29%) LGSCs demonstrated this pattern as well as 6 of 25 (24%) SBOTs. Therefore, exophytic papillary projections without cystic component is thought to be a characteristic MRI feature of serous ovarian neoplasms but not specific to SBOT.³⁶

Figure 5.

32-year-old with serous borderline ovarian tumour. Sagittal (a) and axial (b) T2W MR images demonstrate exophytic high signal intensity papillary projections (white arrows) without a cystic component on the surface of left ovary. The ovarian parenchyma is distorted with a triangular shape (black arrows), including small follicles and a benign cystic lesion (c) located anteriorly. U, Uterus.

On MRI, papillary projections exhibit T2 high signal intensity (SI) with T2 low SI stalk in a branching architecture (Figure 6).⁸ Indeed, being familiar with histopathological features of BOTs may facilitate the understanding of radiological findings. On MR images, the T2 hypointensity of the internal branching corresponds to the fibrous stalk on histopathology, and the T2 hyperintensity of papillary projections reflects the peripheral oedematous papillae with superficial epithelial cells, the loose structure of papillary projections, and fluid permeating multiple spaces between papillary clusters.^8,43

Figure 6.

50-year-old with serous borderline ovarian tumour. Sagittal (a) and axial (b) T2W MR images demonstrate high signal papillary projections (yellow arrows) on the surface of left ovary. Central fibrous stroma of the papillary projections is seen as T2 low signal intensity internal branching as black lines and dots (red arrows) in the lesion. Distorted ovarian parenchyma including several follicles is seen at the edge of the lesion (white arrows). U, uterus.

On T2WI, borders of the endophytic papillary projection may not be clearly seen due to similar SI with surrounding cystic fluid. Presence of T2 hypointense fibrous stroma in the centre of the papilla is a helpful tip in the diagnosis of BOT (Figure 7a and b).^13,43 Tanaka et al.¹⁶ described this appearance as like a sea anemone and deemed it as a “hallmark” feature of surface SBOTs. Those projections might also be assumed to have a frond-like appearance which can be seen as branching papillary projections especially on T2WI (Figure 8).

Figure 7.

Two different patients with serous borderline ovarian tumours (SBOT). (a, b, c) 18-year-old with SBOT. A large endophytic papillary projection (arrows) with subtle borders is seen in sagittal (a) and axial (b) T₂-weighted MR images. Note the T2 hypointense fibrous stroma in the centre. In contrast-enhanced T₁-weighted fat-suppressed image (c), borders can be seen clearly due to contrast enhancement (arrow). (d–f) 35-year-old with SBOT. Axial (d) T₂-weighted MR image demonstrates endophytic papillary projections (arrow) with intermediate signal intensity in a cystic ovarian lesion. High b-value (b 800) diffusion-weighted image (e) and ADC map (f) show restricted diffusion in the papillary projections (arrows).

Figure 8.

Two different patients with serous borderline ovarian tumours (SBOT). (a) Axial T2W MR image demonstrates “sea anemone sign” representing multiple papillary projections (red arrow) in the SBOT. Normal crescent-shaped ovarian parenchyma is seen at the edge of the cystic lesions in both ovaries (yellow arrows). (b) Sagittal T2W MR image demonstrates frond-like appearance representing a branching papillary projection (red arrow) in the SBOT. Corresponding schematic illustrations of both signs are shown at the right bottom corners.

Papillary projections may demonstrate restricted diffusion on diffusion-weighted imaging (DWI) (Figure 7d–f).⁴³ However, ADC is known to be significantly higher in papillary projections in borderline tumours compared to solid tissue in malignant ovarian tumours.³⁶ Nevertheless, it is difficult to make a standardisation amongst MRI scanners to determine a cut-off value. According to published results, the guide range of ADC value is 1.62–1.70 × 10⁻³ mm²/s in BOTs and 0.87–1.25 × 10⁻³ mm²/s in malignant ovarian tumours.³⁶

Contrast-enhanced MR imaging is useful for accurately differentiating inconspicuous small papillary projections and solid elements from a cystic component since they show avid enhancement (Figure 7c).³⁸

An important finding that can be seen in SBOT is the presence of normal ovaries or at least some normal ovarian tissue, completely or partially surrounded by the solid tumour (Figure 9).^24,44 Ovarian stroma preservation is displayed as hypointense ovarian capsular margin and multiple follicles on T2 due to exophytic growth pattern of SBOT, which grows on the surface of the ovary without invading the underlying stroma.⁴⁵ In a recent study, ovarian preservation was shown in 58% of SBOTs.⁴⁵ Due to mass effect, the ovarian stroma may get distorted or compressed and may lose its normal morphology. This feature can be of use to the surgeon in cases of previous unilateral oophorectomy, as it may allow consideration of cystectomy rather than oophorectomy in patients with the desire to retain fertility.⁴²

Figure 9.

Three different patients with serous borderline ovarian tumours. Sagittal (a) and axial (c) T2W MR images demonstrate exophytic papillary projections (yellow arrows) arising from the outer surface of the ovary. Axial T2W MR image in (b) demonstrate two cystic lesions with endophytic papillary projections (red arrows) in (b). Note that preserved ovarian parenchyma (white arrows) is clearly seen in all patients. A low signal rim is delineated around the ovarian margin, or between the lesion and normal ovarian stroma.

Another feature associated with SBOT is the presence of peritoneal implants which are seen in about 30% of cases (Figure 10).⁸ High correlation is reported in pathology specimens between exophytic tumour on the ovarian surface and synchronous implants.² In clinical practice, peritoneal implants of SBOTs create diagnostic and clinical challenges. Although there are several studies on histopathological features of those implants, imaging studies are very limited.^9,10 Nougaret et al⁴⁶ studied the imaging differences between SBOTs and LGSCs including the CT features of peritoneal disease. They found that the presence of nodular peritoneal disease pattern and calcification in the peritoneal disease were associated with invasive peritoneal lesions (i.e. LGSC).⁴⁶ Although the administration of gadolinium-based contrast media and the addition of DWI can improve the detection of peritoneal implants on MRI, the invasiveness of the implants, cannot be accurately determined by MRI.^8,33,47 Besides, a recent genetic study on implants of SBOTs revealed that non-invasive and invasive implants show distinct differentially expressed genes with genetic heterogeneity of the invasive implants, and morphology is insufficient to classify peritoneal invasive implants.⁴⁸ Therefore, future studies are warranted on especially preoperative imaging of those implants to unravel the behaviour and predict outcome.

Figure 10.

62-year-old with serous borderline ovarian tumour-micropapillary variant (non-invasive low-grade serous carcinoma) of the left ovary. Sagittal (a) and axial (b) T2W MR images demonstrate exophytic high signal intensity papillary projections (yellow arrows) with a T2 low-to-intermediate signal intensity solid portion (white arrows). There are few small deposits (red arrows) seen on both images on the peritoneal surface of pouch of Douglas which were confirmed to be non-invasive implants pathologically.

Accompanying ascites may also be a secondary finding at SBOT, but it is an unreliable indicator for predicting the presence of BOT, because it could be seen in both borderline and malignant tumours.^8,13

Differential diagnosis

Differentiation of SBOT from the other borderline subtypes and malignant counterparts may be difficult since some of the features overlap. The typical MRI findings of those tumours are summarised in Table 1.

Table 1.

Summary of typical imaging features of borderline and malignant ovarian tumours

Feature	Borderline ovarian tumours (BOT)						Malignant ovarian tumours
	Serous BOT	Mucinous BOT	Endometrioid BOT	Seromucinous BOT	Clear cell BOT	Borderline Brenner tumour	Low-grade serous carcinoma
Laterality	Bilateral (1/3)	Unilateral	Unilateral	Bilateral (1/3)	Unilateral	Unilateral	Bilateral (2/5)
Predominant morphology	Cystic	Cystic	Mixed	Mixed	Solid	Cystic	Mixed
Multilocularity	Rare	Yes	No	No	No	Yes	Yes
Papillary projections	Yes	No	Yes	Yes	No	Yes	Yes
Exophytic growth pattern	Yes	No	No	No	No	No	Rare
T2 hypointense internal branching pattern	Yes	No	No	Yes	No	No	No

Mucinous borderline ovarian tumours

MBOTs are the second most common type of BOTs which often present as unilateral large multilocular cystic mass (mean tumour size about 20 cm) containing mucinous fluid (Figure 1c and f).^4,11 Zhao et al reported the honeycomb loculi, densely aggregated numerous loculi of 5–10 mm, which form as a result of multiple grouped septations, as a characteristic feature of MBOTs.⁴⁹ Regarding the SI of the cystic contents, the varying SI on both T1WI and T2WI corresponds to the differing viscosities of mucin, which is called “stained glass appearance” (Figure 11).²⁸ Thick mucinous material may increase the T1 SI and decrease the T2 SI of the cyst contents.⁵⁰

Figure 11.

28-year-old with mucinous borderline ovarian tumour. Large cystic mass with multiple septa and loculations is demonstrated on coronal (a), axial (b) T2W MR images and axial T1W fat-suppressed image (c). A slightly thickened septum (arrow) is seen on the axial images (b, c). Note that there are no papillary projections and the cystic locules (stars in c) have different T1 signal intensities consistent with the differing viscosities of the mucinous content.

Differentiation between serous and mucinous BOTs can also be a diagnostic dilemma occasionally, particularly when mucinous tumours include vegetations and serous tumours include irregular septations.⁴⁰ Bazot et al⁴⁰ found that the most significant features predictive of MBOT on MRI were grouped septa, multilocularity, >10 septa, loculi of different SI and irregular thickened septa, whereas bilaterality, mixed or predominantly solid morphology and the presence of vegetations especially when exophytic or displayed high T2 SI were most significant predictive features of SBOT. Amongst those variables, vegetations and multilocularity were independent MRI features to differentiate SBOT from MBOT.⁴⁰

Other less frequent borderline ovarian tumours

This group includes endometrioid, seromucinous, clear-cell and borderline Brenner tumours. Very little is known about MR imaging features of those subtypes since they are very rare. Endometrioid BOT has been reported to present as solid-cystic lesions often including haemorrhagic cysts, with papillary projections displaying low SI on T2WI.⁵¹

For seromucinous BOTs, several MRI features such as papillary solid nodule, T2 high SI solid portion and T2 low SI core was reported with high specificity (over 93%).⁵² It may be difficult to distinguish SBOT from seromucinous BOT owing to some overlapping imaging features. According to the study of Kurata et al, exophytic growth pattern was found only in SBOT when compared with seromucinous subtype.⁴¹ Papillary solid nodule was present more frequently in SBOT. Regarding other MRI features such as nodule in cyst appearance, multilocularity, T2 high SI solid portion and T2 low SI core, there was no significant difference between two groups.⁴¹

Clear cell BOTs are very rare and demonstrate as solid or predominantly solid masses owing to adenofibromatous pattern, but without papillary architecture.^11,53

Borderline Brenner tumours are large cystic masses with papillary projections protruding into the cyst lumen, however also cases with multiloculated cystic mass including thickened septa have been reported.^11,54,55

Amongst those rare subtypes, especially endometrioid, seromucinous and clear cell BOTs have been proved to have strong associations with endometriosis.^11,51,56

Malignant invasive ovarian tumours

As the prognosis of SBOT is far superior to malignant invasive tumours, distinction of those tumours from LGSC and high-grade serous carcinoma is critical.⁸ Several studies compared imaging findings of SBOTs with malignant counterparts.³⁶ Nougaret et al⁴⁶ proposed that bilateral ovarian masses, the presence of peritoneal disease and greater solid tumour volume were significant predictors of LGSC when compared to SBOTs. According to the study of Kawaguchi et al³⁶, the predominantly solid lesions were far more frequent in LGSC than in SBOT (43% vs 8%). Moreover, all predominantly cystic LGSCs were multilocular including mural nodules, while predominantly cystic SBOTs were either unilocular (30%) or multilocular (70%).³⁶ Papillary growth pattern with internal branching was significantly more frequent in SBOTs than in LGSC (72% vs 0%).³⁶ Exophytic growth pattern was also observed in LGSC as well as SBOT however, without internal branching.³⁶ In addition, the SI of solid components on T2WI and ADC values were significantly lower in LGSCs than in SBOTs.³⁶

For differentiation of BOTs from malignant ovarian tumours, Lu et al⁵⁷ proposed that ADC value of the tumour and normalised ADCs regardless of b values were significantly higher in BOTs than malignant tumours however this study included serous, mucinous and endometrioid BOTs and all subtypes of malignant epithelial ovarian tumours.

Clear cell carcinoma (CCC) of the ovary may present as a unilocular large cyst with papillary projections or mural solid component⁵⁸ and thus can be a mimicker of SBOT which present as a unilocular cyst with endophytic papillary projections (Figure 12).^59,60 The ADC values of solid portions tend to be lower in CCC compared with SBOT.^36,61

Figure 12.

Two different patients with single mural nodule in a large cystic pelvis mass. Sagittal (a) and axial (c) T₂-weighted MR images of the first patient with serous borderline ovarian tumour demonstrates a small papillary projection with T2 hypointense signal intensity internal branching (arrows). Sagittal (b) and axial (b) T₂-weighted images of the second patient shows T2 intermediate signal intensity mural solid tissue without internal branching pattern (arrows). Pathological evaluation confirmed clear cell carcinoma of the ovary.

MRI advancements and new technologies

Different advanced MRI methods have been investigated in the evaluation of BOTs in recent studies. In a study constructed by Ma et al⁶², the utility of MR spectroscopy was examined to differentiate borderline from malignant epithelial tumours and N-acetylaspartate/Choline (NAA/Cho) ratio was found to be a reliable biomarker. For BOTs, a significant elevation of NAA peak has been defined as a characteristic feature.⁶²

In a recent study, histogram analyses derived from DWI, diffusion kurtosis imaging, intravoxel incoherent motion (IVIM) and dynamic contrast enhanced (DCE) MRI were assessed for differentiation of borderline epithelial ovarian tumours (BEOT) from malignant epithelial ovarian tumours (MEOT).⁶³ According to results of that study, MEOTs tended to have right-skewed distributions (higher skewness values of ADC, diffusion coefficient, pure diffusion coefficient), while BEOTs tended to have left-skewed distribution (lower skewness values of ADC, diffusion coefficient, pure diffusion coefficient).⁶³ These trends corresponded with the characteristics of the two types of tumours and seemed to be promising for getting information noninvasively for differentiation of both entities.

DCE-MRI is an advanced non-invasive technique that can reflect the vascularity and microcirculation of tumours, thereby playing an important role in tumour detection and characterisation.⁶⁴ Li et al investigated quantitative parameters of DCE for differentiating benign, borderline, and malignant ovarian tumours.⁶⁵ They found that volume transfer constant (K^trans) and rate constant (k_ep) were higher in malignant tumours than in BOTs, and higher in BOTs than in benign tumours. In addition, the K^trans was a better indicator than the others to differentiate borderline from malignant tumours (AUC: 0.743).⁶⁵ Similarly, a recent study compared DCE and IVIM parameters of SBOT and early serous ovarian cancer (eSOCA) in rat models and showed that SBOTs presented significantly lower values of K^trans and extracellular extravascular volume fraction (v_e) and a significantly higher value of true diffusion (D) compared with eSOCAs.⁶⁶ When DCE-MRI was combined with IVIM-DWI, diagnostic performance was better than that of DCE alone.⁶⁶

Radiomics is an emerging research area which extracts a wealth of data using tools based on standard of care or advanced radiological imaging. It may capture the spatiotemporal tumour complexity in a comprehensive and non-invasive manner.⁶⁷ Song et al⁶⁸ used radiomics features extracted from DCE-MRI pharmacokinetic protocol and reported promising results for discriminating among benign, borderline and malignant ovarian tumours. Zhang et al⁶⁹ showed that radiomics features extracted from MRI were highly correlated with ovarian cancer classification and prognosis of the patient. In that study, computed model performed better than the radiologist in distinguishing malignant adnexal masses from benign tumours.⁶⁹ The most common diagnostic error among the radiologists was classifying BOTs into the benign group in that study,⁶⁹ which shows the clinical need in machine-learning algorithms in this area. Therefore, both radiomics and computer-aided models in differentiation of SBOTs appears promising and merits further evaluations and developments for future clinical applications.

Surgical management

The management of SBOT differs from that of benign and malignant counterparts. Limited surgery with cystectomy or unilateral salpingo-oophorectomy may be sufficient for patients with SBOT.^15,70 However, the patient’s age and fertility desire play a decisive role in the surgical decision.¹² In cases with bilateral cystectomy, it should be noted that there is a high risk of recurrence which is up to 31%.⁷¹ There may also be a role for surgical staging and debulking for SBOT of the micropapillary type, which has been associated with invasive peritoneal implants and an increased risk of recurrence.⁴³ Routine lymphadenectomy is not recommended in SBOT but is offered if frozen-section examination is consistent with micropapillary BOT.^72,73

Prognosis

BOTs are known to have a better prognosis than malignant tumours even with intraperitoneal spread or lymph node metastasis.²⁴ Their 5- and 10-year survival rates have been reported as 95 and 85%, respectively.²⁴ Approximately 11% of the patients with BOT relapse in the follow-up.⁷¹ The tumour related prognostic factors for disease recurrence were found as micropapillary structure, microinvasion, peritoneal implantation and FIGO stage.⁷⁰ According to the literature, invasive implants appear to have a higher recurrence rate and worse prognosis than non-invasive implants.⁷⁰ However, lymph node involvement did not seem to affect survival or recurrence.⁷⁴ Moreover, as the BRAF-mutant status has been shown to be associated with an improved outcome, this result could potentially be a guide for a better prognosis through new therapeutic targets in the future.⁴⁶

Conclusion

Borderline ovarian tumours form an intermediate group of neoplasms between benign and malignant ovarian tumours. They have a different prognosis and management compared to their malignant counterparts. Accurate characterisation of these tumours is important in surgical planning since fertility-sparing surgery can be offered to young patients with early-stage disease. Serous and mucinous subtypes are the most frequent amongst all borderline tumours. Both of them have distinct morphological MRI features which can be readily appreciated on conventional MR sequences such as T2WI. SBOTs are distinct entities clinically and radiologically with abundant papillary projections, T2 low signal internal branching, and preservation of the ovarian stroma. MRI can reveal the complex architecture of SBOT and helps in preoperative differentiation of SBOT from benign or malignant tumours. Being aware of the typical MRI features allows for accurate characterisation and appropriate management.