Imaging of Pediatric Ovarian Tumors: A COG Diagnostic Imaging Committee/SPR Oncology Committee White Paper

Abstract

Ovarian tumors in children are uncommon. Like those arising in the adult population, they may be broadly divided into germ cell, sex cord and surface epithelium subtypes however germ cell tumors comprise the majority of lesions in children, whereas tumors of surface epithelial origin predominate in adults. Diagnostic workup, including the use of imaging, requires an approach which often differs from that required in an adult. This paper offers consensus recommendations for imaging of pediatric patients with a known or suspected primary ovarian malignancy at diagnosis and during follow-up.

INTRODUCTION

Ovarian tumors are uncommon in childhood with an estimated annual incidence of 2.6 per 100,000 girls^1,2. The tumors are typically benign with a malignancy rate ranging between 9.5 and 30%^1,3–5. However, this range is likely skewed high due to surgical recruitment bias^3,6–8. Interestingly, the reported malignancy rate has been decreasing^9–11. This suggests that the increased utilization of and advancements in imaging have led to the detection of a greater number of benign tumors. Although this can sometimes lead to unnecessary workup or surgery, large benign tumors can result in torsion.

Most ovarian tumors in children arise from either germ cells, sex cord (ovarian stroma) or surface epithelium. The different types of ovarian tumors and differentiating clinical and imaging features are included as Table 1. Mature teratomas are the most common ovarian neoplasm in children while dysgerminomas are the most common malignant tumor.

TABLE 1.

Differential Diagnosis of Ovarian Masses in Children

Tumor type	Cell origin	Imaging features	Tumor Markers	Clinical features
Teratoma	GCT	Fat, calcium fluid	CA 19-9 (AFP in immature)	Rare association with NMDA receptor encephalitis
Yolk sac tumor	GCT	Mixed cystic and solid. May have dilated vessels seen as flow voids on MRI	AFP
Dysgerminoma	GCT	Solid; septa enhances but with low signal on T2 weighted sequence	LDH	Associated with gonadal dysgenesis; predilection for nodal spread
Choriocarcinoma	GCT	Solid; hemorrhagic	β-hCG	Nongestational tumors occur but are very rare in children
Polyembryoma	GCT			Rare
Sertoli-Leydig	Sex cord stromal	Variable; May be solid mass; may have low signal intensity on MRI	AFP	Virilization; Associated with DICER-1 mutations
Juvenile granulosa cell	Sex cord stromal	Sponge-like appearance	Inhibin
Fibroma		Low MRI signal; delayed or variable enhancement		Associated with pleural effusions and ascites (Meigs syndrome)
Cystadenoma/ Cystadenocarcinoma	Surface epithelial	Single or multiple cysts. Malignancy risk if mural nodules or other cystic complexities.	CA-125	Can be of mucinous or serous origin. Carcinoma is rare in children.
Borderline tumor	Surface epithelial	Mildly complex cystic lesion; may have papillary projections
Gonadoblastoma	Hybrid of GCT and Sex cord components	solid mass; calcifications may be present however paucity of data		Precocious puberty; association with gonadal dysgenesis where Y chromosome present
Small cell carcinoma	Origin unclear	Rare; too little data		Hypercalcemia

AFP [Alpha-fetoprotein]

β-hCG [human chorionic gonadotropin, beta subunit]

LDH [lactate dehydrogenase]

Girls with an ovarian tumor may have chronic symptoms such as recurrent abdominal pain, anorexia, nausea and vomiting. At least half present with acute abdominal pain caused by ovarian torsion or tumor rupture^12,13. Girls with sex cord stromal tumors tend to present with smaller tumors due to hormonal production. Examples include isosexual precocious puberty in juvenile granulosa cell tumor and virilization with hirsutism in Sertoli-Leydig cell tumors. Rare presentations with relatively high specificity include Anti-N-methyl D-aspartate [NMDA] receptor encephalitis in the setting of an ovarian teratoma and elevated serum calcium, polydipsia, and polyuria in patients with small cell carcinoma of the ovary.

STAGING

Surgical staging systems are used when assessing ovarian tumors. The International Federation of Gynecology and Obstetrics (FIGO) staging system and the American Joint Committee on Cancer (AJCC) tumor-node-metastasis (TNM) staging systems have been used to stage ovarian tumors in adults. While the FIGO system has historically been used for children, both it and the AJCC TNM staging system were designed for the most common ovarian malignancies in adult women, those of epithelial origin^14,15. However, most childhood ovarian tumors are germ cell tumors (GCT) which do not warrant as an aggressive surgical approach. To address this discrepancy, in 2004 the Children’s Oncology Group (COG) developed a surgical staging procedure for ovarian GCT^16,17. Evidence supports the safety of the COG staging system’s less aggressive surgical approach and its prognostic capability¹⁸. While its adoption has been slow, the COG staging remains the more relevant one for the pediatric population.

Building on this work and others, the Malignant Germ Cell International Consortium (MAGIC) was formed in 2009. MAGIC analyzed patient data from both the COG and the Children’s Cancer and Leukemia Group (United Kingdom) which resulted in the introduction of a further refined and validated risk stratification for children^19–21. In contrast to the child or the adult, there is no universally accepted classification scheme for adolescents with a GCT. This remains a management challenge for this population^22–25. One of several important goals of MAGIC is the standardization of staging and treatment specifically for the adolescent patient with GCT by bringing together multiple cooperative groups and including both adult and pediatric experts²⁶. This ongoing work will likely yield data that is expected to underlie future guidance.

In contradistinction to GCT, both ovarian sex cord stromal tumors and tumors of epithelial origin are still staged based on the FIGO staging system²⁷. Thus, it is important to attempt to distinguish these tumor types prior to surgical resection. Imaging is used to perform this preoperative assessment and can help in triage and surgical planning, including potentially, ovarian sparing approaches^6,28.

Imaging at Diagnosis:

In a child with a suspected ovarian mass, transabdominal ultrasound (US) is the modality of choice for initial imaging investigation [GRADE A, SOR 1.00] ^3,29,30. US is useful to identify an ovarian lesion, determine its composition and its neoplastic potential and differentiating it from functional findings such as a dominant follicle, corpus luteum, or hemorrhagic cyst. Any suspicious lesion should be measured as size has been shown to be an independent predictor for neoplasm³¹. Large ovarian lesions may be difficult to visualize in their entirety and some lesions may remain undefined on US³². Additional criteria to distinguish benign from malignant ovarian lesions exist for adults. These criteria have not been validated in the pediatric popluation^33,34

It is highly preferred that the transabdominal US be obtained with a full bladder, especially when ovarian torsion is being considered in the differential diagnosis. In older, sexually active girls, transvaginal sonography may be pursued. Transvaginal US allows for more detailed imaging of the uterus and ovaries as compared to transabdominal imaging. This is due to the fact that the US probe can be placed closer to the organ of interest. While transvaginal US is routinely recommended for adults, it is not performed in younger children due to the size of the probe and its invasive nature. There have been no studies comparing the performance of the two approaches in children. However, most pediatric radiologists do not believe that imaging is compromised using transabdominal approach due to the smaller size of the patient, the relative decreased amount of abdominal fat, and the larger size of most ovarian tumors.

When interpreting an exam, a radiologist may find the recently developed Ovarian-Adnexal Reporting and Data System (O-RADS) system to be of value as a guide. The O-RADS system was developed by an international multidisciplinary working group and included representatives from the Society for Pediatric Radiology. Its stated goal is to standardize terminology so that radiologists can provide more consistent interpretations and increase their accuracy in assigning a risk of malignancy to ovarian massess³⁵. Although the O-RADS scoring system has not been validated in the pediatric population, its lexicon is relevant and pediatric radiologists should be familiar with it in order to optimally communicate risk in appropriate cases such as presence of complex cysts [GRADE B, SOR 2.00] ^36–38.

MRI of the abdomen and pelvis is recommended for further evaluation of an ovarian neoplasm or suspected neoplasm [GRADE A, SOR 1.5]. ^39–42 .

Additionally, any cystic lesion larger than 7.5 cm in longest axis, not visualized in its entirety (regardless of size), or with any solid components should be imaged by MRI [GRADE C, SOR 1.90] ^7,43,44 .

MRI is used to characterize the tumor, differentiate benign from malignant lesions, identify sites of spread, and plan surgery. In a series of 27 MRI exams in 18 patients, Emil et.al. found that MRI may characterize a mass as neoplastic with a sensitivity and specificity of 89% and 94%, respectively and directly influenced the surgical approach in 74% of exams⁴¹. When a malignant or a potentially malignant adnexal mass is encountered, MRI is best for assessment of adnexal invasion and evaluation of the pelvic sidewall³. Additionally, most ovarian masses have characteristic features which are, in some cases, pathognomonic and in others, suggestive⁴⁵. Recently, Jenssen, et al, reported that in a cohort of 30 girls with ovarian lesions who underwent MRI, all lesions were accurately characterized as benign or malignant⁴⁶. One notable limitation of MRI is that it cannot be relied on for differentiation of mature vs immature teratomas⁴⁷. Indeed, a teratoma may potentially harbor more aggressive elements such as yolk sac tumor without any imaging manifestations and reliance on tumor markers and complete surgical staging is needed in many cases⁴⁸. Finally, Marro, et al. have shown that findings on MRI significantly change the surgical plan in almost one-third of children with ovarian masses when compared with interpretation of US alone⁴⁹.

The recommended MRI protocol is included as Table 2.

TABLE 2.

Pelvic MRI at diagnosis. [Abdomen can be scanned with either CT or MRI.]

Plane	Sequence	Contrast phase	Coverage	Required/Optional	Comment
Axial	T2 FSE	Pre	Pelvis	Required	• Do not fat suppress as the fat enables good contrast
Coronal	T2 FSE or single shot echo	Pre	Abdomen and pelvis	Recommended	• Recommended. • Ideal for assessing kidneys for obstruction or for identifying intraabdominal extension or metastatic disease
Sagittal	T2 FSE	Pre	Pelvis	Optional	• Sagittal plane may facilitate assessment of endometrium if thickened from hormonally secreting tumor
Axial	DWI	Pre	Pelvis	Recommended	• Paucity of data. Cannot be used in isolation to suggest malignancy
Axial	In-phase & opposed phase imaging or Dixon method	Pre	Pelvis	Required	• Can assist in detecting very small amounts of fat or tiny teratomas that may be present in patients with NMDA encephalitis
Axial	T2 FSE with spectral fat saturation	Pre	Pelvis	Required	• Key sequence to detect fat. i.e. teratomas. • As an alternative, precontrast SE T1 with and without fat saturation is acceptable
Axial	T1 with spectral fat saturation Either spoiled gradient echo (GRE) or spin echo-based sequences	Pre and Post	Pelvis	Required	• Used to assess enhancement • GRE technique is faster and can give information on contrast kinetics though little use in children. • Spin echo technique offers improved image quality

Intravenous contrast media is recommended for all patients being imaged with MRI to evaluate an ovarian lesion [GRADE B, SOR 1.30]. Administration of intravenous contrast has shown some value in small studies to assist in differentiating benign from malignant ovarian tumors in children^{46,50 46,51} . Our recommendation is based on the belief that that contrast may help to identify mural nodules within a cystic tumor, distant or regional lymph nodes, or peritoneal nodules.

Although MRI is preferred, CT of the abdomen and pelvis is an acceptable alternative modality in children who cannot undergo MRI [GRADE D, SOR 1.40]. In adults, CT has been shown to be 81% sensitive and 87% specific for the diagnosis of ovarian carcinoma⁵¹. Accuracy is improved when it is used in combination with US⁵². There are no similar trials in children. In the setting of a teratoma, either CT or MRI scan may be diagnostic.

If CT is performed, IV contrast is required to assess an ovarian lesion [GRADE A, SOR 1.20]. The use of oral contrast is more controversial. The authors recognize that recent data has questioned the ubiquitous use of oral contrast for abdominal imaging^53–55. However, adnexal masses are poorly represented in these studies. Moreover, it has been suggested that a cystic adnexal lesion may be confused for an unpacified bowel loop^3,56. Thus, oral contrast is recommended for CT evaluation of pelvic masses in girls [GRADE D, SOR 2.40].

If an ovarian malignancy is present, a CT scan of the chest is recommended to identify pulmonary metastatic disease [GRADE D; SOR 1.50]. Metastatic work-up may vary based on the suspected site of systemic metastasis, the type of malignant components in the teratoma or the type of ovarian malignancy⁵⁷. Imaging of the lungs may be important as malignant GCT’s can result in maligant pleural effusions or pleural nodular implants⁵⁸. However, due to the rarity of the tumor, incidence is unclear. Yolk sac tumors can also spread to the peritoneum, liver or lymphatics. Other tumors at high risk for metastasis include choriocarcinomas. Dysgerminomas have a predelection for nodal metastasis.

In patients presenting with NMDA receptor encephalitis, dedicated pelvic imaging is required as there is a strong association with ovarian teratomas [GRADE A, SOR 1] ^59,60. We suggest starting with pelvic US and if negative, considering pelvic MRI for increased sensitivity [GRADE C, SOR 2.6]. In patients with NMDA receptor encephalitis and without imaging evidence of ovarian tumor, there are reports of subsequent development of tumor raising speculation that at least some cases harbored an occult ovarian tumor at initial presentation^59,61,62. Presence of GCT in the mediastinum in the setting of NMDA receptor encephalitis has been reported but is exceedingly rare⁶³

Criteria for tumor response assessment:

There are no published guidelines formalizing tumor response on imaging that are specific to ovarian tumors in children. RECIST guidelines have shown mixed performance in evaluation of ovarian tumors in adults and have been supplanted by assessment of tumor makers. However, RECIST criteria is still used for patients on study protocols^{64,65 66}.

Post-Treatment surveillance imaging:

The goal of imaging surveillance in girls with treated ovarian tumors is to detect either local recurrence or metastatic disease before symptoms present. There are no formal guidelines for the post-treatment imaging surveillance strategy in children treated for ovarian tumors. Proposed guidelines are based on tumor type. Data from an adult cohort of patients with sex cord stromal tumors suggests recurrence is unlikely after three years and that post treatment metastatic lung disease is unlikely in the absence of localized recurrence⁶⁷. For dysgerminomas, the Society of Gynecologic Oncologists recommends imaging every 2-4 months for the first two years if patient’s initial tumor marker levels were not elevated. Later recurrences occur but are rare^68,69. Adult granulosa cell tumor can recur ten years post-treatment and require long-term follow-up though juvenile granulosa cell tumors may be less aggressive^70–72. Serum tumor markers such as inhibin may also be followed for detection of recurrence⁷³.

Imaging follow-up of the patient with a resected ovarian teratoma is controversial. Veneris, et al. observes that most adult patients with GCT receive 5-10 CT scans during the first five years after diagnosis though the authors note this may not be appropriate in children^25,30. Some authors have suggested that routine imaging surveillance in the asymptomatic patient is not useful⁷⁴. Others have advocated for ultrasound-based surveillance. One large multicenter study identified metachronous tumors in 3.2% of all patients with an ovarian tumor and 4% of all patients diagnosed with a mature teratoma⁷⁵. Additionally, tumor recurrence occurred in 4.8% of girls with an ovarian tumor and 1.1% of patients with a mature teratoma⁷⁵. Most metachronous disease and recurrences were detected by routine follow-up US with a median time between initial diagnosis and metachronous or recurrent disease of 12 months and an upper range extending to 80 months ⁷⁶. Other retrospective studies have suggested an overall recurrence risk as high as 23%⁷⁷. Based on this data, we recommend that imaging surveillance be performed with either pelvic US or MRI in all girls with a resected ovarian tumor [GRADE C, SOR 1.40] ^78–82.

The Future of Imaging:

Future approaches to imaging of ovarian tumors in children may make use of radiomics and incorporate machine-learning algorithms to decipher benign from malignant masses and possibly predict histology. This is based on progress in deploying these techniques in adult patients with ovarian and other cancers though, to date, no such studies in children with ovarian tumors have been published^83,84. Although there are some case reports describing the utility of contrast enhanced US in cases of pediatric ovarian torsion, its role in evaluation of pediatric ovarian masses still needs to be explored.^85,86

Advanced imaging techniques such as diffusion weighted imaging (DWI) is widely used in oncologic MRI. Early attempts at its use in ovarian tumors in adults and children were met with disappointment as the normal ovary intrinsically restricts diffusion and any intralesional fat may also show restricted diffusion^46,87. However, more recently, investigators have found use in the detection of malignant ovarian tumors and detection of tumor implants in children³².

Abbreviation

NMDA

Anti-N-methyl D-aspartate

FIGO

International Federation of Gynecology and Obstetrics

AJCC

American Joint Committee on Cancer

TNM

Tumor-Node-Metastasis

GCT

Germ Cell Tumor

COG

Children’s Oncology Group

MAGIC

Malignant Germ Cell International Consortium

SOR

Strength Of Recommendation

O-RADS

Ovarian-Adnexal Reporting and Data System

DWI

Diffusion Weighted Imaging