Background : About 25% of adnexal masses remain indeterminate after transvaginal ultrasonography, as evidenced by the published literature. According to a multicenter study conducted in France by Huchon et al., surgery is mainly indicated based on tumor size and ultrasonographic appearances and a high rate of benignity is finally found at pathology (83.3%). MRI is the most accurate second-line imaging technique for characterizing adnexal masses with an accuracy>95%. However, among physicians, a debate exists on the relevance of MR imaging to characterize adnexal masses possibly due to the absence of standardization of the MR report. In 2013, the first MR scoring system for adnexal masses named "ADNEXMR score" was developed and found to be accurate, with an AUROC>0.94 to distinguish benign from malignant adnexal masses. Indeed, our hypothesis is that ADNEXMR score that relays the radiologist's suspicion of malignancy to clinician has the potential aim of improving patient management, in particular by limiting the number of patients undergoing inappropriate surgery.

Main objective: To compare the rate of inappropriate surgery that occurs when patients are managed according to the therapeutic strategy based on ADNEXMR score with the therapeutic strategy performed in clinical routine.

Primary endpoint: The rate of inappropriate surgical intervention during the two first months after MR imaging (i.e. unnecessary diagnostic surgery for benign lesions and incomplete staging for borderline or invasive cancer).

Study design:

• The study is a prospective multicenter randomized diagnostic study. 606 patients will be included by 9 hospitals (17 centers).
• Total duration of the study is 56 months with 32 months for patient inclusions.
• At D0, following informed written consent, patients will be randomized in one of two groups (intervention group or control group). In the intervention group, patients will undergo a pelvic MR imaging (1.5T or 3T) as routinely performed, including morphological sequences (T2, T1 with and without fat suppression and T1 after dynamic gadolinium injection) and functional sequences (perfusion and diffusion-weighted sequences). Prospectively, a senior radiologist analyzes the different MR criteria to characterize adnexal masses. The reader will classify the mass using ADNEXMR SCORING system and the patient will be managed according to the score. In control group, the complex adnexal mass will be managed according to the standard strategy and treatment plan routinely used by the multidisciplinary team. Clinical, biological, ultrasonographic and/or MR data and the type of treatment will be recorded. If a surgery is required, surgical procedure will be performed within the two months after initial diagnostic imaging (as routinely recommended).

At M6/M12, follow-up will be performed. If the lesion has increased or changed with suspicious criteria, surgery will be performed as routinely recommended in both groups.

At M24, adnexal masses status will be recorded by clinical and imaging follow up.

Context and rational Ovarian cancer is the fifth leading cause of death among neoplastic diseases in women in developed countries (1). The 5-year survival rate is 90% for patients at stage I disease at diagnosis but only 2% are diagnosed at this stage because of the non-specific nature or total absence of symptoms in the initial phase of the disease (12). Survival rate strongly depends on complete surgical resection during the first surgery with a 5-year survival rate about 20% for incompletely resected stage III tumors (2). Thus, preoperative diagnosis strategy must be improved. Four multicentric randomized trials (3-6) have underlined the lack of specificity of the classic diagnosis strategy consisting in the combination of measurement of CA125 and transvaginal ultrasonography (TVS) to detect and characterize adnexal masses. In PCLO trial, 30 patients underwent a surgery for every cancer diagnosed (3). According to a multicenter study conducted in France (9), surgery is mainly indicated on tumor size and ultrasonographic data and a high rate of benignity is finally found at pathology (83.3%) as previously demonstrated (13). Recent trials have shown that inappropriate interventions in women with benign adnexal lesions lead to significant morbidity and mortality. In PCLO trial, among the 1080 women with false positive results (i.e benign lesions) who underwent surgery, 163 (15%) had at least one serious complication (3). All women have operative and anaesthesic risks, which are increased in menopausal women who suffer from diabetes of high pressure blood (8). Moreover, in pre-menopausal women, fertility is usually a main issue and the negative impact of surgery is well known. Two surgical procedures exist for the management of adnexal masses. The most conservative one, the laparoscopy which has become the gold standard in the treatment of benign adnexal tumors and the laparotomy which remains the standard in the treatment of malignant tumors. However, even laparoscopy may induce severe adhesions and unnecessary damage to the ovarian tissue (14). According to Huchon and al., the high rate of inappropriate procedures was not observed only for benign tumors but also for borderline and invasive malignancy with incomplete staging in 16% and 50% respectively, which is the main prognostic factor of ovarian cancer (9). Inadequate surgical procedures may worsen the prognosis because of no fully inspection of peritoneal cavity, tumor morcellation, or cystic rupture especially for surgeons without lower expertise in this field. Bristow et al. demonstrated that both low-volume hospitals (HR 1.08, 95% CI 1.01-1.16) and low-volume physicians (HR 1.18, 95% CI 1.09-1.28) were associated with decreased disease-specific survival (7). Thus, preoperative diagnosis is crucial to refer patients to gynecologist oncologist in a specialized tertial cancer center if there is a suspicion of ovarian cancer.

An alternative evidence-based approach to the presurgical diagnosis of adnexal tumors is to use Logistic regression model (LR2) or Simple Rules developed by the International Ovarian Tumor Analysis (IOTA) group as demonstrated by a recent updated meta-analysis. Irrespective of which method used there is a group of tumors which remains difficult to classify after TVS. International Ovarian Tumor Analysis (IOTA) group and others teams estimate at 25% the percentage of adnexal masses that remains indeterminate and where it is evident that accurate second stage tests would be of value in this group of masses (15,16). In this setting, MR imaging was proven to be the most accurate second-line imaging technique (17) and the most cost effective intervention for ultrasonographically indeterminate masses (16): Firstly, the risk of malignancy for these masses is small (lower than 20% of malignancy), particularly in premenopausal women. Secondly, by removing the need for US follow up, MR imaging offers the possibility of earlier diagnosis of cancer in that minority of women with malignant masses (17,18). In the literature, the accuracy of MRI for distinguishing malignant from benign indeterminate adnexal masses ranges from 83% to 93% (18-20), compared to 63%-80% with ultrasonography (21,22). Recently, functional imaging techniques adding new criteria (23) to conventional MRI improve the interpretation, reproducibility and characterization of indeterminate adnexal masses, resulting in an accuracy higher than 95.3% (19,24,25) However, among physicians, a debate exists on the relevance of MR imaging to characterize adnexal masses possibly due to the absence of standardization of the MR report. In this way, only 25% of surgeons performed MR imaging before surgery explaining the high incidence of pre-operative misdiagnoses (9).

In 2013, the first MR scoring system for adnexal masses named "ADNEXMR SCORING system" was developed (10). The features included in this scoring system are well established, as each has previously been used to distinguish benign from malignant masses (26-31). ADNEXMR SCORING system had good discriminatory characteristics with an AUROC of 0.981/0.964 in both the training and validating set (10). A preliminary external validation was performed on the validating set with 4 external readers that suggest a high diagnostic performance (AUROC 0.980, 0.954, 0.955, 0.973 for both two senior and two juniors readers, respectively). For these four subsequent radiologists, the observed malignancy rates were less than 2 %, less than 5%, between 5% and 90% and over 90 % for ADNEXMR-SCORE 2, 3, 4 and 5, respectively. Therefore, this classification should be easily adopted by radiologists in pelvic imaging to improve report standardization. Finally, ADNEXMR SCORING system should influence pelvic mass management. With a score 4 or 5, ADNEXMR SCORING system predicts malignancy with 93.5% of sensitivity (58/62) and 96.6% of specificity (258/267). Thus, the risk of malignancy is high and the patient should be referred to a cancer center to undergo an appropriate surgical procedure at the first attempt. With a score ≤ 3, the risk of malignancy is minimal and the patient may benefit from more imaging follow-up or conservative treatment depending of clinical symptoms or infertility history. Inspired by the BI-RADS classification for breast tumors, the ADNEXMR scoring system is a simple and accurate five-category classification with accuracy higher than 95%. ADNEXMR Scoring system includes functional data that allows an excellent tissue characterization allowing not miss any cancer in case of score ≤ 3 in tumors lower than 6cm.

Indeed, our hypothesis is that ADNEXMR score that relays the radiologist's suspicion to clinician has the potential aim of improving patient management especially limiting the number of inappropriate surgery. Up to date, no research tests the clinical impact of any MR scoring system for the management of adnexal masses. Thus, this research would be the first one with the finality to homogenize clinical practices limiting inappropriate procedures that impact on (1) morbidity and fertility (32,33), especially when patient unnecessary undergoing surgery for benign lesions and (2) mortality when patient do not undergo a complete surgery for ovarian cancer at the first attempt (34,35).

If the study demonstrates our hypothesis, the results of the study will have an important impact on clinical practice, by increasing patients' safety, associated to possible major impact on fertility due to the avoidance of unnecessary surgery, together with significant savings for the health services. Moreover, this score could help clinician in planning the surgery accordingly and referring patients to a specialized tertial cancer center in case of ovarian cancer for a better prognosis.

Summary of the known and foreseeable benefits and risks for the research participants The control group undergo standard of care treatment and therefore there is no associated risk or benefit in this group. The intervention group will all undergo MR, which may or may not have been standard of care in each case. There are no significant risks associated with MR scanning apart from the minor discomfort or inconvenience of undergoing the scan. The categorization into the AdnexMR score group is highly unlikely to be a risk to the patient as this has been validated in a large retrospective cohort of patients (10). However, there is a possible risk if a patient with a cancer is incorrectly categorized as score 3 and may not undergo oncologic cytoreductive surgery as their initial treatment. This theoretical risk exists but is very low because none invasive cancer was categorized as score 3 in the large cohort (n= 497) previously published (10).

Primary objective To compare the rate of inappropriate surgery that occurs when patients are managed according to the therapeutic strategy based on ADNEXMR SCORING system with the therapeutic strategy performed in clinical routine according to European Society of Medical Oncology guidelines (11) (i.e. ultrasonography, CA125 +/-MR imaging).

Reference standard: surgical with histology / clinical and imaging follow-up at 24 months.

Secondary objectives To compare the routine therapeutic strategy and the therapeutic strategy based on ADNEXMR SCORING system.

a) On therapeutic morbidity and mortality On delayed ovarian cancer diagnosis.

Description of research methodology Experimental plan A prospective multicenter randomized diagnostic study. At D0, following informed written consent, patients will be randomized in one of two groups (intervention group or control group).

In the intervention group, patients will undergo a pelvic MR imaging (1.5T or 3T) as routinely performed, including morphological sequences (T2, T1 with and without fat suppression and T1 after dynamic gadolinium injection) and functional sequences (perfusion and diffusion-weighted sequences). Prospectively, a senior radiologist analyzes the different MR criteria to characterize adnexal masses. The reader will classify the mass using ADNEXMR SCORING system and the patient will be managed according to the score. In control group, the complex adnexal mass will be managed according to the standard strategy and treatment plan routinely used by the multidisciplinary team. Clinical, biological, ultrasonographic and/or MR data and the type of treatment will be recorded. If a surgery is required, surgical procedure will be performed within the two months after initial diagnostic imaging (as routinely recommended).

At M6 and M12, follow-up will be performed. If the lesion has increased or changed with suspicious criteria, surgery will be performed as routinely recommended in both groups.

At M24, adnexal masses status will be recorded by clinical and imaging follow up. This visit marks the end of the study for the patient.

Total duration of the study is 56 months with 32 months for patient inclusions.

Number of centres participating 9 hospitals (17 centers). Identification of the subjects

For this research, the subjects will be identified as follows:

Centre No. (3 numerical positions) - Selection order No. of the person in the centre (4 numerical positions) - surname initial - first name initial This reference is unique and will be retained for the entire research period.

Randomisation

Centralized blocked randomization list will be prepared by the clinical research Unit (URC-EST). The investigators at each hospital will obtain the randomized strategy allocation using internet (CleanWeb, Telemedecin Technologies, S.A.S). Randomization will be stratified on hospitals.

Procedures D0: Gynecologist oncologist or radiologist will ask to the patient to be included in this study. Following informed written consent, patients will be randomized in one of two groups (intervention group or control group) In intervention group: Patients will undergo a pelvic MR imaging in a maximal delay of one month after inclusion (1.5T or 3T) as routinely performed, including morphological sequences (T2, T1 with and without fat suppression and T1 after dynamic gadolinium injection) and functional sequences (perfusion and diffusion-weighted sequences). Prospectively, a senior radiologist) analyzes the different MR criteria to characterize adnexal masses. The reader will classify the mass using ADNEXMR SCORING system and the patient will be managed according to the score, which will be given in the MR report and given to the patient by the clinician in a delay of 15 days after MR imaging.

For masses lower than 4cm: Score ≤3: Follow up / Score = 4: Diagnostic surgery / Score 5 = Oncologic cytoreduction For masses between 4 and 6 cm: Score ≤3: No surgery except for suspicion of dermoid cysts and endometriomas / Score = 4: Diagnostic surgery / Score 5 = Oncologic cytoreduction For masses higher than 6 cm: Score ≤2: No surgery except for dermoid cysts and endometriomas / Score 3 and 4: Diagnostic surgery / Score 5: Oncologic cytoreduction. For extra-adnexal mass: Score = 1 (however have to be treated according to radiologist conclusion).

In control group, the complex adnexal mass will be managed according to the standard strategy and treatment plan routinely used by the multidisciplinary team.

• For suspected endometriomas or dermoid cysts lower than 4cm or functional cysts: Follow up
• For suspected invasive malignant: Oncologic cytoreduction
• For all other lesions: Diagnostic surgery

Clinical, biological, ultrasonographic +/- MR data and the type of treatment will be recorded.

Between M1 and M2: for both, if a surgery is required, surgical procedure with pathological analysis will be performed within the two months after initial diagnostic imaging (as routinely recommended). If an oncologic cytoreduction is decided, pre-operative imaging staging will be performed as routinely recommended.

Follow-up visit (M6 +/- 15D and M12 +/- 15 days)

M6 and M12 (+/- 15 days): Clinical and imaging follow up will be performed, the choice of the imaging is under the appreciation of the clinician.

If the lesion has increased or changed with suspicious criteria, surgery with pathological analysis will be performed as routinely recommended in both groups.

End of research visit (M24 +/- 15 days)

M24 (+/- 15 days): Adnexal mass status will be recorded by clinical and imaging follow up and a phone interview will be done if the patient doesn't come to her M24 visit. This imaging follow-up marks the end of the study for all the remaining patients

Expected length of participation and description of the chronology and duration of the research.

Total study duration: 56 months Duration of recruitment: 32 months Duration of participation for each patient: 24 months

Inclusion Criteria:

• Patient ≥ 18 years old, with no upper limit

• With sonographically indeterminate or complex adnexal mass. Complex or indeterminate adnexal mass is defined as :

  • an adnexal lesion with a solid component which must be characterized or
  • a cystic adnexal lesion for which a solid component cannot easily be formally excluded by ultrasonograghy

• Patient covered by health insurance

• Informed written consent

Exclusion Criteria:

• Pregnant and breast feeding women (relative contra indication for gadolinium injection), pregnancy test will be performed before inclusion for women of childbearing age
• Pacemaker, ferromagnetic materials, or foreign body at risk of mobilization or any other contra-indication to MR imaging.
• Intolerance to iodinated or gadolinium contrast agents, or severe renal insufficiency (GFR <30 ml/min/1.73m²).
• Infertile women (women who unsuccessfully tried to have child during 2 years)
• Pelvic pain which conduct to a surgical indication regardless of imaging results
• Patient deprived of liberty or under legal protection measure