Epithelial ovarian cancer and brain metastases: might the BRCA status, PARP inhibitor administration, and surgical treatment impact the survival?

OBJECTIVE: To evaluate disease characteristics and survival according to BRCA status, administration of poly-(ADP-ribose) polymerase inhibitors (PARPi), and surgery in patients with ovarian cancer and brain metastases. METHODS: This is a monocentric retrospective cohort of patients with ovarian cancer and brain metastases treated between 2000 and 2021. Data were collected by a retrospective review of medical records and analyzed according to: (1) BRCA mutation; (2) PARPi before and after brain metastases; (3) surgery for brain metastases. RESULTS: Eighty-five patients with ovarian cancer and brain metastasis and known BRCA status (31 BRCA mutated (BRCAm), 54 BRCA wild-type (BRCAwt)) were analyzed. Twenty-two patients had received PARPi before brain metastases diagnosis (11 BRCAm, 11 BRCAwt) and 12 after (8 BRCAm, 4 BRCAwt). Brain metastases occurred >1 year later in patients who had received previous PARPi. Survival was longer in the BRCAm group (median post-brain metastasis survival: BRCAm 23 months vs BRCAwt 8 months, p=0.0015). No differences were found based on BRCA status analyzing the population who did not receive PARPi after brain metastasis (median post-brain metastasis survival: BRCAm 8 months vs BRCAwt 8 months, p=0.31). In the BRCAm group, survival was worse in patients who had received previous PARPi (median post-brain metastasis survival: PARPi before, 7 months vs no-PARPi before, 24 months, p=0.003). If PARPi was administered after brain metastases, survival of the overall population improved (median post-brain metastasis survival: PARPi after, 46 months vs no-PARPi after, 8 months, p=0.00038).In cases of surgery for brain metastases, the prognosis seemed better (median post-brain metastasis survival: surgery 13 months vs no-surgery 8 months, p=0.036). Three variables were significantly associated with prolonged survival at multivariate analysis: BRCA mutation, multimodal treatment, and ≤1 previous chemotherapy line. CONCLUSIONS: BRCA mutations might impact brain metastasis occurrence and lead to better outcomes. In a multimodal treatment, surgery seems to affect survival even in cases of extracranial disease. PARPi use should be considered as it seems to prolong survival if administered after brain metastasis.

Circulating tumor DNA as a biomarker for predicting progression-free survival and overall survival in patients with epithelial ovarian cancer: a systematic review and meta-analysis.

OBJECTIVES: Circulating tumor DNA (ctDNA) is emerging as a potential prognostic biomarker in multiple tumor types. However, despite the many studies available on small series of patients with ovarian cancer, a recent systematic review and meta-analysis is lacking. The objective of this study was to determine the association of ctDNA with progression-free-survival and overall survival in patients with epithelial ovarian cancer. METHODS: An electronic search was conducted using PubMed (MEDLINE), Embase, CENTRAL (Cochrane Library), and CINAHL-Complete from January 2000 to September 15, 2023. To be included in the analysis the studies had to meet the following pre-specified inclusion criteria: (1) evaluable ctDNA; (2) progression-free-survival and overall survival reported as hazard ratio (HR); and (3) the patient population had epithelial ovarian cancer at the time of ctDNA detection. We evaluated the association of ctDNA with progression-free survival and overall survival. Secondary outcomes focused on sub-group analysis of genomic alterations and international Federation of Gynecology and Obstetrics (FIGO) stage. RESULTS: A total of 26 studies reporting on 1696 patients with epithelial ovarian cancer were included. The overall concordance rate between plasma-based and tissue-based analyses was approximately 62%. We found that a high level of ctDNA in epithelial ovarian cancer was associated with worse progression-free survival (HR 5.31, 95% CI 2.14 to 13.17, p<0.001) and overall survival (HR 2.98, 95% CI 1.86 to 4.76, p<0.0001). The sub-group analysis showed a greater than threefold increase in the risk of relapse in patients with positive HOXA9 meth-ctDNA (HR 3.84, 95% CI 1.57 to 9.41, p=0.003). CONCLUSIONS: ctDNA was significantly associated with worse progression-free survival and overall survival in patients with epithelial ovarian cancer. Further prospective studies are needed. PROSPERO REGISTRATION NUMBER: CRD42023469390.

Preventing Bone Loss in Breast Cancer Patients: Designing a Personalized Clinical Pathway in a Large-Volume Research Hospital.

BACKGROUND: We assess the impact of bone health clinical management in breast cancer (BC) patients receiving adjuvant endocrine therapy and design a personalized clinical pathway to reduce bone loss in an Italian research hospital. METHODS: The primary endpoint was to assess (through the process improvement organizational method) the clinical pathway that post-surgical BC patients prescribed with endocrine therapy undergo to prevent bone loss. The secondary endpoint was to design a personalized clinical pathway for a prompt implementation of guidelines, to assess and possibly prescribe antiresorptive therapy. RESULTS: During the first year of the execution of the new Diagnostic Therapeutic Assistance Pathway, a 60% increase in Dual-Energy X-ray Absorptiometry evaluations within 30 days and a 39.5% increase in antiresorptive therapy prescription within 90 days (since the prescription of endocrine therapy) were shown, thus increasing patients' compliance. CONCLUSION: Case managers and bone health specialists in this context can improve patients' adherence to therapies and bone health, helping physicians to expand their collaboration.

Distribution and prognostic role of BRCA status in elderly ovarian cancer patients.

OBJECTIVE: In the era of target therapy and personalized medicine, BRCA mutational status has a major influence on survival in ovarian cancer patients. Our aim is to verify if the poorer prognosis of elderly ovarian cancer patients can be related to the biology of the tumor beyond their own morbidities and/or suboptimal treatments. METHODS: This is a retrospective single-institution study evaluating prognosis of patients with a diagnosis of ovarian cancer and known BRCA status. We collected clinical and surgical characteristics and the distribution of BRCA mutational status according to age groups. RESULTS: 1840 patients were included in the analysis. The rate of BRCA mutated decreased over age-range from 49.7% in patients aged <50 years to 18.8% in ≥80 years old women. The prognostic role of BRCA status on survival is maintained when focusing on the elderly population, with improved Disease Free Survival (27.2 months vs 16.5 months for BRCA mutated and wild type respectively, p = 0.001) and Cancer Specific Survival (117.6 months vs 43.1 months for BRCA mutated and wild type respectively, p = 0.001) for BRCAmut compared to BRCAwt patients. In the multivariable analysis, among elderly women, upfront surgery and BRCA mutation are independent factors affecting survival. CONCLUSIONS: Elderly patients experiment a poorer prognosis due to multiple factors that include both their medical condition and comorbidities, under-treatment and most importantly disease characteristics. We found that beyond disparities, BRCA mutation is still the strongest independent prognostic factor affecting both the risk of recurrence and death due to disease.

Sentinel-node biopsy in apparent early stage ovarian cancer: final results of a prospective multicentre study (SELLY).

AIM: To evaluate the sensitivity and specificity of sentinel-lymph-node mapping compared with the gold standard of systematic lymphadenectomy in detecting lymph node metastasis in apparent early stage ovarian cancer. METHODS: Multicenter, prospective, phase II trial, conducted in seven centers from March 2018 to July 2022. Patients with presumed stage I-II epithelial ovarian cancer planned for surgical staging were eligible. Patients received injection of indocyanine green in the infundibulo-pelvic and, when feasible, utero-ovarian ligaments and sentinel lymph node biopsy followed by pelvic and para-aortic lymphadenectomy was performed. Histopathological examination of all nodes was performed including ultra-staging protocol for the sentinel lymph node. RESULTS: 174 patients were enrolled and 169 (97.1 %) received study interventions. 99 (58.6 %) patients had successful mapping of at least one sentinel lymph node and 15 (15.1 %) of them had positive nodes. Of these, 11 of 15 (73.3 %) had a correct identification of the disease in the sentinel lymph node; 7 of 11 (63.6 %) required ultra-staging protocol to detect nodal metastasis. Four (26.7 %) patients with node-positive disease had a negative sentinel-lymph-node (sensitivity 73.3 % and specificity 100.0 %). CONCLUSIONS: In a multicenter setting, identifying sentinel-lymph nodes in apparent early stage epithelial ovarian cancer did not reach the expected sensitivity: 1 of 4 patients might have metastatic lymphatic disease unrecognized by sentinel-lymph-node biopsy. Nevertheless, 35.0 % of node positive patients was identified only thanks to ultra-staging protocol on sentinel-lymph-nodes.

Therapeutic Potential of Tisotumab Vedotin in the Treatment of Recurrent or Metastatic Cervical Cancer: A Short Report on the Emerging Data.

Cervical cancer is the fourth most common type of cancer in women worldwide. It is associated with a high death rate, despite the fact that it is a nearly 100% preventable disease because of very effective primary and secondary preventive strategies. Advanced and recurrent disease is uncurable with a high relapse risk and the second-line therapies are limited with modest response rates and short durability. Investigating alternative mechanisms of action is crucial because of the high request for effective new therapies. Tisotumab vedotin (TV) is the first antibody-drug conjugated to target a cell surface-expressed tissue factor, and preliminary data in patients with metastatic and recurrent cervical cancer have been promising. In addition, the trials showed a favorable tolerability profile, with limited incidence of grade 3 or worse adverse events. According to the data of ENGOT-cx6/GOG-3023/innovaTV 204, the US Food and Drug Administration granted expedited approval of TV on September 20, 2021, for women with recurrent or metastatic cervical cancer. Actually, two other trials testing TV alone or in combination with other agents are ongoing. ENGOT-cx8/GOG-3024/innovaTV 205 is a Phase Ib/II trial of TV in combination with platinum or bevacizumab or pembrolizumab, in patients with recurrent or metastatic cervical cancer who have not received prior systemic therapy or who have progressed after no more than two prior systemic therapies. ENGOT-cx12/GOG-3057/InnovaTV 301 is a Phase 3 trial of TV vs investigator's choice chemotherapy in patients with advanced or recurrent cervical cancer who had received no more than 2 prior chemotherapy lines. The outcomes of these two trials will potentially confirm and reinforce the use of TV as a new standard of care in advanced or recurrent cervical cancer.

Ovarian cancer beyond imaging: integration of AI and multiomics biomarkers.

High-grade serous ovarian cancer is the most lethal gynaecological malignancy. Detailed molecular studies have revealed marked intra-patient heterogeneity at the tumour microenvironment level, likely contributing to poor prognosis. Despite large quantities of clinical, molecular and imaging data on ovarian cancer being accumulated worldwide and the rise of high-throughput computing, data frequently remain siloed and are thus inaccessible for integrated analyses. Only a minority of studies on ovarian cancer have set out to harness artificial intelligence (AI) for the integration of multiomics data and for developing powerful algorithms that capture the characteristics of ovarian cancer at multiple scales and levels. Clinical data, serum markers, and imaging data were most frequently used, followed by genomics and transcriptomics. The current literature proves that integrative multiomics approaches outperform models based on single data types and indicates that imaging can be used for the longitudinal tracking of tumour heterogeneity in space and potentially over time. This review presents an overview of studies that integrated two or more data types to develop AI-based classifiers or prediction models.Relevance statement Integrative multiomics models for ovarian cancer outperform models using single data types for classification, prognostication, and predictive tasks.Key points• This review presents studies using multiomics and artificial intelligence in ovarian cancer.• Current literature proves that integrative multiomics outperform models using single data types.• Around 60% of studies used a combination of imaging with clinical data.• The combination of genomics and transcriptomics with imaging data was infrequently used.

Verification of the prognostic precision of the new 2023 FIGO staging system in endometrial cancer patients - An international pooled analysis of three ESGO accredited centres.

BACKGROUND: Recently, the new 2023 International Federation of Gynecology and Obstetrics (FIGO) staging system for endometrial cancer (EC) critically integrating new pathological and molecular features was published. The present study evaluated the clinical impact of the new 2023 FIGO staging system by comparing it to the previous 2009 system. METHODS: This is an international, pooled retrospective study of 519 EC patients who underwent primary treatment (and molecular characterisation) at three European Society of Gynaecological Oncology (ESGO) accredited centres in Austria/Italy. Patients were categorised according to the 2009 and the 2023 FIGO staging systems. Stage shifts were analysed and (sub)stage specific 5-year progression-free (PFS) and overall survival (OS) rates were calculated and compared. Different statistical tests were applied to evaluate the prognostic precision of the two FIGO staging systems and to compare them to each other. RESULTS: (Sub)stage shifts occurred in 143/519 (27.6%) patients: 123 upshifts (23.7%) and 20 (3.9%) downshifts. 2023 FIGO staging system identified a stage I cohort with a notably higher 5-year PFS rate compared to 2009 (93.0% versus 87.4%, respectively). For stage II disease, the 5-year PFS rate was similar in the 2023 and the 2009 FIGO staging systems (70.2% versus 71.2%, respectively). The two new molecularly defined 2023 FIGO substages IAmPOLEmut and IICmp53abn displayed distinct, particularly favourable and adverse oncologic outcomes within early stage disease, respectively. A remarkably lower 5-year PFS rate for stage III patients was revealed in the 2023 FIGO staging system compared to 2009 (44.4% versus 54.1%, respectively). All applied statistical tests confirmed a more accurate prediction of PFS and OS by the 2023 FIGO staging system compared to 2009. CONCLUSION: The new 2023 FIGO stating system led to a substantial stage shift in about one quarter of patients leading to a higher prognostic precision. In early stage disease, the new substages added further prognostic granularity and identified treatment relevant subgroups.

Unexpected finding of a rare pathogenic germline BRCA1 variant in an intrahepatic cholangiocarcinoma using the Oncomine Focus DNA assay: clinical and diagnostic implications.

BACKGROUND: Cholangiocarcinoma (CCA) is a malignant tumor arising from the epithelial cells of the bile ducts and is the second most common liver cancer after hepatocellular carcinoma. Recently, our Institution launched a Comprehensive Genomic Profiling (CGP) program (named FPG500 program), set up to provide a complete molecular characterization through the TruSight Oncology 500 High Throughput (TSO500HT) solution and samples that do not reach pre-set sample quantity and/or quality thresholds required for TSO500HT, are addressed to Oncomine Focus DNA Assay (OFA) and the Archer's FusionPlex Lung Panel (AFL). METHODS AND RESULTS: Here we report the case of a patient with iCCA enrolled in the FPG500 program and screened by the orthogonal workflow (OFA/AFL). Although BRCA1 is not among the genes declared in the OFA panel, we unexpectedly detected a pathogenic variant in this gene (c.5278-2del, rs878853285). CONCLUSIONS: This case highlights the diagnostic capabilities of CGP, now widely used in both clinical practice and academic setting. The incidental involvement of BRCA1 focuses attention on the role of BRCA genes in biliary tract cancers. Finally, as an orthogonal test confirmed the germline origin of BRCA1 c.5278-2del variant, the germline implications of CGP need to be considered.

Lung cancer multi-omics digital human avatars for integrating precision medicine into clinical practice: the LANTERN study.

BACKGROUND: The current management of lung cancer patients has reached a high level of complexity. Indeed, besides the traditional clinical variables (e.g., age, sex, TNM stage), new omics data have recently been introduced in clinical practice, thereby making more complex the decision-making process. With the advent of Artificial intelligence (AI) techniques, various omics datasets may be used to create more accurate predictive models paving the way for a better care in lung cancer patients. METHODS: The LANTERN study is a multi-center observational clinical trial involving a multidisciplinary consortium of five institutions from different European countries. The aim of this trial is to develop accurate several predictive models for lung cancer patients, through the creation of Digital Human Avatars (DHA), defined as digital representations of patients using various omics-based variables and integrating well-established clinical factors with genomic data, quantitative imaging data etc. A total of 600 lung cancer patients will be prospectively enrolled by the recruiting centers and multi-omics data will be collected. Data will then be modelled and parameterized in an experimental context of cutting-edge big data analysis. All data variables will be recorded according to a shared common ontology based on variable-specific domains in order to enhance their direct actionability. An exploratory analysis will then initiate the biomarker identification process. The second phase of the project will focus on creating multiple multivariate models trained though advanced machine learning (ML) and AI techniques for the specific areas of interest. Finally, the developed models will be validated in order to test their robustness, transferability and generalizability, leading to the development of the DHA. All the potential clinical and scientific stakeholders will be involved in the DHA development process. The main goals aim of LANTERN project are: i) To develop predictive models for lung cancer diagnosis and histological characterization; (ii) to set up personalized predictive models for individual-specific treatments; iii) to enable feedback data loops for preventive healthcare strategies and quality of life management. DISCUSSION: The LANTERN project will develop a predictive platform based on integration of multi-omics data. This will enhance the generation of important and valuable information assets, in order to identify new biomarkers that can be used for early detection, improved tumor diagnosis and personalization of treatment protocols. ETHICS COMMITTEE APPROVAL NUMBER: 5420 - 0002485/23 from Fondazione Policlinico Universitario Agostino Gemelli IRCCS - Università Cattolica del Sacro Cuore Ethics Committee. TRIAL REGISTRATION: clinicaltrial.gov - NCT05802771.

Dual inhibition of CDK12 and CDK13 uncovers actionable vulnerabilities in patient-derived ovarian cancer organoids.

BACKGROUND: High grade serous ovarian cancer (HGSOC) is highly lethal, partly due to chemotherapy resistance and limited availability of targeted approaches. Cyclin dependent kinases 12 and 13 (CDK12/13) are promising therapeutic targets in human cancers, including HGSOC. Nevertheless, the effects of their inhibition in HGSOC and the potential synergy with other drugs are poorly known. METHODS: We analyzed the effects of the CDK12/13 inhibitor THZ531 in HGSOC cells and patient-derived organoids (PDOs). RNA sequencing and quantitative PCR analyses were performed to identify the genome-wide effects of short-term CDK12/13 inhibition on the transcriptome of HGSOC cells. Viability assays with HGSOC cells and PDOs were performed to assess the efficacy of THZ531 as single agent or in combination with clinically relevant drugs. RESULTS: The CDK12 and CDK13 genes are deregulated in HGSOC and their concomitant up-regulation with the oncogene MYC predicts poor prognosis. HGSOC cells and PDOs display high sensitivity to CDK12/13 inhibition, which synergizes with drugs in clinical use for HGSOC. Transcriptome analyses revealed cancer-relevant genes whose expression is repressed by dual CDK12/13 inhibition through impaired splicing. Combined treatment with THZ531 and inhibitors of pathways regulated by these cancer relevant genes (EGFR, RPTOR, ATRIP) exerted synergic effects on HGSOC PDO viability. CONCLUSIONS: CDK12 and CDK13 represent valuable therapeutic targets for HGSOC. We uncovered a wide spectrum of CDK12/13 targets as potential therapeutic vulnerabilities for HGSOC. Moreover, our study indicates that CDK12/13 inhibition enhances the efficacy of approved drugs that are already in use for HGSOC or other human cancers.

Radiomics and Radiogenomics of Ovarian Cancer: Implications for Treatment Monitoring and Clinical Management.

Ovarian cancer, one of the deadliest gynecologic malignancies, is characterized by high intra- and inter-site genomic and phenotypic heterogeneity. The traditional information provided by the conventional interpretation of diagnostic imaging studies cannot adequately represent this heterogeneity. Radiomics analyses can capture the complex patterns related to the microstructure of the tissues and provide quantitative information about them. This review outlines how radiomics and its integration with other quantitative biological information, like genomics and proteomics, can impact the clinical management of ovarian cancer.

Back to the future: The impact of oestrogen receptor profile in the era of molecular endometrial cancer classification.

PURPOSE: The aim of this study is to evaluate the impact of the oestrogen receptor (ER) profile on oncologic outcomes in the new endometrial cancer (EC) risk classification. METHODS: Immunohistochemistry (IHC) analyses were performed in a retrospectively reviewed large series of ECs to assess the presence/absence of oestrogen receptors (ER0\1+ or ER2+\3+) and other molecular factors (i.e. p53 mutation, p53mut; and mismatch repair mutational status, MMRd (mismatch repair deficient) versus MMRp (mismatch repair proficient)), histopathologic and clinical outcomes. ER status was correlated with molecular, histologic, clinical and prognostic data. RESULTS: 891 EC patients were included in the study (211 ER0\1+ and 680 ER2+\3+). The ER0\1+ phenotype was associated with an unfavourable clinicopathological profile (i.e. grading, histotype, lymphovascular space invasion (LVSI), stages, etc.). Simple regression showed that risk class, p53mut, and ER0/1+ impacted on both disease-free survival (DFS) and overall survival (OS) (p < 0.05). In the ER0/1+ population, p53mut no longer influenced DFS and OS (p > 0.05). In multiple regression, age, high and advanced/metastatic risk classes influenced survival outcomes (p < 0.05), but lost significance in the ER0/1+ population (p > 0.05). ER-positivity retained a remarkable prognostic impact even after stratification of the population according to the European Society of Gynaecological Oncology, the European Society for Radiotherapy and Oncology, and the European Society of Pathology (ESGO/ESTRO/ESP) 2021 risk classes and molecular classification. ER0/1+ intermediate, high-intermediate, high and advanced risk versus ER2+/3+ intermediate, high-intermediate, high and advanced risk classes showed statistically different OS and DFS (p< 0.001). ER0/1+ status was associated with a worse prognosis when associated with MMRp, MMRd and p53mut compared to the same molecular classes associated with ER2+/3 (p < 0.001). CONCLUSIONS: We demonstrated that ER status has a significant impact on oncologic outcomes, regardless of risk class and p53/MMR status. Based on our results, we recommend the inclusion of ER assessment in featured EC risk classification system.

Recent progress in the use of pharmacotherapy for endometrial cancer.

INTRODUCTION: Endometrial cancer (EC) is the most common gynecological cancer in developed countries. The ESGO/ESTRO/ESP updated evidence-based guidelines in 2020, introducing molecular classification to guide EC treatment. The genomic-based approach has identified four prognostic subgroups of EC. Each of these may benefit from a tailored treatment depending on the molecular profile, the histotype, and stage of disease for the adjuvant and the metastatic/recurrent setting. Several clinical trials are now ongoing to identify the best treatment according to the molecular profile of EC. AREAS COVERED: This review analyzes tailored treatment for EC according to the molecular profile, both in the adjuvant and in the metastatic/recurrent setting. The authors review the results of clinical studies and highlight ongoing trials. EXPERT OPINION: Several new agents are under evaluation in order to personalize EC treatment according to specific molecular profiles in the adjuvant, advanced, and recurrent settings. Clinical trials investigating the impact of molecular classification have yielded encouraging results. EC can no longer be considered a single tumor entity susceptible to a single treatment modality but rather be split into four distinct types, requiring tailored treatments.

A Computational Framework for Comprehensive Genomic Profiling in Solid Cancers: The Analytical Performance of a High-Throughput Assay for Small and Copy Number Variants.

In January 2022, our institution launched a comprehensive cancer genome profiling program on 10 cancer types using a non-IVD solution named the TruSight Oncology 500 Assay provided by Illumina®. The assay analyzes both DNA and RNA, identifying Single-Nucleotide Variants (SNV)s and Insertion-Deletion (InDel) in 523 genes, as well as known and unknown fusions and splicing variants in 55 genes and Copy Number Alterations (CNVs), Mutational Tumor Burden (MTB) and Microsatellite Instability (MSI). According to the current European IVD Directive 98/79/EC, an internal validation was performed before running the test. A dedicated open-source bioinformatics pipeline was developed for data postprocessing, panel assessment and embedding in high-performance computing framework using the container technology to ensure scalability and reproducibility. Our protocols, applied to 71 DNA and 64 RNA samples, showed full agreement between the TruSight Oncology 500 assay and standard approaches, with only minor limitations, allowing to routinely perform our protocol in patient screening.

Deep-Learning to Predict BRCA Mutation and Survival from Digital H&E Slides of Epithelial Ovarian Cancer.

BRCA 1/2 genes mutation status can already determine the therapeutic algorithm of high grade serous ovarian cancer patients. Nevertheless, its assessment is not sufficient to identify all patients with genomic instability, since BRCA 1/2 mutations are only the most well-known mechanisms of homologous recombination deficiency (HR-d) pathway, and patients displaying HR-d behave similarly to BRCA mutated patients. HRd assessment can be challenging and is progressively overcoming BRCA testing not only for prognostic information but more importantly for drugs prescriptions. However, HR testing is not already integrated in clinical practice, it is quite expensive and it is not refundable in many countries. Selecting patients who are more likely to benefit from this assessment (BRCA 1/2 WT patients) at an early stage of the diagnostic process, would allow an optimization of genomic profiling resources. In this study, we sought to explore whether somatic BRCA1/2 genes status can be predicted using computational pathology from standard hematoxylin and eosin histology. In detail, we adopted a publicly available, deep-learning-based weakly supervised method that uses attention-based learning to automatically identify sub regions of high diagnostic value to accurately classify the whole slide (CLAM). The same model was also tested for progression free survival (PFS) prediction. The model was tested on a cohort of 664 (training set: n = 464, testing set: n = 132) ovarian cancer patients, of whom 233 (35.1%) had a somatic BRCA 1/2 mutation. An area under the curve of 0.7 and 0.55 was achieved in the training and testing set respectively. The model was then further refined by manually identifying areas of interest in half of the cases. 198 images were used for training (126/72) and 87 images for validation (55/32). The model reached a zero classification error on the training set, but the performance was 0.59 in terms of validation ROC AUC, with a 0.57 validation accuracy. Finally, when applied to predict PFS, the model achieved an AUC of 0.71, with a negative predictive value of 0.69, and a positive predictive value of 0.75. Based on these analyses, we have planned further steps of development such as proving a reference classification performance, exploring the hyperparameters space for training optimization, eventually tweaking the learning algorithms and the neural networks architecture for better suiting this specific task. These actions may allow the model to improve performances for all the considered outcomes.