Objective A novel classification system of high-grade serous ovarian carcinoma based on gross morphology observed at pre-treatment laparoscopy was recently defined. The purpose of this study was to identify radiographic features unique to each morphologic subtype.
Methods This retrospective study included 109 patients with high-grade serous ovarian cancer who underwent pre-operative computed tomography (CT) scanning and laparoscopic assessment of disease burden between 1 April 2013 and 5 August 2015. Gross morphologic subtype had been previously assigned by laparoscopy. Two radiologists independently reviewed CT images for each patient, categorized disease at eight anatomic sites, and assessed for radiographic characteristics of interest: large infiltrative plaques, mass-like metastases, enhancing peritoneal lining, architectural distortion, fat stranding, calcifications, and lymph node involvement. Demographic and clinical information was summarized with descriptive statistics and compared using Student's t-tests, χ² tests, or Fisher exact tests as appropriate; kappa statistics were used to assess inter-reader agreement.
Results Certain radiographic features were found to be associated with gross morphologic subtype. Large infiltrative plaques were more common in type 1 disease (88.7% (47/53) vs 71.4% (25/35), p=0.04), while mass-like metastases were more often present in type 2 disease (48.6% (17/35) vs 22.6% (12/53), p=0.01). Additionally, radiographic presence of disease at the falciform ligament was more common in type 1 morphology (33.9% (19/56) vs 13.2% (5/38), p=0.02).
Conclusion Morphologic subtypes of high-grade serous ovarian cancer were associated with specific CT findings, including the presence of large infiltrative plaques, mass-like metastases, and falciform ligament involvement.
- Ovarian Cancer
- Cystadenocarcinoma, Serous
Data availability statement
Data are available upon reasonable request. Deidentified participant data are available from the corresponding author (Anil K. Sood, email@example.com, ORCID 0000-0003-4242-1762) upon reasonable request.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
WHAT IS ALREADY KNOWN ON THIS TOPIC
A morphology-based classification of high-grade serous ovarian cancer, wherein tumors are stratified into two subtypes based on differences in gross appearance, was recently described. These morphologic subtypes have distinct clinical and molecular features, suggesting possible implications for personalized oncologic care.
WHAT THIS STUDY ADDS
This study validates the concept that the morphologic subtypes of high-grade serous ovarian cancer are associated with CT findings, and thus may be identified non-invasively.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
Though additional work is required for clinical translation, this study prompts further research into the radiographic correlates of the morphologic subtypes of ovarian cancer. As understanding of these morphologic subtypes unfolds, knowledge of their radiographic appearance may aid in treatment planning.
Ovarian cancer is the fifth most common cause of cancer-related death in women in the USA.1 Despite tireless research efforts and the emergence of new therapies over the past decade, the prognosis of patients with advanced-stage ovarian cancer remains poor, with a 5-year survival rate of 30%.2 Most patients respond to first-line therapy,3 but many quickly develop relapsed disease that becomes progressively resistant to therapy.4 Intertumoral and intratumoral heterogeneity has been recognized as a hallmark of ovarian cancer and likely contributes to the high rate of treatment resistance.5 Even within the singular histology of high-grade serous carcinoma, the most common and most lethal ovarian malignancy,6 some subtypes portend better outcomes. For example, patients with germline BRCA1/2 mutations or other deficiencies in the homologous recombination pathway are known to have disease that is sensitive to poly ADP-ribose polymerase (PARP) inhibitors7 and have improved survival.8
To better understand how the morphologic variability of high-grade serous ovarian cancer may also be clinically relevant, Handley et al9 identified and described two gross morphologic subtypes based on pre-treatment laparoscopic assessment. Type 1 disease was defined as infiltrative disease that distorts surrounding anatomy, while type 2 morphology was described as exophytic with well-defined borders. Extensive multi-omics analysis of type 1 and type 2 tumor samples revealed distinct clinical outcomes and molecular profiles.9 Given the widespread use of computed tomography (CT) prior to treatment in patients with ovarian cancer,10 11 we asked whether these gross morphologic subtypes could be detected and distinguished on CT imaging. Specifically, we aimed to identify radiographic features unique to each morphologic subtype of high-grade serous ovarian cancer.
Since 1 April 2013, all patients presenting to our institution with suspected advanced-stage ovarian, fallopian tube, or primary peritoneal cancer who are considered possible candidates for primary tumor-reductive surgery have been offered laparoscopic assessment to determine the feasibility of optimal resection. Laparoscopic assessments are video recorded. Demographic and clinical data for this patient population are collected prospectively.12
This retrospective cohort study included patients who underwent laparoscopic assessment of disease burden at a single large referral center between 1 April 2013 and 5 August 2016 and had videos reviewed to determine morphologic subtype; all patients had treatment-naïve high-grade serous ovarian cancer and had pre-operative CT imaging available for review. This mature cohort was selected because of the availability of morphologic subtype data and adequate follow-up time (median observation time 4.09 years, IQR 2.26–5.83 years). Consecutive sampling was employed. As sample size was limited by the number of patients included in the earlier study, an a priori power analysis was not performed. Clinical and demographic information was extracted from a secure database on 23 September 2021. This study was approved by our Institutional Review Board (PA16-1010), and a waiver of informed consent was obtained. Patient confidentiality was maintained in accordance with the Health Insurance Portability and Accountability Act.
Gross morphologic subtype (type 1 or type 2) had been previously assigned for these patients based on review of video recordings of pre-treatment laparoscopy by at least two physicians. Type 1 disease was defined as deep and infiltrative, characterized by architectural distortion, raised plaques, and miliary disease. Type 2 morphology was described as superficial and exophytic, with well-defined borders and normal surrounding tissue. Morphologic subtype was considered to be predominant if that subtype was identified at the majority of anatomic sites where disease was present (among the diaphragm, peritoneum, pelvic peritoneum, and omentum).9
For training purposes, after being provided with descriptions of type 1 and type 2 morphologic subtypes, one radiologist viewed laparoscopic videos and CT images of representative cases of type 1 and type 2 morphology and developed descriptors for corresponding imaging findings. Then, two radiologists independently performed a preliminary review of CT images for each patient, categorizing disease as type 1 or type 2 at eight anatomic sites (diaphragm, peritoneum, pelvic peritoneum, omentum, falciform ligament, mesentery, porta hepatis, and liver) if present. Following this preparatory review, gynecologic surgeons and radiologists collaborated to identify specific characteristics of interest which were thought to be radiologically identifiable and potentially distinct between the two morphologic subtypes. These characteristics were: large infiltrative plaques (thick rind of tissue abutting/infiltrating adjacent organs; Figure 1A), mass-like metastases (lesions larger than 2 cm; Figure 1B), enhancing peritoneal lining, architectural distortion (spiculated appearance of the implants causing tethering of adjacent structures; Figure 1C), fat stranding (wispy areas in the omentum), calcifications, and lymph node involvement (Online supplemental table 1). Two radiologists then independently performed formal assessments of the CT images of each patient in the cohort and recorded the presence or absence of these characteristics. In some cases, no response was selected, at the discretion of the reading radiologist.
For each imaging review, a research assistant presented the imaging studies to two of the three radiologists specializing in gynecologic oncologic imaging, who then evaluated the imaging studies independently on a picture archiving and communication system (PACS) workstation. Reads were performed independently; kappa statistics (range −1 to 1) were calculated to assess inter-reader agreement. Radiologists were blinded to morphologic subtype assigned by laparoscopy and other clinical data. Bias was minimized by independent review of images by multiple radiologists and blinding to clinical outcomes.
The reviewed pre-operative CT imaging included scans performed at our own institution (39.4% (43/109)) as well as outside facilities (60.6% (66/109)). For this reason, imaging quality and specifications were heterogeneous. However, all the pre-operative CT scans were obtained in the portal venous phase of contrast enhancement where applicable. Maximum slice thickness was ≤5 mm for all patients. Detector pitch ranged from 0.45:1 to 1.75:1. Intravenous contrast was utilized in nearly all cases (97.2% (106/109)); contrast agents included iohexol, iopamidol, and iodixanol. Oral contrast, such as iohexol or barium, was given in 88 cases (80.7%). On average, imaging was completed 20±22 (mean±SD) days prior to surgery.
For statistical analysis, morphologic subtype was defined as the predominant morphologic subtype assigned by laparoscopy in the prior study.9 Demographic and clinical characteristics were described by frequencies and percentages and compared between morphologic types 1 and 2 by unpaired Student's t-tests for continuous variables and χ² square or Fisher exact tests for categorical variables. Radiographic presence of disease at the evaluated anatomic locations was defined as at least one radiologist recording type 1 or type 2 disease on evaluation of CT images. Absence of disease was defined as the converse, with neither radiologist indicating type 1 or type 2 disease at the given location. Radiographic presence of a feature of interest was defined as at least one radiologist noting presence of that feature. Absence of a feature was defined as both radiologists documenting absence of the feature. Radiographic characteristics were compared between type 1 and type 2 morphologic subtypes using χ² square or Fisher exact tests. Statistical significance was defined as p<0.05. Statistical analyses were performed using Stata/MP v16.0 (StataCorp LLC) and GraphPad Prism v8.0.0 (GraphPad Software).
In accordance with the journal’s guidelines, we will provide our data for independent analysis by a team selected by the editorial team for the purposes of additional data analysis or for the reproducibility of this study in other centers if such is requested.
During the study period, 566 patients were considered candidates for laparoscopic assessment of suspected advanced-stage ovarian cancer. Of these, 272 underwent laparoscopic assessment of disease burden and 176 were confirmed to have high-grade serous ovarian carcinoma. A total of 112 videos were reviewed to assign morphologic subtype, as previously reported by Handley et al.9 One hundred and nine had pre-operative CT scans of the abdomen and pelvis available for review and were thus included in the study (Online supplemental figure 1).
The study population comprised 56 patients (51.4%) with type 1 morphology, 38 (34.9%) with type 2 morphology, and 15 (13.8%) with no predominant type. Demographic and clinical characteristics of the population are presented in Online supplemental table 2. All 109 patients had high-grade serous ovarian carcinoma by virtue of study inclusion; most had primary ovarian (86.2% (94/109)) and stage IIIC (83.5% (91/109)) disease. Demographic traits were similar between patients with type 1 and type 2 disease, except for ethnicity. Clinical characteristics such as stage, platinum sensitivity, progression-free survival (16.1 vs 17.2 months, p=0.51), and overall survival (22.9 vs 33.0 months, p=0.46) did not differ significantly between the two subtypes. Although there was no difference in the timing or approach of tumor-reductive surgery, several surgical outcomes differed between patients with type 1 and type 2 morphology (Table 1).
Radiographic presence of disease at a given anatomic site was not associated with morphologic subtype, with the exception of the falciform ligament (Table 2); falciform involvement was more common in type 1 morphology (33.9% (19/56) vs 13.2% (5/38), p=0.02). Inter-reader agreement (κ) between the two radiologists assessing radiographic presence of disease ranged from 0.22 to 0.58 (mean±SD 0.38±0.12) at the specified locations.
Two radiographic features of interest were found to be associated with morphologic subtype (Table 3). Specifically, large infiltrative plaques were more common in type 1 disease (88.7% (47/53) vs 71.4% (25/35), p=0.04), and mass-like metastases were more often seen in type 2 disease (48.6% (17/35) vs 22.6% (12/53), p=0.01). Representative images are shown in Figure 2. Inter-reader agreement (κ) between radiologists documenting these radiographic characteristics ranged from 0.12 to 0.79 (mean±SD 0.43±0.28).
Summary of Main Results
In the context of recent efforts to understand the gross morphologic variations in ovarian cancer, we aimed to identify radiographic features of type 1 and type 2 morphologic subtypes of high-grade serous ovarian carcinoma. The primary finding of this study is that there are indeed radiologic differences observed in patients with high-grade serous ovarian cancer that correspond to gross morphological subtypes identified during laparoscopic assessment. Large infiltrative plaques on CT were more common in type 1 disease, as was radiographic presence of disease involving the falciform ligament. Mass-like metastases were more often seen in type 2 disease.
Results in the Context of Published Literature
Historically, high-grade serous ovarian cancer has not been categorized according to gross morphology; however, a wide range of gross appearances can be easily observed at the time of laparoscopic assessment or tumor-reductive surgery. In a word, both the gross and histologic appearances of high-grade serous ovarian cancer have been described as variable. The primary tumor may range from solid to cystic and subcentimeter to massive, with or without varying types of necrosis.13 Despite these observations by surgeons and pathologists alike, there have been few attempts to define or understand the array of gross morphologic subtypes prior to the classification system of Handley et al.9
As would be anticipated, the radiographic appearance of ovarian cancer is also variable.14 Though there is great interest in optimizing the use of imaging in assessment and treatment of high-grade serous ovarian cancer,15 16 no previous research has assessed the utility of imaging in evaluating the morphologic characteristics of ovarian cancer. Here, we found that type 1 and type 2 morphologic subtypes of high-grade serous ovarian cancer have distinct radiographic features. Large infiltrative plaques on CT scans, found more frequently in type 1 disease, likely represent the deep, infiltrating disease and raised plaques described on laparoscopy by Handley et al.9 Mass-like metastases seen on imaging in patients with type 2 morphology may parallel its exophytic gross appearance.
Strengths and Weaknesses
This study provides imaging correlates of gross morphologic subtypes by CT imaging. We evaluated a sizeable population with complete, centralized clinical data. Our patient population was representative of patients with ovarian cancer in the USA, with similar age of diagnosis,2 racial and ethnic breakdown,17 and BRCA mutation status18 to that of the broader population; thus, our results are expected to be externally valid.
One limitation of this study was inter-reader agreement in assessments of CT scans. CT images were obtained at different institutions, on different scanners, and with different imaging techniques. Furthermore, the imaging studies were not tailored to stage ovarian cancer. Some did not have oral contrast, and some had slice thicknesses up to 5 mm. This was likely compounded by the challenges of elucidating and describing peritoneal involvement by CT. This study was also limited by its sample size, which was restricted by the number of patients previously assessed for morphologic subtype. Although the sample size was adequate to detect the differences reported here, there may be additional distinctions which were not recognized owing to suboptimal power.
Implications for Practice and Future Research
Epithelial ovarian cancer is traditionally categorized by histologic subtype (serous, mucinous, endometrioid, etc), as well as by stage and grade.5 Although these broad classifications currently guide treatment and give general information regarding prognosis,19 patients with advanced-stage high-grade serous ovarian cancer may experience quite variable disease courses. Overall survival in the cohort of ovarian cancer patients in The Cancer Genome Atlas Ovarian Cancer ranges quite broadly from 1 week to 13 years at last follow-up. These differences are likely caused by a complex combination of patient and disease characteristics including genetic profile, tumor microenvironment, and other yet-to-be-determined factors. Even within a single tumor, genotypic20 and phenotypic21 heterogeneity is present. These nuanced molecular distinctions have been shown to correlate with clinical outcomes22–24 and treatment response.25–27
This study represents the first step toward using CT imaging to classify high-grade serous ovarian cancer by gross morphologic subtype. Efforts to further understand the implications of these subtypes and refine their detection are required. However, given the initial reports of clinical and molecular significance by Handley et al,9 we anticipate radiographic detection of morphologic subtype to be clinically impactful. For example, patients with type 2 morphology had a significantly higher estimated blood loss and operative time during tumor reductive surgery. If these findings are confirmed, they may be relevant to primary surgical planning, in which case it would be valuable to identify morphologic subtype pre-operatively. Likewise, molecular signatures associated with each subtype (eg, enriched PI3K/AKT/mTOR signaling in type 1 tumors) may indicate utility of particular targeted therapies.
Additional work in an expanded patient population to improve inter-reader agreement is planned. Future studies would ideally be prospective in nature, and should expand the roster of radiographic characteristics in order to more comprehensively define the radiographic appearance of type 1 and type 2 high-grade serous ovarian carcinoma.
We were able to successfully identify radiographic correlates of morphologic subtypes of high-grade serous ovarian cancer. As our understanding of these morphologic subtypes unfolds, knowledge of their radiographic appearance may aid in treatment planning.
Data availability statement
Data are available upon reasonable request. Deidentified participant data are available from the corresponding author (Anil K. Sood, firstname.lastname@example.org, ORCID 0000-0003-4242-1762) upon reasonable request.
Patient consent for publication
This study involves human participants and was approved by The University of Texas MD Anderson Cancer Center Institutional Review Board, PA16-1010. A Waiver of Informed Consent was granted by the Institutional Review Board because this was a retrospective data review that involved no diagnostic or therapeutic intervention, and no direct patient contact.
We thank Kelly Rangel, Supervisor of Clinical Studies, Department of Gynecologic & Reproductive Medicine, The University of Texas MD Anderson Cancer Center for her indispensable research support. We thank Amy Ninetto, Scientific Editor, Research Medical Library, The University of Texas MD Anderson Cancer Center, for editing the manuscript. We thank Kelly Kage, Medical Illustrator, Department of Diagnostic Imaging, The University of Texas MD Anderson Cancer Center for her illustrations. Preliminary results of this study were presented at the 53rd Annual Meeting on Women’s Cancer, The Society of Gynecologic Oncology, Phoenix, AZ, March 18-21, 2022.
KIF and KFH contributed equally.
Contributors The authors contributed to this work as follows. KIF: conceptualization, formal analysis, investigation, writing – original draft, writing – review & editing, visualization. KFH: conceptualization, methodology, formal analysis, investigation, writing – original draft, writing – review & editing, funding acquisition. DG: methodology, investigation, writing – review & editing. TTS: investigation, writing – review & editing. SJ: investigation, writing – review & editing. SMP: investigation, writing – review & editing. MMS: investigation, data curation, writing – review & editing. BMF: methodology, formal analysis, writing – original draft, writing – review & editing. NDF: conceptualization, methodology, writing – review & editing. PRB: conceptualization, methodology, investigation, resources, writing – original draft, writing – review & editing, visualization, supervision. AKS: conceptualization, methodology, resources, writing – original draft, writing – review & editing, supervision, funding acquisition. Katherine I. Foster and Katelyn F. Handley contributed equally to this paper, warranting joint first authorship. AKS is the guarantor of the overall work and accepts full responsibility for the finished work and conduct of the study, had access to the data, and controlled the decision to publish.
Funding This research was supported in part by the MD Anderson Ovarian Cancer Moon Shot; the National Institutes of Health through grants CA016672 (MD Anderson’s Cancer Center Support Grant), CA217685, CA101642, and CA209904; the American Cancer Society; the Ovarian Cancer Research Alliance; and the Frank McGraw Memorial Chair in Cancer Research. KFH is supported by a training fellowship from the Gulf Coast Consortia, through the Computational Cancer Biology Training Program (CPRIT Grant No. RP170593). Funding sources were not involved in study design, collection, analysis, or interpretation of data, writing the report, or decision to submit for publication.
Competing interests AKS is a consultant for AstraZeneca, GSK/Tesaro, KIYATEC, ImmunoGen, Iylon, Merck, and Onxeo and is a shareholder of BioPath.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.