Molecular and therapeutic stratification of endometrioid ovarian carcinoma
View/ Open
Date
31/07/2021Author
Stanley, Barbara
Metadata
Abstract
Epithelial ovarian carcinomas (EOC) comprise five main histological subtypes, each displaying
distinct pathological, molecular and clinical characteristics. Endometrioid ovarian carcinomas
(EnOC) account for 10% of EOC and have been historically under-investigated. They typically
present as early stage, grade 1 or 2 (low grade) tumours arising from endometriosis, and are
associated with excellent clinical outcomes. However, Grade 3 (high grade) EnOC, as well as
the even rarer de-differentiated carcinomas, can be challenging to differentiate from high grade
serous ovarian carcinomas (HGSOC) based on morphology alone. Through the refinement
of EnOC diagnostic criteria, several studies have now demonstrated that many previously
diagnosed high grade EnOC are in fact HGSOC. This is further supported by gene expression
profiling studies demonstrating that a proportion of high grade EnOC cluster together with
HGSOC. As such, true high grade EnOC are increasingly rare and are associated with poor
prognosis. WT1 immunohistochemistry (IHC) is a useful tool to discriminate high grade EnOC
(WT1 negative (WT1-ve)) from HGSOC (WT1 positive), reducing inter-observer variation.
To date, clinical and molecular characterisation of EnOC has been confounded by the
inclusion of historically misclassified HGSOC in older studies. Mutational analysis performed
by more recent studies have either only been applied to low grade EnOC, or lack information
on grade or diagnostic criteria used. As a result, the molecular landscape and clinical
behaviour of EnOC, in particular high grade EnOC, is not well defined.
In this study, tumours historically diagnosed as EnOC were identified through the Edinburgh
Ovarian Cancer Database. Contemporary pathology review was performed utilising WT1 and
p53 IHC. WT1-ve EnOC of all grades, and WT1-ve tumours with high grade serous and
undifferentiated morphology were identified and included in the primary analysis. Clinical
characteristics of the primary cohort were extracted from the database. Survival analysis was
performed and responses to chemotherapy and endocrine therapy recorded. 63 tumours from
the primary cohort underwent DNA extraction and whole exome sequencing (WES);
comprising all WT1-ve tumours with mutant p53 expression on immunohistochemistry
(p53mut(IHC)) (n=28), all WT1-ve high grade carcinomas with p53 wild-type expression on
immunohistochemistry (p53wt (IHC)) (n=12) alongside a randomly selected subset of WT1-ve
p53wt (IHC) low grade EnOC tumours (23 of 87 cases, 26.4%). Supervised mutational and copy
number analysis was performed across 75 commonly mutated genes previously reported in
endometrial, ovarian or pan cancer studies and molecular subgroups were identified.
Unsupervised clustering analysis validated these molecular subgroups. Hormone receptor
expression levels (oestrogen receptor (ER), progesterone receptor (PR) and androgen
receptor (AR)) were evaluated as histoscores in the primary cohort. Multivariable survival
analysis, accounting for stage, residual disease, decade of diagnosis. grade and age, was
performed on resulting molecular and hormone receptor subgroups.
Between May 1980 and December 2013, 125 WT1-ve tumours were identified. Overall five year
disease specific survival (DSS) was 73.2% with the most favourable prognosis in those with
early stage disease. Five year DSS in patients with advanced stage p53wt (IHC) low grade EnOC
was 50.0%. Late relapses beyond five years were common in early stage disease. Patients
with stage IV disease had poor prognosis with median DSS of less than one year. Radiological
and CA125 response rates to platinum based chemotherapy in evaluable tumours was 44.5%
and 69.3%, respectively. Median duration of endocrine therapy in evaluable tumours was 317
days (range 35 – 615 days). Of the 61 tumours successfully sequenced, TP53 mutations
(TP53mut) were the most common (45.9%); followed by mutations in EnOC-associated genes
(ARID1A (41.0%), CTNNB1 (31.1%), PTEN (24.6%) and PIK3CA (23.0%)). Only TP53mut
status was independently associated with shortened DSS (HR=0.35, 95% CI 0.14-0.83,
P=0.018). Copy number analysis revealed significantly more alterations in the TP53mut tumours
compared to TP53wt tumours (P <0.0001), with a particular enrichment of variation across
EnOC-associated genes in the TP53mut subgroup with no EnOC-associated gene mutations.
The majority of the primary cohort expressed ER and PR whereas AR expression was low. A
PR histoscore of >150, when compared to a PR histoscore ≤ 150, was found to be
independently associated with DSS, whereas no associations were observed with ER or AR
expression levels. In particular, patients with stage II EnOC and a PR histoscore of >150
displayed a ten year DSS of over 90%.
Through this study, TP53 mutation status and a PR histoscore of greater than 150 were
identified as independent predictors of survival. This demonstrates EnOC to be a
heterogeneous disease with distinct molecular and hormone receptor subgroups that
demonstrate differential clinical outcome. Patients with TP53 mutated and/or low PR
expression EnOC have inferior prognosis and the development of novel therapeutic agents
should be focused on these groups which display the greatest unmet need.