ABSTRACT
Epithelial ovarian cancer is the deadliest gynecologic malignancy among women, with less than half of women
surviving five years after diagnosis. Black women with ovarian cancer have worse survival compared to White
women. The causes of these disparities remain elusive as prior research suggests that it is not entirely due to
differential access to care or guideline-adherent treatment. Thus, it remains a high priority to uncover approaches
to reduce mortality and improve survival, especially for Black women. However, the majority of research on
ovarian cancer is from White women, which has hindered the discovery of novel factors important for prognosis
in underrepresented minority groups. Here, we will leverage epidemiologic, molecular, and outcome data from
well-established observational studies to investigate the methylomic basis of ovarian cancer survival disparities
between Black and White women. DNA methylation provides a unique opportunity to investigate disparities as
lifestyle and sociocultural conditions that are disparate between racial/ethnic groups may manifest as alterations
in tumor DNA methylation, resulting in phenotypic differences between populations. We hypothesize that Black
women will have a different composition of methylation patterns or a unique tumor DNA methylation signature
associated with poorer survival compared to White women. To limit contributions of disease heterogeneity, we
will focus on the most common and one of the deadliest histotypes, high-grade serous ovarian carcinoma
(HGSOC). Among 239 Black and 478 White women with HGSOC, tumor DNA methylation will be measured
using the Illumina MethylationEPIC array. Frist, data dimension reduction methods will be used to determine
tumor DNA methylation signatures that are associated with survival among the overall study population and
among Black and White women separately, identifying differentially methylated regions associated with
outcomes that are specific to each race and those that are shared across race. These signatures will be validated
among an additional 200 Black and 200 White women with HGSOC. Second, the association between pre-
diagnostic exposures that have the potential to alter DNA methylation states (e.g., age at diagnosis, smoking
status) and outcome-associated DNA methylation signatures will be investigated to determine which factors may
be informative for preventive strategies. As DNA methylation is highly tissue specific, the DNA methylation
signature of the tumor is a weighted mixture of the methylation signature for each of the cells within the tumor.
Therefore, in the third aim, we will infer cell composition from tumor DNA methylation data using a novel cell
mixture deconvolution method, and examine whether inferred cell composition is associated with risk of mortality.
Comparing DNA methylation across populations has important applications as this work will advance the
discovery of molecular targets among an underrepresented racial minority group, aiding in clinical decision-
making and informing the development of personalized therapies to reduce mortality in Black women and
ultimately reduce the survival gap between Black and White women with ovarian cancer.