Introduction: Ovarian cancer (OC) remains the major killer in gynecological oncology. In Denmark,
the 5-year survival rate is around 30%, which is primarily caused by advanced stage at diagnosis,
recurrence of disease and ultimately resistance to chemotherapy. The few vague symptoms
explain why the majority of patients are diagnosed in advanced stage leaving few curative options
despite progress in surgery and medical treatment.Evaluation of treatment effect is another major challenge, especially due to the spread of OC
throughout the peritoneal cavity in the form of carcinosis, which is difficult to measure in
centimeters according to the international RECIST criteria.
Finding biomarkers that could contribute to the management of OC by monitoring treatment
response, detecting recurrence, distinguishing benign from malignant pelvic masses, and
potentially diagnose OC at an earlier stage has been of major interest, but no reliable candidate
has been identified. The currently used biomarker, CA125, does not meet the criteria of an ideal
marker, as a large proportion of OC expresses little or no CA125, and the sensitivity and specificity
is poor.Aberrant methylation of CpG islands is a general feature of cancer cells and occurs in almost all
malignant tumors. The methylated genes often examined represent developmental regulatory
genes important for embryogenesis, including the homeobox (HOX) genes. Methylation of the
HOXA9 gene has particularly been associated with OC.
Part of the methylated DNA from the tumor is shed into the circulation and is detectable as
circulating tumor DNA (ctDNA), which can be measured with high sensitivity and at low analytical
variation. Analysis of plasma would overcome the intratumoral heterogeneity, it is a timely
measurement, convenient to the patients, and sufficient material is easily accessed.
The aim of this thesis was to investigate the clinical potential of methylated HOXA9 with respect to
diagnosis, prognosis, and treatment monitoring.
Materials and methods: The studies are based on different cohorts of patients with OC. Tissue as
well as plasma from newly diagnosed patients (FIGO stage I-IV) was included for evaluation of the
diagnostic potential of methylated HOXA9 and for validation of the assays and method used. To
study the clinical aspects of methylated HOXA9 ctDNA (meth-HOXA9) during treatment patients
with recurrent OC were included.DNA was extracted from 3 x 15 µm tissue specimens fixed in formalin and embedded in paraffin
(FFPE) and from 4 mL plasma. Following bisulfite conversion methylated HOXA9 was analyzed
using methylation specific droplet digital PCR. Detection of methylated HOXA9 was reported as a
percentage of total cell free DNA and as a binary variable (detectable and undetectable).
Paper I: For diagnostics, methylated HOXA9 was detected in 93.2% (82/88) of patients with OC, in
87.5% (14/16) of borderline tumors, in 16.7% (3/18) of benign tumors, and in none of the 16
normal ovaries. Using receiver operating characteristic (ROC) analysis methylated HOXA9 had a
diagnostic accuracy of 98%. Paper II: In plasma, 59.5% (47/79) of newly diagnosed OC patients had detectable meth-HOXA9,
15 of which had FIGO stage I-II (detection rate=37.5%) and 32 had FIGO stage III-IV disease
(detection rate=82.1%). The sensitivity was increased by 25% in patients with FIGO stage I-II using
assays targeting both the sense and antisense strand of the DNA compared to targeting one DNA
strand only.Paper III: In 126 patients with recurrent OC and detectable vs undetectable meth-HOXA9 at
baseline the median overall survival was 8.9 vs 17.9 months. In patients with increased vs
undetectable meth-HOXA9 after one treatment cycle the median overall survival was 5.3 vs 33
Conclusion: This thesis demonstrates that meth-HOXA9 is highly cancer specific and might serve as
a general marker of ovarian malignancy since the majority of patients with OC have detectable
meth-HOXA9 in tissue and in plasma. Consecutive analyses of meth-HOXA9 during chemotherapy
are clinically feasible, significantly related to poor survival, and have potential in treatment
monitoring with the perspective of guiding clinical decision making.