Multidrug Resistance-Linked Gene Signature Predicts Overall Survival of Patients With Primary Ovarian Serous Carcinoma
Abstract
Purpose:
This study assesses the ability
of multidrug resistance (MDR)-associated gene expression patterns to
predict survival
in patients with newly diagnosed carcinoma of the
ovary. The scope of this research differs substantially from that of
previous
reports, as a very large set of genes was evaluated
whose expression has been shown to affect response to chemotherapy.
Experimental Design:
We applied a customized TaqMan
Low Density Array, a highly sensitive and specific assay, to study the
expression profiles of 380 MDR-linked genes in 80
tumor specimens collected at initial surgery to debulk primary serous
carcinoma.
The RNA expression profiles of these drug
resistance genes were correlated with clinical outcomes.
Results:
Leave-one-out cross-validation was used to
estimate the ability of MDR gene expression to predict survival.
Although
gene expression alone does not predict overall
survival (P=0.06), four covariates (age, stage, CA125 level and surgical
debulking)
do (P=0.03).
When gene expression was added to the
covariates, we found an 11-gene signature that provides a major
improvement
in overall survival prediction (log-rank statistic P
less than 0.003). The predictive power of this 11-gene signature was
confirmed by dividing high and low risk patient
groups, as defined by their clinical covariates, into four specific risk
groups
based on expression levels.
Conclusion:
This study reveals an 11-gene signature
that allows a more precise prognosis for patients with serous cancer of
the ovary treated with carboplatin- and
paclitaxel-based therapy. These 11 new targets offer opportunities for
new therapies
to improve clinical outcome in ovarian cancer.
