The group, led by Paul Spellman, PhD, from the US Department of Energy’s Lawrence Berkeley National Laboratory in California and now a guest researcher at Oregon Health & Science University in Portland, report a new overall picture of ovarian cancer genomics, including:
96% of tumors sequenced had mutations in TP53.
Nine genes were mutated at much lower frequency (2-6%) but found repeatedly, including BRCA1, BRCA2, NF1, RB1, and CDK12
Ovarian cancer subtypes can be split into four subtypes based on gene expression patterns (Interestingly none of these correlated with better or worse survival.).
Pathway analysis demonstrated that homologous recombination DNA repair – which depends on BRCA1/2 gene function – was compromised in approximately half of the tumors tested, suggesting that these tumors may be vulnerable to PARP inhibition
Ovarian cancer has a much higher frequency of genomic rearrangements, both small and large, than has been seen in other cancer types.
The team also discovered a 193-gene expression signature that predicted survival in an independent test set within TCGA’s samples, as well as in three independent, previously published expression data sets.
Remarkably, Dr. Spellman and colleagues also found that epigenetic silencing of BRCA1/2 genes, found in approximately 11% of the samples, is biologically different than mutational silencing of the genes, found in approximately 20% of the samples.
As reported recently by others, the TCGA team found that patients whose tumors carried a BRCA1/2 mutation had better overall survival than those whose tumors were BRCA1/2 wild type. Surprisingly, the tumors that had epigenetically silenced BRCA behaved like the wild type tumors, not like the mutant BRCA1/2 tumors, and were associated with poorer survival.
Finally, the team noted that thats there are a number of genes affected by either mutation or amplification that can be targeted with existing therapies or drugs in development. Even though each of those changes are rare individually – occurring in just 1-2% of patients – together they comprise a substantial proportion of ovarian cancer patients, perhaps as much as 20%.
The authors therefore suggest that new approaches to ovarian cancer treatment may be close for the patients whose tumors are driven by those rare events.
Functional Analysis at the Genome-wide Level
In the second paper, published this week in the Proceedings of the National Academy, a group of Boston researchers described an enormous effort to understand the functional necessity of more than 10,000 genes for cancer cell survival.
The investigators, led by William Hahn, MD, PhD, of Dana-Farber Cancer Institute, designed a loss-of-function study in which they systematically silenced 11,194 genes in 102 human cancer cell lines using short-hairpin RNAs (shRNAs).
The effort, which they’ve named Project Achilles, uncovered 54 genes essential for survival of one kind of tumor cell or another.
In some cases, the genes were essential only in particular genetic backgrounds. For example, mTOR protein function was required ibky in cells that carry a mutation in the PIK3CA gene. Cells that were PIK3CA-normal could lose mTOR function without a substantial decrease in cell survival.
Similarly, the investigators found that some genes had an effect on only one or a few cancer types, suggesting that those genes were related to the cell’s lineage and differentiation.
Ovarian Cancer Driven by PAX8
But the real meat of the paper comes in the team’s analysis of genes essential for ovarian cancer. In that sub-study, they re-examined 582 genes tagged as potentially interesting in their first pass.
Depending on how they re-analyzed those data, they came up with between 5 and 50 interesting genes: 22 genes appeared essential by three statistical analyses of the shRNA data; 50 appeared in at least one of those statistical analyses and are amplified in ovarian cancer tumors, and; 5 appeared in at least one of the shRNA statistical analyses and are differentially expressed in ovarian cancer relative to normal ovarian tissue.
The only gene that appeared in all three groups is PAX8. Prior research has shown that PAX8, which is a transcription factor involved in organogenesis of the Müllerian system, the thyroid, and the kidney, is overexpressed in ovarian cancer.
Dr. Hahn and colleagues found that as well. Twenty-one of 25 ovarian cancer cell lines overexpressed PAX8 in their experiments, and those cell lines were particularly sensitive to loss of PAX8 function.
In one confirmatory experiment, the team introduced previously untested shRNAs against PAX8 into eight ovarian cancer cell lines. Of those, six had a 50% or greater drop in cell survival after PAX8 silencing. Similarly, they saw a drop in cell viability in kidney and endometrial cancer cell lines that overexpress PAX8.
Finally, when the team re-examined the 489 patient samples collected and analyzed by TCGA, they found that PAX8 was amplified in 16% of the ovarian cancer tumors.
Altogether the data suggest that inhibition of PAX8 activity could have a substantial impact on ovarian cancer treatment. The caveat to that, of course, is that PAX8 is a transcription factor and those have been notoriously difficult to inhibit. There has been some recent progress on that front, though.
The best part of the study, in my view, though, is still the basic fact that the team has taken the leap from cataloging to testing. And it’s been no small feat. They generated more than 22 million individual measurements of the impact of shRNAs on survival. A jaw-dropping number.