NEW YORK (GenomeWeb) – Researchers at Stanford University studying the development of breast cancer in individuals have found that early lesions are genetically heterogeneous and that mutations in a well-known cancer gene, PIK3CA, induce cell growth but don't always lead to cancer.
The study, published online today in Genome Medicine, proposes the mutational analysis of pre-cancerous lesions in the clinic and raises questions about the utility of PIK3CA as a diagnostic marker of cancer.
For their study, the scientists, led by Robert West and Arend Sidow at Stanford University School of Medicine as well as Serafim Batzoglou at Stanford's Department of Computer Science, built lineage trees for samples from the same patient in order to reveal their phylogenetic relationships.
Pre-invasive lesions, they found, were genetically quite heterogeneous, both internally and in relation to each other. They also discovered that mutations in the PIK3CA kinase gene, which is mutated in a substantial fraction of breast cancers, occurred independently in separate lesions, yet was sometimes absent from the tumor of the same patient, suggesting they might be "comparatively benign inducers of cellular proliferation."
PIK3CA is the target of several drugs that are currently in development, but "our finding that PIK3CA may recur multiple times at various stages of tumor or neoplastic development suggests that it is more of a moving target than one would like,” Sidow said in a statement.
While many studies have analyzed the mutational spectrum of fully-fledged tumors and metastases, only a few have looked at pre-invasive lesions, or neoplasias. Those lesions, which include hyperplasias and carcinomas in situ and are often considered precursors to full-blown cancer, are increasingly being detected by breast cancer screening programs, so it would be important to determine which ones might develop into invasive disease.
To shed new light on this, the Stanford team performed deep targeted sequencing on 66 samples from six breast cancer patients, including 18 carcinomas and 34 neoplasias. Samples included breast neoplasias, hyperplasias, ductal carcinoma in situ, invasive ductal carcinoma, and metastases.
For an earlier study, published two years ago in Genome Research, they had already analyzed 31 samples − including neoplasias, normal tissue, and carcinomas − from the same six patients by whole-genome sequencing.
For the current study, they added 38 samples, mostly pre-invasive lesions, and decided to focus on a limited number of targets — a total of 1,185 single-nucleotide variants — in order to estimate their variant allele frequencies more accurately. The SNVs they chose were a random subset from the variants discovered in the original project and the original phylogenetic trees and included mutations found in PIK3CA.
After amplifying their targets by multiplexed PCR, starting with 10 nanograms of genomic DNA from the original samples and 2 nanograms from the new samples, the researchers constructed Illumina libraries and sequenced them on a HiSeq 2000.
Based on the variant allele frequencies they determined for each SNV, they then constructed phylogenetic trees for all samples from the same patient.
Many pre-invasive neoplastic lesions were heterogeneous. For example, twenty of those that were histologically similar to normal epithelial cells had either no common ancestors or had a shared common ancestor with other samples high in the tree.
Heterogeneity also occurred within pre-invasive lesions. In five such samples, for example, the researchers detected two independent lineages with distinct SNVs.
Overall, the phylogenetic trees "cast the traditional concept of tumor progression into a new light," they wrote, noting that "pre-invasive neoplastic lesions are not the direct precursors of invasive lesions but independent clonal proliferations of an ancestral cell … just as the carcinomas are. By tracing the lineage from the root of each tree to the [invasive ductal carcinoma] ancestor, a progression of the cellular lineage from mildly to strongly proliferative and eventually invasive phenotypes can be inferred."
Sidow further told GenomeWeb that the group's study "confirms that there are some neoplastic lesions, particularly those with atypia, that bear a lot of the genomic change also present in the carcinoma. Most other lesions appear not to be burdened by high mutational load and represent proliferations independent of the carcinoma lineage."
Genomic heterogeneity is often not visible by histological analysis, so characterizing pre-invasive lesions would allow doctors to stratify lesions into normal and aberrant ones, "with clearly distinct choices for treatment and monitoring as a result," the authors wrote.
For example, if several lesions were detected by imaging, most or all of them should be biopsied and analyzed for genetic heterogeneity. For a single lesion, sub-lesions could be analyzed to reveal heterogeneity. "In either scenario, understanding the genetic and lineage heterogeneity among the samples and having the chance to identify those sites, if any, that have evolved a high mutation burden, will improve monitoring and treatment for that specific patient," they wrote.
The researchers also found that mutations in PIK3CA can arise several times independently in the same patient and do not always persist in their tumor. In their previous study, the researchers had identified two missense mutations in the PIK3CA gene in four patients, the origin of which they were now able to assess in greater detail.
Four patients carried the same missense mutation in PIK3CA, and in addition, two of the patients carried a different missense mutation. But according to the phylogenetic analysis, the same mutation occurred independently in different samples, arising at least three times in two patients, and at least twice in another patient. Also, in two of the patients, the PIK3CA mutation was only present in their pre-invasive neoplasias, but absent from their invasive tumors.
In light of these results, the researchers wrote, "it is worth considering whether [the] PIK3CA H1047R [mutation] is really a driver of neoplasia at this stage of cancer development or a comparatively benign inducer of proliferation alone."
If it turns out that the mutation is indeed a driver of benign proliferation, "then its diagnostic value is questionable," Sidow said.
Still, PIK3CA, which is part of a pathway that controls cell proliferation, may be a good drug target in patients where all tumor cells have mutations that activate PIK3CA or upstream components of the pathway, he said, adding that "we don't know until the clinical trials have been performed."