NEW ORLEANS – Based on data from an in-depth molecular analysis, researchers are questioning a longstanding assumption that all primary precancerous breast lesions are the root cause of tumors that recur in the same breast.
The findings, presented at the American Association for Cancer Research on Sunday by Tanjina Kader, a postdoctoral researcher at the Peter MacCallum Cancer Center in Melbourne, Australia, suggest that a subset of patients with ductal carcinoma in situ (DCIS) develop breast cancers that are actually new primary tumors, genomically unrelated to the earlier precancerous lesion. Kader and colleagues discovered this by comparing the genomic profiles of DCIS cases that recurred either as DCIS or invasive breast cancer, and by also comparing the genomic features of DCIS cases that recurred and didn't recur.
Their analysis showed that around 18 percent of ipsilateral breast cancers — tumors that occur in the same breast after surgical removal of DCIS — had unique genomic profiles compared to the precancerous lesion. "These tumors are new primaries … and suggest a high-risk environment or a genetic predisposition [for breast cancer] for these patients," Kader said. "These patients may be highly susceptible to developing new tumors."
Kader said that this type of phylogenetic analysis should be conducted more readily to better distinguish which patients' DCIS have advanced to clonal breast cancer and which patients have a non-clonal new primary tumor in the same breast.
"We are proposing that in the future, instead of treating all the recurrent patients the same way, we have to find out first whether there is a group of patients that fits into [this] non-clonal" type for whom "surgical resection may not be sufficient," Kader said, adding that these patients should receive genetic counseling to assess their inherited cancer risk or undergo more aggressive prevention strategies, such as bilateral mastectomy or endocrine therapy.
A diagnosis of DCIS means that the patient has abnormal cells in the milk ducts that haven't spread to surrounding tissue. These abnormal cells are considered a precursor to invasive cancer, though 75 percent of DCIS cases don't recur. But 25 percent of patients with DCIS do experience recurrence in the same breast; half of the time this is another DCIS and the other half it is invasive breast cancer.
To avoid DCIS becoming invasive breast cancer, patients typically receive breast-conserving surgery (lumpectomy) with or without radiation therapy or endocrine therapy. Or, if their DCIS covers a large area or if a biopsy shows abnormal cells at the edge of the tissue specimen, patients can have a mastectomy.
Doctors are able to catch DCIS more readily now with the help of mammography. But since most DCIS cases won't recur, some oncologists believe patients with these precancerous lesions are overtreated. Testing for certain chromosomal and genetic alterations shared by the primary DCIS and recurrent tumors present an enticing approach for predicting which patients are at low risk for invasive breast cancer recurrence and can have de-escalated treatment. However, Kader said this study suggests that these biomarkers wouldn't identify the subset of patients who are at risk for developing new primary tumors in the same breast.
Testing an assumption
To conduct their analysis, Kader and colleagues micro-dissected and extracted DNA from 67 pairs of samples, with each pair including a sample from a primary DCIS and another from a recurrent breast tumor. Half the samples from recurrent tumors were invasive breast cancer. While the researchers did not have any matched normal samples, they did have a set of 32 samples from non-recurrent DCIS cases that were treated with wide local excision and had a minimum of seven years of follow-up data.
They analyzed the chromosomal changes and genetic mutations in these samples using a targeted sequencing panel, whole-exome sequencing, and low-coverage whole-genome sequencing. Researchers performed statistical and phylogenetic analysis to compare the differences in genomic profiles between DCIS cases that recurred and those that didn't. They tracked how DCIS evolved to recur in the same breast, and ultimately, tested the assumption that all recurrent tumors are descended from a primary DCIS.
Kader compared this type of clonality analysis to a tree. "The trunk of the tree [represents] clonal events or shared DNA events," she said. "And the branches of the tree represent a tumor-only event."
Using whole-exome sequencing, the researchers distinguished between clonal tumors that shared DNA mutations or chromosomal abnormalities with the primary DCIS and non-clonal tumors that didn't. Overall, Kader reported that 82 percent of the recurrent cases were clonal and 18 percent were independent, new primaries.
She highlighted that clonal recurrences tended to have changes in chromosome 5, 11, 17, and 20, and TP53 mutations that they shared with the primary DCIS. These biomarkers didn't show up in de novo tumors, however. In fact, these de novo tumors had similar genomic profiles to the non-recurrent DCIS cases.
"This data suggests that these four chromosomal changes and TP53 mutation status could be a biomarker to predict recurrences of primary DCIS," she said. "But even though we get excited about predictive biomarkers, this particular finding [about the subset of non-clonal tumors] raises the question whether we should actually use [such] predictive biomarkers for DCIS recurrence."
She further estimated that using these biomarkers to make decisions about treatment might end up undertreating around 15 percent of patients who are at risk for developing new primary tumors. These non-clonal patients, according to Kader, if they can be identified using the type of phylogenetic analysis conducted by her group, may be eligible for mastectomy even if their tumors are small, and should consider genetic testing to assess if they have inherited a genetic alteration that increases their breast cancer risks.
Kader acknowledged that phylogenetic analysis requires whole-exome sequencing, which isn't readily performed in the clinic presently. "It has to be high-depth [whole-exome sequencing]," Kader said. "If you have lower-depth genome sequencing, then we won't be able to do this phylogenetic analysis. But the analysis pipeline is pretty simple to follow."
Another limitation of this analysis was that the researchers used only chromosomal changes when conducting the phylogenetic analysis. While these are the main genetic changes in breast cancer, Kader and colleagues didn't factor in all the tumor-specific mutations due to limited genetic information on normal cells.
A new era in pre-cancer research
Kader's research is significant in that it is the first time this subset of non-clonal ipsilateral breast tumors has been identified. "This is actually our first finding that non-clonal tumors actually exist," she said, adding that further research is needed to understand the biology of these tumors.
She noted that recently published data from the Breast PreCancer Atlas, which is being conducted within the larger National Cancer Institute-funded Human Tumor Atlas Network, suggests the important role the tumor microenvironment and the immune system play in DCIS's transformation into invasive breast cancer.
During a plenary session on Friday, Avrum Spira, global head of Johnson and Johnson's Lung Cancer Initiative and a professor of medicine, bioinformatics, and pathology at Boston University Medical Center, also highlighted that research out of the NCI's pre-cancer atlas has shown that the immune microenvironment of DCIS may be a predictor of relapse after surgery.
Historically, precancerous lesions have been hard to study. Unlike metastatic tumors, samples from pre-cancer lesions are more challenging to procure in certain tumor types like lung cancer. Moreover, samples from these lesions aren't banked at research hospitals.
And when there are samples, they are paraffin-embedded and small, making molecular profiling difficult, Spira said. The lack of access to samples has hindered the types of in-depth analysis needed to determine which precancerous lesions will progress to invasive cancer and identify biological targets that would stop the pre-cancer from becoming full-blown cancer.
But over the last few years, five pre-cancer atlas consortia have launched in the US with NCI support and in other countries. Spira estimated that there are now 20 medical centers in five countries that are profiling precancerous lesions of at least nine different tumor types. "What's remarkable is that more than 2,000 patients have been profiled across these five consortia, and they're using the latest and greatest molecular techniques to profile very small, often paraffin-embedded tissues," he said. "That really speaks to how far molecular biology has come, and the informatics has come along, as well."