SAN ANTONIOExpression of a set of 21 genes can be analyzed
from ordinary, paraffin-embedded tissue specimens and can predict which newly
diagnosed, node-negative, tamoxifen-treated breast cancer patients are likely
to have distant disease recurrence within 10 years, according to data reported
at the 26th Annual San Antonio Breast Cancer Symposium (abstract 16). The assay
is less useful in node-negative patients who have not been treated with
Soonmyung Paik, MD, director of pathology, National Surgical
Adjuvant Breast and Bowel Project (NSABP), said that the Oncotype DX
genomic assay developed by Genomic Health, Inc. (Redwood City, California)
provides critical information for a substantial group of patients and
represents a major technological advance because it can be done using routinely
available diagnostic biopsy tissue, unlike other genome-based assays that
require special handling such as snap freezing in liquid nitrogen.
Steven Shak, MD, chief medical officer at Genomic Health,
told ONI: "We need better prognostic markers to stratify patients into
low- vs high-risk groups to decide who gets more aggressive therapy. There is
only 43% concordance among pathologists using conventional risk factors such as
tumor grade, patient age, and tumor size."
Two previous studies showed the feasibility of using this
new multigene reverse transcriptase polymerase chain reaction (RT-PCR) assay on
standard, paraffin-embedded tissue to identify gene expression profiles that
correlated with the likelihood of recurrence. "To confirm these preliminary
results, we performed a validation study using blocks from an NSABP trial," Dr.
The assay produces a "recurrence score" of 0 to 100. Dr.
Paik found that 10-year breast cancer recurrence rates were 6.8% in patients
with recurrence scores below 18 (low risk), 14.3% for scores 18 to less than 31
(intermediate risk), and 30.5% for scores of 31 or higher (high risk).
Statistical significance was P < .0001 for low-risk vs high-risk
patients. In this population, 51% of patients were low risk, 22% were
intermediate risk, and 22% were high risk.
The NSABP assay validation was a blinded, prospective study
using stored tissue samples from 668 patients who were node-negative,
ER-positive, and treated with tamoxifen. Patients had enrolled in the NSABP
B-14 trial from 1982 to 1988 and been tracked since that time, with a median
follow-up of 14 years.
The NSABP database included 2,617 tamoxifen-treated
patients. Patients were eligible for the validation study if tumor blocks were
available and contained at least 5% invasive cancer (n = 668). RNA was
extracted from three 10-µ sections for each patient.
Dr. Shak told ONI that the Oncotype DX assay
includes 5 reference genes and 16 cancer-related genes (see Table 1). "This
study was the first validation of a multigene panel in oncology," he said.
The final assortment of genes included in the RT-PCR assay
was selected from 185 cancer-related genes identified in preliminary studies.
These affected recurrence-free survival in various ways. For example, shorter
recurrence-free survival was associated with higher expression of the HER-2
adaptor Grb7 (P < .0001) and of the proliferation-related gene survivin
(P = .0001). Longer recurrence-free survival was associated with higher
expression of progesterone receptor (P = .0002).
The primary study endpoint was distant recurrence-free
survival, which was significantly higher in the low-risk patient group. In
multivariate analysis, the recurrence score remained a significant predictor of
the likelihood of relapse, but neither age nor tumor size was significant, when
compared with the recurrence score (Table 2).
"The genomic analysis generated a continuous predictive
model (see Figure). When the recurrence score is more than 30, the chance of
recurrence within 10 years is 25% ± 5%," Dr. Paik said.
During the discussion period, Dr. Paik was asked whether the
assay had been validated in a similar way in the placebo-treated patients in
the NSABP study. He said that those studies are ongoing.
Meeting attendees seemed generally impressed with the
report, but one pointed out that using this technology would still result in
misclassification of 1 of every 10 patients in the low-risk group.
Related work was reported by Francisco J. Esteva, MD, and
colleagues from the University of Texas M.D. Anderson Cancer Center (abstract
17). They tested the Oncotype DX assay using tissue from 149 patients
with node-negative breast cancer who had surgery at that institution between
1978 and 1995, both with and without tamoxifen treatment. Dr. Esteva found no
significant relationship of recurrence score with 10-year distant
recurrence-free survival (P = .16). The M.D. Anderson study did confirm
the assay’s reliability for capturing known markers in paraffin-embedded
"Recurrence score was not predictive of recurrence in
node-negative patients who did not receive adjuvant tamoxifen," Dr. Esteva
said. Since the Oncotype DX analysis uses an algorithm based on patients
treated with tamoxifen, Dr. Esteva said that it may be good at predicting
response to tamoxifen but not at more general prediction.
"Unexpectedly, higher nuclear grade correlated with lower
risk of distant recurrence-free survival (P = .02)," he said, adding
that this might have confounded the attempt to stratify patients by recurrence
During the discussion, an attendee asked how generalizable
these conclusions are, since the population of node-negative breast cancer
patients who do not get tamoxifen is very small. "Our results are not
generalizable," Dr. Esteva said. "This was a very selected population. The
results of the NSABP study presented earlier are more generalizable."
According to Genomic Health, the Oncotype DX breast
cancer assay is now commercially available as a clinical laboratory service.