Multigene Assay Predicts Breast Ca Recurrence

SAN ANTONIO—Expression 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 tamoxifen.

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. Shak said.

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 tissue.

"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 score.

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.