Hopeful
04-01-2010, 02:51 PM
J Clin Oncol. 2010 Mar 15; Epub ahead of print, J Staaf, M Ringnér, J Vallon-Christersson, G Jönsson, PO Bendahl, K Holm, A Arason, H Gunnarsson, C Hegardt, BA Agnarsson, L Luts, D Grabau, M Fernö, PO Malmström, OT Johannsson, N Loman, RB Barkardottir, A Borg
Supplementary editorial provided by OncologySTAT
TAKE-HOME MESSAGE
Investigators identified 3 distinct HER2 molecular subtypes, one of which was independently prognostic for poor outcomes.
STUDY IN CONTEXT
HER2-positive breast cancers generally confer a poor prognosis, having variable response to chemotherapy. Certain HER2-positive tumors start out, or later become, resistant to targeted therapy with trastuzumab. Thus, improved stratification of HER2-positive tumor subtypes is needed. In this study by Staaf et al, gene expression analysis was used to identify 3 subtypes of HER2-positive disease that were associated with distinct clinical outcomes. One subtype, found to be associated with significantly worse clinical outcome, was used to construct a predictive gene signature (HER2-derived prognostic predictor [HDPP]). Validation studies showed that HDPP was strongly and independently prognostic for poor outcomes, regardless of other factors such as basal-like histology, estrogen-receptor (ER)−negative status, positive lymph node status, and high-grade disease.
Hierarchical clustering analysis of expression data from 10,377 genes (Lund HER2 set) was used to identify the 3 HER2 subgroups (clusters). Tumors associated with cluster 2 had worse clinical outcomes than did those associated with clusters 1 and 3. Cluster 2 tumors tended to be ER negative and to overexpress steroid response genes (steroid response–positive phenotype). Tumors in cluster 3 were associated with better overall survival (OS) and distant metastasis–free survival (DMFS) than were those associated with the other 2 clusters. Cluster 3 tumors tended to be smaller in size and associated with fewer involved lymph nodes, but they were also highly proliferative, high-grade tumors with active PI3K signaling. Cluster 1 tumors had similar outcomes to those in cluster 3, but they were less proliferative, of lower grade, had less active PI3K signaling, and displayed an extracellular matrix–positive phenotype.
Based on the gene expression clustering, a 158-gene prognostic predictor (HDPP) was constructed. Tumors could now be separated into 2 groups with distinct prognoses, one with poor and one with good OS and DMFS outcomes (log-rank P < .00006 and P < .00009, respectively). In multivariate analyses, HDPP was independently prognostic for both the entire 10-year (P = .02) and the 5-year (P = .05) follow-up periods.
Validation studies of HDPP were carried out using several prognostic breast cancer expression datasets. Using the Nederlands Kanker Instituut (NKI) and Erasmus Medical Center (EMC) datasets, HDPP identified a subgroup of patients with good OS and/or DMFS outcomes, independent of ER status. In the NKI dataset, HDPP demonstrated superior capability for prognostic stratification of HER2 tumors when compared with the 70-gene signature (MammaPrint) and the 21-gene recurrence scores (Oncotype DX).
In a separate experiment using Affymetrix data (n = 1881), HDPP was also independently prognostic and identified a subgroup of 203 ER-negative, HER2-positive tumors with good DMFS. HDPP was prognostic, regardless of ER and node status, histologic grade, tumor size, and patient age, for both the entire (P = .0002) and the 5-year (P = .0004) follow-up periods.
Analysis evaluating the power of HDPP to predict response to treatment showed a significant association for patients of any HER2 status receiving adjuvant tamoxifen. In HER2-positive patients who received neoadjuvant chemotherapy, the majority of patients with residual disease were classified as HDPP poor, while those achieving a pathologic complete response were classified as HDPP good.
In this study, Staaf et al constructed a HER2-derived gene signature that was independently prognostic for tumors, conferring a good vs a poor prognosis in HER2-positve breast cancer. HDPP signature may represent a tool for improved outcome prediction in HER2-positive breast cancer.
Hopeful
Supplementary editorial provided by OncologySTAT
TAKE-HOME MESSAGE
Investigators identified 3 distinct HER2 molecular subtypes, one of which was independently prognostic for poor outcomes.
STUDY IN CONTEXT
HER2-positive breast cancers generally confer a poor prognosis, having variable response to chemotherapy. Certain HER2-positive tumors start out, or later become, resistant to targeted therapy with trastuzumab. Thus, improved stratification of HER2-positive tumor subtypes is needed. In this study by Staaf et al, gene expression analysis was used to identify 3 subtypes of HER2-positive disease that were associated with distinct clinical outcomes. One subtype, found to be associated with significantly worse clinical outcome, was used to construct a predictive gene signature (HER2-derived prognostic predictor [HDPP]). Validation studies showed that HDPP was strongly and independently prognostic for poor outcomes, regardless of other factors such as basal-like histology, estrogen-receptor (ER)−negative status, positive lymph node status, and high-grade disease.
Hierarchical clustering analysis of expression data from 10,377 genes (Lund HER2 set) was used to identify the 3 HER2 subgroups (clusters). Tumors associated with cluster 2 had worse clinical outcomes than did those associated with clusters 1 and 3. Cluster 2 tumors tended to be ER negative and to overexpress steroid response genes (steroid response–positive phenotype). Tumors in cluster 3 were associated with better overall survival (OS) and distant metastasis–free survival (DMFS) than were those associated with the other 2 clusters. Cluster 3 tumors tended to be smaller in size and associated with fewer involved lymph nodes, but they were also highly proliferative, high-grade tumors with active PI3K signaling. Cluster 1 tumors had similar outcomes to those in cluster 3, but they were less proliferative, of lower grade, had less active PI3K signaling, and displayed an extracellular matrix–positive phenotype.
Based on the gene expression clustering, a 158-gene prognostic predictor (HDPP) was constructed. Tumors could now be separated into 2 groups with distinct prognoses, one with poor and one with good OS and DMFS outcomes (log-rank P < .00006 and P < .00009, respectively). In multivariate analyses, HDPP was independently prognostic for both the entire 10-year (P = .02) and the 5-year (P = .05) follow-up periods.
Validation studies of HDPP were carried out using several prognostic breast cancer expression datasets. Using the Nederlands Kanker Instituut (NKI) and Erasmus Medical Center (EMC) datasets, HDPP identified a subgroup of patients with good OS and/or DMFS outcomes, independent of ER status. In the NKI dataset, HDPP demonstrated superior capability for prognostic stratification of HER2 tumors when compared with the 70-gene signature (MammaPrint) and the 21-gene recurrence scores (Oncotype DX).
In a separate experiment using Affymetrix data (n = 1881), HDPP was also independently prognostic and identified a subgroup of 203 ER-negative, HER2-positive tumors with good DMFS. HDPP was prognostic, regardless of ER and node status, histologic grade, tumor size, and patient age, for both the entire (P = .0002) and the 5-year (P = .0004) follow-up periods.
Analysis evaluating the power of HDPP to predict response to treatment showed a significant association for patients of any HER2 status receiving adjuvant tamoxifen. In HER2-positive patients who received neoadjuvant chemotherapy, the majority of patients with residual disease were classified as HDPP poor, while those achieving a pathologic complete response were classified as HDPP good.
In this study, Staaf et al constructed a HER2-derived gene signature that was independently prognostic for tumors, conferring a good vs a poor prognosis in HER2-positve breast cancer. HDPP signature may represent a tool for improved outcome prediction in HER2-positive breast cancer.
Hopeful