Genetic Heterogeneity in HER2 Testing Revisited: How Nonuniform Are “Genetically Heterogeneous” Tumors? A Study of 1550 HER2 FISH Cases.
Janet I Malowany, Julita Mazurkiewicz, Martha Wood, Martin C Chang. Mount Sinai Hospital, Toronto, ON, Canada; Univ. of Toronto, ON, Canada
Background: Testing for HER2 in invasive breast carcinomas is both prognostic and predictive. The 2007 ASCO/CAP guidelines interpret HER2/CEP17 ratios as positive (>2.2), negative (<1.8) and equivocal (1.8-2.2). A subsequent guideline (2009) defined “genetically heterogeneous” (GH) tumors as those with 5-50% (non-clustered) of nuclei with a ratio of >2.2. This study characterizes cases tested for HER2 with GH data, and the statistical distribution of amplified nuclei within a tumor.
Design: We identified 1550 cases (2009-2010) in which FISH for HER2 amplification was performed at our center on invasive breast carcinomas. The FISH protocol was standardized, using Vysis HER2 and CEP17 probes, with manual counts by the same technologists (either JM, MW; or both) according to ASCO/CAP Guidelines (min. 20 nuclei/case, up to 120). Clusters of amplified cells were scored separately. Data on every cell nucleus counted were retrieved and analyzed by linear regression of the Q-Q plot against a normal distribution. This provides an estimate of normality of the tumor population.
Results: Of 1550 FISH tests for HER2, 1097 (71%) were nonamplified, 265 (17%) were amplified, and 188 (12%) were equivocal. A total of 450 (29%) had GH. Among these GH cases, the ratio was nonamplified in 283 (63%), amplified in 17 (4%), and equivocal in 150 (33%). The amplified subpopulation in GH tumors was larger if the overall ratio was close to 2.2 (Figure 1). However, the number of nuclei >2.2 in a GH tumor did not correlate with deviation from a normal distribution (Figure 2).
Conclusions: “Genetic heterogeneity”, as defined by ASCO/CAP (2009), does not distinguish between true intratumoral nonuniformity and HER2 amplification within a normal distribution. Studies using statistical approaches are needed to determine the prevalence of true intratumoral subpopulations, and their clinical significance.
Monday, February 28, 2011 8:30 AM
Platform Session: Section C, Monday Morning