BALTIMORE—Some clinical trials of HER2-targeted therapies have found that patients assessed to have HER2-negative breast cancer nonetheless derive benefit. Is this a real benefit stemming from unrecognized biological factors in the adjuvant setting? Or is it an artifact of assay methodology producing false-negative HER2 results? Soonmyung Paik, MD, of the National Surgical Adjuvant Breast and Bowel Project (NSABP) and Michael Press, MD, PhD, of the Norris Comprehensive Cancer Center debated this topic at the 2008 Era of Hope meeting.
Contrasting with current dogma
Speaking in support of the use of HER2-targeted therapies in at least some cases of HER2-negative breast cancer, Dr. Paik discussed the NSABP B31 adjuvant trial, which tested the addition of trastuzumab (Herceptin) to chemotherapy among patients with HER2-positive disease. The latter was defined as 3+ staining by immunohistochemistry (IHC) or a ratio exceeding 2.0 by fluorescence in situ hybridization (FISH), as determined by any lab in the US or Canada. Central retesting showed that 10% of cases were negative for HER2 by both assays.
“Surprisingly, when we looked at efficacy or the degree of benefit from trastuzumab, we found that both HER2-negative and HER2-positive patients had the same degree of benefit,” Dr. Paik commented. Moreover, the parallel trial, NCCTG N9831, had similar findings.
A plot of the observed risk reduction in both trials as a function of FISH ratio yielded a flat line, he continued, meaning that “everybody got benefit, regardless of how much HER2 DNA or protein they had.” This result contrasts with the current dogma regarding response to HER2-targeted therapies, which is drawn from the metastatic setting and proposes an absolute threshold for response at a FISH ratio of 2.0. “If you follow that logic, patients with tumors that have a ratio of 1.99 should not respond, while [those with tumors with a ratio of] 2.01 will respond,” he commented, saying that does not intuitively make sense.
One proposed explanation for the B31 findings, according to Dr. Paik, is that the negative results obtained in the central lab were incorrect. However, these results were determined by three board-certified breast pathologists, and the assays have stood up to multiple tests scrutinizing their validity, he contended. For example, a microarray analysis of 753 HER2-positive and -negative cases from the B31 trial confirmed that there was no HER2 amplification in the negative tumors, and the specificity of the results ruled out the possibility that the results are due to degradation of mRNA in the paraffin blocks.
Furthermore, he noted, the false-positive rate for the central lab was lower than that reported by other trials, and the assay methodology used was rigorous enough that it served as the basis for the current ASCO College of American Pathologists (CAP) guidelines for HER2 testing.
Another possible explanation that the investigators considered for the response to trastuzumab in HER2-negative patients was tumor heterogeneity, with the possibility that local and central labs sampled tissue blocks from different parts of the tumor with different HER2 status. But in 67 HER2-negative cases in which both the primary tumor and a lymph node were tested by FISH, the node was positive in only two cases (3%). “That means that there is no possibility that our data can be explained by heterogeneity within the same tumor,” Dr. Paik asserted.
One might note the overlap in HER2 mRNA levels in positive and negative cases in B31 and argue that the results were due to degradation of mRNA as the result of formalin fixation and paraffin embedding, he observed. However, he pointed out, a compilation of 10 microarray studies from Oncomine looking at HER2 mRNA levels in fresh or frozen tumors showed that there is almost always overlap between amplified and nonamplified cases.
Preclinical data suggest that the correlations between level of HER2 amplification and response to trastuzumab (Emlet DR et al: Mol Cancer Ther 6:2664-2674, 2007) and lapatinib (Konecny GE et al: Cancer Res 66:1630-1639, 2006) are in fact linear, he observed. This finding, in turn, suggests that the current generation of IHC and FISH assays yielding a sharp cutoff is nonlinear, which could also be an explanatory factor.
In fact, Dr. Paik noted, HER2 expression by breast cancer is a continuous variable, with the distribution overlapping the currently used cutoffs for IHC and FISH, as shown by an RT-PCR analysis of 20,000 cases. Nevertheless, additional analyses of B31 tumors showed that HER2 content did not predict benefit from trastuzumab, regardless of whether the level was measured by a DNA, mRNA, or protein assay.