A study led by Mayo Clinic researcher Edith Perez has found that including additional HER-family proteins in the standard HER2 companion diagnostic assay does not improve prediction of the effectiveness of trastuzumab treatment in breast cancer patients.
The findings, Perez told ProteoMonitor, counter hopes held by many breast cancer researchers that conventional HER2 testing could be made more effective by shifting to a multi-protein panel.
Testing of HER2 gene or protein levels via fluorescence in situ hybridization or immunohistochemistry assays, respectively, is standard clinical practice to determine which breast cancer patients are likely to respond to treatment with trastuzumab, marketed by Roche as Herceptin. US Food and Drug Administration-approved HER2 tests include the IHC-based Dako Herceptest and Roche/Ventana Pathway test; and Abbott/Vysis's FISH-based PathVysion HER-2 DNA Probe Kit.
Only 20 percent to 25 percent of HER2-positive patients, however, respond to Herceptin treatment. When Herceptin is combined with chemotherapy, the response rate rises to over 50 percent. Hoping to improve on those numbers, Perez and her colleagues examined more than 1,400 tumor samples to investigate what effect adding the proteins HER3, HER4, EGFR, ER, and PTEN to HER2 protein expression testing might have.
Their research, which they presented last week at the 2011 San Antonio Breast Cancer Symposium, found that including the proteins HER3, EGFR, ER, and PTEN failed to improve the predictive power of HER2 testing. The researchers also investigated the usefulness of HER4, but that data, Perez said, is still being validated.
The most significant finding, she noted, was that measuring HER3 added nothing to the test’s predictive power.
“Many people thought that HER3 was going to be a very important marker because of preclinical data that suggested that dimerization of HER2 with HER3 could be a big driver of cell growth,” Perez said. Her team found, however, that “HER3 expression was not associated with any tumor characteristics. Whether a tumor had a high or low level of HER3, there was absolutely no impact on its sensitivity to trastuzumab.”
In addition to looking at new markers, the researchers also tested the use of automated quantitative analysis, or AQUA, to measure protein expression. This technique, which was developed by Yale University researcher Robert Camp and is offered commercially by biotech firm HistoRx, uses fluorescence-labeled antibodies and computerized image analysis to provide a more quantitative, reproducible measurement of protein expression than traditional IHC, which relies on the subjective judgments of individual pathologists.
The hope, Perez said, was that quantitative measurements of HER2 expression might be a better predictor of Herceptin response than traditional IHC-based guidelines, which simply define as HER2-positive any patient showing HER2 expression in more than 10 percent of their cells. No improvement in the test’s predictive power was seen using the AQUA method, however, she said.
The results of the study are “a disappointment,” Nita Maihle, a professor of Obstetrics, Gynecology, and Reproductive Sciences at Yale, told ProteoMonitor. “They looked at all the usual suspects – which is a good idea, looking at all four members of the HER family in addition to ER and PTEN – and what they found is basically that nothing correlates.”
In her own research, Maihle has been studying the possibility that the performance of EGFR testing might to some extent be hindered by EGFR isoforms not picked up by the test (PM 5/20/2011). These concerns are also applicable to HER2 testing, she said, noting that “certainly there are multiple soluble and intracellular or truncated forms of HER2 … which haven’t been looked at in the context of trastuzumab responsiveness yet.”
Although the Mayo study didn’t investigate the possible role of protein isoforms, Perez said she was familiar with Maihle’s work on the question and open to exploring it in future studies.
Perez also said she has established a collaboration with George Mason University researcher Emanuel Petricoin to investigate using protein phosphorylation as a measurement of cancer signaling that could be used to predict drug response. Petricoin has been pursuing such work for several years using the reverse phase protein array technology he developed with GMU colleague Lance Liotta (PM 11/18/2011).
Perez said, though, that she is highly skeptical of protein phosphorylation work done to date due to the questions surrounding sample collection.
“Many people have tried to do phosphorylation work, and it is a pity that people have failed to realize that phosphorylation of proteins can change in a few minutes after surgery has been performed,” she said. “It’s really hard to understand the validity of what has been published so far. We can all keep saying that phosphorylation is important, but we don’t really have any way to truly measure it.”
Perez’s comments reflect growing concern among scientists about phosphoproteomic research. The issue has recently impacted high-profile projects like the National Cancer Institute's Clinical Proteomic Technologies for Cancer initiative, which has seen significant discussion about whether samples the group had obtained from the NCI-funded Cancer Genome Atlas would be appropriate for phosphoproteomic analysis.
Perez was interested in partnering with Petricoin in part, she said, because “he has devoted his career to figuring out” how to effectively measure protein phosphorylation. Most recently, he and Liotta developed a fixative for phosphoproteomics that, he told ProteoMonitor, could resolve many of the problems associated with both flash-freezing and formalin-fixing clinical samples for phosphoproteomic analysis (PM 9/2/2011). Called TheraLin, the fixative is offered for sale through Theranostics Health, the firm Petricoin and Liotta founded to commercialize the RPPA technology.
“We will see what happens,” Perez said. “But up to now no one has been able to [measure protein phosphorylation] in a reliable way.”
Have topics you'd like to see covered in ProteoMonitor? Contact the editor at abonislawski [at] genomeweb [.] com.