It's estimated that half of all patients prescribed trastuzumab — a monoclonal antibody-based drug for the treatment of HER2-positive breast cancer — do not respond to it. Instead, they are exposed to a whole host of potential side effects, including cardiomyopathy and pulmonary toxicity, while experiencing no apparent benefit.
There are several US Food and Drug Administration-approved companion diagnostic assays that interrogate a patient's HER2 status — via immunohistochemistry or fluorescence in situ hybridization — to determine whether they are eligible for treatment with trastuzumab, which is marketed as Herceptin. Patients who test HER2-positive with IHC, FISH, or both qualify for trastuzumab therapy, though that won't necessarily mean they will respond to the drug. Compounding the issue, says Yale University School of Medicine's David Rimm, is the fact that neither assay is sensitive or specific enough to distinguish responders from non-responders. In addition, there is a lack of protocol standardization among pathology labs, despite guidelines from the American Society of Clinical Oncology and other groups.
"The assays that are used generate a fair bit of uncertainty. There's concern about the quality of the assays. There's concern about pre-analytical variables," Rimm says. "Different labs give different results and [that's] ... why you get the New York Times articles that we've seen over the last year about whether a patient was really HER2-positive or not and whether the FISH or IHC worked."
To better predict which patients are likely to respond to trastuzumab, Rimm says that a quantitative, multiplexed companion diagnostic will be needed. His group is being funded by the National Cancer Institute to develop just that.
"The underlying hypothesis of our grant was that if we looked at different molecules down the pathway associated with HER2 signaling, we would be able to more specifically find patients who are able to respond to the drug," Rimm says. He and his team are interrogating DNA and RNA copy number as well as gene and protein expression levels within the PI3-kinase pathway — a downstream target of HER2 — in patients previously characterized by their response to trastuzumab in clinical trials.
In 2007, Rimm's group presented data demonstrating the accuracy of a multiplexed, automated quantitative analysis — or AQUA — assay of estrogen, progesterone, and four HER2 family proteins for predicting trastuzumab response in a training set at the American Association for Cancer Research annual meeting. They also described a multivariate logistical model for predicting trastuzumab response during metastasis. Rimm and his colleagues found that the assay "was superior to any one protein, traditional IHC, or FISH for prediction of trastuzumab response." Overall, Rimm says, the results showed that "you can build a model that's more sensitive and more specific for predicting response to Herceptin by using five or six proteins than you can with just one."
While the team is still in the model-building stage, Rimm says that any diagnostic test that results from his team's work would build on its success with AQUA and would likely incorporate a DNA copy-number analysis component. It would also have to be validated in a large cohort.
Discovery-based research for companion diagnostic development is often neglected by industry — especially outside of the co-development setting, Rimm says, because most firms "tend to shy away from things that aren't near-term to revenue generation." But, once "we get to the point where it looks like it's working in validation, I think that there will be industry uptake," he adds.
Things are also moving along in Rimm's push to distinguish trastuzumab responders among HER-2 positive patients. "I think in the future — especially as more HER2-targeted drugs come to market — that will start to be important," he says.