NEW YORK – Researchers found that local and central labs provided discordant HER2 expression results for nearly one-third of patients with solid tumors in a clinical trial for a HER2-targeted drug, suggesting that trial sponsors must carefully weigh the benefits and drawbacks of confirmatory biomarker testing when determining eligibility in such precision medicine trials.
Investigators conducted the HER2 expression testing within a Phase I study testing the activity of Jazz Pharmaceuticals and Zymeworks' HER2-targeted bispecific antibody zanidatamab in patients with locally advanced and metastatic solid tumors. Although inter-lab variability has been a long-recognized problem with HER2 testing conducted to determine which breast and gastric cancer patients should receive HER2-targeted treatment, newer therapies are being evaluated across a wider variety of tumor types, often in a tissue-agnostic fashion. In the latest study, published in Molecular Cancer Therapeutics earlier this month, researchers considered the impact HER2 testing discordance can have amid broadening utility of HER2-targeted drugs and evolving HER2 testing parameters.
Funda Meric-Bernstam, chair of the department of investigational cancer therapeutics at MD Anderson Cancer Center and senior author of the study, noted that pharma companies take different approaches when identifying patients for trials of HER2-targeted treatments. Some rely only on local labs to do the testing, while others require local lab results be confirmed via centralized testing, and still others funnel all testing to one central lab. Some studies may involve assessment of HER2 expression on archival tissue, while others may involve patients undergoing new biopsies for testing.
Each has its advantages and disadvantages. "The disadvantage of the pretreatment biopsy approach is the cost involved with biopsy and potential delays in initiating treatment related to biopsy," Meric-Bernstam explained. "The disadvantage of relying on prior local testing is that it may have variability in the assay used. If we go with central confirmation of previous local testing, that has a disadvantage of delays in initiating a trial due to sample acquisition as well as the central testing. It is really unclear what the best approach to this is."
To partake in the zanidatamab trial, patients had to have HER2 overexpressing tumors. They first had to undergo local testing for HER2 expression or amplification or ERBB2 amplification — using immunohistochemistry, fluorescence in situ hybridization, or next-generation sequencing — followed by confirmatory testing via IHC or FISH at a central laboratory on new or archival tissue. The researchers wrote in their paper that a wide range of local lab tests were used to gauge HER2 expression and identify patients for the trial, but that "this lack of standardization reflects a real-world clinical scenario."
Fifty-two out of 70 patients in the trial had their HER2-expression status centrally confirmed on a new biopsy sample. In this group, 30.8 percent of patients had discordant results between local and central testing, including 41 percent of those with gastric or gastroesophageal junction tumors, 22 percent of breast patients, and 29 percent of colorectal cancer patients. Eight patients who had local testing and central HER2 reevaluation on the same tissue sample did not have discordant results. Patients needed to have centrally confirmed HER2 expression to enroll on the trial, and eight patients who were initially deemed HER2-positive by local testing became ineligible to enroll after central testing found no HER2 expression.
HER2 expression results can differ between labs for a variety of complex reasons, from variability in testing to changing tumor biology. Previous studies in gastric cancer have suggested that patients can become resistant to HER2-targeted treatment through loss of HER2 expression, for example. But the researchers in this study observed HER2 testing discordance in some patients who had not received a prior HER2 targeted therapy before trying to enroll in the zanidatamab study, suggesting the discordance may be due to test differences. While one-third of 30 patients who had received an anti-HER2 treatment other than zanidatamab between positive local HER2 results and central confirmation ended up HER2-negative with central testing, six of 22 patients who had a different type of treatment between the two tests turned up HER2-negative via central testing.
When Meric-Bernstam and colleagues evaluated concordance of the IHC staining intensity across local and central lab testing, they found that some patients had different HER2 expression scores from different biopsies. Among 30 HER2-positive patients with a IHC 3+ HER2 expression score via local testing, nine patients, or 30 percent, had lower HER2 IHC staining by central lab testing.
An IHC 2+ score is considered "equivocal" for HER2 overexpression and needs to be confirmed using FISH testing. IHC scores of 1+ or 0 are considered HER2 negative. In 15 patients with IHC 2+ scores on local testing in this study, 66.7 percent had discordant central lab results. In this group, central testing deemed two patients as IHC 3+, seven patients as IHC 1+, and one patient as IHC 0. One patient who was IHC 1+ on local testing was IHC 2+ by central testing.
In an evaluation of local FISH amplification versus central lab results, 14 of 23 patients with ERBB2 amplification on local FISH also demonstrated HER2 protein overexpression (IHC 3+) or equivocal expression (IHC 2+), while nine had low HER2 protein immunostaining (IHC 1+ or 0).
"We saw discordance across a variety of different tumor types, and we had discordance in both HER2 overexpressing as well as HER2 lower-expressing patients," Meric-Bernstam said, adding that even among patients tested with NGS, there was some discordance in the results.
"It highlights that whether we're looking at expression by IHC or amplification by FISH or NGS, this discordance can occur across tumor types, and it is not limited to patients that had intervening HER2 targeted therapy," she continued.
The results also highlight the challenges of assessing high or low HER2 expression to determine patient eligibility for treatments that require more nuanced biomarker interpretation. For example, the approval in the US last year of AstraZeneca and Daiichi Sankyo's Enhertu (trastuzumab deruxtecan) for HER2-low breast cancer patients, defined as those with IHC 1+ HER2 expression scores or those with IHC 2+ scores and a negative FISH test, has raised calls for greater standardization of HER2 testing between labs and pathologist test interpretations.
While zanidatamab is being developed for any level of HER2 expression, Meric-Bernstam noted that this study is part of a recent push to consider testing concordance across tumor types. Similarly, researchers from the DESTINY-PanTumor02 trial of Enhertu in solid tumors also considered HER2 testing concordance.
Meric-Bernstam noted that researchers are beginning to consider a wider range of HER2 expression in trials of HER2 targeted treatments. "Previous studies would look at HER2 amplified or overexpressing patients, or the traditional HER2-positive patients, but now we have to start looking at lower expressors as well," she said. "A lot more information is needed across tumor types. We're still scratching the surface to try to understand both how often these biomarkers change and what is the implication for this for different therapies."
In practice, repeat testing for HER2 expression has become common for breast cancer patients, specifically to confirm results prior to beginning treatment, Meric-Bernstam noted, but the practice is less common for patients with other types of HER2-expressing tumors like colorectal cancer.
"We definitely need to be able to better counsel patients as we get more of that data to know how often a patient that's HER2 non-expressing becomes HER2-expressing and how often a patient that's HER2-high-expressing becomes HER2-low expressing, because different treatments will likely have different efficacy in those scenarios," she said.