Anne Paxton
February 2025—HER2 in predictive biomarker testing, HPV-related cancer testing guidelines, and the art of applying polygenic risk scores will be some of what’s heard in USCAP 2025 sessions next month in Boston.
The meeting’s Maude Abbott Lecture, “From Revolution to Evolution: HER2 Lessons for Predictive Biomarkers,” promises to tell “an incredible, still-evolving story,” says speaker Kimberly Allison, MD, professor of pathology and director of breast pathology at Stanford Medicine.
Her lecture will explore HER2 testing as a key case study of principles underlying predictive biomarker testing in oncology and pathology.
“It’s going to be the story of how a biomarker became high impact and changed our field, how it created, for the first time, a new predictive pathology role where we had to create standards and be accountable to them so the right patients were eligible for the right treatments. But it also has taught us the challenges around setting these standards and how there are sometimes gray zones for which we need standardized strategies to resolve,” says Dr. Allison, who is the expert panel co-chair of the ASCO/CAP ER/PR and HER2 testing in breast cancer guideline updates.
“We learned over time that if we just said, ‘That’s a gray zone; it’s equivocal by both tests,’ the medical oncologists didn’t know what to do with patients. They would say, ‘I have to decide. Help me out. What’s the truth?’ That was a big part of the HER2 guideline’s updates over time—addressing equivocal results and unusual in situ hybridization results like groups two through four.”
Where it’s recently become complex in HER2 is the low end of the spectrum of HER2 scores, Dr. Allison says. This is because “today in the metastastatic breast cancer setting you can use an antibody drug conjugate targeted to the HER2 protein to successfully treat some patients even if HER2 is not overexpressed or amplified. Patients with low levels of protein expression appear to still receive benefit from this newer therapy, trastuzumab deruxtecan.”
The issue, she says, is that it is likely quite rare for a breast cancer to truly have no HER2 protein at all, but IHC as a test is likely not reliable or sensitive enough to discriminate cases with no expression from those with very low levels. The clinical trials that expanded HER2 IHC testing for this indication did not in fact include patients who had no HER2 expression by IHC, “so it is still not clear if this is a reliable and predictive new threshold to treat or not treat.”
HER2 testing for antibody drug conjugates is generating a number of questions about biomarkers in general, Dr. Allison says, and that may call for a different way of thinking about some biomarkers. “Are non-gene amplified protein expression levels truly predictive in the antibody drug conjugate setting, and at what specific level? Do we even need them for some indications?” It’s a challenge, she says, when pathologists are expected to adopt testing in the setting of more limited evidence. “An unreliable biomarker is a bad idea for us to try to attempt to standardize and reliably report. We are very good at determining HER2 overexpressed/amplified cases from non, but the very low protein expression discrimination is affected by so many variables, including tissue preservation. We need to be able to understand and express the uncertainty in some test results.”
Reporting standards and terminology are another issue. Even though HER2 “low” and “ultralow” have become common terms among treating clinicians as a shorthand for specific HER2 IHC scoring categories in nonamplified cases, “there are concerns that these terms are misleading since they don’t really refer to any new categories or breast cancer subtypes,” Dr. Allison says. Therefore, these terms have not been adopted formally in the updated guideline. “That was a big question for our last HER2 guideline update,” she says. But “if we adopt new terminology for every indication for the same biomarker, we could end up with some very complicated reporting.”
Dr. Allison is currently helping to update the CAP breast cancer biomarker reporting templates so that the detailed staining patterns of each scoring category are more clearly spelled out and the information clinicians need to decide treatment is present. “We need to be able to rapidly adjust our reporting so that when a new indication is approved, our reports can be quickly adjusted to ensure the information needed is there,” she says.
With continually evolving indications and biomarker developments, there is a need for more pathogists to stay on top of the latest. Dr. Allison hopes to “spread awareness about what matters for patient treatment” where predictive biomarkers are involved. “We need to continue to evolve our knowledge if we’re interpreting these predictive tests.”
“We’re gatekeepers, and it’s important to do our best to standardize all aspects of testing, including how we’re scoring and reporting,” she continues. “The CAP’s biomarker reporting templates can also be designed to help standardize reporting these biomarkers in ways that both educate pathologists, ensuring they are scoring correctly, and make it clear to treating oncologists the details they need to decide treatment.”
Dr. Allison sees the HER2 story as exemplifying the leadership path that pathology needs to take as biomarkers evolve. “It’s going to get even more complicated and we’ll have to continue to keep up with the pace of change, but we also need to communicate to the oncology community what our limits are. And I think CAP is interested in trying to get involved in clinical trial design earlier, with pathology in mind.”
“If you’re going to apply a biomarker for prediction of treatment benefit, you need to think about the practical, simple way it can be scored and reported by everyone,” she says. Artificial intelligence tools will help, but pathologists have the final say. “Pathologists are still the gatekeepers and we will continue to be, even if we have fancy tools that help us.”
Standardization will also be the focus of a CAP companion society session on the updated CAP guidelines for HPV-related cancers.
At the 2024 USCAP annual meeting, four guidelines were the focus: diffuse gliomas, PD-L1 and tumor mutation burden testing of patients with lung cancer, HER2 testing in breast cancer, and mismatch repair and microsatellite instability testing for immune checkpoint inhibitor therapy.
Two guideline updates were chosen for discussion in this year’s companion society session, “both around the role of HPV and cancer but in different cancer types,” says Russell Broaddus, MD, PhD, session chair and moderator and chair of the Department of Pathology and Laboratory Medicine at the University of North Carolina School of Medicine. Also planned is a brief recap of the guideline process. “This is not something that just drops from heaven. These guidelines are built. And a big thing I stress is that these are evidence-based guidelines that are not opinion. I think we know that as pathologists, but it doesn’t get stressed enough,” he says.
Developing the mismatch repair and microsatellite instability testing guideline, which Dr. Broaddus chaired, was a multidisciplinary effort. One of the medical oncologists who was part of the expert group, and serves on multiple guideline committees in many disciplines, told Dr. Broaddus how much he enjoyed the CAP evidence-based process, where “opinion is weighed against published evidence. He enjoyed that and respected it. It’s unique to what pathology does,” Dr. Broaddus says.
Two original guidelines in need of updates are those on lower anogenital squamous terminology for HPV-associated lesions (LAST, published in 2012) and HPV testing in head and neck carcinomas (published in 2017).
“If immunohistochemistry is involved as your biomarker, which it probably is with p16 IHC and HPV detection, then sometimes there can be interpretive differences,” Dr. Broaddus says. “One person’s focally positive is another person’s negative, for example.”
Christina S. Kong, MD, medical director and vice chair for clinical affairs in the Department of Pathology at Stanford University, will address in the CAP session how anogenital squamous intraepithelial lesion nomenclature, with specific attention to tiered morphological diagnosis, informs clinical management. She will also discuss how the use of biomarkers can impact interobserver variability in the diagnosis of these lesions.
For the updated lower anogenital squamous terminology guideline, which is still in progress, the plan is to reaffirm the use of the two-tiered nomenclature system of low-grade and high-grade intraepithelial squamous lesion (LSIL/HSIL). However, instead of making the -IN (-intraepithelial neoplasia) classification optional, Dr. Kong says, the plan is to make it a requirement, “because in discussions with the clinical providers on our panel, it was evident that they still use -IN classification to guide management.” In particular, there are scenarios in which an -IN2 patient may be managed less aggressively than an individual with -IN3. “These differences in management are informed by the fact that, despite attempts to better standardize the -IN2 diagnosis, this remains a biologically equivocal category that encompasses proliferations with behavioral and morphologic overlap with LSIL as well as more aggressive lesions,” she says. Dr. Kong co-chairs with Anne M. Mills, MD, the lower anogenital squamous terminology for HPV-associated lesions guideline update expert panel.
The updated LAST guideline will also address recommendations on biomarker testing, including best practices for p16 utilization. In its review of the literature, the expert panel found that LAST biomarker recommendations have been followed inconsistently since their introduction in 2012. For example, some studies reported a high percentage of cases of HSIL (CIN2) that are p16 negative. “Based on the 2012 LAST guidelines, p16 was supposed to be used to adjudicate cases of potential CIN2. A negative p16 stain would strongly favor LSIL or non-HPV lesion while positive p16 would support a diagnosis of HSIL (CIN2) in the appropriate morphologic context. Seeing these studies that had a lot of cases diagnosed as HSIL (CIN2) that were p16 negative, it became clear that pathologists are not consistently following the original LAST guideline,” Dr. Kong says.
The expert panel found it did not always agree with the morphologic diagnoses or p16 interpretation as shown in the figures. Many p16-negative CIN2, for example, were morphologically more suggestive of benign and metaplastic processes to many of the panel members. “That’s what led us to the realization that morphology is a faulty gold standard since there is not an agreed upon set of criteria for the diagnosis of -IN1 versus -IN2 and -IN3. We also noticed variability in what is considered a positive p16 stain. If you want to use biomarkers to reduce interobserver variability, a reliable gold standard is needed to accurately classify lesions and a standard definition of what constitutes a positive stain,” Dr. Kong says.
“The expert panel, in talking about this, affirms that p16 is a good, negative prognostic marker,” she continues. “When it’s negative, it’s helpful in saying it’s not high-grade disease. In contrast, positive p16 can be seen with LSIL and HSIL, and that is when standardized morphologic criteria become important in reproducible classification.”
The panel, recognizing the variability in diagnosis, would like to get a better sense of where pathologists are setting their thresholds, she says. “What percentage of HSILs should reasonably be negative for p16? The expert panel thinks it should be a very small percentage. But in practice, do pathologists think it should be higher? Based on the literature review, there appears to be wide variability.” These are some of the challenges and questions that will be addressed in the updated LAST recommendations.
In the same session, James S. Lewis Jr., MD, senior associate consultant at Mayo Clinic in Arizona and professor of laboratory medicine and pathology, Mayo Clinic Alix School of Medicine, will present highlights of the HPV testing in head and neck carcinomas guideline update, which is pending publication in Archives of Pathology & Laboratory Medicine.
For the guideline update, the expert panel performed a systematic review of the large amount of new published data on high-risk HPV in head and neck carcinomas, and there are several areas in which the data has led to changes in recommendations. The changes, in particular, focus on “fencing in” p16 immunohistochemistry, with much more application of HPV-specific testing in oropharyngeal squamous cell carcinoma patients, in cervical cytology specimens, and in specimens from metastatic SCC of unknown primary.
“p16 is a great surrogate of transcriptionally active high-risk HPV here in the U.S. where HPV-attributable rates approach 70 to 80 percent, but worldwide, rates are lower,” Dr. Lewis says. “Do we think p16 immunohistochemistry is suitable as a standalone surrogate marker in every oropharyngeal SCC patient across the world? Certainly we do not.” When the attributable fraction is 50 percent or less, high-risk HPV RNA in situ hybridization or high-risk HPV DNA PCR is recommended testing for all p16-positive patients, says Dr. Lewis, who is co-chair with William Faquin, MD, PhD, of the HPV testing in head and neck carcinomas guideline update expert panel.
HPV-specific testing is now recommended for all fine-needle aspiration specimens from patients with metastatic SCC to cervical lymph nodes. p16 immunohistochemistry is discouraged. “Bill Faquin says that when p16 immunohistochemistry is applied to FNA cell block material, for unknown reasons it performs poorly compared with HPV-specific testing for predicting high-risk HPV status,” Dr. Lewis says.
When p16 immunohistochemistry is applied to FNA cell block material, it has a low sensitivity compared with HPV-specific testing for predicting high-risk HPV status. In addition, HPV-specific testing is recommended for all patients with metastatic SCC of unknown primary, regardless of location in the neck or tumor morphology.
The sinonasal tract is an anatomic site where 25 to 30 percent of SCC are high-risk HPV associated. “The data in recent years clearly shows a prognostic benefit for patients with transcriptionally active high-risk HPV-associated SCCs,” Dr. Lewis says. “The new guideline recommends across-the-board HPV-specific testing for all of these patients in routine clinical practice.”
Not all HPV-specific tests are the same, to be sure. DNA in situ hybridization, as demonstrated by Keung, et al., in a 2020 paper, lacks sensitivity for high-risk HPV in low copy number cell line specimens (Keung ES, et al. Arch Pathol Lab Med. 2020;144[3]:344–349). The updated guideline recommends either high-risk HPV RNA in situ hybridization or DNA PCR as the HPV-specific test for p16-positive patients and strongly discourages the use of DNA in situ hybridization.
“Only 85 to 90 percent of high-risk HPV-associated oropharyngeal SCC are caused by HPV 16, so the panel recommends that the HPV-specific test have broad enough coverage of the other high-risk types that uncommonly cause these cancers, including HPV 18, 31, and 33. The updated guideline recommendations serve to tighten up the clinical testing in these patients and provide patients and treating physicians with consistent information in their pathology reports,” Dr. Lewis says.
One related area to be addressed in the CAP session, Dr. Broaddus says, is professional practice gaps—the difference between what is and what should be. He cites a few examples: “Are we overusing p16 immunohistochemistry when we should be using HPV testing? Are we generalizing primary tumor to metastasis? Is it okay to assess the metastasis if we don’t have the primary site?”
Bogdan Pasaniuc, PhD, will present a Timely Topics Lecture on polygenic risk scores for genomic medicine.
He is professor of pathology and laboratory medicine at UCLA and founding director of the Center for Computational Biomedicine, Department of Genetics, Penn Institute for Biomedical Informatics, Perelman School of Medicine at the University of Pennsylvania.
The polygenic risk score is a tool to predict individuals with high genetic predisposition (or risk) for common diseases, he says. “Think about it as heart disease or breast cancer, or any type of diseases that are common enough in the population such that, because they’re common, they’re multifactorial, so there are many different genes that interact to put someone on a spectrum of high risk for a particular disease.”
In breast cancer, for example, BRCA1 and 2 have “high effects,” he says, “but there are dozens, if not hundreds, of other genes that each have a tiny effect. Each of those lesser mutations doesn’t move the needle too much.” However, if a person has all of the risk variants and the risk is adding up, “that puts those individuals at higher risk for that disorder.” That’s the polygenic risk score.
“As of yet, they don’t do a good job of identifying which people are at high risk or not,” Dr. Pasaniuc says of the scores, but they are becoming more reliably predictive for various diseases and disorders such as diabetes, heart disease, and breast and prostate cancers, among others (for example, the conditions studied by the eMERGE Network: https://emerge.study/conditions).
Though pathologists don’t interact now with the polygenic risk score, they are likely to do so in the near future, he predicts. “For example, in heart disease, there are already polygenic risk scores that are able to identify the five or 10 percent of people who are in the high-risk category.” Clinical trials are underway now to determine, for those who are at higher risk based on their polygenic risk, “whether getting prescribed statins will reduce their risk for developing heart disease in the long run.”
Cost, complex logistics, and ethical implications are barriers to polygenic risk score use in clinical practice, he says. Another issue is what to do with the scores. “For some of these traits, say status for heart disease, there’s something to do, but for some other traits there isn’t much to do,” he says. It’s more of an information item that provides an idea about prioritizing which of the polygenic risks should be acted on.
In the USCAP lecture, Dr. Pasaniuc says, “I’ll try to make the case that there’s a huge benefit for trying to solve all these questions one at a time, and because there’s a lot of information there to be gleaned from polygenic scores to make better informed decisions for every patient, I’ll make the case that it’s to the medical system’s benefit, as well as the patient’s and clinician’s benefit, to have genetic risk profiles for every patient in the hospital.”
“You can imagine,” he continues, “the pathologist looks at one’s genome and identifies whether that person has a rare or a genetic disorder. That’s already happening with rare genetic disorders, the one in 10,000 or 100,000, which are defined as a mutation that knocks out a gene and that most of the time leads to some phenotype or some disease but is rare.”
That same principle holds for polygenic risk scores as well, he says. “Imagine the same thing happening for one gene, one disease at a time. A pathologist will look at one’s genome, and now, instead of looking at just one gene they will interpret the entire genetic contribution under polygenic risk, or for a particular disease like polygenic breast cancer. That’s where the pathologist will interact with it or assign diagnoses or risk profiles, if you will.”
“We’re imagining this genetically informed risk assessment that we provide for specialty clinicians will essentially let them make statements about how that particular patient should be diagnosed, treated, and followed up,” Dr. Pasaniuc says.
In the United States, a central concern is equity of access, he says. “We have huge problems in this country in terms of who even gets to be seen, to get access to health care. A second issue is that certain groups or sociodemographic strata of the population are participating less and less in genetic-risk related studies because of the history of misuse of genetics.”
“So there are many obstacles to overcome with respect to essentially making sure that tools like polygenic risk scores are accessible to everyone.” But, he adds, “There’s a lot happening on that front here in the U.S.”
Anne Paxton is a writer and attorney in Seattle.