Breast cancer is a heterogeneous disease that develops and progresses because of alterations in genes that control cell growth, proliferation, and differentiation.1 To ensure optimal management, each patient diagnosed with the disease should receive treatment that is guided by the molecular composition of his or her tumor. Although such a scenario is still not fully realized, biomarkers are beginning to play an important role in preparing the way for precision treatment in breast cancer. In particular, biomarkers are increasingly being used to predict patient outcomes and help determine the most appropriate type of systemic therapy to be used. Reflexive testing for biomarkers in newly diagnosed cases of breast cancer includes estrogen receptor (ER) and progesterone receptor (PR) status for selecting patients who should receive endocrine treatment, and human epidermal growth factor receptor 2 (HER2) status for identifying patients who are likely to benefit from anti-HER2 therapy.
One of the challenges that many oncology practices face is developing best practices that will allow the efficient and cost-effective incorporation of biomarker testing for prognostic and predictive purposes into the care of patients with breast cancer. The development of these best practices was the focus of an informative and insightful conversation between a pathologist and a medical oncologist. Harris S. Goodman, MD, is a community-based pathologist in San Francisco, CA, who is affiliated with urban, mid-sized hospitals, including Saint Francis Memorial Hospital and Alameda County Medical Center. Alan M. Kramer, MD, is a community-based medical oncologist in private practice in San Francisco. He specializes in treating patients with breast cancer, and is affiliated with several hospitals, including California Pacific Medical Center, Saint Francis Memorial Hospital, and the University of California at San Francisco. In this publication, Dr Goodman and Dr Kramer explain how they work together as part of the cancer care team to deliver individualized treatment to patients. Because these 2 experts are in community-based practices, their perspectives on the development of best practices to effectively incorporate biomarker testing into precision treatment for patients with cancer—especially those with breast cancer—will be of value to other community medical oncologists and pathologists.
Biomarkers are increasingly being used to predict patient outcomes and help determine the most appropriate type of systemic therapy to be used.
Impact of Biomarker Testing on Cancer Therapy
Dr Goodman and Dr Kramer agreed that biomarker testing has had a huge impact on the way they manage patients with breast cancer and often drives treatment strategies. Dr Goodman provided a pathologist’s perspective, “It is interesting. When I first started practicing pathology, our goal was to simply make the most accurate and precise diagnosis, and then characterize the tumor in terms of its size, grade, involvement in neighboring structures, and so on. That has really changed recently. We are expected to do both prognostic and predictive testing, and that includes biomarker testing. Breast cancer is unique in that we do biomarker testing reflexively. Dr Kramer does not need to ask me to do ER, PR, HER2, or Ki-67 testing. I will automatically do that for all breast cancer cases and include our findings as part of the pathology report.”
Dr Kramer provided a medical oncologist’s perspective, “I see patients being considered for neoadjuvant therapy, and sometimes for adjuvant therapy. If it is for adjuvant or neoadjuvant therapy, I usually look at the pathology report to see whether they are ER-/PR-positive. If they are Ki-67-positive, that is sometimes helpful in classifying them; I then look for HER2 expression.”
Today, the main challenge for oncologists and pathologists is the identification of the appropriate predictive biomarkers for the selection of optimal treatment.
The comments by these experts reflect the fact that the standard of care for patients with breast cancer has evolved significantly from empirical treatments based solely on the tumor’s clinical and pathologic characteristics to the use of targeted approaches based on the molecular profile of the tumor, especially the so-called “driver mutations.” Today, the main challenge for oncologists and pathologists is the identification of the appropriate predictive biomarkers for the selection of optimal treatment, which can help avoid treatment-associated side effects and minimize the overall cost of therapy.2
To improve on current standards for breast cancer prognosis and prediction of the benefit of therapy, Parker and colleagues developed a risk model that incorporates gene expression–based intrinsic subtypes of breast cancer, including luminal A, luminal B, HER2-enriched, and basal-like.3 As shown in Table 1, these subtypes are characterized based on their immunohistochemical and molecular profiles with resulting options for treatment.2,3 Dr Goodman noted, “Certain biomarker testing is linked to certain therapeutic agents. Rather than do a whole bunch of different tests, I will ask Dr Kramer, ‘Which agents are you considering using on the patient?’ Then I will perform the appropriate predictive biomarker test.”
Prognostic biomarkers in breast cancer have come a long way. In recent years, 5 novel gene expression prognostic tests for breast cancer have been developed: MammaPrint®, MapQuant Dx™, Oncotype DX®, PAM50, and the Theros Breast Cancer Index. The rationale for developing multigene-based prognostic tests is not only to add prognostic and predictive information to conventional biomarkers, but to provide more reliable and reproducible techniques than immunohistochemistry-based assays, thereby reducing technical errors and subjective interpretation.4 One of the first commercially available and US Food and Drug Administration (FDA)-approved signatures was the 70-gene MammaPrint assay, which stratifies patients into low- or high-risk categories for distant metastases at 5 years. More recently, the 21-gene Oncotype DX assay was developed to estimate the risk for relapse in ER-positive, node-negative breast cancer and the chemosensitivity of these tumors. This test divides patients into 3 groups on the basis of their recurrence score (RS): low risk (RS <18), intermediate risk (RS 18-30), or high risk (RS >31).5
The PAM50 test defines the 4 major intrinsic subtypes of breast cancer through the analysis of 50 classifier genes and 5 control genes. Along with the identification of subtypes, PAM50 has been shown to be an independent predictor of survival in patients with the disease.6 PAM50 generates a numerical score (risk of recurrence) that, along with clinical features, estimates the risk for relapse at 10 years in postmenopausal women with stage I/II node-negative or stage II node-positive (1-3 positive lymph nodes) and HR-positive breast cancer.7 A summary of the main features of the 3 most common prognostic biomarker tests for breast cancer is shown in Table 2.2
Dr Kramer provided his perspective on the use of prognostic biomarkers, “In the adjuvant setting, if the lymph nodes are negative and the patient is ER-positive, we are using Oncotype DX to determine prognosis and to see which patients might benefit from hormone therapy alone or chemotherapy plus hormone therapy. If patients are node-positive and ER-positive, we frequently utilize MammaPrint to determine whether they are in a low-risk category and can avoid chemotherapy. That is how we usually approach patients.”
Best Practices in Communication Within the Oncologist-Pathologist Team
A significant issue related to biomarker testing in patients with cancer is effective communication between the medical oncologist and the pathologist. The experts did not fully agree on the topic of who drives the discussion and makes the final decision about which biomarkers to test in a given patient.
Dr Goodman noted, “I would say that we have a collegial discussion. Dr Kramer asks me to perform biomarker testing. It is my job to determine whether the available specimens are adequate and appropriate for testing. If he is looking at PD-L1 therapy, to be efficient, I may ask him which agent he is considering. Dr Kramer is the one treating the patient, so he drives the requests. I evaluate the appropriateness of the request and the specimen, and then determine the laboratory that will perform the testing.”
Dr Kramer had a somewhat different view, “We used to piecemeal it. Now that we are asking for more than 5 biomarkers, we are using outside reference laboratories more frequently. I usually direct it; it is not directed by the pathologist. The medical oncologists will request it, because we have to obtain authorization through our office. We fill out a form and send it in to pathology. Then, the specimen gets sent out. The order would not routinely come from pathology.”
Dr Goodman countered, “When I get the oncologist’s request, I evaluate the specimen to make sure it will be useful before it is actually sent out. Most of the time, it is because we participate in acquisition of the specimen, but occasionally it is not. Then, I will let Dr Kramer know. I will say, ‘You know, this is not likely to give you an answer, and we may want to look at a different sample or perform another biopsy.’”
“Now that we are asking for more than 5 biomarkers, we are using outside reference laboratories more frequently.” -Alan M. Kramer, MD
This conversation clearly indicated that decisions are made in collaboration between the medical oncologist and the pathologist, with required input from both specialists. How does the discussion occur as to whether biomarker testing in a particular patient is appropriate and which biomarkers are important? The experts explained that these conversations occur during a tumor board, at meetings of cancer committees, and through consultations conducted by e-mail or phone, as well as in person.
Dr Goodman remarked, “Dr Kramer will stop by my office to have a face-to-face discussion, or he will call me. We also have 2 tumor boards a week, so the conversation may occur at the tumor board. Because our hospital is small, it is easy for Dr Kramer to stop by my office; I think that is an advantage to working at Saint Francis Memorial Hospital. Although, at larger institutions, there is the benefit of having a specific molecular tumor board. I think both approaches work. The main thing is that the patient gets the right testing on the right tissue for the right reasons.”
Dr Kramer added, “At California Pacific Medical Center, we have a molecular tumor board where we present cases to determine whether a patient is qualified for any clinical studies based on molecular expression. We have the research team in the room and we discuss molecular studies from Foundation Medicine. Representatives from the company are on the line and they discuss their findings. Sometimes, they direct us to clinical trials that patients may be candidates for.”
Moreover, a recent study showed that input from several disciplines (ie, medical and surgical oncology, pathology, radiology, and nursing) into case reviews stimulates optimal decision-making, reinforcing the importance of multidisciplinary tumor boards and informal communications involving the entire cancer care team.8
“The main thing is that the patient gets the right testing on the right tissue for the right reasons.” -Harris S. Goodman, MD
The Process of Biomarker Testing: Practical Issues
Time is an important issue for many providers—the time it takes to collect a biopsy specimen and send it to the laboratory; to process the specimen and obtain results from biomarker testing; to interpret results; and to discuss treatment options with oncologists so they can advise their patients. This process typically takes 7 to 10 days or sometimes even longer, depending on the type of testing being performed. If next-generation sequencing (NGS) is included in the testing, the process may take 2 weeks or longer.
Sharing his expertise on the topic, Dr Goodman said, “From my perspective, the way it typically works is the biopsy is obtained and the very next day, I get the slide. I look at it and make a diagnosis. Because I know from the order that Dr Kramer is going to be the treating oncologist, I will call him and ask, ‘Do you want a specific biomarker testing done?’ He will usually say ‘yes,’ and then I will send the block off that day. It cannot be any faster than that. Once the block is sent out, whether to Foundation Medicine or NeoGenomics, we wait for the results. The other thing I do, especially in cases where we are not sure of the diagnosis in the computed tomography suite, is divide the specimen into 2 parts. Now, I have 2 paraffin-embedded tissue specimens. If I need to perform testing to clarify whether the tumor is a metastasis or a primary, and if it is a metastasis, what the primary source is, I can be doing that on 1 block while the other block is out for molecular biomarker testing. Having 1 particular tissue for 2 different types of testing, allowing it to be done parallel, reduces the turnaround time. However, it still ends up being 1 to 2 weeks, especially with some of the molecular pathology tests.”
Is 1 to 2 weeks acceptable to the treating oncologist? Dr Kramer replied, “It depends on the condition of the patient. If the patient looks like he or she needs treatment, we cannot wait and must use our best guess in terms of systemic treatment. If the patient looks like they are stable and can wait a few weeks, then we wait.”
“I think that creating 2 paraffin tissue blocks instead of one is a significant benefit when we are dealing with a small amount of tissue and ordering tests using different methodologies.” -Harris S. Goodman, MD
Can anything be done to shorten that interval or accelerate the process from the time that the biopsy specimen is collected to the time that an actionable mutation has been identified on which a treatment decision can be made? Dr Goodman offered the following recommendations, “I think that creating 2 paraffin tissue blocks instead of one is a significant benefit when we are dealing with a small amount of tissue and ordering tests using different methodologies (ie, immunohistochemistry, fluorescence in situ hybridization, reverse transcription polymerase chain reaction). Even if it is just half a core in each block, I can do all the necessary testing. If I had 2 cores of identical tumor tissue and I really just needed 1, I am not wasting tissue. I find that creating 2 blocks initially is very helpful. Then, calling the oncologist the day I have the slide on my microscope and finding out what his or her plans are for the patient, also helps to streamline things.”
The availability of sufficient tissue to perform all requested biomarker testing is also a problem cited by many pathologists involved in predictive testing. There have been a number of unique approaches to this issue, including measuring circulating mRNA in peripheral blood (particularly for HER2-overexpressing breast cancer),9 as well as investigating tumor-infiltrating lymphocytes as a correlate to clinical responsiveness, especially in the neoadjuvant treatment of triple-negative breast cancer.10
When asked about liquid biopsies, Dr Goodman said, “My feeling is it is a one-tailed test in that if you send it out and you get a positive hit, and it is positive for an actionable mutation, then it is useful. If it comes back negative, I would feel uncomfortable. Obviously, what to do with it is up to the oncologist, but I would not be comfortable saying there is nothing there.”
Dr Kramer agreed, “I was involved in a clinical trial looking at liquid biopsies for patients with lung cancer. They looked at it prospectively. I did not find it particularly helpful for those patients. I do not feel there was a high degree of sensitivity. I think they are still trying to sort things out.”
Cost and Value in Biomarker Testing
Although it is evident that associations between relevant genes and corresponding targeted drugs have been discovered, it is also clear that biomarker-guided treatment needs to be cost-effective as well as clinically beneficial to be incorporated into the treatment paradigm for patients with cancer.11 In a recent study, economic evaluations were reviewed for biomarker-drug associations listed in the FDA Table of Pharmacogenomic Biomarkers in Drug Labeling. Of 137 pharmacogenomic associations identified in the table, 57% were economically in favor of biomarker testing, of which 30% were cost-effective and 27% were actually associated with a cost-savings.11 The study authors developed a model for evaluating pharmacogenetics-guided treatment versus standard treatment (Figure).11 This model predicts that if the pharmacogenetics-guided treatment reduces costs and achieves a better clinical outcome than the standard treatment, the pharmacogenetics-guided strategy would be preferred over standard therapy. In contrast, if the pharmacogenetics-guided option costs more but is less clinically effective than standard therapy, the preferred option would be standard therapy versus pharmacogenetics-guided treatment.11 When 1 treatment costs more but is clinically more effective than the other, the payer’s “willingness-to-pay threshold” will ultimately determine cost-effectiveness and whether a molecular biomarker test will be covered.
Dr Goodman noted, “I would say that I try to be very cost-effective and efficient with the ordering of these tests. I do not order all of the PD-L1 antibody clones available—only the ones that the oncologist tells me he or she is thinking of using. Ultimately, the hospital has to collect their component billing or they end up eating it—for lack of a better word—if it is not paid. The reality is that costs do not affect me significantly, but I am still aware of them and try to be cost-effective in all that I do.”
Dr Kramer concurred, “Foundation Medicine obtains insurance authorizations, so the patient is not stuck with the bill. If a patient is stuck with the bill, they will frequently perform the test for free. I have not had a patient stuck with a bill. Although actionable targets are few and far between, patients who have gone through several lines of therapy and various treatments still want to have molecular biomarker testing (and even NGS) performed so they have some understanding of the biology of their tumor. However, we are starting to see a lot of variants that we do not know how to interpret. I think all of this is still evolving.”
“Patients who have gone through several lines of therapy and various treatments still want to have molecular biomarker testing (and even NGS) performed so they have some understanding of the biology of their tumor.” -Alan M. Kramer, MD
Meeting the Challenges of Incorporating Biomarker Testing into Clinical Practice
The experts offered their perspectives on the greatest challenges related to the incorporation of biomarker testing into their clinical practices. Dr Goodman said, “One of the greatest challenges is obtaining enough tissue to perform biomarker testing. Tumor heterogeneity also remains a challenge. We are doing biopsies that result in a small sample of tissue, and intra-tumor heterogeneity is always in the back of my mind. If I am performing expensive biomarker testing on a tiny tissue sample, is it representative of the entire tumor present in the patient, including metastases? Certainly, cost containment is also an issue. The cost of biomarker testing is huge compared with the cost of performing a routine pathologic examination with diagnosis. How do we deal with the rapidly expanding costs associated with biomarker testing? These are the big challenges for me.”
“As I view it, the preordering of tests and refraining from ordering tests that would not be of use to the patient are among the best practices in biomarker testing.” -Harris S. Goodman, MD
Dr Kramer stated, “I think we are going in the right direction to try to understand the biology of the tumor microenvironment, but we are certainly not there yet. Many times, we receive reports where we do not understand the significance of all the different mutations that the patients may have, none of which may be targetable. I think we need to try to understand the biology of tumors and to see what the tumor cell-immune cell interactions are. I would like to see assays of the environment to determine how that affects genetic instability. Also, how these things affect immunocompetence. Presently, there is no comprehensive assay of the immune system.”
The experts were asked about their best practices in dealing with these challenges—concepts that could be adopted by other community-based practices. Dr Goodman said, “From a pathologist standpoint, it is the appropriate management of a limited amount of tissue. It is very easy to waste tissue, and as a consequence not have it available for the tests that the oncologists need to help care for their patients. As I view it, the preordering of tests and refraining from ordering tests that would not be of use to the patient are among the best practices in biomarker testing.”
Dr Kramer said, “I think ‘best practice’ is a term that is still evolving. We do not really know what the best practice is, as yet. It is such a moving target. A best practice today may not be a best practice a year from now. I think the biomarkers for breast cancer have remained the same for years. That really has not changed much. The identification of biomarkers for lung cancer is where there has been quite substantial improvements. If you can identify a molecular biomarker, and there is a targeted therapy, that is very helpful. It depends on the solid tumor. Then, trying to understand how to treat patients with immunotherapy, incorporating that into the mix is important. These things all need to be factored in, so you must have the data in terms of selecting patients. In lung cancer, the best practice is to start with tyrosine kinase inhibitors, because lung cancer is an oncogenic driven tumor, and its biology is different than non-oncogenic tumors. Certainly, we have seen that to be the case, so we understand the biology a little bit better. I would like to see that happen in breast cancer as well, but we are not there yet.”
Many of the best practices discussed in this newsletter are tumor-agnostic, so that they can be applied to biomarker testing (including NGS and other comprehensive genetic profiling) across multiple tumor types, including lung cancer, metastatic melanoma, breast cancer, and other solid tumors, as well as hematologic malignancies. Because molecular biomarker testing is increasingly being used to inform targeted therapy and immunotherapy in patients with cancer, these best practices are relevant for the entire cancer care team, including medical and surgical oncologists, pathologists, pulmonologists, interventional radiologists, oncology nurses, and other healthcare professionals involved in the care of patients.
- Toss A, Piacentini F, Cortesi L, et al. Genomic alterations at the basis of treatment resistance in metastatic breast cancer: clinical applications. Oncotarget. 2018;9:31606-31619.
- Toss A, Cristofanilli M. Molecular characterization and targeted therapeutic approaches in breast cancer. Breast Cancer Res. 2015;17:60.
- Parker JS, Mullins M, Cheang MCU, et al. Supervised risk predictor of breast cancer based on intrinsic subtypes. J Clin Oncol. 2009;27:1160-1167.
- Prat A, Ellis MJ, Perou CM. Practical implications of gene-expression-based assays for breast oncologists. Nat Rev Clin Oncol. 2011;9:48-57.
- Goncalves R, Bose R. Using multigene tests to select treatment for early-stage breast cancer. J Natl Compr Canc Netw. 2013;11:174-182.
- Chia SK, Bramwell VH, Tu D, et al. A 50-gene intrinsic subtype classifier for prognosis and prediction of benefit from adjuvant tamoxifen. Clin Cancer Res. 2012;18:4465-4472.
- Filipits M, Nielsen TO, Rudas M, et al. The PAM50 risk-of-recurrence score predicts risk for late distant recurrence after endocrine therapy in postmenopausal women with endocrine-responsive early breast cancer. Clin Cancer Res. 2014;20:1298-1305.
- Soukup T, Lamb BW, Sarkar S, et al. Predictors of treatment decisions in multidisciplinary oncology meetings: a quantitative observational study. Ann Surg Oncol. 2016;23:4410-4417.
- Wu Y, Meng Q, Yang Z, et al. Circulating HER-2 mRNA in the peripheral blood as a potential diagnostic and prognostic biomarker in females with breast cancer. Oncol Lett. 2018;16:3726-3734.
- Ruan M, Tian T, Rao J, et al. Predictive value of tumor-infiltrating lymphocytes to pathological complete response in neoadjuvant treated triple-negative breast cancers. Diagn Pathol. 2018;13:66.
- Verbelen M, Weale ME, Lewis CM. Cost-effectiveness of pharmacogenetic-guided treatment: are we there yet? Pharmacogenomics J. 2017;17:395-402.