Publication
Article
Author(s):
Lynette M. Sholl, MD, discusses why cancer pathologists still want better tissue samples.
Cancer pathology has undergone radical changes in recent years. Analysis of a seemingly small number of tumor types, classed almost entirely by anatomic site, has become analysis of dozens of tumor types, classified by a combination of genomic aberrations and location. Pathologists who once performed largely visual analysis now supplement their eyes and experience with hard data from a variety of assays.
But one fact remains constant: Your pathologist still wants better tissue samples. In the lung cancer field, for example, shortcomings in specimens provided to laboratories are a significant cause of molecular testing failures, according to findings from a survey conducted by The International Association for the Study of Lung Cancer (Figure).1
“It’s a cliché, but we always do want better samples. There are all these challenges to working with tissue that just don’t exist for other diagnostics like radiology, which gives you this digital image that loads right into the record, never degrades, and follows patients around forever. If a biopsy is too small, it doesn’t give you the data you need. If it’s not processed correctly, it can get lost. If it’s placed in the container carelessly, it can stick to the side, dry up, and fall apart before you can analyze it. And if any of these things happen, you either get limited pathologic info, or you go back to the patient and set up another procedure,” said Lynette M. Sholl, MD, in an interview with OncologyLive®.
Sholl, an associate professor of pathology at Harvard Medical School and a pathologist at Brigham and Women’s Hospital, both in Boston, Massachusetts, has spent much of her 20-year career focusing on integrating molecular pathology into oncology practice. A lung cancer specialist, she has played a leading role in developing protocols for tissue handling and workflows as well as in launching and validating assays and sequencing panels.
At the same time, Sholl has been an advocate of communication between pathologists and other oncology specialists to make the most of the treatment opportunities that molecular diagnostics can provide for patients. On June 25, Sholl will cochair the 6th Annual International Congress on Oncology & Pathology™: Toward Harmonization of Pathology and Oncology Standards, a virtual, interactive conference hosted by Physicians’ Education Resource®, LLC (PER®) (Sidebar).
“People think of pathology as generating very black and white results and sometimes it does, but sometimes there’s a tremendous amount of uncertainty because the specimen wasn’t well fixed or it’s an intermediate-grade tumor or any number of other problems,” Sholl said. “There are countless variables that can create uncertainty in pathological analysis, and oncologists should have a sense of what those variables are.”
The conference also offers an opportunity to explore biomarkers important to the treatment landscape in a range of tumor types. Sholl believes that the genetic literacy so central to molecular diagnostics is lacking in the oncology community. “It is essential to understand the basics of mutations—germline vs somatic—and [the implications for] an individual patient when we report out specifics around those mutations, such as what a variant allele fraction means,” she said.
Sholl, a graduate of Stanford University School of Medicine, has been specializing in pathology since her days as a resident at Brigham and Women’s Hospital in the mid-2000s. Her fascination with analyzing samples, however, started in childhood. Her mother occasionally brought home microscopes from the community hospital in Pennsylvania where she worked as a medical technologist and let Sholl view specimens they collected together.
“We’d look at things like water from the creek near our house, and we’d find things like protozoa,” Sholl said. “It didn’t lead me straight to medicine, but it may have laid the foundation for my ultimate career choice. I did have a predilection for looking down a microscope from the earliest preclinical coursework we had in medical school. I felt like I got what was going on, in contrast to a couple of my classmates who would have a panic attack whenever a microscope was placed in front of them.”
That predilection for looking down a microscope ultimately led to a career that balances clinical practice with research focusing on identifying pathologic, immunohistochemical, genetic, and genomic markers that will improve the classification of lung cancer and provide predictive information regarding prognosis and therapy.
“My research focuses on practical applications, ways to use molecular information as a ground truth that informs H&E [hematoxylin and eosin] histology to make it more accurate and informative, to allow people to distinguish between things that look very similar and parse out fine details of what they’re actually seeing,” Sholl said. “For instance, if you can take specific molecular markers that you see more commonly in one tumor type vs another and then find a protein correlate that you can find via immunohistochemistry [IHC], you can identify the tumor type in a day rather than waiting for molecular testing, and that’s a really powerful way to aid in diagnosis.”
Sholl helped to do just that with ALK gene rearrangements that predict response to ALK inhibitors in patients with non–small cell lung cancer (NSCLC). She worked to correlate ALK gene fusions identified by next-generation sequencing (NGS) panels with ALK-related protein expression that could be detected by IHC.2,3 Her laboratory also was the first to validate the clinical utility of IHC testing for ROS1 gene rearrangements in NSCLC.4
“It was a significant discovery because immunohistochemistry is routinely available to pathology labs around the world while most of them do not yet have access to [molecular methods] or FISH [fluorescence in situ hybridization] that had previously been necessary to identify targetable ALK rearrangements,” said Jason Hornick, MD, PhD, director of surgical pathology and the immunohistochemistry laboratory at Brigham and Women’s Hospital and a professor of pathology at Harvard Medical School.
“Even in places that have next-generation sequencing, using immunohistochemistry to identify actionable genetic alterations has a significant advantage: time. The most efficient labs take 2 weeks to return next-generation sequencing tests, but we can return immunohistochemistry results in a day. That difference can be important for patients with aggressive metastatic cancers.”
Sholl’s efforts to improve biomarker testing go beyond IHC testing. Since 2012, she has served as associate director of the Center for Advanced Molecular Diagnostics (CAMD) at Brigham and Women’s Hospital, where she helped validate and launch droplet digital polymerase chain reaction (ddPCR) assays.
The technology leverages advances in microfluidics to emulsify DNA with oil into thousands of droplets that can then be analyzed via PCR and flow cytometry.5 Sholl worked with Geoffrey R. Oxnard, MD, and Cloud Paweletz, PhD, from Dana-Farber Cancer Institute and Neal I. Lindeman, MD, from CAMD on the assays.
The team developed a ddPCR assay to detect EGFR sensitizing and resistance mutations in plasma from patients with NSCLC and a similar assay for identifying BRAF V600 mutations in specimens from patients with metastatic melanoma. Investigators from these institutions were the first to demonstrate that serial genotyping using ddPCR of cell-free plasma DNA can signal the development of resistance before clinical presentation.5
Findings from further clinical analysis have demonstrated that ddPCR technology can be used to detect EGFR L858R mutations and exon 19 deletions in DNA isolated from plasma and formalin-fixed paraffin-embedded tissue with a faster turnaround time than NGS.
In February 2022, Sholl and colleagues reported a median turnaround time for ddPCR of 6 days (range, 1-10) on tissue specimens and 4 days (range, 1-11) on plasma samples compared with 13 days (range, 6-133) with NGS testing using the OncoPanel tool. Data from the ddPCR approach was 100% concordant with samples analyzed through OncoPanel.6
However, the sensitivity of ddPCR was 53% on cell-free tumor DNA isolated from plasma, a lower level than that observed in other studies of the technology. Investigators theorize that differences in patients’ cancer stage and prior treatment may have affected the assay’s detection performance. By contrast, the OncoPanel’s sensitivity was 97.8% on single nucleotide variants and 100% on all but 2 insertion and deletion variants (FLT3 and NPM1).
In addition to diagnostic technologies, Sholl has focused on improving workflows to enable more extensive biomarker testing. One of these efforts is the LUNGCOR histology protocol, in which lab samples from lung biopsies are prospectively identified for biomarker testing so that additional unstained slides can be prepared and reserved for further study. In an analysis of 300 specimens collected at Brigham and Women’s Hospital, the LUNGCOR protocol resulted in an increase in the availability of tissue for molecular testing and a reduction in the risk of sample inadequacy through further handling.7
Sholl also is involved in helping to translate diagnostic information into clinical practice. She is scheduled to serve as the pathology principal investigator on the Lung Cancer Mutation Consortium’s next major study, LCMC4 (NCT04712877). The planned umbrella trial will screen samples from patients with early-stage (IA2-III) NSCLC who are candidates for neoadjuvant therapy for 10 actionable mutations and then assign therapy based on whether there is a match. The study, which is not yet recruiting participants, aims to enroll 1000 patients undergoing treatment at cancer centers in 14 states.8
The landscape for treating patients with NSCLC in the neoadjuvant setting is changing. In March 2022, the FDA approved the combination of nivolumab (Opdivo), a PD-1 inhibitor, plus platinum-doublet chemotherapy for patients with resectable NSCLC (tumors ≥ 4 cm or node positive). The decision marks the first time the FDA has approved a neoadjuvant therapy for early-stage NSCLC.9
The approval was based on findings from the phase 3 CHECKMATE 816 trial (NCT02998528), in which 358 patients with stage IB, II, or IIIA tumors were randomly assigned to receive platinum-doublet chemotherapy with or without nivolumab every 3 weeks for up to 3 cycles. The median event-free survival was 31.6 months (95% CI, 30.2-not reached) with the nivolumab-containing regimen compared with 20.8 months (95% CI, 14.0-26.7) for chemotherapy alone (HR, 0.63; 97.38% CI, 0.43-0.91; P = .0052). The pathologic complete response rate was 24% (95% CI, 18.0-31.0) with nivolumab plus chemotherapy and 2.2% (95% CI, 0.6-5.6) with chemotherapy alone.9
Sholl said the key considerations for the newly approved therapy are “less about the biomarker and more about developing a robust process for response assessment following immunotherapy in the neoadjuvant setting.”
Given both the shortage of pathologists10 and the relative difficulty of obtaining tissue rather than blood for testing, investigators are exploring strategies for replacing labor-intensive aspects of anatomical or tissue-related pathology with automated tests from clinical pathology. Among the most promising strategies is analyzing circulating tumor DNA (ctDNA) from blood samples to identify the mutations that should drive treatment.
“The impact has been hard to quantify, but I suspect that we have already had some decrease in primary molecular diagnostics requests on tissue. Oncologists will often check ctDNA before tissue because you can draw the blood faster than you can schedule a biopsy, and if the patient has a sufficient burden of ctDNA in their blood, the oncologists will get the relevant predictive biomarker information from the plasma sample,” said Sholl.
She noted, however, that current limitations of ctDNA testing restrict their ability to replace assays performed on tissue samples. “The big caveat,” Sholl said, “is that not all tumors shed enough DNA into the blood for good analysis. If a patient has a tumor that doesn’t shed much, a plasma test is not going to be informative. So one of the challenges that clinicians face is knowing when they do or do not have an informative ctDNA test, and it’s not always easy to tell. You can pick up [information] that may be common in, say, myeloid lineages and incorrectly think that you are capturing some information about your patient’s lung cancer, for instance. You need some degree of sophistication in terms of what you can act on as a predictive biomarker in the blood and when you need to proceed to getting a proper tissue sample to profile.”
In most cases, Sholl believes that ctDNA tests are optimally used as complements rather than substitutes for tissue-based assays.
“Tissue and the ctDNA plasma tests can give you different information, so ctDNA info can be a valuable addition,” she said. “There can be tremendous heterogeneity in tumor biology at different tumor sites. Patients may have multiple different relevant mutations in different places, and you’re not going to get that from tissue taken from 1 site. On the other hand, with tissue profiling, you know more about the tumor biology, and, of course, you know exactly where the tumor with that biology is. With plasma tests, you have to infer where those variants you’re seeing are coming from.”
The value of ctDNA testing is likely to grow as the testing industry matures. “Among the big challenges with ctDNA testing in clinical practice right now is that it’s decentralized and unstandardized and not very user friendly. You put the sample in the mail, and when the results come back, they’re often in a PDF or something else that isn’t searchable and isn’t easy to enter into the fixed fields of the medical record, so there’s a lot of information that’s not getting used like it should because it’s hard to find,” Sholl said. “As that improves, the tests will be better utilized.”
Sholl’s work occasionally brings her into direct contact with patients, sometimes when an oncologist wants help explaining a particularly complex pathology report or when a patient has questions the oncologist cannot answer. For the most part, however, Sholl’s clients are other physicians who rely upon the information she provides. Working like this, at one step removed from the patient, may partially insulate pathologists from the intense emotions of treating patients with cancer, but only partially.
“You do have a bit more distance than the medical oncologist or the surgeon would, but you’re not immune,” Sholl said. “I see literally thousands of potentially tragic situations over the course of the year. When I’m rendering this new diagnosis of lung cancer, and not just lung cancer but hideous metastatic lung cancer, and then I look down at the demographic information and see it’s a 28-year-old woman with a young child, it’s hard. And sometimes, I get 10 of those in a row. And it’s terrible.”
For pathologists, the challenge is to understand exactly what information different specialists need and to provide it in a way that allows them to make the best treatment decisions. “I think, as a pathologist, you have to demonstrate a level of humility and willingness to seek additional opinions because the complexity of the human body is sort of mind boggling…So reaching out to colleagues throughout the department is often really critical.”