Widespread implementation of next-generation sequencing, along with more patients undergoing such testing, has significantly increased the number of gene variants detected. Consequently, clinical laboratories are facing new quandaries. One is how often to reinterpret data to classify a variant, especially if the variant initially was found to be of unknown significance. Another is when to contact providers or affected patients with any new information.
To try to help, the American College of Medical Genetics and Genomics (ACMG) last year released two policy statements (Genet Med 2019;21:769-71; Genet Med 2019;21:1267-70). One recommended that clinical laboratories have policies and protocols for variant-level reevaluation and case-level reanalysis that keep pace with new developments in population databases and bioinformatics. A second suggested recontacting patients is a shared responsibility among ordering providers, laboratories, and patients.
Neither document put a definitive timetable in place for these actions, which was deliberate, said Josh Deignan, PhD, FACMG, a co-author of both and an associate director of the University of California, Los Angeles Molecular Diagnostics Laboratories.
“We didn’t want to hold anybody to a specific standard, like every variant or every case has to be reviewed every year, two years, or three years,” Deignan said. Instead, the authors felt laboratories should set their own policies without restrictions based on their size, resources, and scope of practice. Deignan and his co-authors also underscored that reclassifications should be expected, and that the reports that labs produce “are really only good as of the date on the report, which is based on our available resources and knowledge at the time,” he said.
Is Every 2 Years Optimal?
Exactly how often findings should be reinterpreted is a matter of opinion. “A lot of people would argue that maybe a year is not enough time, and probably five years is too much time,” said Deignan. “I would say every two years is appropriate.”
The Advanced Diagnostics Laboratory at Children’s Medical Center in Dallas reinterprets its entire historical database of variants every 2 years, based on how swiftly data are changing, said Jason Park, MD, PhD, DABCC, the lab’s clinical director and an associate professor of pathology at UT Southwestern Medical Center. In doing so, lab staff use automatic and manual means of reviewing identified variants and searching databases like ClinVar for updated information.
A study of pediatric epilepsy patients that Park co-authored supports the 2-year time frame, at least in that population (JAMA Pediatr 2019;173:e182302). Park and colleagues reviewed genomic test results from July 2012–August 2015 for 309 patients, 185 of whom had relevant variants. Upon reinterpretation of the tests in May 2017, 67 of the 185 (about 36%) had a change in variant classification, 21 of whom experienced a change of diagnosis.
Other labs with a focus on cancer, including the Genetics and Solid Tumor Diagnostics Laboratory at the University of Washington, in Seattle, reevaluate data every year, said lab director Tina Lockwood, PhD, DABCC, DABMGG. “We’re learning so much right now that things change pretty rapidly.”
Deignan’s lab at UCLA doesn’t abide by a set timetable. Instead, staff will reinterpret information upon request from a provider, or periodically mine the literature for updates. The Children’s Hospital of Eastern Ontario, in Canada, has a similar protocol, said genetic counselor Caitlin Chisholm, MS, CGC. The lab generally reinterprets variants only when prompted, such as a request from a provider, or if a variant is identified in an unrelated proband. With such prompts, it will reinterpret variants of unknown significance (VUS) and likely pathogenic variants after 1 year, and pathogenic variants after 3 years. Her group published a flowchart they use to determine when to reinterpret, which the Canadian College of Medical Geneticists has used as background in discussions about formulating guidelines (Genet Med 2018;20:365-8).
“We identify thousands of variants every year, so it’s just not feasible to reinterpret all of them all of the time,” Chisholm said. “The lab has to pick and choose which variants are most likely to have a change in classification … That’s where we spend a lot of time.”
European guidelines regarding reinterpretation are “very broad and context-nonspecific,” according to the authors of a paper surveying practices among the nine certified genetics laboratories in the Netherlands (BMC Med Genomics 2019;12:17). “There is general consensus that laboratories do not have the duty to actively, periodically reinterpret genetic data,” said co-authors Julia El Mecky, a joint doctoral student at the University Medical Center Groningen, in the Netherlands, and at the University of Southampton, in the United Kingdom; Lennart Johansson, PhD, a postdoctoral researcher at the University Medical Centre Groningen; and Irene van Langen, MD, PhD, a professor of clinical genetics and head of the ethical legal and social implications program at the Groningen center. Indeed, their study revealed that Dutch laboratories reinterpret genetic variants reactively, such as when requested by a clinician or when they detect a particular variant in a new patient that previously was found in other, unrelated patients. These labs believe the main responsibility to initiate reinterpretation of genetic data lies with patients, in tandem with their clinicians.
Based on their study, the authors came up with five considerations they felt guidelines should address: Should laboratories as a routine practice actively reinterpret variants? How does reinterpretation initiated by a laboratory relate to patient expectations and consent? When should reinterpreted data be considered clinically significant and communicated from laboratory to clinician? Should reinterpretation, reanalysis, or a new test be conducted? How are reclassifications perceived, and how might this affect laboratory practice?
However, while guidelines might serve as a goal to streamline service provision and ensure good practice, “the questions that laboratories deal with in practice are much more nuanced and context-dependent than those that are, and arguably could or should be, addressed in policy or guidelines,” said El Mecky, Johansson, and van Langen in an email exchange. The Groningen center lab is investigating possibilities that might facilitate standard periodic reinterpretation of variants. In addition, the lab is developing pathways to more easily allow clinicians, in consultation with patients and families, to request a reinterpretation or reanalysis in so-called cold cases, where no diagnosis had been made.
A Duty to Contact Patients?
None of the sources CLN interviewed contact patients directly—all share any new information only with the ordering provider or clinic. They recommended that both consent forms and direct counseling sessions inform patients that interpretations can change based on new information and encourage patients to update their contact information as appropriate.
“Everybody agrees on the principle that if the lab knows new information that would impact how someone takes care of a patient, the provider wants to know that and tell that to the patient,” Chisholm said, “but when you get into the nitty-gritty of how to make that work, it gets a lot more complicated.”
Chisholm tells patients who receive a VUS report the onus is on them to call the lab back in a year or two to trigger a reinvestigation. “That works out well for patients who are highly motivated and engaged, and sophisticated and savvy with the medical system, but the majority of patients never call us back.”
Clinical laboratories face several challenges in both reinterpretations and recontacts. One is information. ClinVar is one of the only public databases for multiple labs, Park said, but not every lab submits material. In some cases, information available on a variant in a database has been reported by just one lab, which potentially causes diagnostic uncertainty. And, people from underrepresented minorities are much less represented in the databases, too, Lockwood noted. Efforts like the National Institutes of Health’s All of Us research program, which aims to gather data from 1 million diverse Americans, should help.
The Dutch laboratories surveyed cited workload as a factor impeding active, periodic reinterpretation, with the process seen as “laborious and unamenable to complete automation,” although El Mecky, Johansson, van Langen, and their co-authors suggested that reclassifying only variants for which clear, new evidence is present regarding a pathogenic or benign effect could benefit patients without exhaustive database searches.
Lack of reimbursement for the time and effort spent on reanalysis is another concern, Deignan said. “There’s a sense that we should make sure people have updated classifications and reports when possible, and to do what’s best for patient care, but other than a re-review of the entire case, for which there is a CPT code, a one-off variant reevaluation isn’t typically thought to be a reimbursable activity.”
Some labs are looking for advanced software tools to help. For example, a portal through which she could push variant updates automatically to providers or patients would be helpful, Chisholm noted. While no computational barriers exist to reinterpreting variants on a regular basis, getting the diagnosis correct can only be solved through clinical trials and better basic science research directed at disease models of those specific variants, Park cautioned.
Eventually, said Deignan, there will be enough information in ClinVar and other databases that labs will discover that variants new to them may already have been vetted by several others.
Whole-genome sequencing also is changing the field, Lockwood said. “We’re going from studying 1.5% of the genome to three billion base pairs—it’s a pretty significant jump,” she observed. “We still don’t know that much, but this isn’t a problem that’s going away anytime soon, and the more sharing we can do across the entire community, the faster we’re really going to be able to implement broad genomic medicine across different disease indications or healthy population screening.”
Karen Blum is a freelance medical/science writer in Owings Mills, Maryland. +Email: firstname.lastname@example.org