Skip to main content
Premium Trial:

Request an Annual Quote

At AACR-FDA Workshop, Oncologists Grapple With Balancing Risks, Benefits of DPYD Testing

Premium

NEW YORK – By the end of a controversial workshop Thursday on whether to test patients for genetic abnormalities that put them at risk of severe chemotherapy-related toxicities, oncologists seemed willing to accept that, at a minimum, they should test to identify those who completely lack the ability to break down these medications and are likely to die from an overdose.

However, oncologists remained unsure about how to manage patients whose ability to process these chemotherapies, called fluoropyrimidines, may not be clearly understood because they're compound heterozygous carriers of two different DPYD gene variants and the ability of those variants to influence drug response is not well validated.

"I think this workshop has made clear that testing can save lives, but that there's still much more work to be done," said Patricia LoRusso, associate cancer center director of experimental therapeutics at the Yale School of Medicine and chair of the workshop, entitled "To Test or Not to Test — That is the Question: DPD Deficiency and Weighing Potential Harms," which was hosted by the US Food and Drug Administration and the American Association for Cancer Research.

Throughout the meeting, doctors and patient advocates supportive of testing patients for risk of DPYD-related toxicities before prescribing fluoropyrimidines like Xeloda (capecitabine) and 5-FU (fluorouracil) repeatedly challenged experts who said there's not enough evidence to support genotype-guided dosing reductions and that available genetic tests are not as reliable as they'd like in identifying all patients with harmful variants.

Xeloda and 5-FU, "in our head spaces, are not replaceable" in settings like adjuvant treatment of stage III colon cancer or biliary duct cancer, Ravin Garg, a hematologist-oncologist at Maryland Oncology Hematology, said at the meeting, noting that these treatments are considered critical for eradicating micro-metastases and extending survival. "If [patients] have a heterozygosity for a DPYD*2A allele and we do an a priori 50 percent dose reduction, are we also compromising the benefit for those patients?"

The FDA-AACR workshop came about at the urging of Karen Merritt, who began advocating for pretreatment DPYD testing after her mother died after her first 5-FU infusion due to an unknown deficiency of the dihydropyrimidine dehydrogenase (DPD) enzyme. "The oncologist on call said there was no way to know that this would happen when we started this treatment," Merritt, a founding member of the Advocates for Universal DPD/DPYD Testing (AUDT), said at the meeting. "I now know and learned that that is incorrect. She could have been tested for DPD deficiency."

Approximately 1 in 1,000 patients carry two copies of a variant in the DPYD gene that result in the absence of the DPD enzyme that is necessary to process fluoropyrimidines. Between 3 percent and 8 percent of the general population has one copy of a variant associated with lower levels of the DPD enzyme, and these variants are more common among African Americans. Fluoropyrimidines can quickly build up in patients with DPD deficiencies, increasing toxic exposure.

The Clinical Pharmacogenetics Implementation Consortium (CPIC), an internationally recognized guidelines body, recommends testing for four well-studied DPYD variants: DPYD*2A, DPYD*13, DPYD D949V, and DPYD HapB3. The group advises avoiding fluoropyrimidines entirely in patients with a complete deficiency and reducing the starting doses of 5-FU or Xeloda by 50 percent in patients with a partial DPD deficiency and increasing doses gradually while closely monitoring patients for toxicities.

Recognizing the lack of standardized DPYD testing, the Association for Molecular Pathology in concert with CPIC and other expert groups last year published a list of seven DPYD variant alleles for which labs "must test" and an expanded list of variants labs can consider including when designing their tests. Tier 1 "must test" alleles have a known effect on protein function, an appreciable minor allele frequency, and reference materials for assessing labs' proficiency in detecting them, and are technically feasible to detect in a clinical lab. Tier 2 variants meet at least one of these criteria. These groups don't recommend labs include variants of unknown significance or variants with uncertain function in their test panels, and caution against full gene sequencing, which can identify more VUS.

Despite the availability of CPIC guidelines and an expert consensus "must test" variant list, groups that shape oncology treatment practices in the US — the National Comprehensive Cancer Network and the American Society of Clinical Oncology — don't recommend pretreatment DPYD testing. At the meeting, Alan Venook, head of the gastrointestinal oncology program at the University of California, San Francisco, and vice chair of the NCCN committees responsible for anal and colon cancer treatment guidelines, acknowledged that genotype-guided dosing can reduce the risk of toxicities but said that "evidence cited in support of maintained efficacy with dose-reduced fluoropyrimidines … is simply not convincing."

The FDA approved 5-FU in 1962 and then Genentech's Xeloda in 1998 as the first oral chemotherapy. These are widely used backbone chemotherapies, particularly for treating colorectal and breast cancer, and in certain settings, like earlier-stage colorectal cancer, oncologists said there aren't alternative treatments. At the same time, fluoropyrimidines are known to cause serious adverse events in 20 percent to 30 percent of patients, and DPD deficiencies can be a major cause of these toxicities.

In 2003, the FDA updated labeling advising against prescribing Xeloda in those with a known DPD deficiency. Spurred by petitions from physicians and patient advocates, the agency updated the labels for both Xeloda and 5-FU in 2015 and 2016, respectively, to warn of the risk of serious or fatal adverse reactions in those with low or absent DPD activity.

Over the last two years, the agency updated the labels of these drugs yet again (see here and here) with more information about the risk of DPD deficiency, including recommending that doctors use their clinical judgment to decide whether to test patients for DPYD abnormalities and discuss risks and testing with patients. The labels now state that "no [fluoropyrimidine] dose has been proven safe for patients with complete DPD deficiency," and in those with a partial DPD deficiency, "there are insufficient data to recommend a specific dose."

Reasons to not test

At the meeting, the relatives of DPD-deficient patients who succumbed to fluoropyrimidine toxicities argued that the FDA's labeling language doesn't go far enough, that without testing, it's impossible to predict whether a patient has a complete DPD deficiency. Without a testing recommendation from the FDA and without guidance from NCCN or ASCO, however, only a few cancer centers and hospitals in the US have implemented DPYD testing, typically after losing patients with DPD deficiencies to fluoropyrimidine toxicities, such as Dana-Farber Cancer Institute and, more recently, Mass General Brigham.

At MedStar Health, a healthcare organization serving the Baltimore-Washington metropolitan area, DPYD is available for physicians to order, but there's no institutional policy around pretreatment testing. It's up to the doctor to decide whether or not to order this testing, Max Smith, a clinical pharmacogenomics specialist at MedStar, said at the meeting.

Sam Abdelghany, executive director of oncology pharmacy services at the Smilow Cancer Hospital in Connecticut, noted that his institution struggled to find a test panel that is comprehensive enough to not give patients a false sense of security with a negative result, has a turnaround time that allows the results to inform treatment decisions, and the results of which can be integrated into physicians' clinical decision support systems.

The hospital has settled on a test panel from the Mayo Clinic that gauges eight clinically significant DPYD variants but Abdelghany said the test is expensive and isn't integrated into electronic medical records. While pharmacogenomics test reimbursement is improving among commercial and government payors, he noted that there's still no local Medicare coverage for DPYD testing in Connecticut, though that might change this year. But as of now, he said the hospital has around a 70 percent denial rate.

At the Mayo Clinic, meanwhile, there's been a shift from performing the eight-variant DPYD panel, which didn't include all the Tier 1 variants in guidelines, to doing full sequencing. Both tests cost several hundred dollars and can take up to 10 days to return results. Because Mayo is doing full sequencing, however, oncologists are receiving results reporting variants, for which DPD enzyme activity is not clear but a starting chemo dose reduction could still be recommended. Because of these uncertainties, breast cancer and gastrointestinal oncologists at Mayo are not uniformly performing pretreatment DPYD testing, though it is discussed with patients, said Christina Wu, a GI oncologist there.

The main uncertainty that practicing oncologists raised again and again during the meeting had to do with the evidence around whether genotype-guided dose reductions would sacrifice treatment efficacy, particularly for patients with earlier-stage disease receiving adjuvant chemotherapy that could potentially cure them.

One study cited throughout the meeting by supporters of pretreatment DPYD testing and naysayers alike was a 2023 paper in the Journal of Clinical Oncology by Knickman et al. In that study, researchers considered the overall survival and progression-free survival of DPYD variant carriers who received reduced doses of fluoropyrimidines against a matched cohort of patients with normal DPYD who received normal doses. The analysis showed that survival outcomes between these groups were not statistically different.

For supporters of pretreatment DPYD testing, this is just one in a growing body of published data that provides confidence that genotype-guided chemo dose reductions are unlikely to sacrifice treatment efficacy, but it didn't satisfy Venook and others. Venook pointed out that progression-free survival between the groups was "approaching" a statistically significant difference with a p-value of 0.053, "but not quite." He maintained that "to use this as an argument for proving equal efficacy is disingenuous," but that the NCCN panel is not nitpicking over the available evidence. "We're looking at the same data and coming to different conclusions, probably because we're looking for different findings or a different level of certainty from the data."

Venook further asserted that studies used to argue for universal pretreatment DPYD testing often include patients with a range of cancers, typically advanced disease, who are receiving a variety of combination therapies and that the studies are vastly underpowered to address efficacy. Dose reductions recommended by CPIC guidelines aren't specific to 5-FU and Xeloda, he added, even though in the case of the latter, the amount of active drug that enters the body can be affected by factors like patients' diet and other medications.

At UCSF, Venook worked closely with Bani Tamraz, an associate professor of clinical pharmacy, to set up pharmacogenomics testing, and the test panel in that program includes two of the best-studied DPYD variants. In an interview about DPYD testing at UCSF last year, Tamraz said that since the launch of that program in May 2023, there have been 31 orders to test for the two DPYD variants as of January 2024.

By comparison, DPYD testing uptake has been more robust at Dana-Farber Cancer Institute, which is often held up as a model of how to implement pretreatment testing, as it includes a multidisciplinary advisory group, physician education, and clinical decision support that is set up to ensure doctors receive test results before initiating chemotherapy. Since launching the program in fall 2022, DPYD testing for eight variants has found 156 of 3,018 tested patients, or around 5 percent, to be abnormal fluoropyrimidine metabolizers. After 10 months of running the program, more than 90 percent of patients undergo DPYD testing before their first dose of 5-FU or Xeloda, compared to 26 percent when the program began.

During his 35-year career, Venook said he has treated thousands of patients with 5-FU and has had one patient die from treatment-related toxicities. "These are extremely uncommon events," he said in an interview last year. At the meeting, Venook said that he does discuss DPYD testing with his patients and often performs it, and while he knows what to do with the results, other oncologists might not.

Despite the FDA recommending that doctors discuss the risk of adverse events and DPYD testing with patients, agency officials at the meeting suspected this may not be happening widely. At Maryland Oncology Hematology, there have been patients with DPD deficiencies who have died from fluoropyrimidine toxicities, and Garg acknowledged at the meeting that DPYD testing should be discussed with patients.

But he also noted that at the community practice level, CPIC guidelines have been challenging to translate into clinical practice, especially for carriers of compound heterozygous DPYD variants who may not all experience severe toxicities to these chemotherapies. Moreover, patients react differently to the information on DPD-related toxicity risks. Even after becoming aware of the data on DPD, Garg has had patients who still want treatment fearing that cancer may remain in their body with a reduced dose.

Asal Sayas, who was diagnosed with metastatic colorectal cancer at age 35, recalled at the meeting how she screamed, "No!" when her oncologist suggested a dose reduction after learning she had experienced an infusion reaction that put her in a wheelchair. "I was so afraid to tell him about my reaction. I thought they would take the chemo away from me because I knew that cancer was going to kill me," she said.

"Dying from cancer is horrific," Merritt, the AUDT founder, acknowledged, but at the same time, she said her 73-year-old mother with an unknown DPD deficiency "would not have wanted to die from her cancer medication."

During the meeting, Dan Hertz, an associate professor of pharmacy at the University of Michigan College of Pharmacy, challenged the FDA's labeling language that there is insufficient evidence to recommend dosing changes for patients with partial deficiency. "We know from all the evidence we have that if you give a patient with partial deficiency standard doses in fluoropyrimidines, they've got a 70 percent risk of severe toxicity and a 3 percent risk of death," he said. Moreover, proponents of DPYD testing noted that CPIC guidelines recommend starting with a reduced dose in those with partial DPD deficiencies and then increasing it gradually while closely tracking toxicities.

Venook countered that this is hard to do at a busy oncology practice. "Practically, it's exceedingly difficult to reliably say we can dose escalate," he said. "We're under extreme pressure to have the chemotherapy orders ready to go. You can't have patients occupy the chair for hours waiting for the pharmacy to change the dose of the 5-FU."

These arguments against testing — that there's a lack of evidence that genotype-guided dosing won't sacrifice efficacy, that test turnaround times are too long, or that complete DPD deficiencies are rare — didn't sit well with Scott Kapoor, whose brother, Anil Kapoor, died from a 5-FU overdose due to a DPYD variant that wasn't included in testing. "Anybody who is opposed to this, they should go to the bedside, they should look and see what an overdose of 5-FU does to a patient," Kapoor said. "They should sit there and watch them. It's the same pattern over 20 days what it does to the body. It burns the body completely. You should first look at what it does to a patient [with an] overdose before you oppose this."

Grappling with uncertainty

Throughout the meeting, participants had a variety of recommendations for how to reduce uncertainty when it comes to DPYD testing. Merritt was optimistic that stronger test recommendations from the FDA would prod guidelines bodies to take a positive stance on testing as well as improve reimbursement and increase standardization of the variants tested by labs. Garg said he wants "codification" of actionable DPYD variants and the 5-FU and Xeloda dose reductions that go with it.

If oncologists are waiting for a definitive, prospective randomized-controlled trial to ease their worries, that's not feasible, said University of Michigan's Hertz, who is leading an effort to mollify their concerns with real-world data instead. So far, 20 centers conducting DPYD testing are submitting data within the real-world project, called the Oncology Pharmacogenetics Real-world Evidence Consortium, but anyone doing this testing can participate. The consortium's aim is to more precisely answer questions around the impact of DPYD-informed dosing on toxicities and efficacy.

Hertz is also starting a separate observational study comparing drug concentrations in those with and without DPYD variants. If the drug concentrations are the same in those who received genotype-informed dose reductions and those who got standard doses, then that would suggest that dose adjustments were properly calibrated. This is the best way to demonstrate whether dosing guidelines are appropriate, Hertz said, since efficacy data will be very challenging to collect.

Michael Pacanowski, director of the division of translational and precision medicine at the FDA's Office of Clinical Pharmacology, said that present unknowns about DPYD test performance could be alleviated if labs brought their tests through FDA review. "There is a lot of variability that remains. As a consumer and a citizen, I would expect to have a [test] sent out and have the same result regardless of where I happen to present," Pacanowski said.

This is not an expectation patients can have right now, given the intensely competitive genetic testing market and the state of test regulation. As of 2020, there were more than 170,000 marketed genetic tests, and most of them are performed at labs and aren't regulated by the FDA. When the agency acted last year to extend its oversight over lab tests, pathology and lab industry groups sued, challenging its authority to regulate the industry. At the meeting on DPYD testing, agency officials said they couldn't talk about lab test regulation because of the lawsuits but urged industry to bring tests in for review.

While FDA regulation wouldn't remove variability in test performance, it might make the differences between marketed tests more transparent through labeling. For example, the FDA authorized 23andMe's direct-to-consumer BRCA1/2 test initially in 2018 to detect three variants associated with increased cancer risk that show up most often in those of Ashkenazi Jewish descent and then expanded the test's indication to 41 variants in 2023.

Like the DPYD gene, BRCA1 and BRCA2 are highly variable genes, with more than 4,000 variants known to increase cancer risk. There are tests on the market, some with FDA authorization and others without, that do a far more comprehensive job of detecting them than the FDA-approved 23andMe test. The FDA would counter that the strengths and weaknesses of 23andMe's test are known in labeling, though those against FDA regulation would argue that including this information in FDA labeling doesn't necessarily raise a physician's or patient's understanding of appropriate testing.

Moreover, variability among labs doesn't stop with the variants tests can detect, since there may also be disagreement on a variant's classification: whether it increases risk, is benign, or if its risks are unknown. Despite these unknowns, NCCN over the years has recommended testing BRCA1/2 and other hereditary cancer risk genes for broader swathes of the population based on cancer type and family history. And cancer centers have integrated testing, genetic counseling, and other procedures to identify patients who need genetic testing to gauge their risk for hereditary cancers.

Paul Kluetz, deputy director of the FDA's Oncology Center of Excellence (OCE), said the agency should have more meetings like this with NCCN and ASCO to raise awareness of DPD deficiency and what the FDA says about DPYD testing in fluoropyrimidine drug labeling.

Drug "labeling is a living document and is flexible and can change over time," Richard Pazdur, the founding director of the OCE, said at the start of the meeting, a sentiment that was echoed by other FDA officials throughout the day's discussions.

The present meeting closed, however, without any definitive commitments from the FDA to strengthen the labeling language on pretreatment DPYD testing. It's unclear if such an update would sway oncologists.

The FDA's Pacanowski pointed out that the FDA requires pretreatment genetic testing for very few drugs. In most cases, the agency gathers data on test performance from clinical trials of treatments before they hit the market. When the agency has updated labels with testing recommendations for drugs that have been on the market for a while, for example, for Bristol Myers Squibb and Sanofi Aventis' antiplatelet drug Plavix (clopidogrel) or the anticoagulant warfarin, it was controversial, with guidelines bodies not recommending routine testing and cardiologists not adopting testing.

In the case of Plavix, the agency included a black box warning about the availability of CYP2C19 testing to identify poor metabolizers of treatment and reviewed and cleared CYP2C19 tests. Still, "the cardiology community did not know what to do with that information," Pacanowski said. "It was a very complicated situation."

The Plavix experience begs the question: Even if the FDA updates the labels of 5-FU and Xeloda strongly urging DPYD testing and grants market authorization to DPYD tests, will oncologists adopt it? As the FDA pushes ahead with efforts to regulate lab developed tests, the lab community often reminds the agency that tests they perform constitute laboratory medicine and that "the FDA doesn't regulate the practice of medicine." 

One way to get around the testing controversy is to advance safer drugs. Sayas, who is still on 5-FU five years after her metastatic colorectal cancer diagnosis and has experienced a slew of toxicities, said, "I cannot wait for the day that we no longer have to use 5-FU because we have better drugs approved by the FDA."