Science

How AI Can Help Treat (and Cure) Your Dog’s Cancer

GOOD BOY!

A tech startup wants to use AI to help veterinarians give targeted drugs for a dog’s cancer treatment—but is it a helpful tool or are they barking up the wrong tree?

An illustration including a photo of a dog and code
Photo Illustration by The Daily Beast / Getty

When Cory Padilla felt swollen lymph nodes in the neck of his dog, Copper, he knew it was time to take his best friend to the vet. Padilla told The Daily Beast that Copper, a beagle and German shepherd mix, was also experiencing bouts of diarrhea and his eyes looked “droopy”—but cancer wasn’t on the pet owner's radar.

“I thought it could be another disease that hounds get,” Padilla said. But the veterinarian confirmed the pet parent’s worst fears after aspirating Copper’s swollen lymph nodes and diagnosing him with lymphoma, one of the most common cancers in dogs.

“My heart hit the floor,” Padilla said. He was right to be worried: If left untreated, canine lymphoma kills most dogs within three to four weeks.

ADVERTISEMENT

Padilla immediately went online to learn all that he could about the disease and treatment options. That’s how he discovered ImpriMed, a company launched in 2017 that uses artificial intelligence and comprehensive analysis of live cancer cells to make personalized drug response AI-based predictions for dogs with lymphoma.

A small brown and black dog smiles at the camera.

Cory Padilla's dog, Copper.

Courtesy of Cory Padilla

Once ImpriMed receives the live cancer cells at their lab in California, they can conduct flow cytometry testing, a technique that helps veterinarians differentiate lymphoma subtypes, cell analysis, as well as drug sensitivity testing, which ranks 13 anticancer drugs that are commonly used for treating lymphoma and leukemia.

“This approach is not just about choosing the right treatment,” ImpriMed CEO and co-founder Sungwon Lim told The Daily Beast. “It’s also about avoiding spending on treatments that may not be effective for your pet’s specific cancer. In the long run, this can potentially save both time and money, reducing the trial-and-error process often associated with cancer treatment.”

Lim explains that ImpriMed’s predictions are made using proprietary AI that rely on the outcome of treatment effectiveness collected from patients across the country, as well as data pulled from the live cancer cells to give clients and veterinarians personalized drug-response predictions for their dogs.

“I thought that it was amazing that AI took [the cancer] on a cellular level and told you in a panel what chemo treatments were the best or most effective for your individual dog,” Padilla said.

“I Thought It Was Going to Kill Her.”

Canine lymphoma is traditionally treated with chemotherapy using a protocol called CHOP that involves a combination of four drugs: doxorubicin, cyclophosphamide, vincristine, and prednisone.

“We know that 90 to 95 percent of dogs will respond to CHOP and that’s been proven unequivocally in the literature that CHOP should be the gold standard of treatment,” Jack O’Day, a veterinary oncologist based in Fairfax, Virginia, who uses ImpriMed for precision treatment in his practice, told The Daily Beast.

However, it’s not perfect. CHOP doesn't take into account the uniqueness of each patient—or their owner’s wallet. The treatment ranges between $10,000 to $25,000 and not all dogs respond to the protocol in the same way. “I wanted to do CHOP, but we were talking $500 to $600 on average per week,” Padilla said. “I don’t have that kind of money.”

That’s what made ImpriMed’s service so appealing. It was able to provide Padilla insights with Copper’s lymphoma type and ranked drug options in order of effectiveness.

“If you know what is the most effective then you can choose to make decisions based on that instead of throwing darts in the dark,” Padilla said, adding he was able to save both time and money in treating Copper.

While it shows promise, ImpriMed doesn’t come without a cost of its own. The average ImpriMed Prediction Profile, the portion of the testing that ranks anticancer drugs, runs pet owners between $1,000 and $1,500. That can be prohibitively expensive for many dog owners.

Lim said that their service is offered to vet clinics “at a set price.” But the final cost is ultimately determined by each individual clinic.

Susie Araiza, another pet owner in Texas, told The Daily Beast that she didn’t find the ImpriMed Prediction Profile particularly helpful in navigating treatment options after her dog Precious experienced a lymphoma relapse. For example, the first drug recommended by ImpriMed made Precious anemic enough to require an emergency iron injection from her veterinarian.

A small black and white dog lays in a dog bed

Susie Araiza's dog Precious.

Courtesy of Susie Araiza

“We took her off of it and she was back to normal,” Araiza said. “So to me, even though I appreciated (the panel results), I thought it was strange that the very first drug…she couldn’t take it.”

Araiza said the other top drug recommended on Precious’ ImpriMed panel—doxorubicin—put her dog in the hospital for over a week. In the end, she said she would have rather put the money she spent on the panel toward treatment.

“I thought it was going to kill her,” Araiza recalled. “The oncologists said ‘We’re probably not ever going to give her that again.’”

The Rise of VetGPT

Araiza’s experience, and that of other pet owners who have detailed their experiences in social media groups dedicated to canine lymphoma, begs the question: Can this panel be effective without seeing the full clinical picture of an animal?

Veterinary oncologist Daniela Korec told The Daily Beast that she primarily uses ImpriMed in her practice for flow cytometry—an advanced type of cell analysis—not predictive drug panels. “I see what comes back [on the Prediction Profiles] and I think it's important to point out that there are still a lot of predictive abilities with lymphoma from our knowledge of testing these drugs so often,” she said.

Korec adds that while none of her patients have done Prediction Profiles with ImpriMed, she is supportive of the technology with the caveat that owners also consider her expertise as a clinician who is treating the animal—and not just the information printed on the report.

A black labrador sits in a chair next to a female veterinarian.

Veterinary oncologist Daniela Korec.

Courtesy of Daniela Korec

Parminder Basran, associate research professor and medical physicist at Cornell University’s College of Veterinary Medicine, also cautions against relying solely on AI for the trajectory of patient care—especially when it comes to broad topics like cancer.

“These are the tough questions that medical veterinary experts are best equipped to handle, not an AI system,” Basran, whose research lab explores the use of AI in veterinary healthcare, told The Daily Beast.

Basran calls the use of AI models in veterinary medicine “exciting and certainly very interesting,” but admits there is “a lot of work to do” when it comes to trusting AI systems in both veterinary and human medicine.

While pet owners, veterinary oncologists, and AI experts in the medical field have varied confidence in ImpriMed’s predictions, Lim stands by his product—though he acknowledges the challenges that can still arise when treating the animals.

“Unfortunately, some dogs are resistant to all available drugs so the drug that we predict to have the highest likelihood of success will still not elicit a positive response,” he said. “Furthermore, like all medical tests and AI predictions, our drug response predictions are sometimes incorrect.”

But Lim believes the only way to push modern medicine forward is to continually test the limits, which is why the ImpriMed team uses real-world statistics to determine if their product can provide clinical benefits in everyday use and utilizes post-market clinical research for studies.

In December 2023, the company released a study using combined data from 60 dogs independently treated across the country, which resulted in threefold longer patient survival and fourfold higher complete remission rates in patients treated in high concordance with ImpriMed’s AI predictions. It’s news that excites Lim.

For the first time, pet owners and veterinarians have a tool that could potentially target an animal’s specific cancer with the creation of an AI-predicted panel of effective drugs—but there is work to be done when it comes to both accessibility and accuracy.

While alluring, for many pet owners, treating their dog’s disease, not to mention splurging on predictive AI technology, will never be a reality because cancer—and the research that advances curative treatments—comes at a cost. But still, when it comes to our pets, there’s nothing many of us won’t do to help them.

Lim added, “Our goal is to support veterinary oncologists and their clients with the best possible tools for fighting cancer in their beloved pets.”

Got a tip? Send it to The Daily Beast here.