In 2024, an estimated 6 million pet dogs will be diagnosed with cancer in the United States. As someone who has lost several family dogs to cancer, I know how heartbreaking this diagnosis will be for each of those dogs’ owners.
But I also know that these dogs hold the key to accelerating the development of precision treatments that will save the lives of countless cancer patients, human and canine.
The striking concordance between human and canine cancers—from genetics to incidence rates—means that we can leverage the genomic data from the millions of dogs who get cancer every year to accelerate the development of precision treatments that will save the lives of dogs and people.
The idea of using canine cancers as models for human cancers is known as comparative oncology and has already been used to develop several breakthrough cancer drugs. But to tap into the full potential of comparative oncology we need to study canine cancers at scale, which means bringing precision medicine tools out of the research lab and to the point of care in the thousands of veterinary clinics across the United States.
Consider this: If we were to sequence the tumors of just 1% of the pet dogs who are diagnosed with cancer next year, we would increase the total number of canine tumors ever sequenced more than 10-fold.
This alone is exciting given that dozens of precision cancer drugs, including ibrutinib, acalabrutinib, and selinxor, have already been developed using relatively small and context-limited canine cancer datasets from university veterinary schools.
If we massively increase the quantity and quality of canine cancer data through clinical sequencing we can expect it to play an important role in bringing new precision cancer drugs to market, even if nothing about the drug development process changes.
But the truly great opportunity in building clinico-genomic datasets from real-world cancer patients only emerges at a scale that is impossible in a solely laboratory-driven comparative oncology paradigm.
The plummeting costs of genomic sequencing means that it is now possible to sequence tumors from the millions of canine cancer patients seen by America’s roughly 30,000 veterinarians each year.
If we sequence just 1% of the tumors of all the pet dogs that are diagnosed with cancer next year, there will be more than twice as many canine tumor sequences as human sequences.
The billions of resulting data points from this effort can be analyzed by artificial intelligence to help match canine cancer patients to precision treatments that are most likely to benefit them given their breed, age, and other characteristics. Moreover, this same data can be used to identify new indications for existing precision cancer treatments that may also benefit humans and can expedite the clinical trial process.
In other words, we now have the technology and foundational scientific knowledge to launch a continuous and distributed clinical trial for canine patients across a wide variety of cancers.
In addition to enrolling pet dogs in site-specific limited cohort trials at research labs, we can collect more comprehensive data from millions of dogs that visit the veterinary clinic each year and dramatically accelerate the development of new precision treatment that will benefit both dogs and their owners.
This isn’t just a nice idea—we already have evidence showing it can work.
Over the past few years, my colleagues and I have collaborated with researchers from several leading institutions including Stanford AI Health, Tufts University, the Broad Institute of MIT and Harvard, and the University of Georgia on studies that have borne out this theory in practice.
These studies drew from data collected from pet dogs whose cancer was diagnosed and treated at one of our 1,350 partner veterinary clinics. Last year, we published the results from several of those studies, which have resulted in the largest clinico-genomic dataset of canine cancer data from real world patients.
Our results show that this approach can in fact significantly increase the survival of canine cancer patients by matching them with precision treatments, that canine cancer data can identify promising new indications of existing cancer drugs for humans, and that dog and human cancers are even more genetically similar than previously known.
Taken together, the results make it clear that we’ve barely begun to tap the potential of this new approach to comparative oncology. Millions of human and animal lives are at stake and we can’t afford to let this opportunity go to waste.
The role of dogs in the development of precision cancer treatments
Companion canine cancer patients are about as perfect an animal model for human cancer as we could hope for.
They are at least 85% genetically similar to humans, they have a similar microbiome and immune system, they spontaneously develop similar types of cancers, they have similar metastatic profiles, their cancer develops more rapidly than humans allowing for easier study, and the incidence of cancer is estimated to be about 10 times higher than in humans.
They also happen to be the only large animal where cancer treatments are considered routine, which creates an unprecedented opportunity for us to study this deadly disease.
The research community has been aware that companion canine cancer patients are a useful model for human cancer for decades, but comparative oncology has only risen to prominence since the turn of the new millennium as the true extent of this overlap came into focus.
This was driven in large part by new tools that revealed the genetic, immunological, and metabolic aspects of human cancers, which provided fertile ground for exploring the depth of the link between human and canine cancers.
In 2003, the National Cancer Institute launched the Canine Comparative Oncology Genomics Consortium, which built a first-of-its-kind biorepository of canine cancer tissues to kickstart the genetic exploration of canine cancers.
Since then, the research institutes participating in the consortium have revealed the deep genetic similarities between canine and human cancers, which has cemented the role of companion animal cancer models in the age of precision cancer treatments.
This year marked the 20th anniversary of the NCI’s consortium, which has been by virtually every measure a resounding success. But it has also underscored the urgent need for dramatically more genomic data from companion canine cancer patients to deliver on the full potential of comparative oncology.
As an example, earlier this year, my colleagues and research collaborators at One Health, the Broad Institute of MIT and Harvard, and the University of Georgia published a paper in Scientific Reports detailing the results from the largest-ever genomic sequencing study of canine tumors.
The study collected clinico-genomic data from 671 canine cancer patients at 200 veterinary clinics from May 2019 until September 2020. One of the major results of that study was the revelation that canine and human cancers were even more genetically similar than previously known following the identification of multiple previously unknown mutational hotspots shared between human and canine cancers.
What is so remarkable about this study is that it shows the incredible potential of enlisting veterinarians to run distributed continuous comparative oncology clinical trials and collecting genomic data from canine cancer patients at massive scale.
To put these results in perspective, the NCI’s consortium has collected data from a total of 738 dogs across 17 clinical trials during the past 20 years. By bringing sequencing out of the lab and into the clinic, our research team was able to collect genomic data from 90% as many dogs more than 10 times faster and reveal important new insights about human and canine cancers.
At the time our study concluded in September 2020, only around 2,000 canine tumors had ever been genomically sequenced. That study alone increased the total number of sequenced canine tumors by 33%. Since the study ended, we have sequenced the tumors of more than 4,000 pet dogs in total. In other words, in less than 4 years we have managed to double the total number of canine tumors that have ever been sequenced.
The NCI’s program has been instrumental in bringing comparative oncology to the forefront of modern cancer research and the data from these canine patients continues to help accelerate numerous precision small molecule treatments through the clinical trial process.
Yet the ability of the consortium’s participants to collect more data is subject to many of the same restraints as human clinical trials such as budgetary constraints and enrollment challenges.
We can overcome these challenges and accelerate the development of new precision cancer treatments that depend on canine data by meeting canine cancer patients where they are: in the veterinary clinic.
It’s a dog’s world
Today, there are roughly 25,000 tissue samples from human tumors that have been genetically sequenced and made available to the research community.
This is about 6 times more than the total number of canine tumors that have been sequenced, despite the fact that it is far more complex and expensive to obtain the sequencing data from a human tumor.
If we sequence just 1% of the tumors of all the pet dogs that are diagnosed with cancer next year, there will be more than twice as many canine tumor sequences as human sequences. But what would we do with all that data?
Earlier this year, my colleagues and research collaborators at One Health and Stanford AI Health published a paper in Nature Precision Oncology that provides a glimpse of the enormous potential in creating unprecedentedly large canine cancer datasets through distributed continuous clinical trials at veterinary offices.
In that study, we examined the outcomes of 2,119 companion canine cancer patients and the prognostic effects of genomic alterations in a subset of 1,108 of those pets. What we found was astounding.
First, the data identified several key oncogenes shared by humans and dogs—including TP53 and PIK3CA—that can reliably predict a canine cancer patient’s prognosis based on its genetic mutation profile and small molecule treatment.
Second, and most exciting, was the discovery that several small molecule drugs developed for humans resulted in dramatically improved—and statistically significant—survival times in canine patients with cancers that were different from the drug’s original indication.
In other words, by using artificial intelligence to analyze large scale genomic datasets from pet dogs with cancer, we were able to identify new lifesaving applications for existing cancer drugs.
Our results show that this approach can in fact significantly increase the survival of canine cancer patients by matching them with precision treatments, that canine cancer data can identify promising new indications of existing cancer drugs for humans, and that dog and human cancers are even more genetically similar than previously known.
The reason this is possible with pet dogs is because veterinarians, unlike human doctors, are given wide latitude in the types of treatments they can prescribe to canine cancer patients.
There is no established ‘standard-of-care’ for many types of canine cancer and these patients are often treated based on data from published literature and clinician preference.
Moreover, clinical trials in canine cancer patients aren’t constrained by conventional phased clinical trial designs, which means that a veterinarian can offer investigational new drugs to pet dogs if the data looks promising for treating the dog’s cancer.
But this only works if the veterinarian has genetic data on their canine patient’s cancer and is aware that a precision treatment exists and may prove beneficial to a pet dog with a particular genetic and phenotypic profile.
And now, for the first time, they do.
By collecting genomic data from the millions of canine cancer patients seen by tens of thousands of veterinary offices in the United States and around the world, we have created the world’s largest clinico-genomic canine cancer dataset that veterinarians can use to identify recommended treatments for their canine cancer patients based on the results from a rapidly growing group of other pet dogs whose genetic tumor profile and outcomes have been recorded in the database.
This is good news for canine cancer patients, who can now have the option to receive precision cancer treatments that their veterinarians didn’t know existed or didn’t realize had additional indications.
The study we published in Nature this year showed how this can meaningfully increase the survival times for pet dogs, but it can also reduce the time to market for new precision cancer treatments for humans, too.
The dismal stats around the development of new cancer drugs are well known. We constantly hear how 90% of of new cancer drugs never progress from mice to market and the ones that do frequently take over a decade and cost north of $1 billion to develop.
The primary reason for this is due to the slow process of identifying promising new drugs to bring to a clinical trial, which is followed by the considerable expense and difficulty of recruiting patients to enroll in the trial. 92% of cancer patients never enroll in a clinical trial and the data from the ones that do is extremely difficult to use due to restrictions around the sharing of patient data from laws like HIPAA.
As we saw in our study of more than 2,000 dogs, we can help overcome many of these challenges in the drug development process by leveraging clinico-genomic data collected by veterinarians in the course of caring for their canine cancer patients.
When we collect this data at scale, we can use artificial intelligence to quickly identify the most promising and unknown new indications for new and existing cancer drugs, which drug developers can use to fast track those new indications to human clinical trials.
There have been more than a dozen precision cancer treatments brought to clinical trial—and eventually to market—over the past decade on the strength of data from canine patients in clinical trials run by the NCI and others. And now, drug developers are already using data collected from this new approach to safely accelerate the drug development process.
The global pharmaceutical company Eisai, for example, has developed a precision cancer drug called Eribulin, which was originally developed to treat breast cancer and liposarcoma. But canine cancer data from One Health’s platform confirmed that it may be effective in treating Angiosarcoma and Epithelioid Hemangioendothelioma as well, and now the drug is being studied in a Phase II clinical trial at Massachusetts General Hospital to expand its indication for these two cancers.
Our experience building the largest dataset of clinico-genomic canine cancer data over the past few years has revealed the enormous potential in enlisting our pet dogs in the fight against cancer.
The plummeting cost of genomic sequencing combined with the rapidly increasing sophistication of artificial intelligence has finally enabled us to collect and analyze genomic canine cancer data at a pace that would have been unthinkable only a few years ago.
Each life lost to cancer is a tragedy and we—as a field—have the opportunity and obligation to use these new tools to accelerate the development of life saving drugs that will save countless lives on both ends of the leash.
The author is CEO and co-founder of The One Health Company, the leader in precision medicine for dogs with cancer. She has co-authored several papers on comparative oncology in journals including Nature Precision Medicine, Scientific Advances, and Veterinary and Comparative Oncology. Christina holds a patent for the treatment of canine cancers by targeting tumor mutations and has been recognized as a Young Global Leader by the World Economic Forum.
