Innovation Insights: Calviri’s cancer diagnostic, Zoetis Imagyst expansion, Idexx, Antech, Urine-sniffing dogs, and High Agency
Periodically I’ll share some of the latest developments in animal health including emerging startup companies, noteworthy innovations, trends, and impactful scientific publications. You’ll also get occasional book or podcast recommendations.
I value your input- if you have a newsworthy item or resource you think should be featured, please share it! Let’s keep our finger on the pulse of innovation in animal health together!
Zoetis’ Imagyst expands its capabilities to AI cytology
Computer Vision models are popping up everywhere in veterinary diagnostics. It was only a few years ago that cytology went digital in veterinary medicine. Fueled by innovative folks like Lacuna Diagnostics, point of care application launched simultaneously, with small affordable slide scanners installed in hospitals allowing rapid cytology interpretation from remote clinical pathologists in under two hours. The next step in the journey was predictable: computer vision AI model evaluation of cytology slides from point of care scanners. Zoetis has been moving this direction for some time with the Vetscan Imagyst analyzer.
Imagyst is a basically a set of cloud-based software applications that provide AI-powered analysis built around a simple Grundium ocus digital slide scanner. Till now, those applications were limited to things like AI fecal, AI blood smear, AI urine sediment, AI dermatology. Cytology slides made from lymph nodes and masses were still scanned to a pathologist for review.
Zoetis just announced the upcoming launch of AI Masses on June 2, 2025, a model that also evaluates lymph node aspirates. Similar functionality is forthcoming from the Idexx Invue Dx instrument, but a timeline has not been announced.
In a time when tech is changing this rapidly it’s easy to get numb to these kinds of advancements. But this is a big step for veterinary medicine, with potential to change not only clinical medicine but the field of pathology. Will these applications replace clinical pathologists or will we just need fewer of them? What does the future of pathology and radiology look like in light of the many companies bringing innovation to bear? There are many questions to be answered and much to evaluate regarding this application, but there’s no doubt that this is an innovative and important advancement.
A novel cancer screening approach from Calviri detects Stage 1 cancer
A number of companies in human and animal health are tackling the challenge of multicancer early detection testing, often referred to as a “liquid biopsy”. Approaches vary, with many assays looking at circulating tumor DNA (ctDNA), nucleosome quantification, VOCs, or cell-free DNA (cfDNA) fragment analysis. While these tests tend to have relatively high specificity, sensitivity at the early stage, (exactly when you’d want to detect cancer) is often quite poor.
Calviri takes a novel approach, both in their cancer vaccine and in the diagnostic outlined in this study. Rather than focusing on detection of tumor-derived material, their method targets the immune response to neoantigens that arise from RNA processing errors, translation mistakes, and protein misfolding. These REDNS (RNA-error derived neoantigens), are hypothesized to be shared across tumor types and highly immunogenic, offering a better signal for early detection than low-abundance tumor DNA.
Calviri’s thesis is that these antibodies serve as biologically amplified indicators of cancer, especially at its earliest stages, and that this immune signal can be more reliably detected than the direct products of a tumor. This approach also only needs a very small amount serum that could be run on leftover serum from routine bloodwork, as compared to sequencing-based tests which require uncomfortably large blood volumes (10+ mls).
In this study, Calviri created a high-density microarray containing 125,000 peptides and screened sera from 283 dogs (including stage 1 cancers and verified-cancer-free controls) to detect antibody binding signatures indicative of cancer. They then developed 2 multiclass models. The simple count-based model had a sensitivity of 68-98% and the neural network model had a sensitivity of 60-88%. Both achieved high specificity (92-99%). Below is the performance for the simple count-based model. (HAS= hemangiosarcoma, LYM= Lymphoma, MCT= Mast Cell Tumor, OSA= Osteosarcoma, STS= Soft Tissue Sarcoma).
Urine-sniffing dogs and worms, living together, mass hysteria
Early cancer detection is being tackled by many companies in human and animal health, using a variety of methodologies. Sequencing of circulating tumor DNA (ctDNA), cell-free DNA quantification, nucleosome quantification, Fragment End Analysis, Mysterious Idexx Test, and many other approaches are being utilized to bring accurate, affordable tests to market. One of the more fascinating and off-the-wall approaches is by detecting Volatile-Organic-Compounds (VOCs). VOCs are organic chemicals that readily evaporate into the air at normal temperature and pressure, and are found in many substances. They are being evaluated as promising biomarkers for a number of condition in medicine. Weirdly, so far the approaches to detecting of VOCs in cancer have utilized worms racing across petri dishes (Oncotect) and urine-sniffing dogs. In this UK-initiative, a team of highly trained dogs have been purported to be able to detect the presence of cancer in a person from sniffing their urine, in just 10 seconds. Dogs have up to 300 million olfactory receptors in their noses, compared to just 6 million in humans. Is the future of early cancer detection a lab full of worms and Doodles huffing urine? I don’t think so. Even worms have down days (I think, right?). And it just feels weird to call the lab for results and ask “What are the dogs saying today? I mean, can I have my results, please?”. But the underlying and important concept here is that VOC detection holds potential for cancer detection. And it turns out there are some companies working on doing this in an objective way. More on that in future issues!
The first autologous prescription product to receive USDA-CVB approval
Elias Cancer Immunotherapy has received approval from the USDA-CVB for the treatment of canine osteosarcoma, the first product of its kind to achieve this regulatory hurdle. The decision to use the therapeutic requires cancer tissue sampling prior to surgery to develop this personalized immunotherapy that stimulates the patient’s immune system to kill cancer cells. Veterinary Medicine needs more tools for treating cancer in dogs. Kudos to Elias for this achievement.
AI scribes are improving physician-patient communication
AI scribes are popping up everywhere in human and vet med. For good reason. If I was back in practice I’d love to have a tool that would take the laborious and mind-numbing task of documentation off my hands. While there is still much to be learned about the use, accuracy, and advantage of these tools, early papers like this one show overall positive effects with improved physician-patient communication and reduced documentation burden. Notable takeaways from this evaluation of 10,000 physicians across 303,266 encounters:
-A majority of physicians reported a highly positive experience with AI scribes, with all patients reporting a positive to neutral impact of AI scribe usage on the quality of their visit.
-The biggest barrier to AI scribe use was no direct integration with documentation software (sounds familiar, huh?)
-AI scribe use most prominent in mental health (50% of visits), emergency medicine (36% of visits), and primary care (34% of visits)
Idexx adds critical capabilities to 75k analyzers- Cortisol testing in the clinic
Cortisol testing is used to identify dogs with hypoadrenocorticism (Addison’s) and hyperadrenocorticism (Cushing’s). Historically, cortisol testing has not been available on in-clinic analyzers. Getting quantitative results required sending out blood for testing. This is inconvenient, but okay for dogs with Cushing’s disease as the goal in testing is usually long-term management and not an emergency. However, dogs with Addison’s disease can present with an acute crisis, with collapse and shock secondary to electrolyte imbalance and metabolic acidosis. This crisis can resemble many other severe diseases, lending Addison’s the nickname of “The Great Pretender”. Idexx’s addition of a cortisol test to the Catalyst One analyzer gives clinicians a new tool for emergency cases as well as long term management of Cushing’s patients.
Antech launches a heartworm and vector-borne disease test- No thumbs required!
Dogs get heartworms and vector-borne diseases (Lyme, Ehrlichia sp., Anaplasma sp.). We’ve been testing for them for a long time. I well remember waiting for those little blue dots to appear in the window of the Idexx snap tests back in my days as a mixed animal vet. Idexx offers the Snap 4Dx Plus, a lateral flow snap test. Zoetis has the VetScan FLEX4 lateral flow test. Antech now joins the fray with their own truRapid FOUR, a lateral flow test that is stable at room temperature with no snapping required. I always took pride in my snapping abilities. The big companies are just talking all the fun out of stuff these days…….
VET-DINO: Self-supervised AI model learns anatomy from multiple-radiograph views
In veterinary radiology there is much criticism right now regarding the lack of transparency and data about the various AI applications available commercially. In this paper, the Mars Next Generation Technologies team discuss VET-DINO, a self-supervised approach to veterinary radiology that leverages multi-view veterinary studies to better understand anatomical structures, developing an understanding of 3D structures from 2D projections. Trained on 5 million radiographs from 668k dog studies, the multi-view VET-DINO performed better than single-view VET-DINO, and untuned DINOv2. Here are some of the team’s other notable publications:
StudyFormer : Attention-Based and Dynamic Multi View Classifier for X-ray images
Who Goes First? Influences of Human-AI Workflow on Decision Making in Clinical Imaging
Using AI on biopsy images to distinguish between feline intestinal lymphoma and IBD
Distinguishing between early intestinal lymphoma and IBD in feline intestinal biopsies is a long-standing challenge for pathologists. The core of the issue is that early lymphoma can look very similar to Inflammatory Bowel Disease (IBD), with subtle differences in subjective features such as numbers of lamina propria lymphocytes and intraepithelial lymphocytes.
In this study, an AI model was developed to quantify intraepithelial and lamina propria lymphocytes in H&E-stained intestinal biopsies from cats in order to provide a reproducible and objective assessment of feline intestinal lymphocytes. This is a great way to objectively evaluate something that we humans cannot, and the authors intended to provide the pathologist with an AI-based tool rather than replace the pathologist with a black-box model. The application here is very focused and niche, not considering other intestinal disease or lymphoma types. Overall, the model with pathologist oversight succeeded in providing a more objective and reproducible count of lymphocytes, particularly in cases where pathologists alone had low interobserver agreement.
Recommended
“High agency is such an important idea because the more agency an individual or society has — the more problems they can solve.”
This is a great read. It will cause you to evaluate your life, career, network, and current trajectory. Very motivational. A good wakeup call. Thanks for the reference, Aaron Massecar!
*****Please note that I am not affiliated with any of the companies mentioned in this newsletter, nor do I receive any financial benefit from endorsing them. My interest is purely based on their innovative contributions to the industry.*****
