ImpriMed AI Chemotherapy Shows Longer Disease Control in Canine Lymphoma

ImpriMed Reports Longer Disease Control and Lower Daily Treatment Costs With AI-Guided Chemotherapy Selection in Dogs With Relapsed Lymphoma

ImpriMed, a precision veterinary-medicine company using artificial intelligence (AI) and live-cell drug-sensitivity testing, has reported new retrospective findings suggesting that closer alignment between AI-predicted and administered chemotherapy treatments is associated with longer disease control and lower estimated treatment costs per day in dogs with relapsed B-cell lymphoma.

Presented at the 2026 Veterinary Cancer Society (VCS) Annual Conference and announced on September 29, 2026, the analysis examined data from 60 dogs whose treatment outcomes had been evaluated in an earlier clinical study. Dogs in the higher treatment-prediction concordance group had a median progression-free survival (PFS) of 49 days, compared with 21 days in the lower-concordance group.

The economic analysis estimated treatment costs of US$117.09 per progression-free day in the higher-concordance group, compared with US$188.53 in the lower-concordance group—a difference of approximately 38%.

The findings offer an important clinical and commercial signal for precision veterinary oncology. However, the retrospective design, small cohort and limited cost accounting mean that the results should be interpreted as an association rather than proof that AI-guided treatment independently improves outcomes or reduces total treatment expenditure.

Key Findings: Clinical Outcomes and Treatment Economics

The reported PFS difference was statistically significant (p = 0.0028). The daily treatment-cost difference, however, did not reach conventional statistical significance (p = 0.075). This distinction is important when interpreting the economic findings.

The higher-concordance group also received fewer chemotherapy drugs despite having longer follow-up. The result is consistent with the hypothesis that better-informed treatment selection may help avoid less effective drug choices.

An important economic qualification: Total estimated treatment costs were higher in the higher-concordance group because treatment continued for longer. The reported advantage concerns the estimated cost per day without progression—not lower total treatment spending.

How ImpriMed’s Technology Works

ImpriMed’s Personalized Prediction Profile (PPP) combines laboratory testing of a patient’s live cancer cells with computational prediction models. The workflow involves:

1. Sample collection: A fine-needle aspirate is collected from an affected lymph node.

2. Ex vivo drug-sensitivity testing: Live cancer cells are exposed to selected chemotherapy drugs under laboratory conditions.

3. Tumour profiling: Laboratory and clinical information is combined to characterise the patient’s cancer.

4. AI-based prediction: Machine-learning models estimate the likelihood of clinical response to different drugs.

5. Clinical decision support: The resulting report supports veterinary oncologists in selecting treatment, alongside clinical assessment, patient factors and owner preferences.

ImpriMed’s current commercial offering includes individual-drug response predictions and separate prediction services for first-line CHOP-based treatment in eligible patients. The company’s website reports more than 300 veterinary oncologists ordering its services, over 250 collaborating hospitals and more than 15,000 canine lymphoma tests ordered and reported. These are company-reported operational metrics, not independently audited market-share figures.

Scientific Evidence: Building on the 2024 Clinical Study

The new economic analysis draws on a peer-reviewed clinical study published in Frontiers in Oncology in February 2024.

That study evaluated machine-learning predictions for 10 commonly used chemotherapy drugs in dogs with relapsed B-cell lymphoma. It reported that patients receiving drugs with stronger alignment to the prediction models experienced better clinical outcomes across several measures.

The response-rate comparison for complete remission was statistically significant (p = 0.002), whereas the overall response-rate comparison was not (p = 0.12). The duration-of-complete-response comparison was also not statistically significant (p = 0.10) and was based on a limited number of complete responders in the lower-matching group.

The 2024 study was an open-cohort clinical evaluation, not a randomised controlled trial. The new cost analysis is a retrospective reanalysis of this earlier clinical dataset and should not be described as a separate prospective validation of the technology.

Economic Interpretation and Study Limitations

The cost analysis included chemotherapy drug costs and the PPP report, using the price list of one confidential metropolitan referral hospital. Costs were standardised to a 30 kg dog to reduce variation from body weight.

The analysis excluded veterinary visits, diagnostic tests, examinations and other associated costs. Consequently, it does not provide a complete estimate of the total economic burden of lymphoma treatment.

The cost-per-day metric is also inherently influenced by PFS. A patient who remains progression-free longer can have a lower average daily cost even if the total treatment bill is higher. Key limitations include:

• A retrospective, non-randomised comparison.

• A relatively small cohort of 60 dogs.

• Potential differences in patient characteristics, treatment protocols and clinical decision-making between concordance groups.

• Cost estimates derived from one hospital’s price list.

• Exclusion of substantial components of real-world veterinary expenditure.

• No demonstration that the observed economic association generalises across hospitals, treatment settings or lymphoma subtypes.

Further prospective, multi-centre research with broader economic accounting would help establish the reproducibility and practical cost-effectiveness of AI-guided treatment selection.

Strategic Implications for Veterinary Oncology

The findings reinforce a broader shift from standardised rescue-chemotherapy selection toward patient-specific decision support.

For veterinary oncologists, the potential value lies in identifying drugs with a higher predicted probability of response when treatment options are limited and disease has returned after initial chemotherapy.

For pet owners, more informed treatment selection could potentially improve the value obtained from costly oncology care. However, the study does not establish that the technology lowers the total cost of treatment.

For the animal-health industry, the opportunity extends beyond canine lymphoma. The integration of live-cell assays, clinical outcome databases and machine learning could become a model for precision therapeutics in other veterinary oncology indications.

Commercial adoption will depend on clinical validation, laboratory turnaround times, reimbursement or owner affordability, integration into oncology workflows and evidence that the testing cost is justified by meaningful patient outcomes.

Analyst View

ImpriMed’s new findings provide a useful bridge between precision veterinary oncology and treatment economics. The observed median PFS of 49 versus 21 days is clinically notable, while the approximately 38% lower estimated cost per progression-free day presents a potentially attractive economic signal.

The next important evidence milestone will be prospective, independently replicated research that measures clinical outcomes, quality of life, total expenditure and treatment decisions across multiple veterinary oncology centres.

Conclusion

ImpriMed’s retrospective analysis adds an economic dimension to the growing evidence base for AI-supported veterinary oncology. By associating treatment-prediction concordance with longer progression-free survival and lower estimated daily costs, the findings support continued investigation of personalised chemotherapy selection for relapsed canine lymphoma.

For the veterinary oncology sector, the key question is now whether these promising associations can be reproduced in larger, prospective studies and translated into demonstrable improvements in both patient outcomes and the overall affordability of cancer care.