A rare neuron-like cell may spare dogs a jaw resection

A Cornell-led team reports that single-cell RNA sequencing of canine oral tumors turned up a rare cell population, carrying what one senior author called an unexpected neuron-like molecular signature, that is unique to canine acanthomatous ameloblastoma ^[5][6]. That matters because CAA is benign but locally aggressive and destructive, and it overlaps enough with malignant, metastatic canine oral squamous cell carcinoma to produce misdiagnoses in close to a third of cases ^[1][2][3].

The cost of getting it wrong is anatomical. Surgery is the standard treatment for CAA, often involving removal of a section of the jawbone ^[4]. Dr. Santiago Peralta, a senior author and associate professor of dentistry and oral surgery at Cornell's College of Veterinary Medicine, framed the diagnostic problem as a consequence of drug options rather than a curiosity: the goal is to shrink tumors to the point of minimal or no surgery, and "that's when it becomes super critical to have an absolutely accurate diagnosis because one molecule will be effective in one tumor but not in the other" ^[12][13].

The molecule in question comes out of the same cells. One of the genes expressed in the rare neuron-like population is ERBB4 ^[6], a member of a gene family already well studied in human cancers including certain forms of breast cancer ^[7]. Working with Taranjit Gujral, a systems biologist at the Fred Hutchinson Cancer Center at the University of Washington, the team screened a panel of dozens of therapeutic compounds against live CAA tumor tissue, and neratinib, a known ERBB4 inhibitor, suppressed tumor growth ^[9]. Cornell's summary describes the hit as an FDA-approved drug targeting ERBB4 that reduced CAA tumor growth and is ready for clinical trials ^[8]; by elimination that drug is neratinib ^[10].

What the announcement does not carry is arithmetic. There is no effect size, no number of tumors or dogs, no dosing, and no sensitivity or specificity for the new diagnostic separation of CAA from COSCC ^[14]. Live tumor tissue in a screening plate is a long way from a treated patient, and a hit inside a panel of dozens of compounds says more about the panel than about durability. The specimens came from Peralta's own biopsy archive ^[15], which is the right material for discovery and the wrong material for claiming a test works prospectively.

The translational argument is the more interesting part of the pitch. The team notes that the human analog of CAA is relatively rare, which makes funding hard, and that human trials require stringent protocols and take years to start and finish ^[16]. Dogs are easier to work with, and the researchers hope a successful treatment would both help canine patients and inform the analogous human disease ^[17]. The study, with doctoral student Andreas Stephanou as first author, was published in Molecular Therapy Oncology ^[11].

Watch for the canine trial itself, and for whether ERBB4 expression in the rare cell population predicts which dogs respond rather than merely marking which tumor is which ^[18]. Watch also for a prospective read of the diagnostic on fresh biopsies, and for whether tumors shrink enough to change the surgical plan, which is the only endpoint that saves a jaw ^[4].