Antibody drug conjugates only deploy their cytotoxic payload when immunohistochemistry (IHC) staining hits a threshold, so a routine score that used to be eyeballed is now a treatment and trial enrollment decision. AI vendors are moving in to standardize the read.
A test pathologists have run for decades is suddenly a gatekeeper for drugs that cost thousands of dollars a month. The mechanism underneath the AI pathology story is not automation. It is repricing. Antibody-drug conjugates, or ADCs, are targeted therapies that carry a cytotoxic payload only to cells expressing a chosen antigen. ADCs have converted a commodity tissue stain called immunohistochemistry (IHC) into a high-stakes diagnostic decision. AI is entering the pathology lab because the number now has to be right.
That shift is what Ibex Medical Analytics co-founder Yossi Mossel is describing when he says AI can free pathologists' "cognitive fuel" from manual visual scoring in a Q&A with Technology Networks. The framing is a vendor's, and it is worth holding at arm's length. The independent evidence underneath it is the question that matters.
IHC works by staining tumor tissue so pathologists can see how much of a target protein, an antigen, is sitting on or in the cells. In breast cancer the most established version scores HER2, a receptor that drives tumor growth and that older targeted therapies like trastuzumab already depend on. The score is a number on a 0-to-3+ scale, assigned by a human looking down a microscope. The newer wrinkle is that ADCs use that same number as a gating read. A drug like trastuzumab deruxtecan, marketed as Enhertu, only deploys its payload if HER2 is present at sufficient density on the tumor.
That coupling is the unit-of-account change. A routine stain that used to influence which of two older therapies a patient got now influences whether an entire class of newer therapies applies at all. And because the score is still read by eye, two pathologists looking at the same slide can land in different HER2 categories, a problem the field has known about for years. Variability that used to be a clinical inconvenience is now a trial-enrollment and prescribing liability.
Two reference points give the AI-IHC story independent grounding beyond the vendor's pitch. The first is a multireader study published in JCO Precision Oncology evaluating Ibex's fully automated HER2 IHC scoring AI in breast cancer. The second is the company's IVDR certification announcement for its HER2 breast biomarker scoring solution, which puts the product in front of European regulators under the in vitro diagnostic regulation framework rather than as a research tool. An international multi-site study announced via BusinessWire in October 2024 reached a similar conclusion on scoring improvement. The IVDR anchor does not equal FDA clearance; for the U.S. market the relevant question is whether the FDA has cleared any specific Ibex AI product, which is not established by the public record available here.
What changes for the pathologist if these tools actually take hold is not displacement so much as workflow redesign. A lab that adopts automated IHC scoring still has a pathologist in the loop, but the pathologist's job shifts from primary scorer to validator and exception handler. The cognitive fuel framing is half-right. The manual score gets handed to the algorithm, but the interpretation, the override, and the accountability stay human.
The honest counterweight is that ADCs themselves are a maturing therapeutic class. Several high-profile ADC programs have produced mixed phase 3 readouts, and the pipeline of antigen-density-gated therapies is not guaranteed to keep expanding at the current pace. If the ADC pipeline shrinks, the economic forcing function behind AI-IHC scoring shrinks with it, and the demand case for automated scoring weakens. That is the falsifier for the unit-of-account story. The price has to keep being repriced for the AI response to remain justified.
What to watch: FDA decisions on AI-assisted IHC scoring tools under the 510(k) and De Novo pathways, ASCO/CAP guidance updates on HER2 scoring thresholds in the ADC era, and whether reimbursement codes follow automated IHC reads. The pathology lab's microscope is not going away. What is changing is who has to be right, and how often, when the slide decides whether a several-thousand-dollar-a-month drug is worth starting.