Some years ago, a team asked which patients with Barrett’s esophagus are likely to progress to cancer. Their assay measures a handful of protein markers by immunofluorescence, folds in tissue architecture, and returns a risk score [1]. Years later, it is one of the very few multiplexed immunofluorescence tests in routine care. I admire it. I also think it is the clearest illustration of why many protein tissue panels don’t scale. It shows what it costs to do it right once. Multiply by every indication that needs one.
A test is only worth the decision it changes.
A diagnostic panel is not a reagent. It is a reagent plus a question, a patient population, a clinical decision, and the evidence that the decision gets better when you use it. Each piece has an owner, and it is rarely whoever made the reagent.
The drug belongs to pharma. The indication belongs to clinicians. The decision belongs to the oncologist. The assay developer owns the antibodies, the protocol, and the hope that someone else’s trial will need them. Evidence for one indication can cost more than many programs ever raise.
Dark can always get darker.
Underneath the business problem sits a physical one. H&E has a floor and a ceiling: unstained tissue at one end, a fully stained nucleus at the other. Chromogenic IHC works the same way, from no brown to saturated brown. Pathologists spent a century calibrating to that bounded scale. A 2+ in Boston is roughly a 2+ in Seoul.
Immunofluorescence has neither. The floor is the tissue’s own glow and the operator’s exposure; dark can always get darker. The ceiling is the camera, not the chemistry; bright can always get brighter. Numbers are relative to the instrument, the protocol, and the day.
The signal must stand out from the glare, reproducibly, including when expression is low and the clinical question depends on it. They can be rescaled, normalized, and re-thresholded until they agree with whatever you want to see. Convenient for a paper. Corrosive for a standard.
This is solvable, but only on purpose. Chromogenic stains got their anchors for free by giving up dynamic range. Fluorescence needs to build them: reference standards on every slide, calibrated intensities, and cells whose expression is known.
Every module is a new company.
The field’s answer has been the modular panel: a small marker set per question. Some cover immune cells. Others chase a niche. The overhead doesn’t shrink. Each module needs its own development, manufacturing, QC, analytical validation, clinical studies, billing code, and payer fight. Fixed cost stays high as each niche market shrinks. It isn’t impossible. That Barrett’s assay proves it. But thirty indications means thirty starts from zero. That is a franchise, not a platform.
What sequencing figured out.
Genomics faced the same trap and got out. A base call means the same thing in every lab. Copy numbers come in whole numbers. Spike-ins were developed to provide floor and ceiling, so its measurements travel. That opened a different path. Comprehensive tumor-profiling assays have been cleared as a single test covering hundreds of genes, with companion-diagnostic claims for several drugs, and have added claims since [2]. Analytical work once. Each new drug added a claim, not a new product. Validation compounded.
What proteins need.
Proteins need the same two things. A fixed scale: anchors on every slide so a number means the same thing across instruments, labs, and years. And a panel broad enough to reuse across indications, so manufacturing, QC, and analytical validation are paid once, not once per module.
A word of caution: a broad panel does not make clinical validation free. Every indication still needs its own evidence. What changes is the price of the next one: an added claim on an existing test, not a new product line.
Numbers that travel.
Physics told us how many molecules we can see. Budgets told us which ones we chose to see. The third constraint is the least discussed: a measurement is only worth as much as its numbers can travel, to the next lab, the next panel, and the next patient. Until protein measurements can make that trip, every panel will start from zero, and medicine will keep rebuilding the same foundation.
Sources
-
Critchley-Thorne RJ, et al. A tissue systems pathology assay for high-risk Barrett’s esophagus. Cancer Epidemiology, Biomarkers & Prevention 25, 958-968 (2016).
-
U.S. Food and Drug Administration. Premarket approval for a comprehensive genomic profiling companion diagnostic (P170019) (2017).