A peptide PET tracer caught CD147 in eleven patients. The biomarker FDG follows did not correlate.

A research peptide PET tracer reflected the abundance of a tumor surface protein called CD147 across eleven cancer patients on whole-body imaging, and reflected it more accurately than FDG, the workhorse cancer PET tracer that has been the default for thirty years, the developers wrote in Advanced Science ↗ on May 26. The strongest correlations the new tracer hit were with CD147 itself (r equals 0.80, P less than 0.001), with the lactate transporter MCT1 (r equals 0.86), and with the proliferation marker Ki67 (r equals 0.80). The one biomarker it did not correlate with was GLUT1 (P equals 0.07), which is the glucose transporter that FDG follows.

A PET tracer is a short-lived radioactive molecule that you inject and then scan; the scanner sees where the molecule went. FDG (fluorodeoxyglucose) is a glucose lookalike that piles up in cells running hot on glucose, which most cancer cells do. The new tracer, called [68Ga]Ga-DOTA-AP9, is a short peptide labeled with gallium-68 that binds to CD147, a surface protein that turns up on many cancer cells and is involved in invasion and metabolic rewiring. CD147 is also known as EMMPRIN, the extracellular matrix metalloproteinase inducer, and it has been pursued as a therapeutic target by antibody-drug conjugates and small molecules for over a decade. Imaging it non-invasively, on a living patient, on a single scan, is what the new tracer is for.

The preclinical work is what you would expect from a tracer paper: high binding affinity in vitro, specific uptake in CD147-positive A375 melanoma tumors implanted in mice, low uptake in CD147-negative controls. The translational step is the eleven-patient pilot. The cohort had diverse cancer types and varying CD147 expression levels by biopsy. Each patient was imaged on a uEXPLORER, the total-body PET system that captures the whole human body in one acquisition. Tumor lesions reached a maximum standardized uptake value of 10.1 at thirty minutes after injection. Standardized uptake value is the standard PET intensity number; ten is high.

The GLUT1 result is the headline. Three of the four biomarkers tracked AP9 uptake tightly: CD147 (the intended target), MCT1 (the lactate exporter), and Ki67 (proliferation). The fourth, GLUT1, did not. The interpretation the paper offers is that AP9 reads the lactate-loading metabolic axis (CD147 plus MCT1 plus proliferation) without crosstalk to the glucose-uptake axis (GLUT1) that FDG reads. In some lesions, the two tracers gave divergent uptake patterns. That is the value proposition. FDG tells you which tumors are burning glucose. AP9, in this paper, tells you which tumors are pushing lactate out and turning over CD147 at the same time.

The clinical question the paper raises but does not answer is whether AP9 stratifies patients better than FDG for any actual treatment decision. Eleven heterogeneous patients on a pilot scan is not powered for survival or response endpoints. The authors stop at correlation strength and uptake-pattern divergence. What comes next is cancer-type-specific cohorts, then prospective treatment-selection comparisons. CD147-targeted therapeutics in development (including monoclonal antibody-drug conjugates and small-molecule inhibitors) all need a way to identify patients whose tumors actually express the target. An imaging companion does that.

The broader pattern the paper sits in is that the peptide PET tracer menu past FDG keeps growing. Somatostatin analogs (DOTATATE) for neuroendocrine tumors, PSMA-617 for prostate, FAPI peptides for fibroblast activation across solid tumors. Each one reads a target-specific layer of tumor biology that FDG cannot. AP9 adds a CD147-specific reader to that menu, and the same molecule class supports a parallel therapeutic axis: when a peptide PET tracer maps where the target is, a peptide therapeutic tagged with a different isotope can deliver radiation to the same place. That is the theranostic pairing that lutetium-177 DOTATATE already runs in neuroendocrine cancer and PSMA-617 already runs in prostate.

What this study is not: a survival readout, a treatment-response prediction, or a cancer-type-specific efficacy claim. Eleven patients is a phase-zero pilot. The signal here is that a peptide tracer caught what a small-molecule tracer missed in a head-to-head subset, and the missed information was clinically meaningful enough to publish. The clinical use case is hypothesis. The correlation strength is news.