Blood C3 fell by 90 per cent and PD-1 therapy still worked; the tumour's own C3 was the one that mattered

Two sources of one protein

Complement C3 is usually discussed as a blood protein: the liver makes it, it circulates, and it takes part in the innate immune response. The Nagoya University group reports that tumours have their own supply. In mice engineered so that cancer-associated fibroblasts cannot make C3, colorectal and lung tumours resisted anti-PD-1 treatment and filled with the M2-like macrophages that suppress an immune attack. Cutting the liver's contribution, about 90 per cent of the C3 in circulation, left the response to the antibody intact.^[1]

If the C3 that decides the outcome is made a few micrometres from the tumour cells, then measuring C3 in a blood sample answers a different question from the one a clinician wants answered. That is the practical shift here: the quantity moves from something a routine assay already reports to something that requires tissue. I read this as the strongest claim the paper can carry, and it is a claim about where to look. A second reading is available: stromal C3 may be a marker of a broader fibroblast state that itself drives the response, in which case C3 travels with the biology without being the lever.^[1]

What the mice establish

The mechanism the authors set out runs through iC3b, the fragment left when C3 is broken down. iC3b signals through complement receptor 3 and holds immunosuppressive myeloid cells out of the tumour. When the team intervened in that pathway in mice whose tumours had stopped responding, sensitivity to PD-1 blockade returned. This is the part of the work with a control and an intervention behind it, and it is the part I would expect to survive replication.^[1]

The human evidence sits at a different level. In patient tumour tissue, higher stromal C3 expression went with fewer M2-like macrophages and better immunotherapy outcomes, a correlation measured in samples. There is no randomisation and no prespecified endpoint. The reported split is stark: roughly half of the patients with high local C3 responded to treatment and none of those with lower levels did. A stark split in a small series is exactly the kind of result that shrinks as the series grows. Preclinical work of this quality earns a trial; it does not stand in for one.^[1]

What would move this to the clinic

The next observable step is narrow and datable. A prospective cohort would have to measure stromal C3 in biopsies taken before checkpoint therapy starts, fix the cut-off in advance, and report response rates by C3 level. If a study of that shape reports by the end of August 2027 and the separation holds, local C3 becomes a candidate criterion for deciding who receives a checkpoint antibody. Without such a study the finding stays a hypothesis about patient selection.^[1]

For a patient on a checkpoint antibody today, none of this changes the next appointment. What it changes is the question a pathologist might one day be asked: a question about the tumour's own stroma in place of a question about the blood. The therapy itself has not moved; the place we would look in order to predict it has.^[1]