The most abundant cells in a tumour —
and they have not picked a side
Cancer immunotherapy is almost always told through T cells. Yet dissociate a tumour and count: T cells are only 4.3%. What fills the space is the myeloid compartment — 85.1%: neutrophils and macrophages.
A T cell has picked a side. A myeloid cell has not. The same macrophage can eat a cancer cell or lay down the vessels that feed it. What decides that — and can a drug change it?
Not just in words: you open 33,594 mouse tumour myeloid cells, re-derive the paper's six subtypes from its own markers, test with independent genes whether those labels earn their names, measure whether M1/M2 is really one axis, and check the whole conclusion against spatial transcriptomics.
How you'll learn
We re-derive the six subtypes of Fig. 7 using the paper's own markers. M1-like monocytes 11.46% → 28.07%, circulating monocytes 27.34% → 13.75% — both claims reproduce.
Proving a name with the genes that gave the name is circular. We test with different genes: circulating monocytes lead three of four suppression axes, while 'M1-like' tops the ROS axis but comes last on pathogen sensing.
If M1/M2 were one axis, the two scores would correlate at −1. They correlate at −0.179, and 21.9% of cells are high on both. It is not one axis.
The paper's Cd47 claim is about cancer cells, and this dataset has none. Elane and Mpo read zero, so NETs are invisible here. We label the untestable as untestable.