Cell-cycle–interferon coordination changes across proliferative states in TNBC/basal-like breast cancer

Cohort-wide pathway correlations can hide strong relationships that differ across tumor states. We asked whether this occurs in TNBC/basal-like breast cancer by ordering tumors on prespecified Hallmark pathway axes and comparing pathway correlations in fixed low and high states under complete-family error control. The dominant result was a recurrent proliferation-interferon pattern: three independently replicated relationships showed weaker or reversed cell-cycle/interferon correlation at higher E2F or G2/M state. Two additional, biologically distinct patient relationships also replicated, linking IL-6/JAK/STAT3 state to apoptosis-cholesterol coordination and IFN-gamma state to EMT-IFN-alpha coordination, without implying a shared molecular mechanism. Thirty-two patient relationships survived structural checks and five replicated in the same direction. Cell-line analyses provided complementary constraints: CRISPR tested selective single-gene dependency associations, whereas GDSC tested preclinical drug-response associations. None of 895,224 valid CRISPR axis-gene tests survived global correction, whereas four of 31,610 GDSC axis-drug tests did, on NOTCH, Complement, or Apoptosis axes; three involved MEK1/2 inhibitors. These drug-response associations do not validate the patient correlations, but they motivate prospective testing of pathway states as preclinical stratification variables. Practically, a weak pooled correlation should not be read as no association when a biologically defensible ordering axis exists.