BackgroundTumor-associated　macrophages　(TAMs)　are　pivotal　components　of　the　breast　cancer　(BC)　tumor　microenvironment　(TME),　significantly　influencing　tumor　progression　and　response　to　therapy.　However,　the　heterogeneity　and　specific　roles　of　TAM　subpopulations　in　BC　remain　inadequately　understood.MethodsWe　performed　an　integrated　analysis　of　single-cell　RNA　sequencing　(scRNA-seq)　and　bulk　RNA　sequencing　(RNA-seq)　data　from　BC　patients　to　comprehensively　characterize　TAM　heterogeneity.　Utilizing　the　MetaTiME　computational　framework　and　consensus　clustering,　we　identified　distinct　TAM　subtypes　and　assessed　their　associations　with　clinical　outcomes　and　treatment　responses.　A　machine　learning-based　predictive　model　was　developed　to　evaluate　the　prognostic　significance　of　TAM-related　gene　expression　profiles.ResultsOur　analysis　revealed　three　distinct　TAM　subgroups.　Notably,　we　identified　a　novel　macrophage　subtype,　M_Macrophage-SPP1-C1Q,　characterized　by　high　expression　of　SPP1　and　C1QA,　representing　an　intermediate　differentiation　state　with　unique　proliferative　and　oncogenic　properties.　High　infiltration　of　M_Macrophage-SPP1-C1Q　was　significantly　associated　with　poor　overall　survival　(OS)　and　chemotherapy　resistance　in　BC　patients.　We　developed　a　Random　Forest　(RF)-based　predictive　model,　Macro.RF,　which　accurately　stratified　patients　based　on　survival　outcomes　and　chemotherapy　responses,　independent　of　established　prognostic　parameters.ConclusionThis　study　uncovers　a　previously　unrecognized　TAM　subtype　that　drives　poor　prognosis　in　BC.　The　identification　of　M_Macrophage-SPP1-C1Q　enhances　our　understanding　of　TAM　heterogeneity　within　the　TME　and　offers　a　novel　prognostic　biomarker.　The　Macro.RF　model　provides　a　robust　tool　for　predicting　clinical　outcomes　and　guiding　personalized　treatment　strategies　in　BC　patients.