Abstract Background: Categorizing breast cancer HER2/ERBB2 expression as “positive” or “negative” is no longer sufficient, with evidence that treatment response and outcomes are associated with HER2 “low” status and HER2 intratumoral heterogeneity. We hypothesized that interrogating HER2 heterogeneity (HER2het) across multiple spatial resolutions would more accurately capture HER2 diversity and be associated with clinical outcomes. Methods: We interrogated tumor cell and microenvironmental features by profiling 1,113,204 single cells in tissue sections from 171 HER2+/HER2low cancers via custom 25-marker high dimensional multiplexed immunofluorescence (HDmIF) using NeoGenomics MultiOmyx, with adjacent section HER2 immunohistochemistry (IHC). We developed novel metrics to concurrently: 1) interrogate HER2 heterogeneity at four spatial resolutions; 2) use machine learning to translate HER2 IF to IHC, termed ‘HAIQu’ (HER2 Automated Immunofluorescence Quantification), scoring HER2 IF expression according to ASCO/CAP guidelines; 3) delineate HER2 signaling phenotypes at the tumor cell level based on six HER2-related proteins; 4) evaluate immunophenotype of 23 immune cell types. We evaluated the association of these novel HER2het metrics with patient clinicopathologic features, recurrence-free survival (RFS), overall survival (OS), and diverse antibody markers representing tumor cell intrinsic processes and tumor-immune microenvironment (TME). Results: 1166 regions of interest were analyzed from 208 unique tumors profiled. Median follow-up from diagnosis was 143 months and 98.9% (n=183/185) received HER2-directed therapy in the (neo)adjuvant or metastatic setting. Our HAIQu scoring system effectively translated adjacent section HER2 IF to IHC with 97.9% concordance between HAIQu and clinical IHC scoring. Single-cell phenotypic analyses of 392,984 HER2+/PanCK+ tumor cells’ concurrent expression of six HER2-positive breast cancer related proteins (HER2, HER3, EGFR, pAKT, ER, KI67) using an unsupervised neural network-based self-organizing map approach resulted in 7 HER2 signaling cell phenotypes. Most patient samples are dominated by a single cell cluster but, intriguingly, Cluster 1 cells (EGFR-low) predominate in tumors with high HER2 cell membrane heterogeneity (ANOVA p=0.003). Evaluation of immunophenotype in hormone receptor-negative, HER2+ tumors, demonstrated significant association with immune cluster and recurrence-free survival (RFS; log-rank p=0.024) with zero RFS events among immune-high versus median survival of only 53.5mo among immune low-PDL1 low tumors. A multivariable Cox proportional hazards model including single cell HER2-heterogeneity (only significant metric on univariate), receptor subtype, and immunophenotype cluster demonstrated significant association with RFS (overall model log-rank p=0.005) and each significantly contributed to the model (all p< 0.05). Conclusions: We present novel metrics of HER2 heterogeneity via HDmIF, which offer detailed characterization of the diversity of HER2 expression in a large, clinically-annotated cohort with long-term follow-up. Identification of a strong association between immunophenotype and RFS supports further investigation of the highly immune activated subsets of ER-/HER2+ breast cancer. Strong correspondence of HER2 IF and IHC and our HAIQu methodology offers a pathway to translation of HER2het metrics to clinical practice. Table 1. Association of HER2 Heterogeneity Metrics with Recurrence-Free Survival Citation Format: David Tallman, Anna Juncker-Jensen, Harry Nunns, Kevin Gallagher, Heather LeFebvre, Karen Yamamoto, Katharine Collier, Mark Vater, Ava Strahan, Mathew Cherian, Ashley Pariser, Preeti Sudheendra, Bhuvaneswari Ramaswamy, Margaret Gatti-Mays, Ainura Kyshtoobayeva, Zaibo Li, Daniel Stover, Kai Johnson. Novel Metrics of HER2 Heterogeneity in HER2-Positive and HER2-Low Breast Cancer via High Dimensional Multiplexed Immunofluorescence Spatial Profiling [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO2-14-03.