PO.BCS01.17 · 生物信息与计算
HPlot：一种用于肿瘤微环境免疫异质性空间分析的新型定量框架
HPlot: A novel quantitative framework for spatial profiling of Immune heterogeneity in tumor microenvironments
作者与单位 Authors & Affiliations
摘要 Abstract
中文摘要
免疫细胞和肿瘤细胞的空间组织在疾病进展和治疗反应中发挥关键作用。现有的空间分析方法常将组织区域视为均一的隔室，无法量化免疫组成和功能状态如何跨肿瘤边界变化。这限制了检测与治疗结局相关的空间生物标志物的能力。
我们开发了H-Plot，一种用于高分辨率分析空间免疫异质性的定量框架。该方法兼容全切片成像和现代单细胞空间生物学平台。H-Plot整合三个组成部分：
（1）细胞预测与分类：使用基于机器学习的单细胞图像分析或源自空间转录组学的细胞亚型分型，生成高保真的细胞图谱；
（2）细胞功能富集分析：通过局部功能程序检测识别具有生物学意义的结构，如肿瘤区域、淋巴聚集体和三级淋巴结构（TLS）；以及
（3）细胞空间分析：测量空间关系并计算距生物结构边界（如肿瘤边缘）的逐层距离，以捕捉跨肿瘤-免疫界面的空间梯度。
将H-Plot应用于ABACUS膀胱癌数据集，结果显示治疗应答者在肿瘤边界附近表现出更高的淋巴细胞富集和更平滑的免疫层过渡，而无应答者则表现出浅表或碎片化的浸润。在TEMPUS mCRPC队列中，治疗前淋巴细胞邻近性降低和富集层减弱与较差的临床结局相关，并与基于RNA的免疫特征和组织学评估一致。H-Plot提供了一个可重复、可解释且可视化就绪的框架，用于量化肿瘤-免疫空间异质性。通过将复杂的空间排布转化为逐层定量分析和直观可视化，H-Plot实现了跨患者、跨治疗和跨肿瘤类型的系统性比较。该框架支持空间生物标志物发现，并为转化肿瘤学研究提供了一个可扩展的工具。
查看英文原文 English abstract
Spatial organization of immune and tumor cells plays a critical role in disease progression and therapeutic response. Existing spatial analysis methods often treat tissue regions as uniform compartments and fail to quantify how immune composition and functional states change across tumor boundaries. This limits the ability to detect spatial biomarkers associated with treatment outcomes.
We developed H-Plot, a quantitative framework for profiling spatial immune heterogeneity with high resolution. The method is compatible with whole-slide imaging and modern single-cell spatial biology platforms. H-Plot integrates three components:
(1) Cell prediction and classification using machine-learning-based single-cell image analysis or spatial transcriptomics-derived cell subtyping to generate a high-fidelity cellular map;
(2) Cell-function enrichment analysis to identify biologically meaningful structures such as tumor regions, lymphoid aggregates, and tertiary lymphoid structures (TLS) through localized functional program detection; and
(3) Cell spatial profiling, which measures spatial relationships and computes layer-wise distances from biological structure boundaries (e.g., tumor margins) to capture spatial gradients across the tumor-immune interface.
Applying H-Plot to the ABACUS bladder cancer dataset revealed that treatment responders exhibited higher lymphocyte enrichment adjacent to tumor borders and smoother immune-layer transitions, whereas non-responders showed shallow or fragmented infiltration. In the TEMPUS mCRPC cohort, reduced pre-treatment lymphocyte proximity and weaker enrichment layers were associated with poorer clinical outcomes and aligned with RNA-based immune signatures and histological assessments.H-Plot provides a reproducible, interpretable, and visualization-ready framework for quantifying tumor-immune spatial heterogeneity. By converting complex spatial arrangements into layer-wise quantitative profiles and intuitive visualizations, H-Plot enables systematic comparison across patients, treatments, and tumor types. This framework supports spatial biomarker discovery and offers a scalable tool for translational oncology research.
利益披露 Disclosure
C. Huang,
Pfizer Inc. Employment, Stock, Stock Option, Travel.
A. M. Gonzalo,
AIgnoistics Employment, Stock, Stock Option.
S. Laturnus,
AIgnoistics Employment, Stock, Stock Option.