A 3D Atlas Characterizes Molecular Features of Colorectal Cancer Morphology

Histopathology informs the clinical management of advanced solid tumors, as many characteristic features of tumors have diagnostic and prognostic value. To investigate the molecular processes that underlie the various morphologies present within a tumor, Lin, Wang, Coy, and colleagues created a spatial tumor atlas of human colorectal cancer (CRC) through image-based single-cell analysis of patient samples, integrating high-plex cyclic immunofluorescence (CyCIF) imaging (a method for multiplexed imaging) with hematoxylin and eosin (H&E) staining. Although high-plex tissue imaging techniques often use tissue microarrays (TMA), a comparison of feature measurements from assessing either TMA cores or whole-slide images (WSI) highlighted the superior statistical power of the latter approach, demonstrating that analysis within the small field of view of TMA cores could not adequately estimate features such as cell abundance. Machine learning classifiers of CyCIF intensities were trained on H&E-based pathology annotations, revealing that disease-relevant morphologies often co-occurred within the same sample and corresponded to specific molecular patterns of marker expression. Moreover, transitions from one morphologic class to another were common and accompanied by a molecular gradient, suggesting that observations of intratumoral heterogeneity may arise, in part, via sampling of different regions along a continuous gradient. Molecular transitions were also apparent in tumor buds, classically defined as clusters of four or less tumor cells surrounded by stroma along the invasive front. 3D reconstruction of tumor buds indicated that, rather than discrete 2D clusters, tumor buds were connected as 3D fibrillar structures extending from the tumor mass and terminating in bud-like structures that protruded into stroma as well as mucinous spaces. Similarly, tertiary lymphoid structures also formed interconnected networks, which could be heterogeneous in cell type and cell state. Notably, analysis of PD1–PDL1 interactions revealed that PDL1⁺ myeloid cells, rather than tumor cells, were the primary mediators of PD1⁺ T-cell immunosuppression. Overall, this spatial 3D atlas of CRC reveals insights into important structural features and the molecular patterns that underlie these morphologies.

Lin JR, Wang S, Coy S, Chen YA, Yapp C, Tyler M, et al. Multiplexed 3D atlas of state transitions and immune interaction in colorectal cancer. Cell 2023;186:363–81.

Note:Research Watch is written by Cancer Discovery editorial staff. Readers are encouraged to consult the original articles for full details. For more Research Watch, visit Cancer Discovery online at http://cancerdiscovery.aacrjournals.org/CDNews.