Pseudobudding: ruptured glands do not represent true tumor buds

Tumor budding (TB) is a strong biomarker of poor prognosis in colorectal cancer and other solid cancers. TB is defined as isolated single cancer cells or clusters of up to four cancer cells at the invasive tumor front. In areas with a large inflammatory response at the invasive front, single cells and cell clusters surrounding fragmented glands are observed appearing like TB. Occurrence of these small groups is referred to as pseudobudding (PsB), which arises due to external influences such as inflammation and glandular disruption. Using a combination of orthogonal approaches, we show that there are clear biological differences between TB and PsB. TB is representative of active invasion by presenting features of epithelial‐mesenchymal transition and exhibiting increased deposition of extracellular matrix within the surrounding tumor microenvironment (TME), whereas PsB represents a reactive response to heavy inflammation where increased levels of granulocytes within the surrounding TME are observed. Our study provides evidence that areas with a strong inflammatory reaction should be avoided in the routine diagnostic assessment of TB. © 2023 The Authors. The Journal of Pathology published by John Wiley & Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland.


Introduction
Tumor budding (TB) is defined as isolated single cancer cells or clusters up to four cancer cells located at the invasive margin of colorectal cancer (CRC) [1]. TB is firmly established as a strong and independent predictor of lymph node metastasis, disease recurrence, and cancerrelated deaths [2]. Pseudobudding (PsB), in contrast to TB, is defined as isolated single cancer cells or clusters of up to four cells at the invasive front that occur due to gland disruption and external influences such as inflammation [3]. PsB may be confused with true TB, yet their biology likely differs. No studies to date have investigated the molecular backgrounds of these two features and whether they are truly biologically distinct. Here we investigate TB and PsB using spatial transcriptomic, multiplex immunofluorescent, immunohistochemical, threedimensional (3D), and super-resolution-based approaches to gain insight into their underlying biology.

Cases and tissue material
This study included resection material from 21 CRC cases at the Radboud University Medical Center (RadboudUMC), where TB and PsB regions were observed (supplementary material, Table S1). Pathologists (SKO, IDN) were guided and selected TB and PsB regions based on a review of cases according to the International Tumor Budding Consensus Conference (ITBCC) 2016 guidelines and using the formal definition of PsB [1,3]. A consort diagram detailing patients and how the samples were used in our study is provided in supplementary material, Figure S1. Ethics guidelines were followed for using archival material for research. This study was approved by the Ethics Committee (CMO) of the RadboudUMC (approval number 2017-3603).

Multiplex immunofluorescence (mIF)
Serial sections (4 μm) of TMAs used for DSP were stained using an Opal™ 7 tumor-infiltrating lymphocyte (TIL) panel and myeloid-derived suppressor cell (MDSC) panel (Akoya Biosciences, Marlborough, MA, USA, supplementary material, Table S1), respectively, and analyzed as described previously [9]. The same ROIs as those selected for the DSP analyses (supplementary material, Figure S2) were selected within these serial sections, and only the regions within the ROIs were included in our mIF analysis.

Transmission electron microscopy (TEM)
Samples from three CRC cases where fresh formalinfixed resection material representing regions of TB and PsB was available were processed and visualized using TEM as described previously [10].

Differences in cellular morphology through 3D visualization and validation using TEM
The PsB region was found to be characterized in three dimensions by cellular debris surrounding a discontinuous epithelial lining ( Figure 1B, supplementary material, Video S1). Single cells within the disrupted area were observed as elongated and dissociated with lagging remnants of cell-to-cell adhesions. This may be due to inflammatory cells disrupting the epithelial lining and causing the cells to dissociate. In contrast, within the TB region we did not observe a disrupted epithelial lining ( Figure 1A, supplementary material, Video S2). Single cells and cell clusters were observed as intact and displayed a more linearly oriented invasive pattern.
TEM was performed to validate the differences observed in three dimensions and reveal the cellular ultrastructure of both features. A pseudobud ( Figure 1D) is observed to be elongated and detached from a larger glandular structure. The cell's membrane is ruptured and surrounded by cellular debris. In contrast, a single tumor

TS Haddad et al
bud is revealed to have an intact cell membrane with little to no perturbations in the surrounding tissue structure ( Figure 1C).

Phenotypic comparison of TME using mIF
To profile the cellular composition of the TME within the TB and PsB regions, we performed mIF on TMA serial sections using TIL and MDSC panels to quantify immune cell populations (Figure 2A,B). We observed no significant differences in T-cell, B-cell, or natural killer (NK) cell populations ( Figure 2C). In contrast, we observed higher densities of granulocytes in PsB regions compared with TB regions ( Figure 2D). We also observed significant differences in monocytes, macrophages, and antigen-presenting cell populations (supplementary material, Figure S3).
Distinct transcript profiles and molecular pathways differentiate between TB and PsB regions with validation using IHC Following preprocessing and normalization, we profiled the expression of over 1,100 unique cancer related gene targets across 321 total samples, of which 166 and 155 represented the tumor (PanCK+) and TME Tumor budding versus pseudobudding in colorectal cancer 21 (PanCKÀ), respectively. We found 47 and 24 DEGs to be statistically significant for TB and PsB regions, within the tumor and TME, respectively ( Figure 3A and supplementary material, Table S1). Supervised clustering with the top 100 DEGs separated the TB and PsB regions (supplementary material, Figure S4).  Figure 3 Legend on next page.

Tumor budding versus pseudobudding in colorectal cancer 23
GSEA demonstrated that there were differing molecular mechanisms between TB and PsB regions. Within the tumor segment of TB regions, we observed a significant enrichment of pathways involved in cell adhesion and motility, hepatocyte growth factor (HGF) receptor (MET), PDGF, and PI3K-Akt signaling, as well as matrix remodeling, metastasis, and epithelialmesenchymal-transition (EMT) ( Figure 3B). Within the tumor segment of PsB regions, we observed a significant enrichment of pathways involved with the cell cycle, DNA damage repair, and mitochondrial metabolism ( Figure 3B). Within the TME segments, we observed a significant upregulation of pathways in TB similar to those of the tumor segments, while in PsB regions we saw a significant enrichment of processes involved with host defense peptides and T-cells ( Figure 3B). A full list of GSEA on DSP is provided in supplementary material, Table S1. GSEA for Gene Ontology (GO) revealed processes involved in epithelial migration, extracellular matrix (ECM) organization, and cell cycle progression (supplementary material, Figure S5). FN1, involved in migration and metastatic processes, was found to be upregulated in TB regions along with structural components such as collagens IαI, IαII, IIαI, IIIαI, VαI, VαII, VIαIII, and XIαI and crosslinker glycoproteins such as tenascin C (TNC) ( Figure 3A).
Spatial deconvolution of immune cell populations within the TME showed a significantly higher abundance of fibroblasts and fewer neutrophils within the TB regions ( Figure 3C). Ki67, a marker of proliferation, was significantly upregulated in PsB regions ( Figure 3A) and validated with IHC staining (p < 0.001, t-test, Figure 4A). In borderline cases where TB and PsB may be difficult to distinguish, Ki67 staining could be used to help identify TB or PsB (supplementary material, Figure S6). Significantly higher stromal FN1 expression assessed using IHC can be observed in TB regions compared to PsB (p < 0.001, t-test, Figure 4B). Within the TME of PsB regions, keratin 18 is part of cellular debris, which is not observed in TB ( Figure 4C).

Discussion
Through a comprehensive analysis of TB and PsB, our findings clearly established these two features as biologically independent with distinct morphology, microenvironment, and transcriptomic features.
Within TB regions we observed higher RNA expression of extracellular structural components. Coupled with more fibroblasts and FN1, TB regions appeared to have increased deposition of ECM compared to that of PsB. CRCs in which major components of collagen are upregulated have been associated with advanced staging, metastasis, and poor prognosis [12][13][14][15][16][17]. In vitro knockdown of FN1 in CRC cell lines significantly inhibited cellular migration and invasion [18]. TNC may also play a role in driving tumor progression [19]. Based on this evidence, tumor regions exhibiting excessive deposition of ECM may possess significant invasive and metastatic potential [20,21].
PsB regions had more granulocytes within the TME, with a subset of them being neutrophils. Considering the nature of PsB regions where a discontinuous epithelial lining can be observed, granulocytes may play a role in causing glandular disruption. This may also explain the observation of keratin 18-positive cellular debris within the TME because neutrophils can cause damage to surrounding tumor tissue [22].
Within PsB regions, we observed an upregulation of genes associated with cellular division and DNA replication and significantly higher expression of nuclear Ki67 compared to TB. The lack of Ki67 expression is a reported attribute of TB and a hallmark of EMT by increasing cellular survival within the stroma during migration [2]. TB and EMT have also been inextricably linked [2,23]. PsB does not appear to exhibit the same hybrid EMT phenotype observed in TB, which provides further evidence that PsB is disseminated due to glandular disruption.
A study conducted by Oishi et al presented cancer gland rupture (CGR) as a potential risk factor for lymph node metastasis (LNM) in early-stage CRC [24]. Moreover, CGR was invariably associated with deep invasion and, to a lesser extent, lymphovascular invasion and TB. No multivariate analysis was performed to establish CGR as an independent risk factor, and associations between CGR and EMT were not investigated. In our study, we performed molecular expression analyses in cancer regions where PsB was present but did not find an association with EMT at either the transcript or protein level. Through a direct comparison with PsB regions, we verify the association of TB regions with EMT and cancer invasion, as have several others [2].
TB regions showed activation of the c-MET/HGF and PDGF pathways, which promote tumor invasion, metastasis, and angiogenesis [25,26]. Potential targeting or inhibition of MET-driven pathways may slow tumor progression [27,28]. PsB regions showed activation of DNA damage repair pathways and host defense peptides, which may be correlated with damage to the tumor   Tumor budding versus pseudobudding in colorectal cancer 25 epithelium caused by immune cells within regions where the host defense is activated. These pathways and mechanisms could be important for the development of future novel biomarkers. This study provides a framework for further research on mechanisms of cancer invasion, tumor heterogeneity, and immune response/activation at the invasive tumor front. This study has given us insight into tumor heterogeneity and the aspects of tumors that make them particularly invasive and migratory/metastatic in CRC. Previous studies showed that TB regions were characterized by a phenotypic switch from the tumor bulk through the acquisition of migratory characteristics, decreasing cell proliferation, and with TB regions having EMT-positive signatures [29]. Our study characterized TB regions with the same phenotype while highlighting heterogeneity across the invasive front by identifying PsB regions characterized by cell proliferation and lacking EMT-like or migratory signatures.
We did not score PsB in a large series to observe its prognostic effect, nor did we identify specific IHC markers to distinguish TB and PsB. Studies should be conducted to investigate the prognostic value of PsB in a large cohort. Our study did not take into account temporal changes, and, as is inherent in histopathology, looked at a single moment in time which is after the tumor specimen had been excised. At this particular moment, our findings provide evidence to suggest that TB and PsB are features representing differing biological processes. Our findings were made in the context of cancerspecific gene targets and not a whole transcriptome panel. We acknowledge that TB and PsB may be difficult to distinguish when assessed in practice. However, the presence or absence of Ki67 might be a useful tool in cases of doubt (supplementary material, Figure S6).
Several studies confirmed the independent prognostic value of TB in stage II CRCs and for LNM in early-stage CRCs [30][31][32][33]. Accurate identification of true tumor buds at the invasive front is important because TB has been found valuable in multiple clinical scenarios: (1) for determining the risk of LNM in patients with early-stage CRC and (2) for identifying patients with high-risk stage II CRC [2]. High-grade TB in early-stage CRC may warrant radical surgery with lymph node dissection. High-grade TB in stage II CRC patients might indicate a potential benefit from adjuvant therapy [2]. We have provided evidence that areas with PsB should be avoided during clinical assessment of TB because their inclusion may lead to an artificially high TB score. With ITBCC guidelines determining TB score based on a range of 0-10 tumor buds, every tumor bud should be accurately identified for effective clinical decision-making.  Table S1. Clinical demographic data of patients, tumor-infiltrating lymphocyte (TIL) panel, myeloid-derived suppressor cell (MDSC) panel, full NanoString data analysis (DEGs, tumor, and TME), full GSEA results (tumor and TME).

SUPPLEMENTARY MATERIAL ONLINE
Video S1. Pseudobudding in 3D Video S2. Tumor budding in 3D Tumor budding versus pseudobudding in colorectal cancer 27