In 1985, a study of colonic adenocarcinoma in rats reported a peculiar feature at the invasive border of differentiated tumors[17,18]. Using both light and electron microscopy, Gabbert et al[18] observed neoplastic glands irregularly arranged into small strands and cords. In addition, they noted discontinuous small aggregates or single tumor cells which ultrastructurally did not possess junctional complexes, often had incomplete desmosomes, missing or rudimentary basement membranes and absent or incomplete brush borders. They determined that at the invasive tumor front of colorectal cancer, differentiated tumors could acquire an undifferentiated phenotype. Their observation is credited for pioneering the concept known as de-differentiation at the invasive margin, which today is commonly referred to as “tumor budding”.

Tumor budding is a histological feature diagnosed at high magnification and is defined as single cells or clusters of up to four or five cells at the invasive tumor front[19]. Budding cells can be spotted using standard HE-stained tissue slides and their visualization is further facilitated using pan-cytokeratin stains (Figure 1A and B)[7]. The process of tumor budding is likened to EMT observed during gastrulation in embryonic development[7]. In the early phase of EMT, epithelial cells reduce intercellular contacts and cell-matrix contacts and reorganize the cytoskeleton to form cell membrane ruffles (lamellipodia) or cytoplasmic protrusions. Migration ensues and new cell-matrix contacts are formed providing cells with an anchorage for the contraction of the cell body. In colorectal cancer, tumor budding is a highly dynamic process giving temporal heterogeneity to the tumor.

Figure 1 The invasive front of colorectal cancer. A: HE staining of colorectal cancer (× 40 magnification) showing tumor buds (arrows) at the invasive front; B: Pan-cytokeratin staining (CK22) highlighting tumor buds at the invasive front of colorectal cancer (× 40 magnification). Colorectal cancers with different tumor border configurations upon evaluation of HE staining at low magnification (× 5); C: Infiltrating tumor border configuration; D: Pushing tumor border configuration; E: Low (× 5); F: High (× 20) power magnification of double immunostaining for pan-cytokeratin (CK22) and anti-CD8 antibody highlighting “attackers” (tumor buds, brown) and “defenders” (CD8+ T-lymphocytes, red) at the invasive front of a colorectal cancer with infiltrating tumor border configuration.

Budding cells have been credited with the properties of malignant stem cells including the potential for redifferentiation both locally and at sites of distant metastasis and marking, what appears to be, the first histological event in tumor cell migration and invasion. Supporting this hypothesis further is the presence of “pseudopodia-like” cytoplasmic protrusions in tumor buds which have been identified by both electron microscopy and recently by immunohistochemistry with pan-cytokeratins[17,20]. These podia appear to be in direct contact with the adjacent interstitial tissue suggesting their formation occurs during tumor cell migration. Moreover, Shinto et al[21] recently suggested that cytoplasmic pseudo-fragments could be used as a marker for an activated budding phenotype that is associated with cell motility and increased invasiveness independent of the extent of budding. Not surprisingly, tumor buds have been shown to over-express proteins involved in extracellular matrix degradation and to under-express adhesion molecules. Previous studies on EMT and events occurring at the invasive tumor front implicate, in particular, the Wingless-INT (WNT) signalling pathway in the process of tumor budding evidenced by increased β-catenin immunohistochemical staining in tumor buds, a concomitant loss of E-cadherin and over-expression of laminin5γ2 along with activation of transcriptional repressors SLUG, and ZEB1[22,23]. Over-expression of urokinase plasminogen activator receptor (uPAR), matrix metalloproteinase-7 and -9 (MMP7, MMP9), matrilysin, CD44, Ep-CAM, and extensive staining of β(III)-tubulin, a major constituent of microtubules, have all been reported[20,23-30] suggestive of the invasion and migration potential of tumor buds. Tumor buds seem to over-express CXCL12, a stromal cell-derived factor whose receptor CXC4 is involved in chemotaxis and angiogenesis[31]. In addition, we recently documented the over-expression of the putative colorectal CSC marker ABCG5 within tumor buds leading to a poorer outcome of patients including those with node-negative disease (Hostettler, World Journal of Gastroenterology, in press). Whether a sub-population of tumor buds may in fact represent malignant stem cells is still an open question which necessitates further investigation.

Since tumor budding appears to play a critical role in the initiation of metastasis, several authors have investigated the potential of this feature to predict dissemination of tumor cells to regional lymph nodes. A significant association between tumor budding and lymph node positivity has been consistently demonstrated correlating with tumor aggressiveness and more advanced TNM stage[32-43]. Tumor budding is frequently associated with poorly differentiated tumors, and with the presence of vascular and lymphatic invasion independently of disease extent[44-48]. Local tumor recurrence and distant metastasis to the lung and liver are also more commonly observed in patients with tumor budding[36,39,48-50] and additionally represent a reproducible prognostic factor in stage II patients[51]. Recently, Suzuki et al[52] found that tumor budding and venous invasion were significant predictors of local and distant metastases in patients with T1 stage colorectal cancers. Xu et al[53] demonstrated an increased rate of tumor budding in colorectal carcinomas with the aggressive micropapillary component. The presence of tumor budding has repeatedly been linked to poor clinical outcome, underlined by the adverse effect on overall survival independently of TNM stage[47,51,54].

Tumor budding is closely linked to tumor growth pattern, a feature described by Jass et al[55] in 1987 which led to the proposal of an alternative prognostic classification system for rectal cancers[55,56]. The diagnosis of either a pushing (or expanding) or infiltrating tumor border configuration can be made at low magnification and is reproducible among pathologists thereby underlining its usefulness as a prognostic indicator (Figure 1C and D)[7]. The pushing tumor border is one in which margins are reasonably well-circumscribed and often associated with a well-developed inflammatory lamina. In contrast, the infiltrative tumor border is characterized by widespread dissection of normal tissue structures with loss of a clear boundary between tumor and host tissues.

Several studies have confirmed that an infiltrative tumor border configuration has a significant adverse prognostic impact in colorectal cancer and may predict local recurrence[57,58]. Our study group has also recently provided evidence for the improved stratification of stage II colorectal cancer patients based on the diagnosis of tumor border configuration. In particular, the 5-year survival rates for patients with stage II tumors decreased substantially from 80% in those with a pushing margin to 62.7% in patients with an infiltrating growth pattern, a survival rate similarly found in patients with stage III disease[59]. Considering that patients with stage III tumors are generally considered for adjuvant therapy[60], the implications of these findings suggest that stage II patients with an infiltrating tumor margin should perhaps be considered for post-operative therapy. The addition of tumor border configuration to TNM stage improved the prognostic classification of colorectal cancer patients by 17.9%. Since the presence of tumor budding can be strongly specific for an infiltrating, rather than a pushing/expanding growth pattern, it is not surprising that loss of cell-adhesion molecule E-cadherin and apoptosis activating factor-1, a pro-apoptotic molecule and over-expression of uPA and uPAR have all been reported as significant predictors of an infiltrating tumor border configuration in colorectal cancer[9,61].

The presence of conspicuous peritumoral lymphocytic (PTL) inflammation, viewed as a distinctive “encapsulating” connective tissue mantle cap at the invasive front of colorectal cancer, is inversely correlated with the presence of tumor budding and positively associated with improved survival[19,65,66]. Jass[8] demonstrated that PTL infiltration in rectal cancer decreased with more advanced Dukes’ stage to 53%, 28% and 13% with Dukes’ A, B and C cases, respectively. In addition, the significantly worsened prognosis in patients lacking PTL inflammation at the tumor border was highlighted, while patients with moderate or pronounced infiltration performed significantly better independently of disease stage. The results have also been confirmed by other study groups[67]. However, the presence of PTL inflammation at the invasive front does not appear to be an independent prognostic factor in patients with this disease[59]. Nonetheless, PTL inflammation seems to be intimately linked with abundant CD8+ tumor infiltrating T-lymphocytes, further implicating tumor immunity in the defense against colorectal cancer.

Most studies to date confirm that a high rate of tumor infiltrating lymphocytes (TILs), in particular those located intra-epithelially characterized by CD4+ and CD8+ tumor-associated antigens are beneficial for patient outcome[68,69]. An abundant TIL count appears to be linked to earlier Dukes’ stage, decreased local recurrence rate following curative surgery and improved overall and disease-free survival time both in non-metastatic and metastatic patients undergoing hepatic resection[10,68-73].

Galon et al[13] evaluated by gene expression profiling and immunohistochemistry, the type, density and location (whether at the invasive margin or the tumor centre) of TILs in a large number of cases. They evaluated CD3, CD8, granzyme B and memory CD45RO T cells and demonstrated a significant independent and positive effect of TILs on both recurrence and survival. Pages et al[15] performed a comprehensive analysis of TILs focusing on early metastatic invasion. They found, by RT-PCR on 75 cases that mRNA levels of CD8, granzyme B and granulysin were significantly greater in patients without vascular emboli, lymphatic and perineural invasion (collectively known as VELIPI) compared to those with these features and that CD45RO+ cells had independent prognostic value[15]. Diederichsen et al[74] showed that a low CD4/CD8 ratio by flow cytometry was an independent prognostic factor for prolonged survival. In addition, Milasiene et al[75] evaluated inter-epithelial CD3, CD4, CD8, CD20 and CD16 and found that increased levels of all these markers, particularly of the natural killer cell marker CD16 led to significantly improved overall outcome[11,75]. Moreover, regulatory T-cells expressing FOXP3+ has been shown to correlate with improved outcome independently of TNM stage[16,76].

In addition to T cells in colorectal cancer, a growing number of studies have demonstrated the clinical impact of dendritic cells, mast cells, macrophages and neutrophils on survival. An improved survival time and a preventative effect of mast cells on local recurrence and distant metastasis in patients with rectal tumors with high mast cell counts have been identified[77-79]. Further, the significant benefit of mast cell number on tumor progression in colorectal cancer was highlighted by Gounaris et al[80] who reported that depletion of mast cells whether by pharmacological means or through generation of chimeric mice with genetic lesions in mast cell development led to remission of existing polyps. Moreover, Halazun et al[81] found that an elevated neutrophil/lymphocyte ratio led to a poorer survival time and higher rate of recurrence in colorectal cancer patients undergoing surgery for liver metastasis.

Dendritic cells are the most potent antigen-presenting cells and as such are now one of the many important tools for tumor immunotherapy. Evidence is accumulating which suggests that the presence of dendritic cells may be of significant benefit to patients with colorectal cancer[82]. Using immunohistochemistry for CD83, Suzuki et al[83] described the presence of mature dendritic cells at the invasive margin of cancer stroma and demonstrated by light and electron microscopy their formation into clusters with lymphocytes, the majority of which were CD45RO+ T cells. They conclude that mature CD83+ dendritic cells at the invasive margin promote T-cell activation for the generation of tumor specific immunity. Using electron microscopy, tumor-infiltrating dendritic cells were found to make contacts among themselves, with TILs and tumor cells. The presence of dendritic cells was found predominantly in early compared to later disease stages and mostly located in tumor surrounding tissue[84]. Dadabayev et al[12] demonstrated that dendritic cells were significantly correlated with intra-epithelial CD4+ and CD8+ lymphocytes. Recently, HLA-DR expressed constitutively on antigen-presenting cells such as dendritic cells and macrophages has also been found to correlate with the presence of TILs and PTLs as well as improved patient outcome[85].