Diagnosis of Submucosal Invasive Colorectal Carcinoma (pT1 Colorectal Cancer): Overview of Histopathological and Molecular Markers to Predict Lymph Node Metastasis of Submucosal Invasive Colorectal Cancer

Fig. 12.1

Representative images of submucosal invasive colorectal cancer (SICRC) at (a) low and (b) high magnifications. (c) Desmin immunostaining. Note that muscularis mucosa is clearly found, although the narrow area is focally ruptured. SID: 470 μm

../images/461529_1_En_12_Chapter/461529_1_En_12_Fig2_HTML.png — Fig. 12.2

Representative images of submucosal invasive colorectal cancer (SICRC) at (a) low and (b) high magnifications. (c) Desmin immunostaining. Note that muscularis mucosa is ruptured. SSID: 1430 μm

12.2.2 Tumor Budding

Tumor budding reflects detachment of tumor cells at the invasive front of CRC and is defined as the presence of single tumor cells or clusters of up to five cells [20, 21], although a cut-off value of four cells has also been used [14]. This difference in cut-off cell number may be important for distinguishing tumor budding from the novel histopathological parameter poorly differentiated cluster (PDC), which is defined as at least five or six tumor cells (discussed later). Tumor budding is assessed at high magnifications and should not be confused with cytoplasmic fragments of tumor cells, which cannot be easily assessed at low magnifications [22]. Many studies have highlighted the prognostic power of tumor budding using different cohorts, scoring systems, and assessment methods. Indeed, numerous studies have shown tumor budding to be a strong and independent predictor of lLNM of SICRC [23–26]. Many studies therefore strongly recommend evaluating tumor budding in addition to other histopathologic predictors of LNM of SICRC, including poor tumor differentiation, lymphovascular invasion, and SID >1000 μm [14, 21]. Tumor budding can be subclassified according to the number of tumor buds as low-grade (0–9 buds) or high-grade (>10 buds) [27]. The agreement between tumor bud scoring on tissue sections stained with hematoxylin and eosin versus a pan-cytokeratin antibody is poor. Hematoxylin and eosin staining is recommended over pan-cytokeratin staining for assessment of tumor budding in routine pathological diagnosis [14, 28]. Representative images of tumor budding are shown in Figs. 12.3 and 12.4.

../images/461529_1_En_12_Chapter/461529_1_En_12_Fig3_HTML.jpg — Fig. 12.3

Low-grade tumor budding at the invasive front

../images/461529_1_En_12_Chapter/461529_1_En_12_Fig4_HTML.jpg — Fig. 12.4

High-grade tumor budding at the invasive front

12.2.3 Poorly Differentiated Clusters

A recent study showed that the presence of PDCs, defined as at least five cancer cells with no gland formation, may predict the metastatic potential of SICRC [29, 30]. In addition, assessment of PDCs may be more reproducible than that of tumor budding. Barresi et al. showed that PDC identification combined with SID measurement helps identify most SICRC cases with lymph node metastases [30]. PDCs occur sequentially to tumor budding, according to the definitions of each process [29, 30]. The difference in the definitions of tumor budding and PDCs is based on the number of tumor cells within the tumor cluster [14, 29], and it is occasionally difficult to differentiate tumor budding from PDCs. The usefulness of identifying tumor budding to predict SICRC is limited by its low standardization and high interobserver variability [14, 29]. Even though tumor budding has been considered a strong indicator of LNM of SICRC, PDCs may also be a useful predictor [29, 30]. Ryan et al. showed that although PDC grade and extramural vascular invasion are independent predictors of LNM, tumor budding is the best predictor of disease-free survival of patients with mismatch repair-deficient SICRC [31]. If tumor budding cannot be assessed accurately, the PDC grade may be used as a prognostic surrogate [31]. According to previous studies, tumor budding and PDC may be the best histological markers predicting LNM [14, 21, 29, 30].

12.2.4 Lymphovascular Invasion

Despite being considered one of the best histological predictors of LNM of SICRC [32], lymphovascular invasion is not necessarily correlated with LNM in routine pathological diagnosis. Although Yasuda et al. showed that vascular invasion, tumor budding, and SID were significant risk factors for LNM, lymphovascular invasion was not predictive of LNM [33, 34]. According to their study, a combination of histological markers can be used to identify patients requiring additional surgery after endoscopic resection [33]. However, it is not clear what combination of histological markers is most powerful.

12.2.5 Desmoplastic Reaction

Desmoplastic reaction (DR), which is characterized by stromal fibrosis of invasive carcinoma, is observed around cancer cells that have invaded beyond the muscularis mucosae [35]. DR is defined as stromal fibroblasts expressing desmin but not smooth muscle actin (SMA) [35]. Whereas DR is not an important risk factor for LNM, it is a useful histological marker to assess tumor invasion into the submucosal layer [35].

12.3 Biomarkers

Biomarkers that predict LNM and facilitate development of better therapeutic strategies are needed [36–38]. Makino et al. examined whether TP53 overexpression is a novel genetic marker predictive of LNM of early invasive CRC [39]. According to recent opinion, however, predicting LNM by TP53 overexpression alone is difficult.

12.3.1 MicroRNAs

MicroRNAs (miRNAs) are small noncoding RNA molecules of 20–25 nt (nucleotides) that regulate gene expression via transcriptional repression and mRNA degradation [40]. Recent studies have shown that altered miRNA expression is a key event closely associated with CRC [12, 40]. Recently, Ozawa et al. used data from The Cancer Genome Atlas to identify five miRNAs (miR-32, miR-181B, miR-193B, miR-195, and miR-411) with significant differential expression in SICRC with versus without LNM [12]. The expression signature of these five miRNAs distinguished CRCs with from those without LNM, with an area under the receiver operating characteristic curve of 0.77. They concluded that the expression of these five miRNAs can be used to identify high-risk SICRC with greater accuracy compared with currently used pathologic features.

Moreover, Fujino et al. demonstrated that the expression levels of miR-125a-5q, miR-125b, miR-155, miR-342-3p, miR-100, and miR-30a-5p were significantly lower in patients with than in those without metastatic CRC [41]. Among these miRNAs, downregulation of miR-100 and miR-125b was closely associated with LNM of SICRC. In particular, miR-100 promoted metastasis by upregulating the expression of targeting the messenger RNA of mTOR, IGF1R, Fas, and XIAP. Thus, the authors suggested that miR-100 and miR-125b may be novel biomarkers of LNM of SICRC.

12.3.2 Anticipated New Markers

Mori et al. examined the expression pattern of mRNAs using a comparative proteomics approach based on isobaric tags for relative and absolute quantification to identify novel biomarkers predicting LNM in CRC patients [42]. In their multivariate analysis, high ezrin protein and mRNA expression in CRC samples was an independent predictor of LNM.

Somatic copy number alterations (SCNAs) are a limiting step in tumor invasion or metastasis during tumor development and have helped elucidate tumor invasion and metastasis [42]. We investigated the role of SCNAs in colorectal adenoma, intramucosal cancer, and early invasive CRC using hierarchical cluster analysis [43]. Accumulation of SCNAs was seen in early invasive CRC compared with colorectal adenoma and intramucosal cancer, suggesting that SCNAs play a major role in invasion of the submucosal layer [43, 44]. Although SCNA cannot be used as a practical parameter that is not easily measured, SCNA identification may help predict LNM of SICRC. Further studies are needed in the near future.

12.3.3 Cancer-Associated Fibroblasts and Epithelial–Mesenchymal Transition

A recent study showed that tumor budding is closely associated with epithelial–mesenchymal transition (EMT), which is characterized by transformation of epithelial clusters into mesenchymal cells [45]. However, Yamada et al. suggested that tumor budding located at the tumor invasive front does not result from EMT [46]. Although there is controversy regarding the association between tumor budding and EMT, the hypothesis that tumor budding is linked to EMT has been supported by many investigators [45–47]. Signals derived from the tumor microenvironment characterized by small cell clusters and the surrounding mesenchymal cells at the invasive front may play a significant role in the development of a pro-budding phenotype [45]. Several transcription factors, including Snail, Slug, Twist, and ZEB1, have been shown to induce E-cadherin expression directly by targeting its promoter [45, 48]. It has been hypothesized that mesenchymal cells expressing these EMT-related proteins contribute to tumor invasion and metastasis [46].

Recent studies have shown that cancer-associated fibroblasts (CAFs) play an important role in tumor invasion and metastasis, as well as carcinogenesis [49, 50]. CAF-related proteins are expressed in fibroblasts involved in the invasion process [49, 50]. Although many CAF-related proteins, including α-SMA, CD10, podoplanin, fibroblast-specific protein 1, platelet-derived growth factor α and β, adipocyte enhancer-binding protein 1, and the EMT-related proteins ZEB1 and TWIST1 [49, 50] have been identified as target markers of tumor invasion, the associations of these proteins with LNM of SICRC have not been examined. In our previous study, we examined the expression patterns of CAF- and EMT-related proteins in SICRC using hierarchical cluster analysis and classified their expression patterns into four subgroups [49]. Subgroup 1 was characterized by high expression of α-SMA, AEBP1, and ZEB1 and low expression of CD10, podoplanin, and TWIST1; subgroup 2 by high expression of α-SMA, CD10, podoplanin, FSP1, AEBP1, ZEB1, and TWIST1; subgroup 3 by significantly high expression of α-SMA, podoplanin, AEBP1, and ZEB1; and subgroup 4 by high expression of α-SMA, podoplanin, FSP1, AEBP1, ZEB1, and TWIST1 and low expression of CD10. In both univariate and multivariate analyses, subgroup 2 was correlated with LNM (p < 0.01). Next, we examined the associations between individual biomarkers and lymph node metastasis using multivariate analysis and found that high CD10 expression was associated with the presence of LNM, although the association did not reach statistical difference (p = 0.06). We suggested that the expression patterns of CAF- and EMT-related proteins at the front of submucosal invasion help predict LNM of SICRC.

12.4 Conclusion

We present an overview of the prediction of LNM in patients with SICRC. Histological factors including a SID >1000 μm, high histological grade (poorly differentiated adenocarcinoma and unfavorable histology), and lymphovascular invasion are traditionally used independent risk factors for LNM in patients with SICRC; however, overtreatment cannot be prevented using these criteria. Tumor budding, which is defined as small clusters consisting of less than five or six tumor cells, and PDCs are also important histological indicators of LNM of SICRC. On the other hand, it is hoped that novel biological predictors, rather than histological parameters, for SICRC are identified. To this end, several miRNAs have been found to be predictive of LNM of SICRC, including miR-32, miR-181B, miR-193B, miR-195, miR-411, miR-100, and miR-125b. Finally, the expression patterns of CAF- and EMT-related proteins may be strong candidate markers predicting LNM of SICRC. Additional novel markers are needed in the future.