Numerous previous studies have investigated the clinicopathological characteristics of early colorectal cancer types. As a result, a consensus is forming in regards to the clinical treatment of T1 cancer types (1–3). According to the Colorectal Cancer Treatment guidelines, T1 colorectal cancer types are recommended for endoscopic treatment if they are papillary adenocarcinoma or tubular adenocarcinoma with a vertical invasion of ≤1,000 µm and without any lymphovascular invasion (3).

T2 cancer types, which invade deeper layers as compared with T1 cancer types, are considered an early stage of advanced cancer. T2 cancer types are considered to be the developmental transition from early to advanced cancer. Although several retrospective studies on the association between morphology and tumor characteristics of T2 colorectal cancer types have been performed in Japan, there remains no consensus on the subject (4–8). This may be due to advanced cancer types exhibiting highly modified morphologies as compared with early cancer types.

Nevertheless, numerous lesions are thought to clearly derive from depressed type early cancer (Fig. 1). The present study proposed novel morphological categories for T2 colorectal cancer types, based on the growth and development route. The primary aim of the present study was to clarify the association between morphology and malignant potential of T2 colorectal cancer types.

The present study proposed novel morphological classifications of T2 colorectal cancer with reference to growth and development route, and subsequently investigated their respective features.

Since numerous previous studies have investigated the growth and development route of T1 stage cancer, a consensus is forming on this issue (1,2,9,10). Growth and development classifications for depressed, flat and protruded types were constructed based on their various morphologies (Fig. 2). By acontrast, several previous studies on T2 cancer types have been previously performed (4–8); however, limited debate on this topic has occurred compared with that of T1 cancer types.

The morphologies containing depressed areas on tumor surfaces were dominant among T2-T4 cancer types, thus in the clinical setting they were conventionally treated as an identical morphology, often termed ‘Bormann type 2’ cancers in Japan. However, biological difference among the ‘Bormann type 2’ cancers were considered to exist, thus, several morphological classifications were proposed by Japan and the west. George et al (5) classified T2 cancers into polypoid and non-polypoid types based on whether the marginal mucosa is elevated or not. This resulted in 45% of tumors being classified as polypoid, half of which included adenomatous components within the lesions, and 48% were classified as non-polypoid, which rarely included adenomatous components. The remaining 7% included morphologies that exhibited both polypoid and non-polypoid characteristics. Non-polypoid tumor types often exhibited wild-type K-ras and had a significantly higher rate of postoperative recurrence compared with polypoid tumors, which suggested that non-polypoid tumors were more malignant. In a previous study of T2 colorectal cancer tumor types ≤2 cm, Goi et al (6) reported that 63% of small, advanced colorectal cancer types were non-polypoid, while ~25% of normal advanced tumors of ≥2 cm were non-polypoid tumors. The non-polypoid vs. polypoid classifications used by these researchers assumed that non-polypoid lesions were associated with the de novo cancer route (10) and that polypoid lesions were associated with the carcinogenic route that followed the adenoma-carcinoma sequence (9). Nevertheless, since these classification systems were focused on remnant intramucosal lesions, certain researchers may feel they should be restricted to use on early-stage cancer types. Another problem is that these non-polypoid vs. polypoid classifications were based on resected pathological specimens, thus, it cannot be used for treatment determination.

In consideration of these problems, the present study classified tumor types focusing on growth and development route based on preoperative endoscopic findings. Specifically, the laterally spreading and protruded types of lesions were firstly classified as groups B and C, respectively, because each were clearly derived from laterally spreading type and protruded type early lesions. The remaining depressed and ulcerative types of lesions were classified based on whether the margins have normal mucosal tissue or not. Depressed type lesions, which are thought to be de novo cancer-derived, were placed into their own category.

Of all the types in the present classification system, depressed type T2 cancer showed characteristic clinicopathological features. They exhibited a significantly smaller tumor size and higher incidence of lymphovascular invasion. These features have extremely strong resemblance to the clinicopathological features of depressed type T1 cancer (1,2). However, no significant differences were observed in terms of lymph node metastasis and distant metastasis. This may be a result of the fact that positive rates for T2 cancer lymph node metastasis and distant metastasis were relatively low at 25.0 and 3.2%, respectively. Therefore, this issue requires further study with a larger number of cases for analysis.

In conclusion, compared with the other types, depressed type T2 cancer exhibited a smaller tumor size, higher rate of lymphatic involvement and venous involvement, which suggested a more malignant nature of these lesions.