METHODS

Patients

Eighty six colorectal tumours (64 tubular or tubulovillous adenomas and 22 carcinomas extending no further than the upper submucosal region) were resected surgically or endoscopically from 59 patients (44 men and 15 women; age range, 40–83) at Niigata University Hospital. Histological classification was determined according to the criteria of the Japanese Research Society for Cancer of the Colon and Rectum.¹⁴ Patients with familial adenomatous polyposis, hereditary non-polyposis colorectal cancer, or inflammatory bowel disease were excluded from our study.

Assessment of macroscopic form

The extirpated specimens were stretched on to styrofoam and immersed in 10% buffered formalin. The macroscopic form was determined according to the following three types: polypoid, over 3 mm in height; flat, 3 mm in height or less; and depressed, lower than the surrounding normal mucosa. The specimens were then embedded in paraffin wax, and histological diagnoses were made on routine haematoxylin and eosin stained sections.

Immunohistochemistry

We applied the streptavidin–biotin, technique using a Histofine SAB-PO kit (Nichirei Corporation, Tokyo, Japan) for immunostaining. Serial sections heated in a microwave oven in 0.01M sodium citrate buffer (pH 5.9) for five cycles of three minutes were incubated with the anti-Ki-67 antibody (MIB-1; Zymed Laboratories, San Francisco, USA), anti-Bcl-2 antibody (clone 124; Dako, Copenhagen, Denmark), anti-Bak antibody (TC102; Oncogene Research Products, Cambridge, USA), or anti-p53 antibody (PAb1801; Oncogene Research Products). The colour was developed with 3,3`-diaminobenzidine (DAB) solution. The sections were counterstained with haematoxylin and mounted. Negative controls were established by replacing the primary antibodies with phosphate buffered saline. As internal positive controls, the proliferative zone cells of normal crypts were used for Ki-67, lymphocytes within the lamina propria for Bcl-2, and epithelial cells of normal crypts for Bak. For p53, a colon cancer sample with a high degree of nuclear immunoreactivity for this oncoprotein was used as a positive control.

Detection of apoptosis

DNA fragmentation associated with apoptosis was identified by means of the 3` end DNA labelling (TUNEL) method using the ApopTag in situ apoptosis detection kit (Oncor, Gaithersburg, USA), according to the method described by Gavrieli et al.¹⁵ The steps were performed according to the manufacturer's instructions, with minor modifications. The DAB reaction, counterstaining, and setting up of negative controls were performed using the same procedures as those used for immunohistochemistry. Positive controls were obtained by DNase treatment of the sections.

Evaluation

Cytoplasmic staining was considered positive for Bcl-2 and Bak. A semiquantitative evaluation was performed, the reproducibility of which was confirmed by two of the authors. Because various degrees of staining were found, the proportion of positive to whole tumour cells on each section was scored as follows: 0, < 1%; 1, 1–25%; 2, 26–50%; 3, 51–75%; and 4, > 75%. For p53, only nuclear staining was considered positive; it was evaluated by a slightly modified method (previously described by Kobayashi and colleagues⁸) as follows: –, negative; 1+, sporadic positive; and 2+, diffuse positive.

Cells showing nuclear staining were considered positive for Ki-67, and those with homogeneous or granular nuclear staining, in addition to nuclear fragments located inside the epithelial cells, were defined as apoptosis positive. Positive cells to a total of at least 1000 tumour cells were counted in several fields of specimens in which several longitudinally sectioned crypts could be evaluated. The apoptotic index (AI) and Ki-67 index (KI) were defined as the percentage of apoptosis and Ki-67 positive cells, respectively.

Statistical analysis

The unpaired t test was used to compare tumour sizes in each group. The AIs, KIs, and the Bcl-2 and Bak scores in groups classified for macroscopic form were compared using the Mann-Whitney U test. The χ² test was used to compare the results of p53 staining. p Values < 0.05 were regarded as significant.

RESULTS

Table 1 details the clinicopathological features of each macroscopic form of the tumours investigated. The mean size of polypoid tumours was significantly greater than that of flat and depressed ones (p < 0.01 and p < 0.05, respectively). The ratio of carcinoma in depressed tumours was higher than that in polypoid and flat ones.

Ki-67

Increased numbers of Ki-67 positive cells were seen in neoplastic areas, mainly in the upper part of the crypts (superficial pattern, fig 1B), or throughout the crypts (diffuse pattern, fig 2B). The distribution of Ki-67 positive cells differed slightly among the three macroscopic forms. The superficial pattern was shown in 60.7%, 87.2%, and 72.7% of polypoid, flat, and depressed tumours, respectively. Tumours with a diffuse pattern of Ki-67 positivity were significantly larger (p < 0.01) than those with a superficial pattern (data not shown).

Apoptotic cells

Apoptotic cells were found throughout the neoplastic crypts and were also found sporadically in the tumour epithelium (fig 1C). The AIs tended to decrease with the progression of the tumours (2.48, 2.06, and 1.76 in adenoma, intramucosal carcinoma, and submucosal carcinoma, respectively), but the difference was not significant.

Bcl-2 and Bak

The proportion of Bcl-2 positive cells varied in each tumour. The following scores were obtained among the 86 tumours: a score of 0 was found in four cases; a score of 1 in 15 cases; a score of 2 in 17 cases; a score of 3 in 15 cases; and a score of 4 in 35 cases (figs 1D and 2C). The positive cells were mainly located in the basal side of the tumour in cases with low scores. The percentage of Bak positive cells was high in most of the tumours (fig 1E), having a score of 4 in 63 of the 86 tumours.

p53

Table 2 shows the p53 immunoreactivity in each macroscopic form. The overexpression of p53 did not correlate with any of the macroscopic forms.

Relation between macroscopic form and apoptosis

The KIs and the Bak scores for each macroscopic form were similar (table 3 and fig 3). In contrast, the mean AI of depressed tumours was significantly lower than that of polypoid (p < 0.01) and flat (p < 0.05) type tumours. The ratio of KI to AI (KI/AI) in polypoid tumours was significantly lower than that in the other types of tumour (p < 0.01). The Bcl-2 score became significantly lower as the tumours became flatter or took on a depressed form.

DISCUSSION

In our present study, we focused on the possible role of cell proliferation and loss in early colorectal tumorigenesis, particularly in terms of macroscopic form. There are several proteins in the Bcl-2 family, and their expression differs according to the histological type of the tumour.¹⁶ Among them, antiapoptotic Bcl-2 and proapoptotic Bak are major proteins that have been reported to show characteristic expression patterns in colorectal neoplasia.^10–12,17 To clarify how they are involved in the morphogenesis of colorectal adenomas and early carcinomas, we investigated the immunohistochemical expression of these two proteins, together with Ki-67 and apoptosis.

“Antiapoptotic Bcl-2 and proapoptotic Bak are major proteins that have been reported to show characteristic expression patterns in colorectal neoplasia”

We found that both the AI and the Bcl-2 score were significantly higher in polypoid tumours than in the other two macroscopic forms. Although a direct inverse relation between Bcl-2 expression and apoptosis (which would be expected in view of the fact that the Bcl-2 protein can block apoptosis)¹⁸ has been found in some organs/tissues,^19,20 this may not hold true during the morphogenesis of colorectal tumours, as has been reported in the lung²¹ and ovary.²² The results of our present study showing that the KI/AI ratio is lowest in polypoid tumours suggests that polypoid adenomas and carcinomas do not grow simply because of increased proliferation and reduced apoptosis. As hypothesised for gastric neoplasia,⁶ the diffuse expression of Bcl-2 and a superficial distribution of proliferating cells may contribute to the formation of a polypoid configuration in colorectal tumours. This is supported by our results of a higher Bcl-2 score in polypoid tumours, 60.7% of which showed a superficial distribution of Ki-67 positive cells, than in the other types of tumour.

Although flat and/or depressed tumours of the large intestine have became apparent in recent years, the mechanism by which the non-polypoid configuration is formed is unknown. Unlike previous reports,^9,13 the proliferative index did not vary significantly with the different macroscopic forms or histopathological scores in our study. This may be because the tumours investigated in our study were smaller and earlier than those in previous reports. Consequently, our results might suggest that, in the extremely early phase of colorectal tumorigenesis, the relative increase of the KI/AI ratio following a significant reduction of AI contributes to the growth of flat and, especially, depressed tumours with high malignant potential, perhaps through some unknown mechanisms unrelated to Bcl-2,⁹ or through the accumulation of genetic changes.

We used the TUNEL technique in our present study to identify apoptotic cells. One known shortcoming of this method is that any type of DNA fragmentation, including that of necrotic cells, might be detected.²³ However, because extensive necrosis caused by severe inflammation or ischaemia rarely occurs in the background mucosa without inflammatory bowel disease, we think that there were very few necrotic cells in the very small tumours investigated in our study. Furthermore, to exclude contaminating necrotic cells as much as possible, the positive cells that were sloughed off from the epithelial tubules were not evaluated.⁹

The wild-type p53 protein prevents cells that have DNA damage from entering the cell cycle, which results in apoptosis in some cells,²⁴ and the p53 gene often undergoes mutation in the transition from large adenoma to carcinoma of the colorectum.²⁵ Our previous report indicated that sporadically expressed p53 protein is probably wild type and induces apoptosis in response to active cell proliferation.⁸ Because most (97.7%) of the tumours investigated in our present study showed no or sporadic expression of p53, the activation of apoptosis and those proteins involved in the apoptotic pathway in early colorectal tumorigenesis may be mediated directly by wild-type p53. However, more detailed molecular analysis would be necessary to determine the involvement of p53.²⁶

In conclusion, to investigate the possibility that cell proliferation and death are related to the early morphogenesis of colorectal neoplasia, we demonstrated that the frequency of apoptosis and Bcl-2 expression differed significantly according to the macroscopic form. Further investigations may be required once more detailed mechanisms of apoptosis are elucidated.