Materials and methods

We examined 55 surgically resected invasive gastric adenocarcinomas that were confined to the gastric wall (not invading the adjacent organs) and their normal tissues from the surgical pathology files of the first department of pathology, Kochi Medical School and its affiliated hospitals from 1994 to 1999. The definitions used for histological classification were based on the criteria of Lauren¹³; 15 D-type and 40 I-type tumours were identified. Depth of tumour invasion was classified as submucosa (27 tumours; six D-type, 21 I-type), muscularis propria (13 tumours confined to the muscularis propria; five D-type, eight I-type), and subserosa (15 tumours; four D-type, 11 I-type). We classified the tumours confined to the submucosa as early cancers, and those invading the muscularis propria and subserosa as advanced cancers.

Immunohistochemical studies were performed by the streptavidin–biotin method using the Histofine SAB-PO(M) kit (Nichirei, Tokyo, Japan). Three monoclonal antibodies against CD34, CD31, and low molecular weight cytokeratins (LMW-CKs; cytokeratins 8 and 18) were used. Table 1 lists the monoclonal antibodies and staining procedures. We examined immunoreactivity for CD31 in all of the tumours and their normal tissues, to distinguish CD34 positive stromal cells from vascular endothelial cells, which are positive for both CD34 and CD31.¹⁴ Vascular endothelial cells were used as the internal positive control of immunostaining for CD34 and CD31. (We did not apply digital subtraction.) Similar to our recent study regarding gastric cancer, immunostaining for LMW-CKs was also performed, to recognise the tumour border in every specimen examined.¹⁵

After immunostaining, we examined CD34 positive stromal cell distribution in gastric adenocarcinomas and normal tissues of the 55 cases. Regarding the tumours, the number of CD34 positive stromal cells was classified into two groups, namely: (+), tumours having CD34 positive stromal cells in the tumour stroma (fig 1A; note: fig 1B, C shows the expression of CD31 and LMW-CKs, respectively, at the same site as fig 1A); and (−), tumours having no CD34 positive stromal cells in the tumour stroma (fig 2). Statistical analysis was carried out using Fisher's exact probability test and p values < 0.05 were considered to be significant.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

CD34 positive stromal cells in diffuse-type advanced gastric adenocarcinoma tissue. Staining for (A) CD34, (B) CD31, and (C) low molecular weight cytokeratins (LMW-CKs). Large numbers of CD34 positive stromal cells are detected in the tumour stroma. Immunoreactivity for CD31 is shown to distinguish CD34 positive stromal cells from vascular endothelial cells. Immunoreactivity for LMW-CKs is shown to recognise the diffuse-type gastric adenocarcinoma cells.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

CD34 immunostaining in intestinal-type advanced gastric adenocarcinoma tissue. CD34 is positive in vascular endothelial cells; no CD34 positive stromal cells are seen.

Results

Table 2 summarises the results.

In all of the 55 cases examined, CD34 positive stromal cells were mainly in the perivascular area of the normal submucosa, muscularis propria, and subserosa. CD34 positive stromal cells were not detected in the lamina propria but they were distributed in the areas adjacent to the muscularis mucosa. Results concerning the normal gastric wall were the same as the recent report by Kim et al,¹⁶ who found that CD34 was expressed in stromal cells around vessels and muscle bundles in the gastric wall.¹⁶

In the 15 D-type gastric adenocarcinomas examined, eight tumours had CD34 positive stromal cells in the tumour stroma, whereas the remaining seven tumours had no CD34 positive stromal cells in the tumour stroma (table 2). All of the eight tumours were advanced cancers, and had a large number of CD34 positive stromal cells (fig 1A; note: fig 1B, C shows the expression of CD31 and LMW-CKs, respectively, at the same site as fig 1A). Eight of the nine advanced cancers had CD34 positive stromal cells in the tumour stroma classified as (+), whereas all of the six early cancers had no CD34 positive stromal cells and were classified as (−) (p = 0.0014).

In contrast, all 40 I-type gastric adenocarcinomas examined had no CD34 positive stromal cells classified as (−), regardless of tumour depth (fig 2).

Regarding the 28 advanced cancers, eight of the nine D-type adenocarcinomas were classified as (+), whereas all of the 19 I-type adenocarcinomas were classified as (−) (p = 0.0000029).

Discussion

CD34 positive stromal cells were absent in the tumour stroma of well and moderately differentiated adenocarcinomas of the colorectum.¹² However, in our present study, we detected CD34 positive stromal cells in the D-type advanced gastric adenocarcinomas. To our knowledge, this is the first report to detect large numbers of CD34 positive stromal cells in malignant epithelial tumour stroma.

CD34 positive stromal cells, namely dendritic interstitial cells, are distributed in the connective tissue surrounding mammary acini,⁴ salivary gland acini,⁵ thyroid follicles,⁶ hair follicles,⁷ sweat glands,⁸ and endocervical and deep endometrial glands of the uterus,⁹ whereas no CD34 positive stromal cells are detected in the lamina propria of the colorectum¹² and stomach. In the colorectum¹² and stomach, CD34 positive stromal cells are found only in the submucosa, subserosa, muscularis propria, and muscularis mucosa. These results suggest that CD34 positive stromal cells are essential for the maintenance of gastrointestinal mesenchymal elements including smooth muscles and vessels, but not for normal gastrointestinal epithelial cell maturation and proliferation.

During haematopoietic differentiation, the expression of CD34 decreases and terminally differentiated cells do not express CD34.¹⁷ Accordingly, in view of the distribution of CD34 positive stromal cells, the stroma of D-type gastric adenocarcinoma invading the muscularis propria and subserosa may be immature, whereas that of I-type gastric adenocarcinoma may be mature. Further studies including molecular and cell biological analyses are needed to confirm the biological meaning of CD34 expression.

D-type gastric carcinoma-like carcinomas are also reported in the colorectum, although at a very low frequency.¹⁸ Further examination of CD34 expression in tumour stromal cells should be performed in malignant neoplasms of other organs showing a D-type gastric carcinoma-like growth.

In conclusion, D-type advanced gastric adenocarcinomas had large numbers of CD34 positive stromal cells in the tumour stroma, whereas D-type early gastric adenocarcinomas had no CD34 positive stromal cells in the tumour stroma. In contrast, regardless of tumour depth, I-type gastric adenocarcinomas had no CD34 positive stromal cells in the tumour stroma. These results suggest that CD34 expression is associated with progression of diffuse carcinomas, and that absence of this expression is also seen in intestinal carcinomas that are progressing. To elucidate the clinical relevance of these immunohistochemical findings, further clinicopathological investigations are needed.