| Japanese Journal of Clinical Oncology | Pages |
Macroscopic Features at the Deepest Site of Tumor Penetration Predicting Liver Metastases of Colorectal Cancer
Introduction
Materials and Methods
Patients
Macroscopic and Histopathological Examination
Statistical Analysis
Results
Characteristics of Streak Type Tumor
Predictive Value for Liver Metastasis
Cumulative Survival Rate
Discussion
Acknowledgment
References
Macroscopic Features at the Deepest Site of Tumor Penetration Predicting Liver Metastases of Colorectal Cancer
Liver metastasis is the gravest prognostic factor in colorectal cancer. To identify a reliable indicator for liver metastasis, we evaluated macroscopic features and seven established histopathological findings at the cut section containing the deepest penetration using univariate and multivariate analyses in 417 colorectal cancers. Macroscopic features were divided into two types, streak type and non-streak type, according to the presence or absence of white streak(s) at the advancing margin of tumor invasion. Streak type was observed in 109 patients (26%). The frequency of liver metastasis in streak type tumors (56%) was significantly higher than that in non-streak type tumors (13%) (p < 0.001). The white streak corresponded histologically with cancer cells showing focal dedifferentiation with marked stromal and perivascular fibrosis extending towards the serosa or adventitia. In 343 curatively treated patients, univariate analysis showed that recurrent liver metastasis was significantly associated with macroscopic features, venous invasion, focal dedifferentiation and lymph node metastasis. Multivariate analysis disclosed that macroscopic features and lymph node metastasis were independent indicators of liver metastasis. These macroscopic features, corresponding histologically to stromal behavior against invading cancer cells, are a simple and useful indicator of liver metastasis of colorectal cancer.
Introduction
Liver metastasis is found in approximately one-third of patients with colorectal cancer and is an important cause of death (1,2). Prediction of liver metastasis is important since high-risk patients may benefit from systemic or regional chemotherapy after operation (3-5).Liver metastasis of colorectal cancer correlates with several histopathological findings including lymph node metastasis (6), deeper tumor penetration (7), venous invasion (8,9), differentiation of tumor invasive region (10,11), growth pattern (12,13) and host inflammatory cell reaction (14,15) and molecular biological properties including laminin (16), CD44 variants (17), matrix metalloproteases (18) and sialyl-Lewis X (19). However, these factors have rarely been utilized clinically, because of their low specificity in identifying liver metastasis and the complexity and impracticality of examinations.
Recently, we reported that `focal dedifferentiation' (20), a histopathological finding at the invasive front of primary colorectal cancer, was associated with liver metastasis. This finding was mainly defined by the morphology of cancer cells dissociating from the glandular structure and invading with solitary or trabecular nests, whereas stroma around focal dedifferentiation also frequently exhibited desmoplastic features with marked fibrosis and neovascularization, which suggested the activation of stromal reaction against invading cancer cells. Consequently, an overall estimate of morphological changes of both cancer cells and stroma may predict more accurately the metastatic potential of a primary tumor than an estimate of only cancer cells.
To estimate more accurately an overall feature formed by both cancer cells and stroma, we investigated the macroscopic features at the cut section of primary tumors. In examining and evaluating a large number of patients by multivariate analysis, we found a characteristic macroscopic feature at the tumor invasive region, which was a more simple and reliable indicator of liver metastasis than established histopathological factors of colorectal cancer.
Materials and Methods
Patients
A total of 553 patients who underwent surgery for colorectal cancer at the National Cancer Center Hospital between July 1984 and December 1988 were originally reviewed. Of these, 35 patients with synchronous or metachronous multiple cancers, 12 with familial adenomatous polyposis coli or chronic inflammatory bowel disease, 28 on whom there was incomplete clinical information of at least 5 years' follow-up, 33 who died of other diseases and 28 who underwent polypectomy were excluded from this study. A total of 417 patients were finally entered into the study. Of these eligible patients, 37 had simultaneous liver metastases at the time of surgery and 343 were curatively treated cases, of whom 65 developed liver metastasis during their follow-up period.
The diagnosis of liver metastasis was confirmed by ultrasonography, computed tomography (CT), magnetic resonance imaging (MRI) or histological examination on laparotomy.
Macroscopic and Histopathological Examination
Using color photographs of cut sections containing the deepest portion of tumor penetration, macroscopic findings were divided into two types, streak type and non-streak type (Figs 1 and 2). Streak type was featured by an invasive pattern with one or more white streak(s) extending more than 1 cm beyond the muscularis propria towards the serosa or adventitia. All samples were examined independently by two observers (M.I. and A.O.) and any differences were resolved by a joint review.
Figure 1. Macroscopic features at the cut section containing the deepest site of tumor penetration in colorectal cancer. Streak type is defined as an invasive pattern showing one or more white streak(s) (arrows) extending beyond the muscularis propria toward the serosa or adventitia. Non-streak type is defined as an invasive pattern without any white streak.
Figure 2. Color photographs at the cut section containing the deepest point of tumor penetration after fixation in 10% formalin solution (left) and their sections with HE staining (right). Streak type (a, b) and non-streak type (c) are shown. Seven histopathological variables were examined: histological type, depth of tumor penetration, lymphatic invasion, venous invasion, focal dedifferentiation (20), inflammatory cell infiltration and lymph node metastasis. The location of venous invasion was divided into two groups: intramural (submucosal and muscular layer) and extramural (beyond the muscular layer). Cases in which venous invasion was observed in both the intramural and extramural layers were classified in the extramural group.
Statistical Analysis
An unconditional logistic regression analysis was performed to assess the univariate and multivariate effects of various parameters on liver metastasis (21). Histological type was evaluated as a nominal variable and other variables were evaluated by single-order variables in the model, as shown in the codes in Table 3. This model selection was based on the forward stepwise procedure (enter limit 0.05). Survival rates were calculated by the Kaplan-Meier method (22) and survival curves were compared by the log-rank test (23).
Results
Characteristics of Streak Type Tumor
Of 417 patients, 109 (26.1%) had streak type tumor and 308 (73.9%) had non-streak type tumor. The frequency of liver metastasis was 56.0% in streak type tumor and 13.3% in non-streak type tumor, showing a statistically significant difference (p < 0.001, Table 1). The streak type tumor was characterized by deeper tumor penetration (p < 0.05), frequent venous invasion (p < 0.05) and high degree of focal dedifferentiation (p < 0.05) (Table 2).
Table 1
| Macroscopic feature | Liver metastases | p value | |
| Positive (n = 102) |
Negative (n = 315) |
||
| Streak type (n = 109) | 61 (56.0%) | 48 (44.0%) | |
| <0.001 | |||
| Non-streak type (n = 308) | 41 (13.3%) | 267 (86.7%) | |
Table 2
| Variable | Macroscopic feature | p value | |
| Streak (n = 109) |
Non-streak (n = 308) |
||
| Location | |||
| Colon | 50 | 141 | N.S. |
| Rectum | 59 | 167 | |
| Size (cm) | |||
| <5.0 | 55 | 160 | N.S. |
| [ge]5.0 | 54 | 148 | |
| Histological type | |||
| Well | 50 | 149 | |
| Moderate | 56 | 138 | N.S. |
| Poor | 2 | 15 | |
| Others | 1 | 6 | |
| Depth of tumor penetration | |||
| Within muscularis propria | 2 | 51 | |
| Beyond muscularis propria but not involving serosa | 55 | 132 | <0.05 |
| Beyond muscularis propria involving serosa | 52 | 125 | |
| Lymphatic invasion | |||
| Absent | 38 | 126 | N.S. |
| Present | 71 | 182 | |
| Venous invasion | |||
| Absent | 47 | 197 | |
| Intramural | 21 | 66 | <0.05 |
| Extramural | 41 | 45 | |
| Focal dedifferentiation | |||
| None or mild degree | 54 | 210 | <0.05 |
| Moderate or severe degree | 55 | 98 | |
| Host inflammatory cell infiltration | |||
| Mild degree | 83 | 208 | N.S. |
| Moderate or severe degree | 26 | 100 | |
The white streak, which was macroscopically observed at the invasive region of streak type tumor, corresponded histopathologically to marked stromal and perivascular fibrosis extending towards the serosa or adventitia (Fig. 3). In approximately one-third of cases with streak type tumor, cancer cells invaded the extramural vein in the white streak, filling the vein and adhering to the vein wall. In these regions, invading cancer cells frequently showed focal dedifferentiation, exhibiting desmoplastic stromal reaction, where fibroblasts and endothelial cells with enlarged nuclei and basophilic cytoplasm were found (Fig. 4).
Predictive Value for Liver Metastasis
In 343 patients undergoing curative surgery, univariate analysis showed that recurrent liver metastasis was associated with the macroscopic features (p < 0.001), venous invasion (p < 0.05), focal dedifferentiation (p < 0.05) and lymph node metastasis (p < 0.001) (Table 3). The sensitivity, specificity and predictive value of macroscopic features as an indicator of liver metastasis were 60, 85 and 79%, respectively, and the false-positive and false-negative rates were 12 and 10%, respectively. Multivariate analysisrevealed that the macroscopic features and lymph node metastasis were independent predictors of liver metastasis (Table 4). The predictive value of liver metastasis based on the macroscopic features and lymph node metastasis is shown in Table 5. The frequency of recurrent liver metastasis of patients having no lymph node metastasis with streak type tumor was 18.2% compared with 7.3% for those with non-streak type tumor. The frequency of recurrent liver metastasis of patients having lymph node metastasis with streak type tumor was 66.7% compared with 18.3% for those with non-streak type tumor.
Table 3
| Variable | Code | Liver metastases | ||
| Positive ( |
Negative ( |
|||
| Macroscopic features | ||||
| Streak type | 1 | 27 | 34 | <0.001 |
| Non‐streak type | 2 | 38 | 244 | |
| Histological type | ||||
| Well | 1 | 28 | 146 | |
| Moderate | 2 | 32 | 119 | N.S. |
| Poor | 3 | 3 | 10 | |
| Others | 4 | 2 | 3 | |
| Depth of tumor penetration | ||||
| Within muscularis propria | 1 | 4 | 48 | |
| Beyond muscularis propria | ||||
| but not involving serosa | 2 | 25 | 123 | N.S. |
| Beyond muscularis propria | ||||
| involving serosa | 3 | 36 | 107 | |
| Lymphatic invasion | ||||
| Absent | 0 | 26 | 191 | |
| Intramural | 1 | 9 | 43 | <0.05 |
| Extramural | 2 | 20 | 43 | |
| Focal dedifferentiation | ||||
| None or mild degree | 0 | 32 | 183 | <0.05 |
| Moderate or severe degree | 1 | 33 | 95 | |
| Inflammatory cell infiltration | ||||
| Mild degree | 0 | 55 | 213 | N.S. |
| Moderate or severe degree | 1 | 10 | 65 | |
| Lymph‐node metastasis | ||||
| None | 0 | 15 | 158 | |
| 1–4 | 1 | 37 | 99 | <0.001 |
| >4 | 2 | 13 | 21 | |
Figure 3. Histopathological findings for the white streak. Cancer cells invaded extramural veins, almost filling the veins and adhering to the venous wall, exhibiting desmoplastic stromal reaction with marked perivascular fibrosis. (a and b, HE stain, ×40).
Figure 4. Focal dedifferentiation. This characteristic refers to cancer cells which focally lost glandular differentiation and invaded showing solitary and trabecular nests with desmoplastic stromal reaction at the invasive front of colorectal adenocarcinoma (HE stain, ×200).
Table 4 Table 5
Variable
Coefficient
Standard error
p value
Exp (coefficient)
Macroscopic feature
1.6251
0.3249
<0.001
5.0791
Lymph node metastasis
1.1000
0.2266
<0.001
3.0041
Constant
-4.0871
0.4859
Lymph node metastasis
Macroscopic feature
Number of patients
Number of patients with liver metastasis
Absent (Dukes' A/B)
Non-streak type
Streak type151
2211 (7.3%)
4 (18.2%)
Present (Dukes' C)
Non-streak type
Streak type131
3924 (18.3%)
26 (66.7%)
Cumulative Survival Rate
In 343 patients undergoing curative surgery, the 5-year survival rate in patients with streak type tumor (70.3%) was significantly lower than that in patients with non-streak type tumor (89.2%, p < 0.05) (Fig. 5).
Figure 5. Cumulative survival curves in 343 patients undergoing curative surgery for colorectal cancer. *The 5-year cumulative survival rate in patients with streak type tumor was significantly lower than that in patients with non-streak type tumor (log-rank test, p < 0.05).
Discussion
Macroscopic investigation of the cut section of colorectal cancer provided characteristic features at the tumor invasive region and, according to the presence or absence of these macroscopic features, all cases examined were divided into two types: streak type and non-streak type. In our series, streak type tumor was closely related to the liver metastasis and characterized by deep wall invasion, frequent extramural venous invasion and a high degree of focal dedifferentiation. In addition to the lymph node metastasis, these histological parameters were predictors of liver metastasis. Hence it is reasonable that the streak type tumor featured by these histopathological findings was a significant indicator of liver metastasis. Since neither venous invasion nor focal dedifferentiation was an independent predictor of liver metastasis in multivariate analysis, macroscopic features and lymph node metastasis were most important in assessing liver metastases. The patients with lymph node metastasis, categorized as Dukes' C group (24), constituted approximately half of those undergoing curative surgery. Therefore, the macroscopic features combined with Dukes' staging system may help in the identification of patients at high risk of liver metastasis and these patients could then benefit from intensive chemotherapy (3-5). Furthermore, the survival rate of patients with streak type tumor was significantly lower than that of patients with non-streak type tumor. These macroscopic features are also useful as a prognostic indicator of colorectal cancer.
As a histopathological parameter for estimating the nature of the tumor margin, we used focal dedifferentiation (20). This is similar to `tumor budding', which refers to microscopic clusters of undifferentiated cancer cells just ahead of the invasive front (11) or the `infiltrating' type of Jass classification, which is characterized by deep, wide infiltration by isolated, individual tumor cells (13,14). Focal dedifferentiation exhibits a desmoplastic stromal reaction with marked fibrosis and neovascularization in addition to morphological changes of cancer cells at the tumor invasive front.
The white streak, which was macroscopically observed at the invasive region of streak type tumor, corresponded histopathologically to invading cancer cells often involving the extramural vein with marked perivascular fibrosis, where a high degree of focal dedifferentiation was frequently found. Hence the features of streak type tumor were formed by a marked stromal reaction against invading cancer cells, suggesting that a cancer-stromal interaction was important in regulating cancer metastasis. In estimating the morphological features of both cancer cells and stroma, our data suggested that it was more effective to examine the macroscopic appearance than individual histopathological findings, such as venous invasion and focal dedifferentiation.
In conclusion, this study clarified that the macroscopic features, which corresponded to the behavior of stroma against invading cancer cells, correlated well with liver metastasis and survival of patients with colorectal cancer. We hope that these simple and useful macroscopic features will be clinically useful and further contribute to pathological and biological assessment.
Acknowledgment
This study was supported in part by a Grant-in-Aid for the Second Term Comprehensive 10-year Strategy for Cancer Control from the Ministry of Health and Welfare of Japan.
References
This article has been cited by other articles:
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: www-admin{at}oup.co.uk
Last modification: 19 May 1998
Copyright© Japanese Journal of Clinical Oncology, 1998.
CiteULike 
Connotea 
Del.icio.us What's this?
G Brown
Thin section MRI in multidisciplinary pre-operative decision making for patients with rectal cancer
Br. J. Radiol.,
October 1, 2005;
78(Special_Issue_2):
S117 - S127.
[Abstract]
[Full Text]
[PDF]
S. Hirohashi
Inactivation of the E-Cadherin-Mediated Cell Adhesion System in Human Cancers
Am. J. Pathol.,
August 1, 1998;
153(2):
333 - 339.
[Abstract]
[Full Text]
[PDF]
This Article 
Abstract
Alert me when this article is cited
Alert me if a correction is posted
Services
Email this article to a friend
Similar articles in this journal
Similar articles in ISI Web of Science
Similar articles in PubMed
Alert me to new issues of the journal
Add to My Personal Archive
Download to citation manager
Search for citing articles in:
ISI Web of Science (10)
Request Permissions
Google Scholar
Articles by Inomata, M
Articles by Hirohashi, S
Search for Related Content
PubMed
PubMed Citation
Articles by Inomata, M
Articles by Hirohashi, S
Social Bookmarking
What's this?