Fax: (585) 273-1042
Vascular endothelial growth factor, CD68, and epidermal growth factor receptor expression and survival in patients with Stage II and Stage III colon carcinoma
A role for the host response in prognosis
Article first published online: 3 FEB 2003
Copyright © 2003 American Cancer Society
Volume 97, Issue 4, pages 960–968, 15 February 2003
How to Cite
Khorana, A. A., Ryan, C. K., Cox, C., Eberly, S. and Sahasrabudhe, D. M. (2003), Vascular endothelial growth factor, CD68, and epidermal growth factor receptor expression and survival in patients with Stage II and Stage III colon carcinoma. Cancer, 97: 960–968. doi: 10.1002/cncr.11152
- Issue published online: 3 FEB 2003
- Article first published online: 3 FEB 2003
- Manuscript Accepted: 24 SEP 2002
- Manuscript Revised: 23 SEP 2002
- Manuscript Received: 12 AUG 2002
- Cancer Action, Inc., of Rochester/ Gilda's Club of Rochester
- Dr. Robert A. Cooper Jr. Endowed Fund
- James P. Wilmot Cancer Research Fellowship
- vascular endothelial growth factor (VEGF);
- epidermal growth factor receptor (EGFR);
- tumor-associated macrophages (TAM);
- colon carcinoma;
The elucidation of new therapeutic targets of prognostic significance in colon carcinoma is necessary to improve outcomes. In the current study, the authors examined the expression of epidermal growth factor receptor (EGFR) and vascular endothelial growth factor (VEGF) in primary colon carcinoma cases and VEGF in tumor-associated macrophages (TAM)/stroma, and their correlation with survival.
The authors identified 131 consecutive American Joint Committee on Cancer Stage II and Stage III colon carcinoma patients seen at the University of Rochester between 1990–1995. Expression of VEGF, EGFR, and CD68 were examined by immunohistochemistry in paraffin-embedded primary colon tumors and graded as the percentage of cells stained. Data were analyzed using a multivariate Cox proportional hazards model.
VEGF expression in tumor was not found to be significantly associated with survival. However, 42% of the patients expressed VEGF in TAM/stroma. The median survival in this group was 9.7 years versus 4.3 years in the VEGF-negative (TAM/stroma) group (hazards ratio of 0.57, 95% confidence interval [95% CI], 0.34–0.95; P = 0.03). Although TAM infiltration alone was not found to be significant in multivariate analysis, the presence of both CD68 and VEGF (TAM/stroma) was predictive of improved survival (hazards ratio of 0.48, 95% CI, 0.28–0.83; P = 0.006). High grades of EGFR expression (≥ Grade 2) were found to be associated with a trend toward worsened survival.
The greater than twofold increase in median survival associated with VEGF-expressing TAM suggests a hitherto unknown role for this subset of cells in the host response to colon carcinoma and requires further investigation. Overexpression of EGFR may be associated with worsened survival, providing a rationale for trials of anti-EGFR agents as adjuvant therapy. Cancer 2003;97:960–8. © 2003 American Cancer Society.
Colorectal carcinoma is one of the leading causes of cancer death in both developed and developing nations. Worldwide, > 900,000 new cases of colorectal carcinoma are diagnosed each year, and colorectal carcinoma accounts for nearly 500,000 cancer deaths annually.1 Currently, the most important factor predictive for survival is regional lymph node status at the time of initial surgery. However, this is not sufficient to predict outcomes accurately. Approximately 20% of patients with Stage II disease (i.e., without regional lymph node involvement) and approximately 50% of patients with Stage III disease will not be alive 5 years after curative resection.2 Although much has been learned regarding the molecular pathogenesis of colon carcinoma in the past 20 years, the elucidation of markers of prognosis that also could serve as therapeutic targets is necessary to better understand and improve outcomes.
Angiogenesis is essential for tumor growth and metastasis.3 Vascular endothelial growth factor (VEGF) is an important proangiogenic cytokine and is required to initiate the formation of immature vessels by vasculogenesis or angiogenic sprouting.4 Tumor cells are the predominant source of VEGF; however, tumor-associated stroma also has been shown to produce VEGF.5 VEGF expression has been associated with colonic neoplastic progression.6, 7 VEGF also stimulates macrophage migration8 and has been associated with macrophage infiltration in breast carcinoma9 and melanoma.10 To our knowledge the clinical significance of VEGF-expressing macrophages is unknown.
Epidermal growth factor receptor (EGFR) is a transmembrane growth factor receptor tyrosine kinase that transduces growth-stimulatory signals when bound by its ligands, epidermal growth factor and transforming growth factor-α. The EGFR signaling pathway is important in tumorigenesis because it regulates cellular differentiation, proliferation, and angiogenesis.11 EGFR overexpression has been associated with poor prognosis in several solid tumors.12, 13 However, to our knowledge, data regarding EGFR expression in colorectal carcinoma are limited, and study populations have included both early-stage patients and those with metastatic disease.14, 15 In a study of 45 liver metastases and 33 corresponding primary colorectal carcinomas, the absence of EGFR expression was associated with a better survival after partial liver resection.16
Therefore, we investigated the correlation between survival and immunohistochemical expression of VEGF by tumor cells as well as TAM/stroma, and of EGFR by tumor cells in primary colon carcinoma specimens.
MATERIALS AND METHODS
We studied tumor specimens from 131 consecutive patients with American Joint Committee on Cancer (AJCC) Stage II and Stage III colon carcinoma (excluding rectal carcinoma)17 who underwent curative resection at the University of Rochester Medical Center between 1990–1995. One hundred fifty-seven consecutive colon carcinoma patients initially were identified through the tumor registry. Patients were excluded for the following reasons: lack of availability of tumor specimens (n = 13), noncurative resections or metastatic disease (n = 12), and carcinoid tumor (n = 1). Clinicopathologic information on these patients was obtained through the tumor registry database and chart review. The clinical database was prepared by one of the authors (A.A.K.) with no knowledge of the results of immunohistochemical analysis.
An institutional review board approval was obtained to conduct this retrospective study. Patient confidentiality was maintained with the patient database kept in a data file with access limited to study authors.
Analysis of Tumor Specimens
Immunohistochemical stains were performed on freshly cut, formalin-fixed, paraffin-embedded tissue sections. Paraffin-embedded tissues were cut at 4 μm, floated on distilled water at a temperature of 54 °C, and mounted on chemically charged slides. Slides were dried at room temperature for 4 hours and then placed in an oven overnight at 57 °C. Sections were deparaffinized and rehydrated in a series of alcohols and xylene according to established procedures.
Rabbit polyclonal antibody to VEGF (1:50) (Zymed Laboratories, San Francisco, CA), which recognizes the 121-, 165-, and 189-amino acid splice variants of VEGF, and CD68 (1:200) (Dako Corporation, Carpinteria, CA) were employed for immunostaining. Sections for immunostaining were treated with 3% hydrogen peroxide for 10 minutes to quench myeloperoxidase and then cleared in running water followed by a 5-minute rinse in Tris-buffered saline (TBS) at a pH of 7.6. Antigen unmasking with heat retrieval solution (Dako Corporation) at a pH of 6.1 was accomplished by placing slides in preheated Dako TBS in a steamer (Black & Decker, Shelton, CT) for 30 minutes. Slides were mounted on the Dako Autostainer and incubated with antibody to VEGF (1 hour) or CD68 (45 minutes), and staining was completed using Dako Rabbit Envision Plus Kit (Dako Corporation). The sections were removed from the stainer and counterstained with a modified Mayer hematoxylin followed by 10 dips in 0.3% ammonia water. Sections were rinsed in running tap water and mounted with Immunmount (Harvey Instruments, Buffalo, NY). Negative controls were comprised of nonimmune serum matched in parallel to primary conditions. Positive controls for VEGF immunostaining included invasive ductal carcinoma of human breast and glomerular podocytes of human kidney. Positive controls for CD68 immunostaining were a routine sausage block including spleen, colon, and tonsil.
Immunostaining for EGFR was accomplished using the Epidermal Growth Factor Receptor Detection Kit (Dako Corporation) on a Dako Autostainer. The antibody to EGFR is a monoclonal mouse antihuman antibody clone 2-18C9, which recognizes a 170-kilodalton band encoded by the human HER-1 gene. As per the manufacturers' instructions, sections were incubated with peroxide block for 5 minutes followed by a TBS rinse and incubation for 5 minutes with proteinase K, which was freshly diluted to 1:100 before use. After a rinse in TBS, the sections were incubated in the monoclonal EGFR primary antibody for 30 minutes and staining was completed by incubation with monoclonal-labeled polymer for 30 minutes. The detection kit employed a visualization reagent based on dextran technology followed by a chromogen, diaminobenzidine. Slides were counterstained in Mayer hematoxylin blue in 0.3% ammonia water followed by a tap water rinse and mounted using an aqueous media. The positive and negative controls were supplied by the manufacturer. These were comprised of pelleted, formalin-fixed, paraffin-embedded cell lines expressing Grade 2 staining of EGFR (line HT-29) and a negative cell line (CAMA-1).
VEGF expression was examined using light microscopy and both malignant cells and stroma were graded separately as the percentage of cells stained with moderate or greater intensity as follows: negative, < 2%; Grade 1, 2–20%; Grade 2, 20–50%; and Grade 3, > 50%. EGFR and CD68 were graded in a similar fashion with EGFR as the percentage of malignant cells only and CD68 as the percentage of stromal/inflammatory cells only. Cell membrane staining was interpreted with EGFR while predominantly cytoplasmic staining was interpreted with VEGF and CD68. The immunostains were graded by a single pathologist (C.K.R.) who was blinded to clinical outcomes.
The primary endpoint in this study was overall survival, as measured from the time of diagnosis (defined as the date of initial biopsy). The survival function (proportion surviving) and median survival time (with 95% confidence intervals [95% CI]) were estimated using the Kaplan–Meier curve. The Cox proportional hazards model was used to obtain hazards ratios (95% CI) for single variable and multivariate models. The hazards ratio describes the relative risk of reaching the endpoint per unit of time.
One hundred thirty-one colonic adenocarcinoma patients who underwent resection with intent to cure comprised the study population. Patient characteristics are described in Table 1. The mean age of this population was 71 years (± 13.4 years). Sixty-nine patients (53%) were male and 62 patients (47%) were female. A large majority, 108 patients (84%), had T3 tumors. Fifty-one patients (40%) had Stage II disease and 77 patients (60%) had Stage III disease.17 At the time of data analysis, these patients had been followed for a median period of 5.0 years (range, 18 days–10.8 years). The overall survival for the entire study population at 5 years was 54%.
|Patient characteristics||All patients (n = 131)||Patients with VEGF (tumor) negative (n = 86)||Patients with VEGF (tumor) ≥ Grade 1 (n = 45)||Patients with VEGF (TAM/stroma) negative (n = 76)||Patients with VEGF (TAM/stroma) Grade 1 (n = 55)||Patients with EGFR ≤ Grade 1 (n = 104)||Patients with EGFR ≥ Grade 2 (n = 27)|
|Male||69 (53)||46 (53)||23 (51)||35 (46)||34 (62)||57 (55)||12 (44)|
|Female||62 (47)||40 (47)||22 (49)||41 (54)||21 (38)||47 (45)||15 (56)|
|Mean (± SD) age at diagnosis (yrs)||71 (± 13.4)||71 (± 13.1)||69 (± 14.1)||71 (± 14.3)||70 (± 12.2)||71 (± 13.8)||69 (± 12.0)|
|Tumor invasion (no.) (%)|
|T1 or T2||11 (9)||6 (7)||5 (11)||5 (7)||6 (11)||7 (7)||4 (15)|
|T3 or T4||118 (91)||79 (93)||39 (89)||69 (93)||49 (89)||96 (93)||22 (85)|
|AJCC pathologic stage (no.) (%)|
|II||51 (40)||32 (38)||19 (44)||24 (33)||27 (49)||41 (40)||10 (38)|
|III||77 (60)||53 (62)||24 (56)||49 (67)||28 (51)||61 (60)||16 (62)|
|Adjuvant chemotherapy (no.) (%)|
|Yes||38 (29)||24 (28)||14 (31)||21 (28)||17 (31)||27 (26)||11 (41)|
|No||93 (71)||62 (72)||31 (69)||55 (72)||38 (69)||77 (74)||16 (59)|
|Radiotherapy (no.) (%)|
|5-year survival (%)||54||53||55||46||64||55||48|
|Median survival (yrs)||5.5||5.8||5.5||4.3||9.7||5.5||4.5|
VEGF expression was observed in primary tumor cells (Fig. 1) as well as adjacent stroma. No VEGF expression was observed in normal colonic mucosa. VEGF expression in tumor cells was graded separately from VEGF expression in TAM and stromal cells. Forty-five of the 131 patients (34%) expressed VEGF (≥ Grade 1) in tumor cells. Of these, 35 patients (27%) had Grade 1 expression, 7 patients (5%) had Grade 2 expression, and 3 patients (2%) had Grade 3 VEGF expression. In 86 patients (66%), the tumors did not express VEGF (Table 1). The median survival was 5.5 years (95% CI, 3.0–8.9 years) in the patients whose tumors expressed VEGF, which was not significantly different from the median survival of 5.8 years (95% CI, 3.7–10.3 years) observed in patients with an absence of VEGF expression in tumors (P = 0.62).
Fifty-five of the 131 patients (42%) expressed VEGF in stroma, specifically in TAM, with occasional fibroblast staining (Fig. 2A,C). Of these, 49 patients (37%) had Grade 1 expression, 5 patients (4%) had Grade 2 expression, and 1 patient (1%) had Grade 3 VEGF expression. Patients whose tumors expressed VEGF (TAM/stroma) were more likely to be male and to have Stage II disease (Table 1), although these differences were not found to be statistically significant (P > 0.05). The median survival was 9.7 years (95% CI, 5.0 [insufficient data to estimate] years) in the patients whose tumors expressed VEGF (TAM/stroma), which represented a > 2-fold increase over the median survival of 4.3 years (95% CI, 3.0–7.1 years) observed in patients whose tumors did not express VEGF (TAM/stroma) (P = 0.0065). The 5-year survival rate for patients whose tumors expressed VEGF (TAM/stroma) was 64% compared with a survival rate of 46% for those patients whose tumors demonstrated an absence of VEGF expression (TAM/stroma). The proportionality of the survival advantage associated with VEGF expression (TAM/stroma) can be observed in a Kaplan–Meier survival curve (Fig. 3).
Using routine hematoxylin and eosin staining, the great majority of stromal cells expressing VEGF appeared to be macrophages. This finding was confirmed by staining consecutive levels of 130 tumors using CD68 (a macrophage-specific immunostain) and VEGF. Paired levels of both stains were identical with regard to location and similar in staining intensity; however, a greater number of stromal cells were observed to stain for CD68 compared with VEGF (Fig. 2B). In addition, a greater percentage of patients expressed stromal CD68 (73%) compared with stromal VEGF (42%), implying that a subset of macrophages was expressing VEGF. Of these, 49 patients (38%) had Grade 1 expression, 37 patients (28%) had Grade 2 expression, and 9 patients (7%) had Grade 3 CD68 expression in stroma. Thirty-five patients (27%) did not demonstrate CD68 expression in stroma. The median survival was 7.9 years (95% CI, 5.0–9.9 years) in the patients whose tumors expressed CD68 and again, this finding represented a > 2-fold increase over the median survival of 3.1 years (95% CI, 1.4–5.3 years) observed in patients whose tumors were negative for CD68 (P = 0.0075). Similarly, the median survival of patients whose tumors expressed both VEGF and CD68 in stroma (Grades 1–3) was 9.7 years (95% CI, 7.2 [insufficient data to estimate] years), and this finding compared favorably with the median survival of 3.7 years (95% CI, 2.6–5.8 years) reported in patients whose tumors did not express VEGF and CD68 in stroma (P = 0.0011) (Fig. 4).
Sixty of the 131 patients (46%) demonstrated some degree of EGFR expression in tumor cells. Of these, 33 patients (25%) had Grade 1 expression, 20 patients (15%) had Grade 2 expression, and 7 patients (5%) had Grade 3 EGFR expression. Seventy-one patients (54%) had tumors that did not express EGFR. EGFR expression was not observed in adjacent normal colonic mucosa or stromal cells (Fig. 5). Patient characteristics of those whose tumors expressed EGFR were similar to those of the larger study population. The presence or absence of EGFR expression did not appear to be correlated with survival (P = 0.24). However, the median survival of patients with higher grades of EGFR expression (≥ Grade 2) was 4.5 years (95% CI, 1.9 [insufficient data to estimate] years) compared with 5.5 years (95% CI, 4.1–8.9 years) in patients with tumors that either were negative for EGFR expression or demonstrated Grade 1 expression, a finding that was significant in a multivariate analysis (P = 0.05).
Multivariate Survival Analysis
In multivariate analyses, using a Cox proportional hazards model, the following prognostic factors were found to be significantly and independently associated with improved survival: younger age at diagnosis (measured as a continuous variable), Stage II disease, use of adjuvant chemotherapy, a VEGF (TAM/stroma) expression grade ≥ 1, and the presence of both a VEGF (TAM/stroma) and CD68 expression grade ≥ 1 (Table 2). Negative or Grade 1 EGFR expression was reported to be marginally significant as a predictor of improved survival. CD68 expression alone was not found to be significant on a multivariate analysis (P = 0.20). The interaction between VEGF (TAM/stroma) expression and stage of disease also was not found to be significant (P = 0.46).
|Prognostic Factor||Model 1a||Model 2a|
|HRb (95% CI)||P valuec||HRb (95% CI)||P valuec|
|Age||1.03 (1.00–1.05)||0.03||1.02 (1.00–1.05)||0.04|
|Stage III diseased||2.97 (1.71–5.15)||0.0001||2.97 (1.70–5.21)||0.0001|
|Adjuvant chemotherapy||0.45 (0.24–0.84)||0.01||0.41 (0.22–0.79)||0.01|
|VEGF (TAM/stroma) ≥ Grade 1||0.57 (0.34–0.95)||0.03||—||—|
|VEGF (TAM/stroma) and CD68 ≥ Grade 1||—||—||0.48 (0.28–0.83)||0.01|
|EGFR ≥ Grade 2||1.74 (0.93–3.25)||0.08||1.88 (1.00–3.54)||0.05|
The results of the current study demonstrate, unexpectedly, that VEGF expression in TAM/stromal cells is associated with a greater than twofold increase in median survival, and this effect was reported to be significant on both univariate and multivariate analyses. The presence of TAM, as determined by CD68 staining, is associated with a similar improvement in median survival, although this effect was not found to be significant on multivariate analysis. VEGF expression in tumor cells does not appear to correlate with survival. Finally, EGFR overexpression in resected colon tumors appears to confer an adverse prognosis.
The role of TAM in the host response to neoplasia has long been known and debated.18, 19 Activated TAM are capable of tumoricidal activity both directly through the production of tumor necrosis factor-α,20 nitric oxide,21 and reactive nitrogen intermediates,22 and indirectly through the production of cytokines such as interleukin (IL)-12 and IL-1823 and the expression of costimulatory molecules such as CD80 and CD86.24 However, TAM also produce numerous angiogenic and growth factors that promote tumor growth and metastasis, including fibroblast growth factor-2, transforming growth factor-α, and platelet-derived growth factor.25 This dual potential of TAM has been described in the macrophage balance hypothesis.26 The available clinical data also are conflicting. TAM infiltration has been associated with an improved prognosis in patients with carcinoma of the prostate, stomach, and lung, but with a worsened prognosis in patients with breast, cervical, and transitional cell carcinomas.19 In the current study, the presence of VEGF-expressing TAM was found to be associated significantly with favorable outcomes on a Cox regression analysis, although merely the presence of TAM was not. Therefore, it is possible that VEGF-expressing TAM may tip the macrophage balance in favor of an antitumor effect. Although VEGF is well known to be a regulator of tumor angiogenesis, it also stimulates macrophage migration.8 Macrophage recruitment induced by VEGF is mediated via the receptor flt-127 and by induction of monocyte chemoattractant protein-1.28 These pathways differ from those required for its proangiogenic activity. In this context, it is interesting to note that in models of atherosclerosis, VEGF- expressing macrophages recently were demonstrated to further recruit and activate monocytes29 and to promote plaque progression via an inflammatory process.30, 31 These activities appear to be separate from the well documented proangiogenic role of VEGF in atherosclerosis. Immunohistochemical studies previously demonstrated that TAM are a source of VEGF in breast carcinoma.9, 32 However, to the best of our knowledge, no study published to date has correlated VEGF-expressing TAM with survival in patients with resected colon carcinoma. In our analysis, patients with VEGF-expressing TAM were more likely to have Stage II disease. However, this difference was not reported to be statistically significant, and the beneficial effect observed was independent of stage in the multivariate analysis. Therefore, these results clearly challenge prevailing concepts of VEGF function in tumor progression and suggest a possible role for the subset of VEGF-expressing macrophages in the host response to colon carcinoma. Further studies are needed to confirm these findings and to delineate mechanisms for the observed effects.
Conflicting results have been reported regarding the prognostic significance of angiogenesis and VEGF expression in tumor cells. Both low and high microvascular density have been associated with increased survival in colon carcinoma.33, 34 Similarly, in one study, VEGF was not found to demonstrate any correlation with disease recurrence in patients with Stages B2 or C disease,7 but was found to be associated with an increased risk for disease recurrence in patients with Stage II colon carcinoma in another study.35 It therefore is not surprising that our analysis failed to reveal an adverse prognostic effect for VEGF expression in tumor cells. VEGF has six alternatively spliced isoforms, some of which are secreted and some of which are cellularly associated.36 Distinctive roles may exist for various isoforms, but to our knowledge these remain largely unknown. The possibility of differential isoform expression in macrophages and tumor cells exists because the antibody used for immunohistochemical staining in the current study is directed against the carboxyl terminal of human VEGF, which is conserved across the splice variants of VEGF.36 VEGF isoforms have been shown to display distinct activities in promoting tumor angiogenesis at different anatomic sites in vivo,37 although whether such a difference would occur between tumor and stromal microenvironments to our knowledge is not known. It is interesting to note that overexpression of macrophage colony-stimulating factor has been shown to predict for poor prognosis when expressed by ovarian epithelial cells, but to have a protective effect when expressed by stromal cells.38
EGFR expression has been associated with poor prognosis, resistance to chemotherapy, and decreased survival in several tumor types.12 This has led to the development of various approaches to inhibit the EGFR signaling pathway, including tyrosine kinase inhibitors, monoclonal antibodies, and antisense oligonucleotides.39 In colorectal carcinoma, EGFR expression has been correlated with both stage of disease and tumor grade.14, 40 In our analysis, expression of EGFR (≥ Grade 1) was not, by itself, found to be a significant prognostic factor. However, higher grades of EGFR expression (Grades 2-3) were reported to be associated with a worsening of survival. In the current study, the overall survival for the entire population was 54%, which is lower than previously reported figures. It is interesting to note that only 19% of the patients in the current study were age < 60 years, compared with 44% in the National Surgical Adjuvant Breast and Bowel Project (NSABP) C01-04 and other similar studies.41–43 Age is a well known prognostic factor in colon carcinoma, and may have contributed to low overall survival rates, as well as to the decreased use of adjuvant chemotherapy. It is possible that our analysis actually underestimates the true prognostic significance of EGFR because immunohistochemistry assesses total cellular levels of EGFR and not the activated form. If confirmed by larger studies, the results of the current study will provide a rationale for the investigational use of anti-EGFR agents in an adjuvant therapy setting.
VEGF-expressing TAM were observed in 42% of patients with Stage II and Stage III colon carcinoma. This was found to be associated independently with a twofold increase in median survival after curative resection, suggesting that these cells may signify a host response to colon carcinoma. High grades of EGFR expression in tumor cells may be associated with worsened survival. Further studies are necessary to confirm and extend these findings.
The authors would like to thank Ms. Patricia Bourne for excellent technical assistance with immunostaining and Dr. John M. Bennett for reviewing the article.
- 17American Joint Committee on Cancer, American Cancer Society. American Joint Committee on Cancer cancer staging manual. New York: Springer-Verlag, 2002.,
- 42Clinical trial to assess the relative efficacy of fluorouracil and leucovorin, fluorouracil and levamisole, and fluorouracil, leucovorin, and levamisole in patients with Dukes' B and C carcinoma of the colon: results from National Surgical Adjuvant Breast and Bowel Project C-04. J Clin Oncol. 1999; 17: 3553–3559., , , et al.