Fax: (514) 398-7446
VEGF as a predictive marker of rectal tumor response to preoperative radiotherapy
Version of Record online: 12 OCT 2005
Copyright © 2005 American Cancer Society
Volume 104, Issue 11, pages 2517–2521, 1 December 2005
How to Cite
Zlobec, I., Steele, R. and Compton, C. C. (2005), VEGF as a predictive marker of rectal tumor response to preoperative radiotherapy. Cancer, 104: 2517–2521. doi: 10.1002/cncr.21484
- Issue online: 18 NOV 2005
- Version of Record online: 12 OCT 2005
- Manuscript Accepted: 5 JUL 2005
- Manuscript Revised: 9 MAY 2005
- Manuscript Received: 1 FEB 2005
- rectal cancer;
- predictive marker;
- preoperative radiotherapy
Neoadjuvant radiotherapy for rectal cancer may result in tumor downstaging or complete tumor regression leading to greater sphincter preservation. The identification of molecular predictive markers of tumor response to preoperative radiotherapy would provide an additional tool for selecting patients most likely to benefit from treatment. The aim of this study was to determine whether VEGF expression in preirradiation tumor biopsies is a useful predictive marker of tumor response in patients with rectal cancer undergoing preoperative radiotherapy.
Immunohistochemistry for VEGF was performed on 59 preirradiation biopsies from patients with completely responsive (ypT0) or nonresponsive tumors after preoperative radiotherapy. VEGF positivity was evaluated using several scoring methods and the association between VEGF and tumor response was compared. The distribution of VEGF scores was obtained as well as the mean VEGF expression in the two response groups.
The mean VEGF expression in nonresponsive tumors (NR) was significantly greater than in completely responsive tumors (CR) (P = 0.0035). Nearly half (47%) of all CR tumors had a VEGF expression of 10% or less. Eleven tumors were negative (0% immunoreactivity) for the protein and all of these (100%) were complete responders. Fifty-two percent of the NR tumors had VEGF scores of 80% or greater. The four scoring methods used to determine the association between VEGF and tumor response each produced significant results (P < 0.05).
The results of this study indicate that VEGF assessed immunohistochemically from preirradiation tumor biopsies may be a useful marker of rectal tumor response to preoperative radiotherapy. Cancer 2005. © 2005 American Cancer Society.
Neoadjuvant radiotherapy is part of standard care for patients with advanced rectal cancer.1 This treatment has been shown to improve survival and may reduce local recurrence rates versus surgery with or without postoperative radiotherapy.2 In addition, tumor downstaging and complete tumor regression may be achieved with preoperative radiotherapy leading to greater sphincter preservation.3, 4 The ability to predict tumor response from preirradiation biopsies may significantly improve the selection of patients for preoperative radiotherapy.
Vascular endothelial growth factor (VEGF) is a potent mediator of tumor angiogenesis.5 VEGF can be activated in tumor cells by several factors, including oncogenes, tumor suppressor genes, cytokines (IL-1, IL-6), and hypoxia resulting in secretion of proteolytic enzymes and matrix metalloproteases that degrade the basement membrane and extracellular matrix surrounding the tumor.6, 7 These events ultimately lead to endothelial cell migration and the formation of a new vasculature that supports the growth of the tumor and its nutrient requirement.8 In situ hybridization studies show that VEGF mRNA is significantly elevated in many human cancers and is associated with poor clinical outcome and greater aggressiveness of the tumor.5, 9 VEGF has been shown to up-regulate the antiapoptotic gene BCL-2, thereby acting as a survival factor for both endothelial and tumor cells.10, 11 Activation of VEGF also leads to increased vascular permeability of tumor vessels, causing them to be ‘leaky’ and less efficient in their ability to diffuse oxygen.7, 12 This leaky vasculature appears to contribute to less efficient delivery of chemotherapeutic agents to the tumor.10, 13
The aim of this study was to determine, from preirradiation tumor biopsies, the value of VEGF as a predictive marker of rectal tumor response to preoperative radiotherapy.
MATERIALS AND METHODS
Fifty-nine patients with rectal adenocarcinoma were entered into the study and informed written consent was obtained from each. Clinical staging was performed via magnetic resonance imaging (MRI) and endoscopic ultrasonography (EUS). Patients were treated on a preoperative conformal high-dose rate endorectal brachytherapy protocol.14 Patients with abdominal nodal disease or metastases were excluded from the study. Radiation was delivered preoperatively with a flexible, 8-channel endorectal catheter using a high-dose-rate remote after-loading system. A daily fraction of 6.5 Gy was administered over 4 consecutive days to a total of 26 Gy. Each patient was planned by computed tomography (CT) simulation before treatment with the endorectal catheter in place. To obtain optimal conformal dosimetry for each individual tumor, differential loading of the eight channels was performed. Patients underwent surgery 4–8 weeks later, regardless of tumor response.
Tumor response was evaluated pathologically from the postoperative specimens. Complete tumor response was defined as no evidence of residual carcinoma or ypT0.15 Partial response was characterized by the presence of microfoci of residual carcinoma typically measuring from 0.3–0.9 cm in diameter. Nonresponsive tumors have large residual carcinoma with absence of microfoci. Residual tumors ranged from 2–6 cm in diameter. For the purposes of this study, only completely responsive and nonresponsive tumors were evaluated.
Preirradiation, formalin-fixed paraffin-embedded tumor biopsies from all 59 patients were collected. Immunohistochemistry for VEGF was performed using the avidin-biotin complex (ABC; Vector Laboratories, Burlingame, CA) procedure, including heat-induced antigen retrieval procedures. Incubation with polyclonal anti-VEGF antibody (Santa Cruz Biotechnology, Santa Cruz, CA; VEGF-A20, 1:100) was carried out at 37 °C for 1 hour. Negative controls were treated identically with primary antibody omitted. Tissue from glioblastoma was used as the positive control.
Scoring of VEGF Immunohistochemistry
Evaluation of VEGF immunoreactivity was made by two independent observers. The percentage of positive tumor cells was determined by each observer and the average of the two scores was obtained.
Several scoring systems have previously been used to evaluate VEGF positivity.16–18 In this study, the average scores obtained by the observers were used to compare the following scoring methods: 1) negative/positive: negative tumor with 0% VEGF staining versus positive tumor with any degree of staining; 2) 10% cutoff: positive tumor with more than 10% immunoreactive tumor cells; 3) 0, 1+, 2+, 3+: tumor is negative for VEGF (0), has fewer than 20% positive cells (1+), has 20–50% positive staining (2+), or has greater than 50% staining (3+); 4) percentages: the actual percentage of positive tumor cell staining obtained by the observers.
Assessment of VEGF immunoreactivity from preirradiation tumor biopsies was performed blinded to postoperative tumor response.
Patient and tumor characteristics were assessed by the chi-square test. The Wilcoxon Rank Sum test was used to evaluate differences in mean VEGF expression between response groups. P < 0.05 was considered statistically significant. Analysis of VEGF immunoreactivity and response was carried out by the Fisher exact and chi-square tests for scoring methods 1–3. Logistic regression was used to test for differences in VEGF and tumor response in scoring method 4. All analyses were carried out using SAS, 8th ed. (SAS Institute, Cary, NC).
Pathologic evaluation of the irradiated tumor bed postoperatively identified 30 tumors with complete response and 29 with no response to radiotherapy. Patient and tumor characteristics are summarized in Table 1.
|Age in yrs|
|Tumor stage %|
|Nodal status %|
|Tumor response %|
Cytoplasmic immunoreactivity ranged from 0–100%. The mean VEGF expression in nonresponsive (NR) tumors was 63% and was significantly greater than completely responsive (CR) tumors (37.31%) (P = 0.0035). No significant association between age, gender, stage, or nodal status and tumor response was found.
The distribution of VEGF scores for each response group is shown in Figure 1. Nearly half (47%) of all CR tumors were found to have a VEGF expression of 10% or less. Of those, 11 tumors (79%) were negative for the protein (no VEGF expression). All NR tumors showed some degree of VEGF positivity. Fifteen of these 29 tumors (52%) had at least 80% immunoreactivity. Ten NR tumors had more than 90% VEGF expression, whereas only 2 CR tumors (6%) were found in this region. The association between VEGF expression and tumor response produced by each of the four scoring systems is listed in Table 2. All methods yielded a statistically significant association between VEGF immunoreactivity and tumor response (P < 0.05).
|2) 10% cutoff||0.0153a|
|3) 0, 1+, 2+, 3+||0.0026b|
These results appear to indicate that tumors completely responsive to preoperative brachytherapy most often express no or low levels of VEGF in their pretreatment biopsies, whereas nonresponsive tumors are generally highly immunoreactive.
The identification of molecular predictive markers of tumor response to preoperative radiotherapy would provide an additional tool for selecting patients most likely to benefit from treatment. Recently, the role of VEGF in angiogenesis and, particularly, in colorectal cancer has been investigated. Immunohistochemistry studies have shown VEGF to be absent in normal colorectal mucosa, while carcinomas are highly immunoreactive.19, 20 Wong et al.20 investigated the temporal relationship between VEGF expression and tumor progression from adenoma to carcinoma. They found that activation of VEGF was an early event in the adenoma-carcinoma sequence, suggesting that VEGF may be an angiogenesis-initiating factor in the early phase of tumor development. In colorectal carcinoma, no difference in stage-specific VEGF expression has yet been reported. Nozue et al.18 described VEGF status before and after preoperative radiotherapy in locally advanced rectal cancers. They found a greater number of VEGF-positive tumors and more intense VEGF immunoreactivity after treatment. Up-regulation of VEGF has been associated with poor prognosis in patients with colorectal cancer and linked to liver metastasis.21, 22
Hypoxia is a major inducer of VEGF activation, which occurs primarily through the transcription of hypoxia-inducible factor 1-alpha (HIF-1α).7 Tumor growth leads to limitations in oxygen diffusion provided by the host vasculature, creating areas of hypoxia.23 In response to this low oxygen tension, tumor cells either undergo apoptosis or begin to produce VEGF to induce vasculature that will in turn increase oxygen delivery to sustain their survival.24 In addition, VEGF may activate Bcl-2, an antiapoptotic protein.10, 11 This may further contribute to the survival advantage of tumor cells expressing VEGF.
Our results show that low or absent VEGF in preirradiation rectal tumor biopsies is strongly associated with complete tumor response. A comparison of mean VEGF expression shows that nonresponsive tumors are more highly immunoreactive and have a significantly greater overall VEGF expression than completely responsive tumors. Of those tumors negative for VEGF, all (100%) were completely responsive to therapy.
In this study, we further investigated whether a variety of frequently used VEGF scoring methods affect the predictive value of the protein. The overwhelming majority of studies use a scoring method based on the 10% cutoff point.19, 20, 24, 25 Our results demonstrate that VEGF may be predictive of tumor response to preoperative brachytherapy regardless of the scoring system used. However, the selection of the scoring method may have a nonnegligible affect on the final interpretation of the results. More research must be done in the area of scoring methods and how their interpretation may affect the predictive value of the protein.
Although most complete responders are found in the lower end of the distribution of VEGF scores including nearly half with 10% immunoreactivity or less, approximately 26% have more than 80% positive tumor cell staining for VEGF. One explanation for this may lie in the fact that the expression of VEGF is not sufficient for angiogenesis to occur.7 Numerous antiangiogenic proteins are secreted by tumor cells including endostatin, angiostatin, and thrombospondins whose apoptotic action on endothelial cells counterbalances the effects of proangiogenic agents.7, 26 The ‘switch’ or imbalance of pro- and antiangiogenic factors leading to tumor angiogenesis may not have yet occurred in these completely responsive yet highly immunoreactive tumors.6 Similarly, nonresponsive tumors with low VEGF levels may be more antiangiogenic. If so, other mechanisms of radioresistance may be in place in these tumors, such as an imbalance of proliferation versus apoptosis, or deregulated cell-cycle arrest. It may therefore be important to study VEGF in combination with proteins that may have predictive potential such as p53, p27, Bcl-2, or cyclin D and E.27–31
In summary, the results of this study indicate that VEGF assessed immunohistochemically from preirradiation tumor biopsies may be a useful marker in the prediction of tumor response to preoperative radiotherapy.
The authors thank Dr. Nilima Nigam for input and contribution to the editing of this article, Julie Hanck for technical assistance, and Dr. Té Vuong for continued support.
- 15American Joint Committee on Cancer. AJCC cancer staging manual. 6th ed. New York: Springer, 2002.