Customize your JAMA Network experience by selecting one or more topics from the list below.
Boonkitticharoen V, Kulapaditharom B, Leopairut J, et al. Vascular Endothelial Growth Factor A and Proliferation Marker in Prediction of Lymph Node Metastasis in Oral and Pharyngeal Squamous Cell Carcinoma. Arch Otolaryngol Head Neck Surg. 2008;134(12):1305–1311. doi:10.1001/archotol.134.12.1305
Copyright 2008 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2008
To explore the effect of Ki-67 and vascular endothelial growth factor A (VEGF-A) expression on the risks of advanced T category (T3,4) and positive lymph node involvement (N+) in oral and pharyngeal squamous cell carcinoma (SCC) compared with laryngeal SCC.
Immunohistochemical analysis of prospectively recruited patients.
A total of 147 previously untreated patients with different stages of SCC in the oral cavity, pharynx, and larynx.
Main Outcome Measures
Relative risks of T3,4 tumor and N+, a risk ratio comparing risks under high vs low marker expression.
A significant association of Ki-67 and VEGF-A expression with tumor T category was observed for oral and pharyngeal SCC and for laryngeal SCC (P ≤ .006). Regarding nodal status, Ki-67 expression was a significant risk factor for N+ in all tumors (P ≤ .009), whereas VEGF-A expression was related to N+ in oral and pharyngeal SCC only (P < .03). Analytically, Ki-67 expression alone in oral and pharyngeal SCC was associated with a relative risk of N+ of 3.83 (95% confidence interval, 1.22-11.99; P = .009), and additional expression of VEGF-A raised the value to 6.12 (2.09-17.93; P < .001). Moreover, the combined expression of both markers was 3.25 times more effective in predicting N+ for T1,2 tumor compared with T3,4 tumor.
Proliferative status was a common risk factor for N+ in all of the tumors in this series. Exploitation of VEGF-A in lymph node metastasis in addition to proliferation by oral and pharyngeal SCC but not by laryngeal SCC explains the clinical aggressiveness of oral and pharyngeal SCC, especially the early lymphatic invasion. In the management of cervical lymph nodes, combined expression of Ki-67 and VEGF-A may help identify patients at risk for occult metastases. This study suggests anti–VEGF-A therapy, an additional intervention to the classic antiproliferative regimen, for preventing lymphatic progression of oral and pharyngeal SCC.
Oral and pharyngeal squamous cell carcinoma (SCC) is clinically aggressive compared with laryngeal SCC. Most of these tumors show advanced invasion and spread to regional lymph nodes even at an early stage.1,2 A 20% to 50% incidence of positive lymph node involvement (N+) was reported for oral SCC2 and 65% to 80% for pharyngeal SCC.1 In contrast, laryngeal SCC seems to metastasize less avidly, with a much lower incidence of N+, that is, 1% for glottic SCC and 29% for supraglottic SCC.2 Different lymphatic vessel density among these sites3 has been suggested as a major determinant for lymphatic dissemination. Metastasis to a regional lymph node is the first indication of tumor metastasis competence.4 The high propensity for lymph node metastasis remains a challenge in treating SCC of the oral cavity and pharynx because the presence of cervical lymph node disease is associated with up to a 50% decrease in survival.5 Understanding the process involving this lymphatic progression will have a substantial influence on the management of these aggressive tumors.
Neoplastic cells with metastasis capacity acquire distinct cellular capabilities, such as the ability to proliferate without limit, to evade apoptosis, to escape immune surveillance, and to express factors that alter the growth of blood and lymphatic vessels so as to create conduits for tumor metastasis.6 Tumor proliferative status, a reflection of uncontrolled proliferation from loss of cell-cell contact inhibition, is traditionally regarded as an index of tumor aggressiveness. Indeed, reduced expression of E-cadherin, a cell adhesion molecule in epithelial tissue, can be observed in highly proliferative laryngeal and hypopharyngeal SCC with nodal diseases.7,8 Besides head and neck SCC (HNSCC),7-10 a significant association with tumor proliferation, as measured by Ki-67 expression and N+, has been reported for other tumor types.11,12
Tumor growth and metastasis are angiogenesis dependent.4,13 The family of vascular endothelial growth factors (VEGFs), including VEGF-A, VEGF-B, VEGF-C, VEGF-D, and VEGF-F, plays a key role in angiogenesis and lymphangiogenesis.13 Among them, VEGF-A is a key angiogenic factor and is most frequently used by a tumor to switch on its angiogenic phenotype.13 It can also mediate immunosuppression in HNSCC.14,15 Recently, VEGF-A has been shown experimentally to be a potent lymphangiogenic factor to promote lymph node metastasis by activating VEGF receptor 2 on lymphatic vessels16 and recruiting macrophages, which are known to produce another lymphangiogenic factor, VEGF-C/VEGF-D.17 In clinical study, VEGF-A expression is a common feature in most cancers. Several recent publications18-24 support its association with N+ in many types of solid tumors. In fact, differential metastatic gene expression profiles have been documented for laryngeal and hypopharyngeal SCC, that is, overexpression of several metastatic suppressor genes in laryngeal SCC against downregulation of the same genes in hypopharyngeal SCC.25 This difference may relate to the inherent aggressiveness or proliferative status of a cancer cell and the ability of the malignant cell to exploit VEGF-A to create channels for lymph node metastasis.
Many studies3,7,8,25 used tumors of the hypopharynx and the larynx as a natural model to study factors that determine the invasive/metastatic phenotype leading to the sharp difference in clinical courses between these tumors. Ki-67 and VEGF-A are markers of interest for their relevance to lymph node metastasis in HNSCC7-10,20-24 and other solid tumors11,12,18,19 in many recent studies. In this study, we explored how tumor proliferation as measured by Ki-67 and tumor-derived VEGF-A expression affects risks of advanced T category (T3,4) and N+ in oral and pharyngeal SCC compared with laryngeal SCC, We also evaluated whether these markers could explain the early nodal invasion of oral and pharyngeal SCC.
Patients diagnosed as having SCC of the oral cavity, pharynx, or larynx who underwent surgery at the Department of Otolaryngology, Ramathibodi Hospital, were consecutively included. Patients with distant metastases or recurrent diseases were excluded. Informed consent approved by the Ramathibodi Hospital ethics committee was obtained from all the patients. Tumor staging was assigned after surgery according to the American Joint Committee on Cancer classification.26 Patient and tumor characteristics are given in Table 1.
Five-micrometer sections of formalin-fixed, paraffin-embedded tumor specimen were mounted on silanized slides. The sections were deparaffinized and hydrated. Endogenous peroxidase was blocked in 3% hydrogen peroxide. Antigen retrieval was performed by heating specimens in 0.025M citrate buffer (pH, 5.8) using a microwave oven and 4 cycles of 5-minute heating at a mean (SD) of 94°C (1°C). The primary antibodies used were mouse monoclonal antibodies against Ki-67 (clone 7B11, IgG1 isotype, ready-to-use) (Zymed, San Francisco, California) and VEGF-A (monoclonal antibody 293, clone 26503, IgG2b isotype) (R&D Systems Inc, Minneapolis, Minnesota). The optimal concentration of monoclonal antibody 293 for HNSCC staining was 50 μg/mL. Specimens for Ki-67 staining were treated with 0.01% trypsin for 6 minutes at 37°C before nonspecific background blocking, including an avidin-biotin block, followed by 10 minutes of incubation with 10% nonimmune goat serum. For VEGF-A staining, an additional 1 hour of incubation at room temperature (23°C) with 5% bovine serum albumin was required to reduce the cytoplasmic background. A low-ionic-strength (ie, 0.01M) phosphate buffer (pH, 7.2) was used as diluent and washing solution. Treated specimens were incubated at room temperature with anti–Ki-67 antibody for 2 hours and at 4°C overnight with anti–VEGF-A antibody. Reaction of the primary antibody was detected by biotinylated anti–mouse antibody and streptavidin-peroxidase complex using a diaminobenzidine substrate and then was counterstained with hematoxylin. Negative reagent controls were included in all stainings. Anti–Ki-67 stained the nucleus of the proliferating cell, and anti–VEGF-A stained the cytoplasm of the cancer cell.
The fraction of Ki-67–labeled cells was determined using a point-counting technique and a 10 × 10 square graticule (area, 1 cm2) (Olympus, Tokyo, Japan) at magnification ×400. Counting the area with the highest number of positive cells was used rather than random counting because of its reproducibility (2.79% for hot spot counting vs 33.8% for random counting as observed in a preliminary study).27 Approximately 1000 tumor cells were counted. Expression of VEGF-A was quantitated using a computer-based image captured by a charged-couple device camera. A VEGF-A score varying from 0 to 4 was rated on the basis of stain intensity and percentage of cell stained. Libraries of images with 4 different color grades were prepared as a tool for grading. The reliability of these reference libraries was verified by 2 of us (V.B. and N.L.), who blindly rated the same series of 12 specimens. Good agreement was achieved, with a reasonable concordance correlation coefficient of 0.83.28 Images of 10 representative low-power fields (magnification ×100) were graded and averaged.
For this study, we stratified tumors as less proliferative and highly proliferative, with a cutoff point at 50% Ki-67 labeling index. Tumors in each of these categories were further classified as high and low VEGF-A expressers using the median VEGF-A score19 of 2.46 as a cutoff value. Frequency tables were constructed for the χ2 test or the Fisher exact test for the effect of enhanced expression of either or both Ki-67 and VEGF-A on risks of T3,4 tumor and N+ in oral and pharyngeal SCC compared with laryngeal SCC.28 All of the statistical tests were 2-tailed, and P ≤ .05 was considered statistically significant.
This study enrolled 147 patients (male to female ratio, 3.74:1; mean age, 62 years; age range, 27-84 years) (Table 1). Sixty patients had oral SCC, 34 had pharyngeal SCC, and 53 had laryngeal SCC. Most tumors (75.5%) were well differentiated. In staging, 62.6% of the patients had negative lymph nodes, but the distribution of these patients among T categories was more uniform. Oral and pharyngeal SCC was more likely to invade the lymph nodes (47.9%), even at the early stage (18.1%), whereas SCC of the larynx tended to expand locally, with a much lower chance of invading the lymph nodes (18.9%) (Table 2). A significant difference in the clinical courses of tumors from these sites (P < .001) had been verified by means of a χ2 test for mutual dependence among nodal status, T category, and tumor site using a 2 × 2 × 2 contingency table.28 Immunohistochemical study showed that the medians (ranges) for the Ki-67 labeling index and the VEGF-A score were 56.7% (0%-90.0%) and 2.5 (0-4.0), respectively. A cutoff level was chosen to distinguish tumors with high vs low marker expression. Because the distribution curve for the Ki-67 labeling index displayed a bimodal characteristic (kurtosis = 0.06), a dip in the curve at 50% was chosen as the cutoff point. In contrast, the curve for VEGF-A was unimodal (kurtosis = −0.97), with a peak close to the median. The median score was, thus, chosen as the cutoff value. This method of cutoff value selection has been used by other investigators.19 Tumors were stratified according to their patterns of marker expression. Despite the difference in the clinical courses of tumors from the oral cavity and pharynx vs the larynx, the expression of VEGF-A under a high or low Ki-67 index was independent of whether the tumor was of the oral cavity and pharynx or the larynx. A χ2 analysis of the 2 × 2 × 2 contingency table yielded a P = .18 (Table 2).
Because of the multitude of evidence supporting the association of Ki-67 or VEGF-A expression with nodal status and sometimes with tumor T category,7-12,18-24 we analyzed the effect of enhanced expression of either or both markers on T3,4 tumor and N+. Table 3 presents the risks of T3,4 tumor associated with marker overexpression. Under low VEGF-A expression, a high Ki-67 index was a significant indicator of increased risk of T3,4 tumor in oral and pharyngeal SCC (P = .005) and in laryngeal SCC (P < .001). Alternatively, a high VEGF-A score was a significant indicator of risk of T3,4 under low Ki-67 expression (P ≤ .006). However, in the presence of either high VEGF-A or Ki-67 expression, expression of the other marker was irrelevant in determining the T category (P ≥ .15). In assessing the effect of marker overexpression on tumor growth, the relative risk of T3,4 (RRT3,4), a comparative risk ratio under high vs low marker expression, was calculated for single and combined marker overexpression. For oral and pharyngeal SCC, the RRT3,4 was 3.39 (95% confidence interval [CI], 1.29-8.89) for Ki-67, 3.69 (1.37-9.96) for VEGF-A, and 3.70 (1.46-9.40) for combined marker overexpression. The RRT3,4 could not be determined for laryngeal SCC because the risk of T3,4 tumor under low marker expression was 0%.
Associated risks of N+ are detailed in Table 4. A characteristic difference was observed between oral and pharyngeal SCC vs laryngeal SCC. In oral and pharyngeal SCC, a high Ki-67 index under low VEGF-A expression was significantly associated with a higher risk of N+ (P = .009). The risk due to VEGF-A overexpression was nearly significant in less proliferative tumor (P = .07) and was highly significant in highly proliferative tumor (P = .03). Analytically, the RRN+ was 3.83 (95% CI, 1.22-11.99) for Ki-67, 3.07 (0.87-10.85) for VEGF-A, and 6.12 (2.09-17.93) when both markers were highly expressed. In laryngeal SCC, VEGF-A overexpression did not raise the chance of N+ in less proliferative tumor (P >.99) or in highly proliferative tumor (P = .47). Nevertheless, when data were pooled according to proliferative status, a high Ki-67 index was significantly associated with an increased risk of N+ (29.4% vs 0%; P = .008).
To evaluate whether Ki-67 and VEGF-A expression could explain the early lymphatic spread observed in oral and pharyngeal SCC, we compared the likelihood ratio for N+ (LRN+) under different patterns of marker expression in T1,2 vs T3,4 tumor. The LRN+, an index to predict the likelihood of a tumor being found to have N+, is calculated by dividing the percentage of N+ by the percentage of N0. Table 5 suggests that the combined overexpression of Ki-67 and VEGF-A is 3.25 times more effective in predicting N+ in T1,2 than in T3,4 tumor (LRN+ = 6.46 at T1,2 vs LRN+ = 1.99 at T3,4). However, no difference in the LRN+ was observed between early and advanced tumors with enhanced expression of either marker. Of note, more T1,2 (42.6%) than T3,4 (8.5%) tumors were low marker expressers and were relatively nonaggressive for their low LRN+ of 0.22.
Oral and pharyngeal SCC represents a biologically aggressive subset of HNSCC because of its high propensity for lymph node metastasis even at an early T category.1,2 Interest in identifying biomarkers capable of predicting the likelihood of cervical lymph node metastasis from a primary tumor specimen is being developed to enable the separation of patients at high risk for occult metastases from those at low risk. This will provide tremendous advantages for optimal treatment planning. Theoretically, uncontrolled proliferation and expression of growth factors to induce angiogenesis and lymphangiogenesis may contribute to lymphatic spread of the metastasis-competent cancer cells. In this study, we explored how Ki-67 and VEGF-A overexpression relate to risks of advanced growth (T3,4 tumor) and lymph node metastasis (N+) in oral and pharyngeal SCC compared with the less aggressive laryngeal SCC.
Regarding proliferative status alone, oral and pharyngeal SCC with a high Ki-67 index can expand to reach an advanced T category and can invade the lymph nodes even in the absence or at a low level of VEGF-A expression. On the other hand, laryngeal SCC tends to expand locally and is less effective at invading the lymph nodes. This finding confirms the classic belief that tumor proliferative status can reflect the invasiveness and metastatic potential of a cancer cell. In addition, more recent publications8-12 suggest the predictive value of the proliferation marker for lymph node metastasis in many other tumor types. Few questions remain as to why oral and pharyngeal SCC with a high Ki-67 index could effectively invade the lymph nodes even at an early T category, whereas laryngeal SCC, despite expressing the same level of Ki-67, showed less lymphatic aggression. Differences in metastatic gene expression profiles25 and the density of preexisting lymphatic networks3 observed in these tumors3,25 could possibly determine the chance of lymph node metastasis.
Vascular endothelial growth factor A has been demonstrated to promote the growth of early cancer in in vivo models29,30 and clinical studies.31,32 Likewise, we also noted the same effect of VEGF-A on less proliferative tumor, which is usually observed at the early T category. Recently, VEGF-A has been shown to promote tumor lymphangiogenesis and lymph node metastasis in some tumor models, for example, chemical-induced skin cancer using a transgenic mouse model16 and the T241 fibrosarcoma xenograft.17 However, in the 293 ENBA tumor xenograft33 and the transgenic pancreatic tumor model,34 VEGF-A promotes neither lymphangiogenesis nor lymph node metastasis. Variables that determine whether VEGF-A is lymphangiogenic probably include the isoform of VEGF-A, the level of VEGF receptor 2 expression on lymphatic endothelium, and recruitment by VEGF-A of monocytes and macrophages that express VEGF-C/VEGF-D.13 O-charoenrat et al23 demonstrated the co-expression of VEGF-A (isoforms 121 and 165) and VEGF-C in association with N+ in HNSCC. More recent publications on gastric cancer,35 breast cancer,19 and lung adenocarcinoma36 also support this notion.
In this study, a significant association of VEGF-A overexpression with N+ was found only in oral and pharyngeal SCC. Similar findings are documented for oral SCC20-22 but not for laryngeal SCC. In this series, VEGF-A overexpression moderately increased the risk of N+ in less proliferative tumor (RRN+ = 3.07). In highly proliferative tumor, the risk was additive to that due to proliferation (RRN+ = 3.83), to yield an overall RRN+ of 6.12. The underlying mechanism for VEGF-A in promoting lymph node metastasis in less proliferative tumor is unknown. It is likely that the growth-promoting activity of VEGF-A may aid this tumor to expand to gain access to the preexisting peritumoral lymphatics. Immunosuppressive and angiogenic properties of VEGF-A may support survival and growth of the metastatic cells in lymph nodes. In highly proliferative tumor, VEGF-A overexpression may involve lymphangiogenesis, leading to a much greater increase in the risk of lymph node metastasis. However, it remains uncertain whether VEGF-A by itself is lymphangiogenic or simply acts via the induction of other lymphangiogenic factors. Given that most N+ tumors in recent publications have been reported to co-express VEGF-A and VEGF-C19,23,35,36 and that VEGF-A treatment can upregulate VEGF-C in cultured endothelial cells,37 it is likely that lymph node metastasis in highly proliferative tumor may be facilitated by the co-expression of VEGF-A and VEGF-C. However, the expression of VEGF-C was not measured in this study.
Whether the expression of Ki-67 and VEGF-A can predict the chance of early lymphatic spread in oral and pharyngeal SCC is another major point of interest. A relatively high incidence of occult cervical metastases (21%-42%) in clinical stage I and II SCCs of the oral cavity and oropharynx is generally recognized.38,39 The ability to separate patients at high risk for occult diseases from those at low risk will spare the low-risk patients the morbidity of unnecessary treatment while adequately identifying patients who need aggressive therapy. Conventionally, an elective neck dissection is performed in cases of suspected occult disease by means of biopsy or radiologic investigations.40 This method, however, has the disadvantage of an unnecessary surgical procedure because a high percentage of clinically negative necks actually do not harbor metastases (approximately 60%-80%).38,39 To avoid the morbidity of neck dissection in patients in the low-risk group, a watchful waiting principle is adopted. This method also has the disadvantage of delayed detection of the recurrence when the investigations confirming the negative neck in fact produce a false-negative result.40 In this study, we discovered that the combined expression of Ki-67 and VEGF-A displays a greater likelihood (6.46:1) of finding N+ in T1,2 tumor against the chance of that in T3,4 tumor (1.99:1). This LR drops to 0.75 when either marker is highly expressed. Tumor with low marker expression is least likely to present with N+ (LRN+ = 0.22). On the basis of biomarker expression, patients with T1,2N0 SCC of the oral cavity and oropharynx who are at high risk for occult metastases can be distinguished from those at low risk.
In conclusion, these data demonstrate that SCCs of the oral cavity, pharynx, and larynx, although histologically similar, are biologically different. The ability to exploit VEGF-A in addition to proliferation in lymphatic progression makes oral and pharyngeal SCC clinically more aggressive than laryngeal SCC. On this basis, the conventional antiproliferative therapeutic intervention or the novel anti–VEGF-A and the anti–VEGF-C targeting system alone may not be effective enough to prevent lymphatic progression in oral and pharyngeal SCC. In an effort to identify patients who benefit from neck dissection, combined expression of Ki-67 and VEGF-A is particularly effective in revealing N+ in T1,2 tumor and, hence, may help identify patients who may be at considerable risk for occult metastases. A watchful waiting principle may be safely applied for patients whose tumors express low levels of the biomarkers.
Correspondence: Vipa Boonkitticharoen, PhD, Department of Radiology, Ramathibodi Hospital, Mahidol University, 270 Rama VI Rd, Bangkok 10400, Thailand (firstname.lastname@example.org).
Submitted for Publication: October 28, 2007; final revision received April 22, 2008; accepted May 5, 2008.
Author Contributions: Drs Boonkitticharoen, Kulapaditharom, and Leopairut had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Boonkitticharoen, Kulapaditharom, Leopairut, Kraiphibul, and Pochanukul. Acquisition of data: Boonkitticharoen, Kulapaditharom, Larbcharoensub, Cheewaruangroj, and Chintrakarn. Analysis and interpretation of data: Boonkitticharoen and Kulapaditharom. Drafting of the manuscript: Boonkitticharoen. Critical revision of the manuscript for important intellectual content: Kulapaditharom, Leopairut, Kraiphibul, Larbcharoensub, Cheewaruangroj, Chintrakarn, and Pochanukul. Obtained funding: Boonkitticharoen and Pochanukul. Administrative, technical, and material support: Boonkitticharoen, Kulapaditharom, Leopairut, Kraiphibul, Larbcharoensub, Cheewaruangroj, Chintrakarn, and Pochanukul. Study supervision: Boonkitticharoen, Kulapaditharom, and Leopairut.
Financial Disclosure: None reported.
Funding/Support: This work was supported by a grant from Mahidol University, Bangkok, Thailand.
Additional Contributions: We thank Koset Pinpradab, BSc, for technical assistance.