Maxillofacial Surgery, Maxillofacial Reconstruction and Function, Division of Maxillofacial and Neck Reconstruction, Graduate School, Tokyo Medical and Dental University, Tokyo, Japan
Maxillofacial Surgery, Maxillofacial Reconstruction and Function, Division of Maxillofacial and Neck Reconstruction, Graduate School, Tokyo Medical and Dental University, 1-5-45 Yushima, Bunkyo-ku, Tokyo 113-8549, Japan===
The management of occult cervical lymph node metastases originating from oral squamous cell carcinomas (OSCCs) remains controversial. The purpose of this study was to evaluate the value of cyclin D1 gene (CCND1) numerical aberrations in predicting the risk of late lymph node metastases.
Fluorescence in situ hybridization (FISH), using a BAC clone specific for CCND1, was performed on OSCC specimens obtained by fine-needle aspiration (FNA) biopsy from 45 patients with previously untreated TNM Stage I and II (T1-2N0M0) disease who had not undergone elective cervical lymph node dissection.
CCND1 numerical aberrations were observed in 15 (33.3%) of the 45 patients and were significantly associated with the mode of invasion of the primary tumor (P = 0.01) and the presence of occult lymph node metastases (P < 0.001). Twelve of these 15 patients (80%) developed late cervical lymph node metastases within 2 years of surgery for primary OSCCs. All patients with cluster-type amplification of CCND1 developed late lymph node metastases. Multivariate analysis showed that only CCND1 numerical aberrations (risk ratio, 8.685%, 95% confidence interval = 2.232–33.802, P = 0.002) independently predicted late cervical lymph node metastasis.
The presence of lymph node metastases is the most important prognostic factor in squamous cell carcinomas of the head and neck (HNSCCs), including oral squamous cell carcinomas (OSCCs). It is generally accepted that cervical lymph node metastasis in patients with OSCCs indicates a poor prognosis; survival is poorer in patients with cervical lymph node metastasis than in those without, and the possibility of distant metastasis increases when there is cervical node involvement.1, 2 Thus, optimal management of cervical lymph node metastasis is very important to improving survival. There are two approaches to the management of the N0 neck in Stage I and II (T1-2N0M0) OSCCs; one is to perform elective cervical lymph node dissection, the other is to employ a ‘wait-and-see’ policy. The current standard management of the N0 neck is to perform cervical lymph node dissection when the risk of occult cervical metastasis is higher than 20%.3 As a result, up to 80% of patients with a clinically node-negative neck will undergo unnecessary surgery with its associated morbidity. Conversely, although previous studies have reported that wait-and-see groups survived for longer than elective cervical lymph node dissection groups,4, 5 late cervical lymph node metastasis may occur if a wait-and-see strategy is used, and this can have more unfavorable characteristics, such as extracapsular spread and multiple lymph node involvement. Therefore, there currently is no definite consensus regarding the treatment of the N0 neck in patients with Stage I and II OSCCs.
Amplification of the chromosomal region 11q13 has been detected in various human carcinomas, including OSCCs.6, 7 This region encompasses several putative oncogenes, such as protein phosphatase 1 alpha, cyclin D1 (PRAD1/CCND1), FGF4 and FGF3 (formerly known as HST-1 and INT-2), and the EMS1 and SEA genes. Of these, the cyclin D1 gene (CCND1) is the most appropriate target for study because the 11q13 amplicon always includes CCND1, and CCND1 is consistently amplified and overexpressed in tumor cells with 11q13 amplification.8CCND1 encodes a critical cell cycle regulator protein that drives the cell cycle from the G1 to the S-phase. Cyclin D1 protein functions by activating cyclin-dependent kinase (CDK) 4 and 6, leading to phosphorylation of pRb with consequent depression of E2F-mediated transcription and promotion of cell cycle progression.9 It has been assumed that overexpression of this protein may lead to the propagation of unrepaired DNA damage, the accumulation of genetic errors, and a selective growth advantage for altered cells. Experimental studies have shown that overexpression of cyclin D1 protein in fibroblasts induces cellular transformation and genetic instability.10, 11 In HNSCCs, the frequencies of CCND1 amplification and overexpression are approximately 20–40% and 40–80%, respectively, and these genetic abnormalities have been reported to correlate significantly with aggressive tumor growth and a poor prognosis. Therefore, the amplification and overexpression of CCND1 in HNSCCs may be a valuable biologic marker of a poor prognosis, tumor aggressiveness, and recurrence.12–15
Various methods, including Southern blot hybridization, polymerase chain reaction (PCR), and immunohistochemistry, have been used to evaluate diagnostic and prognostic genetic markers in HNSCCs. However, these methods have inherent problems, particularly with OSCCs. Southern blot analysis is often unsuitable because OSCC specimens are usually too small to yield adequate purified DNA. Conversely, although PCR is suitable for use with small amounts of DNA, contamination with normal cells may cause data misinterpretation. Immunohistochemistry demands the availability of antigen-specific antibodies, and cannot identify genetic abnormalities, such as amplifications, deletions, and chromosomal rearrangements. These disadvantages can be largely overcome by using fluorescence in situ hybridization (FISH). This technique does not require purified DNA, and therefore the results are not affected by DNA degradation. Although FISH cannot be used to detect point mutations, it can rapidly and easily identify chromosomal aberrations in very small amounts of tumor tissue, such as samples obtained by fine-needle aspiration (FNA) biopsy.16, 17 FNA biopsies can obtain not only the tumor cells from small samples, but it is also well suited for preparing the slides for FISH analysis. Recently, FISH combined with FNA biopsy has been used in attempts to diagnose breast and prostate cancers based on numerical changes in chromosomes and genes.18, 19 This approach enables numerical aberrations in chromosomes and genes to be detected easily and rapidly in small samples and is a practical and useful tool for preoperative diagnosis.
Since 1998, we have been investigating CCND1 deregulation using FISH on FNA biopsies from primary OSCCs. Our previous studies have revealed that: 1) it is possible to investigate chromosomal aberrations, such as amplifications, deletions, and chromosomal rearrangements, in OSCCs using this technique; 2) CCND1 amplification detected by FISH is a more reliable prognostic indicator than CCND1 overexpression in OSCCs; 3) tumors that exhibit aberrant numbers of CCND1, including gene amplification and multiple copies associated with chromosome 11 polysomy, always overexpress cyclin D1 protein; and 4) such CCND1 numerical aberrations are significantly associated with an invasive tumor phenotype and pathologic lymph node status.20, 21 These results suggest that CCND1 numerical aberrations may be a good predictor of late cervical lymph node metastasis in OSCCs, and that FISH performed on FNA biopsies from the primary tumors would provide a rapid and useful method of identifying patients at high risk of cervical lymph node metastasis. Therefore, in the present study we investigated whether CCND1 numerical aberrations act as a reliable predictor of late lymph node metastasis in Stage I and II (T1-2N0M0) OSCCs.
MATERIALS AND METHODS
Forty-five patients with Stage I and II (T1-2N0M0) OSCCs who underwent partial resection through the mouth without elective cervical lymph node dissection at Maxillofacial Surgery, Graduate School, Tokyo Medical and Dental University (Tokyo, Japan) between April 1998 and May 2004 were selected for this study. No patients had positive surgical margins, preoperative or postoperative treatment, or recurrence at the primary site. All patients gave written informed consent according to our Institutional Review Board guidelines.
The mean age of the patients was 62.5 years (range, 21–89 yrs). The OSCC samples were derived from the tongue (n = 33), lower gingiva (n = 3), upper gingiva (n = 3), buccal mucosa (n = 4), and the floor of the mouth (n = 2). Clinical staging was performed according to the International Union Against Cancer TNM classification22; 29 patients were classified as having Stage I [T1N0M0] disease and 16 were classified as having Stage II [T2N0M0] disease. Cervical lymph node of patient was considered N0 when there was no palpable lymph node by physical examination and the size of the lymph node was <1 cm by both computed tomography and ultrasound without any area suggesting central necrosis or metastasis. Of the 45 patients, 17 developed late cervical metastases, while 28 demonstrated no further evidence of disease after surgery. The median follow-up period was 28.9 months (range, 6.2–67.8 mos). The primary tumors were classified histopathologically as well, moderately, or poorly differentiated according to their cellular differentiation as defined in the World Health Organization (WHO) classification.23 The mode of tumor invasion at the tumor–host border was also classified using a modification of the criteria of Jacobsson et al., which include five categories: Grade 1: well-defined borderline; Grade 2: presence of cords, less marked borderline; Grade 3: groups of cells, no distinct borderline; Grade 4C: diffuse invasion of cord-like tumor type; and Grade 4D: diffuse invasion of diffuse tumor type.24, 25
FNA Biopsy Samples and FISH
Tumor cell sampling and slide preparation for FISH analysis were performed as described previously.20, 21 A suspension of single cells was obtained by aspirating the OSCC with a 21-gauge needle. The cells were soaked in 0.05 M KCl solution for 2 minutes to tear the cell membranes and expose the nuclei. The cells were then fixed by adding an equal volume of methanol/acetic acid (3:1) solution (Carnoy). After centrifugation at 3000 revolutions per minute (rpm) for 10 minutes, the upper layer was exchanged for Carnoy solution. Centrifugation and solution exchange was repeated twice and the resulting upper layer was dripped onto glass slides under steam.
To determine the CCND1 copy number, we used a BAC clone probe specific for CCND1 (Vysis, Downers Grove, IL), labeled with Spectrum Orange, and chromosome 11 centromeric DNA, labeled with Spectrum Green. FISH was carried out as follows. Briefly, the materials on the slides were aged in 2× saline–sodium citrate (SSC)/0.1% (volume/volume [v/v]) NP-40 at 37 °C for 30 minutes and dehydrated through an ethanol series. The slides were then denatured in 70% (v/v) formamide/2×SSC at 75 °C for 5 minutes and again dehydrated through an ethanol series. The probe, denatured at 75 °C for 5 minutes, was placed on the denatured slides, covered with Parafilm (American National Can, Greenwich, CT), and incubated in a humid box at 37 °C overnight. After being washed at 45 °C three times in freshly prepared 50% (v/v) formamide/2×SSC for 10 minutes, SSC for 10 minutes, and 2×SSC/0.1% (v/v) NP-40 for 5 minutes, the slides were counterstained with 4,6-diamidino-2-phenylindole dihydrochloride (DAPI; 1 μg/mL).
An Olympus BX50 epifluorescence microscope (Olympus Corporation, Tokyo, Japan) equipped with ×60 and ×100 oil-immersion objectives and a triple-pass filter for Spectrum Green/Spectrum Orange and DAPI (Vysis) was used to count the fluorescent signals. Overlapping and damaged nuclei were ignored, and only intact nuclei were evaluated. Hybridization signals were counted in 100 interphase nuclei. The copy numbers of CCND1 and chromosome 11 centromere signals were counted for each nucleus.
Evaluation of FISH Analysis
The criteria for defining CCND1 numerical aberrations were as described in our previous report.20 When at least 20% of the nuclei exhibited 3 or more signals for CCND1, the tumor was considered to have a “CCND1 numerical aberration.” When the average CCND1 copy number per chromosome 11 centromere copy number per scored cell was >1, the tumor was defined as showing “CCND1 amplification.” When clusters of CCND1 signals were observed in more than 20% of 100 nuclei, the tumor was considered as showing “cluster-type amplification of CCND1.”
Statistical Analysis of FISH Results
The FISH results were compared with the patient's clinicopathologic information using the chi-square test and the two-tailed Fisher exact test. The clinicopathologic information included the patient's age, gender, tumor site, disease stage, histopathologic grading, presence of late lymph node metastasis, and mode of invasion according to the modified criteria of Jacobsson et al.24 Disease-free survival rates were calculated using the Kaplan–Meier method and statistical significance was determined using the log rank test. Disease-free survival time was defined as the interval between the date of first visit and the date of the development of late cervical lymph node metastases after surgery on the primary tumor. Multivariate survival analysis was performed using the Cox proportional hazards model. The level of significance was set at P < 0.05. All statistical analyses were performed using SPSS 10.5J software (SPSS Inc., Chicago, IL).
FISH Analysis of CCND1 Numerical Aberrations
FISH with a gene-specific probe for CCND1 and a centromere repeat probe for chromosome 11 was used to determine the CCND1 copy number. CCND1 numerical aberrations were identified in 15 of the 45 (33.3%) primary OSCCs. CCND1 amplification was detected in 11 of the 15 tumors that showed CCND1 numerical aberrations (Fig. 1B). Only 4 tumors showed multiple copies of CCND1 associated with chromosome 11 polysomy (i.e., more than 2 centromeric signals per cell) (Fig. 1C). Of the 15 tumors that exhibited CCND1 numerical aberrations, 7 showed cluster-type amplification of CCND1 (Fig. 1D).
Correlation between CCND1 Numerical Aberrations in OSCCs and Clinicopathologic Parameters
The correlations between CCND1 numerical aberrations and clinicopathologic parameters in the OSCC patients are summarized in Table 1. The presence of CCND1 numerical aberrations was significantly associated with a more diffuse invasive pattern (mode of invasion, Grade 4C or 4D) and late cervical lymph node metastasis (chi-square test and two-tailed Fisher exact test for trend, P = 0.01 and P < 0.001, respectively), indicating that CCND1 numerical aberrations are related to aggressive tumor traits in OSCCs. However, there were no significant correlations between the presence of CCND1 numerical aberrations and age, gender, tumor site, clinical disease stage, or cellular differentiation.
Table 1. Correlation between CCND1 Numerical Aberration and Clinicopathologic Parameters
Predictive Value of CCND1 Numerical Aberrations for Late Cervical Lymph Node Metastasis of OSCCs
Late cervical lymph node metastasis occurred in 17 of the 45 patients (38%) during follow-up after treatment of the primary tumor. Of the 30 patients whose tumors exhibited no CCND1 numerical aberrations, only 5 developed late cervical lymph node metastases. Conversely, 12 of the 15 patients (80%) whose tumors showed CCND1 numerical aberrations went on to develop postoperative cervical lymph node metastases. Among these patients, all 7 with cluster-type amplification of CCND1 developed late lymph node metastases within 2 years of surgery for the primary tumor (Table 2). Only 3 of the 15 patients (20%) whose tumors showed CCND1 numerical aberration and 25 of the 30 patients (83%) whose tumors lacked such genetic abnormalities achieved disease-free survival from metastatic disease. The disease-free survival from metastatic disease rate was significantly higher in patients without CCND1 numerical aberrations than in those with such aberrations (log-rank test, P < 0.0001) (Fig. 2).
Table 2. Relation between CCND1 Numerical Aberration and Late Cervical Lymph Node Metastasis
Late lymph node metastasis
CCND1 numerical aberration (+)
CCND1 numerical aberration (−)
Multivariate Cox Proportional Hazards Analysis of Clinicopathologic Factors and the CCND1 Status in Late Cervical Lymph Node Metastasis
A multivariate Cox proportional hazards analysis including clinicopathologic factors and CCND1 status revealed that only the presence of CCND1 numerical aberrations was significantly and independently predictive of late lymph node metastasis (risk ratio = 8.685, 95% confidence interval = 2.232–33.802, P = 0.002; Table 3). As shown in Table 2, the CCND1 status had a predictive success rate of 80% (12 of 15 tumors), with 70.6% sensitivity (12 of 17 tumors) and 89.3% specificity (25 of 28 tumors).
The preoperative detection of cervical lymph node metastases is crucial to the effective management of patients with OSCCs. The introduction of various imaging techniques, such as computed tomography, magnetic resonance imaging, and ultrasound, has led to an improvement in the preoperative detection of lymph node metastasis. However, the sensitivity of these imaging methods and clinical examination is only about 70%.26 Recently, positron emission tomography (PET) as a functional imaging modality has been applied to the detection of primary recurrence and regional and distant metastasis of head and neck neoplasms. However, this new imaging technique still has the limitation that the metastases must be of a minimal size to be detected. Thus, there are no imaging techniques capable of detecting micrometastases originating from Stage I and II (T1-2N0M0) OSCCs. Therefore, the management of the N0 neck in OSCC patients remains controversial. Identification of predictive markers of cervical lymph node metastasis in the primary tumor would be helpful in deciding the optimal treatment for Stage I and II OSCCs, and in achieving a high survival rate.
Many studies have investigated the role of CCND1 in the carcinogenesis of HNSCCs, including oral cancers. These have shown that amplification or overexpression of CCND1 may be a valuable biologic marker of poor prognosis, tumor aggressiveness, and recurrence of these malignancies.14, 15 Our previous studies have also indicated that CCND1 numerical aberrations, as detected by FISH on FNA biopsies of primary OSCCs, are significantly associated with an invasive phenotype and pathological lymph node status, and that these genetic alterations are a reliable prognostic predictor in OSCCs.20, 21 However, few data are available concerning the correlation between CCND1 status and the presence of clinically occult cervical lymph node metastases in HNSCCs,27 and the value of CCND1 status in predicting the risk of occult metastases in patients with Stage I and II (T1-2N0M0) OSCCs has not been sufficiently investigated.
In the present study, we investigated CCND1 numerical aberrations using FISH on FNA biopsies of primary Stage I and II OSCCs and evaluated the CCND1 status as a predictor of cervical lymph node metastasis. CCND1 numerical aberrations were observed in 15 of 45 (33.3%) primary Stage I and II OSCCs and were significantly associated with invasive phenotypes and late cervical lymph node metastasis. A multivariate Cox proportional hazards analysis demonstrated that CCND1 numerical aberrations were the only significant independent predictor of late cervical lymph node metastasis in patients with Stage I and II carcinoma of the oral cavity. Late cervical lymph node metastasis was observed in 12 of the 15 patients (80%) whose tumors showed CCND1 numerical aberrations. In particular, all seven patients whose tumors exhibited cluster-type amplification of CCND1 developed late lymph node metastases after surgery on their primary tumors. The CCND1 status, as measured by FISH on FNA biopsy samples, therefore had a sensitivity of 70.6% (12 of 17 samples), a specificity of 89.3% (25 of 28 samples), a positive predictive value of 80% (12 of 15 samples), a negative predictive value of 83.3% (25 of 30 samples), and an accuracy of 82.2% (37 of 45 samples). These findings suggest that CCND1 numerical aberrations are a reliable predictor of patients at high risk of developing late cervical lymph node metastases from Stage I and II OSCCs, and that analysis of the CCND1 status using FISH on FNA biopsies may be a useful tool for selecting patients who would benefit from elective cervical lymph node dissection.
A number of studies have reported on predictive factors for late cervical lymph node metastasis in patients with Stage I and II HNSCCs, including oral cancers.28–35 These studies examined the predictive value of various clinical and histopathological characteristics of the primary tumor, and showed that tumor thickness is an important factor in predicting late lymph node metastasis in SCC of the tongue.28–30 Although the tumor thickness is measured by simple and easy methods such as palpation and ultrasonography, the cut-off value for tumor thickness remains controversial, and accurate measurement is difficult preoperatively. Recently, changes in gene and protein expression and genetic alterations in the primary tumor have been explored as predictors of occult cervical lymph node metastasis, and coexpression of MMP-9 and TIMP-2, expression of Flt-4, and decreased expression of maspin and E-cadherin were found to be significantly associated with lymph node metastasis.31–35 However, although these biomarkers may be significantly associated with the metastatic potential of OSCCs, they cannot be used preoperatively to decide on later regional treatment. A few studies have simultaneously investigated clinical and histopathological characteristics and biological markers associated with late cervical lymph node metastasis in patients with Stage I and II OSCCs and applied linear discrimination analysis to these factors. Lim et al.35 recently investigated 12 clinicopathologic factors and 10 immunohistochemical biomarkers as predictors of late cervical metastasis in patients with T1-2N0M0 SCCs of the tongue and found that a tumor thickness of more than 4 mm, a mode of invasion grade of 3 or 4, and low expression of E-cadherin were independent predictive factors. Linear discrimination analysis revealed that the combination of these three factors yielded a predictive rate of 77.8%. In another study, Okamoto et al.33 evaluated 10 clinicopathologic and 4 immunohistochemical parameters as predictors of occult metastasis in Stage I and II SCC of the tongue and observed that an endophytic growth pattern, a tumor thickness of more than 4 mm, and overexpression of Flt-4 were significantly predictive. The combination of these three factors provided a predictive rate of 77.8% on linear discrimination analysis. In the present study, we investigated seven clinicopathologic factors and the CCND1 status in Stage I and II OSCCs and found that CCND1 numerical aberrations had a predictive rate of 80% and an accuracy of 82.2%. Thus, the CCND1 status has a predictive value at least as good as the factors identified in the two above-mentioned studies. Although direct comparisons with these previous data must be made with caution due to the different methods of detection used to identify the biological markers (FISH vs. immunohistochemistry) and the different primary sites (general OSCCs vs. SCCs specifically of the tongue), CCND1 numerical aberrations appear to be the most reliable single predictor of late cervical lymph node metastasis in Stage I and II OSCCs.
Our present and previous studies have shown that CCND1 numerical aberrations correlate significantly with an invasive phenotype and metastasis to the regional lymph nodes.20, 21 Our previous studies also found that tumors with CCND1 numerical aberrations always overexpress cyclin D1 protein. Thus, overexpression of this protein may affect the phenotypes of cancer cells. CCND1 encodes an important regulatory protein that promotes cell cycle progression by activating CDK 4 and 6. However, recent studies have indicated that cyclin D1 affects the activity of various other non-CDK-dependent cellular transcription factors, such as estrogen, DMP1, STAT3, and BETA2/ NeuroD.36 Although to our knowledge it is not known how overexpression of cyclin D1 protein affects the invasive and metastatic behavior of cancer cells, its CDK-independent biological activities may contribute to the acquisition of the invasive ability and metastatic potential by tumor cells. However, overexpression of other oncogenes on the 11q13 amplicon, such as EMS1 and TAOS1, may also contribute to the development of invasive and metastatic characteristics in cancer cells.8, 37 Another possibility is that CCND1 numerical aberrations might reflect general genomic instability in cancer cells, and such cells possess a more aggressive phenotype.38 Thus, this issue is still a matter of controversy and further investigations are required.
In conclusion, the current study found that CCND1 numerical aberrations seem to be an excellent predictor of late cervical lymph node metastasis in Stage I and II OSCCs. Evaluating CCND1 numerical aberrations preoperatively by using FISH on FNA biopsy samples may therefore be a useful tool for selecting patients at high risk of late cervical lymph node metastasis who would benefit from invasive procedures. In particular, we suggest that elective cervical lymph node dissection is indicated for patients with Stage I and II OSCC whose primary tumors show cluster-type amplification of CCND1.
Supported by grants 15209069 and 15791161 from the Scientific Research from the Ministry of Education, Science, Sport and Culture, Japan.