Applicability of Preoperative Nuclear Morphometry to Evaluating Risk for Cervical Lymph Node Metastasis in Oral Squamous Cell Carcinoma

Background We previously reported the utility of preoperative nuclear morphometry for evaluating risk for cervical lymph node metastases in tongue squamous cell carcinoma. The risk for lymph node metastasis in oral squamous cell carcinoma, however, is known to differ depending on the anatomical site of the primary tumor, such as the tongue, gingiva, mouth floor, and buccal mucosa. In this study, we evaluated the applicability of this morphometric technique to evaluating the risk for cervical lymph node metastasis in oral squamous cell carcinoma. Methods A digital image system was used to measure the mean nuclear area, mean nuclear perimeter, nuclear circular rate, ratio of nuclear length to width (aspect ratio), and nuclear area coefficient of variation (NACV). Relationships between these parameters and nodal status were evaluated by t-test and logistic regression analysis. Results Eighty-eight cases of squamous cell carcinoma (52 of the tongue, 25 of the gingiva, 4 of the buccal mucosa, and 7 of the mouth floor) were included: 46 with positive node classification and 42 with negative node classification. Nuclear area and perimeter were significantly larger in node-positive cases than in node-negative cases; however, there were no significant differences in circular rate, aspect ratio, or NACV. We derived two risk models based on the results of multivariate analysis: Model 1, which identified age and mean nuclear area and Model 2, which identified age and mean nuclear perimeter. It should be noted that primary tumor site was not associated the pN-positive status. There were no significant differences in pathological nodal status by aspect ratio, NACV, or primary tumor site. Conclusion Our method of preoperative nuclear morphometry may contribute valuable information to evaluations of the risk for lymph node metastasis in oral squamous cell carcinoma.


Introduction
Lymph node metastasis strongly influences the five-year survival rate and prognosis of oral cancer. Besides treating the primary lesion, appropriate management of the cervical lymph nodes is an important part of oral cancer therapy [1][2][3][4][5]. Up to 30% of patients with a clinical N0 neck may still harbor occult metastasis [7], and how this can be best managed remains unclear [5], especially in cases that were clinically or radiographically diagnosed as lymphnode positive but no metastasis had been found in neck dissection [8,9].
Lymph node metastasis is a complex multi-step process and cannot be explained by enlargement only [10]. Radical neck dissection has been considered the standard treatment procedure for cervical lymph node metastasis in oral cancer since Crile first described it in 1906 [11]. However, there has long been controversy over the indications, timing, and methods of neck dissection [2][3][4][5]. A reliable and accurate means of preoperative evaluation of cervical lymph node metastasis is therefore crucial for the correct management of oral cancer [1,4,6,8,12] and risk criteria should be established.
Many studies have reported the risk factors for cervical lymph node metastasis in oral cancer [8,[13][14][15][16][17][18][19]. Most have used a combination of several qualitative methods, including histopathologic parameters [13][14][15][16], gene expression [17][18][19], sentinel lymph node detection [20][21][22][23][24], and imaging techniques [25][26][27][28]. Quantitative analysis of nuclear variations has been undertaken for other types of lesions such as thyroid and breast cancers [29,30]. In a previous study investigating the relationship between nuclear variations of squamous cell carcinoma of the tongue and cervical lymph node metastasis, we performed preoperative nuclear morphometry and found that this quantitative characterization the nuclear features of the primary tumor was a potential criterion for predicting lymph node metastasis in squamous cell carcinoma of the tongue specifically [8]. However, the potential risk for lymph node metastasis in oral squamous cell carcinoma (OSCC) differs depending on the anatomical site of the primary tumor, such as the tongue, gingiva, mouth floor, and buccal mucosa [12,31,32].
In this study, we performed preoperative nuclear morphometry for OSCC cells obtained on preoperative biopsy from primary tumors in the tongue, gingiva, buccal mucosa, and mouth floor. The investigated parameters were the mean nuclear area, mean perimeter, nuclear circular rate, ratio of nuclear length to width (aspect ratio), and nuclear area coefficient of variation (NACV) as an objective parameter of anisonucleosis. The relationships between these parameters and pathologic nodal classification (pN) status, including level of the metastatic lymph nodes, were evaluated retrospectively.

Data collection
Data were retrospectively collected for patients who were histopathologically diagnosed with OSCC and underwent surgical management including neck dissection at the Department of Oral and Maxillofacial Surgery, Nagasaki University Medical and Dental Hospital between January 1986 and January 2001 and the Department of Oral and Maxillofacial Surgery, Shimane University Faculty of Medicine between 1981 and 2012. Recurrent cases were excluded.

Biopsy specimens and pathologic nodal classification
Biopsy was performed in all patients preoperatively and/or prior to neoadjuvant therapy. The biopsy specimens were fixed with 10% neutral buffered formalin for 24 h and were processed for routine paraffin embedded sections, then stained with hematoxylin and eosin.
All the lymph nodes dissected from the biopsy specimens were examined for pN status and level of the metastatic lymph nodes. Cervical lymph node level was determined based on the cervical lymph node metastatic guide [33], as shown in Fig. 1.

Image analysis and nuclear parameter measurements
Images of each section were stored using a standard light microscope (using x10 objective lens) connected to a computerized digital camera. The image data were analyzed by Mac Scope software (Mitani Co., Fukui, Japan) to estimate the various quantitative nuclear features (at least 100 nuclei per case). Nuclear margins were digitally marked under high power view on the computer screen to ensure measurement accuracy [8].
Mean (standard deviation) values of the nuclear area and perimeter were calculated from counts of the pixels capturing the nuclei and their edges. The nuclear circular rate and aspect ratio were automatically calculated to determine variations in shape; briefly, in a round circle, the circular rate and aspect ratio values correspond to 1: if the object is elliptical, the circular rate is ,1 and the aspect ratio is .1. NACV was calculated to express variations in size in individual cases (Fig. 2).

Statistical analysis
To examine differences in patient characteristics between pN-positive and pNnegative patients, we performed t-tests for continuous variables. p values less than 0.05 were considered statistically significant.
Logistic regression analysis was performed to identify the risk factors for nodepositive status. Odds ratios and confidence intervals (based on the Wald test) were also calculated. Candidate risk factors with p values less than 0.1 on the Wald test were selected. From among them, risk factors were determined by variable selection using Akaike's Information Criterion. The optimal cutoff values for measurements were obtained with the minimum p values from the t-tests. All analyses were performed using SAS version 9.3 software (SAS Institute Inc., Cary, NC).

Ethics statement
The Department of Oral and Maxillofacial Surgery, Shimane University Faculty of Medicine was responsible for the biopsy specimens used in this study. The Ethics Committee of Shimane University approved the study (Approval No.: 1286). Patients provided written informed consent for their data to be used in this study.
Histopathologic examination of the preoperative biopsy specimens identified squamous cell carcinoma in all 88 patients: well differentiated in 65, moderately differentiated in 19, and poorly differentiated in 4. Forty-six patents were pN positive, the number and level of the metastatic lymph nodes are shown in Table 2.

Risks for cervical lymph node metastasis according to pN status
Mean nuclear area was significantly larger in pN-positive patients than in pNnegative patients (97.6¡30.2 mm 2 and 73.2¡19.6 mm 2 , respectively, p50.0226), as was mean nuclear perimeter (37.6¡5.7 mm and 32.3¡4.9 mm, p50.0217). However, there were no significant differences between the two groups in relation to nuclear circular rate, aspect ratio, or NACV (Fig. 3).

Discussion
A simple and reliable method for evaluating the preoperative risk for lymph node metastasis would be indispensable in routine clinical practice, and our approach requires no special equipment or staining technique. Several papers discuss that nuclear shape is a critical factor in the characterization of many neoplastic and non-neoplastic proliferations [8], and irregularity of the nuclear shape is one of the morphological characteristics commonly used to determine the type or degree of neoplastic transformation [34]. Recently, in the field of oral and maxillofacial surgery, several studies using morphometric analysis have evaluated the relationship between nuclear morphometry and histological grading [35][36][37], malignancies [39][40][41][42][43], and metastatic potential [14,44]. Size and contour irregularities of the nuclei are important features in the grading of OSCC [35][36][37][38][39][40]. Furthermore, it was reported that the nuclear size was larger in proportion to the grade of malignancy [41,42]. On the whole, among the quantitative morphometric parameters of the nuclei analyzed in this study, nuclear area and nuclear perimeter were significantly larger in pN-positive cases than in pN-negative cases, while the nuclear circular rate was lower in pN-positive cases but not significantly so. These results suggest that malignant nuclei become aspherical, which would be consistent with previous reports [8,[35][36][37][38][39][40][41][42]. Regarding the patient's age, however, it has been reported that low age (,42 years) was associated with the development of cervical lymph node metastasis within a short time frame (#50 days) [45]. In the present study, age,65 years was suggested to be a risk factor for cervical lymph node metastasis in OSCC.
NACV was reported to be the most feasible parameter for predicting risk for lymph node metastasis in thyroid cancer and breast cancer [29,30]. Nuclear pleomorphism is considered the most important feature in the grading of OSCC [35]. However, in the present study, although NACV was higher in pN-positive cases it did not reach statistical significance, and thus it does not appear to be a reliable parameter for predicting risk for lymph node metastasis in OSCC.
As for the relationships we investigated between the quantitative morphometric parameters and nodal staging, nuclear area was significantly larger in pN-positive patients than in pN-negative patients. In addition, pN0 and pN2b status showed a significant difference in nuclear perimeter, nuclear circular rate, and NACV. However, the number of metastatic lymph nodes showed no significant correlation with NACV. Briggs et al. [13] reported that nuclear morphometric measurement was useful for evaluating metastatic potential in early squamous cell carcinoma of the mouth floor. Therefore, we also examined whether the applicability of our approach would be influenced by the site of the primary OSCC tumor. However, we found no significant differences in nuclear morphometric results between pN-positive and pN-negative cases according to sites in the gingiva, buccal mucosa, mouth floor, or tongue (Table 3).
In this study, 36 of 42 patients with pN-negative disease and 40 of 46 patients with pN-positive disease received neoadjuvant chemotherapy, radiotherapy, or both. It is likely that such preoperative therapies might affect the nodal staging. While our technique of preoperative nuclear morphometry using biopsy specimens appears to be applicable to help decide whether the lymph nodes harbor metastases or not, retrospective studies in general are insufficient to discuss a cause-effect relationship for the metastasis from evaluating the lymph node profiles alone (e.g., nodal staging and number of nodes). It is essential, therefore, to carry out a prospective study to verify the applicability of our method for predicting cervical lymph node metastasis. Future studies also need to address the applicability of clinical and biological marker analyses to accurately evaluate metastatic potential in OSCC preoperatively. Recently, the nuclear factor kappa B was reported to be a key protein in multi-step carcinogenesis, lymph node metastasis, and prognosis of oral, head, and neck squamous cell carcinoma [44,46]. The expression of a combination of nuclear factor kappa B or other markers should be examined further.
In conclusion, our method of preoperative nuclear morphometry may contribute valuable information to evaluations of the risk for lymph node metastasis in OSCC.