|
Indian Journal of Cancer, Vol. 47, No. 4, October-December, 2010, pp. 406-411 Original Article Prediction of lymph node metastases by preoperative nuclear morphometry in oral squamous cell carcinoma: A comparative image analysis study S Natarajan, S Mahajan, K Boaz, T George Department of Oral Pathology, Manipal College of Dental Sciences, Mangalore, India Correspondence Address: Code Number: cn10098 PMID: 21131754 DOI: 10.4103/0019-509X.73580 Abstract Context: Prediction of metastases of oral squamous cell carcinoma (OSCC) using incisional biopsy. Aims: To assess the efficacy of nuclear morphometry in predicting the nodal metastases of OSCC and to compare manual and computer-based image analyses. Materials and Methods: Sixteen cases of moderately differentiated squamous cell carcinoma (8 with metastasized lymph nodes and 8 node-negative cases) were examined using manual and computer-based image analyses. Nuclear area, perimeter, circular rate, larger to shorter axis ratio, and coefficient of variance of nuclear area (NACV) were calculated from 100 nuclei of the processed incisional biopsy specimen. Statistical Analysis: The parameters were compared between the node positive and negative cases using Student's t test and with the level of lymph node involved using ANOVA test and Bonferroni's post hoc assessment. Manual and computer-based image analyses were compared by Spearman's correlation. Results: A significant variation between the node positive and negative groups was observed with respect to circular rate (P = 0.031) and NACV (P = 0.036). The nuclear area (P = 0.009), perimeter (P = 0.0001), circularity (P = 0.045), and larger to shorter diameter ratio (LS ratio) (P =0.02) were significantly different in the levels of the nodes involved. There was good a correlation between manual and computer-based analyses (Spearman's rho, 0.616-0.824) in nuclear size parameters and the circular rate had a negative correlation (-0.118, P = 0.664). Nuclear circularity was better assessed by computer-based analysis. Conclusion: Preoperative morphometric assessment of the nuclear features may detect early cellular changes and thus, are useful in predicting nodal metastases in OSCC. Keywords: Image analysis, lymph node metastasis, morphometry, oral squamous cell carcinoma Introduction In oral squamous cell carcinoma (OSCC), the presence of tumor metastases in the regional lymph nodes at presentation is the most significant adverse prognostic factor and a major determinant of poor survival. Precise indicators to predict local recurrence and/or metastasis might thus be very valuable. [1] A number of studies have reported the predictive factor for cervical lymph node metastases in oral cancer. [2],[3] The predictions of lymph node metastasis in most of these studies were made on the basis of histopathologic parameters or imaging techniques. [4] The present study was carried out in order to assess the predictability of preoperative nuclear morphometry in detecting lymph node metastases. Morphometric analysis of nuclear variations of cancer cells were performed using preoperative biopsy specimens manually as well as by computerized image analysis. The relationship between these results and pathologic node classification (pN), including the number and level involved was studied retrospectively and a correlation between manual and computerized image analysis was attempted. Materials and Methods The subject group composed of 16 patients who had been histopathologically diagnosed as having squamous cell carcinoma and surgically treated with techniques including neck dissection at the Department of Oral and Maxillofacial Surgery, of our institute, from January 2001 to June 2006. Patients′ age ranged from 28 to 78 years (average of 59.25 years) and male to female ratio was 7:1. A detailed information of cases, including TNM classification is shown in [Table - 1]. All patients had been diagnosed preoperatively by an incisional biopsy. The tissues were fixed in 10% neutral buffered formalin for 24 h and were processed for routine paraffin embedded sections. Hematoxylin and eosin stained, 5μm thick sections from these processed incisional biopsy specimens were used for morphometric analysis. Image analysis was performed using manual and computer-aided image analyses. The sections were examined under 100Χ objective (Olympus CX 21, Mangalore, INDIA, ; oil immersion). A 200-point "plus" eyepiece reticule was used to examine and measure the larger diameter (D) and the smaller diameter (d) of 100 nuclei at the deepest invading part of the tumor. Using these values, the nuclear area (πdD/4), perimeter [π(D + d)/4], circular rate (1 − 4area/πD2), LS ratio (D/d), and the coefficient of variation of nuclear area (NACV) [100 Χ standard deviation (SD) of nuclear area/mean nuclear area] were calculated using the respective formulae. [4] The image of each field corresponding to manual count was photographed (using Sony cybershot DSCp200, 7.2 megapixel digital camera) and examined using ImageJ software (ImageJ 1.36b, National Institutes of Health, USA, working on Java 1.5.0_06). The ImageJ software was calibrated for the magnification and 100 nuclei were then separated by thresholding under green mode. The corresponding values of the larger diameter, smaller diameter, nuclear area, perimeter, and circular rate were accurately derived. In each nucleus, the mean nuclear area and mean nuclear perimeter were derived by the automated totaling of pixels corresponding to the thresholded nuclei. NACV were calculated accordingly form the above parameters. Circular rate and LS ratio denote the variation in shape, the values of which correspond to 1 in case of a round circle. Circular rate becomes less than 1 and LS ratio more than 1 if the object is elliptical. In this study, all the patients had undergone neck dissection varying from submandibular or supraomohyoid to radical dissection. All the lymph nodes dissected from the specimens were examined in order to determine the pN and the number and the level of involved metastatic lymph nodes. The above-obtained parameters were statistically compared with the lymph node involvement. Statistical differences between the 2 groups of positive and negative lymph node metastasis were determined using independent sample Student′s t test. Analysis of variance (ANOVA) was performed between the level of involvement of positive nodes and each parameter. P values < 0.05 were considered statistically significant. Results The assessment of tumor and host characters of the incisional biopsy specimens using the Anneroth′s system, [5] grouped them into moderately differentiated squamous cell carcinoma. The demographics, pN status, number of nodes involved and the type of neck dissection are shown in [Table - 1]. The comparison of the parameters between the lymph node involved and free cases were compared using Student′s t test, both in manual and computerized analyses [Table - 2]. Among the parameters, the circular rate was significantly lower in pN(+) cases (0.7488 ± 0.036) compared with pN0 cases (0.7825 ± 0.1669) when assessed by computer-based analysis [Figure - 1]. The NACV was significantly higher (P = 0.036) in pN(+) cases than in pN0 cases. In the computerized assessment, the nuclear area and perimeter were higher in pN(+) cases than in pN0 cases but were not statistically significant [Figure - 2]. The L/S ratio also showed higher values in the pN(+) cases by manual analysis. Manual analysis showed a similar trend in all the parameters except the nuclear area for which the mean values were higher in the pN0 cases (95.8377 ± 29.81784) as opposed to the pN(+) cases (93.9334 ± 21.37655). The correlation between each morphometric parameter and pN is shown in [Table - 3a]. The ANOVA of the levels of positive nodes and parameters measured by manual analysis showed significant variation in the SDs of nuclear area (P = 0.009), perimeter (P = 0.0001), LS ratio (P = 0.02), and the values of NACV (P = 0.001). The parameter of circular rate showed a significant difference (P = 0.045) in the computer-based analysis. Bonferroni′s post hoc correction [Table - 3b] to check the variability within the groups showed the maximum variation between pN0 and pN2a for the manual parameters. The circularity rate measured by image analysis also showed a significant difference between pN0 and pN2b (P = 0.001). A positive correlation between the manual and computer-based analyses was observed as indicated by Spearman′s correlation analysis. The values for nuclear area (0.797), perimeter (0.824), and NACV (0.616) were all positive and closer to value ′1′ [Table - 4]. This indicated that the 2 systems give agreeable values. The parameter of circularity in manual and computer image analyses showed a negative correlation between the 2 techniques of analysis with a coefficient of −0.118 (P = 0.664). Discussion Abnormal nuclear morphology is a hallmark of neoplasia, but much remains to be learned about the causes and effects of abnormalities. The 5-year survival of non-lymph node involved vs the lymph node involved cases of squamous cell carcinoma drops from 85% to 41%. [6] OSCC when treated surgically is not always accompanied by cervical neck dissection. The frequency of occult metastasis (false-negative neck) is stated to be 20-45%. [7] The metastasis to lymph nodes manifests when specific cellular and nuclear changes representing a distinctive clone of tumor cells occur. The identification of this clone at an early stage will greatly aid in reducing the morbidity of OSCC cases. The nuclear changes taking place in neoplasia include the genetic content (DNA, RNA content), the transcription factors, nongenetic content (histones and nuclear proteins), and the nuclear membrane changes. Nuclear morphology appears to determine in large part by the nuclear matrix that contains the structural elements of pore complexes, lamina, an internal network of filaments, and nucleoli. The matrix gives the nucleus its 3-dimensional organization and shape, serves as an important role in DNA organization and is in continuity with the cytoskeleton. Altered nuclear morphology of transformed cells, some mutagens, and transformation-related viral protein are known to associate with nuclear matrix changes. In addition, the manipulation of cell and nuclear shape has been shown to affect DNA synthesis and can be well correlated with the quantity of DNA and histones. [8] The nuclear content and therefore the content of DNA and histones are directly proportional to the nuclear size, which is quantitated by the morphometric parameters of nuclear diameter, nuclear area, and perimeter. Tumor cells with greater diameter are known to show a more aggressive pattern of invasion and higher incidence of nodal metastasis. [9] Sarker et al, 1997, [10] found increased nuclear area in relation to positive neck node but could not arrive at a statistically significant value. The SD of the nuclear area and perimeter are a measure of variation of the nuclear size, and consequently, the pleomorphism. The biologic aggressiveness of a tissue is expressed by the irregularity and variance of nuclear shape, which hints toward neoplastic transformation and progression. In the present study, we noted an increased nuclear area, perimeter, and LS ratio values in the pN(+) cases but they were not statistically significant. Thus, these parameters are not effective in predicting the lymph node metastasis. On the other hand, the SDs of the nuclear area (P = 0.009), perimeter (P = 0.0001), and LS ratio (0.02) showed a significant difference between the levels of lymph nodes involved [Table - 3a]. The combination of the SD and the mean nuclear area (NACV) also showed a significant difference between the 2 groups (P =0.036; [Table - 2]). This observation emphasizes that although there is a minimal increase in the size of the nucleus, the variation in shape, indicative of pleomorphism, has better predictive value to detect the metastatic clone in the primary tumor. Tumor invading in small groups with wide spread cellular dissociation and spindling is known to have a higher tendency to metastasize to regional lymph nodes as compared with tumor that invaded with pushing fronts or in solid cords. McGavran and co-workers [11] stated that laryngeal carcinomas with "pushing" margins have a low (12%) incidence of lymph node metastasis, whereas a high (54%) incidence of metastasis is seen in carcinomas with infiltrating margins. Yamamoto et alalso noted that carcinomas that had invaded or infiltrated the host tissues as small aggregates or single spindled cells had a higher metastatic rate (67.7%) than those with pushing margins (13.9%). [12] The measurement of the spindling of the cells is done efficiently by the circular rate. In our study, the circular rate in pN(+) were significantly lower (P = 0.031), which indicated that in the group of metastasized tumor, the nuclei were less spherical and the shapes of nuclei became more elliptic. The circular rate may thus be an indication of anaplastic progression of the disease, which may lead to metastatic potential. On comparison of the manual and the computer-based image analyses techniques, the parameters of nuclear area, perimeter, and the NACV showed similar progression as per the Spearman′s coefficient. The parameter of circularity was significant only in the computer-based image analysis. This indicated that the computer-based image analysis technique measured the subtle variations on the nuclear membrane more accurately. The manual technique calculates the circular rate by incorporation of the longer and shorter diameters of nucleus in a formula to give the circularity. Pixel optical density-based thresholded images accurately measure the nuclear membrane irregularity thus allowing us to harness the hidden data missed by manual morphometric techniques. Conclusion We have highlighted a feasible method to extract valuable morphometric information for predicting lymph node metastasis and the disease progression in squamous cell carcinoma of the oral cavity. Computer-based morphometry is a valuable tool in detecting subtle microscopic changes in the cell. This may be useful to detect the evolution of tumor cells at the earliest, where they exhibit molecular events that permit metastasis, such as the expression of cellular receptors and the production of enzymes to facilitate their access to lymphatic vessels. References
Copyright 2010 - Indian Journal of Cancer The following images related to this document are available:Photo images[cn10098t2.jpg] [cn10098t1.jpg] [cn10098t4.jpg] [cn10098f2.jpg] [cn10098t3.jpg] [cn10098f1.jpg] |
|