J Oral Pathol Med (2015) 44: 680–684 © 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
doi: 10.1111/jop.12280
wileyonlinelibrary.com/journal/jop
Nuclear fractal dimension in oral squamous cell carcinoma: a novel method for the evaluation of grading, staging, and survival Gabriella Mincione1, Marta Di Nicola1, Maria Carmela Di Marcantonio1, Raffaella Muraro1, Adriano Piattelli2, Corrado Rubini3, Enrico Penitente2, Marcello Piccirilli2, Giuseppe Aprile4, Vittoria Perrotti2, Luciano Artese2 Department of Experimental and Clinical Sciences, University “G. d’Annunzio” Chieti-Pescara, Chieti, Italy; 2Department of Medical, Oral and Biotechnological Sciences, University “G. d’Annunzio” Chieti-Pescara, Chieti, Italy; 3Department of Pathology, Polytechnic University of the Marche, Ancona, Italy; 4Private Practice, Rome, Italy 1
Fractal dimension (FD) in tissue specimens from patients with oral squamous cell carcinoma (OSCC) was evaluated. FD values in different stages of OSCC, and the correlations with clinicopathological variables and patient survival were investigated. Histological sections from OSCC and control non-neoplastic mucosa specimens were stained with hematoxylin–eosin for pathological analysis and with Feulgen for nuclear evaluation. FD in OSCC groups vs. controls revealed statistically significant differences (P < 0.001). In addition, a progressive increase of FD from stage I and II lesions and stage III and IV lesions was observed, with statistically significant differences (P = 0.003). Moreover, different degrees of tumor differentiation showed a significant difference in the average nuclear FD values (P = 0.001). A relationship between FD and patients’ survival was also detected with lower FD values associated to longer survival time and higher FD values with shorter survival time (P = 0.034). These data showed that FD significantly increased during OSCC progression. Thus, FD could represent a novel prognostic tool for OSCC, as FD values significantly correlated with patient survival. Fractal geometry could give insights into tumor morphology and could become an useful tool for analyzing irregular tumor growth patterns. J Oral Pathol Med (2015) 44: 680–684 Keywords: fractal analysis; fractal dimension; oral squamous cell carcinoma; prognostic factors
Correspondence: Vittoria Perrotti, Via dei Vestini 31, 66100 Chieti, Italy. Tel: +39-0871-3554083, Fax: +39-0871-3554076, E-mail: v.perrotti@ unich.it Accepted for publication October 2, 2014
Introduction Oral cancer accounts for 2–4% of all cancer cases with 42 440 worldwide estimated new cases in both sexes in 2014 (1). Oral squamous cell carcinoma (OSCC) represents the most frequent type, accounting for more of 90% of all oral carcinomas (2) and is associated with significant mortality and morbidity. OSCC is a complex malignancy where environmental factors, infective agents, and genetic alterations most likely interact. Tobacco use and alcohol consumption are regarded as the main risk factors, and it seems that their effect is synergistic (3). Human papilloma virus (HPV) infection is emerging as the leading risk factor in cancers of the oropharynx and is an independent prognostic factor for survival for these patients (4, 5). Even if most cases of oral cancer are diagnosed on the sixth and seventh decades, with the highest prevalence noted in male patients over 65 years (6), there is concern about the incidence increase in younger patients and in women. Indeed, the male to female ratio has been steadily shifting mainly due to the women increasing smoking habits. Oral squamous cell carcinoma typically presents as a persistent mass, nodule, or indurated ulcer. Symptoms are uncommon in early stages of the disease, but become frequent with advanced local invasion. Behavior of OSCC depends on its site of origin: each anatomic site displays a particular spread pattern and has a different prognosis. Carcinoma of the oral cavity spreads primarily to the lymph nodes of the neck. Metastases may, however, occur through the bloodstream, localizing especially in the lungs. Eliminating risk factors is crucial to prevent recurrences. In spite of the progresses on the understanding of OSCC biological and molecular characteristics, and of the different therapeutic approaches (surgery, radiotherapy, and/or chemotherapy), OSCC patients have poor overall survival rates (~50%) due to frequent recurrences and second primary tumors. Indeed, survival rates have not improved significantly over the past three decades (7). Therefore, early
Prognostic value for FD in primary OSCC Mincione et al.
diagnosis, before spreading of the tumor to other tissues, is an important issue, because prevention may greatly improve the cure rate, up to 90%. Early diagnosis requires a careful monitoring of oral mucosa pre-cancerous lesions, such as leukoplakia and erythroplakia, to ensure the secondary prevention of neoplastic forms. However, to date, in spite of the early clinical monitoring of this malignancy, mortality, and incidence appear to be still unchanged, suggesting the need to identify novel prognostic factors. Recently, fractal geometry has been used to go over the limits of the classical histomorphometrical approaches. Fractal is a geometric shape, characterized by being scalefree, that is, it appears similar whatever the scale used to look at them and is described by mathematic formulas. With fractal geometry, it is possible to quantify the morphological features routinely used by pathologists to qualitatively characterize and describe malignancies, that is, wrinkled borders. An ideal tool to describe a fractal is the fractal dimension (FD). FD is a shape index, and it can be defined as a measure of irregularly shaped objects. The lower the value of FD, the more regular or chaotic is the shape of the objects (8, 9). Therefore, it can be considered a powerful parameter to assess and represent shape variability of particular structures/cells/nuclei in neoplastic tissues as compared to normal counterparts. Indeed, several studies used this approach to describe different tumors, that is, breast carcinoma (10), lung (11), and the oral cavity (12), and its association with known clinicopathological parameters and overall survival. The aim of this study was to investigate FD as a novel prognostic tool for OSCC by correlating FD values with clinicopathological features and survival of patients affected by OSCC.
Material and methods Study design and pathologic examination Sixty-four primary oral squamous cell carcinomas (OSCC) and 10 histologically normal oral samples were obtained from the Department of Medical, Oral and Biotechnological Sciences, University ‘G. d’Annunzio’ Chieti-Pescara, Chieti, Italy, and the Department of Pathology, Polytechnic University of the Marche, Ancona, Italy. The experiments were undertaken with the understanding and written consent of each subject to use the tissues for research purposes and according to ethical principles, including the World Medical Association Declaration of Helsinki. All patients enrolled in this study had a OSCC clinical history and available histology paraffin material. Of the 64 patients with OSCC, 49 were males and 15 females, while in the control group, three were male and seven females. The mean age was 63.6 12.8 years and 50.0 21.9 years for OSCC and control groups, respectively. There was no statistical difference when comparing sex and the mean age in the two groups. The tumor topography, grading, and staging of the samples in the study were shown in Table 1. For all cases, the available hematoxylin–eosin sections were reviewed by two experienced pathologists (LA and CR) in order to confirm the
Table 1 Clinicopathological squamous carcinoma
characteristics
of
patients
with
oral
681
Variable Gender, n (%) Male Female Age (year), mean SD Topography, n (%) Tongue Mucosa geniena Other localizations Stage, n (%) I–II III–IV Grading, n (%) 1 2 3
49 (76.6) 15 (23.4) 63.6 12.8 10 (15.6) 49 (76.6) 5 (7.8) 32 (50.0) 32 (50.0) 24 (37.5) 30 (46.9) 10 (15.6)
original diagnosis and to select the representative neoplastic areas. Two 3-microns sections were prepared from each specimen and were mounted on super frost plus-coated glass slides (Menzel-Gl€aser, Braunschweig, Germany). One of them, stained with hematoxylin–eosin was used for pathological reevaluation, and the other one was stained with Feulgen stain for evaluation of the nuclei. The histological status of the lesions was defined after examination of the complete section under light microscopy, and the neoplasms were staged as: G1 (well differentiated), G2 (moderately differentiated), G3 (poorly differentiated) OSCC. Histological material of 10 patients randomly selected and treated for non-neoplastic lesions was studied as control. Nuclear fractal dimension For standard nuclear histomorphometry, three representative high-power fields (409) of Feulgen stains were randomly selected and photographed from each biopsy. The images were captured with a digital CCD-videocamera and were stored as high quality TIFF files, 1600 9 1200 pixels and 16.7 million colors (24-bit). The microscope and camera settings were kept consistent during the whole procedure. The above-mentioned image files were automatically transformed to binary images (image binarization converts an image to a black and white image suitable for fractal analysis) using ImageJ 1.40 g (Wayne Rasband; National Institute of Health, Bethesda, MD, USA). After selecting a 4.3 9 4.3 cm area on each image, the fractal dimension (FD) was estimated by using the box-counting method by means of ImageJ 1.40 g (Wayne Rasband; National Institute of Health). Statistical analysis Continuous parameters were reported as either mean and standard deviation (SD) or median and interquartile range. Categorical parameters were reported as frequency and percentage. The statistical analysis was performed using nonparametric tests. The Mann–Whitney U-test was used to evaluate statistical significant differences in FD values between groups (cases and control). J Oral Pathol Med
Prognostic value for FD in primary OSCC Mincione et al.
682
Table 2 Fractal dimension values respect to different clinicopathological characteristics of patients with oral squamous carcinoma Variable Gender Male Female Topography Tongue Mucosa geniena Other localizations Stage I–II III–IV Grade 1 2 3
Mean SD
Min–Max
P-value
1.41 0.13 1.38 0.08
1.16–1.67 1.26–1.57
0.568a
1.43 0.12 1.39 0.12 1.49 0.13
1.23–1.67 1.16–1.65 1.30–1.63
0.130b
1.35 0.21 1.44 0.11
1.16–1.67 1.23–1.65
0.003
1.30 0.08 1.41 0.06 1.61 0.04
1.16–1.41 1.27–1.53 1.55–1.67
0.001b
a
Mann–Whitney U-test. Kruskal–Wallis H-test. Statistically significant P-value are shown in bold. b
Cox proportional hazard method was applied to estimate cancer specific survival rate (CSS) and to determine the impact of FD values as independent prognostic factor for OSCC. All statistical analysis was performed using SPSSâ Advanced StatisticalTM 13 (2004, Chicago, IL, USA) software package.
Results Figure 1 Representative images of binarized and feulgen-stained samples from normal oral mucosa (control) and primary oral squamous cell carcinomas with different grade of differentiation (G1, G2, and G3).
The images showed cell nuclei from normal oral epithelium (controls), from well-differentiated OSCC (G1), moderately differentiated OSCC (G2), and poorly differentiated OSCC (G3) (Feulgen, 409). To the side of each image, there was a selected field of the same image after binarization (Fig. 1).
The Kruskal–Wallis H-test or Mann–Whitney U-test, when appropriate, was used to evaluate differences between FD values after stratification of the cases regarding their clinicopathologic characteristics. Post hoc analysis was performed with Kruskal–Wallis multiple comparison test.
Fractal Dimension value in OSCC groups and control The median FD was 1.15 (1.14–1.17) and 1.39 (1.35–1.43) for control and test groups, respectively, with statistically significant differences in FD value between two groups (P < 0.001) (Fig. 2).
Figure 2 Box-whiskers graphs of fractal dimension in control and test groups. Box-whiskers plots show the 25th and 75th percentile range (box) with 95% confidence intervals (whiskers) and median values (transverse lines in the box). A statistical significant differences was found between the two groups (P < 0.001, Mann–Whitney U-test).
Figure 3 Box-whiskers graphs of fractal dimension for stage. Boxwhiskers plots show the 25th and 75th percentile range (box) with 95% confidence intervals (whiskers) and median values (transverse lines in the box). A statistical significant difference was found between groups (P < 0.001, Kruskal–Wallis H-test).
J Oral Pathol Med
Prognostic value for FD in primary OSCC Mincione et al.
value was found between the different degrees of tumor differentiation groups (P = 0.001). Moreover, a statistically significant difference comparing OSCC with control groups (P < 0.001) was found (Fig. 4).
Figure 4 Box-whiskers graphs of fractal dimension for grade. Boxwhiskers plots show the 25th and 75th percentile range (box) with 95% confidence intervals (whiskers) and median values (transverse lines in the box). A statistical significant differences was found between the three groups (P < 0.001, Kruskal–Wallis H-test).
Relationship between Fractal Dimension and stage The clinical tumor stage at diagnosis has an important impact on treatment decisions and consequently on health outcomes. Thus, we evaluated the relationship between FD and stage. The basic statistical parameters of FD for clinicopathological characteristics of patients were shown in Table 2. The results obtained demonstrated a progressive increase of the FD value from stages I and II to stages III and IV, and the differences were statistically significant (P = 0.003). Moreover, a statistically significant difference when comparing OSCC and control groups (P < 0.001) was found. The dot in the graph indicates a outlier value (Fig. 3). Relationship between Fractal Dimension and tumor grading The assessment of relationship between FD and tumor grading was also examined to gain better insight into its prognostic significance in OSCC. The mean FD may differ according to histologic grades. FD values were proved to be low in G1 group (well differentiated) and high in G3 group (less differentiated tumors). A significant difference in the average nuclear FD
Figure 5 Overall survival curves stratified for median of FD values. There is statistically significant difference in outcome between cases with FD value
doi: 10.1111/jop.12280
wileyonlinelibrary.com/journal/jop
Nuclear fractal dimension in oral squamous cell carcinoma: a novel method for the evaluation of grading, staging, and survival Gabriella Mincione1, Marta Di Nicola1, Maria Carmela Di Marcantonio1, Raffaella Muraro1, Adriano Piattelli2, Corrado Rubini3, Enrico Penitente2, Marcello Piccirilli2, Giuseppe Aprile4, Vittoria Perrotti2, Luciano Artese2 Department of Experimental and Clinical Sciences, University “G. d’Annunzio” Chieti-Pescara, Chieti, Italy; 2Department of Medical, Oral and Biotechnological Sciences, University “G. d’Annunzio” Chieti-Pescara, Chieti, Italy; 3Department of Pathology, Polytechnic University of the Marche, Ancona, Italy; 4Private Practice, Rome, Italy 1
Fractal dimension (FD) in tissue specimens from patients with oral squamous cell carcinoma (OSCC) was evaluated. FD values in different stages of OSCC, and the correlations with clinicopathological variables and patient survival were investigated. Histological sections from OSCC and control non-neoplastic mucosa specimens were stained with hematoxylin–eosin for pathological analysis and with Feulgen for nuclear evaluation. FD in OSCC groups vs. controls revealed statistically significant differences (P < 0.001). In addition, a progressive increase of FD from stage I and II lesions and stage III and IV lesions was observed, with statistically significant differences (P = 0.003). Moreover, different degrees of tumor differentiation showed a significant difference in the average nuclear FD values (P = 0.001). A relationship between FD and patients’ survival was also detected with lower FD values associated to longer survival time and higher FD values with shorter survival time (P = 0.034). These data showed that FD significantly increased during OSCC progression. Thus, FD could represent a novel prognostic tool for OSCC, as FD values significantly correlated with patient survival. Fractal geometry could give insights into tumor morphology and could become an useful tool for analyzing irregular tumor growth patterns. J Oral Pathol Med (2015) 44: 680–684 Keywords: fractal analysis; fractal dimension; oral squamous cell carcinoma; prognostic factors
Correspondence: Vittoria Perrotti, Via dei Vestini 31, 66100 Chieti, Italy. Tel: +39-0871-3554083, Fax: +39-0871-3554076, E-mail: v.perrotti@ unich.it Accepted for publication October 2, 2014
Introduction Oral cancer accounts for 2–4% of all cancer cases with 42 440 worldwide estimated new cases in both sexes in 2014 (1). Oral squamous cell carcinoma (OSCC) represents the most frequent type, accounting for more of 90% of all oral carcinomas (2) and is associated with significant mortality and morbidity. OSCC is a complex malignancy where environmental factors, infective agents, and genetic alterations most likely interact. Tobacco use and alcohol consumption are regarded as the main risk factors, and it seems that their effect is synergistic (3). Human papilloma virus (HPV) infection is emerging as the leading risk factor in cancers of the oropharynx and is an independent prognostic factor for survival for these patients (4, 5). Even if most cases of oral cancer are diagnosed on the sixth and seventh decades, with the highest prevalence noted in male patients over 65 years (6), there is concern about the incidence increase in younger patients and in women. Indeed, the male to female ratio has been steadily shifting mainly due to the women increasing smoking habits. Oral squamous cell carcinoma typically presents as a persistent mass, nodule, or indurated ulcer. Symptoms are uncommon in early stages of the disease, but become frequent with advanced local invasion. Behavior of OSCC depends on its site of origin: each anatomic site displays a particular spread pattern and has a different prognosis. Carcinoma of the oral cavity spreads primarily to the lymph nodes of the neck. Metastases may, however, occur through the bloodstream, localizing especially in the lungs. Eliminating risk factors is crucial to prevent recurrences. In spite of the progresses on the understanding of OSCC biological and molecular characteristics, and of the different therapeutic approaches (surgery, radiotherapy, and/or chemotherapy), OSCC patients have poor overall survival rates (~50%) due to frequent recurrences and second primary tumors. Indeed, survival rates have not improved significantly over the past three decades (7). Therefore, early
Prognostic value for FD in primary OSCC Mincione et al.
diagnosis, before spreading of the tumor to other tissues, is an important issue, because prevention may greatly improve the cure rate, up to 90%. Early diagnosis requires a careful monitoring of oral mucosa pre-cancerous lesions, such as leukoplakia and erythroplakia, to ensure the secondary prevention of neoplastic forms. However, to date, in spite of the early clinical monitoring of this malignancy, mortality, and incidence appear to be still unchanged, suggesting the need to identify novel prognostic factors. Recently, fractal geometry has been used to go over the limits of the classical histomorphometrical approaches. Fractal is a geometric shape, characterized by being scalefree, that is, it appears similar whatever the scale used to look at them and is described by mathematic formulas. With fractal geometry, it is possible to quantify the morphological features routinely used by pathologists to qualitatively characterize and describe malignancies, that is, wrinkled borders. An ideal tool to describe a fractal is the fractal dimension (FD). FD is a shape index, and it can be defined as a measure of irregularly shaped objects. The lower the value of FD, the more regular or chaotic is the shape of the objects (8, 9). Therefore, it can be considered a powerful parameter to assess and represent shape variability of particular structures/cells/nuclei in neoplastic tissues as compared to normal counterparts. Indeed, several studies used this approach to describe different tumors, that is, breast carcinoma (10), lung (11), and the oral cavity (12), and its association with known clinicopathological parameters and overall survival. The aim of this study was to investigate FD as a novel prognostic tool for OSCC by correlating FD values with clinicopathological features and survival of patients affected by OSCC.
Material and methods Study design and pathologic examination Sixty-four primary oral squamous cell carcinomas (OSCC) and 10 histologically normal oral samples were obtained from the Department of Medical, Oral and Biotechnological Sciences, University ‘G. d’Annunzio’ Chieti-Pescara, Chieti, Italy, and the Department of Pathology, Polytechnic University of the Marche, Ancona, Italy. The experiments were undertaken with the understanding and written consent of each subject to use the tissues for research purposes and according to ethical principles, including the World Medical Association Declaration of Helsinki. All patients enrolled in this study had a OSCC clinical history and available histology paraffin material. Of the 64 patients with OSCC, 49 were males and 15 females, while in the control group, three were male and seven females. The mean age was 63.6 12.8 years and 50.0 21.9 years for OSCC and control groups, respectively. There was no statistical difference when comparing sex and the mean age in the two groups. The tumor topography, grading, and staging of the samples in the study were shown in Table 1. For all cases, the available hematoxylin–eosin sections were reviewed by two experienced pathologists (LA and CR) in order to confirm the
Table 1 Clinicopathological squamous carcinoma
characteristics
of
patients
with
oral
681
Variable Gender, n (%) Male Female Age (year), mean SD Topography, n (%) Tongue Mucosa geniena Other localizations Stage, n (%) I–II III–IV Grading, n (%) 1 2 3
49 (76.6) 15 (23.4) 63.6 12.8 10 (15.6) 49 (76.6) 5 (7.8) 32 (50.0) 32 (50.0) 24 (37.5) 30 (46.9) 10 (15.6)
original diagnosis and to select the representative neoplastic areas. Two 3-microns sections were prepared from each specimen and were mounted on super frost plus-coated glass slides (Menzel-Gl€aser, Braunschweig, Germany). One of them, stained with hematoxylin–eosin was used for pathological reevaluation, and the other one was stained with Feulgen stain for evaluation of the nuclei. The histological status of the lesions was defined after examination of the complete section under light microscopy, and the neoplasms were staged as: G1 (well differentiated), G2 (moderately differentiated), G3 (poorly differentiated) OSCC. Histological material of 10 patients randomly selected and treated for non-neoplastic lesions was studied as control. Nuclear fractal dimension For standard nuclear histomorphometry, three representative high-power fields (409) of Feulgen stains were randomly selected and photographed from each biopsy. The images were captured with a digital CCD-videocamera and were stored as high quality TIFF files, 1600 9 1200 pixels and 16.7 million colors (24-bit). The microscope and camera settings were kept consistent during the whole procedure. The above-mentioned image files were automatically transformed to binary images (image binarization converts an image to a black and white image suitable for fractal analysis) using ImageJ 1.40 g (Wayne Rasband; National Institute of Health, Bethesda, MD, USA). After selecting a 4.3 9 4.3 cm area on each image, the fractal dimension (FD) was estimated by using the box-counting method by means of ImageJ 1.40 g (Wayne Rasband; National Institute of Health). Statistical analysis Continuous parameters were reported as either mean and standard deviation (SD) or median and interquartile range. Categorical parameters were reported as frequency and percentage. The statistical analysis was performed using nonparametric tests. The Mann–Whitney U-test was used to evaluate statistical significant differences in FD values between groups (cases and control). J Oral Pathol Med
Prognostic value for FD in primary OSCC Mincione et al.
682
Table 2 Fractal dimension values respect to different clinicopathological characteristics of patients with oral squamous carcinoma Variable Gender Male Female Topography Tongue Mucosa geniena Other localizations Stage I–II III–IV Grade 1 2 3
Mean SD
Min–Max
P-value
1.41 0.13 1.38 0.08
1.16–1.67 1.26–1.57
0.568a
1.43 0.12 1.39 0.12 1.49 0.13
1.23–1.67 1.16–1.65 1.30–1.63
0.130b
1.35 0.21 1.44 0.11
1.16–1.67 1.23–1.65
0.003
1.30 0.08 1.41 0.06 1.61 0.04
1.16–1.41 1.27–1.53 1.55–1.67
0.001b
a
Mann–Whitney U-test. Kruskal–Wallis H-test. Statistically significant P-value are shown in bold. b
Cox proportional hazard method was applied to estimate cancer specific survival rate (CSS) and to determine the impact of FD values as independent prognostic factor for OSCC. All statistical analysis was performed using SPSSâ Advanced StatisticalTM 13 (2004, Chicago, IL, USA) software package.
Results Figure 1 Representative images of binarized and feulgen-stained samples from normal oral mucosa (control) and primary oral squamous cell carcinomas with different grade of differentiation (G1, G2, and G3).
The images showed cell nuclei from normal oral epithelium (controls), from well-differentiated OSCC (G1), moderately differentiated OSCC (G2), and poorly differentiated OSCC (G3) (Feulgen, 409). To the side of each image, there was a selected field of the same image after binarization (Fig. 1).
The Kruskal–Wallis H-test or Mann–Whitney U-test, when appropriate, was used to evaluate differences between FD values after stratification of the cases regarding their clinicopathologic characteristics. Post hoc analysis was performed with Kruskal–Wallis multiple comparison test.
Fractal Dimension value in OSCC groups and control The median FD was 1.15 (1.14–1.17) and 1.39 (1.35–1.43) for control and test groups, respectively, with statistically significant differences in FD value between two groups (P < 0.001) (Fig. 2).
Figure 2 Box-whiskers graphs of fractal dimension in control and test groups. Box-whiskers plots show the 25th and 75th percentile range (box) with 95% confidence intervals (whiskers) and median values (transverse lines in the box). A statistical significant differences was found between the two groups (P < 0.001, Mann–Whitney U-test).
Figure 3 Box-whiskers graphs of fractal dimension for stage. Boxwhiskers plots show the 25th and 75th percentile range (box) with 95% confidence intervals (whiskers) and median values (transverse lines in the box). A statistical significant difference was found between groups (P < 0.001, Kruskal–Wallis H-test).
J Oral Pathol Med
Prognostic value for FD in primary OSCC Mincione et al.
value was found between the different degrees of tumor differentiation groups (P = 0.001). Moreover, a statistically significant difference comparing OSCC with control groups (P < 0.001) was found (Fig. 4).
Figure 4 Box-whiskers graphs of fractal dimension for grade. Boxwhiskers plots show the 25th and 75th percentile range (box) with 95% confidence intervals (whiskers) and median values (transverse lines in the box). A statistical significant differences was found between the three groups (P < 0.001, Kruskal–Wallis H-test).
Relationship between Fractal Dimension and stage The clinical tumor stage at diagnosis has an important impact on treatment decisions and consequently on health outcomes. Thus, we evaluated the relationship between FD and stage. The basic statistical parameters of FD for clinicopathological characteristics of patients were shown in Table 2. The results obtained demonstrated a progressive increase of the FD value from stages I and II to stages III and IV, and the differences were statistically significant (P = 0.003). Moreover, a statistically significant difference when comparing OSCC and control groups (P < 0.001) was found. The dot in the graph indicates a outlier value (Fig. 3). Relationship between Fractal Dimension and tumor grading The assessment of relationship between FD and tumor grading was also examined to gain better insight into its prognostic significance in OSCC. The mean FD may differ according to histologic grades. FD values were proved to be low in G1 group (well differentiated) and high in G3 group (less differentiated tumors). A significant difference in the average nuclear FD
Figure 5 Overall survival curves stratified for median of FD values. There is statistically significant difference in outcome between cases with FD value
