Introduction
Worldwide, oral cancer is the sixth most commonly diagnosed cancer, accounting for 2.1 per cent of total cancer cases.Reference Jemal, Bray, Center, Ferlay, Ward and Forman1 Higher prevalence is observed in Asia (especially Pakistan, India, Bangladesh, Sri Lanka and Nepal) than in Europe or USA.Reference Aamir, Mirza, Mirza and Qureshi2, Reference Ferlay, Soerjomataram, Dikshit, Eser, Mathers and Rebelo3 In India, oral cancer is the most common form of cancer among males, while in Pakistan it is the second most common cancer.Reference Joshi, Dutta, Chaturvedi and Nair4, Reference Parkin, Arslan, Bieber, Bouvy, Muir and Owor5 This is attributed to the increased risk for oral cancer development in these regions associated with smoking, and the high consumption of alcohol and smokeless tobacco, paan, gutka, areca nut and betel quid.Reference Bhurgri, Bhurgri, Hussainy, Usman, Faridi and Malik6
Generally, 90 per cent of all oral malignancies exist as oral squamous cell carcinoma (SCC), which globally accounts for an estimated 275 000 new cases per year.Reference Alamgir, Jamal and Mirza7, Reference Farrukh, Syed and Pervez8 Despite the relative ease of diagnosis, oral cancer is associated with a low survival rate, of 40–50 per cent.Reference Martinez-Useros and Garcia-Foncillas9 This is because patients often present with advanced disease. In addition, access to medical care is limited in many parts of the world and treatment strategies are often inadequate, resulting in severe cosmetic problems that reduce quality of life. In Pakistan, oral cancer ranks as the most common malignancy in males and second most common malignancy in females,Reference Bhurgri, Bhurgri, Hussainy, Usman, Faridi and Malik6, Reference Bhurgri10 which is ascribed to the widespread use of cultural forms of smokeless tobacco, areca nut and betel quid.
Although the precise cause of oral cancer is unknown, it is evident that some substantial changes occur at the molecular level before the development of clinically significant malignant lesions. One such change is thought to occur in the expression of the cyclooxygenase-2 (‘COX-2’) enzyme.Reference Amirchaghmaghi, Mohtasham and Mozaffari11 Cyclooxygenase-2 (also known as prostaglandin-endoperoxide synthase 2) is a rate-limiting enzyme involved in arachidonic acid metabolism, which is considered critical to various biological processes, including angiogenesis, inflammation, thrombosis, ovulation and various other pathways of immunological importance. Though in the normal cellular environment cyclooxygenase-2 is expressed in negligible amounts, its expression increases in response to inflammation, growth factors and tumour progression.Reference Wang, Liu, Liu, Ye, Zhang and Yang12, Reference Williams, Mann and DuBois13
Cyclooxygenase-2 has been linked to malignant processes such as increased anogiogenesis,Reference Dannenberg, Altorki, Boyle, Dang, Howe and Weksler14 apoptosis preventionReference Wang, Liu, Liu, Ye, Zhang and Yang12 and metastasis.Reference Dannenberg, Altorki, Boyle, Dang, Howe and Weksler14 Cyclooxygenase-2 overexpression has been observed in oesophageal, colorectal, breast, urinary bladder, lung, stomach and pancreatic cancers.Reference Bartchewsky, Martini, Masiero, Squassoni, Alvarez and Ladeira15–Reference Wang, Colby, Rengel, Fischer, Clinton and Klein21 A study assessing the expression of cyclooxygenase-2 by immunohistochemistry revealed a higher expression in oral SCC as compared to leukoplakia and normal tissues.Reference Amirchaghmaghi, Mohtasham and Mozaffari11 Pandey et al. reported similar findings using reverse transcription polymerase chain reaction, with higher cyclooxygenase-2 expression in malignant tissues as compared to normal tissue or pre-malignant lesions.Reference Pandey, Prakash, Santhi, Soumithran and Pillai22 Increased cyclooxygenase-2 expression has also been related to lymph node metastasis, histological grade and tumour–node–metastasis (TNM) stage in oral SCC.Reference Cao, Zhang, Wu, Liu and Zhang23 Selective inhibition of cyclooxygenase-2 has been achieved by designing cyclooxygenase-2 inhibitors, which resulted in cell proliferation suppression by inhibiting the G0/G1 cell cycle phase, leading to cell cycle arrest and apoptosis.Reference Williams, Mann and DuBois13
Although the role of cyclooxygenase-2 has been explored in relation to oral SCC, information regarding the Pakistani population is lacking. As Pakistan is a high-risk population for oral cancer, we aimed to: investigate the overexpression of cyclooxygenase-2 in oral SCC patients, correlate it with clinicopathological features, and study its effect on overall and disease-free survival.
Materials and methods
Patient selection
A total of 100 patients who had been diagnosed and treated for oral SCC at Aga Khan University Hospital during the years 1991–2004 were recruited in this study. Inclusion criteria were: complete medical history, clinicopathological data, habit history (i.e. data on the use of high-risk substances such as tobacco, betel quid and areca nut), adequate specimen and follow-up details. Exclusion criteria were: incomplete medical records, lack of follow up (with minimum follow-up time set at 60 months) and those who did not undergo surgery at the Aga Khan University Hospital. Ethical approval was obtained from the Ethical Review Committee of Aga Khan University Hospital. Participating patients were informed about the purpose of the study and were asked to provide written informed consent.
Immunohistochemistry
Four-μm thick sections were sliced from formalin-fixed, paraffin-embedded blocks of surgical specimen onto pre-coated slides. Dewaxing occurred at 60 °C for 30–40 minutes, followed by deparaffinisation in xylene and rehydration in a graded water-ethanol series. Antigen retrieval was achieved using Dako Antigen Retrieval Solution diluted at 1.5 ml per 50 ml, at 100 °C for 40 minutes. After cooling at room temperature for 20 minutes, the slides were incubated in buffer solution for 10 minutes. Endogenous peroxidase activity was blocked using peroxidase blocking reagent for 10 minutes. This was followed by rinsing and incubation with primary mouse monoclonal anti-human cyclooxygenase-2 antibody (Clone CX-294; Dako, Glostrup, Denmark) diluted at 1:100 for 30 minutes. Subsequently, sections were: washed and treated with secondary antibody horseradish peroxidase/Flex (Dako) for 30 minutes, rinsed and treated with EnVision+ system of horseradish peroxidase with 3,3'-diaminobenzidine (DAB) (Dako) for visualisation. Haematoxylin was applied for 30 seconds to provide a counterstain. Lastly, the slides were: dehydrated in a water-ethanol series, mounted using mounting medium (Dako); and viewed at 20× and 40× magnification. Experimental controls were run alongside each batch. Brain tumour samples were used as the positive control, and incubation with buffer instead of primary antibody served as the negative control.
Slide evaluation
Independent evaluation was conducted by two pathologists who were unaware of the clinical history and blinded to each other's readings. Any discrepancies arising between them were resolved using a conference microscope.
The following criteria were used for scoring cyclooxygenase-2 positivity, according to the percentage of cells stained: less than 10 per cent = negative, 10 to 20 per cent = mild, more than 20 to 40 per cent = moderate, and more than 40 per cent = strong positivity. As existing literature reports various scoring criteria for cyclooxygenase-2 positivity, with little uniformity among them, we used our own criteria. These criteria were adapted from a study with a similar sample size conducted on the Indian population, which can be considered as genetically close.Reference Mohammad, Ram, Gupta, Husain and Bhatt24
Statistical analysis
Analysis was performed using SPSS® version 22 statistical software. Descriptive analysis was performed, reporting the mean ± standard deviation (SD) or median (interquartile range) for continuous variables, depending on the normality assumption of the variables; the two groups (i.e. cyclooxygenase protein overexpression and no overexpression) were evaluated using the independent t-test or Wilcoxon rank sum (Mann–Whitney test) respectively. For categorical variables, frequency and percentage were reported; the two groups were assessed using the chi-square test of independence if the frequency in each cell was 5 or more, or using the Fisher's exact test if the frequency was less than 5 in any cell.
Follow-up time for each patient was calculated in months. Disease-free survival time was measured from the date of surgery to the date of recurrence (treatment failure), or to the date of last follow up if no recurrence. Overall survival time was measured from the date of surgery to the date of death (treatment failure), or to the date of last follow up if the patient was alive. Patients who were alive at the time of last contact were considered censored observations in overall survival analysis.
Kaplan–Meier survival curves, log-rank tests and Cox proportional hazards regression analysis were used to compare disease-free survival (recurrence) and overall survival (death). A second proportional hazard regression model, adjusted for the other study covariates, was used to examine the independent effect of treatment. Adjusted hazard ratios with 95 per cent confidence intervals were reported. A p-value of less than 0.05 was considered significant.
Results
Patient demographics
The selected 100 patients had all been diagnosed with oral SCC. They had a mean (± SD) age of 52.73 ± 12.68 years, with ages ranging from 20 to 78 years. The majority of patients were male (57 per cent) and aged over 40 years (85 per cent). The most commonly reported primary lesion site was the cheek (64 per cent), followed by the tongue (36 per cent).
The consumption of substances known to be high-risk factors for oral SCC was recorded for our population. The majority of patients (77 per cent) had a positive habit history. Habits included the use of tobacco (smoking and smokeless), areca nut and betel quid in the culturally prevalent forms of paan, supari, chalia, gutka and naswar. Habit pattern assessment showed that 32 per cent of patients were habitual cigarette smokers, and 59 per cent and 28 per cent were betel quid and areca nut chewers respectively. The habit history of 23 patients was either insufficient or not reported. The habitual use of any high-risk substance was not determined to have any effect on cyclooxygenase-2 expression.
Histologically, the most common diagnosis was moderately differentiated tumour (56 per cent) and American Joint Committee on Cancer stage II (31 per cent). According to the TNM staging, stages T2 and N0 were most often diagnosed; none of the patients had distant metastasis.
Upon last follow up, 57 per cent of patients were reported alive and 68 per cent had experienced disease recurrence.
Cyclooxygenase-2 and clinicopathological factors
Cyclooxygenase-2 overexpression was positive (as illustrated in Figure 1a) in 55 per cent of patients and negative (Figure 1b) in the remaining 45 per cent. Based on staining intensity, 20 per cent of positive specimens were mild, 29 per cent were moderate and 6 per cent were strong.
Fig. 1. Photomicrographs of: (a) cyclooxygenase-2 positive oral squamous cell carcinoma (SCC) tissue showing strong brown cytoplasmic staining, and (b) cyclooxygenase-2 negative oral SCC tissue showing haematoxylin counterstain (magnification × 20).
The associations of cyclooxygenase-2 overexpression and patients’ clinicopathological factors are shown in Table 1. There was no significant relationship between cyclooxygenase-2 positivity and any of the clinicopathological factors tested by chi-square, except for overall survival and disease-free survival. Regarding the latter variables, cyclooxygenase-2 positivity was a significant indicator of lower overall survival (p = 0.01) and a higher chance of disease recurrence (p = 0.004).
Table 1. Clinicopathological characteristics*
*Total n = 100. †Age presented as mean ± standard deviation, in years. ‡n = 55; **n = 45. §P-value calculated using chi-square or Fisher's exact test (as applicable). #P < 0.05, indicating significance. COX-2 = cyclooxygenase-2; NA = data not available; AJCC = American Joint Commission on Cancer
Overall and disease-free survival
Table 2 shows the univariate analysis findings for overall survival and disease-free survival. For our study population, mean overall survival time was 146.88 months and disease-free survival time was 78.86 months. Cyclooxygenase-2 positivity was a statistically significant indicator both for overall survival (p = 0.013) and disease-free survival (p = 0.001). Other factors that had a notable impact on overall survival were American Joint Committee on Cancer stage (p = 0.001), surgical margins (p = 0.045), T stage (p = 0.033) and N stage (p < 0.001). Similarly, disease-free survival was also associated with American Joint Committee on Cancer stage (p = 0.005) and N stage (p < 0.001).
Table 2. Univariate analysis for overall and disease-free survival
*P-value calculated using log-rank analysis. †P < 0.05, indicating significance. AJCC = American Joint Commission on Cancer; COX-2 = cyclooxygenase-2
The Kaplan–Meier analysis also supported these findings, as shown in Figure 2. The mean overall survival time for cyclooxygenase-2 positive patients was 120.76 months, as compared to the much longer survival time of 174.53 months for cyclooxygenase-2 negative patients. This was reflected in the mean disease-free survival times, which were 48.02 months and 115.46 months for cyclooxygenase-2 positive and negative patients respectively.
Fig. 2. (a) Overall survival and (b) disease-free survival curves. Cyclooxygenase-2 (COX-2) positive patients had significantly lower overall and disease-free survival rates.
On multivariate analysis, only cyclooxygenase-2 was a significant independent predictor of prognosis (Table 3). Disease-free survival was significantly affected if the patient was cyclooxygenase-2 positive (p = 0.044); however, overall survival was not a significant independent factor on multivariate analysis (p = 0.208).
Table 3. Multivariate analysis for overall and disease-free survival
*P-value calculated using log-rank analysis. †P < 0.05, indicating significance. CI = confidence interval; COX-2 = cyclooxygenase-2
Discussion
Oral cancer is attributed to a multifactorial aetiology involving genetic and environmental factors. Various genetic alterations have been associated with oral SCC. These include alterations in chromosomes 3, 9, 11 and 13, the overexpression of various oncogenes, as well as the inhibition of tumour suppressor genes and other apoptotic factors and DNA repair mechanisms, which together lead to carcinogenesis.Reference Scully, Field and Tanzawa25 Various biomarkers have been correlated with oral SCC incidence and disease progression, suggesting their critical role in tumourigenesis, with one such marker being cyclooxygenase-2. We investigated whether cyclooxygenase-2 was overexpressed in oral SCC patients of Pakistan, and examined the effect this had on patient prognosis. The Reporting Recommendations for Tumour Marker Prognostic Studies (‘REMARK’) guidelines were used as a standard of reporting for this study.Reference Altman, McShane, Sauerbrei and Taube26
Previous studies have suggested that cyclooxygenase-2 is usually up-regulated in pathological conditions such as oral SCC, oesophageal, lung, breast, colorectal and other forms of cancer.Reference Wang, Liu, Liu, Ye, Zhang and Yang12, Reference Bartchewsky, Martini, Masiero, Squassoni, Alvarez and Ladeira15–Reference Ratnasinghe, Tangrea, Roth, Dawsey, Hu and Anver19 Cyclooxygenase-2 overexpression has also been reported in oral pre-malignant lesions. Shibata et al. investigated the expression of cyclooxygenase-1 and cyclooxygenase-2 in patients with dysplasia and SCC, and revealed that cyclooxygenase-2 expression was strongly associated with higher stages of dysplasia rather than SCC.Reference Shibata, Kodani, Osaki, Araki, Adachi and Ryoke27 Another study also demonstrated that cyclooxygenase-2 overexpression influenced progression towards malignancy.Reference Nankivell, Williams, McConkey, Webster, High and MacLennan28
In this study, immunohistochemical analysis revealed cyclooxygenase-2 overexpression in the majority of oral SCC patients (55 per cent), suggesting that this may be common in oral SCC carcinogenesis. The overall demographics of our patients correlate with previous studies on the Pakistani oral SCC population, with buccal mucosa being the most common lesion site and moderately differentiated American Joint Committee on Cancer stage II carcinoma being the most common diagnosis.Reference Alamgir, Jamal and Mirza7, Reference Alamgir, Jamal, Jafarey and Mirza29 There was a higher number of females in our study as compared to previous reports, and the mean age of our population was lower than the reported average of 64 years.Reference Haffner, Laimer, Chaux, Schäfer, Obrist and Brunner30 However, there was no significance of age or gender in terms of cyclooxygenase-2 expression in our population.
We found a slightly higher percentage of cyclooxygenase-2 positive patients (55 per cent), as compared to previous studies, in which cyclooxygenase-2 was found overexpressed in 28 per cent, 42 per cent and 14 per cent of oral SCC specimens using immunohistochemistry.Reference Haffner, Laimer, Chaux, Schäfer, Obrist and Brunner30–Reference Itoh, Matsui, Furuta and Takano32 This relatively high rate of cyclooxygenase-2 in our population may be explained by the heavy consumption of high-risk substances. Continuous abrasion of buccal mucosa associated with chewing tobacco, chalia, gutka and naswar causes inflammation; as cyclooxygenase-2 is one of the key mediators of the process, its overexpression can be attributed to persistent consumption of these factors.
The majority of patients were users of substances that are known oral SCC risk factors, but their habits were not correlated with cyclooxygenase-2 overexpression. There were an almost equal number of cyclooxygenase-2 positive and negative patients when distributed according to their particular habits. Contrary to our results, Moraitis et al. reported higher overexpression of cyclooxygenase-2 in smokers, and suggested that this expression was regulated through the production of prostaglandins E2 and tyrosine kinase.Reference Moraitis, Du, De Lorenzo, Boyle, Weksler and Cohen33 In another study, smoking was associated with the induction of cyclooxygenase-2, which led to the increased production of prostaglandin E2 and thromboxane A2; increased levels of these inflammation mediators can lead to tumour development and progression.Reference Huang and Chen34
Correlating cyclooxygenase-2 expression with any clinicopathological factor did not yield significant results, except for overall survival (p = 0.01) and disease-free survival (p = 0.004). The impact of cyclooxygenase-2 on survival in combination with other factors was observed on the univariate analysis, as cyclooxygenase-2 positive patients had markedly lower overall survival (p = 0.013) and disease-free survival (p = 0.001). However, the multivariate analysis showed that cyclooxygenase-2 positivity had an independent effect on disease-free survival only, while overall survival was affected by a combination of factors, one of which was cyclooxygenase-2 positivity. Moreover, the Kaplan–Meier analysis showed that cyclooxygenase-2 positive patients had lower overall survival and higher recurrence rates than their cyclooxygenase-2 negative counterparts.
Similar to our results, Cha et al. also evaluated oral SCC by immunohistochemistry to assess cyclooxygenase-2 overexpression, and found that cyclooxygenase-2 expression was statistically significant for overall survival on univariate analysis, but this was not replicated on multivariate analysis.Reference Cha, Li and Cha31 A study conducted by Haffner et al., however, revealed that cyclooxygenase-2 overexpression was independently and significantly associated with poor prognosis.Reference Haffner, Laimer, Chaux, Schäfer, Obrist and Brunner30 Other studies have linked cyclooxygenase-2 independently with poor overall survival in pancreatic cancerReference Juuti, Louhimo, Nordling, Ristimäki and Haglund35 and in combination with other variables in ovarian cancer.Reference Athanassiadou, Grapsa, Athanassiades, Gonidi, Athanassiadou and Tsipis36
There has been debate regarding the association between the degree of tumour differentiation and cyclooxygenase-2 expression. An increased expression of cyclooxygenase-2 has been observed in well differentiated tumours, as compared to poorly differentiated tumours, in lung, laryngeal and oesophageal cancer.Reference Ratnasinghe, Tangrea, Roth, Dawsey, Hu and Anver19, Reference Hida, Yatabe, Achiwa, Muramatsu, Kozaki and Nakamura37, Reference Ranelletti, Almadori, Rocca, Ferrandina, Ciabattoni and Habib38 In contrast, other studies have reported an increased expression of cyclooxygenase-2 in poorly differentiated carcinomas of the oesophagus and tongue, rather than in well differentiated tumours.Reference Renkonen, Wolff and Paavonen39, Reference Nozoe, Ezaki, Kabashima, Baba and Maehara40 In our study, the overexpression of cyclooxygenase-2 was not associated with histological classification, American Joint Committee on Cancer stage, gender or TNM staging, which is similar to the findings of Pandey et al.Reference Pandey, Prakash, Santhi, Soumithran and Pillai22 The lack of any statistically significant correlations may be a result of missing patient information or the small sample size, which may not be a true representation of the population.
Having ascertained the role of cyclooxygenase-2 in tumour development and progression, inhibitors of cyclooxygenase-2 have been found to play a critical role in cancer treatment. The inhibition of cyclooxygenase-2 decreases angiogenesis, cell proliferation and invasiveness, G0/G1 cell cycle arrest, p21 expression, and prostaglandin E2 production.Reference Haffner, Laimer, Chaux, Schäfer, Obrist and Brunner30, Reference Toyoshima, Kamijo, Takizawa, Sumitani, Ito and Nagumo41, Reference Zhang, Bian, Li, Wu, Bi and Yan42 Kim et al. demonstrated the anticancer effects of celecoxib, which significantly reduced the invasive potential of cancer cell lines by inhibiting cyclooxygenase-2.Reference Kim, Lee, Kim, Kim, Park and Myoung17 In addition, various anti-apoptotic proteins have radioresistant properties, and the regulation of such factors is controlled by the nuclear factor kappa Β pathway, which is associated with the increased production of cyclooxygenase-2, hence its playing a major role in radioresistance.Reference Chandrasekharan and Simmons43, Reference Johnson, Ivanov and Hei44 The present evidence suggests a role of cyclooxygenase-2 as a prognostic marker and in targeted therapy of oral SCC patients.
• Oral squamous cell carcinoma (SCC) is one of the most commonly occurring cancers worldwide
• Cyclooxygenase-2 has been tested as a prognostic marker in various cancers
• Immunohistochemistry was used to detect cyclooxygenase-2 expression in 100 oral SCC cases in Pakistan
• Increased cyclooxygenase-2 expression is found in oral SCC patients, leading to poor overall and disease-free survival
• Inhibiting cyclooxygenase-2 and its related pathways can help prevent cell proliferation and invasive potential of tumour-promoting factors
Acknowledgements
This work was supported by a grant from the Higher Education Commission, Pakistan, awarded to Dr Syed Muhammad Adnan Ali (grant identification code: PM IPFP/HRD/HEC/2012/4015).
Competing interests
None declared
