Significant association of high-risk human papillomavirus (HPV) but not of p53 polymorphisms with oral squamous cell carcinomas in Malaysia

Abstract

Purpose

The purpose of this study was to evaluate the role of HPV and p53 polymorphisms in oral squamous cell carcinomas (OSCC) affecting Malaysian population.

Methods

We analysed frozen samples from 105 OSCC as well as 105 oral specimens derived from healthy individuals. PCR assays targeting two regions of the virus were used. PCR amplification for the analysis of p53 codon 72 arginine/proline alleles was carried out in a separate reaction.

Results

HPV DNA was detected in 51.4% OSCC samples, while 24.8% controls were found to be HPV positive. HPV was found to be significantly associated with OSCC (P < 0.001, OR = 4.3 after adjustment for habits) when compared to controls. High-risk HPV was found to be significantly associated with OSCC cases (P < 0.05). Demographic profiles of age, gender, race and habits were not associated with HPV presence in cases and controls. However, significantly less HPV positivity was seen in poorly differentiated compared to well-differentiated OSCCs. No significant association was found between HPV positivity and p53 polymorphisms in cases and control groups. Additionally, we found no association of codon 72 polymorphism with oral cancer.

Conclusions

This study indicates that high-risk HPV infection is one of the contributing factors for OSCCs. HPV 16 was the predominant type found in Malaysian patients with OSCC. Further, we did not find any association between p53 codon 72 polymorphism and HPV infection or between the p53 polymorphism and the risk of oral cancer.

Introduction

High incidence of oral cancer in many developing countries is attributable mainly to the habits of tobacco and betel quid chewing and alcohol consumption (Zain et al. 1999). However, oral cancer may also develop without exposure to these risk factors, suggesting that additional causes, such as genetic predisposition, diet and viruses, may play a role in oral tumorigenesis (Enwonwu and Meeks 1995; Scully and Bedi 2000). An etiologic role of HPV infection in the pathogenesis of oral premalignant and malignant lesions has been indicated by the discovery of HPV in oral cancer specimens (D’Costa et al. 1998; Elamin et al. 1998; Fouret et al. 1995). However, studies on their association in the initiation and progression of oral neoplasias have produced conflicting results (Ha and Califano 2004; Miller and Johnstone 2001).

HPV belongs to a large family of viruses, the papova viridae, and so far nearly 100 different types have been described in humans (de Villiers et al. 2004). Some of the HPVs such as 16, 18, 33 and 58 are known to be oncogenic (high-risk) as they play a role in the progression of certain human cancers particularly cervical and anogenital carcinomas (zur Hausen 1996).

The human p53 gene contains a nucleotide polymorphism for amino acid codon 72, exon 4 of the p53 protein that results in either a proline (CCC, p53Pro) or an arginine (CGC, p53Arg) allele (Buchman et al. 1988). Thus, in the general population, each individual inherits a p53 genotype that is homozygous for either arginine (Arg/Arg) or proline (Pro/Pro) or heterozygous (Arg/Pro). This polymorphism was first suggested to play a role in the development of HPV-associated cancer by Storey et al. (1998). They showed that high-risk E6 protein has a selective affinity for the p53 arginine variant, and Arg/Arg individuals were seven times more susceptible to developing HPV-associated cervical cancer than heterozygotes, suggesting that arginine homozygosity represents an important risk factor for development of cervical cancer. Several studies were performed on cervical cancers to confirm an association between this polymorphism and cervical neoplasia, but there were more contradictory findings (Giannoudis et al. 1999; Minaguchi et al. 1998; Rosenthal et al. 1998) than similar results (Agorastos et al. 2004; Zehbe et al. 2001).

Recent studies done on other cancers such as nasopharyngeal (Tsai et al. 2002), stomach (Hiyama et al. 2002) and lung carcinomas (Jin et al. 1995) have shown a higher risk for tumours associated with Pro/Pro homozygotic status and a protective effect of the Arg allele. Few studies done on oral squamous cell carcinomas in different populations found no association of Arg/Pro polymorphism with HPV-related tumorogenesis (Katiyar et al. 2003; Summersgill et al. 2000). Thus, there are conflicting results from the studies of the p53 polymorphism in different types of cancers.

Till date, no studies have been published about HPV infection in normal mucosa or malignant lesions of the oral cavity in the Malaysian population. This study was undertaken to determine the prevalence of HPV in OSCC and oral epithelium of normal individuals and to investigate any association of HPV with demographic variables and cancer status. Further, associations of p53 polymorphisms with risk of oral cancer and HPV infection were also investigated.

Materials and methods

This was a retrospective case–control study that included 105 patients with OSCC and 105 normal individuals who served as controls for determination of association of HPV infection with the oral cancer and also for the evaluation of the role of p53 codon 72 polymorphism.

Oral squamous cell carcinoma specimens

In this study, we used genomic DNA from fresh frozen oral squamous cell carcinoma (OSCC) specimens obtained from tissue banks located at the Oral Cancer Research Collaboration Center (OCRCC) in University of Malaya and the Cancer Research Initiatives Foundation (CARIF). Tissues from the centre of the surgical resection from the OSCCs of only the primary origin were sent to OCRCC or CARIF from various collaborating institutes. Clinicians, scientists and epidemiologists from Ministry of Health Hospitals (six hospitals from Peninsula Malaysia and two from East Malaysia states on the island of Borneo) and three hospitals attached to Institutes of Higher Learning (University of Malaya, Universiti Sains Malaysia and Universiti Kebangsaan Malaysia) acted as data and specimen providers. Informed consent from OSCC patients were obtained with ethical approval from the University of Malaya, Universiti Sains Malaysia, Universiti Kebangsaan Malaysia and endorsed by the Ministry of Health Malaysia.

Buccal swab specimens

The subjects to be used as normal controls were normal individuals in Kota Bharu, Kelantan who registered at the Outpatient Clinic, School of Dental Sciences, Universiti Sains Malaysia and met all the inclusion criteria such as age of at least 18 and in good general health. The written informed consent and ethical approval was obtained from Universiti Sains Malaysia. Relevant data such as age, sex, smoking and alcohol use were collected. Buccal swabs were taken by gently scraping right and left sides of oral buccal mucosa for 20 s in 20 circular motions (10 each side) with a sterile EUROTUBO^® collection swab (Deltalab, Spain) in order to get adequate amount of mucosal cells. The swabs were immediately stored in −20°C freezer until further analysis.

Extraction of DNA from specimens

Sectioning of fresh frozen specimens was done using a Cryostat (Leica CM 1850) preset at −20°C. Reference slides were made and stained with haematoxylin & eosin to confirm the presence of adequate tumour cells before sectioning for DNA extraction was performed. Samples with more than 70% tumour cells in the reference slides were selected for DNA extraction. QIAamp^® DNA Extraction Mini Kit (Qiagen, Germany) was used for extraction of DNA from cancer samples as well as buccal swabs from normal controls, and the procedure followed was according to the manufacturer’s protocol. To ensure DNA quality, a 110-base pair fragment of β-globin gene was detected by PCR amplification (data not shown) in all samples according to the method of Saiki et al. (Saiki et al. 1992).

HPV detection and type identification

The presence of HPV DNA in the samples was detected by PCR assay using degenerate consensus primers from two regions of the virus. The β-globin-positive samples were amplified at the L1 genomic region using the universal primers MY09/MY11 and GP5+/GP6+ in a nested PCR using conditions as per the published literature (Ammatuna et al. 2000; Saini et al. 2007). Two separate PCRs were performed. The first PCR performed using the MY09/MY11 outer primers (Manos et al. 1989), producing an amplicon of 450 bp. This was followed by GP5+/GP6+ PCR (de Roda Husman et al. 1995) that produced an amplicon of 150 bp. Samples were further analysed with another set of degenerate primers from E1 region CPI/CPIIG (Tieben et al. 1993). The CPI/CPIIG PCR used an annealing temperature of 51.1°C for 40 cycles.

PCR products for samples positive for HPV were purified by using Wizard^® SV Gel and PCR Clean-Up System (Promega, USA). PCR products were subsequently sequenced using Big-Dye Terminator sequencing kit (Applied Biosystems) by using one of the PCR primers as a sequencing primer. Sequencing reactions were analysed on the ABI 310 genetic analyser (Applied Biosystems). Obtained sequences were compared with documented virus sequences that were available in GenBank by using the BLAST program server to determine the HPV type in positive samples.

PCR amplification of p53:72Pro/Arg polymorphism

The PCR amplification for the analysis of p53:72Proline/Arginine alleles was carried out in separate reactions using the primers as described by Storey et al. (1998). Initial denaturation at 95°C was done for 5 min followed by 35 cycles of 95°C for 30 s, 61°C for 20 s and 72°C for 20 s, and a final extended synthesis at 72°C was carried out for 4 min. The amplified products of 141 bp for p53Arg and 177 bp for p53Pro were visualized on an ethidium bromide-stained 2% agarose gel (Promega, USA) under UV transilluminator.

Statistical analysis

SPSS 15.0 and Stata 8.0 were used for data entry and data analysis. Quantitative data was expressed as mean (±SD). Qualitative data was expressed in frequency (%). For association between independent variables and the outcome, we used Pearson chi-square or Fisher’s exact test if the expected count of less than five consisted of 20% or more of the cross-tabulated data. Simple logistic regression was used to estimate the crude odds ratios (ORs) along with their 95% confidence intervals (CI) of oral cancer according to HPV exposure variables. To adjust for important confounders, the multiple logistic regression was used, and the adjusted crude OR and their 95%CI were obtained. The level of significance was fixed at P < 0.05 (twosided).

Results

The demographic information and habits of the study subjects are summarized in Table 1. As Malaysia is a multi-ethnic country, the ethnicity of the subjects was included as one of the demographic variables.

Table 1.

Demographic characteristics and personal habits of OSCC cases and controls

Variables		Controls		Cases
		n	%	n	%
Age range	18–29	20	19.0	4	3.8
	30–39	23	21.9	10	9.5
	40–49	33	31.4	16	15.2
	50–59	19	18.1	31	29.5
	60–69	7	6.7	26	24.8
	70 & above	3	2.9	18	17.1
	Total	105	100.0	105	100.0
	Mean age ± SD	42.00 ± 13.99		56.23 ± 12.84
Gender	Male	58	55.2	51	48.6
	Female	47	44.8	54	51.4
	Total	105	100.0	105	100.0
Ethnicity	Malay	82	78.1	47	44.8
	Chinese	13	12.4	15	14.3
	Indians	10	9.5	33	31.4
	Others	0	–	10	9.5
	Total	105	100.0	105	100.0
Habits	Smoking	21	20.0	39	37.1
	Alcohol use	12	11.4	21	20.0
	Betel quid chewing	11	12.5	48	45.7

More than half of the patients with OSCC in this study were 50–69 years old (54.3%). Gender distribution was matched for both OSCC cases and control groups. Malays formed the largest ethnic group in controls as well as in OSCC group though their percentage was much higher in controls (78.1%) than in OSCC cases (44.8%). More smokers, alcohol users and tobacco chewers were seen in OSCC cases group than in control group.

Detection of HPV in cases and controls

In total, 54 OSCC samples were positive for HPV (51.4%), while 26 normal samples were detected positive for HPV (24.8%) using the nested PCR assay with MY and GP primers (Table 2). It was found that prevalence of HPV was 3.21 (95%CI: 1.79–5.77) times more in cases when compared to normal controls (P < 0.001). When adjusted for habits alone like smoking, alcohol drinking and tobacco chewing, the odds of detecting HPV positivity in cases increased to 4.3 (95%CI: 2.15–8.64, P < 0.001) but decreased to 3.57 when adjusted for habits and socio-demographic characteristics. No positive results were obtained by PCR with CP primers.

Table 2.

Association of HPV with cases and controls

	Case n (%)	Control n (%)	Crude OR (95%CI)	P value*	Adjusted ORa (95%CI)	P value**	Adjusted ORb (95%CI)	P value**
HPV positive	54 (51.4)	26 (24.8)	3.21 (1.79–5.77)	<0.001	4.30 (2.15–8.64)	<0.001	3.57 (1.63–7.79)	0.001
HPV negative	51 (48.6)	79 (75.2)	1		1		1
High-risk type	40 (74.1)	11 (42.3)	3.89 (1.45–10.46)	0.006	3.31 (0.99–11.11)	0.052	2.74 (0.70–10.69)	0.140
Low-risk type	14 (25.9)	15 (57.7)	1		1		1

OR Odds ratio

*Simple logistic regression, **multiple logistic regression

^aAdjusted for smoking, alcohol drinking and betel quid chewing

^bAdjusted for age, gender, race, smoking, alcohol drinking and betel quid chewing

In cases group, out of 54 samples positive for HPV, 40 were high-risk HPVs with HPV 16 being the most frequent. Other high-risk types were 26, 31, 33, 35, 45, 51 and 58. The 26 samples positive for HPV in control group showed that 15 of these samples were low-risk HPVs. The type of low-risk HPVs found were 6, 11, 53 and 54 with HPV 6 being most frequent (n = 9,30%). High-risk HPVs were found to be significantly associated with cases (P = 0.006) when compared to control group with an odds ratio (OR) of 3.89 (95%CI: 1.45–10.46) (Table 2). Adjustment for habits reduced the OR to 3.31 (95%CI: 0.99–11.11) and further to 2.74 (95%CI: 0.70–10.69) when adjusted for socio-demographic variables and habits, the high-risk association becoming insignificant (P = 0.14).

Demographic profiles of age, gender, ethnicity and habits (smoking, alcohol use and tobacco chewing) were not associated with HPV presence (P > 0.05) in cases and controls. However, significantly less HPV positivity was seen in poorly differentiated (11.1%) compared to well-differentiated OSCCs (63.8%), although no significant association was seen with tumour site (Table 3).

Table 3.

Logistic regression analyses to predict incidence of HPV in cases

Variables		Number of OSCC (n = 105)	HPV-positive OSCC n (%)	Crude OR (95%CI)*	P value*	Adjusted ORa (95%CI)**	P value**
Age	18–29	4	2 (50.0)	1	0.562	1	0.633
	30–39	10	6 (60.0)	1.25 (0.14–10.94)	0.846	2.33 (0.05–109.04)	0.670
	40–49	16	7 (43.8)	1.87 (0.39–9.01)	0.430	1.83 (0.03–100.35)	0.771
	50–59	31	20 (64.5)	0.97 (0.25–3.77)	0.971	3.94 (0.08–194.18)	0.491
	60–69	26	11 (42.3)	2.28 (0.69–7.44)	0.188	1.24 (0.03–58.34)	0.918
	70 & above	18	8 (44.4)	0.92 (0.27–3.08)	0.891	0.93 (0.02–45.11)	0.976
Gender	Female	54	30 (55.6)	1	0.380	1	0.886
	Male	51	24 (47.1)	1.41 (0.65–3.03)		0.89 (0.19–4.17)
Ethnicity	Malay	47	21 (44.7)	1	0.622	1	0.665
	Chinese	15	9 (60.0)	0.81 (0.21–3.17)	0.761	1.13 (0.19–6.88)	0.894
	Indians	33	19 (57.6)	1.50 (0.30–7.53)	0.623	1.46 (0.36–5.90)	0.596
	Others	10	5 (50.0)	1.36 (0.33–5.61)	0.670	0.39 (0.05–2.99)	0.376
Habits	Smoking	39	18 (46.2)	0.71 (0.32–1.58)	0.411	0.76 (0.16–3.46)	0.729
	Alcohol use	21	12 (57.1)	1.33 (0.51–3.49)	0.567	1.29 (0.29–5.63)	0.733
	Betel quid chewing	48	25 (52.1)	0.90 (0.42–1.94)	0.790	0.52 (0.12–2.31)	0.391
Tumour Site	Cheek mucosa	27	15 (55.6)	1	0.866	1	0.973
	Tongue	29	16 (55.2)	0.98 (0.34–2.82)	0.981	0.76 (0.18–3.32)	0.722
	Gingiva/Gum	17	11 (64.7)	1.47 (0.42–5.13)	0.553	0.88 (0.17–4.57)	0.897
	Floor of the mouth	7	2 (28.6)	0.32 (0.53–1.95)	0.228	0.25 (0.02–2.84)	0.279
	Palate	8	5 (62.5)	1.33 (0.26–6.74)	0.738	0.73 (0.09–6.10)	0.780
	Mandible	3	0 (0.0)	–	–	–	–
	Lip	1	0 (0.0)	–	–	–	–
Differentiation	Well	47	30 (63.8)	1	0.032	1	0.081
	Moderate	49	23 (46.9)	0.50 (0.22–1.14)	0.096	0.44 (0.14–1.38)	0.163
	Poor	9	1 (11.1)	0.07 (0.01–0.62)	0.021	0.05 (0.01–0.81)	0.032

*Simple logistic regression, **multiple logistic regression

^aAdjusted for age range, gender, race, habits of smoking, betel quid chewing and alcohol

Detection of p53 polymorphism in cases and controls

In total, 90 controls and 99 cases were used for polymorphism analysis. A few samples could not be analysed as the amount of DNA left was insufficient to carry out polymorphism studies. Figure 1 shows the result of agarose gel electrophoresis of PCR amplification products for polymorphism detection from 12 OSCC samples. The distribution of arginine, proline homozygotes and Arg/Pro heterozygotes was in Hardy–Weinberg equilibrium (p² + 2pr + r² = 1) for both cases and controls (P > 0.08 and P > 0.2, respectively). Comparison of HPV-negative cases and controls with all HPV-positive cases and controls as well as high-risk HPV-positive cases and controls showed no significant association between HPV infection and p53 codon 72 polymorphisms (Table 4).

Fig. 1.

PCR amplification of arginine (top lanes) and proline (bottom lanes) alleles of codon 72 in exon 4 of the p53 gene in OSCCs showing amplimer sizes of 141 bp for arginine and 177 bp for proline. Sample No. 3, 4, 5, 9 and 11 are proline homozygous (Pro/Pro), while sample no. 2, 7, 10 and 12 are arginine (Arg/Arg) homozygous and 1, 6 and 8 are proline/arginine heterozygous. Lane M: DNA marker. P: positive control, N: negative control

Table 4.

Association of p53 polymorphism with HPV infection

HPV status	Arg/Arg n (%)	Pro/Pro n (%)	Arg/Pro n (%)	P value*
Cases (n = 99)
HPV −	11 (11.1)	14 (14.1)	22 (22.2)	0.312
All HPV +	11 (11.1)	23 (23.2)	18 (18.2)
High-risk HPV +	8 (8.1)	15 (15.1)	16 (16.1)	0.690
Total	22 (22.2)	37 (37.4)	40 (40.4)
Controls (n = 90)
HPV −	23 (25.6)	17 (18.9)	28 (31.1)	0.611
All HPV +	5 (5.6)	6 (6.7)	11 (12.2)
High-risk HPV +	2 (2.2)	2 (2.2)	6 (6.7)	0.523
Total	28 (31.1)	23 (25.6)	39 (43.3)

*Chi square test

Further, no significant association was seen in various p53 polymorphisms between HPV-positive case and controls and HPV-negative cases and controls, or between all cases and controls irrespective of HPV positivity (Table 5). Logistic regression analysis did not show any significant association of p53 polymorphisms with HPV infection in cases and controls (Table 6). Thus, we conclude that none of the p53 polymorphisms predispose to HPV infection or development of oral cancer.

Table 5.

Association of p53 polymorphism with oral cancer

HPV status	n	Arg/Arg n (%)	Pro/Pro n (%)	Arg/Pro n (%)	P value*
All HPV + cases	52	11 (21.2)	23 (44.2)	18 (34.6)	0.351
All HPV + controls	22	5 (22.7)	6 (27.3)	11 (50)
High-risk HPV + cases	39	8 (20.5)	15 (38.5)	16 (41)	0.492
High-risk HPV + controls	10	2 (20)	2 (20)	6 (60)
HPV − cases	47	11 (23.4)	14 (29.8)	22 (46.8)	0.487
HPV − controls	68	23 (33.8)	17 (25)	28 (41.2)
All cases	99	22 (22.2)	37 (37.4)	40 (40.4)	0.173
All controls	90	28 (31.1)	23 (25.6)	39 (43.3)

*Chi square test

Table 6.

Logistic regression analysis for association of polymorphisms with HPV in cases and controls

Group	Polymorphisms	Crude OR (95%CI)	P*	Adjusted ORa (95%CI)*	P**
Cases	Arg/Arg	1	0.313	1	0.522
	Pro/Pro	1.64 (0.56–4.78)	0.366	1.48 (0.47–4.59)	0.501
	Arg/Pro	0.82 (0.29–2.32)	0.711	0.85 (0.29–2.54)	0.784
Controls	Arg/Arg	1	0.615	1	0.873
	Pro/Pro	1.62 (0.42–6.21)	0.487	1.42 (0.34–5.90)	0.622
	Arg/Pro	1.81 (0.55–5.95)	0.332	1.33 (0.36–4.90)	0.668

*Simple logistic regression, **multiple logistic regression

^aAdjusted for age range, gender, race, habits of smoking, betel quid chewing and alcohol

Discussion

In the current study, we investigated the prevalence of HPV DNA in normal mucosa and OSCC by using the PCR method targeting two regions of HPV genome for detection. Our overall HPV detection rate of 51.4% in oral carcinoma using MY/GP nested PCR is in line with the value of 46.5% estimated in a meta-analysis by Miller & Johnstone (Miller and Johnstone 2001). E1 region was targeted using another set of degenerate primers CPI/CPIIG. None of the 105 OSCC cases were found to be positive by this set of primer. The discrepancy in results from two targeted regions could be due to the fact that MY/GP nested PCR was a two-step PCR assay and thus had much higher sensitivity than one-step CP PCR. However, the results of MY PCR alone with that of CP PCR were found to be concordant with none of the two primer pairs able to detect any HPV positivity. This finding indicates that the oral HPV infections were weakly positive and is consistent with prior observations that HPV DNA-positive oral samples generally produce weaker PCR products than HPV-positive cervical specimens (Giovannelli et al. 2002; Ostwald et al. 1994).

HPV 16 was the most frequent type of high-risk HPV found in cases and control group. No HPV 18 was detected in our samples. A previous study on benign cervices and cervical carcinomas in Malaysia had found HPV 16 in 72% of the cervical carcinomas. No HPV 18 was detected in any of the samples (Cheah et al. 2002). It appears that HPV 16 is the most prevalent high-risk type in the Malaysian population and therefore most implicated in carcinoma development. Although HPV 16 and HPV 18 are most commonly implicated in carcinomas (zur Hausen 1996), the non-detection of HPV 18 is not unique to Malaysia. Ibieta et al. (2005) evaluated a Mexican population where HPV 18 was not detected in any of the 51 OSCC samples analysed. A study done in Taiwan on 17 OSCCs for the presence HPV types 6, 11, 16 and 18 sequences detected HPV-16 in 76.4% of the oral carcinomas and in all three cases of papilloma. None of the samples contained HPV 6, 11 or 18 sequences (Chang et al. 1989).

No association was found between various habits and HPV positivity in control and cases group (Table 3). Most studies agree that there is no statistically significant correlation between HPV prevalence and personal habits like smoking, alcohol use and tobacco or betel quid chewing (Badaracco et al. 2000; Baez et al. 2004; Ibieta et al. 2005). One study found that the positive rates of all HPV types and of high-risk HPV types were significantly higher in non-oral habits (OH)-associated OSCC samples than in OH-associated OSCC samples, suggesting that HPVs, particularly high-risk HPVs, might be associated with the development of OSCCs, especially the non-OH-associated OSCCs (Chang et al. 2003). No significant association was found between the habit and non-habit OSCC groups in our study which was similar to the results of Dahlstrom et al. 2003 where it was found that HPV-16 infection did not appear to be more common in never-smokers than ever-smokers with SCC of head and neck (Dahlstrom et al. 2003).

According to our study, highest HPV positivity was found in well-differentiated SCCs with 30 out of 47 cases detected positive for HPV (63.8%). This result was suggestive of significance (P = 0.089) (0.05 < P < 0.10). The percentage of HPV positivity decreased with differentiation as 23 out of 49 moderately differentiated SCC (46.9%) were positive for HPV (P = 0.523). The percentage of HPV positivity decreased further significantly in poorly differentiated SCCs with only one of the nine cases (11.1%, P = 0.016) found positive for HPV (Table 3). One possible reason for a significantly low association with poorly differentiated SCCs could be due to keratinization levels seen at these grades of SCCs. In one study, HPV 16 copy number was higher in keratinizing tumours (well-differentiated tumours) (P < 0.05; Ikenberg et al. 1994). It is also reported that the detection of HPV DNA is significantly associated with the well-differentiated carcinoma, particularly HPV 16 and HPV 6, which results in the keratinization of these lesions (Wilczynski et al. 1988). As this virus is known to infect squamous epithelium and well-differentiated carcinomas appear to resemble their parent tissue and produce keratin, it is more likely for HPV to be more prevalent in well-differentiated SCCs. Further, as HPV is known to interact with wild-type p53 (Scheffner et al. 1990), high frequency of p53 mutations in poorly differentiated SCCs could be another reason for its low prevalence in these types of SCCs. Although HPV infection seems to be associated with more highly differentiated lesions, the evaluation of larger number of tumours of varying degrees of differentiation is needed to confirm this hypothesis.

Our study further demonstrated that there is no association between the p53 codon 72 polymorphism and HPV positivity in control group as well in cases (Table 4) as well as development of oral cancer (Table 5). Several recent studies have re-examined the association of alleles of p53 codon 72 with cervical cancer (Katiyar et al. 2003; Minaguchi et al. 1998; Nishikawa et al. 2000). With the exception of the initial report by Storey et al. (1998) and another by Zehbe et al. (2001), none of these studies found an association between arginine homozygosity and cervical cancer.

Several studies on head and neck and oral cancers have shown similar findings as well. A recent study that investigated squamous cell carcinomas of the head and neck (SCCHN) in 122 patients and 193 healthy subjects found no significant association between the p53 polymorphism with HPV 16/18 transcript and the development of SCCHN (Scheckenbach et al. 2004). In a study done on Indian population, Katiyar et al. (2003) found no association between the p53 codon 72 polymorphism and oral cancer in 13 patients with high-risk positive carcinomas and 20 normal controls from an Indian population. An earlier study performed on 202 patients with oral cancer and 333 controls also found no association between the p53 codon 72 polymorphism and mucosal oncogenic HPVs or between the p53 codon 72 polymorphism and risk of oral cancer (Summersgill et al. 2000).

However, two recent studies on squamous cell carcinomas of oropharynx (SCCOP) have found association of p53 polymorphisms with increased risk for carcinoma. In one study, the increase in risk of SCCOP associated with HPV16 seropositivity was found to be significantly greater for individuals with p53 codon 72 variant genotypes (Arg/Pro and Pro/Pro; OR, 9.2) than for those with the p53 Arg/Arg genotype (OR, 3.9; Ji et al. 2008). Perrone et al. (2007) investigated the role of p53 codon 72 polymorphism with HPV-associated SCCOP and found that the p53 Pro/Pro genotype was a risk factor for developing HPV16-associated SCCOP. In our study, proline homozygosity was higher in cases (37.4%) compared to controls (25.6%; Table 4), but these changes were not found to be statistically significant, therefore do not indicate an increased cancer risk. Based on our results, we may surmise that p53 codon 72 polymorphisms have no association with oral cancer risk, though other known risk factors were present in the cases (smoking, tobacco chewing and alcohol consumption). In order to assess independently the risk of p53 polymorphisms for cancer, only patients without any of these habits ought to be considered, however, due to small sample number this was not possible.

A possible explanation for the increased risk of Pro/Pro genotype is that the change from Arg to Pro modifies the p53 protein, affecting p53’s susceptibility to E6-mediated degradation thus leading to an increased carcinogenic potential of HPV16 (Ji et al. 2008). Arg/Arg genotype 72 has been shown to have greater capacity to induce apoptosis than Pro/Pro genotype (Dumont et al. 2003), while Pro/Pro 72 has greater transcriptional transactivation ability (Thomas et al. 1999) and induces more cell cycle arrest in G1 (Pim and Banks 2004). Thus, it has also been proposed that the protective effect of Arg/Pro heterozygous genotype may be due to a functional combination of the two p53 polymorphic forms, which have different biochemical and biological properties (Perrone et al. 2007).

The conflicting results from various studies may be due to the differences in cancer sites, sample size of cases and controls as well as geographic regions (Beckman et al. 1994). One study reported that frequency of arginine allele increases as the latitude increases, showing ethnic and clinal variation of codon 72 polymorphism. Studies have shown that some sites are more likely to be associated with HPV infection than others (zur Hausen 1996). Similar is the case with head and neck cancers where susceptibility of squamous epithelium to HPV transformation has been found to be more sensitive to HPV infection in tonsillar and oropharyngeal region (Andl et al. 1998; Gillison et al. 2000; Syrjanen 2005; Wilczynski et al. 1998). In contrast, oral cancers that have been found less sensitive to HPV infections may be unlikely to be associated with p53 polymorphisms, as in our case.

Limitation of this study was the retrospective case–control study design used, which could only assess the association or indicate the risk factor but not the causation of the disease. Therefore, we suggest a prospective cohort study for future research.

Conclusion

Our results represent an important preliminary baseline data on HPV related to OSCC in a Malaysian population. Results suggest that HPV infection could be one of several risk factors contributing to oral SCC in Malaysians. HPV 16 is the predominant high-risk HPV type detected. We conclude that none of the p53 polymorphisms predisposes to oral HPV infection or development of oral cancer, although this evaluation may have been limited by sample size.