Expression of nucleotide excision repair genes and the risk for squamous cell carcinoma of the head and neck




Phenotypic differences in the ability to repair genetic damage induced by tobacco carcinogens may reflect genetic differences in susceptibility to squamous cell carcinoma of the head and neck (SCCHN). The objective of this study was to assess the variation in baseline expression of five nucleotide excision repair genes between individuals with SCCHN and cancer free controls.


The authors conducted a hospital-based case–control study of 57 SCCHN patients and 105 cancer free controls. Using peripheral blood lymphocytes, a multiplex reverse transcriptase–polymerase chain reaction assay was used to quantitate in vitro the mRNA levels of five genes (ERCC1, XPB/ERCC3, XPG/ERCC5, CSB/ERCC6, and XPC) involved in the nucleotide excision repair pathway.


The levels of ERCC1, XPB/ERCC3, XPG/ERCC5, and CSB/ERCC6 transcripts were lower in cases than in controls (P =0.0001, 0.096, 0.001, and 0.0001, respectively). In multivariate logistic regression analysis (adjusting for age, gender, race, smoking status, and alcohol use), low expression of ERCC1, XPB/ERCC3, XPG/ERCC5, and CSB/ERCC6 was associated with a statistically significant increased risk for SCCHN (adjusted odds ratios [95% confidence intervals] 6.42 [2.63–15.69], 2.86 [1.39–5.90], 3.69 [1.73–7.90], and 2.46 [1.19–5.09], respectively).


Reduced expression of ERCC1, XPB/ERCC3, XPG/ERCC5, and CSB/ERCC6 is associated with a more than two-fold increased risk of SCCHN. Cancer 2002;94:393–7. © 2002 American Cancer Society.

Squamous cell carcinoma of the head and neck (SCCHN) is the third most common tobacco-induced malignancy in the United States,1 and in parts of the developing world (particularly Southern Asia) SCCHNs are among the most common cancers.2 Although it is well established that smoking and alcohol use are major risk factors for SCCHN,3 only a small fraction of the exposed general population develops SCCHN, suggesting that there may be differences in individual susceptibility to SCCHN.

Benzo[a]pyrene diol epoxide, a classic tobacco carcinogen, induces higher numbers of chromosomal aberrations in vitro in lymphocytes from SCCHN patients than in cancer free controls.4 Benzo[a]pyrene diol epoxide induces genetic damage chiefly through the formation of covalently bound DNA adducts throughout the genome, and these adducts occur in the tumor suppressor gene p53 in a pattern consistent with the mutation patterns of tobacco-induced malignancies.5, 6 These adducts are more prevalent in vitro in lymphocytes from SCCHN patients than in those from cancer free controls.7 The nucleotide excision repair pathway is responsible for the removal and repair of these adducts.8

Both endogenous and exogenous exposures to carcinogens or genotoxic agents cause cell cycle delays allowing repair of DNA damage,9 and DNA repair capacity is central to maintaining normal cellular functions.10 Genetic variants of DNA repair genes11 and reduced DNA repair capacity are thought to contribute to susceptibility to cancer12, 13 including SCCHN.14 Patients with SCCHN are more likely than cancer free controls to have genetic variants of three DNA repair genes XRCC,15XPD,16 and XPC.17 However, the biologic effects of these variants and their relation to the DNA repair phenotype remain to be determined.

We previously reported phenotypic differences in the mRNA levels of several DNA repair genes in lung carcinoma patients and frequency-matched cancer free controls.18 We also reported that SCCHN patients have lower expression level of mismatch repair genes such as hMLH1 than do frequency-matched cancer free controls.19 Therefore, we hypothesize that there are also variations in the baseline expression of genes involved in the nucleotide excision repair pathway in SCCHN patients and cancer free controls. To test this hypothesis, we conducted a pilot case–control study to investigate the role of mRNA levels of five genes (ERCC1, XPB/ERCC3, XPG/ERCC5, CSB/ERCC6, and XPC) in the etiology of SCCHN.


Study Subjects

Subject recruitment was described previously.15 Briefly, from June 1995 until September 1998, patients with newly diagnosed, previously untreated, and histologically confirmed SCCHN (of the oral cavity, oropharynx, hypopharynx, or larynx) were recruited from the registry of the Department of Head and Neck Surgery at our institution. Cancer free controls were selected from the control population of ongoing hospital-based case–control studies of cancer susceptibility at our institution. These controls were identified from enrollees in a managed care organization and were frequency matched to the case subjects on age, gender, race, smoking status, and alcohol consumption. Data on these factors were derived from questionnaires. Subjects who had smoked more than 100 cigarettes in their lifetimes were defined as smokers. Subjects who had drank alcoholic beverages at least once a week for more than 1 year were defined as drinkers. After informed consent was obtained, each subject donated 20 mL of blood in heparinized tubes for biomarker testing.

Multiplex Reverse Transcriptase–Polymerase Chain Reaction Analysis

Blood samples were processed and lymphocyte cultures were performed as described previously.18 After 72 hours of incubation, the lymphocytes were collected and stored at −80 °C for RNA isolation later.18 As described previously, multiplex reverse transcriptase–polymerase chain reaction (RT-PCR) was used to amplify and quantitate the mRNA of 5 nucleotide excision repair genes (ERCC1, XPB/ERCC3, XPG/ERCC5, CSB/ERCC6, and XPC) and β-actin from total RNA.20 β-actin was used as a loading control and internal standard for transcript quantification. After gel electrophoresis separation of the PCR products, computer densitometry was used to calculate the relative level of each band on the gel by comparison with that of the ubiquitous β-actin.

Statistical Analyses

The chi-square test was used to test the differences in distribution of selected variables between the cases and controls. Student t test was used to compare the differences in the relative expression levels analyzed as a continuous variable between groups. Covariance analysis was used to control for the effect of age. For calculation of crude odds ratios and confidence intervals, the median relative expression level in the controls was used as the cutoff point for each gene. Adjusted odds ratios were calculated by fitting logistical regression models with adjustment for age, gender, race, smoking status, and alcohol use. All the statistical analyses were performed with SAS software (version 6.12; SAS Institute Inc., Cary, NC).


Lymphocyte samples from 57 SCCHN patients and 105 control subjects were assessed for baseline in vitro expression of the nucleotide excision repair genes ERCC1, XPB/ERCC3, XPG/ERCC5, CSB/ERCC6, and XPC. The primary tumors were in the oral cavity, oropharynx or hypopharynx, and larynx in 21, 26, and 11 patients, respectively. One patient had simultaneous primaries in the oral cavity and oropharynx. The demographic information on these subjects is presented in Table 1. The average age was 56.0 years for the case subjects (median, 55; range, 25–79) and 54.3 years for the control subjects (median, 56; range, 22–84). Of the subjects, 89.4% of the cases and 87.6% of the controls were non-Hispanic whites; 70.2% of the cases and 60.0% of the controls were male; 59.6% of the cases and 59.1% of the controls were smokers; and 73.7% of the cases and 68.6% of the controls were drinkers. However, none of the differences was statistically significant, suggesting that the two groups were adequately matched on age, gender, race, smoking status, and alcohol use (Table 1). Because only frequency matching was used, these variables were further adjusted for in the multivariate analysis.

Table 1. Demographic Factors and Exposures for Case and Control Subjects
VariableCases (no = 57)Controls(no = 105)P valuea
  • a

    Two-sided chi-square test.

Age (yrs)
Ethnic group
 Non-Hispanic white51(89.4)92(87.6)0.846
Smoking status
Alcohol status

The cases demonstrated lower relative expression of four of the five nucleotide excision repair genes analyzed than did the controls. These differences were statistically significant for ERCC1, ERCC5, and ERCC6 (Table 2) and remained statistically significant when age was controlled for as a continuous variable in covariance analysis (Table 2). There were no significant differences in gene expression levels related to primary tumor location (P > 0.10; data not shown).

Table 2. Relative Expression of Nucleotide Excision Repair Genes in Cases and Controls
GeneMean relative expression ± SD% Differencet testAnalysis of covariance adjusted for age (P value)
  1. SD: standard deviation.

ERCC154.1 ± 14.467.0 ± 19.6−12.9<0.001<0.001
ERCC360.9 ± 16.366.6 ± 22.3−5.70.0960.100
ERCC561.8 ± 19.675.3 ± 27.1−13.50.0010.002
ERCC659.6 ± 19.474.7 ± 25.5−15.1<0.001<0.001
XPC55.8 ± 19.255.9 ± 25.3−0.10.9580.839

Because these five nucleotide excision repair genes participate in the same repair pathway, it is expected that their expression level may be correlated. As shown in Table 3, the expression levels of these genes were statistically significantly correlated in the controls, but the expression level of ERCC3 was not correlated with that of ERCC5 and ERCC6 in the cases.

Table 3. Correlation between Expression Levels of Nucleotide Excision Repair Genes by Cases and Controls
Pearson correlation coefficients/p values

To estimate the risk of SCCHN, the gene expression data were dichotomized into low and high categories based on the median values in the controls. Crude odds ratios and 95% confidence intervals for each gene are presented in Table 4. Low expression of ERCC1 was associated with a 6.01-fold elevated risk for SCCHN, and low expression of ERCC5 was associated with a 3.68-fold elevated risk. Low expression of ERCC3 and ERCC6 also were associated with significant elevated risk (2.51–3.01-fold) for SCCHN. After controlling for age, gender, race, smoking status, and alcohol use in multivariate logistic regression analyses, low expression of ERCC1, ERCC3, ERCC5, and ERCC6 remained associated with significantly increased risks for SCCHN.

Table 4. Risk Estimates for Nucleotide Excision Repair Gene Expression
Expression levelaCasesControlsCrude odds ratio (95% CI)Adjusted odds ratiob (95% CI)
  • CI: confidence interval.

  • a

    Dichotomized based on median value for control subjects.

  • b

    Adjusted for age, gender, race, smoking status, and alcohol use.

 Low49(86.0)53(50.5)6.01 (2.60–13.92)6.42 (2.63–15.69)
 Low43(75.4)53(50.5)3.01 (1.48–6.16)2.86 (1.39–5.90)
 Low45(79.0)53(50.5)3.68 (1.75–7.73)3.69 (1.73–7.90)
 Low52(71.9)53(50.5)2.51 (1.26–5.03)2.46 (1.19–5.09)
 Low23(40.4)52(49.5)0.69 (0.36–1.33)0.66 (0.34–1.30)


Several lines of evidence suggest that SCCHN patients differ from cancer free controls in their ability to repair genetic damage induced by tobacco carcinogens such as benzo[a]pyrene diol epoxide. In this study, we have shown that one explanation for these differences may be lower expression of the nucleotide excision repair genes ERCC1, ERCC3, ERCC5, and ERCC6. Therefore, low levels of those genes may be markers for susceptibility to SCCHN.

These findings are consistent with our previous work demonstrating that SCCHN patients are more sensitive to in vitro chromosomal damage induced by benzo[a]pyrene diol epoxide4 and to in vitro formation of benzo[a]pyrene diol epoxide DNA adducts7 and have a reduced nucleotide excision repair capacity.14 Further studies are needed to determine whether the results of these assays are correlated. For instance, do differences in expression of nucleotide excision repair genes account for differences in nucleotide excision repair capacity, and are expression levels correlated with genetic variants of DNA repair genes? We have found correlations between some of these phenotypic assays in cell lines,21 and Vogel et al. recently demonstrated a correlation between ERCC1 expression levels and DNA repair capacity in 33 subjects.22

These studies and our current study examined phenotypic measures of the DNA damage repair process that may be subject to environmental and laboratory confounders. A genetic marker of risk (or a combination of genetic markers, i.e., genetic profile) would be easier and cheaper to type and independent of many confounders. This study supports ongoing work exploring genetic differences that may underlie these phenotypic differences in the DNA repair process. We recently found that variants of the DNA repair genes XPD,16XRCC1,15 and XPC17 are genetic markers of risk to SCCHN, and we are exploring other DNA repair gene polymorphisms (including ERCC1) as markers of risk for SCCHN.

Because the multiplex RT-PCR assay20 requires only a small amount of total RNA20 and because lymphocytes are the most accessible viable tissue for population-based studies, this combination is a valuable tool for large-scale epidemiologic studies. However, this assay provides only a semiquantitative measurement of relative RNA levels and not absolute quantitation of transcripts. More precise methods are needed to reduce assay variation. The use of high-throughput real-time RT-PCR assays of quantitative gene expression23 and microarray assays for measuring expression of many genes at a time24 will revolutionize the detection of mRNA expression in large epidemiologic studies.


We thank Ms. Margaret Lung for her assistance in recruiting patients, Ms. Yongli Guan, Ms. Yun Zhang, and Mr. Zhaozheng Guo for technical assistance, Dr. Maureen Goode for scientific editing, and Ms. Beverly Lee and Ms. Joanne Sider for article preparation.