Abstract

Background: The expression levels of excision repair cross-complementation group 1 (ERCC1), replication protein A (RPA) and xeroderma pigmentosum group F (XPF) nucleotide excision repair proteins may be important in the response to platin-based therapy in lung cancer patients. It is not known whether ERCC1, RPA and XPF expression levels differ between ever smokers (ES) and never smokers (NS).

Patients and methods: ERCC1, RPA and XPF expression levels were immunohistochemically evaluated in 125 patients with resected lung adenocarcinoma (AC) and carefully reviewed smoking status.

Results: ERCC1 was correlated with XPF (P = 0.001), but not with RPA (P = 0.11). In the univariate analysis, ERCC1 and XPF levels were higher in NS compared with ES (P = 0.004 and P = 0.003, respectively). In the multivariate analysis, the smoking status was predictive of the ERCC1 level [odds ratio (OR) 2.5, 95% confidence interval (CI) 1.03–6.2] after adjustment for variables linked to the smoking status, including age and the presence of bronchioloalveolar (BAC) features. The smoking status was also predictive of both RPA (OR 6.7, 95% CI 1.5–33.3) and XPF levels (OR 12.5, 95% CI 2.9–50) after adjusting for age, sex and BAC features.

Conclusion: In patients with resected lung AC, ERCC1, RPA and XPF expression levels are higher in NS compared with ES.

introduction

About 90% of lung cancer deaths are caused by tobacco smoking [1]. Adenocarcinoma (AC) is the most common type of lung cancer and the least strongly associated with smoking. Over the last several decades, its incidence has increased in relation to other histological types. Although cigarette smoking is the main risk factor for lung cancer, lung cancer in never smokers (NS) ranks as the seventh most common cause of cancer death worldwide, ahead of cancer of the cervix, pancreas and prostate [1].

Exposure to tobacco smoke carcinogens is directly responsible for gene alterations seen in lung cancer. Striking differences between ever smokers (ES) and NS have been found in the mutation rate in the three major genes EGFR, KRAS and TP53 that are involved in the pathogenesis of lung cancers. These differences suggest that lung cancer arises via different molecular mechanisms in NS compared with ES. The TP53 mutational signature (ratio of transitions, transversions and deletions) and the mutational spectrum (distribution of mutations along the gene) are distinct in lung cancers in ES and NS [2]. Mutations in EGFR have emerged as a frequent molecular alteration in lung cancer in NS [3, 4]. Furthermore, EGFR and KRAS mutations exhibit a mutually exclusive distribution [5]. Epigenetic changes such as DNA methylation differ in NS from those described in ES and more specifically in p16, hMLH1 and hMSH2 [6, 7]. Also, chromosomal aberrations in lung cancer have been described as possibly different in NS and in ES [8].

Platinum-based chemotherapy has been used to treat a wide variety of solid tumours including lung, head and neck, ovarian, cervical and testicular cancer for over three decades [9]. Many patients eventually relapse following treatment and become refractory to these anticancer agents [10]. Chemotherapy failure is mainly due to drug resistance in tumour cells. Recent studies have reported that the DNA repair capacity may have a major impact on the onset of drug resistance [11]. The cytotoxic effects of platin-based combinations depend on the formation of platinum–DNA adducts, which cause inter- and intrastrand cross-linking that triggers a series of intracellular events that ultimately result in cell death [10, 11]. DNA adducts are processed and repaired by the nucleotide excision repair (NER) pathway, a complex network of at least 25 polypeptides [12]. The NER pathway contains several steps, including lesion recognition, opening of the double DNA helix around the damage, excision of the DNA strand carrying the platinum adduct and polymerization of a new DNA strand followed by its ligation. Excision repair cross-complementation group 1 (ERCC1) is a component of the NER pathway, which is essential for the repair of platinum–DNA adducts and is associated with cellular resistance to platinum compounds [13–15]. ERCC1 is one of the rate-limiting enzymes in the NER complex, together with its obligate partner, the xeroderma pigmentosum group F (XPF) protein [16]. The ERCC1–XPF heterodimeric protein complex is responsible for the incision that cleaves on the 5′ side of the DNA strand lesion. In addition, the ERCC1–XPF complex must interact with the replication protein A (RPA) for the DNA binding step. In fact, RPA plays a crucial role in damage recognition and in the positioning of the incision steps of excision repair through cooperative DNA binding with ERCC1–XPF such that the nuclease incises only the damaged strand [12, 17]. The ERCC1–XPF structure-specific nuclease also plays a role in the homologous recombination repair of interstrand cross-links [18].

We reported in the IALT-Bio study that patients with completely resected non-small-cell lung cancer (NSCLC) and ERCC1-negative tumours benefited from adjuvant cisplatin-based chemotherapy, whereas patients with ERCC1-positive tumours did not [15]. Other reports confirmed that ERCC1 expression was correlated with a better prognosis in the observation group [19, 20].

We had no information on the smoking status of the patients included in the IALT-Bio study. Therefore, we used samples from another cohort of patients with lung AC who had undergone curative surgery and for whom the smoking status was precisely established. Here, we report results on the expression of ERCC1 and its partners XPF and RPA in ES and NS in lung AC.

patients and methods

clinical data

Using a single institution clinical database (Institut Mutualiste Montsouris, Paris, France), 217 patients with lung AC who had undergone curative surgery from January 1995 to December 2003 were retrospectively identified. Complete clinical data including follow-up information and a detailed smoking history were obtained either by exhaustive electronic or manual query of clinical databases or by postal or personal communication with the patients, their families and at least two treating physicians. This review led to the exclusion of 29 cases due to incomplete clinical records (19 cases), prior treatment with neoadjuvant chemotherapy (two cases) or because the metastatic origin of the lung tumour could not be ruled out (eight cases). The study cohort therefore included a total of 188 chemonaive patients with primary lung AC, for which formol-fixed, paraffin-embedded tumour samples were available. According to the World Health Organisation definition [21], a patient was considered an ES if she or he admitted to having smoked at least 100 cigarettes in his or her life, whereas patients who denied any active tobacco exposure or had smoked <100 cigarettes in their life were defined as NS. The study was carried out according to the national legal regulations.

immunohistochemistry

Whole sections were used for ERCC1 immunostaining, as in the IALT-Bio study [15]. Immunostaining for RPA and XPF was done on tissue microarray (TMA) slides. The TMA construction procedure has been previously described [22].

Immunostaining was carried out using standard procedures following the Vectastain Elite ABC kit instructions. To enhance epitope exposure, deparaffinised slides were heated at 98°C in 10 mmol/l citrate buffer at pH 6 for 30 min. After blockade of endogenous peroxidases and nonspecific sites, slides were incubated with the primary antibody at 4°C overnight (ERCC1 antibody, clone 8F1, dilution 1 : 400 and RPA antibody, clone 9H8, dilution 1 : 200; Lab Vision, Fremont, CA; XPF antibody, clone SMP228, dilution 1 : 400; Abcam, Cambridge, UK,). The chromogen used to localise the antigen was NovaRED (Vector). Slides were counterstained with Mayer's haematoxylin. As a negative control, the staining procedure was carried out by omitting the primary antibody.

immunohistochemical evaluation

The whole sections were evaluated in at least five fields by one observer by light microscopy at ×100, ×200 and ×400 magnifications. The immunostained TMA slides were scanned (VM3 virtual scanner, Siemens, Germany) and registered as electronic archives of high-resolution and high-quality images, thus enabling the study of an identical image for each of the three spots by two independent readers at various magnifications.

As ERCC1, RPA and XPF were ubiquitously expressed in normal cells, the stromal or lung cells in tumour samples were carefully examined for the expression of each protein. Cases without any reactivity in stromal or lung cells were excluded from the analysis. Only nuclear staining of ERCC1, RPA and XPF was validated for the analysis.

The H-score was used for scoring ERCC1 immunostaining in the IALT-Bio study [15]. In order to evaluate semi-quantitatively the staining for each marker in a homogenous and comparable manner, the H-score method was also used for RPA and XPF. Briefly, the percentage of stained tumour cells (assigning a score of 0, 0.1, 0.5 or 1 for 0%, 1%–9%, 10%–49% or ≥50% of stained cancer cells, respectively) and the intensity of staining (scored 0 for the absence of staining to 3 for strong staining) were recorded. The H-score was calculated as the product of the percentage and intensity scores. Consequently, there were nine possible results for the H-score, on a scale of 0 (absence of stained cancer cells) to 3 (strong staining in at least half of the cancer cells).

statistical analysis

Differences between subgroups were assessed using Fisher's exact test or the Mann–Whitney U test. The Kendall tau statistic was used to measure the correlation between immunohistochemical variables. In order to be analysed as a dependent variable in the logistic regression models, the H-scores for each protein were dichotomised into low and high subgroups using the median as a cut-off point. The logistic regression models were developed with the smoking status and characteristics linked to the smoking status as exposure variables. Multivariate analysis of overall survival (OS) was carried out to examine whether protein expression was prognostic after adjustment for clinical and pathological variables in Cox proportional hazard models. All tests were two sided, and a P value of <0.05 was considered statistically significant.

results

patient characteristics

Among the 188 validated cases, 125 (66%) had a value for each protein (160, 138 and 174 cases were assessable for ERCC1, RPA and XPF, respectively) and were included in the analysis. The characteristics of the 125 included patients are shown in Table 1. Patients who were included in the study and the 63 patients who were excluded were comparable in terms of demographic, clinical and histological characteristics.

Table 1.

Expression of the repair proteins according to the main characteristics of patients and tumours

 All patientsa ERCC1
 
P value RPA
 
P value XPF
 
P value 
 N = 125, n (%) Positive Negative  Positive Negative  Positive Negative  
Age at diagnosis (years)    0.28   0.14   0.18 
    ≥63 61 (49) 29 32  34 27  38 23  
    <63 64 (51) 24 40  44 20  48 16  
Sex    <0.0001   0.36   0.18 
    Male 62 (50) 57  36 26  39 23  
    Female 63 (50) 48 15  42 21  47 16  
   0.95   0.77   0.57 
    1 44 (35) 18 26  27 17  31 13  
    2 75 (60) 32 43  48 27  52 23  
    3 6 (5)    
   0.63   0.62   0.52 
    0 82 (67) 34 48  49 33  54 28  
    1 18 (14) 12  13  13  
    2 23 (18) 11 12  15  18  
Stage    0.67   0.76   0.66 
    IA 35 (28) 15 20  21 14  23 12  
    IB 42 (34) 17 25  25 17  29 13  
    II 22 (18) 15  16  14  
    IIIA 24 (19) 12 12  15  19  
Bronchioloalveolar component    0.82   0.17   0.47 
    No 101 (81) 42 59  66 35  71 30  
    Yes 24 (19) 11 13  12 12  15  
Differentiation    0.05   0.18   0.89 
    High 52 (42) 28 24  35 17  37 15  
    Moderate 59 (47) 22 37  32 27  40 19  
    Poor 14 (11) 11  11   
Adjuvant cisplatin-based chemotherapy    0.58     0.22 
    No 111 (89) 46 65  69 42  74 37  
    Yes 14 (11)   12  
Adjuvant radiotherapy    0.59   0.78   0.57 
    No 109 (87) 45 64  67 42  76 33  
    Yes 16 (13)  11  10  
History of malignancy    0.99   0.83   0.49 
    No 94 (75) 41 53  59 35  66 28  
    Yes 29 (23) 12 17  17 12  18 11  
Second cancer    0.99     0.18 
    No 118 (94) 51 67  73 45  79 39  
    Yes 6 (5)    
 All patientsa ERCC1
 
P value RPA
 
P value XPF
 
P value 
 N = 125, n (%) Positive Negative  Positive Negative  Positive Negative  
Age at diagnosis (years)    0.28   0.14   0.18 
    ≥63 61 (49) 29 32  34 27  38 23  
    <63 64 (51) 24 40  44 20  48 16  
Sex    <0.0001   0.36   0.18 
    Male 62 (50) 57  36 26  39 23  
    Female 63 (50) 48 15  42 21  47 16  
   0.95   0.77   0.57 
    1 44 (35) 18 26  27 17  31 13  
    2 75 (60) 32 43  48 27  52 23  
    3 6 (5)    
   0.63   0.62   0.52 
    0 82 (67) 34 48  49 33  54 28  
    1 18 (14) 12  13  13  
    2 23 (18) 11 12  15  18  
Stage    0.67   0.76   0.66 
    IA 35 (28) 15 20  21 14  23 12  
    IB 42 (34) 17 25  25 17  29 13  
    II 22 (18) 15  16  14  
    IIIA 24 (19) 12 12  15  19  
Bronchioloalveolar component    0.82   0.17   0.47 
    No 101 (81) 42 59  66 35  71 30  
    Yes 24 (19) 11 13  12 12  15  
Differentiation    0.05   0.18   0.89 
    High 52 (42) 28 24  35 17  37 15  
    Moderate 59 (47) 22 37  32 27  40 19  
    Poor 14 (11) 11  11   
Adjuvant cisplatin-based chemotherapy    0.58     0.22 
    No 111 (89) 46 65  69 42  74 37  
    Yes 14 (11)   12  
Adjuvant radiotherapy    0.59   0.78   0.57 
    No 109 (87) 45 64  67 42  76 33  
    Yes 16 (13)  11  10  
History of malignancy    0.99   0.83   0.49 
    No 94 (75) 41 53  59 35  66 28  
    Yes 29 (23) 12 17  17 12  18 11  
Second cancer    0.99     0.18 
    No 118 (94) 51 67  73 45  79 39  
    Yes 6 (5)    
a

Patients from whom informations were available.

The median age at diagnosis of the included patients was 63 years (interquartile range 54–72). Patients were well balanced between male and female (62 of 63). The majority of patients were ES (95 of 125 or 76%). Two-thirds (77 of 125 or 62%) had stage I disease. AC with bronchioloalveolar (BAC) features was a minor part of the AC (24 of 125 or 19%). Most patients had not received adjuvant chemotherapy (114 of 125 or 89%) and 16 of 125 (13%) adjuvant radiotherapy.

NS were more frequently women (28 of 30 or 93%) and older (median age 71.5 years, interquartile range 62–76) compared with ES (60 of 95 or 63%, P < 0.001; median age 60, interquartile range 53.5–70, P = 0.004). The tumours in NS exhibited BAC features more frequently (P = 0.03) (10 of 33 or 30%) compared with ES (14 of 95 or 15%).

expression of ERCC1, RPA and XPF

The median H-score was 2 (interquartile range 1–2) for both RPA and XPF, while it was 1 (interquartile range 0.2–2) for ERCC1. Representative examples of cases with strong positive and negative staining for each protein are provided in Figure 1.

Figure 1.

Representative examples of strongly expressing (H-score of 3, left column) and negative (H-score of 0, right column) NSCLC for each repair protein. (A) ERCC1; (B) RPA and (C) XPF. The arrows show positive stromal cells.

Figure 1.

Representative examples of strongly expressing (H-score of 3, left column) and negative (H-score of 0, right column) NSCLC for each repair protein. (A) ERCC1; (B) RPA and (C) XPF. The arrows show positive stromal cells.

There was a correlation between ERCC1 and XPF (tau = 0.24, P = 0.001). No correlation was observed between ERCC1 and RPA (tau = 0.11, P = 0.11).

univariate analysis of ERCC1, RPA and XPF protein expression according to smoking status

The median and interquartile range of the H-scores for ERCC1, RPA and XPF according to the smoking status are shown in Figure 2 and Table 2. Using the Mann–Whitney U test, the ERCC1 H-scores were higher (P = 0.004) in NS (median 2, interquartile range 1–3) compared with ES (median 1, interquartile range 0.2–2). The same comparison showed that XPF was expressed at higher levels (P = 0.003) in NS (median 2, interquartile range 2–3) compared with ES (median 2, interquartile range 1–2). There was no difference in RPA expression between NS (median 2, interquartile range 1–2.75) and ES (median 2, interquartile range 1–2, P = 0.40).

Figure 2.

Distribution of ERCC1, RPA and XPF H-score according to smoking status. Lower and upper limits of boxes indicate the first and third quartiles; circles within boxes indicate the median values; the lines extending from the boxes indicate the range of values. ES: ever smokers, NS: never smokers.

Figure 2.

Distribution of ERCC1, RPA and XPF H-score according to smoking status. Lower and upper limits of boxes indicate the first and third quartiles; circles within boxes indicate the median values; the lines extending from the boxes indicate the range of values. ES: ever smokers, NS: never smokers.

Table 2.

Univariate analysis of ERCC1, RPA, XPF expression according to smoking status

Variable All Ever smokers (n = 95) Never smokers (n = 30) P value (Mann–Whitney U test) 
Median (range) Median (interquartile range) Median (Interquartile range) 
ERCC1 (n = 125) 1 (0–3) 1 (0.2–2) 2 (1–3) 0.004 
RPA (n = 125) 2 (0–3) 2 (1–2) 2 (1–2.75) 0.40 
XPF (n = 125) 2 (0–3) 2 (1–2) 2 (2–3) 0.003 
Variable All Ever smokers (n = 95) Never smokers (n = 30) P value (Mann–Whitney U test) 
Median (range) Median (interquartile range) Median (Interquartile range) 
ERCC1 (n = 125) 1 (0–3) 1 (0.2–2) 2 (1–3) 0.004 
RPA (n = 125) 2 (0–3) 2 (1–2) 2 (1–2.75) 0.40 
XPF (n = 125) 2 (0–3) 2 (1–2) 2 (2–3) 0.003 

The medians were used to distinguish high (positive) and low (negative) expression for each protein. The number of positive cases was 53 (or 42%), 78 (or 62%) and 86 (or 69%) for ERCC1, RPA and XPF, respectively. The proportions of positive cases were higher in NS for both ERCC1 (19 of 30 or 63%) and XPF (14 of 30 or 46%) compared with ES (ERCC1, 34 of 95 or 36%, P = 0.01; XPF, 8 of 95 or 8%, P < 0.0001). There was no difference for RPA (P = 0.40).

multivariate logistic regression analysis of ERCC1, RPA and XPA proteins

The logistic regression method was used to determine whether the dichotomised levels of ERCC1, RPA and XPF expression into subgroups of low and high expressers were explained by the smoking status after adjustment for clinical or pathological variables linked to the smoking status, including sex, age at diagnosis and the presence of BAC features (Table 3). The smoking status was predictive of the ERCC1 level [odds ratio (OR) 2.5, 95% confidence interval (CI) 1.03–6.2, P = 0.04] after inclusion of age and BAC features. Identical results were noted when age was classified into subgroups: <55, 55–64 and >64 as in the IALT-Bio study. The smoking status was not predictive of the ERCC1 level when sex was included in the model. The smoking status was predictive of both RPA (OR 6.7, 95% CI 1.5–33.3) and XPF levels (OR 12.5, 95% CI 2.9–50) after adjusting for age, sex and BAC features.

Table 3.

Logistic regression analysis of ERCC1, RPA and XPF expression according to smoking status

 odds ratio 95% CI low 95% CI high P value 
ERCC1     
    Overall model fit    0.03 
    Exposure variable     
    Age (years) 1.01 0.98 1.05 0.41 
    Sex (male versus female) 0.94 0.4 2.2 0.88 
    Bronchioloalveolar component (yes versus no) 2.4 0.9 6.2 0.11 
    Smoking status (never smokers versus ever smokers) 2.1 0.75 5.9 0.09 
ERCC1     
    Overall model fit    0.01 
    Exposure variable     
    Age (years) 1.01 0.98 1.05 0.41 
    Bronchioloalveolar component (yes versus no) 2.2 0.84 5.7 0.11 
    Smoking status (never smokers versus ever smokers) 2.5 1.03 6.2 0.04 
RPA     
    Overall model fit    0.02 
    Exposure variable     
    Age (years) 0.97 0.92 1.01 0.17 
    Sex (male versus female) 2.3 0.61 8.7 0.22 
    Bronchioloalveolar component (yes versus no) 0.12 0.01 1.01 0.05 
    Smoking status (never smokers versus ever smokers) 6.8 1.5 31.2 0.01 
XPF     
    Overall model fit    0.0004 
    Exposure variable     
    Age (years) 0.98 0.94 1.03 0.49 
    Sex (male versus female) 1.13 0.28 4.4 0.87 
    Bronchioloalveolar component (yes versus no) 0.73 0.20 2.7 0.64 
    Smoking status (never smokers versus ever smokers) 12.3 2.9 51 0.0006 
 odds ratio 95% CI low 95% CI high P value 
ERCC1     
    Overall model fit    0.03 
    Exposure variable     
    Age (years) 1.01 0.98 1.05 0.41 
    Sex (male versus female) 0.94 0.4 2.2 0.88 
    Bronchioloalveolar component (yes versus no) 2.4 0.9 6.2 0.11 
    Smoking status (never smokers versus ever smokers) 2.1 0.75 5.9 0.09 
ERCC1     
    Overall model fit    0.01 
    Exposure variable     
    Age (years) 1.01 0.98 1.05 0.41 
    Bronchioloalveolar component (yes versus no) 2.2 0.84 5.7 0.11 
    Smoking status (never smokers versus ever smokers) 2.5 1.03 6.2 0.04 
RPA     
    Overall model fit    0.02 
    Exposure variable     
    Age (years) 0.97 0.92 1.01 0.17 
    Sex (male versus female) 2.3 0.61 8.7 0.22 
    Bronchioloalveolar component (yes versus no) 0.12 0.01 1.01 0.05 
    Smoking status (never smokers versus ever smokers) 6.8 1.5 31.2 0.01 
XPF     
    Overall model fit    0.0004 
    Exposure variable     
    Age (years) 0.98 0.94 1.03 0.49 
    Sex (male versus female) 1.13 0.28 4.4 0.87 
    Bronchioloalveolar component (yes versus no) 0.73 0.20 2.7 0.64 
    Smoking status (never smokers versus ever smokers) 12.3 2.9 51 0.0006 

Numbers in bold are statistically significant. CI, confidence interval.

overall survival

The median follow-up of the whole cohort was 54 months (range 1–189 months). In the study population, 43 deaths, whatever the cause, had been observed. Median OS was 94 months. Three- and 5-year cumulative OS was 77% and 67%, respectively.

After verification of the validity of the proportional hazards assumption, the effect of all clinical and pathological variables including age, sex, smoking status, pTNM (pathological tumour–node–metastasis), type of surgery, adjuvant chemotherapy, adjuvant radiotherapy, histological differentiation and the presence of BAC features was analysed using the Cox regression method. Only age [hazard ratio (HR) 1.05, 95% CI 1.01–1.08, P = 0.01] and pTNM (HR 1.56, 95% CI 1.13–2.16, P = 0.006) were predictive of OS.

In this model, ERCC1 positivity was not predictive of OS (HR 0.67, 95% CI 0.33–1.38, P = 0.28). Neither RPA nor XPF expression scores were predictive of OS (P > 0.45 for each variable).

The interaction test between ERCC1 and adjuvant chemotherapy was not significant. Only six patients with ERCC1-positive tumours had received adjuvant chemotherapy. In the subgroup of 111 patients, who had not received adjuvant chemotherapy, age (HR 1.06, 95% CI 1.02–1.11, P = 0.002) and pTNM (HR 1.67, 95% CI 1.19–2.34, P = 0.003), but not ERCC1 positivity (HR 0.55, 95% CI 0.25–1.16, P = 0.12), were significant predictors of OS.

discussion

Accumulating evidence points to numerous differences in genetic and epigenetic changes in lung cancer cells between ES and NS [2, 6–8]. Until now, no relationship has been shown for NER protein expression in lung cancer cells according to the smoking status. In the IALT-Bio study, which included 761 patients, no smoking data were recorded [15]. In that study, ERCC1 expression was found to be independently related to age and the histological type. In the study by Zheng et al. [20], there were only 11 (6%) NS, a number that was too small to determine whether ERCC1 expression could be associated with the smoking status. To our knowledge, our study is the first to report the differential expression of ERCC1 and its NER partners RPA and XPF in NS compared with ES, using samples from patients with lung AC, who were consecutively treated by surgery at the same centre and whose smoking history was recorded by cross validating the clinical records. It should be noted that the relatively small number of NS in our cohort (30 of 125 or 24%) is consistent with epidemiological data on lung AC in France [23]. All together, our data indicate that the NER proteins ERCC1, RPA and XPF are differentially expressed according to the smoking status. The NER protein levels were generally higher in NS compared with ES, although a larger cohort and possibly the design of a case–control study are needed for a more precise evaluation of these differences.

We found that ERCC1 was higher in NS compared with ES both in univariate and multivariate analyses after controlling for age and the presence of BAC features—two characteristics which were associated with the smoking status. However, ERCC1 was not associated with the smoking status, when sex was included in the model. The data may indicate that never smoking status was not an independent predictor of ERCC1 after controlling for sex, which was itself strongly associated with ERCC1 in univariate but not in multivariate analysis, a finding previously reported in the IALT-Bio cohort [15]. We cannot rule out that NS status is not an independent predictor of ERCC1 expression, although NS status was the only variable marginally associated with ERCC1 (P = 0.09) in the multivariate model including sex. However, most NS were women, which limits the ability of the multivariate modeling to properly adjust for a confounding effect of sex. Indeed, by removing sex from the analysis, we found that NS status predicted ERCC1 expression.

Smoking status and BAC features were independent predictors of RPA expression by multivariate but not by univariate analysis. While never smoking status increased the likelihood of high RPA expression, presence of BAC features decreased it. The higher proportion of tumours with BAC features in NS confounded the relationship of never smoking status with RPA in the univariate analysis.

XPF was strongly associated with the smoking status both in univariate and multivariate analyses.

The fact that NS status was associated with higher NER protein levels suggests that tumour cell changes common to NS may be responsible for the differential modulation of NER gene expression. At present, little is known about the regulatory mechanisms governing the expression of ERCC1 and other NER proteins. A known level of regulation is the reciprocal down-regulation of protein expression between ERCC1 and XPF. Mouse knockout experiments have shown that a decrease in ERCC1 protein levels induces degradation of its heterodimeric partner XPF and reciprocally [24]. The correlation between ERCC1 and XPF reported in this study is consistent with this reciprocal relationship. It is not known whether down-regulation of protein levels affects primarily one of the two partners, ERCC1 and XPF, or both. Regarding transcriptional regulation, it has been reported that mutant H-Ras or exposure to epidermal growth factor modulates ERCC1 messenger RNA levels [25, 26]. However, these findings need to be confirmed and may depend on the cellular context. They point to the possible involvement of the RAS or epidermal growth factor receptor (EGFR) pathways, which are differentially altered in ES and NS.

Recent reports have established the predictive and prognostic value of ERCC1 expression in tumour cells in resected NSCLC [15, 19, 20]. Here, we report that the interaction term between ERCC1 and chemotherapy was not significantly associated with OS. However, only 14 patients were treated with adjuvant chemotherapy in this cohort. In the left to observation arm (n = 111), the HR associated with ERCC1 positivity (HR 0.55, 95% CI 0.25–1.16, P = 0.12) was consistent with that reported in the IALT-Bio study (HR 0.66, 95% CI 0.49–0.90, P = 0.009) given the lower statistical power in the present study.

As there was a correlation between ERCC1 and XPF levels, it would be useful to determine whether XPF is predictive of a beneficial response to adjuvant chemotherapy as previously demonstrated for ERCC1. Such study is currently in progress using the IALT-Bio cohort.

NS with advanced stage NSCLC may benefit from treatment with EGFR inhibitors [27, 28]. At present, there is no evidence that EGFR inhibitors could improve the prognosis of patients with resected NSCLC. However, ERCC1 may be used as a predictor of a survival benefit from cisplatin-based adjuvant chemotherapy as shown in several studies [15, 29]. It is hoped that adjuvant chemotherapy may be tailored to the molecular profile of tumour cells. Our team is currently involved in a clinical trial to determine whether different therapeutic strategies can be implemented according to ERCC1 expression and EGFR mutations. The smoking status will be an important clinical characteristic in this trial. If confirmed by independent studies, the higher ERCC1 level in NS compared with ES is a finding that may reinforce the decision to treat NS patients with resected lung AC with an EGFR inhibitor instead of cisplatin-based adjuvant chemotherapy.

conclusion

The NER proteins ERRC1, RPA and XPF are differentially expressed according to the smoking status. ERCC1 and XPF are correlated and expressed at a higher level in NS compared with ES. Further studies are needed to confirm these results and to explain precisely the oncogenetic alterations that are associated with ERCC1 expression and the function of other NER proteins.

funding

Institut National du Cancer: PNES Poumon to PF (2006-BC-PNESP) and Projet libre to DP, JCS and PF (0610-301616-122/PL 2006); Ligue contre le cancer: comité du Val de Marne to PF.

Teofil Dutu and Mustapha Erman participated in the TMA confection. Juana Hernandez participated in reading the TMA. The TMA slides were scanned by at European Organisation for Research and Treatment of Cancer by D. Jaminé. Giannis Mountzios and Philippe Girard participated in clinical data collection. The authors thank Lorna Saint Ange for editing.

References

1.
Parkin
DM
Bray
F
Ferlay
J
Pisani
P
Global cancer statistics, 2002
CA Cancer J Clin
 , 
2005
, vol. 
55
 (pg. 
74
-
108
)
2.
Le Calvez
F
Mukeria
A
Hunt
JD
, et al.  . 
TP53 and KRAS mutation load and types in lung cancers in relation to tobacco smoke: distinct patterns in never, former, and current smokers
Cancer Res
 , 
2005
, vol. 
65
 (pg. 
5076
-
5083
)
3.
Sonobe
M
Manabe
T
Wada
H
Tanaka
F
Mutations in the epidermal growth factor receptor gene are linked to smoking-independent, lung adenocarcinoma
Br J Cancer
 , 
2005
, vol. 
93
 (pg. 
355
-
363
)
4.
Pao
W
Miller
V
Zakowski
M
, et al.  . 
EGF receptor gene mutations are common in lung cancers from “never smokers” and are associated with sensitivity of tumors to gefitinib and erlotinib
Proc Natl Acad Sci U S A
 , 
2004
, vol. 
101
 (pg. 
13306
-
13311
)
5.
Tam
IY
Chung
LP
Suen
WS
, et al.  . 
Distinct epidermal growth factor receptor and KRAS mutation patterns in non-small cell lung cancer patients with different tobacco exposure and clinicopathologic features
Clin Cancer Res
 , 
2006
, vol. 
12
 (pg. 
1647
-
1653
)
6.
Divine
KK
Pulling
LC
Marron-Terada
PG
, et al.  . 
Multiplicity of abnormal promoter methylation in lung adenocarcinomas from smokers and never smokers
Int J Cancer
 , 
2005
, vol. 
114
 (pg. 
400
-
405
)
7.
Wang
YC
Lu
YP
Tseng
RC
, et al.  . 
Inactivation of hMLH1 and hMSH2 by promoter methylation in primary non-small cell lung tumors and matched sputum samples
J Clin Invest
 , 
2003
, vol. 
111
 (pg. 
887
-
895
)
8.
Wong
MP
Fung
LF
Wang
E
, et al.  . 
Chromosomal aberrations of primary lung adenocarcinomas in nonsmokers
Cancer
 , 
2003
, vol. 
97
 (pg. 
1263
-
1270
)
9.
Lokich
J
Anderson
N
Carboplatin versus cisplatin in solid tumors: an analysis of the literature
Ann Oncol
 , 
1998
, vol. 
9
 (pg. 
13
-
21
)
10.
Siddik
ZH
Cisplatin: mode of cytotoxic action and molecular basis of resistance
Oncogene
 , 
2003
, vol. 
22
 (pg. 
7265
-
7279
)
11.
Reed
E
Platinum-DNA adduct, nucleotide excision repair and platinum based anti-cancer chemotherapy
Cancer Treat Rev
 , 
1998
, vol. 
24
 (pg. 
331
-
344
)
12.
de Laat
WL
Jaspers
NG
Hoeijmakers
JH
Molecular mechanism of nucleotide excision repair
Genes Dev
 , 
1999
, vol. 
13
 (pg. 
768
-
785
)
13.
Metzger
R
Leichman
CG
Danenberg
KD
, et al.  . 
ERCC1 mRNA levels complement thymidylate synthase mRNA levels in predicting response and survival for gastric cancer patients receiving combination cisplatin and fluorouracil chemotherapy
J Clin Oncol
 , 
1998
, vol. 
16
 (pg. 
309
-
316
)
14.
Shirota
Y
Stoehlmacher
J
Brabender
J
, et al.  . 
ERCC1 and thymidylate synthase mRNA levels predict survival for colorectal cancer patients receiving combination oxaliplatin and fluorouracil chemotherapy
J Clin Oncol
 , 
2001
, vol. 
19
 (pg. 
4298
-
4304
)
15.
Olaussen
KA
Dunant
A
Fouret
P
, et al.  . 
DNA repair by ERCC1 in non-small-cell lung cancer and cisplatin-based adjuvant chemotherapy
N Engl J Med
 , 
2006
, vol. 
355
 (pg. 
983
-
991
)
16.
Evans
E
Moggs
JG
Hwang
JR
, et al.  . 
Mechanism of open complex and dual incision formation by human nucleotide excision repair factors
EMBO J
 , 
1997
, vol. 
16
 (pg. 
6559
-
6573
)
17.
Batty
DP
Wood
RD
Damage recognition in nucleotide excision repair of DNA
Gene
 , 
2000
, vol. 
241
 (pg. 
193
-
204
)
18.
Niedernhofer
LJ
Odijk
H
Budzowska
M
, et al.  . 
The structure-specific endonuclease Ercc1-Xpf is required to resolve DNA interstrand cross-link-induced double-strand breaks
Mol Cell Biol
 , 
2004
, vol. 
24
 (pg. 
5776
-
5787
)
19.
Simon
GR
Sharma
S
Cantor
A
, et al.  . 
ERCC1 expression is a predictor of survival in resected patients with non-small cell lung cancer
Chest
 , 
2005
, vol. 
127
 (pg. 
978
-
983
)
20.
Zheng
Z
Chen
T
Li
X
, et al.  . 
DNA synthesis and repair genes RRM1 and ERCC1 in lung cancer
N Engl J Med
 , 
2007
, vol. 
356
 (pg. 
800
-
808
)
21.
West
R
Tobacco control: present and future
Br Med Bull
 , 
2006
, vol. 
77–78
 (pg. 
123
-
136
)
22.
Dutu
T
Michiels
S
Fouret
P
, et al.  . 
Differential expression of biomarkers in lung adenocarcinoma: a comparative study between smokers and never-smokers
Ann Oncol
 , 
2005
, vol. 
16
 (pg. 
1906
-
1914
)
23.
Devesa
SS
Bray
F
Vizcaino
AP
Parkin
DM
International lung cancer trends by histologic type: male:female differences diminishing and adenocarcinoma rates rising
Int J Cancer
 , 
2005
, vol. 
117
 (pg. 
294
-
299
)
24.
Gaillard
PH
Wood
RD
Activity of individual ERCC1 and XPF subunits in DNA nucleotide excision repair
Nucleic Acids Res
 , 
2001
, vol. 
29
 (pg. 
872
-
879
)
25.
Youn
CK
Kim
MH
Cho
HJ
, et al.  . 
Oncogenic H-Ras up-regulates expression of ERCC1 to protect cells from platinum-based anticancer agents
Cancer Res
 , 
2004
, vol. 
64
 (pg. 
4849
-
4857
)
26.
Andrieux
LO
Fautrel
A
Bessard
A
, et al.  . 
GATA-1 is essential in EGF-mediated induction of nucleotide excision repair activity and ERCC1 expression through ERK2 in human hepatoma cells
Cancer Res
 , 
2007
, vol. 
67
 (pg. 
2114
-
2123
)
27.
Thatcher
N
Chang
A
Parikh
P
, et al.  . 
Gefitinib plus best supportive care in previously treated patients with refractory advanced non-small-cell lung cancer: results from a randomised, placebo-controlled, multicentre study (Iressa Survival Evaluation in Lung Cancer)
Lancet
 , 
2005
, vol. 
366
 (pg. 
1527
-
1537
)
28.
Shepherd
FA
Rodrigues Pereira
J
Ciuleanu
T
, et al.  . 
Erlotinib in previously treated non-small-cell lung cancer
N Engl J Med
 , 
2005
, vol. 
353
 (pg. 
123
-
132
)
29.
Okuda
K
Sasaki
H
Dumontet
C
, et al.  . 
Expression of excision repair cross-complementation group 1 and class III beta-tubulin predict survival after chemotherapy for completely resected non-small cell lung cancer
Lung Cancer
 , 
2008