**3.1 Some HNSCC risk factors are able to inhibit DNA repair**

Epidemiological evidences have demonstrated that alcohol drinking, betel quid (BQ) chewing (especially in South Asia and South-West Pacific area including Taiwan), cigarette smoking, and infection of human papillomavirus are risk factors for HNSCC development (Haddad & Shin, 2008; IARC, 2004). The carcinogenicity of betel nut has been approved by the International Agency for Research on Cancer (IARC), a WHO organization, in 2004 (IARC, 2004), although the molecular mechanism underlying its carcinogenicity is not fully elucidated. In this regard, we have explored the possible effect of arecoline, a major alkaloid in betel nut, on DNA repair activity using HCR. We found that arecoline could inhibit the repair of UV-induced DNA damages, at least partly, through inactivating p53's expression and transactivation activity (Tsai et al., 2008). Besides, we also showed that arecoline could affect mitotic spindles and deregulated mitotic checkpoint, another key guardian of genome integrity (Wang et al., 2010). These results provide molecular explanation for BQ-associated carcinogenicity that has been shown previously by an increase of mitosis errors and micronucleus (MN) in mammalian cells (Lin, 2010). Micronucleus is a typical sign of GIN and is derived from either DNA strand breaks (clastogenic effect) or whole chromosome lagging during mitosis (aneugenic effect) (Norppa & Falck, 2003).

Epidemiological studies also show that the probability of HNSCC development is synergistically increased by simultaneous exposure of BQ, cigarette, and alcohols (Ko et al., 1995; Lee et al., 2005). Regarding the carcinogenic role of cigarette on the aspect of DNA repair, we also found that benzo(a)pyrene (BaP), an important carcinogen in cigarette (IARC, 2010), exhibited negative effects on DNA repair (Lin et al., 2011 manuscript in preparation). The mechanistic study regarding the synergistic effect of arecoline and BaP on regulating DNA repair, especially via p53- and aryl hydrocarbon receptor-dependent pathway, is worthy to be investigated further.

#### **3.2 Alterations of DNA repair genes/activity in HNSCC and the relationship with HNSCC development, treatment, as well as patient's outcome**

GIN is a hallmark of most human malignancies including HNSCC that elevated microsatellite instability, aneuploidy and various genomic alterations have been found by genome-wide analyses (Bockmuhl et al., 1996; Brieger et al., 2003; Friedlander, 2001; Partridge et al., 1999; Sparano et al., 2006), suggesting that GIN may be involved in the development of HNSCC. Some studies also show that DNA repair activity is reduced in the peripheral blood cells of HNSCC patients when compared with normal individuals (Cheng et al., 1998; Paz-Elizur et al., 2006), implying that altered DNA repair genes and/or activity may play a critical role in the development of HNSCC.

Studies using comparative genomic hybridization (CGH) have shown that gene copy numbers at chromosome 11q22-23 (*ATM* locus) are frequently lost in HNSCC (Bockmuhl et al., 1996; Brieger et al., 2003; van den Broek et al., 2007). Lazar et al. also showed loss of heterozygosity (LOH) at 11q23 in 25% (13/52) of primary HNSCC (Lazar et al., 1998). In addition, we have reported that *ATM* mRNA is down-regulated in 81.3% (65/80) of laryngeal and pharyngeal cancers, and further show that lower *ATM* expression (tumor/normal < 0.3) was an independent risk factor for patient's survival (Lee et al., 2011). This is the first study showing that *ATM* expression is a valuable prognostic marker for HNSCC. One study also shows an absent or reduced ATM protein expression in 31.25% (10/32) of oral cancer (He et al., 2008). These results suggest that alteration of *ATM*, either in gene sequence or in expression level may be associated with HNSCC.

Previous investigations showed that LOH of chromosome 17q (*BRCA1* locus) were found in 35% to 56% of laryngeal cancer (Kiaris et al., 1995; Rizos et al., 1998). In contrast, studies using CGH found an overrepresentation of 17q in 9% to 47% of HNSCC (Bockmuhl et al., 1996; Brieger et al., 2003), and one array-CGH reported the gain of 17q21 in 33% (7/21) of oral cancer (Sparano et al., 2006). These controversial results by genome-wide analyses may be due to the physically close localization of *ERBB2* (HER-2/neu) oncogene and the results need to be clarified by specifically looking at the *BRCA1* gene locus. Regarding the expression of *BRCA1* in HNSCC, one study showed that BRCA1 immunostaining positivity was lost in 34% (26/77) of tongue cancers, which might be correlated with early-stage tumor progression (Vora et al., 2003).

The results of genome-wide studies also suggest that genetic alterations at *RAD51* (15q15.1) and *XPC* (3p25) loci may be present in HNSCC (Bockmuhl et al., 1996; Brieger et al., 2003; Partridge et al., 1999; Sparano et al., 2006; van den Broek et al., 2007). Altered RAD51 protein expression has been reported by one pilot study with twelve head and neck cancer patients (Connell et al., 2006). The patients with high RAD51 protein levels in their pre-treatment tumor biopsies demonstrate poorer cancer-specific survival rates than those with lower RAD51 levels (33.3% vs. 88.9% at 2 years; *P* = 0.025). These results suggest that RAD51 expression may influence the outcome of with head and neck cancer patients who receive chemotherapy and radiotherapy (Connell et al., 2006). Other reports regarding altered expression of DNA repair genes in HNSCC include Ku80 (Chang et al., 2006), NBN and ERCC1 (Hsu et al., 2010; Yang et al., 2006). It has been shown that ERCC1 expression is associated with cisplatin resistance (Handra-Luca et al., 2007; Hsu et al., 2010) and NBN is correlated with outcome of advanced HNSCC patients (Yang et al., 2006). Inactivation of the BRCA/FA pathway via promoter methylation has also been described in HNSCC, and may be related to tobacco and alcohol exposure and survival of these patients (Marsit et al., 2004).
