3.1. Tobacco smoking and its interaction in the causation of HCC
The distribution of factors including demographic characteristics of chronic HBV infection, HCV infection and serological evidence between HCC cases and hospital controls, as well as smoking status and alcohol consumption, was showed in Supplementary Table 1. Cases and controls were similarly distributed by age, gender and body mass index. There were, as expected, significant differences (p < 0.05) in gender status and prevalence of HBV and HCV infection markers, because these are the main reasons for HCC. Also, it was interestingly found that smokers were significantly more likely to the causation of HCC than non-smokers (χ2 = 67.131, p = 0.000) and smoking was more common among cases than among controls.
We examined the cigarette smoking-HCC association. Table 1 showed the distribution of HCC cases and hospital controls by smoking habits. The distributions were gender-specific, owing to the expected differences about these activities in terms of prevalence by gender. Smokers were significantly associated with HCC. The corresponding estimates were 3.500 (95% CI = 2.518–4.866) and 2.962 (95% CI = 2.014–4.355) for current smokers and former smokers. In terms of dose, there appeared to be a dose-response relationship between HCC and quantity smoked with adjusted OR estimates of 3.267 and 4.375 for smoking < 2 packs per day and ≥ 2 packs per day for current smoker, respectively. Likewise, significant OR variation was found when considering only a group of subjects among former smoker, the ORs for < 2 packs per day and ≥ 2 packs per day being 2.450 (95% CI = 1.572–3.817) and 4.667 (95% CI = 2.451–8.885), respectively. Altogether, it demonstrated that smoking was associated with an increased risk of HCC according to the adjustment of age and gender, but mutual confounding precludes alcohol intake and valid etiological inferences.
Table 1
Gender-specific distribution of HCC cases and hospital controls by smoking habit, with OR and 95%CI
| HCC cases (n = 300) | | Hospital controls (n = 612) | OR (95%CI) |
Men (n = 198) | Women (n = 102) | Men (n = 468) | Women (n = 144) |
Smoking status (packs/day) | | | |
Never smoker (%) | 54 (27.27) | 66 (64.70) | 288 (61.54) | 132 (91.67) | 1.000 |
Current smoker (%) | 90 (45.45) | 24 (23.53) | 108 (23.08) | 6 (4.17) | 3.500 (2.518–4.866) |
< 2 | 66 (33.33) | 18 (17.65) | 84 (17.95) | 6 (4.17) | 3.267 (2.279–4.682) |
≥ 2 | 24 (12.12) | 6 (5.88) | 24 (5.13) | 0 (0.00) | 4.375 (2.465–7.766) |
Former smoker (%) | 54 (27.27) | 12 (11.76) | 72 (15.39) | 6 (4.17) | 2.962 (2.014–4.355) |
< 2 | 36 (18.18) | 6 (5.88) | 54 (11.54) | 6 (4.17) | 2.450 (1.572–3.817) |
≥ 2 | 18 (9.09) | 6(5.88) | 18 (3.85) | 0 (0.00) | 4.667 (2.451–8.885) |
We further explored the independent and combined effects of smoking and alcohol consumption on HCC risk stratified by age, sex, and body mass index. The distribution of cases and controls by smoking status and the adjusted risks of HCC in relation to smoking status by alcohol intake are shown in Table 2. Compared with never smokers, current and former smokers both had an increased HCC risk of significance. Interestingly, subjects with alcohol drinker had a significantly increased risk. Moreover, stratification revealed that subjects without alcohol drinker had a significant dose-response increase in HCC disease risk. The overall risks of HCC were significantly increased among current smokers smoking < 2 packs per day (OR = 2.800, 95% CI = 1.933–4.055) between HCC patients and controls, and these risks were of similar magnitude among subjects without alcohol drinker, with an estimated effect much higher than that for never smokers (OR = 2.895, 95% CI = 1.640–5.110). Compared with current smokers, former smokers had significant increases in the risk of HCC overall (< 2 packs per day, OR = 3.150, 95% CI = 2.070–4.794) and by subtype of non-alcohol drinking (< 2 packs per day, OR = 2.316, 95% CI = 1.341–4.001).
Table 2
Multiple regression-derived mutually adjusted OR (95%CI) for the association of HCC with smoking habit, by alcohol intake *
| All subjects | | Subjects without alcohol drinker | | Subjects with alcohol drinker |
Cases (n = 300) | Controls (n = 612) | OR (95%CI) | Cases (n = 160) | Controls (n = 306) | OR (95%CI) | Cases (n = 140) | Controls (n = 306) | OR (95%CI) |
Smoking status (packs/day) | | | | | |
Never smoker (%) | 120 (40.00) | 420 (68.63) | 1.000 | 68 (42.50) | 210 (68.63) | 1.000 | 52 (37.15) | 210 (68.63) | 1.000 |
Current smoker (%) | 114 (38.00) | 126 (20.59) | 3.167 (2.289–4.381) | 50 (31.25) | 50 (16.34) | 3.088 (1.915–4.981) | 64 (45.71) | 76 (24.84) | 3.401 (2.168–5.335) |
< 2 | 72 (24.00) | 90 (14.71) | 2.800 (1.933–4.055) | 30 (18.75) | 32 (10.46) | 2.895 (1.640–5.110) | 42 (30.00) | 58 (18.96) | 2.924 (1.774–4.821) |
≥ 2 | 42 (14.00) | 36 (5.89) | 4.083 (2.504–6.660) | 20 (12.50) | 18 (5.88) | 3.431 (1.716–6.862) | 22 (15.72) | 18 (5.88) | 4.936 (2.468–9.870) |
Former smoker (%) | 66 (22.00) | 66 (10.78) | 3.500 (2.353–5.205) | | 42 (26.25) | 46 (15.03) | 2.820 (1.711–4.648) | | 24 (17.14) | 20 (6.53) | 4.846 (2.488–9.438) |
< 2 | 54 (18.00) | 60 (9.80) | 3.150 (2.070–4.794) | | 30 (18.75) | 40 (13.07) | 2.316 (1.341–4.001) | | 24 (17.14) | 20 (6.53) | 4.846 (2.488–9.438) |
≥ 2 | 12 (4.00) | 6 (0.98) | 7.000 (2.573–19.042) | | 12 (7.50) | 6 (1.96) | 6.176 (2.233–17.084) | | 0 (0.00) | 0 (0.00) | N/A** |
* Also controlling for age, gender, body mass index and alcohol intake status, as appropriate |
** The multivariate model did not converge because there were no former smokers with ≥ 2 packs per day in controls and cases. |
Besides, it showed multiple regression-derived, mutually adjusted OR (and 95% CI) for HCC association with cigarette consumption patterns in Table 3. Because the latency of a possible smoking associated with HCC had not yet been determined, we chose to study the impact of daily smoking, regardless of the current or previous smoking status. Data were shown for all subjects, and also were stratified according to serological evidence of HBV and/or HCV infection. Compared with never smokers, the relative risk of HCC for ever-smokers was estimated to be 3.281 (95% CI = 2.462–4.373). Risk was also higher among HBsAg and anti-HCV negative participants, among whom the relative risk of HCC was 2.145 (95% CI = 1.415–3.251) compared with never smokers. Also, it indicated a dose-response relationship with number of cigarettes smoked per day between smoking and HCC overall risk (< 2 packs per day, OR = 2.940, 95% CI = 2.152–4.016; ≥2 packs per day, OR = 4.500, 95% CI = 2.865–7.067), which was positively correlated. Besides, as far as smoking was concerned, the association with HCC risk was stronger in individuals without chronic viral infection than those never smokers (< 2 packs per day, OR = 1.727, 95% CI = 1.090–2.738; ≥2 packs per day, OR = 3.855, 95% CI = 2.021–7.351). These data convincingly suggested a super-multiplicative, interactive effect of smoking in the development of HCC in a statistically significant, dose-dependent manner.
Table 3
Multiple regression-derived mutually adjusted OR (95%CI) for the association of HCC with smoking habit, by HBsAg and/or anti-HCV status*
| All subjects | | Subjects without both HBsAg and anti-HCV | | Subjects with HBsAg and/or anti-HCV |
Cases (n = 300) | Controls (n = 612) | OR (95%CI) | Cases (n = 128) | Controls (n = 324) | OR (95%CI) | Cases (n = 172) | Controls (n = 288) | OR (95%CI) |
Smoking (packs per day) | | | | | |
Never smokers (%) | 120 (40.00) | 420 (68.63) | 1.000 | 60 (46.87) | 212 (65.43) | 1.000 | 60 (34.88) | 208 (72.22) | 1.000 |
Ever smokers (%) | 180 (60.00) | 192 (31.37) | 3.281 (2.462–4.373) | 68 (53.13) | 112 (34.57) | 2.145 (1.415–3.251) | 112 (65.12) | 80 (27.78) | 4.853 (3.234–7.284) |
< 2 | 126 (42.00) | 150 (24.51) | 2.940 (2.152–4.016) | 44 (34.38) | 90 (27.78) | 1.727 (1.090–2.738) | 82 (47.68) | 60(20.83) | 4.738 (3.052–7.354) |
≥ 2 | 54 (18.00) | 42 (6.86) | 4.500 (2.865–7.067) | 24 (18.75) | 22 (6.79) | 3.855 (2.021–7.351) | 30 (17.44) | 20 (6.95) | 5.200 (2.757–9.808) |
* Also controlling for age, gender, body mass index and HBsAg and/or anti-HCV status, as appropriate |
3.2. FASAY for detection of p53 mutations
p53 gene mutation is one of the most common genetic changes in human cancer. The advantage of these mutations (95%) is missense mutations, which are mainly located in the DNA binding domain (amino acids 94–292), with hot spots at codons R175, G245, R248, R249, R273 and R282 [25, 26]. FASAY was proved to reach the highest sensitivity and specificity of p53 mutation detection [27, 28], due to the fact that a larger region of p53 gene in exon 4–10 was tested. Then mutations can be detected in a large number of normal tissues, because there are hundreds of clones in each sample, and simple red and white readouts mean mutations are not easy to ignore [29]. In addition, FASAY can exclude p53 polymorphisms that do not affect their transcriptional activity [30]. To detect the status of p53 in HCC tumors, the tumor mucosa of 300 patients was FASAY examined (Supplementary Table 2). In total, 228 tumors (76.00%) were positive (mutant) by FASAY. Besides, analysis of plasmid sequencing from red yeast colony showed that 210 of 228 plasmids were point missense mutations, and 6 of 228 was frameshift deletion mutation, as well as 9 of 228 were in-frameshift deletion mutations, and 3 of 228 were splicing mutations (Fig. 1A). Moreover, hotspot mutation p53-RS that occurred 126 times in this study, accounting for 55.26% in total mutations. We further evaluated the relation between tobacco smoking and p53-RS mutation. A striking increase in the risk of p53-RS was observed for smokers (Fig. 1B). Furthermore, these differences about p53-RS between smokers and nonsmokers were statistically significant by Chi-square test (56.67% versus 20.00%; χ2 = 38.246, p = 0.000). These results suggested that p53-RS mutation might be associated with the response to cigarette smoking. We also sought to determine whether the presence of p53-RS has a measurable effect on the expression of p53 major transcriptional targets (p21 and Bax), as well as p53-RS-dependent target genes including c-Myc [31] and STAT1 [32] by using qRT-PCR on mRNA from fresh frozen components. For p21 and Bax, we found a decreased profile in p53-RS group compared with the WT p53 group (Figs. 1C and 1D). However, in contrast to WT p53 group, we observed an increased expression about c-Myc and STAT1 in the tumors with p53-RS, showing a loss of transcriptional p53 function and p53-RS gain-of-function (Figs. 1E and 1F).