Table 2.

Logistic-regression models fitted on the training set to screen for SCC at the upper aerodigestive tract based on traditional risk factors or epigenetic markers

Univariate analysesMultivariate analyses with model selection
Risk factorsCrude OR95% CIPAdjusted OR95% CIP
Models based on the traditional risk factors
 Age1.0090.990–1.0280.357
 Male sex5.4052.515–11.618<0.001
 BMI0.9350.875–0.9980.042
 Lifestyle risk factors
  Levels of alcohol drinking1.1991.148–1.252<0.001
  Levels of betel quid chewing1.1311.080–1.185<0.001
  Levels of cigarette smoking1.1761.122–1.233<0.001
  Any one of the above10.5335.766–19.239<0.0014.1721.515–11.4880.006
 Genetic polymorphisms
  ALDH2*1/2*2 or 2*2/2*2, slow1.5630.841–2.9040.158
  ADH1B*1/*1, slow1.9040.881–4.1150.101
  ADH1C*1/*1, fast0.4370.165–1.1610.097
  GSTP1 ile/ile, slow1.9230.767–4.8170.163
  GSTM1 null, inactive0.8220.434–1.5590.549
  GSTT1 null, inactive0.9430.502–1.7730.856
 MCV1.0611.011–1.1140.0151.0601.000–1.1240.049
H. pylori infection1.5530.927–2.6010.094
 Human papillomavirus infection1.4870.598–3.6960.391
 Presence of numerous LVLs in the background mucosae14.9656.176–36.263<0.0019.3843.058–28.795<0.001
Models based on the epigenetic markers
HOXA9 methylation level (%)1.1531.101–1.209<0.0011.1411.086–1.198<0.001
NEFH methylation level (%)1.6701.280–2.179<0.0011.3391.078–1.6640.008
UCHL1 methylation level (%)1.1991.069–1.3450.002
MT1M methylation level (%)1.4741.153–1.8840.002