Table 4 Summary of overall and stratified meta-analysis results

From: Association of p53 codon72 Arg>Pro polymorphism with susceptibility to nasopharyngeal carcinoma: evidence from a case–control study and meta-analysis

Comparisons

Heterogeneity

Model

Forest plot analysis

Egger’s regression analysis

 

Q-value

PH

I2(%)

 

OR

95% CI

POR

Intercept

95% CI

P-value

Overall studies

 Pro vs Arg

15.98

0.06

43.69

Fixed

1.28

1.17–1.40

<0.001

0.69

−1.53 to 2.91

0.49

 ProPro vs ArgArg

16.23

0.06

44.56

Fixed

1.70

1.41–2.04

<0.001

0.75

−1.45 to 2.96

0.45

 ArgPro vs ArgArg

14.95

0.09

39.80

Fixed

1.24

1.07–1.43

0.004

−0.02

−2.33 to 2.28

0.98

 ProPro+ArgPro vs ArgArg

13.55

0.13

33.60

Fixed

1.35

1.17–1.55

<0.001

0.29

−1.94 to 2.54

0.76

 ProPro vs ArgArg+ArgPro

19.05

0.02

52.76

Random

1.54

1.18–2.01

0.002

1.005

−1.20 to 3.21

0.32

Asian studies

 Pro vs Arg

15.78

0.02

55.65

Random

1.30

1.10–1.54

0.002

0.68

−2.24 to 3.60

0.58

 ProPro vs ArgArg

13.94

0.05

49.79

Fixed

1.63

1.34–1.97

<0.001

0.37

−2.50 to 3.24

0.76

 ArgPro vs ArgArg

11.40

0.12

38.62

Fixed

1.31

1.11–1.53

0.001

0.38

−2.15 to 2.93

0.72

 ProPro+ArgPro vs ArgArg

12.18

0.09

42.53

Fixed

1.39

1.20–1.62

<0.001

0.58

−2.07 to 3.24

0.61

 ProPro vs ArgArg+ArgPro

13.40

0.06

47.79

Fixed

1.35

1.15–1.59

<0.001

0.32

−2.42 to 3.06

0.78

Caucasian studies

 Pro vs Arg

0.02

0.87

<0.001

Fixed

1.35

1.03–1.77

0.02

—

—

—

 ProPro vs ArgArg

0.005

0.94

<0.001

Fixed

2.71

1.44–5.09

0.002

—

—

—

 ArgPro vs ArgArg

0.36

0.54

<0.001

Fixed

0.89

0.61–1.31

0.58

—

—

—

 ProPro+ArgPro vs ArgArg

0.22

0.63

<0.001

Fixed

1.13

0.79–1.61

0.50

—

—

—

 ProPro vs ArgArg+ArgPro

0.08

0.77

<0.001

Fixed

2.88

1.57–5.29

0.001

—

—

—

Population-based studies

 Pro vs Arg

10.42

0.16

32.82

Fixed

1.24

1.13–1.37

<0.001

−0.06

−2.71 to 2.58

0.95

 ProPro vs ArgArg

12.99

0.07

46.11

Fixed

1.62

1.34–1.97

<0.001

0.27

−2.56 to 3.11

0.82

 ArgPro vs ArgArg

5.42

0.60

<0.001

Fixed

1.21

1.04–1.41

0.01

−0.57

−2.52 to 1.38

0.49

 ProPro+ArgPro vs ArgArg

5.91

0.55

<0.001

Fixed

1.31

1.13–1.51

<0.001

−0.43

−2.54 to 1.66

0.62

 ProPro vs ArgArg+ArgPro

16.43

0.02

57.39

Random

1.46

1.09–1.96

0.01

0.69

−2.26 to 3.65

0.58

Hospital-based studies

 Pro vs Arg

0.90

0.34

<0.001

Fixed

1.88

1.31–2.69

0.001

—

—

—

 ProPro vs ArgArg

0.23

0.63

<0.001

Fixed

3.02

1.54–5.94

0.001

—

—

—

 ArgPro vs ArgArg

8.35

0.004

88.02

Random

1.65

0.28–9.48

0.57

—

—

—

 ProPro+ArgPro vs ArgArg

5.06

0.02

80.25

Random

2.04

0.61–6.83

0.24

—

—

—

 ProPro vs ArgArg+ArgPro

0.93

0.33

<0.001

Fixed

2.07

1.15–3.73

0.01

—

—

—

Chinese studies

 Pro vs Arg

10.16

0.07

50.80

Fixed

1.23

1.11–1.36

<0.001

−0.17

−4.00 to 3.64

0.90

 ProPro vs ArgArg

10.66

0.05

53.10

Fixed

1.54

1.25–1.89

<0.001

−0.31

−4.26 to 3.64

0.83

 ArgPro vs ArgArg

3.76

0.58

<0.001

Fixed

1.23

1.04–1.45

0.015

−0.80

−2.89 to 1.29

0.34

 ProPro+ArgPro vs ArgArg

4.56

0.47

<0.001

Fixed

1.31

1.12–1.54

0.001

−0.55

−3.11 to 2.00

0.58

 ProPro vs ArgArg+ArgPro

13.03

0.02

61.63

Random

1.36

0.97–1.90

0.07

0.14

−4.10 to 4.39

0.92

  1. Abbreviations: 95% CI, 95% confidence intervals; Fixed, fixed-effect model; OR, odds ratio; PH, P-vaue of heterogeneity analysis. Meta-analysis was performed with comprehensive meta-analysis V2 software in overall studies, Asian, Caucasian, population-based, hospital-based and Chinese studies. Association of p53 codon72 Arg>Pro polymorphisms with NPC was assessed by the estimation of the combined odds ratio (OR), P-value and 95% confidence interval (CI) in five different models: (i) allele contrast (Pro vs Arg), (ii) homozygous comparison (ProPro vs ArgArg), (iii) heterozygous comparison (ArgPro vs ArgArg), (iv) dominant (ProPro+ArgPro vs ArgArg) and (v) recessive (ProPro vs ArgArg+ProPro) model. Heterogeneity between studies was calculated using Cochran’s Q-statistic and I2 values as described earlier.53, 54 Based on heterogeneity or homogeneity among the included studies, the random (Der Simonian and Laird method) or fixed (Mantel–Haenszel's method) model was used to calculate combined OR and 95% CI. Publication bias was assessed from Egger’s regression analysis.
Back to article page