Table 3 Contingency tables for predicting thiopurine intolerance (DIP < 25%) of two-, three-, and four-gene models in pediatric patients with ALL (N = 320).

From: Interplay between IL6 and CRIM1 in thiopurine intolerance due to hematological toxicity in leukemic patients with wild-type NUDT15 and TPMT

(a)

(c)

(e)

STARNUDT15, TPMT

DIP (%)

Total

GVBNUDT15, TPMT, IL6

DIP (%)

Total

GVBNUDT15, TPMT, CRIM1^(r13306435, rs3821169*)

DIP (%)

Total

 ≤ 25

 > 25

 

 ≤ 25

 > 25

 ≤ 25

 > 25

PM + IM

23

57

80

 ≤ 0.3

23

51

74

 ≤ 0.3

31

64

95

NM

21

219

240

 > 0.3

21

225

246

 > 0.3

13

212

225

Total

44

276

320

Total

44

276

320

Total

44

276

320

(b)

(d)

(f)

GVBNUDT15, TPMT

DIP (%)

Total

GVBNUDT15, TPMT, IL6, CRIM1*

DIP (%)

Total

GVBNUDT15, TPMT, IL6, CRIM1^(rs13306435, rs3821169*)

DIP (%)

Total

 ≤ 25

 > 25

 ≤ 25

 > 25

 ≤ 25

 > 25

 ≤ 0.3

23

53

76

 ≤ 0.3

28

60

88

 ≤ 0.3

31

63

94

 > 0.3

21

223

244

 > 0.3

16

216

232

 > 0.3

13

213

226

Total

44

276

320

Total

44

276

320

Total

44

276

320

(g)

Prediction models

Sensitivity (%)

Specificity (%)

PPV (%)

NPV (%)

OR

Relative risk

PAF

NNT

NNG

  

(a) STARNUDT15, TPMT

52.27

79.35

28.75

91.25

4.21

3.286

0.364

5.000

20.000

  

(b) GVBNUDT15, TPMT

52.27

80.80

30.26

91.39

4.61

3.516

0.374

4.618

19.442

  

(c) GVBNUDT15, TPMT, IL6

52.27

81.52

31.08

91.46

4.83

3.641

0.379

4.436

19.181

  

(d) GVBNUDT15, TPMT, CRIM1*

64.64

78.26

31.82

93.10

6.30

4.614

0.498

4.013

14.591

  

(e) GVBNUDT15, TPMT, CRIM1^(r13306435, rs3821169*)

70.46

76.81

32.63

94.22

7.90

5.648

0.580

3.724

12.544

  

(f) GVBNUDT15, TPMT, IL6, CRIM1^(rs13306435, rs3821169*)

70.46

77.17

32.98

94.25

8.06

5.733

0.582

3.673

12.503

  
  1. ALL acute lymphoblastic leukemia, DIP (%) the last-cycle 6-MP dose intensity percentage, PAF population attributable fraction, PPV positive predictive value, NNT number needed to treat, NNG number needed to genotype, NPV negative predictive value, OR odds ratio, M poor metabolizer, IM intermediate metabolizer, STARNUDT15, TPMT classical star (*) allele-based haplotyping of NUDT15 and TPMT genes according to the CPIC guideline, GVB gene-wise variant burden, GVBCRIM1^(rs13306435, rs3821169*), GVB of CRIM1 dependent on IL6 rs13306435 or CRIM1 rs3821169 homozygote, GVB cutoff value of 0.3 was selected as maximized Youden’s index.
  2. Bold number means True Positive (TP) group, True Negative (TN) group and patients with thiopurine intolerance(DIP<25%). For example, In Table 3a, 23 means TP group and 219 means TN group. 44 indicates that number of patients with thiopurine intolerance(DIP<25%) is 44. Additionally, the most strongest statistical parameter numbers related to clinical validity at the bottom of the table was highlighted in bold. For example, prediction model (g) and (e) showed the most desirable Sensitivity(70.46%) among all prediction models. In a similar aspect, model (f) showed the best effectiveness in number needed to trat (NNT, 3.673) and number needed to genotype (NNG, 12.503).
Back to article page