Skip to main content

Table 2 Sensitivity, specificity, accuracy, LR + , LR- and OR for the prediction of pDMI by the preoperative MRI markers iTFD < 6 mm, DOI ≥ 5 mm, APD ≥ 17 mm and iDMI

From: MRI-assessed tumor-free distance to serosa predicts deep myometrial invasion and poor outcome in endometrial cancer

 

iTFD < 6 mm*

DOI ≥ 5 mm*

APD ≥ 17 mm*

iDMI+

p value α

Sensitivity, % (no. of patients) [CI]^

76% (106/139) [68%-83%]

86% (119/139) [79%-91%]

78% (109/139) [71%-85%]

76% (106/139) [68%-83%]

p = 0.049θ

Specificity % (no. of patients) [CI]^

77% (162/211) [70%-82%]

49% (103/211) [42%-56%]

69% (146/211) [62%-75%]

73% (153/211) [66%-78%]

p < 0.001β

Accuracy [CI]^

77% (268/350) [72%-81%]

63% (222/350) [58%-68%]

73% (255/350) [68%-77%]

74% (259/350) [69%-79%]

p < 0.001λ

LR + 

3.28

1.67

2.54

2.77

 

LR-

0.31

0.29

0.31

0.33

 

OR (CI)[p value]

10.6 (6.4, 17.6) [< 0.001]

5.7 (3.3, 9.8) [< 0.001]

8.2 (5.0, 13.4) [< 0.001]

8.5 (5.2, 13.9) [< 0.001]

 

OR (CI) [p value], adjusting for histological risk

10.4 (6.3, 17.4) [< 0.001]

6.1 (3.5, 10.7) [< 0.001]

7.8 (4.7, 13.0) [< 0.001]

8.5 (5.2, 14.0) [< 0.001]

 

OR (CI) [p value] adjusting for histological risk and all listed MRI variables

5.8 (2.9, 11.6) [< 0.001]

2.8 (1.3, 5.7) [0.01]

2.8 (1.5, 5.2) [< 0.001]

1.1 (0.5, 2.4) [0.82]

 
  1. APD, anteroposterior tumor diameter; CI, 95% Confidence interval; DOI, depth of invasion; iDMI, deep myometrial invasion (DMI) based on imaging findings; LR + , likelihood ratio for positive results: LR +  = sensitivity/(1-specificity); LR-, likelihood ratio for negative results: LR- = (1-sensitivity)/specificity; OR, odds ratio; pDMI, DMI based on pathology findings; iTFD, tumor-free distance to serosa based on imaging findings
  2. *Optimal cutoff values for iTFD, DOI and APD based on the receiver operating characteristics analysis (Youden index) for the prediction of pDMI in hysterectomy specimen
  3. +Deep (≥ 50%) myometrial invasion based on standard imaging reading
  4. αCochrans Q-test
  5. ^Clopper–Pearson confidence interval for proportion
  6. θDOI yields significantly higher sensitivity than iDMI (p = 0.02, McNemar test)
  7. βiTFD yields significantly higher specificity than DOI (p < 0.001) and APD (p = 0.045, McNemar test)
  8. λiTFD yields significantly higher accuracy than DOI (p < 0.001, McNemar test)
  9. Binary logistic regression analysis
  10. High-risk histological subtype (endometrioid grade 3 or non-endometrioid) versus low-risk histological subtype (endometrioid grade 1–2) based on preoperative endometrial biopsy/curettage
  11. All significant p values are given in boldface