Skip to main content

Table 1 Patient characteristics

From: A whole slide image-based machine learning approach to predict ductal carcinoma in situ (DCIS) recurrence risk

Clinicopathologic characteristics of patients in the training and validation cohorts

Baseline characteristic

Training cohort (N = 159)

Validation cohort (N = 185)

Difference (p value)

Patient age

 Median age (range), years

57 (30–83)

59 (36–77)

0.30

 Age < 50, n (%)

26 (16.3)

23 (12.4)

 Age ≥ 50, n (%)

133 (83.7)

162 (87.6)

Menopausal status, n (%)

 Pre

31 (19.5)

29 (15.7)

0.35

 Post

128 (80.5)

156 (84.3)

Presentation, n (%)

 Screening

85 (53.5)

120 (64.9)

0.03

 Symptomatic

74 (46.5)

65 (35.1)

Comedo necrosis, n (%)

 No

60 (37.7)

34 (18.4)

< .0001

 Yes

99 (62.3)

151 (81.6)

Radiation, n (%)

 No

117 (73.6)

145 (78.4)

0.30

 Yes

42 (26.4)

40 (21.6)

Grade, n (%)

 1

25 (15.8)

0 (0.0)

< .0001

 2

24 (15.2)

0 (0.0)

 3

109 (69.0)

185 (100.0)

Margins, n (%)

 Negative

154 (97.5)

183 (98.9)

0.31

 Positive

4 (2.5)

2 (1.1)

Tumor size

 Median tumor size (range), cm

1.7 (0.1–14.5)

1.7 (0.2–12.0)

0.74

 Size < 2.0, n (%)

88 (56.4)

101 (55.6)

 Size ≥ 2.5, n (%)

68 (43.6)

84 (45.4)

Survival status, n (%)

 Alive

109 (68.6)

159 (86.0)

0.00

 Dead

50 (31.4)

26 (14.0)

10-year recurrence status, n (%)

 Recurrence free

128 (80.5)

159 (85.9)

0.18

 Recurred

31 (19.5)

26 (14.1)

  1. Descriptive data detailing the training and validation cohort’s clinicopathological variables. The cutoff point for positive margins was 2 mm. In the training cohort, the tumor size of 3 cases was not known and a patient has missing data for margin status and grade. The proportional difference of clinicopathological variables are measured with the chi-square test