Table 3 Results of Cox proportional hazards regression model, with death and recurrence as the response variables

From: p53 mutation, deprivation and poor prognosis in primary breast cancer

 

Death

Recurrence

 

β

P

RR (95% CI)

β

P

RR (95% CI)

Deprivation 10 (vs 1–9)

1.81

0.0003

6.1 (2.3–16.2)

1.31

0.003

3.7 (1.6–8.9)

Tumour size (large vs small)

1.22

0.003

3.4 (1.5–7.5)

0.83

0.012

2.3 (1.2–4.4)

ER (+ vs −)

−2.50

0.041

0.1 (<0.1–0.9)

−1.28

0.056

0.3 (0.1–1.0)

p53 mutation (vs WT)

0.67

0.114

1.9 (0.9–4.5)

0.95

0.009

2.6 (1.3–5.2)

Tumour grade (3 vs 1–2)

0.90

0.087

2.5 (0.9–6.9)

0.18

0.677

1.2 (0.5–2.7)

Nodal status (+ vs −)

1.32

0.005

3.7 (1.5–9.5)

0.94

0.007

2.6 (1.3–5.1)

HER2 status (+ vs −)

−1.96

0.121

0.1 (<0.1–1.7)

−0.58

0.383

0.6 (0.2–2.1)

Triple negative (n vs y)

−1.54

0.229

0.2 (<0.1–2.6)

−0.74

0.325

0.5 (0.1–2.1)

Dep 10 & p53 m (v rest)

2.52

<0.0001

12.4 (4.8–32.3)

2.18

<0.0001

8.8 (3.6–21.7)

Tumour size (large vs small)

1.26

0.002

3.5 (1.6–7.7)

0.90

0.005

2.5 (1.3–4.7)

ER (+ vs −)

−2.65

0.021

0.1 (<0.1–0.7)

−1.47

0.023

0.2 (0.1–0.8)

Tumour grade (3 vs 1–2)

1.07

0.033

2.9 (1.1–7.8)

0.35

0.369

1.4 (0.7–3.0)

Nodal status (+ vs −)

1.49

0.002

4.4 (1.8–11.2)

1.05

0.002

2.9 (1.5–5.6)

HER2 status (+ vs −)

−1.57

0.184

0.2 (<0.1–2.1)

−0.23

0.718

0.8 (0.2–2.8)

Triple negative (n vs y)

−1.55

0.199

0.2 (<0.1–2.3)

−0.77

0.296

0.5 (0.1–2.0)

  1. Abbreviations: ER=oestrogen receptor; RR=risk ratio; WT=wild type.
  2. The upper models have deprivation and p53 mutation status as individual binary input variables, whereas the lower models group deprivation category 10 patients that also have a p53 mutation (versus all other patients) as a binary input variable. The G-statistic χ2⩾51.0 and P<0.0001 in each model, with the upper models having eight degrees of freedom and the lower models having seven. A strong predictor of the response variable (death or recurrence; left and right half of the table, respectively) is one that has a large β-coefficient, whether positive or negative, and a high degree of confidence in the coefficient, P⩽0.05. Such strong predictors are also backed by RR values with associated 95% confidence intervals that exclude unity. Indicators of strong predictors are highlighted in bold type.
Back to article page