Table 2.
Cox regression models for the associations between tumour necrosis percentage and survival in Cohorts 1 and 2.
| Cancer-specific survival | Overall survival | ||||||
|---|---|---|---|---|---|---|---|
| Tumour necrosis percentage | No. of cases | No. of events | Univariable HR (95% CI) | Multivariable HR (95% CI) | No. of events | Univariable HR (95% CI) | Multivariable HR (95% CI) |
| Cohort 1 | |||||||
| <3% | 100 | 15 | 1 (referent) | 1 (referent) | 42 | 1 (referent) | 1 (referent) |
| 3–9.9% | 577 | 145 | 1.73 (1.01–2.94) | 1.76 (1.02–3.04) | 278 | 1.19 (0.87–1.65) | 1.19 (0.85–1.66) |
| 10–39.9% | 327 | 107 | 2.45 (1.43–4.20) | 2.35 (1.34–4.11) | 173 | 1.41 (1.01–1.97) | 1.44 (1.02–2.04) |
| ≥40% | 59 | 28 | 4.24 (2.27–7.94) | 3.22 (1.68–6.17) | 37 | 1.96 (1.26–3.05) | 1.88 (1.19–2.97) |
| Ptrend | <0.0001 | <0.0001 | 0.0010 | 0.0011 | |||
| Cohort 2 | |||||||
| <3% | 61 | 7 | 1 (referent) | 1 (referent) | 22 | 1 (referent) | 1 (referent) |
| 3–9.9% | 105 | 20 | 1.67 (0.71–3.94) | 1.80 (0.71–4.60) | 43 | 1.13 (0.68–1.89) | 1.08 (0.62–1.88) |
| 10–39.9% | 87 | 32 | 3.70 (1.63–8.39) | 1.85 (0.75–4.55) | 47 | 1.78 (1.07–2.94) | 1.10 (0.62–1.96) |
| ≥40% | 31 | 17 | 7.53 (3.12–18.18) | 3.39 (1.28–8.96) | 21 | 3.14 (1.72–5.71) | 1.70 (0.86–3.36) |
| Ptrend | <0.0001 | 0.018 | <0.0001 | 0.19 | |||
HR hazard ratio, CI confidence interval.
Multivariable Cox regression models were adjusted for age (<65, 65–75, >75), sex (male, female), T (1–2, 3–4), N (0, 1–2), M (0, 1), tumour location (proximal colon, distal colon, rectum), year of operation (Cohort 1: 2000–2005, 2006–2010, 2011–2015; Cohort 2: 2006- Jan. 2010, Feb. 2010–2014), lymphatic or venous invasion (no, yes), grade (low-grade, high grade), MMR status (proficient, deficient), and BRAF status (wild-type, mutant). We excluded patients who died 30 days or less after having surgery (N = 37, in Cohort 1 and N = 3 in Cohort 2). Ptrend values were calculated by using the four ordinal categories of tumour necrosis percentage as a continuous variable in univariable and multivariable Cox proportional hazard regression models.