Table 3. Overall survival analysis.
| Variable | Univariate Analysis | Stage-Adjusted Analysis | ||
| HR (95% CI) | P valuea | HR (95% CI) | P valueb | |
| Age at diagnosis (per year increase) | 1.03 (1.01–1.06) | 0.011 | 1.02 (0.99–1.05) | 0.191 |
| Ethnic group (Malays vs Chinese) | 0.39 (0.12–1.27) | 0.127 | 0.68 (0.20–2.33) | 0.943 |
| Ethnic group (Indians vs Chinese) | 0.00 (NE) | 0.00 (NE) | ||
| Ethnic group (Others vs Chinese) | 0.00 (NE) | 0.00 (NE) | ||
| Smoking history (Ever-smoker vs non-smoker) | 0.98 (0.20–4.91) | 0.980 | 0.47 (0.06–3.45)c | 0.458 |
| Family history of cancer (Positive vs Negative) | 5.88 (1.40–24.79) | 0.006 | 7.95 (1.30–48.65)c | 0.025 |
| Presence of comorbidities (Yes vs No) | 1.16 (0.55–2.42) | 0.694 | 0.75 (0.34–1.65) | 0.475 |
| Stage at diagnosis (II vs I) | 4.85 (1.57–14.97) | <0.001 | - | - |
| Stage at diagnosis (III vs I) | 15.42 (6.54–36.36) | - | ||
| Stage at diagnosis (IV vs I) | 28.51 (7.38–110.22) | - | ||
| Tumour differentiation/grade (moderate vs well) | 1.15 (0.48–2.78) | 0.093 | 1.15 (0.47–2.82) | 0.848 |
| Tumour differentiation/grade (poor vs well) | 2.77 (1.05–7.29) | 1.35 (0.46–3.95) | ||
| Tumour type (Non-borderline vs mixed borderline) | 1.75 (0.81–3.80) | 0.153 | 1.33 (0.58–3.07) | 0.500 |
| Ovarian surface involvement (Yes vs No) | 7.80 (3.14–19.35) | <0.001 | 4.14 (1.45–11.80) | 0.008 |
| Lymphovascular invasion (Yes vs No) | 10.25 (4.39–23.92) | <0.001 | 5.58 (2.05–15.21) | 0.001 |
| CA125 (per U/mL increase) | 1.00 (1.00–1.00) | 0.916 | 1.00 (1.00–1.00) | 0.461 |
| Received chemotherapy (Yes vs No) | 4.26 (1.83–9.89) | <0.001 | 1.66 (0.60–4.60)d | 0.335 |
| Received adjuvant chemotherapy (Yes vs No) | 1.18 (0.58–2.41) | 0.646 | 1.01 (0.43–2.38) | 0.974 |
| HER2 status (HER2− vs HER2+) | 1.79 (0.66–4.85) | 0.249 | 1.01 (0.34–2.97) | 0.988 |
| Among HER2+ patients: HER2 amplification ratio (per unit increase) | 0.75 (0.49–1.15) | 0.185 | 0.72 (0.44–1.18)c | 0.193 |
Abbreviation: HR, hazard ratio; CI, confidence interval; NE, not estimable.
a
P values for age at diagnosis, CA125 and HER2 amplification ratio were based on Wald test, and P values for all other variables were based on the log-rank test.
b
Based on Wald test.
c
To interpret with caution as there were <10 deaths in the fitted multivariable model.
d
Departures from proportionality assumption. The time-varying effects of receipt of chemotherapy were further accounted for by including a time-by-covariate interaction term in the Cox model. Based on the extended model, there was no significant association between OS and chemotherapy.