Prognostic factors for OS
Univariate analysis of the training cohort showed that age (P < 0.001, HR = 1.033, 95% CI = 1.023–1.044), SEER stage (regional, P < 0.001, HR = 1.462, 95% CI = 1.177–1.817; distant, P < 0.001, HR = 3.153, 95% CI = 2.318–4.288), T stage (T2, P < 0.001, HR = 0.965, 95% CI = 0.771–1.207; T3, P < 0.001, HR = 1.600, 95% CI = 1.045–2.449; T4, P < 0.001, HR = 2.440, 95% CI = 1.754–3.394; Tx, P < 0.001, HR = 6.557, 95% CI = 0.910-47.217), N stage (N1, P < 0.001, HR = 1.770, 95% CI = 1.347–2.326; N2, P < 0.001, HR = 1.960, 95% CI = 1.478-2.600; N3, P < 0.001, HR = 4.830, 95% CI = 1.795–12.998; Nx, P < 0.001, HR = 0.780, 95% CI = 0.109–5.583), M stage (M1, P < 0.001, HR = 2.135, 95% CI = 1.543–2.956; Mx, P < 0.001, HR = 13.548, 95% CI = 1.871–98.088), tumor size (> 50, P = 0.013, HR = 1.453, 95% CI = 1.081–1.953; Unknown, P = 0.010, HR = 3.240, 95% CI = 1.313–7.997), and radiation (P = 0.001, HR = 1.516, 95% CI = 1.179–1.950) were prognostic factors, as shown in Table 2. Multivariate analysis showed that age (P < 0.001, HR = 1.034, 95% CI = 1.023–1.046), SEER stage (distant, P = 0.007, HR = 2.700, 95% CI = 1.296–5.624), T stage (T2, P = 0.047, HR = 0.726, 95% CI = 0.528–0.997), N stage (N1, P < 0.001, HR = 2.024, 95% CI = 1.442–2.840; N2, P < 0.001, HR = 2.228, 95% CI = 1.557–3.189), M stage (Mx, P = 0.023, HR = 11.076, 95% CI = 1.380-88.911), tumor size (20–49, P = 0.039, HR = 1.342, 95% CI = 1.013–1.778; >50, P = 0.004, HR = 1.752, 95% CI = 1.190–2.580), and chemotherapy (P < 0.001, HR = 0.549, 95% CI = 0.423–0.713) were independent prognostic factors (Table 3).
Table 2
Univariate Cox regression analysis for overall survival in training cohort
Characteristics | Univariate analysis |
HR (95% CI) | P value |
Age | 1.033 (1.023–1.044) | < 0.001 |
Race | | |
White | Reference | |
Black | 1.199 (0.872–1.649) | 0.263 |
Other | 1.197 (0.735–1.950) | 0.469 |
Sex | | |
Male | Reference | |
Female | 0.859 (0.703–1.050) | 0.138 |
Marital status | | |
Married | Reference | |
Unmarried | 0.991 (0.803–1.223) | 0.934 |
Unknown | 0.989 (0.586–1.670) | 0.968 |
Year of diagnosis | | |
2004–2009 | Reference | |
2010–2015 | 1.105 (0.896–1.364) | 0.350 |
Laterality | | |
Left | Reference | |
Right | 0.986 (0.806–1.206) | 0.890 |
Bilateral | - | - |
Grade | | |
Ⅰ | Reference | |
Ⅱ | 1.220 (0.357–4.170) | 0.750 |
Ⅲ | 1.092 (0.350–3.409) | 0.879 |
Ⅳ | 1.117 (0.353–3.534) | 0.850 |
SEER stage | | |
Localized | Reference | |
Regional | 1.462 (1.177–1.817) | < 0.001 |
Distant | 3.153 (2.318–4.288) | < 0.001 |
T | | |
T1 | Reference | |
T2 | 0.965 (0.771–1.207) | < 0.001 |
T3 | 1.600 (1.045–2.449) | < 0.001 |
T4 | 2.440 (1.754–3.394) | < 0.001 |
Tx | 6.557 (0.910-47.217) | < 0.001 |
N | | |
N0 | Reference | |
N1 | 1.770 (1.347–2.326) | < 0.001 |
N2 | 1.960 (1.478-2.600) | < 0.001 |
N3 | 4.830 (1.795–12.998) | < 0.001 |
Nx | 0.780 (0.109–5.583) | < 0.001 |
M | | |
M0 | Reference | |
M1 | 2.135 (1.543–2.956) | < 0.001 |
Mx | 13.548 (1.871–98.088) | < 0.001 |
Tumor size (mm) | | |
< 20 | Reference | |
20–49 | 1.203 (0.936–1.547) | 0.149 |
> 50 | 1.453 (1.081–1.953) | 0.013 |
Unknown | 3.240 (1.313–7.997) | 0.010 |
Radiation | | |
No/Unknown | Reference | |
Yes | 1.516 (1.179–1.950) | 0.001 |
Chemotherapy | | |
No/Unknown | Reference | |
Yes | 0.828 (0.674–1.017) | 0.072 |
Table 3
Multivariate Cox regression analysis for overall survival in training cohort
Characteristics | Multivariate analysis |
HR (95% CI) | P value |
Age | 1.034 (1.023–1.046) | < 0.001 |
Race | | |
White | Reference | |
Black | 1.378 (0.974–1.949) | 0.069 |
Other | 1.140 (0.683–1.901) | 0.614 |
Sex | | |
Male | Reference | |
Female | 0.985 (0.793–1.224) | 0.896 |
Marital status | | |
Married | Reference | |
Unmarried | 0.893 (0.715–1.115) | 0.320 |
Unknown | 1.019 (0.595–1.744) | 0.945 |
Year of diagnosis | | |
2004–2009 | Reference | |
2010–2015 | 1.173 (0.937–1.469) | 0.162 |
Laterality | | |
Left | Reference | |
Right | 1.110 (0.900-1.369) | 0.325 |
Bilateral | - | - |
Grade | | |
Ⅰ | Reference | |
Ⅱ | 1.046 (0.300-3.646) | 0.942 |
Ⅲ | 1.197 (0.379–3.782) | 0.758 |
Ⅳ | 1.272 (0.396–4.082) | 0.685 |
SEER stage | | |
Localized | Reference | |
Regional | 1.213 (0.868–1.696) | 0.256 |
Distant | 2.700 (1.296–5.624) | 0.007 |
T | | |
T1 | Reference | |
T2 | 0.726 (0.528–0.997) | 0.047 |
T3 | 1.106 (0.644–1.898) | 0.714 |
T4 | 1.281 (0.802–2.047) | 0.298 |
Tx | 5.206 (0.389–69.579) | 0.212 |
N | | |
N0 | Reference | |
N1 | 2.024 (1.442–2.840) | < 0.001 |
N2 | 2.228 (1.557–3.189) | < 0.001 |
N3 | 3.868 (0.816–18.329) | 0.088 |
Nx | 0.203 (0.024–1.723) | 0.144 |
M | | |
M0 | Reference | |
M1 | 0.947 (0.462–1.940) | 0.882 |
Mx | 11.076 (1.380-88.911) | 0.023 |
Tumor size (mm) | | |
< 20 | Reference | |
20–49 | 1.342 (1.013–1.778) | 0.039 |
> 50 | 1.752 (1.190–2.580) | 0.004 |
Unknown | 0.515 (0.101–2.620) | 0.424 |
Radiation | | |
No/Unknown | Reference | |
Yes | 1.203 (0.879–1.646) | 0.246 |
Chemotherapy | | |
No/Unknown | Reference | |
Yes | 0.549 (0.423–0.713) | < 0.001 |
Construction of nomogram
The nomogram is constructed based on the independent prognostic factors of the overall survival. As shown in Fig. 2, the variables in the nomogram include age, SEER stage, T stage, N stage, M stage, tumor size, and chemotherapy. Each variable has a corresponding number of points, and the points of all variables are added up to make a total number of points. The total number of points corresponds to the 1-, 3-, and 5-year survival probabilities.