Patient baseline characteristics
Between 2010 and 2016, 2315 metastatic RCC patients were identified from the SEER database based on inclusion criteria. The eligible patients were randomly separated into two sets at a ratio of 1:1, 1158 patients in the training set and 1157 patients in the validation set, respectively. Of all the patients, the median age was 61 (54–68), and the white race (83.8%) was the dominant race. Among the eligible patients, 1629 (70.4%) were males and 1559 (67.3%) were married. The most common histologic subtype was clear cell renal cell carcinoma (84.6%), and the most common Fuhrman grade was Grade III (40.0%). In terms of tumor stage, the majority of tumors were T3 (56.8%), followed by T2 (16.5%) and T4 (12.7%). With respect to treatment, 56.2% of the patients received chemotherapy, 23.3% of the patients received radiotherapy, and 85.6% of the patients underwent cancer-directed surgery. For metastasis, 31.5%, 9.9%, 11.0%, and 61.7% of the patients presented bone, brain, liver, and lung metastasis at diagnosis, respectively.
For the entire cohort, the median follow-up time was 43 months (95% CI, 41–44 months), and the median OS time was 19 months (95% CI, 18–21 months). By the end of the survey, 806 (69.6%) patients in the training set had died, among which 756 (93.8%) died from RCC and 50 (6.2%) died from other causes. The median follow-up time of the training set was 42 months (95% CI, 40–45 months), and the median survival time was 18 months (95% CI, 16–20 months). The demographic and clinical characteristics were summarized in Table 1.
Table 1
Patient demographics and clinical characteristics.
Variables | All patients (n = 2315) | Training set (n = 1158) | Validation set (n = 1157) |
No. (%) | No. (%) | No. (%) |
Age, year | 61 (54–68) | 60 (54–68) | 61 (54–68) |
Race | | | |
White | 1941 (83.8%) | 968 (83.6%) | 973 (84.1%) |
Black | 174 (7.5%) | 92 (7.9%) | 82 (7.1%) |
Other | 200 (8.6%) | 98 (8.5%) | 102 (8.8%) |
Sex | | | |
Male | 1629 (70.4%) | 812 (70.1%) | 817 (70.6%) |
Female | 686 (29.6%) | 346 (29.9%) | 340 (29.4%) |
Marital status | | | |
Married | 1559 (67.3%) | 777 (67.1%) | 782 (67.6%) |
Unmarried | 756 (32.7%) | 381 (32.9%) | 375 (32.4%) |
Histologic subtypea | | | |
CCRCC | 1959 (84.6%) | 982 (84.8%) | 977 (84.4%) |
PRCC | 142 (6.1%) | 56 (4.8%) | 86 (7.4%) |
CHRCC | 27 (1.2%) | 14 (1.2%) | 13 (1.1%) |
SRCC | 168 (7.3%) | 91 (7.9%) | 77 (6.7%) |
CDRCC | 19 (0.8%) | 15 (1.3%) | 4 (0.3%) |
Fuhrman grade | | | |
Grade I | 57 (2.5%) | 26 (2.2%) | 31 (2.7%) |
Grade II | 516 (22.3%) | 247 (21.3%) | 269 (23.2%) |
Grade III | 926 (40.0%) | 482 (41.6%) | 444 (38.4%) |
Grade IV: | 816 (35.2%) | 403 (34.8%) | 413 (35.7%) |
Tumor size, mm | 90 (70–118) | 90.5 (70–120) | 90 (68–115) |
Tumor classification | | | |
T1 | 289 (12.5%) | 141 (12.2%) | 148 (12.8%) |
T2 | 383 (16.5%) | 198 (17.1%) | 185 (16.0%) |
T3 | 1316 (56.8%) | 661 (57.1%) | 655 (56.6%) |
T4 | 293 (12.7%) | 143 (12.3%) | 150 (13.0%) |
TX | 34 (1.5%) | 15 (1.3%) | 19 (1.6%) |
Lymph node status | | | |
N0 | 1618 (69.9%) | 796 (68.7%) | 822 (71.0%) |
N1 | 597 (25.8%) | 310 (26.8%) | 287 (24.8%) |
NX | 100 (4.3%) | 52 (4.5%) | 48 (4.1%) |
Sarcomatoid feature | | | |
Yes | 446 (19.3%) | 220 (19.0%) | 226 (19.5%) |
No | 1697 (73.3%) | 848 (73.2%) | 849 (73.4%) |
Unknown | 172 (7.4%) | 90 (7.8%) | 82 (7.1%) |
Chemotherapy | | | |
Yes | 1302 (56.2%) | 643 (55.5%) | 659 (57.0%) |
No/Unknown | 1013 (43.8%) | 515 (44.5%) | 498 (43.0%) |
Radiotherapy | | | |
Yes | 539 (23.3%) | 255 (22.0%) | 284 (24.5%) |
No | 1776 (76.7%) | 903 (78.0%) | 873 (75.5%) |
Cancer-directed surgery | | | |
Yes | 1982 (85.6%) | 980 (84.6%) | 1002 (86.6%) |
No | 333 (14.4%) | 178 (15.4%) | 155 (13.4%) |
Bone metastasis | | | |
Yes | 729 (31.5%) | 361 (31.2%) | 368 (31.8%) |
No | 1586 (68.5%) | 797 (68.8%) | 789 (68.2%) |
Brain metastasis | | | |
Yes | 230 (9.9%) | 112 (9.7%) | 118 (10.2%) |
No | 2085 (90.1%) | 1046 (90.3%) | 1039 (89.8%) |
Liver metastasis | | | |
Yes | 254 (11.0%) | 126 (10.9%) | 128 (11.1%) |
No | 2061 (89.0%) | 1032 (89.1%) | 1029 (88.9%) |
Lung metastasis | | | |
Yes | 1429 (61.7%) | 723 (62.4%) | 706 (61.0%) |
No | 886 (38.3%) | 435 (37.6%) | 451 (39.0%) |
Follow-up, month | | | |
Median (95%CI) | 43 (41–44) | 42 (40–45) | 43 (40–46) |
Independent Prognostic Factors
Univariate and multivariate Cox regression analysis was carried out to find out independent risk factors of OS. Crude and adjusted HRs were presented in Table 2. After adjusting other risk factors, the results of multivariate Cox regression analysis showed that eight variables were significantly associated with OS, including Fuhrman grade, lymph node status, sarcomatoid feature, cancer-directed surgery, bone metastasis, brain metastasis, liver metastasis, and lung metastasis.
Table 2
Univariate and multivariate Cox regression analysis of OS in the training set.
Variables | Univariate analysis | Multivariate analysis |
HR (95% CI) | P | HR (95% CI) | P |
Age | 1.01 (1.00-1.02) | 0.025 | | |
Race | | | | |
White | Ref. | | | |
Black | 1.26 (0.99–1.61) | 0.06 | | |
Other | 1.08 (0.84–1.38) | 0.561 | | |
Sex | | | | |
Male | Ref. | | | |
Female | 1.07 (0.92–1.24) | 0.383 | | |
Marital status | | | | |
Married | Ref. | | | |
Unmarried | 1.12 (0.97–1.30) | 0.117 | | |
Histologic subtypea | | | | |
CCRCC | Ref. | | | |
PRCC | 1.28 (0.95–1.74) | 0.106 | | |
CHRCC | 1.06 (0.55–2.04) | 0.874 | | |
SRCC | 2.25 (1.78–2.85) | < 0.001 | | |
CDRCC | 2.37 (1.40–4.03) | 0.001 | | |
Fuhrman grade | | | | |
Grade I | Ref. | | Ref. | |
Grade II | 0.81 (0.48–1.35) | 0.411 | 1.07 (0.64–1.81) | 0.792 |
Grade III | 0.94 (0.57–1.55) | 0.809 | 1.38 (0.82–2.31) | 0.223 |
Grade IV: | 1.46 (0.88–2.41) | 0.142 | 1.80 (1.06–3.06) | 0.029 |
Tumor size | 1.001 (1.000-1.002) | 0.002 | | |
Tumor classification | | | | |
T1 | Ref. | | | |
T2 | 1.42 (1.08–1.87) | 0.013 | | |
T3 | 1.47 (1.16–1.86) | 0.002 | | |
T4 | 2.43 (1.83–3.23) | < 0.001 | | |
TX | 5.63 (3.18–9.97) | < 0.001 | | |
Lymph node status | | | | |
N0 | Ref. | | Ref. | |
N1 | 1.87 (1.61–2.18) | < 0.001 | 1.54 (1.31–1.80) | < 0.001 |
NX | 1.85 (1.35–2.53) | < 0.001 | 1.59 (1.15–2.20) | 0.005 |
Sarcomatoid feature | | | | |
Yes | Ref. | | Ref. | |
No | 0.47 (0.40–0.56) | < 0.001 | 0.61 (0.50–0.75) | < 0.001 |
Unknown | 1.05 (0.80–1.38) | 0.709 | 0.84 (0.61–1.15) | 0.282 |
Chemotherapy | | | | |
Yes | Ref. | | | |
No/Unknown | 0.83 (0.72–0.96) | 0.011 | | |
Radiotherapy | | | | |
Yes | Ref. | | | |
No | 0.90 (0.77–1.06) | 0.217 | | |
Cancer-directed surgery | | | | |
Yes | Ref. | | Ref. | |
No | 2.61 (2.19–3.12) | < 0.001 | 2.37 (1.91–2.93) | < 0.001 |
Bone metastasis | | | | |
Yes | Ref. | | Ref. | |
No | 0.80 (0.69–0.93) | 0.003 | 0.68 (0.58–0.80) | < 0.001 |
Brain metastasis | | | | |
Yes | Ref. | | Ref. | |
No | 0.54 (0.44–0.68) | < 0.001 | 0.61 (0.49–0.77) | < 0.001 |
Liver metastasis | | | | |
Yes | Ref. | | Ref. | |
No | 0.54 (0.44–0.66) | < 0.001 | 0.66 (0.54–0.81) | < 0.001 |
Lung metastasis | | | | |
Yes | Ref. | | Ref. | |
No | 0.65 (0.56–0.75) | < 0.001 | 0.65 (0.55–0.76) | < 0.001 |
Nomogram Construction
The nomogram for predicting 1-, 3-, and 5-year OS were constructed on the basis of the significant prognostic factors, which were exhibited in Table 2. As shown in the nomogram (Fig. 2), cancer-directed surgery made the greatest contribution to prognosis, followed by Fuhrman grade, brain metastasis, and sarcomatoid feature. The variables including lymph node status, liver metastasis, and lung metastasis represented moderate impacts on OS. Bone metastasis had the least effect on OS.
An Example Of Using The Nomogram
An example of how to use the nomogram was presented in Fig. 2. A patient was diagnosed as metastatic RCC with Fuhrman grade IV, N1 stage, no sarcomatoid feature, and lung metastasis. He underwent a cancer-directed surgery. Therefore, he would have 68 points for Fuhrman grade IV, 50 points for N1 stage, 0 points for no sarcomatoid feature, 45 points for bone metastasis, and 0 points for cancer-directed surgery, totaling 163 points. The total points corresponded to a 1-year survival probability of 57%, a 3-year survival probability of 20%, and a 5-year survival probability of 8%.
Nomogram Performance
Discrimination and calibration were employed to evaluate the model performance via the training and validation sets. The C-indexes of the nomogram were 0.701 (95% CI, 0.682–0.720) in the training set and 0.675 (95% CI, 0.654–0.696) in the validation set. With respect to the AJCC staging system, the C-indexes were 0.612 (95% CI, 0.591–0.633) in the training set and 0.600 (95% CI, 0.578–0.622) in the validation set, significantly lower than those of the nomogram (P < 0.001). Calibration plots of the nomogram (Fig. 3) revealed excellent consistency between the predicted survival and the actual survival in both sets. The NRI values for 1-, 3- and 5-year follow-up were 0.553 (95% CI: 0.385–0.678), 0.547 (95% CI: 0.380–0.706) and 0.458 (95% CI: 0.284–0.703) in the training set, respectively. For the validation set, the NRI values for 1-, 3- and 5-year follow-up were 0.448 (95% CI: 0.309-0.600), 0.452 (95% CI: 0.305–0.626) and 0.360 (95% CI: 0.186–0.572), respectively. In addition, the IDI values for 1-, 3- and 5-year follow-up were 0.081 (P < 0.001), 0.083 (P < 0.001), and 0.063 (P < 0.001) in the training set and 0.051 (P < 0.001), 0.056 (P < 0.001), and 0.047 (P < 0.001) in the validation set, respectively. These results all exhibited the established nomogram had a better predictive ability of OS in comparison with the AJCC staging system.
Clinical Use
DCA plots showed that our nomogram had greater net benefits in comparison with the AJCC staging system in terms of predicting 1-, 3- and 5-year OS in the training and validation sets (Fig. 4), indicating its clinical usefulness and help to decision-making.