1. Demographic and clinical characteristics of lung cancer patients with synchronous brain metastasis
This study included 29902 patients diagnosed with synchronous brain metastases of lung cancer from 2010 to 2016 in the SEER database (Figure 1). Table 1 listed the demographic and clinicopathological characteristics of patients in the training set (n=26272) and validation set (n=3630). In general, there were slightly more men than women, and most of them were patients over 40 years old. 79.5% of cases were reported as white and 12.2% were black. Most cases were classified into lung adenocarcinoma (51.6%), and SQCC, SCLC and LCLC accounted for 10.6%, 17.0% and 2.3%, respectively. Gleason grade III lung cancer was significantly higher than other grades. Some patients had brain metastases accompanied by liver metastases (20.7%) or bone metastases (33.5%). Very few patients underwent surgery (3.2%), a small number of patients underwent radiotherapy (21.1%), and about half of the patients underwent chemotherapy (56.1%). There was no significant difference in composition between the training group and the validation group.
2. Mortality of early death
Among all lung cancer patients, 27.5% of patients had early deaths, and 22.6% of them were caused by lung cancer (Figure 2A). The early mortality of patients with LCBM is 44.4%, and the early mortality caused by lung cancer was as high as 38.2% (Figure 2B). The early death of patients with LCBM remained stable between 2010 and 2016 (Figure 3A). The early death rate increased significantly with age, whites were slightly higher than that of other ethnic groups and men were higher than that of women (Figure 3B, C, D).
3. Identifying independent factors for early death
Univariate and multivariate logistic regression were used to analyze the risk factors of early death in patients with LCBM in the SEER training group. The results of univariate and multivariate analysis were shown in Table 2 and 3. In univariate analysis, most clinical and pathological characteristics, such as gender, race, age at diagnosis, Gleason grade, histology, T stage, N stage, bone metastasis, liver metastasis, surgery, radiotherapy, chemotherapy and marital status were related to the probability of overall early death. All significant factors were included in the multivariate analysis. Multivariate analysis showed that gender, race, age at diagnosis, Gleason grade, histology, T stage, N stage, bone metastasis, liver metastasis, surgery, radiotherapy and chemotherapy were independent risk factors for predicting overall early death in patients with LCBM. The result of cancer-specific early death is consistent with that of overall early death.
Depending on the multivariate logistic regression analysis model, the risk factor prediction nomogram of the SEER cohort was determined. Figure 4A is an example of using a nomogram to predict the survival probability of a given patient. By calculating each variable point, the total number of points can be attached to the probability of overall early death. Total risk point for most patients in this study is between 750 and 930. Figure 4B showed the predicting nomogram of the probability of cancer-specific early death.
The nomogram showed good prediction efficiency of the probability of early death. The ROC curve used to assess the overall and cancer-specific early death nomogram is shown in Figure 5. The areas under the curve (AUC) of the training group were 0.828 (Figure 5A; overall early death, 95%CI: 0.822-0.833) and 0.800 (Figure 5B; cancer-specific early death, 95%CI: 0.794- 0.806), respectively. And the AUC of the validation group were 0.851 (Figure 5C; overall early death, 95%CI: 0.838- 0.864) and 0.813 (Figure 5D; cancer-specific early death, 95%CI: 0.798-0.828), respectively. The calibration curve of the two nomograms showed that the prediction and observation result in training and validation group were very consistent (Figure 6). In addition, when predicting overall early death and cancer-specific early death from the nomogram, DCAs showed an ideal net benefit for all patients, indicating that the nomogram had good clinical value for predicting the probability of early death (Figure 7). In the hospital cohort, we scored each patient with the nomogram. Only nomogram of cancer-specific early death was used, since the causes of early death of all patients were lung cancer and related factors. When the patients were divided into high score group and low score group, we found that the early mortality in high score group was much higher than that in low score group, and its AUC was as high as 0.719 (Figure 8; 95%CI: 0.545-0.892).