We reviewed the medical records of all adult patients who were diagnosed with histologically proven SCLC between July 2017 and October 2021 at Shandong Cancer Hospital and Institute. Exclusion criteria included incomplete chart information. Patients with other concurrent malignancies or combined-type SCLC (SCLC with squamous cell carcinoma and/or adenocarcinoma) were excluded from this study. Data were collected retrospectively, including physical examination, treatment, chest radiography, contrast-enhanced computed tomography (CT) scans of the chest and upper abdomen, magnetic resonance imaging of the brain, bone radionuclide imaging, and cyto-histopathological results of transthoracic needle biopsy or bronchoscopic biopsy. All patients had no brain metastases on MRI and PCI was not performed. All study patients who had been classified using the two-stage system were restaged according to the UICC for International Cancer Control 7th TNM staging. The current study was performed in accordance with the Declaration of Helsinki and approved by the local ethics committee of the Shandong Tumor Hospital (Jinan, China; approval no. FY2017015). Each patient provided written informed consent for participation.

All patients signed written informed consent for inclusion of their clinical details in the manuscript for publication.

All patients received either cisplatin or etoposide (EP: 30 mg/m² cisplatin on days 1-3 and either 100 mg/m² etoposide on days 1-5 or 100 mg/m² etoposide on days 1-3) or carboplatin and etoposide (CE: carboplatin AUC 5 or 300 mg/m² on day 1 and either 100 mg/m² etoposide on days 1-5 or etoposide 100 mg/m² on days 1-3). The median number of cycles of chemotherapy was 5.

Thoracic radiation therapy (TRT) was administered using intensity modulated radiation therapy technique. The target volume for thoracic radiotherapy included the gross tumor, as defined by the chest CT scan, and the bilateral mediastinal and ipsilateral hilar lymph nodes. This has been our institutional practice though we acknowledge many other centers internationally have moved to omission of elective nodal irradiation and primarily target PET-positive disease with margin. The inferior border extended 3 cm below the carina or to a level including the ipsilateral hilar structures, whichever was lower. The clinical target volume (CTV) was expanded by a margin of 1 to 1.5 cm, and the planning target volume (PTV) included the CTV with a 0.5 cm margin. The radiation was delivered using megavoltage linear accelerators.

TRT was initiated within one week of the second cycle of chemotherapy. Patients receiving once-daily therapy received 1.8-2.0 Gy/day to 42-66 Gy over a period of five weeks. Accelerated twice-daily thoracic radiotherapy involved administration of 1.5 Gy over a period of three weeks.

The follow-up schedule started from the time of the first treatment. Long-term outcomes were obtained from the hospital records or correspondence with patients, families. The last follow-up was conducted on March 31, 2023. CT scans were obtained after the second and fourth cycles of chemotherapy for tumor assessment and every 8-12 weeks thereafter for further follow-up. Response to therapy was independently re-evaluated within 2 weeks after CCRT completion and based on CT scanning, in accordance with the Response Evaluation Criteria in Solid Tumors (RECIST, version. 1.1) (18).

The primary endpoint in this study was BM-free survival time, defined as the period from the initial diagnosis to the date of cerebral metastasis, in patients whose first site of recurrence was the brain.

Descriptive statistics for clinical variables are presented as median and interquartile range (IQR), or as numbers with percentages, as appropriate. Descriptive statistics for BM-free survival time are shown as Hazard ratio (HR) and 95% confidence intervals (95% CI). The actuarial risk of developing BM and BM-free survival time curves were estimated using the Kaplan-Meier method. The log-rank test was used to compare the differences between the groups. Cox proportional hazards regression models were used to develop univariate and multivariate models describing the association between independent variables and BM-free survival time. All reported characteristics were considered in a multivariable Cox model to determine which demonstrated the strongest association with the BM-free survival time. A forward stepwise selection procedure was implemented with a two-sided p-values threshold of ≤ 0.05 for inclusion in the final model. The HR and 95% CI were reported for the variables included in the final model. P-values of ≤0.05 were considered statistically significant. All analyses were performed using SPSS version 22.0 software.

The final multivariable model was converted to a risk score for BM using the following steps: (1) the variable with the smallest regression coefficient in the model was assigned 1.0 risk score points; (2) risk score points were assigned to all other variables in the model by dividing their respective regression coefficients by the smallest coefficient in the model; this quotient rounded to the nearest tenth is the risk score point assignment for that variable; and (3) a total risk score was calculated for every patient by adding the risk score points assigned to each variable in the model (19). The set of total risk scores was divided into three groups, and the Kaplan-Meier survival curves for each juxtaposed to qualitatively assess the risk score’s ability to discriminate between patients with high and low risk of BM. The discriminating ability of the risk score model was measured quantitatively by the c-statistic.

In total, the records of one hundred and eighty-two patients were enrolled and analyzed in this study, among which 182 patients (117 men and 65 women, all Han Chinese) completed the described treatment and developed BM. The patient characteristics are summarized in Table 1. The median follow-up period for all patients was 57 months (range, 30-81 months). The median age at diagnosis was 57 years (range, 32-80 years).

All our patients finally developed BM. The time for the development of BM ranged from 2 months to 60 months, which is a natural process after systemic treatment, with a median time from initial diagnosis to BM development 17 months (25^th and 75th percentiles: 9, 23). IQR = 75^th percentile – 25^th percentile: 14 months. The median ± IQR for the BM-free survival time was 17 ± 14 months. The estimated BM-free survival rates at 12 and 24 months were 65.9% and 17.0, respectively (Figure 1).

The clinical factors evaluated in the univariate analyses to determine their prognostic value for the risk of developing BM are shown in Table 2. Smoking status influenced the BM-free survival time in non-smokers and smokers (P = 0.026, Figure 2). The estimated 12-month and 24-month BM-free survival times were 60.8% and 16.2% in patients who smoked; and 78.8% and 19.2% in patients who did not smoke. There was a significant difference in the BM-free survival time according to the response of PT to CCRT (P  < 0.001, Figure 3). The estimated 12-month and 24-month BM-free survival times were 82.8% and 23.0 in patients with CR; 71.4% and 16.3% in patients with PR; and 28.3% and 6.5% in patients with NR. TNM stage grouping of the initial SCLC showed detrimental effects on the BM-free survival time (P = 0.007, Figure 4). The estimated 12-month and 24-month BM-free survival rates were 87.5% and 41.7% in patients with stage II disease and 61.0% and 12.3% in patients with stage III disease. The risk of developing BM was associated with the T descriptors (P = 0.002, Figure 5). The estimated 12-month and 24-month BM-free survival times were 81.6% and 26.3%, 76.8% and 21.4%, 55.3% and 7.9%, 50.0% and 12.0%, in patients with T1 disease T2 disease, T3, and T4 disease, respectively. N descriptors were also associated with the BM-free survival time (P < 0.001, Figure 6). The estimated 12-month and 24-month BM-free survival times were 88.9% and 50.0%, 77.6 and 26.5, 54.5% and 5.5% in patients with N1, N2, and N3 disease, respectively. We did not describe patients with stage I or N0 disease because only four patients and five patients respectively.

Multivariate Cox proportional hazards regression analysis showed that PT response to CCRT (HR = 1.609, 95%CI: 1.324-1.954, P < 0.001) and N descriptor (HR = 1.564, 95%CI: 1.268-1.929, P < 0.001) were independent factors associated with an increased risk of developing BM (Table 3). The final multivariable model of the risk factors for BM-free survival time with the corresponding risk score points is shown in Table 3. This model included two characteristics related to BM-free survival (N descriptors and PT response to CCRT). Risk score points were assigned as shown for each of the two listed factors, with a total risk score calculated per patient as the sum of the risk score points for all two factors. Table 4 shows the risk score calculations for the 7 most commonly observed combinations of factors in our patient population. For instance, as shown in the third row of Table 4, an individual with stage N3 disease (5.9 risk score points), and complete response of PT to CCRT (0.0 risk score points) would be assigned a risk score of 5.9 points (risk score = 5.9 + 0.0 = 5.9).

Risk score groups were created empirically and referring to Brent^’s study (20) as follows: (1) risk score of 4.0 or less (low risk); (2) risk score of 4.1 to 8.0 (moderate risk); and (3) risk score greater than 8.0 (high risk). In univariate analyses, the differences were statistically significant among the three groups. The estimated 12-month BM-free survival rates across groups were 91.7%, 66.3%, and 25.0%, respectively. The estimated 24-month BM-free survival rates across the groups were 39.6%, 12.2%, and 0.0%, respectively (P < 0.001, Figure 7). The actuarial risk of BM development in the three groups is shown in Figure 8.