The median post-SRS follow-up period for 124 censored observations (8.4%) was 6.5 (IQR; 1.1-20.6) months, with 1358 patients (91.6%) having died as of the end of June 2020. The overall median survival time (MST) after SRS was 6.2 (95% CI; 5.8-6.8) months. The respective actuarial post-SRS survival proportions were 27.7%, 10.9%, 4.6%, 2.5% and 1.6% at the 12th, 24th, 36th, 48th and 60th post-SRS months. Among the 1358 deceased patients, the causes of death could not be determined in 26, but were confirmed in the remaining 1332 to be non-brain diseases in 1136 (85.3%) and brain diseases in 196 (14.7%) patients. Among the total 1482 patients, follow-up MR imaging was performed at least once in 1238 (83.5%). Among the 1482 patients, 553 (37.3%) underwent salvage SRS, generally for newly-appearing lesions (499 patients, 92.3%) and, less commonly, for recurrence of a treated lesion (54 patients, 3.7%), while salvage WBRT was administered to 98 (6.6%).
Cohort study
The post-SRS MST was slightly longer in the group with 10-19 BMs (6.7 [95% CI; 5.9-7.3] months) than in that with ≥20 BMs (5.7 [95% CI; 5.0-6.4] months), as shown in Fig. 1, A. Although the post-SRS MST difference reached statistical significance (hazard ratio [HR]; 1.121, 95% CI; 1.005-1.250, p=0.040), the actual MST difference was only 1.0 month. The respective actuarial post-SRS survival proportions in the patients with 10-19 BMs were 29.0%, 11.6%, 5.2%, 2.2% and 1.4% at the 12th, 24th, 36th, 48th and 60th post-SRS months, while the corresponding percentages in patients with ≥20 BMs were 25.6%, 9.7%, 3.5%, 2.6% and 1.5%. In patients with 10-19 BMs, MSTs were 16.6 months for Modified Recursive Partitioning Analysis (M-RPA) 1+2a, 11.2/2b and 4.7/2c+3, respectively (p<0.0001) (Figure 2, A) [11,14]. Also, in patients with ≥20 BMs, the MSTs were 31.0, 9.5 and 4.5 months for M-RPA 1+2a, 2b and 2c+3, respectively (p<0.0001) (Fig. 2).
Among various pre-SRS clinical factors examined, univariable analysis demonstrated female gender, KPS ≥80%, neurologically asymptomatic, synchronous presentation, controlled primary cancer and no extra-cerebral metastases to be factors significantly favoring longer survival in both of the number groups, i.e., 10-19 BMs and ≥20 BMs, as shown in Table 3. For all clinical factors significantly impacting survival, both HRs and p-values were similar in the two groups. No impacts, on survival, of age, cumulative BM volume, the volume of the largest BM, minimum and maximum irradiation doses, or the primary cancer and prior treatments, i.e., pre-SRS, WBRT or surgery, were detected.
Case-matched study
As shown in Fig. 1, the post-SRS MSTs were the same in the two BM number groups, 5.9 (95% CI; 5.2-7.2) months for those with 10-19 BMs and also 5.9 (95% CI; 5.1-6.8) months for the group with ≥20 BMs. The respective actuarial post-SRS survival proportions in the patients with 10-19 BMs were 26.7%, 10.5%, 5.2%, 1.9% and 0.9% at the 12th, 24th, 36th, 48th and 60th post-SRS months, while the corresponding values in patients with ≥20 BMs were 27.0%, 11.0%, 4.1%, 2.9% and 1.5% (HR; 1.017, 95% CI; 0.890-1.164, p=0.80).
The crude and cumulative incidences of neurological death, neurological deterioration, local recurrence, repeat SRS, salvage WBRT and SRS-related complications did not differ significantly between the two BM number groups (Table 4).
Factors relating to high incidences of SRS-related complications
Univariable analyses of the cohort dataset (1482 patients with ≥10 BMs) showed the presence of neurological symptoms, the cumulative and largest BM volumes, minimum dose and 12 Gy-brain volume to be significantly related to higher incidences of SRS-related complications, as shown in Table 5. Nevertheless, none of these five clinical factors reached statistical significance on multivariable analyses. For this study on SRS-related complications, we used receiver operating characteristics analysis to determine the cut-off value for dividing the continuous variables into two groups, i.e., 4.15 mL (area under the curve [AUC]; 0.536)/cumulative volume, 9.94 mL (AUC; 0.461/the largest BM volume, 20.00 Gy (AUC; 0.538)/minimum dose and 42.8 mL (AUC; 0.563)/12 Gy-brain volume. For convenience, the nearest integers were applied. The cut-off value for the cumulative volume of 4 mL was incompatible with the largest BM volume of 10 mL. Therefore, we also used 10 mL as the cut-off value for the cumulative BM volume.