Demographics:
A total of 205 patients (110 male and 95 female) aged between 18 and 85 years (median 61 years) were included in this study. Primary tumors were distributed within common incidences with NSCLC being the most frequent (44.9%), followed by breast cancer and melanoma (11.2% and 14.1% respectively, see Table 1).
Table 1
Demographics. NSCLC = non-small-cell lung cancer, RCC = renal cell cancer, CUP = cancer of unknown primary, GTR = gross total resection
No total | | 205 |
Gender, n(%) | Male | 110(53.7) |
| Female | 95(46.3) |
Primary, n(%) | | |
| Lung (NSCLC) | 92(44.9) |
| Breast | 23(11.2) |
| Melanoma | 29(14.1) |
| RCC | 5(2.4) |
| CUP | 6(2.9) |
| Others | 50(20) |
Location, n(%) | | |
| Supratentorial | 170(82.9) |
| Infratentorial | 35(17.1) |
| Eloquent | 52(25.4) |
GTR, n(%) | | |
| Surgeon-estimated | 194(94.6) |
| MRI-defined | 159(77.6) |
Eighty-three percent of the resected metastases were located supratentorially versus 17.1% in the cerebellum. 74.6% of the BM occupied non-eloquent areas of the brain.
A complete resection could be achieved in 77.6% according to early postoperative MRI, with an expected discrepancy to the surgeons’ intraoperative rating of gross total resection (GTR) of 94.6%.
Only a small proportion of the patients showed transient postoperative worsening of their neurological status (12.7%, n = 26), whereas 7.8% (n = 17) improved in neurological status immediately after surgery and 79.5% (n = 163) were found to be stable.
Median follow-up (FU) of the included patients amounted to 10 months (IqR 4–25). Patients who died during the follow-up were followed for a median of 7 months (IqR 7–16) and patients that were still alive at time of analysis had a median follow-up of 62 months (IqR 4–25).
Frailty:
Preoperatively, the patients demonstrated a median CFS of 3 (“managing well”, IqR 2–4) that could be preserved postoperatively and during the 3–6 months follow-up.
KPS was 80 preoperatively (median, IqR 80–90) and increased to 90 (median, IqR 80–90) postoperatively, remaining stable at the follow-up (IqR 80–100) at 3 to 6 months.
The preoperative CFS correlated significantly with the postoperative CFS (p < 0.001) and the score at the 3–6 months follow-up (p < 0.001). Spearmen’s test revealed a moderate correlation between pre- and postoperative CFS (r = 0.629) and a weak correlation of preoperative CFS and the performance at the 3–6 months follow-up (r = 0.309; see Fig. 1).
Monovariate analysis revealed a lower preoperative CFS of male patients with a median CFS of 4 (IqR 3–4) compared to a CFS of 3 (IqR 2–4) in women (p = 0.026). Postoperatively, a comparable performance (male: female – 3:2 (both IqR 2–4), p > 0.05.) was noticed and in the 3–6 months follow-up male patients performed worse with a median CFS of 3 (IqR 2–4) compared to a CFS of 2 (IqR 1–4) in female patients (p = 0.047).
Neither the number of resected BM (1–3) nor the experience level of the surgeon (resident vs. consultant) nor the location of the metastasis showed a significant influence on the CFS. Only BM in eloquent locations were associated with lower CFS and KPS preoperatively (CFS/KPS: p = 0.008/p = 0.007) and postoperatively (p = 0.009/p = 0.008), but this correlation was lost at follow-up. If a new neurological deficit was recorded after surgery, CFS and KPS scores worsened significantly (p < 0.001).
The Cox regression analysis revealed a significant influence of the KPS on OS preoperatively (HR 1.267 per 10 point-step, CI 1.132–1.395), postoperatively (HR 1.142 per 10 point-step, CI 1.070–1.225, p < 0.001) and at 3–6 months FU (HR 1.320 per 10 point-step, 1.221–1.420, p < 0.001). The CFS demonstrated an even stronger prediction of outcome preoperatively (HR 1.3 per step, CI 1.157–1.460, p < 0.001), postoperatively (HR 1.394 per step, CI 1.258–1.545, p < 0.001) and at 3–6 months follow-up (HR 1.421 per step, CI 1.270–1.590, p < 0.001) (see Fig. 2).
In a forward stepwise Cox regression model of the data analyzing KPS and CFS pairwise, CFS was superior at predicting the clinical course at all assessments: pre- and postoperatively as well as at 3–6 months FU (p < 0.001 for each pair, see Fig. 3).
The strongest predictor for the clinical course in the forward stepwise model of all performance scores was the CFS at the 3–6 months FU (HR 1.421, CI 1.270–1.590, p < 0.001). Patients’ age showed a significant impact on the clinical course with a HR of 1.029 (p < 0.001) as well as the preoperative tumor volume (HR 1.013, p = 0.004) and the number of brain metastases (HR 1.240, p = 0.007).