We identified 54 patients who met the inclusion criteria of our study. The median age was 67 years (range 42–82). 54 spinal segments were operated on. KPS before surgery was 60 or lower in 22 patients (22/44; 50%). In 10 patients, presurgical KPS could not be assessed due to the retrospective nature of the study and missing documentation. 15 out of 51 patients (29%) had severe comorbidities resulting in an ACE-27 score of 2 or 3. Conversely, 15 patients (29%) had no relevant comorbidities (ACE-27 score 0). 14 patients (26%) were obese (BMI ≥ 30). Body mass index was significantly higher for patients with ACE-27 scores 2 or 3 (28.7 ± 4.43) as compared to 0 or 1 (25.7 ± 3.39; p < 0.01). The median follow-up time was 63 months (95% CI: 49–94 months). Further details are given in Table 1.

The most common indication for surgery was vertebral instability (n = 40/54; 74%) followed by neurologic symptoms (n = 5/54; 9%). In another 5 cases, surgery was indicated due to both instability and neurological symptoms, while 4 patients received spinal surgery for pain relief (8%).

In nearly all patients, RT was indicated due to the postsurgical setting (n = 51/54; 94%). In two patients, RT was also indicated due to instability (4%), while one patient received RT due to persistent pain after surgery (2%). Although pain was not the reason that RT was indicated, 38 patients (70%) reported pain before RT, while only 11 patients (20%) reported some kind of persistent neurologic symptoms, including pain, after surgery (Table 2). Further details regarding treatment modalities are outlined in Table 2.

Spinal lesions or PTV involved the junctional zone in most patients (n = 35/54; 65%). 43 patients (80%) had fractures of varying degrees.

Average SINS score was 10.3 ± 2.5 before RT. Most patients (n = 41/54; 76%) had potentially instable lesions, while only 4 patients (7.5%) had stable lesions. Further details regarding spinal scores and stability of bone lesions can be found in Table 1. Most patients had a Bilsky score of 1 (n = 28; 53%), while there were nearly equal numbers of patients with a Bilsky score of 0 (n = 12; 22%) or at least 2 (n = 13; 25%). Bilsky scores did not differ significantly between patients who reported improved or no pain and those with persistent or increased pain after RT (p < 0.2).

Surgical procedures used were spondylodesis (n = 22, 41%), “tumor debulking” with laminectomy (n = 15, 28%), spondylodesis combined with laminectomy (n = 12; 22%), or kyphoplasty (n = 5; 9%). Of these, 40 (74%) surgeries were performed before or within 30 days of diagnosis. Blood loss as estimated by the anesthesiologist was significantly different between groups defined by surgical procedure (p < 0.05). However, pairwise Wilcoxon tests did not show significant differences between two specific groups. In contrast, there were no significant differences in perioperative blood transfusions between groups (p = 0.4). The severity of postoperative complications also did not differ significantly between groups (p = 0.3). Median time from surgery to discharge was 11.5 days (95% CI 9–16 days), while median time between surgery and start of RT was 34.5 days (95% CI 29–46 days).

Of 54 patients, 22 did not receive systemic therapy prior to RT, while the most common systemic therapy prior to RT consisted of bortezomib plus glucocorticoids, either with or without an additional agent (n = 21; 66%). 6 patients (11%) received systemic therapies and RT simultaneously. More details on systemic therapy regimens prior to RT can be seen in Table 3.

Each operated spinal segment received postoperative RT with a median total dose of 30 Gy (range 20–44 Gy) in 10 fractions (range 5–22 fractions). In most patients (n = 48/54; 89%), more than one vertebra was irradiated. We observed a significantly better prognosis for patients who did not receive a standard RT regimen (either 30 Gy/10 fractions or 20 Gy/5 fractions; p < 0.05). These patients received normofractionated RT with doses ranging between 30 and 44 Gy. During the entire disease course, 32 patients received bisphosphonate therapy (n = 32/52; 62%).

The 3‑ and 5‑year overall survival (OS) was 74.9% (95% CI: 63.5–88.4%) and 58% (95% CI: 44.5–75.6%), respectively. Median survival was not reached, 75% survival was 34.3 months (95% CI: 28.7–95.4 months).

Good to adequate presurgical Karnofsky performance scores (KPS ≥ 70) showed a trend toward improved OS (mean OS: 109 vs. 68.2 months; p < 0.1), while patients with preradiotherapy KPS ≥ 70 had significantly better OS (median OS: 39.8 months vs. not applicable (NA); p < 0.01). McNemar’s test showed significant improvements in KPS after surgery (KPS ≥ 70: 50% vs. 79.2%; p < 0.01).

RT showed significant improvements in pain, as 38 patients (71%) reported some degree of pain after surgery and before RT. However, the majority of patients reported no pain (n = 35/53; 66%) or decreased pain (n = 14/53; 26%) after RT. Only two patients (4%) reported increased pain after RT. Acute toxicities were sparse (n = 13/54; 24%) and no higher-grade adverse events (Common Terminology Criteria for Adverse Events (CTCAE) grade 3) occurred (Table 4). There was neither interruption of RT nor delay of the first systemic therapy cycle after RT due to hematological toxicities. We observed no local recurrence within the treatment field.

While there was no significant difference in OS depending on age (p = 0.8), bisphosphonate therapy (mean OS: 108.5 vs. 70.4 months; p < 0.05) and a Bilsky score of 0 or 1 (mean OS: 104 vs. 62 months; p < 0.05) were associated with significantly better survival. Patients with a body mass index below 30 had a trend towards improved prognosis (mean OS: 107.1 vs. 74.2 months; p = 0.1). Corresponding Kaplan–Meier plots and risk tables for factors that showed a trend in survival differences can be seen in Fig. 1. We observed only two SRE (4%).

Kaplan–Meier plots for overall survival when stratifying patients for a Bilsky score before radiotherapy (RT); b bisphosphonate therapy; c Karnofsky performance score (KPS) before RT; d body mass index. Bisphosphonate therapy, pre-RT KPS ≥ 70 and low Bilsky scores of 0 or 1 proved to be significant prognostic factors in univariate analysis. Lower BMI does show a trend towards improved overall survival, albeit not statistically significant at p < 0.05

Median progression-free survival was 26.5 months (95% CI: 20.9–48.4 months). KPS before RT (median PFS: 13.8 vs. 29.7 months; p < 0.01), revised International Staging System (rISS) stage before RT (median PFS: 26.5 vs. 59.7 months; p < 0.05), bisphosphonate therapy (median PFS: 13.8 vs. 38.2 months; p < 0.001), and Bilsky score of 0/1 (median PFS: 29.7 vs. 15.9 months; p < 0.01) were significantly associated with PFS (Fig. 2).

Kaplan–Meier plots for progression-free survival when stratifying patients for a Bilsky score at baseline (BL) before radiotherapy (RT); b bisphosphonate therapy; c Karnofsky performance score (KPS) before RT; d revised International Staging System (rISS) stage. All abovementioned factors proved to be significant prognostic factors for progression-free survival in univariate analysis

In the multivariate Cox model for OS, a Bilsky score of 2 or 3 showed a significantly increased hazard ratio (HR 3.89; 95 CI 1.4–10.7; p < 0.01). Obesity showed a statistically non-significant trend towards increased HR (HR: 2.57; 95% CI 0.94–7.1; p = 0.067).

In the multivariate Cox model for PFS, a Bilsky score of 2 or 3 showed a significantly increased hazard ratio (HR 2.78; 95 CI 1.27–6.1; p < 0.05). Obesity showed no sign of an increased HR (HR: 1.16; 95% CI 0.55–2.4; p = 0.7).