Stability and survival analysis of elderly patients with osteolytic spinal bone metastases after palliative radiotherapy

The medical records of 322 elderly patients with SBM treated with palliative RT at the University Hospitals of Heidelberg and Mainz and the German Cancer Research Center were retrospectively assessed. Patient data from cancer registries of participating centers were collected to include patients aged 70 years or older receiving palliative RT for SBM between March 2000 and January 2014. The diagnosis of SBM was based on imaging procedures such as computed tomography (CT), magnetic resonance imaging (MRI) or bone scintigraphy. Inclusion criteria of this analysis were an age ≥70 years, an osteolytic phenotype of SBM and location of metastases in the thoracic or lumbar spine.

To assess the stability of osteolytic SBM, the validated Taneichi bone stability score was based on the planning CT scans and two consecutive follow-up CT scans conducted at 3 months (follow-up 1) and 6 months (follow-up 2) after RT. The applied scoring system is based on purely radiological criteria and is therefore a simple method for classifying osteolytic SBM in stable and unstable lesions [14]. For all patients included in this dataset, stability assessments were performed by a board-certified radiologist: Osteolytic SBM were rated on a scale from A to G based on radiological risk factors such as the degree of vertebral body tumor affectation, costovertebral joint or pedicle destruction. Based on previous experiences and publications, type A to C lesions were classified as stable and type D to G lesions as unstable (Fig. 1; [5, 6, 15]). In the case of multiple osteolytic lesions per vertebral body, only the most severe lesion was scored. In 57% of the patients, the planning target volume (PTV) comprised more than one tumor affected vertebral body.

For all patients, RT was planned based on planning CTs and performed over one or several 6 MV photon fields. The PTV included the vertebral body/bodies affected by the metastatic spread and the neighboring ones in the cranial and caudal directions. The most common fractionation scheme was 10 × 3 Gy, followed by 5 × 4 Gy and 20 × 2 Gy (Table 1). The indication for palliative RT was individually determined based on disease stage and the patients’ performance status. None of the study patients received additional surgery or other invasive procedures. Further information concerning additional nonsurgical treatments is provided in Table 1.

Most patients exhibited unstable SBM prior to RT according to the Taneichi score (i.e., subtypes D–G; 57%, n = 183), and 75% of those patients initially presented with associated pain symptoms (n = 138). During the follow-up period, RT stabilized primary unstable bone lesions in 13% (n = 23/183) and 17% (n = 31/183) of the patients as assessed at the first and second follow-up, respectively (p < 0.001 for each follow-up, McNemar’s test). When referring only to the patients with unstable SBM still alive at follow-up 1 and 2, the corresponding stabilization rates were 19% (23/118) and 40% (31/78), respectively. Only 1 out of 139 patients with initially stable SBM was observed with a deterioration of stability and was classified unstable in both follow-up examinations. The Bowker test showed significant departure from symmetry in the joint distribution pattern of the scored Taneichi subtypes prior to irradiation and at both follow-up examinations (Tables 2 and 3) (p < 0.001 at both follow-ups). At 6 months after palliative RT, highest stabilization rates of primary unstable osteolytic SBM were seen in breast cancer patients (Table 4). In contrast, patients with osteolytic SBM from poorly represented tumor entities in our study population such as malignant melanoma, hepatic cell cancer (HCC), head and neck cancers, gynecologic and genitourinary malignancies mainly exhibited a very poor outcome in terms of stabilization (Table 4). Univariate ordinal logistic regression at follow-up 1 and 2 revealed that several factors correlated with a significant negative stabilization probability of initially unstable SBM after palliative RT compared to breast cancer histologies, among them the presence of pathological fractures before the initiation of RT, lung cancer histology or histology of the group of poorly represented tumors (summarized in Table 5). The chance for reaching stabilization and re-ossification in patients with unstable SBM and a KPS below 70% was relatively poor (5%; 3/65) compared to those patients with a KPS of 70% and above (24%; 28/118). Pathological fractures were commonly diagnosed prior to RT (21%; n = 68), while post-RT fractures including increasing sintering of previously collapsed vertebrae were evident in only 5% of patients (n = 17). In 94% of those post-RT fractures (16/17), SBM were initially rated unstable.

For the complete patient cohort, relatively poor survival was evident after palliative RT. Overall survival rates 3, 6 and 12 months after RT amounted to 66% (95% CI 61–72%), 48% (95% CI 43–54%) and 29% (95% CI 24–34%); the median survival following RT was 5.4 months (95% CI 4.4–7.2 months).

To analyze the association of age with overall survival, we stratified the study patients into 3 age groups (i.e., 70–74 years, 75–79 years and ≥80 years) and found a strong tendency towards worse survival after RT for patients ≥80 years, albeit not reaching statistical significance (p = 0.06; Fig. 2). The corresponding median overall survival was 7.6 months (95% CI 4.8–9.0 months) for patients aged 70–74 years, 5.0 months (95% CI 3.6–9.2 months) for patients aged 75–79 years and 4.3 months (95% CI 3.2–6.3 months) for patients aged 80 years or older.

In comparing overall survival for patients with the 4 most common primary tumor histologies (i.e., breast, renal, colorectal and lung cancer) accounting for 81% of the study population (261/322 patients), we found significantly better survival rates for breast cancer patients and a significantly worse prognosis for lung and colorectal cancer patients (p = 0.003; Fig. 3).

Using multivariate analyses, KPS was found to be the strongest predictive factor for overall survival after RT (p < 0.001, multivariate Cox regression analysis; Table 6). In patients with KPS <70%, median overall survival was only 1.8 months (95% CI 1.6–2.3 months) compared to 9.1 months (95% CI 7.6–10.5 months) for patients with KPS ≥70% (p < 0.001; Fig. 4). Furthermore, the number of SBM, the application of chemotherapy and a tumor histology from the pooled group of poorly represented entities were significant prognostic factors in the multivariate analysis (Table 6). In patients with only 1 SBM, median overall survival reached 7.4 months (95% CI 5.4–9.2 months) compared to 3.9 months (95% CI 3.3–6.3 months) in those patients with more than 1 SBM (p = 0.029; Fig. 5). Administration of chemotherapy extended median overall survival from 4.9 months (95% CI 3.9–7.2 months) to 5.8 months (95% CI 4.4–8.7 months; p = 0.046).

In the univariate analysis, the only further predictive factor associated with a poor overall survival was tumor histology, with lung and colorectal cancer histologies correlating with a reduced survival (Table 6). Patients’ age, the prevalence of additional extraskeletal metastases, the location of SBM, the concurrent administration of bisphosphonates and the presence of pathological fractures prior to RT were not statistically significant predictors of overall survival (Table 6).