The demographic variables included were age at the time of surgery and gender. There were 41 men and 42 women with a mean age of 55 years (range, 12 to 72 years) (Table 1). For the purpose of analysis, patients were dichotomized based on their age into ≤ 55 years (n = 42, 51%) and > 55 years (n = 41, 50%). Preoperative blood parameters included were haemoglobin level, white blood cell count and albumin level. The mean preoperative haemoglobin level was 122 g/L (range, 85 to 166 g/L), white blood cell count was 5.8 × 10⁹/L (range, 1.8 × 10⁹/L to 11 × 10⁹/L) and albumin level was 39 g/L (range, 26 to 50 g/L). Bone metastasis was the initial presentation of cancer in 25 patients (31%). Thirty-two patients (39%) had bone metastases at presentation. The median time from the diagnosis of primary cancer to bone metastasis was 3 years (range, 4 to 306 months). The diagnosis of pelvic bone metastasis was made based on a biopsy in 23 patients (28%) and on imaging in the rest.

The cancer-related variables included were primary cancer type, number and location of bone metastasis, size of the MBD, presence of skip lesions, soft tissue extension, pathological fracture and presence and location of visceral metastasis. The most common primary cancers were of the kidney (n = 18, 22%) and thyroid (n = 13, 16%). For the purpose of analysis, the primary cancer types were grouped into favourable types (kidney, breast, thyroid or hematolymphoid) and unfavourable types.1213) Pelvic bone metastasis was solitary in 23 patients (28%). Fifty percent (n = 41) of the patients had more than one site of bone metastasis. The spine (66%) was the common location when compared to the extremities (upper limb, 7%; lower limb, 11%) in patients with multiple bone metastases. The most common location in the pelvic bone was the iliac wing (n = 42, 51%). The size was documented as the largest diameter from the pathology report in patients who underwent en bloc resection or from cross-sectional imaging in other patients. For the purpose of analysis, patients were dichotomized based on the tumor size with a cut-off value of 6.8 cm, the mean size of the tumor (range, 2 to 16 cm). Fifteen patients (19%) had skip lesions and 60 patients (60%) had soft tissue component of the metastatic disease. Twenty-three percent of the patients had a pathological fracture (n = 19). Visceral metastases were present in 43 patients (52%), lung being the most common site (n = 32). The radiographic pattern was lytic in 88% (n = 72) and normal in 7% (n = 6).

The treatment-related variables included were performance of preoperative embolisation, administration of radiation, American Society of Anesthesiologists (ASA) physical status, surgical margin, type of reconstruction, histological margin, duration of surgery, and blood loss. Preoperative embolisation was done in patients with hypervascular tumors. Embolisation was usually done with gelfoam particles within 24 hours prior to the surgery (Table 2). Preoperative embolisation of the tumor was done in 53% of the patients (n = 43). Twenty-one patients (26%) had received radiation therapy previously. Eleven patients (17%) were class I, 39 (62%) were class II, 12 (19%) were class III and 1 (2%) was class IV, based on the ASA physical status. The decision to perform surgery was taken in a multidisciplinary tumor board on an individual patient basis. Surgery was performed for palliation in a majority of the patients, but also with the intent to reduce tumor burden and cure in a few. No standard criteria were followed, but en bloc resection was generally attempted in patients with metastases from a favourable primary site, solitary metastasis, where feasible, and in whom longer life expectancy was expected. The surgical margins were rated as en bloc in 32 (39%) and curettage in 48 (58%). The mean time from the diagnosis of bone metastasis to the surgery on the pelvic bone was 1 month (range, 0 to 10 months). The method of reconstruction was based on the location of the metastasis. No reconstruction was performed for the metastases involving the pubis or the ischium. Cement with or without Steinman pins was used to reconstruct the ilium. Cemented total hip replacement arthroplasty (THR) with Burch-Schneider Reinforcement cage was used to reconstruct the acetabulum. Only cement was used to reconstruct the pelvic bone in 26 patents (33%), cement and Steinman pins were used in 21 patients (26%) and no reconstruction was performed in 21 patients (26%). Cemented THR was used to reconstruct the hip in 12 patients (15%) which included pasteurised autograft in 2 patients. Histological margins were identified using the pathology report. Negative histological margins were achieved in 19% (n = 13). The mean blood loss was 2.74 L (range, 0.2 to 30 L). The mean duration of surgery was 160 minutes (range, 30 to 560 minutes). In general, postoperative follow-up evaluations were performed at 2 weeks after surgery and then with 3 to 6 months of intervals. The follow-up schedule and modality varied according to each patient's condition. Postoperative radiation was administered to 27 patients (33%). Local recurrences were detected using the available imaging modalities such as X-ray, computed tomography (CT), and magnetic resonance imaging (MRI) scans and positron emission tomography (PET) scan.

The outcome measures of this study were complications of surgery, local tumor progression and survival. Complication was defined as any deviation from the ideal postoperative course that was not inherent in the procedure and that did not comprise local progression.14) Local progression was defined as appearance of a new lesion or radiological progression of the existing lesions during follow-up by means of an X-ray, CT scan, MRI scan, PET scan or a bone scan. A patient with no signs of local recurrence at the last imaging was considered censored, meaning that the outcome did not occur by the last moment of follow-up. We only considered the first notice of local recurrence in our time-to-event analyses. Survival was defined as the period between the date of surgery and the date of death.

We presented continuous variables as means with standard deviations and categorical variables as frequencies with percentages. We used a chi-square test for categorical variables. We used log-rank analysis to compare local progression and survival between groups. We used Kaplan-Meier plots to demonstrate the outcomes over time and calculate 2-year, 3-year, and 5-year probabilities. A p-value of less than 0.05 was considered to indicate statistical significance. Statistically significant variables on univariate analysis were subjected to multivariate analysis using Cox regression models. Statistical analyses were performed using the IBM SPSS ver. 21.0 (IBM Co., Armonk, NY, USA).

In all, 13 patients (16%) developed 15 complications. Complication rates were higher in patients who were above the age of 55 years (25% vs. 7.5%, p = 0.034), had albumin levels ≤ 39 g/L (30% vs. 2.5%, p = 0.001) and underwent THR for periacetabular disease than those in whom THR was not done (5/12 [42%] vs. 1/15 [7%], p = 0.030).

The median duration of hospital stay was 14 days (range, 0 to 116 days). Three patients died within 1 month of the surgery; two of them within 24 hours of surgery, one of whom sustained severe blood loss and developed disseminated intravascular coagulation. Six patients developed surgical site infection, which was the most common complication in our series. Five of them developed early surgical site infections, of which 4 needed surgical debridement and one was treated with antibiotics. Surgical debridement was done three times in 1 patient who later developed infectious spondylitis, twice in 1 patient and once in 2 patients. Late surgical site infection was seen in 1 patient who ended up having resection arthroplasty. One patient developed postoperative pneumonia treated with antibiotics. Dislocation of the hip occurred in 2 patients who underwent THR, one of which was managed with closed reduction whereas open reduction was needed in the other patient in whom dislocation was subsequently noted at the last follow-up. Two patients needed postoperative embolisation to control bleeding from the surgical site, one of whom had preoperative embolisation more than 48 hours prior to the surgery. In the other patient, embolization was required due to pseudoaneurysm of the external iliac artery. Electrolyte imbalance and scrotal swelling was noted as a complication and corrected in one patient.

The local progression rate for the series was 46% (n = 37). The median time to local progression was 26 ± 14.3 months (Fig. 1). The median time from local progression to death was 13 months (range, 0 to 81 months). The 2-year and 5-year local progression-free survival were 52.6% ± 6.4% and 36.4% ± 7.6%, respectively. The local progression rate was higher in patients with visceral metastases (57% vs. 33%, p = 0.032). This might be due to the higher rate of en bloc resection with negative margin in patients without visceral metastases. However, neither the surgical margins nor the histological margins were found to have a significant influence on the local progression of the disease.

Local progression was identified with a sophisticated imaging modality such as CT, MRI or a PET scan in the majority, whereas local progression identified on X-ray was confirmed with one of the sophisticated imaging in 2 patients. Biopsy was not performed to confirm local progression. Local progression was seen in the tumor bed or in the tumor bed margin in 27 patients (73%), away from the tumor bed in 6 patients (16%) and in the soft tissue in 4 patients (11%).

Univariate analysis of possible factors associated with local progression revealed that the absence of skip lesions (p = 0.006), absence of visceral metastasis (p = 0.009) and metastases from the breast, thyroid or hematolymphoid malignancies (p = 0.031) were significantly associated with longer local progression-free intervals (Table 3). However, on multivariate analysis, only the presence of skip lesions (p = 0.017) and the presence of visceral metastasis (p = 0.027) were found to be significant factors.

Local progression necessitated treatment in 29 patients. Eight patients underwent repeat surgery and 7 received systemic treatment. In all, 17 patients received radiation therapy including 2 patients who underwent repeat surgery and 1 patient who received systemic treatment. Retreatment rates were significantly higher in patients with single bone metastasis (p = 0.013) and in patients with initial tumor size ≤ 6.8 cm (p = 0.021). The mean survival time from local progression was longer in patients who received retreatment than those who did not receive retreatment (17 months vs. 5 months).

At the time of analysis, 55 patients (66%) had died of disease and 28 were alive. The median survival of all patients was 24 months. The 2-year and 3-year survival rates were 52.5% ± 5.9 % and 35.6% ± 6 %, respectively. For the 55 patients who died of disease, the mean time from surgery to death was 22 months (range, 0 to 95 months) (Fig. 2). The mean follow-up of the 28 survivors was 36 months (range, 2 to 145 months).

On univariate analysis of possible factors associated with survival, favourable primary cancers (of the kidney, breast, thyroid or hematolymphoid) (p = 0.014), absence of visceral metastases (p = 0.008), and albumin levels of more than 39 g/L (p = 0.003), were significantly associated with a better survival (Table 4). On multivariate analysis, all the 3 variables retained statistical significance: favourable primary cancers with a p-value of 0.014, absence of visceral metastasis with a p-value of 0.017 and albumin levels of more than 39 g/L with a p-value of 0.009.