A total of 143 records were identified through database searching and cross referencing retrieved articles. After removing duplicates, 123 records were screened for eligibility (Fig 1). After reviewing the title and abstracts, 55 records were excluded. From the remaining 68 records, full-text articles were reviewed for eligibility and further 19 were omitted. Amongst the 19 excluded articles, 16 did not report on any outcomes for specific treatment of pelvic ABC and 3 had a follow up of less than one year.
The final number of studies included in the systematic review was 49, of which 29 were case reports and 20 were case series, spanning from 1979 to 2017. From the 49 studies, data of 194 patients was extracted into an excel sheet.
Treatment groups and patients included per study
A large number of treatment options was identified and grouped into 5 major treatment groups: observation, adjuvant, minor surgery, major surgery and combined. These major treatment groups were further subdivided into eleven distinct treatment subgroups. The number of studies and patients included in each treatment group are described in Table 1. Some studies appear in more than one treatment group.
Within the 5 major treatment groups, more than half of the patients 113/194 (58%) were treated either with adjuvant 57/194 (29%) or minor surgery 56/194 (29%) whereas observation was the least used treatment accounting for only 13/194 ( 7%) patients.
Looking at treatment subgroups, most patients were treated with curettage 45/194 (23%) followed by selective arterial embolization (SAE) 38/194 (20%). Amongst the latter, 18 (47%) patients had only one SAE session, 10 (26%) patients had two and further 10 patients had three sessions of SAE. Diverse biomaterial like gel foam, latex particles, biospheres, Isobutyl 2 cyanoacrylate (IBCA), polyvinyl alcohol (PVA) and stainless-steel coils were used for the embolization.
Marginal excision was the third most commonly used treatment 27/194 (14%) whereas wide excision was used only in 5 cases. The terms marginal and wide derive from the 'Enneking classification of surgical margins' which is a standard reported measurement of surgical margins in osseous and soft tissue tumor surgery . Various materials have been described to fill the bony voids after major or minor surgery like autologous bone grafts, allografts, bone cement, bioactive glass, plexur M biocomposite, arthrodesis with plating and vascularized fibular graft.
The combined treatment groups included any type of surgery along with either SAE, sclerotherapy or radiotherapy. The combined group minor surgery + adjuvant was the fourth most used treatment (26/194). Most of these patients received SAE as neoadjuvant treatment.
Extended curettage was used in 11/194 cases and reportedly done through high speed burr, phenol, or ethanol washes. Sclerotherapy (10/194 cases) was either performed percutaneous or through a mini open approach under fluoroscopy and general anesthesia. The sclerosants used were ethibloc, demineralized bone matrix (DBM) combined with autologous bone graft, 32 P chromic phosphate and polidocanal. The least used methods included radiotherapy (5/194), wide excision (5/194) and SAE with sclerotherapy (4/194). Radiotherapy has been used as neoadjuvant but also as adjuvant and treatment for recurrence. The maximum dosage received by any patient was 6000 Gy (Gray).
Demographics, radiological and functional outcomes were summarized according to the treatment subgroup.
Patient demographics per treatment subgroup are described in Table 2. From the 194 patients, 91 (47%) were females, 101 (52%) were males and gender was not reported for 2 (1%) patients. The overall median age was 15 years, ranging from 1.5 to 64 years old. The median age among the different treatment subgroups was very similar and within the second decade of life except for the wide excision group.
The mean time from the start of the symptoms to diagnosis was 10 months with a range from 2 weeks to 10 years. The median follow up was 3 years, with the longest follow up of 53 years in a 10-year-old with stage 2 ABC in the sacrum treated with extended curettage and radiotherapy .
Main tumor characteristics in terms of size, anatomic location within the pelvis and Enneking staging according to treatment group and subgroup are described in Table 3. Only 8 (4%) patients presented with a recurring ABC and the rest were all primary de novo lesions.
The information on tumor size for individual patients was available for only 50/194 patients (25%); but for 22 (11%) patients, tumor size was described in categories of less than, up to, or greater than 5 cm . The minimum reported tumor size was 3 cm and the maximum 30 cm. Treatment groups with an average size above 15 cm were all surgical (either as standalone treatment or combined with adjuvant therapy) whereas tumors in the observation group were on average 4.5 cm and were followed up with a median of 2.8 years.
Most tumors (126/194) were located within one pelvic zone, with a majority of them in the ischio-pubic area (zone 3). Only 13/194 occupied three zones: ilium, acetabulum and ischiopubic area in all except one case (zones 1+2+3). The most frequent combination for tumors within two zones was acetabulum and ischiopubic area (zones 2+3) followed by ilium and acetabulum (1+2), and ilium and ischiopubic (zones 1+3). Tumors involving both ilium and sacrum (zones 1+4) were found only in 4 cases.
The Enneking staging divides benign bone tumors into 3 categories: latent (stage 1), active (stage 2), and aggressive (stage 3). Over 50% of the cases were Enneking stage 3 and 30% were stage 2. The staging was not mentioned for 30 (15%) patients.
Radiological and clinical outcomes per treatment group are detailed in Table 4. Given that the radiological and functional outcomes reported in the retrieved studies were remarkably inconsistent, heterogenous and in occasions extremely vague, we tried to categorize the outcomes following our best interpretation and understanding of what was reported by the authors. Only one study  described the outcomes using the musculoskeletal tumor rating scale (MSTS), Toronto extremity salvage score (TESS) and the 36 item short form health survey (SF-36).
Our primary outcome focused in the ossification. In total,116 patients (60%) had complete ossification, 19 patients (10%) had partial and 5 patients (2%) had no evidence of ossification. The ossification status was not available in 20/194 cases: in 18 patients (9%) it was not reported, and two patients died during or shortly after surgery.
Clinical outcomes related to function, pain and recurrence. Patients were reported as functional in 140 (72%) cases, mild functional derangement was seen in 11 patients (6%), moderate in 6 (3%), severe in 4 (2%) and in 41 (21%) the functional outcome was not reported.
At the end of the follow up, 112 patients (58%) had reportedly no pain, whereas mild pain was mentioned in 16 (8%) patients, moderate pain in 5 (2%) patients, and in 50 patients (30%) pain was not reported.
Overall, the incidence of recurrence was low. Recurrence was reported in 21 (10.8%) patients and not reported in 15 patients (8%), whereas the remaining patients had no recurrence at the time of the last follow up. The longest time to recurrence was 6 years in a 13-year-old male patient with an ABC in the pubis extending to acetabulum treated with extended curettage . Recurrence was treated mostly by curettage or excision and autologous bone grafting. Few recurrent cases were managed with sclero-embolic treatments instead of surgery and 3 cases were treated with radiotherapy[23,24]. Curettage had a relatively high number of recurrences compared to the other treatments, however the unbalanced group size and low number of patients in some subgroups precludes a proper statistical analysis.
The number of reported complications is low. Hemorrhage was reported in 6 (3%) patients who were treated with surgery (5 patients) and sclerotherapy (1 patient). Of these, one patient died secondary to massive bleeding. Neurological complications were described in 13 (7%) patients and include hypoesthesia, bowel and bladder dysfunction or residual loss of bowel, bladder control, neuroma formations, and involvement of sacral roots, sciatic nerve, obturator nerve and caudal sac. 11 (6%) patients with neurological complications had ABC located in zone 4 (sacral) while in the remaining 2 were in zones 2 and 3. One patient with a tumor in zone 2 and 3 had partial encasement of the bladder with the ABC mass resulting in subtotal cystectomy. Leg length discrepancy (0.6 – 3cm) was reported in 4 patients (2%) after marginal resection of ABC. Curettage in the sacrum (pelvic zone 4) was performed in 12 (26%) cases and had frequent reports of hemorrhage, bowel injury, dural tear and loss of bowel bladder control. 9/21(42%) patients had complications after treatment for recurrence. Capana et al reported 3 of these cases in which the primary treatment was radiotherapy or combination of radiotherapy with surgery, recurrence was observed and treated with curettage or radiotherapy. These patients were left with residual complications of crural nerve palsy, quadriceps, psoas deficit and femoral head necrosis.
In this series of 194 cases reported in the literature, two patients died of treatment-related complications. A 51-year-old female with a 25 x 25 x 20 cm ABC in the right ilium extending into the ischium. The patient was treated surgically and died of profuse perioperative bleeding . The second case was a 48-year-old female with a 27.6 x 22.4 x 15.9 cm lesion in the left ilium with a history of embolization, sclerotherapy and repeated surgery. The authors performed preoperative embolization and percutaneous intralesional alcohol injection followed by staged near total resection combined with cryosurgery and off label systemic treatment with bevacizumab, a monoclonal anti-vascular endothelial growth factor (VEGF) antibody, to reduce bleeding. The patient developed clinical heart failure, prompting discontinuation of bevacizumab followed by a rapid regrowth of the pelvic lesion. After seven surgical sessions, the patient's condition no longer allowed further interventions and she was treated with radiotherapy. The patient died 2 years after ABC was first diagnosed .
Aneurysmal bone cyst in the pelvis has also been described in 2 pregnant women. A pelvic bone lesion in pregnancy can be a difficult situation for the obstetrician to determine the effect of parturition on the pelvic bone. The first reported case was in 1986 of a right iliac aneurysmal bone cyst measuring 15 cm in a 19-year-old primigravida. At the time of the elective caesarian, the mass was excised through a separate lumbar approach. No recurrence or complication was reported. The second reported case was a 6.1 x 2.9 x 6 cm lesion in the left ilium of a 26-year-old primigravida who had a successful elective caesarian section and the ABC was observed until the last follow-up.
We further analyzed tumor characteristics, clinical and radiological outcomes according to their pelvic zone location and the results are summarized in Table 5. While zone 3 was overall the most common tumor location 48/194 (25%), when taking in consideration tumors with overlapping zones as well, most patients 74/194 (38%) had affection of zone 1 – ilium; of these 14 had partial ossification and the functional outcome was not reported in half of them. Tumors in zone 4 sacrum had more neurological complications reported like bladder and bowel dysfunction. Moderate residual pain was mentioned in 4/194 patients located in zones 4, 2+3 and 1+2+3. Moreover severe FD reported in 4/194 were located in zone 4 (3 cases) and zone 1+3 (1 case).