2.1 Study design and setting

A waiver of consent for this single-institution retrospective study was approved by our institutional review board.

2.2 Study selection criteria

The electronic records of 2630 patients receiving treatment for osseous metastatic disease between 2005 and 2021 at our institution were screened. We included adults (age ≥18) with metastatic disease and pathologic fracture of the acetabulum. Metastatic disease of the acetabulum with associated histologic classification was confirmed from core biopsy or surgical pathology reports. An attending orthopaedic oncologic surgeon adjudicated all radiographic examinations for included patients to ensure consensus. Exclusion criteria were age <18 years, non-myelomatous primary tumors of bone, an absence of acetabular metastatic disease or pathologic fracture, insufficient radiographic records, index procedures performed at outside institutions, non-reconstructive primary procedures, and non-neoplastic indications for reconstruction (Fig. 3). Because percutaneous and minimally invasive procedures were infrequently performed for these fractures at our institution, they are not included in the present analysis. This rendered a cohort of 81 pathologic acetabular fractures in 78 patients with acetabular metastatic disease, and a review of patient demographics, primary malignancy histology, location and type of acetabular fracture, oncologic treatment, and disease course was conducted.

Close

Fig. 3 Study selection criteria. ^†Records obtained from the Sidney Kimmel Comprehensive Cancer Center. *Records obtained from the Division of Orthopaedic Oncology at The Johns Hopkins Hospital.

Export to PPT

2.3 Description of treatment

Patients received nonoperative management alone or open reconstruction following conservative management of their pathologic acetabular fracture. Nonoperative fracture management consisted of radiotherapy to the affected hip and/or antiresorptive therapy (e.g., bisphosphonates, denosumab, or both). Radiotherapy was administered for 65 fractures (80%) at a mean (SD) dose of 30 (14) Gy administered in a mean (SD) of 10 (6) fractions. Antiresorptive therapy was used for 64 fractures (79%). For patients with fractures deemed candidates for reconstruction, 33 (97%) trialed at least 1 nonoperative modality prior to surgical reconstruction (Table 1). Patients were considered to potentially benefit from surgery if they could not bear weight, experienced intractable pain, had pelvic discontinuity, worsened with conservative measures, or had an anticipated life expectancy greater than 3 months.¹³ Decision-support tools, such as PATHFx, may augment this determination.²⁴

Open reconstruction consisted of total hip arthroplasty, utilizing the extensive posterolateral approach with the patient in true lateral decubitus. Gross tumor was resected, and the quality of remaining bone stock and deficient pelvic areas were assessed before choosing the type of reconstruction (Table 2). Acetabular reconstruction consisted of 26 antiprotrusio cages (76%) and 8 porous tantalum acetabular cups (24%). A Harrington modification was used in 6 fractures (18%), adding rebar support to the reconstruction with a mean (SD) of 2 (1) retrograde pins. Acetabular cement was used in 30 reconstructions (88%), with intralesional cementation of an acetabular defect or cementing of an acetabular cage or component. There were no intraoperative complications. There were no occurrences of neurovascular injury in either cohort of this analysis. Postoperative complications included five infections and one hematoma requiring irrigation and debridement, four patients with dislocation (three underwent revision for recurrent hardware dislocation), and two patients undergoing revision to exchange one or more components of the acetabular prosthesis. A full list of reoperations is provided in supplemental materials (Table S2).

2.4 Follow-up

Patients were followed postoperatively at 2, 6, and 12 weeks, and records were maintained through subsequent appointments until date of death or most recent contact. At the time of analysis, 14 patients (18%) were still living with a mean follow-up of 3.4 years (range, 1–13 years), with 6 surviving patients (43%) achieving at least 2 years of follow-up. Dates of death were confirmed using public death records. In patients with osseous metastatic disease, median survival is often less than 6 months,²⁵ with fewer alive by 2 years.²⁶ The shorter relative follow-up in our study is clinically relevant for this population.

2.5 Variables and outcome measures

Our primary outcomes included visual analog scale scores for pain, performance status as measured using the Eastern Cooperative Oncology Group (ECOG) scale,²⁷ and independent ambulation. The visual analog scale ranged from 0 to 10, with 0 being no pain and 10 being maximal pain. All mentions of pain refer specifically to the affected hip. The ECOG scale ranges from 0 to 5, with 0 being fully active and 5 being deceased, although the scale has clinical utility only to a maximum of 4, which was employed in our study. Independent ambulation was defined as the ability to walk without the assistance of a wheelchair or hospital staff. These were recorded at admission; discharge; and 2-, 6-, and 12-week follow-up time points. Independent ambulation and pain control of the affected hip were recorded at the time of final contact (Table 3). Secondary outcomes were rates of dislocation and surgical revision for prosthesis loosening or failure (Table S2). Survival from the diagnosis of bone metastases until time of death was also recorded.

2.6 Statistical analysis

All data were analyzed using bivariate statistical analysis and are provided as descriptive statistics. An α of 0.05 was considered statistically significant, with 2-sided tests for all analyses. Student's t-tests were used for continuous variables and chi-squared tests were used for dichotomous variables, with Fisher's exact test for variables with fewer than 5 observations. A multivariable logistic regression with Firth's correction for small-sample-size bias reduction²⁸ was conducted to generate adjusted odds ratios (aOR) to assess for factors associated with clinically meaningful changes, defined as a minimum 3-point improvement in pain at discharge,²⁹ 1-point improvement in ECOG performance status at discharge,³⁰ or restoration of ambulation by final follow-up (Table 4). Firth's correction was necessary, as ORs become inestimable if a model parameter becomes asymptotic with complete or quasicomplete separation of data.²⁸ Clinically relevant variables (patient age, sex, body mass index, histologic cancer classification, visceral or bone metastases, radiotherapy, Harrington classification, fracture displacement, performance status, and open reconstruction) were analyzed in the regression. An exploratory univariable logistic regression is provided in the supplemental materials (Table S3). Kaplan-Meier survival analysis was performed for both treatment courses (Fig. 4A). A log-rank test was conducted to determine equality of survivor functions, and a Cox proportional hazard regression was performed controlling for variables included in the multivariable regression. Competing risk regression of death or acetabular revision was conducted using the cumulative incidence method for patients undergoing open reconstruction (Fig. 4B). Stata SE 17.0 (StataCorp LLC, College Station, TX; 2021) was used to conduct all analyses.

Close

Fig. 4 Fig. 4-A Kaplan-Meier product limit survival curve for patients undergoing open acetabular reconstruction (n = 31) or nonoperative management (n = 47) of pathologic acetabular fractures. Median survival was 1.6 years (95% CI, 0.9–3.2 years) for patients undergoing open reconstruction and 2.0 years (95% CI, 1.1–3.2 years) for patients managed nonoperatively. Log-rank test for equality of survivor functions showed no difference in survival (P = 0.64). Cox proportional adjusted hazard ratio was 1.2 (95% CI, 0.5–2.6) for patients undergoing open reconstruction (P = 0.68). Fig. 4-B Competing-risk regression of surgical revision of the acetabular component or death in patients undergoing open reconstruction (n = 31) as cumulative incidence. The 10-year cumulative incidence of acetabular revision was 15%.

Export to PPT

3.1 Baseline characteristics and survival

In the setting of osseous metastases, 78 patients (37 female patients [46%]) with a median age of 60 years (range, 31–79 years) presented with 81 pathologic acetabular fractures: 34 underwent open reconstruction and 47 underwent nonoperative management (Table 1). The most common histologic class of primary malignancy was carcinoma (62 patients [79%]), most commonly arising from the breast (14 patients [18%]) and kidney (13 patients [17%]). A list of primary malignancy locations is available in the supplemental materials (Table S1). Concomitant visceral and osseous metastases were present in 39 (50%) and 57 (73%) patients, respectively. For Harrington classification of acetabular lesions, 28 (35%) were Class I, 22 (27%) Class II, 21 (26%) Class III, and 10 (12%) Class IV. At presentation, 25 patients (32%) were performance status ≥3 and 36 fractures (44%) were displaced. Open reconstruction was performed more frequently for displaced fractures (P < 0.01), Harrington class III or IV acetabula (P < 0.01), and patients with a performance status ≥3 at hospitalization (P = 0.00) (Table 1). Patients were comparable by age, sex, body mass index, visceral or osseous metastases, and antiresorptive or radiotherapy utilization (Table 1).

Median survival in patients undergoing open reconstruction (n = 31) was 1.6 years (95% CI, 0.9–3.2 years), compared to 2.0 years (95% CI, 1.1–3.2 years) in patients managed nonoperatively (n = 47) (adjusted hazard ratio [aHR], 1.2; 95% CI, 0.5–2.6; P = 0.68) (Fig. 4A). The 10-year cumulative incidence of acetabular revision was 15% (Fig. 4B).

3.2 Pain

Average patient-reported pain levels were higher at admission in those designated for open reconstruction than nonoperative management (8.3 vs. 5.4; P = 0.00), with no difference at discharge (3.6 vs. 3.8; P = 0.82) (Table 3, Fig. 5A). Following open reconstruction, patients reported a larger decrease in pain, on average, after hospitalization (−4.6 vs. −1.7; P = 0.00), with more patients experiencing a clinically meaningful reduction (P = 0.03).²⁹ Radiotherapy (aOR, 23.3; 95% CI, 2.8–194.2; P < 0.01) and open reconstruction (aOR, 8.3; 95% CI, 1.4–50.6; P = 0.02) were associated with clinically meaningful reductions in pain at discharge (Table 4). No differences were noted between groups for average pain ratings at follow-up (Table 3; Fig. 5A).

Close

Fig. 5 Plots of mean (Fig. 5-A) visual analog scale scores (0–10) for pain and (Fig. 5-B) ECOG performance status scores (0–4) for patients with fractures undergoing open acetabular reconstruction (n = 34) or nonoperative management (n = 47). Standard error of the mean is depicted as vertical error bars. Frequency of observations provided in parentheses above each point. ECOG, Eastern Cooperative Oncology Group.

Export to PPT

3.3 Performance status

In patients undergoing reconstruction, mean performance status was worse at admission (2.8 vs. 1.4; P = 0.00) and discharge (2.2 vs. 1.8; P = 0.04) compared with those undergoing conservative management (Table 3; Fig. 5B). Higher scores translate to decreased performance. A larger average improvement was observed at discharge following open reconstruction (−0.6 vs. +0.3; P = 0.00). We adopted the definition of clinically meaningful as a 0.5-point average change in aggregate performance status³⁰; thus, the average improvement following reconstruction was meaningful. Because the ECOG scale is integer-based, we defined a 1-point change as clinically meaningful on an individual-patient basis. More individual patients experienced a clinically meaningful improvement in performance status following open reconstruction than following nonoperative management (21 [62%] vs. 6 [13%]; P = 0.00) (Table 3). Performance status ≥3 (aOR, 5.0; 95% CI, 1.3–19.7; P = 0.02) and open reconstruction (aOR, 10.9; 95% CI, 11.7–71.0; P = 0.01) were associated with clinically meaningful improvements in performance status at discharge (Table 4). Two weeks after discharge, patients who had undergone open reconstruction had worse mean performance status than those treated conservatively (2.2 vs. 1.7; P = 0.02), with no differences at 6-week (P = 0.25) or 12-week (P = 0.39) follow-up (Table 3; Fig. 5B).

3.4 Ambulation

At admission, fewer patients undergoing reconstruction compared to nonoperative management could ambulate (6 [18%] vs. 36 [77%]; P = 0.00), but by final follow-up, there was no significant difference (24 [71%] vs. 38 [81%]; P = 0.28) (Table 3). Open reconstruction (aOR, 7.5; 95% CI, 1.9–29.8; P < 0.01) was associated with a restoration of ambulation by final follow-up (Table 4).

3.5 Subgroup analyses

In comparing patients with similar degree of acetabular involvement, more Harrington Class III and IV acetabular lesions had an improvement in pain (no change/worsened: 45% [n = 14] vs. improved: 55% [n = 17]), marginally fewer had an improvement in performance status (no change/worsened: 52% [n = 16] vs. improved: 48% [n = 15]), and marginally more improved in ambulatory status (no change/worsened: 48% [n = 15] vs. improved: 52% [n = 16]). For patients with radiosensitive lesions, marginally fewer experienced an improvement in pain (no change/worsened: 51% [n = 20] vs. improved: 49% [n = 19]), fewer had an improvement in performance status (no change/worsened: 67% [n = 26] vs. improved: 33% [n = 13]), and fewer improved in ambulatory status (no change/worsened: 72% [n = 28] vs. improved: 28% [n = 11]). For patients with chemosensitive lesions, more experienced an improvement in pain (no change/worsened: 44% [n = 22] vs. improved: 56% [n = 28]), fewer had an improvement in performance status (no change/worsened: 66% [n = 33] vs. improved: 34% [n = 17]), and fewer improved in ambulatory status (no change/worsened: 68% [n = 34] vs. improved: 32% [n = 16]).