Evaluating fixation and bone graft integration at two years post-surgery in uncemented acetabular revisions with large amounts of allograft bone

Abstract

Aims

Impaction bone grafting of the acetabulum to treat segmental and cavitary defects has been shown to be successful for uncemented acetabular revisions. Concerns remain about graft integration and implant stability when severe bone defects require large amounts of bone graft. This study evaluates bone graft density and implant migration in patients undergoing uncemented acetabular revision with screw fixation and impaction bone grafting using large bone graft volumes.

Methods

In this exploratory study, nine patients undergoing acetabular revision surgery were evaluated to assess bone graft volume, bone mineral density (BMD), and implant migration using dual-energy CT and CT-based micromotion analysis, performed directly postoperatively and at 6, 12, and 24 months of follow-up.

Results

The mean volume of bone graft used was 64.0 cm³ (SD 36.9) and the BMD in the graft increased from a mean of 317 mg/cm³ (SD 96.4) postoperatively to 466 mg/cm³ (SD 104.2) at 24 months (p = 0.002), while iliac BMD changed from 96 mg/cm³ to 111 mg/cm³ (p = 0.258). The median total translation at 24 months was 1.6 mm (IQR 0.82 to 2.7). Cups with graft volumes ≥ 50 cm³ exhibited slightly larger proximal translation than those with < 50 cm³, with median 1.5 mm (IQR 1.4 to 1.6) compared with 0.6 mm (IQR 0.30 to 1.2).

Conclusion

Uncemented acetabular revisions using large allograft volumes exhibit a migration pattern that is consistent with other revision techniques. Despite the large amount of bone graft used, bone density increases over time.

Cite this article: Bone Jt Open 2026;7(2):266–274.

Introduction

The use of impaction bone grafting (IBG) in acetabular revisions is an established technique with well-documented long-term success in several independent case series.^1-4 The technique involves filling of the cavitary and segmental defects, as well as the entire enlarged acetabular cavity, with impacted morselized bone graft.⁵ In IBG, the impacted bone works as a void filler, and the transplanted bone restores the deficient bone stock when remodelled.^6-10

Controversy exists as whether large bone defects can be treated with this technique in uncemented revisions, where the IBG technique involves bone grafting of large defects (≥ one femoral head allograft) and limited host bone contact.¹¹ For a successful revision arthroplasty, the implant should have at least 50% surface contact with intact host bone.¹² However, there is still debate as to the required extent of contact and the level of implant stability required to achieve osseointegration when large amounts of bone graft are used.^5,12 Several studies have reported a bone grafting technique for uncemented acetabular revisions that does not adhere to the 50% host bone paradigm but still has excellent mid- and long-term clinical outcomes.^3,13,14 How the bone graft is remodelled and the magnitude of implant motion after these types of surgeries have not been rigorously evaluated using modern CT technologies.

Modern visualization techniques make it possible to assess indirectly bone-healing processes in vivo, in place of the traditional gold standard histological and micro-CT techniques.^6,15-19 Bone graft healing can be assessed by measuring changes in bone density as a proxy for bone formation in the bone graft.²⁰ The dual-energy high-resolution CT technique enables spatial assessments of bone density,²¹ providing more-reliable information than the traditional dual-energy radiograph analysis, which is constrained by its 2D resolution.²² Another important proxy of bone graft healing is implant migration. Radiostereometric analysis (RSA) has been used for decades to assess implant stability, and increased implant migration is associated with a higher long-term risk of revision.^23,24 Recently, high-precision implant migration measurements using clinically available CT examinations, even without implanted reference markers, have become available; for example, CT-RSA.^25-27 The feasibility of using these techniques in the follow-up of impaction bone grafting of the acetabulum has recently been demonstrated for cemented revisions.²⁰ In the context of uncemented acetabular revisions using large amounts of bone graft, these techniques have not yet been employed to address the still unresolved questions of bone graft integration and implant stability.

The aim of this study was to evaluate whether bone mineral density (BMD) in the grafted bone increases over time, to quantify implant migration when large volumes of bone graft are used, and to determine the influence of graft volume on implant migration in uncemented acetabular revisions.

Methods

Patients

This retrospective study was performed at Linköping Universit Hospital, which is a tertiary referral centre for complex hip arthroplasty surgery in the South-eastern Health Care district of Sweden.²⁸ We conducted a screening of the Swedish Arthroplasty Register (SAR) for patients who had undergone revision total hip arthroplasty at our institution (procedure codes NFC 09-99) from April 2018 to May 2020, following the implementation of routine pre- and postoperative CT examinations for complex hip revisions in 2018. During this period, 106 patients underwent reoperations after total hip arthroplasty. Of these, 46 patients underwent implant revision and were selected for a manual review of their medical records and radiological examinations (Figure 1). The inclusion criteria were: CT examinations directly after the surgery and at 24 months follow-up; operated with an uncemented hemispherical cup with IBG; and at least one femoral head bone allograft was used for IBG. The final cohort comprised nine patients who underwent acetabular revision surgery with an uncemented hemispherical cup and large amounts of IBG (Table I). Bone defects were classified according to Paprosky based on radiological assessment.¹¹

Fig. 1.

Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) diagram of patient inclusion in the study. BMD, bone mineral density; DAIR, debridement, antibiotics, and implant retention.

Table I.

Patient characteristics (n = 9).

Variable	Data
Female, n (%)	5 (55.6)
Median age, yrs (IQR)	75.0 (63.0 to 84.0)
Median BMI, kg/m2 (IQR)	25.1 (23.5 to 26.4)
ASA grade, 29 n (%)
I	2 (22.2)
II	4 (44.4)
III	3 (33.3)
Cause of revision, n (%)
Aseptic loosening	8 (88.9)
Pseudotumor and osteolysis	1 (11.1)
Paprosky classification, n (%)
2A	2 (22.2)
2B	1 (11.1)
2C	1 (11.1)
3A	1 (11.1)
3B	4 (44.4)
Mean bone graft used, cm3 (SD; range)	64.0 (36.9; 31.7 to 144.7)
Implant used, n (%)
Trident HA multihole	1 (11.1)
Trident tritanium	8 (88.9)
Median cup size, mm, (IQR)	58 (50 to 66)
Median number of screws (IQR)	5.00 (4.00 to 5.00)
Mean blood loss, ml (SD)	467 (332)
Femoral component revised, n (%)	3 (33)
Head size, mm, n (%)
28 (in dual-mobility)	4 (44)
32	2 (22)
36	3 (33)
Bearing, n (%)
Ceramic-PE	5 (56)
CoCr-PE	4 (44)
Previous surgeries, n (%)
1	8 (89)
3	1 (9)

ASA, American Society of Anesthesiologists; CoCr, cobalt-chrome; HA, hydroxyapatite; PE, polyethylene.

Surgeries

All surgeries were performed by two senior orthopaedic surgeons (JS, LP) specialized in hip revision surgery, using a posterior approach and standardized bone-grafting protocol.⁴ Briefly, the acetabulum was prepared with hemispherical reamers before trialling with the smallest acetabular shell that, when possible, allowed press-fit contact with host bone between at least two opposing sections of the remaining acetabulum. The deformed acetabular cavity was then filled with bone graft so as to form a hemispherical bone graft bed for the implant. All grafts used were fresh-frozen, nonirradiated, allogenous femoral heads harvested according to institutional standard practice from patients undergoing primary hip arthroplasty surgery and stored at -70°C. During surgery, morselized graft was prepared from the allograft femoral head using a bone mill (Howex, Sweden) to produce bone chips between 3 mm and 5 mm in size which were then repeatedly rinsed in saline. The impaction procedure involved impaction with the backside of the hammer handle into cavities, followed by impaction using a reamer undersized by 2 mm compared with the final implant. The final implant was then seated and secured in position, first with screws directed inferiorly. Additional bone graft was impacted in the gap between the host bone and implant before more screws were inserted in a crisscross pattern to compress evenly the bone graft bed while minimizing changes in implant position. To facilitate screw fixation in multiple directions, we used a specific 3D digital planning protocol and software, Sectra 3D Joint arthroplasty (Sectra, Sweden).³⁰ All patients were operated on with two different versions of the Trident Acetabular Revision System (Stryker Orthopaedics, USA) (Table I). In one case, the metal cup was placed in a position to optimize fixation in the bony socket and a Link acetabular component (Waldemar Link, Germany) was cemented into the revision shell, to optimize anteversion and inclination of the cemented cup. In another case, a metal mesh was secured to the posterior aspect of the acetabulum, to ensure containment of the bone graft.

Graft volume calculation

The open-source software 3D slicer³¹ was used to segment the IBG and to calculate the bone graft volume (cm³) on available postoperative CT-scans. The borders of the bone graft were defined and marked manually to distinguish between the surrounding host bone and the implant on every fifth CT slide.²⁰ The software was used to approximate the IBG volume in the remaining slides and manually corrected when needed. No correction was made for the cancellous bone screws passing through the IBG. This evaluation was performed by two reviewers (JB, JSk) and consensus was reached for each case.

Bone mineral density analysis

Bone mineral density (BMD) and implant motion were assessed using dual-energy CT postoperatively and at six, 12, and 24 months of follow-up for eight patients. The CT scans were performed with routine clinical settings using 128-section, dual-source CT scanners (Somatom; Siemens Healthcare, Germany). BMD measurements (mg/cm³) using 100-kV scans without metal artifact reduction protocols. For the BMD analysis, the Mindways qCT PRO v. 2 software (Mindways, USA) was used, as described previously.²⁰ In brief, calibration scans with the Mindways’ dedicated phantom were performed with the clinically used CT protocols before analysis (institutional routine). Regions of interest (ROIs) were positioned manually in: 1) the IBG apical to the cup; and 2) native surrounding bone in the ilium bone apical to the IBG (Figure 2). BMD was measured directly after surgery, and at six, 12, and 24 months postoperatively.

Fig. 2.

Regions of interest were manually positioned in the qCT PRO software. Shown are the bone graft apical to the cup (dotted line) and host bone in the ilium apical to the bone graft (solid line).

Implant motion analysis

For CT-based RSA (CTMA; Sectra), imaging parameters from available CT scans that were closest to the following settings were used: tube voltage in range of 100 kVp to 150 kVp with tin filtration; tube current, 69/32 mAs; slice thickness, 1 mm; increment, 0.675 mm; pitch, 0.6 rotation time, 0.5 s; and collimation, 32 × 0.6 mm. Metal artifact reduction algorithms were used. Evaluation of all CT scans was performed by a certified biomedical scientist at Sectra AB. Before the analysis, the most-suitable settings for registration of the bone were identified at 100 Hounsfield units (HUs) and at 2,200 HU for the implants in this patient cohort. Using the CT-RSA software, two rigid bodies were defined: 1) the pelvic bone as the reference rigid body; and 2) the acetabular component implant as the migrating rigid body.³² Migration of the acetabular component over time was obtained in 6° of freedom (translations along and rotations around the X, Y, and Z axes). In this report, all measurements of translation are related to the centre of rotation (CoR) of the cup relative to the pelvic bone between immediately postoperative (mean 3.1 days (SD 2.3)) and 24 months postoperatively. We present values for total translation and proximal translation, to allow comparisons with previous studies.²³ To estimate the intraclass correlation for absolute agreement, data analysis for a subset of randomly selected patients was repeated after six weeks and calculated using the intraclass correlation (ICC 2.1). The measurement of maximum total point motion was not available in the CT-RSA system at the time of the analysis. Bone graft volume measurements, measurements of bone density, and CT-RSA were performed blinded for any background parameters.

Clinical outcome

When the information was available in the patient records, patient-reported hip function was assessed preoperatively and at six, 12, and 24 months postoperatively according to the Merle d'Aubigné-Postel rating scale.^33,34

Statistical analysis

Descriptive data are presented as either means (SD) or medians (IQR), depending on the distribution of the variables. Comparisons between groups were made using paired t-tests for bone graft volume and the Mann-Whitney U test for cup migration, following an evaluation of the data distribution. The level of significance was set as a two-sided p-value of 0.05. All statistical analyses were conducted using the SPSS Statistics v. 27 software (IBM, ISA).

Ethics and dissemination

The study was approved by the Swedish Ethical Review Authority (Dnr: 2021-04684) and performed in compliance with the Declaration of Helsinki.³⁵

Results

A mean of 64.0 cm³ (SD 36.9) of bone graft was used. In all, four of the patients had Paprosky type 3B defects and were treated with on average 87.9 cm³ (SD 45.6) of bone graft to fill the whole acetabular cavity. In three of these patients, ≥ 50 cm³ bone graft was used.

Changes in bone mineral density

The BMD in the graft increased from 317.1 mg/cm³ (SD 96.4) postoperatively to 465.6 mg/cm³ (SD 104.2) at 24 months, p = 0.002. The BMD in the surrounding native bone changed from 95.5 mg/cm³ postoperatively to 110.8 mg/cm³ at 24 months (p = 0.258) (Table II).

Table II.

Bone mineral density (mg/cm³) measurements over time.

	Bone graft, mean (SD)				Native bone, mean (SD)
Variable	N	Postop	24 months	p-value	Postop	24 months	p-value
All patients	8*	317.1 (96.4)	465.6 (104.2)	0.002	95.5 (35.7)	110.8 (58.0)	0.258
≤ 50 cm3	4	363.5 (106.1)	449.3 (140.3)	0.023	70.8 (24.3)	78.6 (36.3)	0.537
> 50 cm3	4	270.6 (68.4)	482.0 (70.1)	0.012	120.3 (27.5)	143.1 (61.4)	0.405

^* For one patient, bone mineral density measurements at 24 months were not available.

Implant motion 

The median total translation was 1.6 mm (IQR 0.82 to 2.7) and the median proximal translation, the parameter typically presented in the literature, was 1.4 mm (IQR 0.58 to 1.9) (Figure 3). Graft volumes ≥ 50 cm³ were associated with slightly larger proximal translation than those with < 50 cm³, median 1.5 mm (IQR 1.4 to 1.6) compared with 0.6 mm (IQR 0.30 to 1.2), p = 0.063 (Table III). The intraobserver correlation of CT-RSA measurements was excellent, ICC 0.99. There was one patient with proximal translation > 2.29 mm after six months, which increased to 2.97 mm at 24 months (Figure 3). This patient had not undergone any revision procedure and there were no signs of loosening on CT exams at six years postoperatively. The Merle d'Aubigné-Postel scores increased from 10.0 preoperatively to 16.0 at one year postoperatively.

Fig. 3.

Proximal translation distances of the revision cups. The colour coding refers to the preoperative Paprosky-type defect of each individual patient.

Table III.

Median translations of the revision cups, measured at the centre of rotation of the cup, between postoperative and 24 months.

	Bone graft volume, median (IQR)
Direction	≤ 50 cm3 (n = 4)	> 50 cm3 (n = 5)	p-value
Proximal	0.58 (0.30 to 1.15)	1.53 (1.42 to 1.57)	0.063
Medial	-0.20 (-0.5 to 0.08)	0.10 (0.03 to 0.77)	0.413
Posterior	0.13 (-0.05 to 0.35)	0.60 (0.00 to 0.76)	0.556
Total translation	0.82 (0.54 to 1.22)	1.72 (1.62 to 1.78)	0.032

Upon visual evaluation of postoperative and two-year CT-RSA, examinations we made an unexpected observation. In seven of the nine patients, the revised cup exhibited migration relative to the proximal screws inserted into the ilium, with the migration ceasing upon the liner’s contact with the screw head protruding from the shell (Figure 4, Supplementary Material).

Fig. 4.

Coronal CT-view of the revised cups in A) 50-year old female, B) 63 year-old female, and C) 80-year-old male. The upper row shows the postoperative image and the lower row shows the image 24 months after surgery. The pelvic bone demonstrates signs of restoration of the cortical bone structure in areas of preoperative protrusion (A and B). Note that cup migration appears to be halted by the outer edge of the liner insert (polyethylene in A, and cobalt-chrome in B and C) making contact with a screw head that protrudes from the shell.

Clinical outcome

No cups had been revised at five years follow-up. The median Merle d'Aubigné-Postel score increased from preoperative score of 11.0 (IQR 11.0 to 14.0) to 17 (IQR 17.0 to to 18.0) at two years postoperatively.

Discussion

Acetabular revisions in the presence of bone loss remain challenging due to the difficulties associated with achieving initial implant stability and long-term construct durability, especially for severe segmental bone defects (Paprosky type 3). Many different techniques to handle these types of defects have been described in the literature, with varying surgical outcomes reported.³⁶ We and others^1-4 have reported on the use of bone grafting in combination with uncemented hemispherical cups, demonstrating excellent long-term survival,^2-4 even when large amounts of bone graft are used and with limited host bone contact.^4,37 Based on the technical development of implants and the possibility for variable angular screw fixation, we have used the bone grafting technique for larger and larger bone defects. In this study, we aimed to investigate whether BMD in the grafted bone increases over time, to quantify implant migration when large volumes of bone graft are used, and to determine the influence of graft volume on implant migration.

Changes in bone mineral density

In this study, we found increasing bone graft density after two years. Our findings can be compared with other studies that showed similar results, albeit only in combination with cemented cups and in Paprosky type 2B defects.^20,22 Therefore, our study is the first to report on bone graft integration in Paprosky type 3 defects in combination with uncemented cups using modern radiological techniques. We used roughly six-times larger graft volumes than previously reported in uncemented cup revisions.^9,38 One recent study used titanium graft cages, IBG, and cemented cups²⁰ using similar amounts of bone graft (mean of 40 cm³ compared with mean of 64 cm³ in the present study) and found an increase in BMD from 378 mg/cm³ postoperatively to 466 mg/cm³ after two years, as compared with the increase in BMD from 317 mg/cm³ postoperatively to 466 mg/cm³ after two years in our study. Even in the group with ≥ 50 cm³ of bone graft and Paprosky type 3B defects in our study, the bone graft density increased to similar values even though it started at a lower density (Table II). Several other studies have looked at bone graft density using conventional DEXA, with discrepant results.^22,39,40 The limited spatial resolution obtained with the DEXA used in those studies makes it difficult to discriminate measurements in areas of impacted bone graft from those in the host bone and other tissues, such that the results are challenging to interpret.⁴¹

Implant motion

Most of the implants showed minimal translation during the first two years post-surgery, whereas more proximal translation was associated with the larger bone graft volumes and the increasing acetabular defects. One implant showed proximal translation of 2.97 mm at the two-year follow-up, although there were no radiological or clinical signs of loosening at six years post-surgery. In primary surgery, the relationship between migration at two years and subsequent aseptic loosening is well established.²³ However, in complex revision surgery using bone grafts, this relationship depends on multiple factors, such as the bone defect, graft quality and preparation, impaction technique, and implant choices.⁴² The median proximal translation of 1.4 mm reported in the present study is consistent with the proximal translation distance reported for the cementing technique (1.5 mm).²⁰

The reasons for the observed differences between our experiences and those related to the early findings of Paprosky and others^12-14,37 are likely multifactorial. One possible explanation relates to a biomechanical nature rather than the biological nature proposed by Paprosky. The modern, porous-coated multi-hole cups used in our study (Trident Acetabular System, Stryker Orthopaedics) allow for improved direct stabilization of the cup. Screw fixation with variable angles facilitates fixation in multiple directions, thereby further improving the initial fixation (median of five screws used) (Figure 5). According to Figure 11B in the original classification published by Paprosky et al,¹¹ only one screw is visible that is fixating the hemispherical cup in the host bone, while the remaining screws are used to stabilize the structural bone graft that is meant to support the cup. Recent studies in experimental^43,44 and clinical⁴⁵ settings suggest that screw fixation in multiple planes is beneficial for long-term implant survival in uncemented revisions. Using the CT-RSA technique, we made the observation that the acetabular shell migrates proximally until the liner hits the screw head (Figure 4). This finding is surprising, as the common perception is that the screws facilitate compression of the cup against the underlying bone/graft. In our setting, with large amounts of bone graft in the defect apical to the cup, the screws seem to function more like a distance screw than a compression screw, thereby preventing the cup from proximal migration along the joint forces into the grafted bone.⁴⁶ To our knowledge, this phenomenon has not previously been described and warrants further investigation.

Fig. 5.

Multiple screws in multiple directions allow direct stabilization of the cup in the patient’s host bone, ensuring stability during osseointegration.

Our study has limitations related to its retrospective design and single-centre nature. We used the information available from the medical records and Picture Archiving and Communications System acquired as part of the clinical routine, without following a specific study protocol. We followed a predefined study protocol and the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines to collect and report these data.⁴⁷ Eight patients were excluded due to the lack of sufficient follow-up CT examinations. This selection might have led to bias and limited generalizability. The retrospective design also hampered the use of standardized measurement techniques for our outcome variables of bone graft volume, bone mineral density, and implant migration. However, we used a modern CT analysis technology for the bone graft volume measurements, which has been shown to be feasible in several previous studies.^20,48,49 For the assessment of implant migration, we applied the novel CT-RSA measurement technique to clinical CT examinations, albeit without double examinations, which are otherwise routinely performed to allow calculations of measurement error.²⁷ However, the method has previously been validated in total hip arthroplasty using double measurements.²⁰ Measuring bone volume using 3D slicer shows an excellent ICC > 0.95,^48,49 prompting us to abstain from formal double measurements. Measuring the BMD of the bone graft has been used previously in only one study,²⁰ which showed that measurement accuracy was highly dependent upon the placement of the ROIs in the graft. To limit this source of bias, we selected the largest possible ROI while avoiding the host bone, the implant, and the screws. Great care was taken to place the ROI consistently in the same location for each measurement. To increase the level of precision, ROI placements were validated by a second observer, and adjusted when deemed necessary.

In conclusion, uncemented acetabular revisions using large amounts of bone graft exhibit a migration pattern that is consistent with that seen with the cementing technique. Despite the large amount of bone graft used, the bone density increases over time. Uncemented acetabular revisions with multiplanar screw fixation appear to be a viable option even for revisions with challenging defects, such as the Paprosky type 3 defect.

Take home message

- Impaction bone grafting in uncemented acetabular revision leads to increased bone density over time, even when large graft volumes are used.

- Implant migration is consistent with the cemented impaction bone grafting technique.

- Larger graft volumes (≥ 50 cm³) may be associated with greater proximal migration.

Data Availability

The datasets generated and analyzed in the current study are not publicly available due to data protection regulations. Access to data is limited to the researchers who have obtained permission for data processing. Further inquiries can be made to the corresponding author.