1 Introduction

Periprosthetic fractures (PPF) of the femur represent a significant complication following total hip arthroplasty (THA). The cumulative incidence for PPF is reported as 3.5%.¹ However, NJR data for England and Wales have shown a two-fold increase in the number of revisions performed for this indication since 2006.² Given the ageing population and associated increasing frequency with which THA is being performed, the number of PPF is expected to rise in the future.^3,4 These fractures can be difficult to manage due to both patient factors and operative challenges. The Vancouver Classification represents a commonly utilised system with which to describe fracture pattern and guide management. Type B fractures are the most common (approximately 80% of cases),⁵ with subtype B2 – i.e. a fracture around the stem with a loose implant but preserved bone stock – accounting for approximately 50% of type B fractures.¹ The widely accepted surgical management of such fractures is conversion to a long uncemented tapered fluted stem with diaphyseal fit.⁶ This technique is not without issues, however, including proximal stress shielding, stem subsidence and the need for sufficient distal bone stock for fixation.⁷ In the case of B2 fractures around a cemented implant, this further complicates management by altering the biomechanics and surgical technique.⁵ Specifically, removal of the cement mantle is required, and this potentially increases operative time, bleeding and risk of perforating the femur.⁸ As an alternative, open reduction internal fixation (ORIF) alongside cement-in-cement revision has been utilised. This involves preservation of the original cement mantle and can therefore potentially reduce operative time and bleeding, both of which are beneficial in frail patients. This technique has been widely accepted for a number of indications⁹ with biomechanical¹⁰ and clinical validation.¹¹ In the context of polished tapered stems, which have excellent long-term survivorship,² the cement is likely to have been well fixed prior to fracture¹² and this provides a further reason to avoid revision of the mantle. The supporting evidence for cement-in-cement revision for PPF is not robust, however, with limited published data.⁷ Furthermore, to our knowledge no studies within the literature have directly compared uncemented and cement-in-cement revision. This paper, therefore, sets out to examine the outcomes of both procedures. The primary outcome was survivorship of implants. Secondary outcomes were operative time, complications, patient reported outcomes and radiographic analysis.

2 Methods

Institutional approval was obtained for this study. Our prospectively collated database (Bluespier, UK) was analysed to identify all patients undergoing surgical intervention for a femoral PPF from 2008 to 2018. This enabled a minimum of two years follow-up. Radiographs and operation notes were then examined by two independent observers to identify cases of Vancouver B2 fractures in primary arthroplasties. Cases were excluded if they were not a B2 type or the primary stem was uncemented. Age, gender, American Association of Anethesiologists (ASA) score, body mass index (BMI), operative time, blood transfusion requirement, pre- and post-operative haemoglobin (Hb) level and length of hospital stay was collected for all patients. Follow-up occurred post-operatively at regular intervals until fracture union was achieved, and annually thereafter with clinical and radiological assessment. Complications and survivorship of implant and patients were recorded. Oxford Hip Score (OHS) at last follow-up was established.

3 Results

Seventy-two cases were identified which matched the inclusion criteria (Fig. 1). Forty-three patients were in the uncemented group and 29 in the cement-in-cement. Mean follow-up was 4.1 years (range 0.1–13.4 years). There was no statistical difference in follow-up period between the two groups (p = 0.586). Twenty-eight patients died during the study period at a mean of 2.4 years post-operative. Three deaths occurred in the peri-operative period; the remainder were for causes unrelated to their surgery. There was no difference in baseline demographics between the two groups (Table 1).

Close

Fig. 1 CONSORT diagram illustrating patients excluded from final analysis.

Export to PPT

Seventeen patients (39.5%) in the uncemented group underwent concomitant acetabulum revision compared to two (6.9%) in the cement-in-cement group. Regarding fixation method in the cement-in-cement group, 24.1% utilised cables only, with the remaining 75.9% using a plate and cables. Operative time was significantly shorter in the cement-in-cement group (140 cf 160 min, p = 0.046), however there was no difference in transfusion requirements, change in Hb or length of hospital stay. There was no difference in OHS between the groups (p = 0.779, Fig. 2). Twenty-one patients (48.8%) in the uncemented group developed at least one complication. Five patients (11.6%) suffered multiple complications. In the cement-in-cement group 10 complications occurred in 10 patients (34.5%). This difference did not reach statistical significance (p = 0.084). On review of individual complications, however, dislocation rate was significantly higher in the uncemented group (18.6% cf 0%, p = 0.014). This was the only individual complication to reach statistical significance. Three perioperative deaths occurred in the uncemented group, compared to zero in the cement-in-cement cohort. Regarding the three deaths, the pre-operative ASA score was 3 in two patients and 4 in one. Operative time was greater than the mean in two cases (210 and 240 min). Primary causes of death were recorded as myocardial infarction, sepsis and femoral fracture. Clinical outcomes are presented in Table 2 and complications listed in Table 3.

Close

Fig. 2 Oxford hip scores for both groups. There was no significant differences observed (p = 0.779).

Export to PPT

Radiographic analysis identified two non-unions in the uncemented group; one case of significant osteolysis of the proximal femur, one infected non-union. There were no non-unions in the cement-in-cement group. There was no significant difference in median time to radiological union between groups (cement-in-cement = 6 months cf uncemented = 5 months, p = 0.171). There was a statistically significant difference in mean subsidence between groups, however, with 0.4 mm in the cement-in-cement group compared to 2.3 mm in the uncemented cohort (p = 0.015). There were no cases of aseptic loosening in either group.

Five patients in the cement-in-cement (17.2%) and eight in the uncemented group (18.6%) underwent re-revision. Five (62.5%) of the re-revisions in the uncemented group were of the acetabular component only; four for instability and one for aseptic loosening. There were three revisions for PJI; two underwent a two-stage revision and one a DAIR procedure without exchange of the stem. In the cement-in-cement group, three (60%) of the re-revisions were for further periprosthetic fracture; two B1 type and one C type. Both B1 fractures occurred when cables were used without additional plate fixation at three- and six-months post-operative. All underwent revision of fixation without altering the femoral component. There were two revisions for infection; one underwent washout and removal of the plate, and the other a two-stage revision. Both are infection free at last follow-up. Regarding the two further PJI in this group who did not undergo revision, one died prior to revision surgery and the other is being managed with suppressive antibiotics due to being unfit for further surgery. Reason for re-revision is presented in Table 4. Survival analysis did not reveal any difference between groups for revision (p = 0.955) or survival time (p = 0.572). The Kaplan-Meier analysis for all-cause revision is presented in Fig. 3.

Close

Fig. 3 Kaplan-Meier analysis. There was no difference between groups for re-revision (p = 0.955) or survival time (p = 0.572).

Export to PPT

4 Discussion

Periprosthetic fracture represents a significant complication following THA with predictions that the incidence will increase.^3,4 Management of these fractures is technically challenging and long operative times can be a cause for concern in frail patients. The commonest fracture subtype – Vancouver B2 – has a long history of being managed with an uncemented tapered fluted stem.^5,6 In cemented stems, removal of the mantle to facilitate this can add operative time and blood loss.⁷ As an alternative, cement-in-cement revision of the femoral component has been proposed. This technique has been validated with biomechanical¹⁰ and clinical studies,¹¹ however there is limited evidence specifically for PPF.⁷ To our knowledge, this is the first study directly comparing the outcomes of cement-in-cement with long uncemented revision for Vancouver B2 fractures.

As anticipated, there was a significant reduction in operative time for the cement-in-cement group. Given the reduction in femoral preparation required this is perhaps not surprising. The theoretical advantage of lower blood loss was not borne out in the transfusion requirements or change in Hb level, however, with both groups having a relatively equal post-operative transfusion volume and reduction in Hb. Whilst cement-in-cement revision in isolation can reduce bleeding by avoiding removal of the cement mantle, performing an extensile posterolateral approach to the femur to apply a plate will naturally increase blood loss. This potentially explains the similar transfusion rates between groups. There was also no difference in hospital stay. This could be explained by the similar rehabilitation both groups would have to undergo prior to discharge. Both cohorts achieved similar levels of function at last follow-up based on OHS.

One of the theoretical disadvantages of cement-in-cement revision for PPF is the risk of non-union. This is due to concerns that the endosteal blood supply is reduced as a result of canal preparation and cementation.² The blood flow pattern changes from centrifugal to centripetal and it is unknown whether this recovers.² This is of particular significance in ORIF procedures as periosteal blood supply can also be affected if multiple cerclage wires are used or there is excessive soft tissue stripping intra-operatively. In our series, we did not identify any cases of non-union in the cement-in-cement group. Regardless, the aforementioned surgical factors should still be considered when performing this technique. These results are comparable to the literature; in a series of 23 cement-in-cement cases Bryant-Evans et al.⁷ had only one non-union. In the uncemented group, there were two non-unions; one infective and another with significant osteolysis of the fracture site.

In the polisher taper stem, disruption of the mantle leads to a reduction in radial compressive forces in the cement and hoop stresses in the bone, ultimately leading to subsidence until stability is established.² Following revision, stem subsidence was also not an issue for the cement-in-cement group, with only 0.4 mm subsidence on average and a maximum of 2 mm, which was significantly lower than the uncemented group. This indicates a stable fixation was achieved through ORIF and cement-in-cement revision. Finally, there were no causes of aseptic loosening in either group.

There was a comparable re-revision rate between groups (17.2% cement-in-cement, 18.6% uncemented). Reasons for revision differed, however. In the cement-in-cement group, the majority of cases were revision of the fixation due to a further PPF. The femoral component was not revised in these patients. Given that two B1 fractures occurred in patients where only cables were used, this potentially indicates that additional plate fixation should be considered. By comparison, 50% of re-revisions in the uncemented group were for instability. This suggests that revision of the acetabular component to either a dual mobility or constrained cup should be considered, depending on the patients physiological age and activity level, when undertaking revision to a long uncemented taper stem to improve stability. Revision of the femoral component was only required in one case in each group, both for prosthetic joint infection.

Complication rates varied between the two cohorts (34.5% cement-in-cement, 48.8% uncemented) but did not reach statistical significance. Dislocation rate did reach statistical significance, however, with a notably higher rate in the uncemented group (18.6% cf 0%). The cause for this is uncertain, but may be a result of post-operative femoral version and soft tissue tension being more closely matched to pre-operative levels given the implants were comparable and the cement mantel retained. Intra-operative femoral fracture occurred in 4.7% of the uncemented group. These occurred at the tip of the stem and were identified on the post-operative radiographs. No further intervention was required aside from 6 weeks of reduced weight-bearing. Such fractures are a recognised complication from diaphyseal fitting stems in revision THA and our rate is favourable compared to other studies in the literature.¹³ There was a greater number of peri-operative deaths in the uncemented group (7.0% cf 0%). This is interesting as cement-in-cement for PPF has been proposed as an alternative to uncemented revision in frail, low-demand patients¹⁴ for whom a shorter operation would be advantageous to reduce morbidity and mortality. In our study there was no difference in baseline demographics, however. One of the deaths occurred in a patient with a high ASA score, and two patients had operative times above the mean.

There are limitations to this study. Firstly, there was a small sample size. This is perhaps not surprising when investigating a subgroup of a relatively uncommon complication but it does lead to potential issues with bias as a result. In particular, it can be difficult to determine what role the choice of surgical procedure had in the complication and peri-operative death rate without larger cohort studies to confirm or refute our findings. Although both groups were comparable in terms of baseline characteristics, such as BMI, ASA, age and gender, there was no propensity matching performed as this would have further reduced sample size. Finally, there was no randomisation of what procedure was performed. Rather, surgeon preference dictated the choice of operation. The effect of this selection bias is difficult to quantify. In addition, the follow-up is a minimum of 2 years but longer follow-up would be needed to confirm the ultimate outcome of the cement-in-cement technique.

This study has found that cement-in-cement revision for Vancouver B2 fractures has a lower operative time, and a trend towards a reduction in some complications, such as dislocation. There was certainly no increase in revision rate or fracture non-unions following this procedure and functional outcomes were comparable in this short-term follow-up.

5 Conclusion

With appropriate patient selection, both cement-in-cement and long uncemented tapered stem revision represent appropriate treatment options for Vancouver B2 fractures.

Funding/sponsorship

No external funding was obtained for the completion of this study.

Author contributions

Ian W. Kennedy: Data curation; Formal analysis; Investigation; Methodology; Project administration; Software; Supervision; Writing - original draft; Writing - review & editing.

Alex Hrycaiczuk: Data curation; Investigation; Writing - review & editing.

Nigel Ng: Data curation; Investigation; Writing - review & editing.

Owen Sheerins: Data curation; Investigation; Writing - review & editing.

Sanjeev R. Patil: Data curation; Supervision; Resources.

Bryn G. Jones: Conceptualization; Resources; Supervision; Writing - review & editing.

Andrew Stark: Resources; Supervision; Writing - review & editing.

Dominic R.M. Meek: Conceptualization: Investigation: Resources; Supervision; Writing - review & editing.