Over the past decades, nonmodular stemmed acetabular cups have gained popularity for the reconstruction of periacetabular tumor defects because of their wide availability, intraoperative flexibility, relatively fast and easy implantation, and the possibility of allowing early weight-bearing and rapid postoperative mobilization1,2. Nevertheless, as with any periacetabular reconstruction technique, the risks of dislocation (3% to 31%), aseptic loosening (0% to 16%), and periprosthetic joint infection (PJI) (10% to 50%) remain substantial. These complications commonly necessitate revision surgery, resulting in an even higher risk of complications and morbidity1–11.

The LUMiC prosthesis (Implantcast) was introduced in 2008 for the reconstruction of extensive periacetabular defects. This modular device consists of a stem that sits in the remaining ilium, in line with the weight-bearing axis of the pelvis, and an acetabular cup that is connected to the stem12. The stem and cup are equipped with a sawtooth junction allowing for rotational adjustment of the cup. In a previous study2, we reported promising short-term complication and implant survival rates compared with other techniques1,4,13–15.

In the current study, we aimed to assess the mid-term results of this implant, in a larger multicenter cohort. Therefore, we evaluated (1) the complications and associated risk factors, (2) the reasons for implant revision, and (3) the cumulative incidence of implant revision at 2, 5, and 10 years.

Materials and Methods

Approval for conducting this study was obtained from the scientific committee of the Leiden University Medical Center (LUMC). The committee waived patients’ informed consent (W.22.002/2022-029). Participating centers obtained approval by their local ethical review board.

Study Design, Setting, Participants

In this international, multicenter, observational retrospective study, we assessed all patients in whom an internal hemipelvectomy was performed for a bone tumor and in whom the LUMiC prosthesis was used for reconstruction of the defect during the period of 2008 to 2022. The minimum potential follow-up was 24 months; patients who died within 24 months after implantation were included. Fourteen tertiary referral centers participated. All patients had a periacetabular tumor defect (including P2 according to the modified Enneking classification16) in which the medial ilium was preserved as described in our previous work2,4. One hundred and sixty-six patients (87 female, 52%) with a median age of 57 years (range, 10 to 81 years) were included. The median follow-up for censored patients was 4.2 years (interquartile range [IQR], 2.6 to 7.6 years). The indication for reconstruction in 114 (69%) of the patients was resection of a primary malignant bone tumor, while 46 (28%) had been treated for metastatic carcinoma (Table I). Twenty-six (16%) of the patients had ≥1 previous surgeries at the same site (Table I). All patients received prophylactic antibiotics according to the local protocol: either as a single dose (n = 25, 15%) or over 24 hours (n = 55, 34%), 3 to 5 days (n = 53, 33%), or >5 days (n = 29, 18% [of 162 with data on antibiotic duration]) (Table II). According to the modified Enneking classification, 17 (11%) of the patients underwent type P1b+2 resection; 83 (52%), type P2 resection; 50 (31%), type P2+3 resection; and 11 (7%), type P1b+2+3 resection4,16. In 60 (41%) of patients (146 with data), an extra-articular resection of the hip joint was performed.

TABLE I - Study Population (N = 166)* Variable Values Sex 166  Male 79 (48%)  Female 87 (52%) BMI†(kg/m 2 ) 25 [22-28] ASA score 165  1 40 (24%)  2 81 (49%)  3 44 (27%) Smoking 139  Yes, currently 20 (14%)  Yes, formerly (stopped >6 mo.) 17 (12%) Diabetes 12/152 (8%) Indication for reconstruction 166  Primary malignant tumors 114 (69%)   Chondrosarcoma 67 (40%)   Osteosarcoma 18 (11%)   Ewing sarcoma 11 (7%)   Soft-tissue sarcoma 9 (5%)   Multiple myeloma 4 (2%)   Other 5 (3%)  Metastatic carcinoma 46 (28%)  Benign aggressive lesions 5 (3%)   GCTB 2 (1%)   Chondroblastoma 1 (1%)   Chondromyxoid fibroma 1 (1%)   Yeast infection‡ 1 (1%)  Other 1 (1%) Previous surgery at same site 26 (16%)  Internal hemipelvectomy or partial pelvic tumor resection 9 (35%)  Total hip arthroplasty 13 (50%)  Curettage (GCTB/osteoblastoma) 4 (15%) Soft-tissue involvement 91/158 (55%) Pathological fracture at diagnosis 30/162 (19%) Neoadjuvant chemotherapy 55/163 (34%) Neoadjuvant radiation therapy 28/163 (17%) Adjuvant chemotherapy 52/162 (32%) Adjuvant radiation therapy 30/159 (19%)

*The values are given as the number, with the percentage in parentheses, except where otherwise noted. BMI = body mass index, ASA = American Society of Anesthesiologists physical status, and GCTB = giant cell tumor of bone.

†BMI values are given as the median, with the interquartile range (IQR) in square brackets.

‡Suspicious lesion in a patient with known multiple myeloma. Histology identified no tumor localization but a yeast infection.

TABLE II - Prosthesis and Surgical Details Variable No. (%)* Antibiotic administration 161  Cephalosporins 100 (62%)  Beta-lactam 13 (8%)  Cephalosporins + clindamycin 12 (8%)  Cephalosporins + metronidazole 8 (5%)  Glycopeptides 10 (6%)  Cephalosporins + aminoglycosides 6 (4%)  Glycopeptides + β-lactam 6 (4%)  Other 6 (4%) Modified Enneking resection type 161  P1b+2 17 (11%)  P2 83 (52%)  P2+3 50 (31%)  P1b+2+3 11 (7%) Use of computer-assisted surgery 49/166 (30%) Concomitant proximal femoral reconstruction 56/165 (34%)  Silver-coated proximal femur 40/56 (71%) Cemented LUMiC stem 30/165 (18%) Cup size 163  50 mm 62 (38%)  54 mm 49 (30%)  60 mm 52 (32%) Dual-mobility cup 107/164 (65%) Silver-coated cup 36/166 (22%) Cemented femoral component 62/164 (38%) Use of Trevira tube 48/163 (29%)
Preoperative Planning, Surgical Details, Procedure

The general surgical and procedural details were previously described12. Although not available in all participating centers, the leading center prefers the use of intraoperative navigation to optimize stem placement. Press-fit fixation of an uncemented stem was preferred unless adequate press-fit fixation of the stem was not obtained, or if bone quality was inadequate. Conventional and dual-mobility cup articulations were available and were used at the surgeon’s discretion, although the dual-mobility cup was preferred on the basis of previous results2. Depending on the surgeon’s preferences, a Trevira attachment tube (Implantcast) was used to reattach soft tissues. Usually, early partial weight-bearing (with use of 2 crutches) was allowed under supervision of a physiotherapist. At 6 to 12 weeks, this was gradually increased to full weight-bearing. Combined flexion and external rotation was avoided. In some centers, orthoses were used.

Generally, patients with a suspected PJI underwent a DAIR (debridement, antibiotics, and implant retention) procedure, including intraoperative culturing and a thorough debridement, followed by at least 2 weeks of intravenous antibiotics. The standard antibiotic treatment regimen spanned a minimum of 12 weeks, depending on the isolated microorganism(s) and the susceptibility pattern. In some cases, eradication of the PJI was not achieved, resulting in chronic antibiotic suppression or a draining fistula, as described in our previous paper11.

Variables

Patient records were reviewed to obtain demographics, surgical details, reconstruction details, complications, and functional outcomes at the last date of follow-up. Incision types were divided into 2 groups: a single iliofemoral (“question mark”) incision or a star-shaped incision. Pelvic resection types were divided into 2 groups: P1b+2 and P2 or P1b+2+3 and P2+3. Revision was defined as any surgical procedure in which (part of) the implant was removed or replaced. Complications and the reason for implant revision were categorized according to the Henderson classification17.

Statistical Analysis

Competing risks models18 were used to estimate the cumulative incidences of implant revision and reoperations. A competing risks model with 3 competing events was used to estimate the cumulative incidences of mechanical failure and infection, with death and local recurrence as competing events. A second competing risks model with 2 competing events was employed to estimate the cumulative incidence of any complication, with death as a competing event.

Cause-specific Cox hazard regression models were estimated to study the effect of prognostic risk factors on time to dislocation and time to PJI. Cause-specific hazard ratios and 95% confidence intervals are reported. The proportion of complications was compared among different subgroups using chi-square analysis. Analyses of data were performed using SPSS version 25.0 (IBM) and RStudio version 4.2.119. The R Studio package “cmprsk” was used to estimate the cumulative incidence of implant revision and reoperations. The level of significance was set at p < 0.05.

Results Complications, Implant Revision, Risk Factors

During the study period, 82 (49%) of the patients underwent ≥1 reoperation (Table III). In total, 165 reoperations were performed, of which 104 (63%) were within 6 months.

TABLE III - Complication and Revision Rates, Time to Revision, and Reconstruction Status Among Revised Cases Complication* No. (%) Time to Revision After Implantation (mo) Reconstruction Status at Latest Follow-up (No.) Patients Revisions Total 82/166 (49%) 32/166 (19%) 0-99 H1A (dislocation) 31 (19%) 5 (3%) 0-45 Revision LUMiC (3), resection arthroplasty (1), hindquarter amputation (1)† H1B (aseptic wound dehiscence) 7 (4%) 0 (0%) H2A (aseptic loosening <2 yr after implantation) 1 (1%) 1 (1%) 12 Resection arthroplasty (1) H2B (aseptic loosening ≥2 yr after implantation) 4 (2%) 4 (2%) 53-77 Custom-made implant (4) H3A (implant breakage or wear) 1 (1%) 1 (1%) 32 Revision LUMiC (1) H3B (periprosthetic osseous fracture) 2 (1%) 2 (1%) 0, 9 Revision LUMiC (1), resection arthroplasty (1) H4A (PJI <2 yr after implantation) 36 (22%) 11 (7%) 0-20 Revision LUMiC (2), resection arthroplasty (6), custom-made implant (1), hindquarter amputation (1), spacer (1) H4B (PJI ≥2 yr after implantation) 5 (3%) 4 (2%) 35-65 Revision LUMiC (1), resection arthroplasty (2), rotationplasty (1) H5A (soft-tissue progression of tumor) 2 (1%) 0 (0%) H5B (osseous progression of tumor) 13 (8%) 4 (2%) 0-99 Hindquarter amputation (3), resection arthroplasty (1) Other 12 (7%) 0 (0%)

*H = Henderson classification, and PJI = periprosthetic joint infection.

†One patient had revision LUMiC for dislocation but later underwent amputation due to osseous progression of tumor.

Dislocations (Henderson 1A) occurred in 31 (19%) of the patients; 21 (13%) had a single dislocation and underwent closed or open reduction, and 10 (6%) had recurrent dislocations. The first dislocation occurred within 1 month in 2 (6%) of the 31 patients, between 1 and 6 months in 4 (13%), and later in 5 (16%). Patients who had previous surgery at the same site had a higher dislocation risk than those without previous surgery at the same site (cause-specific hazard ratio [HRCS], 3.0 [95% confidence interval (CI), 1.5 to 6.4]; p < 0.01) (Table IV). Utilization of the dual-mobility cup (HRCS, 0.6 [95% CI, 0.3 to 1.2]; p = 0.17) or the Trevira tube (HRCS, 0.7 [95% CI, 0.3 to 1.6]; p = 0.38) was not significantly associated with dislocation. Dislocations occurred in 15 (26%) of 57 patients with conventional cups compared with 16 (15%) of 107 with dual-mobility cups (p = 0.08). Dislocations occurred in 7 (15%) of 48 patients with reconstruction with a Trevira tube versus 24 (21%) of 115 without (p = 0.35). Dislocations occurred in 3 (10%) of 29 patients with reconstruction with a Trevira tube and dual-mobility cup versus 11 (30%) of 37 who had neither (p = 0.06). Five implants (3% of patients) were revised for instability. Four conventional cups (2%) were exchanged for a dual-mobility cup, and 1 (1%) resection arthroplasty was performed because of recurrent instability and poor oncological prognosis.

TABLE IV - Univariate Cause-Specific Cox Proportional Hazards Regression Model for Prognostic Factors for the Occurrence of Dislocation and PJI* Possible Risk Factors Dislocation P Value PJI P Value HRCS (95% CI) HRCS (95% CI) Sex  Female†  Male 1.8 (0.89-3.77) 0.10 1.2 (0.66-2.21) 0.54 Age 1.0 (0.99-1.05) 0.07 1.0 (1.00-1.04) 0.05 BMI 1.0 (0.96-1.10) 0.49 1.0 (0.98-1.10) 0.21 ASA classification  I†  II 1.3 (0.56-3.20) 0.52 1.2 (0.57-2.73) 0.59  III 0.8 (0.28-2.51) 0.76 1.5 (0.62-3.49) 0.39 Smoking NA  No†  Yes 0.5 (0.16-1.68) 0.27 Diabetes NA  No†  Yes 1.9 (0.69-5.63) 0.20 Previous surgery at same site  No†  Yes 3.0 (1.47-6.41) 0.003 0.8 (0.34-1.89) 0.61 Incision type NA  Single incision†  Star-shaped incision 1.3 (0.69-2.57) 0.39 Proximal femoral resection  No†  Yes 0.8 (0.35-1.64) 0.48 1.2 (0.66-2.26) 0.53 Type of pelvic resection  P1b+2 and P2†  P2+3 and P1b+2+3 1.4 (0.70-2.95) 0.32 2.5 (1.35-4.72) 0.004 Surgical duration (hr) NA 1.1 (0.98-1.29) 0.09 Blood loss (L) NA 1.1 (0.84-1.32) 0.67 Dual-mobility cup  No†  Yes 0.6 (0.29-1.23) 0.17 NA Use of silver-coated cup NA  No†  Yes 2.1 (1.11-4.04) 0.02 Use of silver-coated proximal femur NA  No†  Yes 0.2 (0.07-0.49) <0.001 Use of computer-assisted surgery  No†  Yes 0.7 (0.38-1.92) 0.71 0.8 (0.37-1.54) 0.44 Use of Trevira tube  No†  Yes 0.7 (0.30-1.59) 0.38 1.6 (0.87-3.05) 0.13

*PJI = periprosthetic joint infection, HRCS = cause-specific hazard ratio, 95% CI = 95% confidence interval, BMI = body mass index, ASA = American Society of Anesthesiologists physical status, and NA = not applicable.

†Reference category.

Early aseptic loosening (Henderson 2A) of the stem occurred in 1 patient (1%), who previously had reconstruction with use of an allograft and total hip replacement that had failed as a result of PJI. After 5 years without further reconstruction, a cemented LUMiC prosthesis was implanted; it was removed 12 months later because of aseptic loosening.

Late aseptic loosening (Henderson 2B) occurred in 4 (2%) of the patients, 3 with an uncemented implant and 1 with a cemented implant. All underwent revision with use of a custom-made implant, 4.4 to 6.4 years after implantation. No complications preceded the aseptic loosening, and none of these patients had undergone reconstruction previously.

Intraprosthetic dissociation (Henderson 3A) occurred in 1 patient (1%), who had persistent subluxation. During revision surgery 32 months after implantation, the LUMiC dual-mobility liner appeared to have dissociated. The stem was well fixed and was left in place, and the liner and cup were revised.

Periprosthetic fracture at implantation (Henderson 3B) occurred in 2 (1%) of the patients; the fractures consisted of a periprosthetic fracture of the ilium around the uncemented stem. One patient underwent successful revision to an uncemented LUMiC stem, implanted slightly more dorsal and lateral in the remaining ilium, utilizing fresh bone stock. One was treated conservatively, but the fracture did not consolidate, resulting in revision to a custom-made prosthesis at 9 months. As a result of dislocation, revision surgery was performed to increase the offset. However, during the procedure, it turned out that the custom-made prosthesis had loosened because of poor bone quality, leading to implant removal and resection arthroplasty.

PJI (Henderson 4) occurred in 41 (25%) of the patients. In 22 (54%) of the patients, PJI occurred within 6 weeks; in 4 (10%), between 6 and 12 weeks; in 2 (5%), between 12 and 24 weeks; and in 13 (32%), at >24 weeks postoperatively. The success rate of ≥1 DAIR procedures was 50% (11 of 22) for patients with an early PJI between 0 and 6 weeks, 75% (3 of 4) with PJI between 6 and 12 weeks, 100% (2 of 2) with PJI between 12 and 24 weeks, and 69% (9 of 13) with PJI at >24 weeks postoperatively. Of the patients with infection following reconstruction, 17 (71%) of 24 without a Trevira tube were successfully managed with DAIR procedures versus 8 (50%) of 16 with a Trevira tube (p = 0.18).

The median duration of the index surgery was 5.5 hours (IQR, 4.3 to 6.5 hours) in patients who developed PJI and 4.8 hours (IQR, 3.6 to 6.5 hours) in those who did not develop PJI (p = 0.13). Surgical duration was not associated with PJI risk (Table IV). The PJI risk was lower for patients with a concomitant proximal femoral reconstruction with silver coating compared with those without silver coating (HRCS, 0.2 [95% CI, 0.07 to 0.5]; p < 0.01). Nine (23%) of 40 with silver coating developed PJI compared with 8 (80%) of 10 without silver coating (p < 0.01) (data on silver coating available for 50 of 56 patients with proximal femoral reconstruction). Resections that included the P3 region had an increased PJI risk (HRCS, 2.5 [95% CI, 1.4 to 4.7]; p < 0.01). Median blood loss did not differ between patients with PJI (1.9 L [IQR, 1.0 to 2.5 L]) and those without (1.5 L [IQR, 1.0 to 2.3 L]) (p = 0.90). Ultimately, 15 (9%) of the patients underwent revision because of PJI. One had a previous reconstruction (pedestal-cup prosthesis) that had failed because of PJI, and the others did not have previous reconstructions. Four underwent a new reconstruction (3 were revised to a new LUMiC prosthesis during 1-stage [n = 2] or 2-stage [n = 1] revision, and 1 received a custom-made implant). Others underwent resection arthroplasty (n = 8), implant removal and spacer implantation (n = 1), hindquarter amputation (n = 1), or rotationplasty (n = 1) (Table III).

Local recurrence (Henderson 5B) occurred in 13 (8%) of the patients, leading to implant removal in 4 (2%) of the cases (3 hindquarter amputations, 1 resection arthroplasty).

Cumulative Incidence of Implant Revision, Reconstruction Status, and Functional and Survival Outcomes

The cumulative incidence of implant revision for mechanical reasons (Henderson 1 to 3) at 2, 5, and 10 years was 4% (95% CI, 2% to 8%), 9% (95% CI, 4% to 15%), and 12% (95% CI, 6% to 20%). For PJI (Henderson 4), the rates were 7% (95% CI, 4% to 11%), 10% (95% CI, 5% to 16%), and 11% (95% CI, 6% to 18%) (Fig. 1). For mechanical reasons and PJI (Henderson 1 to 4), the rates were 11% (95% CI, 7% to 17%), 18% (95% CI, 12% to 25%), and 24% (95% CI, 16% to 33%), respectively.

Fig. 1:

Cumulative incidence of LUMiC revision for mechanical complications (Henderson 1 to 3) and PJI (Henderson 4), using a competing risks model.

The cumulative incidence of reoperation for any complication at 2, 5, and 10 years was 44% (95% CI, 36% to 51%), 52% (95% CI, 43% to 60%), and 58% (95% CI, 47% to 67%) (Fig. 2).

Fig. 2:

Cumulative incidence of reoperations for any complication, using a competing risks model.

During the study period, 24 LUMiC reconstructions (14% of the 166 patients) were removed. Four (2%) were removed for tumor progression via hindquarter amputation (n = 3) and resection arthroplasty (n = 1). One additional patient underwent revision for dislocation, but later underwent hindquarter amputation due to tumor progression. Nineteen reconstructions (11%) failed; 11 patients (7%) underwent resection arthroplasty, 5 (3%) were revised to a custom-made prosthesis, 1 (1%) underwent hindquarter amputation, 1 (1%) received a cement spacer, and 1 (1%) underwent rotationplasty (Table III). In 160 (96%) of the patients, limb salvage was achieved. Fifty (30%) were able to walk without mobility aids, 47 (28%) used 1 crutch, 41 (25%) used 2 crutches, and 11 (7%) were not able to walk with crutches (149 with available data).

At the time of the most recent follow-up, 86 (52%) of the patients were alive without disease, 31 (19%) were alive with disease, 41 (25%) had died of disease, and 8 (5%) had died of other causes. The 5-year overall survival was 67% (95% CI, 58.6% to 75.4%).

Discussion

In the current study, we assessed the mid-term clinical outcomes of patients who underwent reconstruction for periacetabular tumor defects with use of the LUMiC prosthesis. To our knowledge, this is the largest oncological pelvic reconstruction series to date, and we found a substantial reoperation risk but a relatively low revision risk for mechanical complications. Dislocation and PJI remain the major concerns in the early postoperative period.

The dislocation rate (19%) in our cohort is comparable with that found for other stemmed acetabular implants, such as the pedestal-cup and the ice-cream cone prostheses (15% to 26%)1,4,8,20,21. Previous surgery at the same site was associated with a higher dislocation risk. This is in line with conventional total hip arthroplasty and might be attributable to the compromised supporting soft tissues22,23. We found no association between dislocation risk and the use of a Trevira tube, although use of the Trevira tube might enhance the stability of the construct12. The dislocation rate for dual-mobility cups (15%) was substantially higher than the 4% we previously found among 24 dual-mobility cups, which might be attributable to the longer follow-up of the dual-mobility articulations2. Although the dislocation risk was not significantly higher for conventional cups (26%), we believe that it is reasonable to continue utilizing the dual-mobility cup. First, previous studies on re-revisions for dislocation in hip-revision surgery have shown promising results for dual-mobility articulations24–26. Second, with the exception of a single intraprosthetic dissociation, no cup-specific complications were observed. Caution should be taken when comparing dislocation rates in the literature, since it is unclear if all dislocations (including those managed with closed reduction) are being reported or only those that require revision surgery21. Furthermore, most prior studies did not include patients with failed previous reconstructions, while these had a higher dislocation risk in our study (36% versus 11%)5–7,20,21. To reduce the dislocation rate, a postoperative abduction orthosis could be of value; Erol et al. found a 10% dislocation risk among 21 patients with LUMiC pros