Hydroxyapatite (HA) coating has been advocated to improve osteointegration due to its osteoconductive potential and to achieve early stability between the prosthesis and the bone ¹2). It has been reported that total hip arthroplasty (THA) using HA-coated prosthesis shows excellent clinical and radiological results including the presence of good osteointegration with regard to short-term result3). Some midterm follow-up studies have also shown good clinical and radiological results with THA using HA-coated acetabular cup^4,56). The authors stated that the rates of polyethylene (PE) wear and acetabular osteolysis were relatively low in contrast to other studies.

However, high failure rate of the HA-coated acetabular cups has been reported in a number of midterm series^{7,8,9,10,11,12,13}14). Numerous authors described a series of cup failures attributable to osteolysis, PE liner wear, and aseptic loosening^15,1617). Retrieved cup studies and animal experiments indicated that the HA coating resorbs as time passes after implantation^18,19,2021).

There are few long-term clinical studies that report more than 20 years of follow-up about HA-coated acetabular cup, and the mechanisms of failure remain uncertain.

The purpose of this study was to report the long-term outcome and the failure mechanism of cementless THA using HA-coated acetabular cup.

We obtained the approval from institutional review board of Yonsei University College of Medicine to search surgical database of our institution to identify cases for the current study. From January 1992 to May 1994, 123 consecutive primary cementless THAs using the HA-coated acetabular cup (Atoll®; Landos, Villeurbanne, France) were performed by one of the authors. Twenty-two hips over 65 years old in index operation were not included in this study. Thirty-five hips were not included because of follow-up loss or incomplete radiographic records. Consequently, 66 hips in 58 patients were included in this study and reviewed retrospectively. The preoperative diagnosis was osteonecrosis of the femoral head in 52 hips (78.8%); femoral neck fracture in 7 hips (10.6%); secondary osteoarthritis in 5 hips (7.6%); and fused hip in 2 hips (3.0%). There were 44 males and 22 females. The patients'age ranged from 20 to 65 years at the time of index operation (average 43 years). The mean height was 164 cm (140-182 cm) and the mean weight was 62 kg (49-82 kg). The mean body mass index (BMI) was 23.1 kg/m² (18.6-28.7 kg/m²). The average duration of follow-up was 18.3 years (10.4-23.6 years) (Table 1).

The acetabular cup was a titanium-alloy hemispherical prosthesis and it had smooth surface coated with HA. Multiple ports within the hemisphere allowed of fixation with titanium-alloy screws. A conventional PE liner and 28 mm chrome-cobalt head were used for articulation in all cases. The femoral stem is an anatomical titaniumalloy prosthesis (Euroform® anatomical femoral prosthesis; Landos).

All procedures were performed by standard posterolateral approach. The acetabular component was inserted using press-fit technique; the diameter of the implant was 1-2 mm larger than the diameter of the last reamer used in preparation of the acetabular bed. Additionally, 2 or 3 screws were inserted in superolateral or superoposterior screw holes. The femoral stem was also inserted using press-fit technique without cement. The range of motion was checked in the operation field and the wound was closed with suction drain. The patients were allowed weight bearing on the affected limb as tolerated, and they generally used 2 crutches for the first 6 weeks after the operation.

All patients were requested to visit at the hospital at 6 weeks, 3, 6, and 12 months postoperatively and yearly thereafter, according to a predetermined protocol. Clinical and radiographic assessments were done at each visit. Clinical assessment was done using Harris hip score and it is determined by an individual other than the operating surgeon22). Standardized radiographs including anteroposterior and lateral hip radiographs were obtained at each follow-up.

Each acetabular component was evaluated for the presence and progression of radiolucent lines and osteolysis at the bone-implant interface according to DeLee and Charnley23). A continuous radiolucent line above 2 mm in width at the bone-implant interface was accepted as radiographic evidence of impending failure according to the criteria of Dorr et al24). Evidence of acetabular migration was measured on serial radiographs according to Massin's criteria25). Acetabular migration was considered as movement >2 mm in the horizontal or vertical directions and change of inclination angle >5°. The acetabular cup loosening was defined if migration could be demonstrated. Osteolysis was defined as a localized or cystic-like scalloped radiolucent lesion adjacent to the prostheses and distinguished from bone resorption by stress-shielding in the serial radiographs. The most recent follow-up radiograph and the one taken shortly after index operation were examined to determine PE wear by the modified technique of Livermore et al26).

Statistical analysis was performed using the SPSS software (version 18.0; SPSS Inc, Chicago, IL, USA). The survival analysis was performed by the Kaplan-Meier method. We defined end point as any failure that required a reoperation of acetabular component due to progressive osteolysis, PE wear or loosening.

The mean Harris hip score was 54.7 (range, 0-77) preoperatively, and it was 94.1 (range, 66-100) at the final follow-up. In the patients undergoing reoperation, the mean Harris hip score just before the reoperation was 81.8 (range, 47-100).

Six hips (9.1%) had a radiolucent line in DeLee and Charnley zone 1, 7 hips (10.6%) in zone 2, and 10 hips (15.2%) in zone 3. Five hips (7.6%) had a radiolucent line in all three zones: less than 2 mm at all zones. Forty-seven hips (71.2%) showed evidence of acetabular osteolysis. With regard to the location of osteolysis, 28 hips (42.4%) was in zone 1, 33 hips (50.0%) in zone 2, and 19 hips (28.8%) in zone 3. For the extent of osteolytic lesions, 14 hips (21.2%) had an osteolytic lesion in two zones and 11 hips (16.7%) in three zones. PE wear was measurable on the most recent anteroposterior radiographs. The mean linear PE wear was 0.29 mm/year (range, 0.07-0.75 mm/year) in all cases. In cases requiring reoperation, the mean linear PE wear was 0.34 mm/year (range, 0.14-0.75 mm/year).

Two hips required circumferential wiring due to a calcar crack, but were healed without subsidence of the stem. One hip had event of dislocation at postoperative 3 months, and there was no more dislocation in further follow-up. Heterotopic bone formation was observed in 6 hips (9.1%), assessed by Brooker's criteria27), but anyone had pain or showed limitation of motion.

Thirty-nine hips (59.1%) defined as a failure that required a reoperation of acetabular component for any reason. Acetabular cup loosening was implicated in 18 hips (27.3%) radiographically. Fifteen cups (22.7%) were exchanged and 3 patients were lost at the followup (Table 2). Reoperation was performed in 21 hips (31.8%) for progressive osteolysis and PE wear. Eleven cups (16.7%) were exchanged because loosening was confirmed at operation field. Ten hips (15.2%) were performed only curettage and allobone graft without cup exchange because there was no evidence of loosening. But out of these cases, 7 hips (10.6%) required rereoperation due to delayed cup migration after reoperation (Fig. 1). Finally, 33 cups of 66 hips (50.0%) were exchanged because of loose cup (Fig. 2). The Kaplan-Meier method showed the survival rate of the acetabular cup to be 97.0% at 5 years, 74.1% at 10 years, 46.3% at 15 years and 34.8% at 20 years for any failure that required a reoperation of acetabular component (Fig. 3).

The survival rate of the smooth HA-coated press-fit acetabular cup decreased dramatically from 97.0% at 5 years to 46.3% at 15 years in our study (Fig. 3). The results are similar to other series of reports that the smooth HA-coated acetabular cup usually failed^{7,8,9,10,11,12,13}14). The effect of HA coating on long-term bone ongrowth is not known. However, retrieved cup studies or animal experiments indicated that the coated HA was resorbs as time passes and the loss of HA could be caused by several mechanisms, such as osteoclast activity during bone remodeling, abrasion, chemical dissolution or delamination^18,19,2021).

On radiographic review of our cases, it appears that both mechanical and biological factors were important in the development of acetabular cup loosening as the failure mechanisms of HA-coated prostheses. In the cases of early acetabular loosening within ten years, radiographs showed radiolucent line less than 1 mm before migration of cup. And these patients have experienced sudden pain suggesting cup migration as the ultimate failure mechanism. In retrieved cups at revision surgery, HA coating was limited on small portion on smooth surface of cups and the extensive loss was occurred on cup-HA junction (Fig. 4). This finding is different to porous coated cups. In the porous coated cup, radiolucent line more than 2 mm in all zones was evidence of cup loosening. This means that HA-coated cup loosening could abruptly occurred by mechanical force on weak bone-implant integration. In addition, abrasion by HA particles separated from the surface of cup could lead to increased wear of the PE liner, periprosthetic osteolysis and early loosening. This mechanism might be responsible for the inferior survival of HA-coated cup seen in other studies^{15,16,17,18,19}20).

A great number of studies have reported excellent long-term survival about the HA-coated stems^7,1114). For the femoral component, focal periprosthetic osteolysis could be overcome because the cone-shaped geometry of the proximal femur might encourage osteointegration. However, the acetabular geometry is hemispherical shape and it is known to have little resistance to mechanical loads and shear force^18,1920). Ten hips were performed only curettage and allobone graft during reoperation because there was no evidence of loosening. However, 7 hips out of these had re-reoperation for delayed cup migration, especially in the cases of removal of screws. Because the fixation was not enough to stabilize the cup into the bone, the acetabular cup might depend solely on the screws to resist the shear stress²⁸29). This means that the stability of HA-coated cup in vivo has been incomplete.

Our cases of late loosening over ten years might be associated with development of osteolysis (Fig. 2). This pattern was similar to porous coated cup. In this study, acetabular osteolysis was seen in 47 hips (71.2%). In Blacha's study15) about 65 hips, he observed acetabular osteolysis in 18 hips (27.7%) and 15 hips (23.1%) were revised because of loosening or progressive osteolysis. The report showed the 9-year survival was only 59%. Several authors have raised concerns regarding the high frequency of acetabular osteolysis¹⁵17). In contrast, Oosterbos et al.30) reported 10-year survival rates in excess of 95% with the ABG I prosthesis. Baker et al.31) demonstrated that only 17% of hips in their series of 69 hips with ABG I prostheses had shown acetabular osteolysis at a mean of 15 years. In our series, the acetabular osteolysis occurred less in patients with longterm survival. It is not clear why frequency of acetabular osteolysis varies so widely among these series. In reports showing excellent results, the wear rate and the degree of osteolysis was relatively low. Because many factors including mechanical and biological elements could be cause of periprosthetic osteolysis, the difference might be related to the difference of patient's factor, surgical procedure, and design of prosthesis. A PE wear rate as threshold of osteolysis is known 0.1 mm/year32), and numerous authors have reported acetabular PE wear rates between 0.24 and 0.32 mm/year in studies suggesting the failure of HA coated cup^15,1617). The mean annual PE wear rate was 0.29 mm/year in our study, and it might also have contributed to the high frequency of osteolysis.

This study has several limitations. First, it is a retrospective study which is dependent on past documentation completely. Second, the sample size was relatively small and the rate of follow-up loss was relatively high. Third, there was no control group for direct comparison. Nevertheless, the strengths of the study include a long-term follow-up over 20 years and consistent surgical method which was performed by one surgeon.

We conclude that the long-term survival rate of THA using HA coated acetabular cup was unsatisfactory, and it was attributed to vulnerable property of HA coating and osteolysis.