Dual-Energy X-Ray Absorptiometry (DEXA) Evaluation of the Bone Remodeling Effects of a Low-Modulus Composite Hip Stem After 2 Decades of Follow-Up

Abstract

Background: The Epoch FullCoat Hip Stem (Zimmer) was an isoelastic composite femoral stem developed to address stem stiffness concerns. Purpose: We sought to evaluate the long-term bone mineral density (BMD) of a cohort of patients who underwent total hip arthroplasty (THA) using the Epoch isoelastic stem and having more than 2-decade follow-up. Methods: We conducted a retrospective chart review of all patients who were study subjects at our institution in a multicenter prospective trial for the Food and Drug Administration of the Epoch implant in the mid-1990s. Through this, we identified 16 patients who had dual-energy X-ray absorptiometry (DEXA) scans, with which we could determine BMD preoperatively and at 3 points postoperatively. Of these, 5 agreed to participate in the study (the others were deceased, unable or declined to participate, or were lost to follow-up) with mean follow-up of 22 years. These participants underwent clinical and radiographic evaluation consisting of a Harris hip score, anteroposterior (AP) pelvis and AP and lateral hip X-rays, and DEXA evaluation of both hips. BMD in the 7 Gruen zones at last follow-up was compared with immediate postoperative and 2-year follow-up. Results: At last follow-up, all stems were well-fixed with signs of extensive osteointegration. In proximal Gruen zones 1 and 7, patients underwent a decrease in BMD with more modest losses in Gruen zone 1. All patients demonstrated an increase in BMD in zones 2 through 6 at latest follow-up, except for 1 patient in Gruen zone 6. BMD changes were not limited to the first 2 years of follow-up. Conclusion: This small follow-up cohort study found excellent long-term clinical results, no plain radiographic signs of notable stress shielding, and general maintenance of BMD at a follow-up of over 20 years for this isoelastic stem. Long-term bone remodeling after implantation of the isoelastic stem resulted in increased BMD in Gruen zones 2 through 6, suggesting that composite implant designs may still have a role in THA.

Introduction

Aseptic loosening of the femoral component following cemented total hip arthroplasty (THA) was the stimulus to investigate options for biologic implant fixation. In the 1980s, early results with cementless femoral components demonstrated success with extensively porous-coated stems designed to provide for tissue in-growth over the entire length of the implant and proximally porous-coated implants for tissue ingrowth into the metaphyseal portion of the implant [10,11,23,26]. Implantation of a metal prosthesis into the femur creates a complex load-sharing environment between the metal prosthesis and the native proximal femur [17,24]. If stresses seen by the proximal femur are substantially decreased, stress shielding may occur; this has been recognized as a possible long-term concern following THA with uncemented metal femoral components.

Novel approaches to address stem stiffness concerns include so-called “isoelastic stems” made from polymeric or composite materials [22,35]. Unfortunately, nearly all composite stems met with failure for 2 reasons: the stem’s inability to withstand physiological stresses (leading to implant fracture) and micromotion at the bone-implant interface [3,7] with resultant ingrowth failure and osteolysis. One stem design saw clinical failures ranging from 36% [35] to 58% [22] related to composite material fatigue and another saw 92% fail secondary to aseptic loosening [1]. Although composite material designs were once thought to be the “future of joint replacement” [36], the high rate of clinical failure of most of these devices [1,3,22,35] shifted interest away from this implant design concept.

The Epoch FullCoat Hip Stem (Zimmer, Warsaw, IN, USA), an isoelastic composite femoral stem, is unique in that it was clinically successful and did not experience the high rates of mechanical failure of other isoelastic stems. In fact, it demonstrated outstanding clinical and radiographic results in early [13,19], intermediate [2], and long-term [16] studies. The original version of this stem was a fully porous-coated implant with a cobalt chromium (CoCr) core, outer porous surface with titanium metal fibers with an intervening polyaryletherketone (PAEK) layer which generated a bending stiffness 25% of CoCr and 50% of Titanium alloy equivalents [13,16]. The CoCr core size increased proportionally as stem diameter increased more proximally to match the bending stiffness seen in the native femur [25]. The proximal geometry of the implant was designed for a high degree of metaphyseal fit.

As part of a Food and Drug Administration (FDA)-approved investigative trial, this stem was implanted at 11 institutions across the United States (and 10 international sites) in the mid-1990s. Six-year results including clinical, radiographic, and histologic outcomes from this population were previously reported by Akhavan et al [2] in 2006; 2 stems were available for histologic evaluation and demonstrated evidence of excellent osseous integration. Hartzband et al [16] reported outstanding clinical and radiographic results at 10 years, which included 100% survivorship. The incidence of thigh pain ranged from 3% to 6% at 2 to 6 years and was 0% after 8 years. Only 3 failures of this isoelastic composite stem have been reported, all of which occurred at the interface of the newer PEEK/CoCr core (PAEK was replaced by PEEK with the second generation in 2006). Saltzman et al [31] reported 2 failures in the setting of infection, and Bartelstein et al [4] reported 1 failure associated with high energy trauma using the original PAEK composite stem.

The purpose of this study was to evaluate the long-term results of a subset of patients who underwent THA using this isoelastic composite femoral stem with a greater than 2-decade follow-up; our specific focus was on long-term changes in bone mineral density (BMD) in the proximal femur. Specifically, we asked the following: (1) how does BMD in the proximal femur change with time and with Gruen zone location; and (2) does BMD in the proximal femur stabilize after 2 years of implantation with this isoelastic, composite femoral component?

Methods

We performed a retrospective chart review of all patients who were study subjects at our institution in a multi-center FDA prospective trial of the Epoch FullCoat Hip Stem that was initiated in the mid-1990s. To evaluate the potential impact of this stem design on BMD following implantation, the original protocol for this study provided for a subgroup of patients who had preoperative and postoperative dual-energy X-ray absorptiometry (DEXA) scans as reported previously [2]. Institutional Review Board permission was granted for this study.

Initial inclusion criteria for the patients in the FDA study included: non-inflammatory arthritis of the hip, weight of <100 kg, and age between 21 and 75 years. Between December 1994 and December 1998, 28 patients received a THA using the isoelastic composite femoral stem (Fig. 1) of which 16 were enrolled in the DEXA portion of the trial. All femurs were classified as A or B as defined by Dorr et al [9]. All procedures were performed by a single joint replacement surgeon (Victor M. Goldberg, MD, now deceased) at our institution through a standard posterolateral approach. All patients received 28 mm diameter femoral heads and conventional polyethylene liners, which were sterilized with gamma radiation in nitrogen. All patients were kept 50% weight bearing with 2 crutches for 6 weeks before progressing to full weight bearing without assistive device, which was accomplished by 8 weeks post-op for all patients. Standardized clinical (Harris Hip Score [15]) and radiographic evaluation (AP pelvis, AP and lateral hip plain radiographs) was performed preoperatively and at 3, 6, 12, and 24 months postoperatively.

Fig. 1.

The Epoch (Zimmer, Warsaw, IN, USA) isoelastic composite femoral stem. PAEK Polyaryletherketone.

The protocol for patients enrolled in the DEXA portion of the prospective FDA trial involved standardized DEXA scans preoperatively, within 1 week postoperatively, and then at approximately 3, 6, 12, and 24 months postoperatively. The initial study protocol incorporated strict positioning by the same radiology technician using the same DEXA scanner among other guidelines. A Lunar DPX absorptiometer scanner was used for all DEXA scan evaluations. Measurements of BMD were made at each of the zones defined by Gruen et al [14] in the operative hip preoperatively, immediately postoperatively, and thereafter.

We attempted to contact all 16 DEXA patients from the FDA trial. Of these, 5 agreed to participate in long-term follow-up (2 were deceased, 1 was unable, 1 declined to participate, and 7 were lost to follow-up) (Table 1). The mean follow-up was 22.0 years (range 21.2 to 22.6 years). Three patients received a 28-mm alumina femoral head and 2 received a CoCr femoral head. Two patients received Harris-Galante porous II (HGPII), and 3 received Trilogy acetabular components.

Table 1.

Patient demographics and HHS.

Patient	Sex	Age a (years)	Diagnosis	BMI (kg/m2)	f/u (years)	Femoral osteolysis	HHS pre-op	HHS final f/u	Contralateral THA?	Comments
1	F	50	OA (DDH)	23.9	22.4	Limited (zone 1)	61	100	Y	No revisions or complications.
2	F	42	OA (DDH)	18.3	22.6	None	38	97	Y	Periprosthetic fracture 8 weeks post-op, fixed with cerclage plate and screws. Plate removed 2 years after ORIF.
3	M	44	OA	30.3	22.3	Extensive (zone 1, posterior proximal)	67	100	Y	No revisions or complications.
4	M	62	OA	29.1	21.4	Limited (zones 1 and 7)	54	94	N	No revisions or complications.
5	M	45	OA	24.4	21.2	None	64	99	Y	No revisions or complications

HHS Harris Hip Scores, BMI body mass index, THA total hip arthroplasty, OA osteoarthritis, DDH developmental dysplasia of hip, ORIF open reduction internal fixation.

^aAt time of index procedure.

The 5 participants in our study underwent standard clinical and radiographic evaluation consisting of a Harris hip score, standard anteroposterior (AP) pelvis, AP and lateral hip X-rays, and DEXA evaluation of BMD for both hips. The senior author (M.J.K.), who was not the operating surgeon for these patients, performed all clinical evaluations and evaluated X-rays for signs of loosening, substantive osteolysis, and polyethylene wear. BMD values were compared with immediate postoperative measurements. The positioning protocols used in the previous DEXA studies were closely adhered to by our most senior DEXA radiology technician. BMD in the 7 Gruen zones for the operative and contralateral hip was determined by DEXA scan using a Lunar PRODIGY Absorptiometer and the manufacturer’s orthopedic software. At final follow-up, BMD in the 7 Gruen zones was compared with immediate postoperative and 2-year follow-up values. Percentage change was calculated and change in BMD was plotted versus time from immediate postoperative measurements to each subsequent follow-up.

Results

At the time of last follow-up, all stems were well fixed with signs of extensive osteointegration (Fig. 2a, b). Only 1 patient (Patient 2) required reoperation for any reason. This patient suffered a displaced distal periprosthetic fracture from a fall at 8 weeks post-op. The stem was well fixed, and the fracture was treated by open reduction and internal fixation. The cable and plate construct used to fix the fracture was removed 2 years later. Limited osteolysis related to polyethylene wear was noted in 2 patients—neither of whom were symptomatic. One patient had substantial osteolysis related to polyethylene wear and revision has been recommended. At last follow-up, median Harris Hip Score was 99 (range 94–100) at just over 2 decades of survivorship. In all, 4 of 5 patients had the contralateral hip replaced, with 1 patient (patient 4) having retained their native proximal femur at the time of latest follow-up.

Fig. 2.

(a) 6-week post-op anteroposterior radiograph of patient with Epoch stem. (b) 21.2-year follow-up anteroposterior radiograph of patient with Epoch stem. Patient has acetabular wear and limited bone resorption in zone 7 above level of lesser troch than could be due to osteolysis and/or stress shielding. Cortical and cancellous bone density in other zones and remainder of zone 7 appear well maintained.

In Gruen zone 1, patients demonstrated an overall decrease or little change in BMD, though 1 patient experienced a notable increase from initial post-op to the latest follow-up (Fig. 3a). Overall BMD changes in Gruen zone 1 were modest, with only 2 patients demonstrating losses greater than the 1% loss per year estimated to occur with normal aging [30] (Table 2).

Fig. 3.

(a) BMD change seen in Gruen zone 1. (b) BMD change seen in Gruen zone 2. (c) BMD change seen in Gruen zone 3. (d) BMD change seen in Gruen zone 4. (e) BMD change seen in Gruen zone 5. (f) BMD change seen in Gruen zone 6. (g) BMD change seen in Gruen zone 7. BMD bone mineral density.

Table 2.

Percent change of bone mineral density from immediate post-op to 22-year dual-energy X-ray absorptiometry follow-up.

Patient	GZ 1	GZ 2	GZ 3	GZ 4	GZ 5	GZ 6	GZ 7
1	−23.2	20.2	44.3	10.0	35.2	13.0	−33.7
2	−27.6	49.8	37.7	54.9	45.3	29.0	−32.9
3	34.1	46.6	160.5	17.2	52.2	18.2	−35.1
4	−10.7	12.1	26.5	4.3	20.4	−0.5	−48.9
5	−7.3	6.5	38.4	3.9	19.4	6.5	−44.6

When considering the change from the 2-year DEXA scan to latest follow-up, 2 patients demonstrated continued decrease in BMD while 3 patients—including 2 who initially demonstrated a decrease from 0 to 2 years—demonstrated an increase in BMD in Gruen zone 1 (Table 3).

Table 3.

Percent change of bone mineral density from 2- to 22-year dual-energy X-ray absorptiometry follow-up.

Patient	GZ 1	GZ 2	GZ 3	GZ 4	GZ 5	GZ 6	GZ 7
1	−18.9	25.8	58.1	3.4	90.6	16.4	−25.8
2	−7.8	41.6	96.9	31.6	6.8	58.7	−19.1
3	23.9	16.9	59.6	9.2	22.5	7.3	−17.9
4	7.1	23.1	17.1	10.3	21.3	22.3	−13.5
5	5.0	6.3	38.7	6.5	18.8	22.1	−6.7

All patients demonstrated an increase in BMD in Gruen zones 2, 3, 4, 5, and 6 from the initial post-op scan until last follow-up (Fig. 3b–f) except for patient 4 who demonstrated a small (0.5%) decrease in Gruen zone 6 (Table 2). We found that for all patients in our cohort, BMD in Gruen zone 7 decreased with time in and continued to decrease even after 2-year follow-up (Fig. 3g, Table 3). This was consistent with prior reports with short-term follow-up [2,6,30,34].

Discussion

This small cohort follow-up study reports the clinical and radiographic outcomes of 5 patients with a low-modulus composite femoral component after more than 2 decades. There were no mechanical failures of these composite stems and all 5 patients had excellent long-term clinical outcomes. Despite the extensively porous-coated design of this stem and concerns about distal fixation and related stress shielding, we observed consistent plain radiographic signs of osteointegration over the distal, mid, and proximal portion in our study patients. We also observed consistent DEXA scan evidence of increases in BMD in Gruen zones 2 through 6 and limitation of decreases in BMD exclusively to zone 7 and to a lesser extent zone 1. These increases in BMD occurred despite the potential age-related decreases in BMD in the proximal femur that one would anticipate over the mean 22-year follow-up. Age-related loss in BMD can be reasonably approximated to occur at 1% per year [30].

Limitations of this study include the small size of the study group, changes in DEXA scanner technology that occurred over the long follow-up period, and lack of a control group. Although the initial FDA study protocol enlisted 16 patients to undergo serial DEXA scans over the initial study period, only 5 patients were available for follow-up evaluation at a mean of 22 years postoperatively. Given the small percentage of the initial study group that was available for long-term follow-up, we acknowledge the selection bias introduced into this study regarding clinical outcomes; however, the X-ray and DEXA changes should still be considered important at such lengthy follow-up. Upgrades in DEXA scanner technology implemented at our hospital over the 22-year period also could have affected the results of this study. Since highly reproducible, serial BMD measurements are essential for evaluation and treatment of patients with osteoporosis, elimination of technical variables that could affect DEXA scan results is important. As a result, the DEXA scanners used for this study underwent detailed calibration and validation when they were upgraded, which should have minimized variations associated with these technology upgrades to less than 1%. Several studies have been performed evaluating appropriate cross-calibration when upgrading to a new DEXA scanner technology in order to eliminate measurement error due to this factor [5,12,28]. To our knowledge, there are no studies specifically evaluating the Gruen zone software used in this study or that directly compare the DPX to PRODIGY scanners. Importantly, both scanners used for data collection in this study are from the same manufacturer, which may result in less variance in BMD values than scanners from different manufacturers [29]. One published study has shown 4% to 6% change in BMD during a cross-calibration study [5] between the DPX-L (a slightly later GE model than the DPX) and the PRODIGY (used in this study), yet still others show no or minimally significant differences [12]. There are multiple conversion factors [5,12,28] in the literature that were considered, but these calculations were complex, varied, and would not affect the overall data trend given no statistical analysis was performed for significance—so they were not used. Additionally, although variation in patient positioning, and use of a different technician could have introduced measurement error, we tried to minimize this by having all studies performed by our most experienced radiology technician who has extensive experience in these studies. And finally, ideally, the patient’s unreplaced contralateral hip would have served as a control to determine changes in BMD related to the patient’s isoelastic composite stem in comparison to the normal age-related changes in BMD. However, 4 of these 5 patients underwent contralateral THA at a variable period of time after their study THA, which made comparison with the contralateral hip challenging to conduct.

To address the problem of aseptic loosening of cemented femoral components in the 1980s, 2 different philosophies for biologic fixation of the femoral component evolved. One of these was the extensively porous coated, cylindrical, diaphyseal fitting stem that was designed to primarily obtain distal fixation with some potential for tissue in-growth over the entire length of the implant. The other was the metaphyseal fitting, proximally porous coated implant that was intended to provide for tissue ingrowth into the metaphyseal portion of the implant [10,11,23,26]. The propensity for extensively porous coated stems to achieve solid fixation in the diaphysis has, however, raised concerns about proximal stress shielding, since this scenario results in bypassing of physiologic loads normally seen by the more proximal regions of the femur. The extent of this phenomena has been reported with several different extensively coated stems. Engh et al [11] reported “pronounced” stress shielding (defined subjectively as significant radiographic bone loss distal to both Gruen zone 1 and 7) in 23% of their extensively porous coated AML patients [11]. We did not see radiographically significant stress shielding of this degree in any patient in this study.

More quantitative assessment of changes in BMD after THA using DEXA scans are few in number, and are typically limited to short-term follow-up and evaluation of proximally porous-coated stems. In a randomized controlled trial that assessed changes in BMD at 2 years post-op with DEXA scans, McDonald et al [23] showed no differences in BMD at 2 years postoperative (except in zone 7) between a proximally porous-coated stem vs a second-generation extensively porous-coated stem. BMD in zone 7 was reduced by 24% in the fully coated stem vs 15% in the proximally coated stem.

We are unaware of any published studies that have used DEXA scans to evaluate other extensively coated devices that we can use to reasonably compare with the results of our study. There is universal agreement that all femoral components result in a variable degree of loss of BMD in Gruen zone 7. The implant evaluated in this study, however, resulted in changes in BMD primarily limited to zone 7 as well as remarkably consistent increases in BMD in Gruen zones 2 through 6, a finding infrequently seen with either proximally coated or extensively coated stems [6, 11, 30].

Although clinically significant stress shielding has been reported with the use of extensively porous-coated devices, it has been suggested that this effect seems to be limited and most often associated with large diameter stems (>15mm) and osteopenic patients. In these situations, the increased elastic modulus and associated stiffness of these large diameter, extensively porous-coated, cobalt chromium alloy stems, relative to the native femur, likely plays an important role in development of stress shielding. Based on this assumption, the low-modulus design of the stem used in this study would be expected to minimize the stress shielding effects of these diaphyseal canal filling implant designs. This concept is supported by the results of our study which demonstrated consistent increases in BMD in zones 2 through 6, which was maintained over the average 22-year follow-up.

Our long-term observations of localized decreases in BMD to the proximal Gruen zones 7 and, to a lesser extent, zone 1, as well as uniform increases in BMD in Gruen zones 2 through 6, suggest that effective loading of the proximal femur occurs with this low-modulus stem. The implant used in this study is somewhat unique and unlike other extensively coated stems in that it has a metaphyseal filling geometry that is identical to a clinically successful proximally coated femoral component. In a prospective randomized trial of this isoelastic composite stem vs a proximally porous-coated cementless stem with an identical proximal geometry (Zimmer Anatomic), Karrholm et al [19] reported more favorable maintenance of BMD in the proximal Gruen zones at 2-years post-op with this stem, similar to our observations in the current study. Based on their study, Karrholm et al [19] suggested that despite the extensively coated design of the Epoch stem, it appeared to load the proximal femur more evenly than the Anatomic stem, which had an identical metaphyseal fitting proximal implant geometry. The results of our study support their conclusion that the extensively porous-coated Epoch stem has biomechanical characteristics more typical of a proximally porous-coated implant. Based on their multi-variant analysis, they also suggested that postoperative changes in proximal Gruen zone BMD were related to stem design factors and distal Gruen zone (3 and 5) BMD was related to patient factors (ie, body weight, age, sex). It is certainly possible that the Epoch stem provides both the metaphyseal loading characteristics of a proximally coated implant and additional stress shielding mitigation in the diaphyseal regions of the femur due to its low-modulus, composite material distal stem design.

Unlike our long-term observations with an isoelastic composite stem, Boden et al [6] demonstrated consistent decreases in BMD in multiple Gruen zones in both the operative and contralateral non-replaced side following THA with a proximally porous-coated, tapered cementless stem. These decreases in BMD continued bilaterally for upward of 14 years postoperatively, suggesting that the cementless femoral component used in their study had a substantial impact on stress-related bone remodeling in both the mid and proximal Gruen zones. Studies with other proximally coated cementless stems have demonstrated that changes in BMD are frequently limited to the more proximal Gruen zones 1 and 7 [19,23,37]. These changes in BMD in zones 1 and 7 are similar to our long-term observations with the low-modulus extensively coated stem evaluated in this study, suggesting that adequate proximal loading of the femur is still maintained with this implant and behavior is similar to many proximally porous-coated implants.

Although short-term follow-up studies have suggested that changes in BMD stabilizes 2 years after THA [21,27], other studies with a wide variety of cementless stems designed to achieve proximal fixation have shown that changes in BMD may continue for considerably longer following THA [6,16]. These long-term changes can be secondary to the effects of normal aging, as well as stress-related bone remodeling following THA. In the study of Boden et al [6], longitudinal decreases in BMD in the contralateral unreplaced side presumably represent changes that occurred in relation to normal aging in their study patients. These presumed age-related decreases in BMD varied across the different Gruen zones and ranged from approximately 3 to 7% over their 10-year study period. Peitgen et al [30] reported continued decreases in BMD in zones 1, 4, 6, and 7 from 12 to 21 years postoperatively in a large group of cementless tapered titanium stems designed to achieve proximal fixation. In our study patients, despite the over 2 decades of potential age-related decline in bone density, BMD was noted to actually increase substantially in Gruen zones 2 through 6, while decreases in zones 1 and 7 were relatively modest in all patients, suggesting that this composite stem may promote more physiologic loading of the proximal femur and actually limit stress-shielding.

Maintenance and increase in BMD over time for all but the most proximal Gruen zones using the extensively porous-coated femoral component seems counterintuitive since fixation can—and radiographically did—occur distally in our study subjects. In contrast, other studies evaluating changes in BMD with implants designed to achieve proximal fixation have shown unanticipated decreases in BMD in the proximal and mid Gruen zones, suggesting that physiologic metaphyseal loading was not met with these devices [6,30].

As a result, despite the clinical success of proximally coated metaphyseal fitting tapered stems, long-term concerns regarding stress shielding may still exist. Clearly, “normal” physiologic loading of bone after THA is determined by a complex interaction between location of ingrowth, location and extent of endosteal contact of the implant in the proximal femur, stiffness of the stem, and other implant design and patient-related factors. Although serious clinical consequences of stress shielding are infrequent and considered by some to be primarily theoretical [8,18,33], the extension of indications for THA to younger patients and increasing life expectancy mandates continued concern regarding future implant designs and the potential long-term risk of stress shielding [32].

In summary, the results of this follow-up study in a cohort of 5 patients showed that this isoelastic composite femoral stem produced excellent long-term clinical outcomes, no plain radiographic signs of notable stress shielding, and general maintenance of BMD at a mean follow-up of 22 years. As is typically seen with all cementless femoral components, continued loss of BMD in Gruen zone 7 occurred with time in our 5 patients. However, unlike several other extensively porous-coated [20,37] and proximally porous-coated implants designed to obtain proximal fixation [6,30], long-term bone remodeling after implantation of this isoelastic composite stem actually resulted in increased BMD in Gruen zones 2 through 6 in our 5 patients. Although the Zimmer Epoch stem is no longer available, the findings of this unique isoelastic porous-coated composite design [25] suggest that similar composite implant design concepts may still have a role in the future of hip arthroplasty.