- Email: [email protected]
Stress Shielding Around Radial Head Prostheses
SCIENTIFIC ARTICLE
Stress Shielding Around Radial Head Prostheses Cholawish Chanlalit, MD, Dave R. Shukla, MB, BCh, James S. Fitzsimmons, BSc, Kai-Nan An, PhD, Shawn W. O’Driscoll, MD, PhD Purpose Stress shielding is known to occur around rigidly fixed implants. We hypothesized that stress shielding around radial head prostheses is common but nonprogressive. In this study, we present a classification scheme to support our radiographic observations. Methods We reviewed charts and radiographs of 86 cases from 79 patients with radial head implants from both primary and revision surgeries between 1999 and 2009. Exclusion criteria included infection, loosening, or follow-up of less than 12 months. We classified stress shielding as: I, cortical thinning; II, partially (IIa) or circumferentially (IIb) exposed stem; and III, impending mechanical failure. Results Of 26 well-fixed stems, 17 (63%) demonstrated stress shielding: I ⫽ 2, II ⫽ 15 (IIa ⫽ 12, IIb ⫽ 3), and III ⫽ 0. We saw stress shielding with all stem types: cemented or noncemented; long or short; and straight, curved, or tapered. The only significant difference was that stems implanted into the radial shaft had less stress shielding than stems implanted into the neck or tuberosity (P ⫽ .03). The average follow-up was 33 months (range, 13–70 mo). Stress shielding was detectable by an average of 11 months (range, 1–15 mo). The pattern of bone loss was similar in 16 of 17 cases (94%), starting on the outer periosteal cortex. The 3 cases with circumferential exposure of the stem (stage IIb) averaged 2.6 mm (range, 1– 4 mm) of exposed stem. Stress shielding never extended to the bicipital tuberosity, and there were no cases of impending mechanical failure. Conclusions Stress shielding around radial head prostheses is common, regardless of stem design. However, it is typically minor, nonprogressive, and of questionable clinical consequence. (J Hand Surg 2012;37A:2118–2125. Copyright © 2012 by the American Society for Surgery of the Hand. All rights reserved.) Type of study/level of evidence Therapeutic IV. Key words Loosening, radial head prosthesis, radiographic bone loss, stress shielding. HE USUAL INDICATIONS for radial head replacement are displaced, comminuted fractures of the radial head, especially when associated with complex elbow or forearm instability.1,2 Patients who
T
From the Biomechanics Laboratory, Department of Orthopedic Surgery, Mayo Clinic, Rochester MN; and the Department of Orthopedics, Faculty of Medicine, HRH Princess Maha Chakri Sirindhorn Medical Center, Srinakhrinwirot University, Bangkok, Thailand. Received for publication October 4, 2011; accepted in revised form June 21, 2012. ThisstudywasfundedbytheMayoFoundation.S.W.O.receivesroyaltiesfromAcumed,LLC,which is related to the subject of this work. The Mayo Foundation also receives royalties from this company. Corresponding author: Shawn W. O’Driscoll, MD, PhD, Mayo Clinic, 200 First Street, SW, Rochester, MN 55905; e-mail: [email protected]. 0363-5023/12/37A10-0024$36.00/0 http://dx.doi.org/10.1016/j.jhsa.2012.06.020
2118 䉬 © ASSH 䉬 Published by Elsevier, Inc. All rights reserved.
require treatment for this type of injury are often young and active.1–3 Thus, there are many years during which complications could develop after prosthetic replacement of the radial head. Stress shielding refers to bone loss around an implant in response to altered (diminished) mechanical stress.4 –7 This phenomenon is seen with well-fixed implants in which bone resorption occurs in regions where load transfer across the implant bypasses a portion of bone. There is no clear documentation of any statistical correlation between stress shielding and complications involving implants in the hip, where this phenomenon has been most studied. However, potential complications discussed in the literature include reduced implant longevity and increased fracture risk.4,6,8–11
RADIAL HEAD PROSTHESES STRESS SHIELDING
2119
FIGURE 1: A Anteroposterior x-ray in neutral rotation optimizing dorsal and volar cortex visualization showing no stress shielding. B Lateral view optimizing medial and lateral cortex visualization showing no stress shielding. C Stress shielding was quantitated by measuring periosteal (PL) bone loss and exposed stem (ES). Copyright © 2012 Mayo Foundation. Used with permission.
Popovic et al12 observed lucencies around 31% of bipolar radial head prosthetic replacements at midterm follow-up. However, the authors attributed bone resorption to polyethylene wear, rather than to stress shielding. We performed the present study to investigate the prevalence and radiographic characteristics of stress shielding around radial head prostheses. Our hypothesis was that stress shielding around radial head prostheses is common but nonprogressive. In this study, we present a classification scheme to support our radiographic observations. MATERIALS AND METHODS We conducted a retrospective review study evaluating 86 consecutive cases from 79 patients on whom the senior author (S.O.D.) performed insertion, removal, or revision of a radial head replacement between August 1, 1999, and March 30, 2009. After we obtained patient consent and institutional review board approval, 42 cases in 40 patients met exclusion criteria, including infection, loosening, inadequate follow-up (⬍ 12 mo), having had a custom-made prosthesis, or having an-
other risk factor for bone loss (ie, chronic regional pain syndrome, particle-induced osteolysis resulting from a disengaging bipolar implant). Indications for radial head prosthetic replacement were acute or chronic radial head fracture associated with elbow or forearm instability in 8 cases and radiocapitellar arthritis in 3 cases; revision in 11 cases for overstuffed, loose, malaligned, or disengaged radial head; prior resection in 3 cases; and osteochondromatosis in 1. An observer who was a trained orthopedic surgeon (C.C.) reviewed radiographs and records of the remaining 26 cases of radial head arthroplasty in 26 patients. Multiple variables were reviewed, including type of radial head prosthesis, operative procedure, and radiographic appearance. Types of radial head prostheses investigated included cemented and noncemented stems and long and short stems. In addition, we examined straight, curved, and tapered stems. We performed radiographic assessment immediately postoperatively and at 6 weeks, 3 months, 6 months, 12 months, and annually thereafter. We obtained anteroposterior (AP) and lateral plain film
JHS 䉬 Vol A, October
2120
RADIAL HEAD PROSTHESES STRESS SHIELDING
radiographs of the elbow in a consistent manner according to standard x-ray practices for elbow radiographs, with the patient seated on a stool and the upper extremity resting on a table. We took the AP radiograph with the shoulder flexed forward, the elbow extended as much as possible, and the forearm in supinated or neutral rotation, depending on the range of the patient’s forearm rotation. We took lateral radiographs with the forearm in neutral rotation. These positions allowed the 4 sides of the radial neck to be imaged with plain films, as shown in Figure 1. The beam was centered on the midpoint of the antecubital crease, perpendicular to the elbow. We obtained the lateral radiograph with the elbow flexed to 90° and the shoulder internally rotated while the arm and elbow were supported on the table. We documented the timing, location, and pattern of bone loss from stress shielding. The distal extent of bone loss on the periosteal and endosteal surfaces was measured from the collar of the implant. We digitized the radiographic images in Digital Imaging and Communications in Medicine format and measured them using QREADS Clinical Image Viewer (Mayo Clinic, Rochester, MN), a picture archiving and communication system. It was possible to infer a 3-dimensional image from the plain films by imagining that in each view different cortices of the proximal radius were visible. We measured 2 cortical bone lines in each image: periosteal bone loss, representing the outer cortex, and endosteal bone loss. We assessed periosteal loss by examining the existing periosteum and imagining a continuation of where that line would lie if it were intact. We measured endosteal length loss, representing the inner cortex loss, in the same manner (Fig. 1). From the data, we classified bone loss from stress shielding according to 2 variables: location and severity. We identified the cortices as lateral, medial, dorsal, and volar, according to the anatomic position.13 Overall severity of bone loss was qualified as follows: stage I: cortical bone thinning; stage IIa: partially exposed stem; stage IIb: circumferentially exposed stem; and stage III: mechanical or impending failure (ie, actual or impending loss of stem fixation or periprosthetic fracture). We defined cortical bone thinning as any visible loss of the width of the cortex but with some endosteal bone remaining around the stem. If the endosteal cortex was breached, the severity was classified as stage II (exposed stem). The labeling of a case as having mechanical or impend-
FIGURE 2: Cumulative graph displaying the sequential detection of stress shielding. We detected the first case of stress shielding at 1 month postoperatively. The latest detection, when all cases had presented, was 15 months. Copyright © 2012 Mayo Foundation. Used with permission.
ing failure would have been done retrospectively. For example, if an implant had failed as the result of stress shielding, the image taken before failure would have qualified as stage III. Statistical analysis We calculated the average lengths of exposed stem (in millimeters) and compared them by performing a nonparametric repeated-measures analysis using a Friedman test. We used a paired sign test to compare the amounts of exposed stem among the 4 cortices (ie, anterior, posterior, medial, and lateral). For both analyses, P ⱕ .05 was considered significant. RESULTS Of the 26 patients, 14 were male and 12 were female; 15 cases underwent primary arthroplasty and the other 11 were revisions. The average follow-up period was 33 months (range, 13–70 mo). Radiographic assessments Seventeen cases (63%) had evidence of stress shielding on plain films. We initially detected stress shielding an average of 7 months postoperatively and as early as 1 month postoperatively (Fig. 2). All cases of stress shielding had presented by 15 months, the progression of which is shown in Figure 3. Cortex involvement Of the 17 cases, there was lateral cortex involvement in 12 (71%), volar cortex involvement in 11 (65%), dorsal cortex involvement in 10 (59%), and medial cortex
JHS 䉬 Vol A, October
RADIAL HEAD PROSTHESES STRESS SHIELDING
2121
FIGURE 4: Average lengths and standard error bars of exposed radial head prosthetic stem by cortex. Cortices with different lowercase letters are significantly different from each other (P ⫽ .01). Copyright © 2012 Mayo Foundation. Used with permission.
involvement in 5 (29%). The amount of exposed stem varied depending on which cortex was being measured (P ⫽ .02). Paired sign test analysis revealed that the amount of exposed stem in the medial cortex was significantly less than in the lateral and volar cortex (P ⬍ .02), whereas the lateral and volar cortices were not significantly different from each other (P ⫽ .09). The length of exposed stem in the dorsal cortex was similar to that of the medial cortex (P ⫽ .18). Initiation of stress shielding In 7 patients, we observed stress shielding to affect 1 cortex in isolation before involving other cortices. In 5 patients (71%), the lateral cortex experienced bone loss first and in isolation. In 1 patient, the volar cortex was affected first and in isolation; the medial cortex was also affected first in 1 patient. For all patients, the average length of exposed stem in any cortex averaged 2 mm (range, 0 –5 mm) (Fig. 4).
FIGURE 3: Progression of stress shielding detected at 3, 6, 9, 12, and 15 months postoperatively. Copyright © 2012 Mayo Foundation. Used with permission.
Pattern and severity of stress shielding The shape pattern of stress shielding can be conveyed by measuring periosteal bone loss and endosteal bone loss. Periosteal bone loss was consistently more extensive than endosteal bone loss (Figs. 5, 6). For the 3 patients with circumferential bone resorption (stage IIb), the mean length of exposed stem averaged 2.3 mm (range, 0 – 4 mm) (Table 1). Stress shielding did not extend to the bicipital tuberosity. We noticed no common pattern of bone loss in any cases involving cemented stems. To assess for progression of bone resorp-
JHS 䉬 Vol A, October
2122
RADIAL HEAD PROSTHESES STRESS SHIELDING
FIGURE 5: Example of stress shielding beginning in the lateral cortex. A Immediately postoperatively, there was good contact between solid bone and the collar. B Stage I stress shielding was apparent 2 months postoperatively in the lateral cortex. C, D Stage IIb stress shielding 6 and 18 months postoperatively, respectively. Circumferential exposure of the proximal stem was present but was not progressive. Copyright © 2012 Mayo Foundation. Used with permission.
tion, we analyzed the 6 implants with the longest follow-up. Only 1 had evidence of progression (case 16), but the maximum length of exposed stem was 3 mm in any cortex. Stress shielding appeared to be nonprogressive in the other 2 (Table 2) and stabilized by a mean of 23 months (range, 13–27 mo).
Influence of revision Five cases that experienced stress shielding were the result of revision surgery. We statistically analyzed these cases separately. There were no significant differences in the amount of exposed stem in any cortices between the revision and primary cases (P ⫽ .11–.83).
Stress shielding by stem type We saw stress shielding with all 5 prosthetic designs used in this series. No single implant design (ie, by manufacturer) was more predisposed to developing stress shielding when all manufacturers were statistically compared (P ⫽ .08). We also saw stress shielding with cemented as well as cementless prostheses, bipolar and monopolar, regardless of stem length, shape, and surface texture. Some groups were too small to derive valid statistical conclusions, but the only statistically significant factor was the use of a long cemented stem, which had less stress shielding (P ⫽ .03). However, that stem required resection of 19 or 22 mm of head and neck to a level just proximal to the bicipital tuberosity. Therefore, we cannot directly compare it with implants with conventional stem lengths.
DISCUSSION This study demonstrated that stress shielding around well-fixed radial head prostheses is common; it occurred in 17 of 26 prostheses (65%). We noted it with all 5 stem designs studied, whether cemented or porous-coated, and regardless of stem shape, length, and surface texture. However, it was typically minor and nonprogressive, and it never extended to involve the bicipital tuberosity. In this series, no implants appeared to be at risk of failure as a result of radiographic bone loss from stress shielding. We confirmed solid fixation of the stems in a number of cases that required revision. We observed consistent patterns of bone loss. Resorption always began on the periosteal surface. Even when resorption progressed to involve the
JHS 䉬 Vol A, October
RADIAL HEAD PROSTHESES STRESS SHIELDING
FIGURE 6: Three-dimensional reconstructed computed tomography image demonstrating bone loss resulting from stress shielding (case 13). Copyright © 2012 Mayo Foundation. Used with permission.
endosteal surface, periosteal bone loss always exceeded endosteal bone loss. We consistently noted the radiographic appearance of a convex outer surface of the proximal radial neck. Consistent with Wolff’s law4 –7 of bone remodeling, load transfer across the radiocapitellar joint to the radial head prosthetic stem dissipates into the endosteal cortex. This biomechanical concept explains why the periosteal cortex resorbed before the endosteal cortex. This characteristic pattern of bone loss is valuable when distinguishing between stress shielding and bone loss that is caused by infection, loosening, or osteolysis. In our opinion, the latter 3 conditions can cause endosteal or concave erosions, which are not seen in stress shielding. We analyzed bone loss around the prostheses separately for each of the 4 quadrants represented on the AP and lateral radiographs. Cortical involvement was variable; the lateral cortex was most frequently involved, the medial cortex least frequently involved, and dorsal and volar cortical involvement fell in between. Whereas bone loss began in an asymmetrical fashion, it never in-
2123
volved 1 cortex excessively without starting to advance circumferentially until the proximal stem was exposed. We noticed this recurring pattern in 16 of the 17 cases (94%). Endosteal bone loss never exceeded 5 mm in any 1 cortex without involving at least 1 other cortex. In 3 of 17 cases (18%), bone resorption was sufficient to cause circumferential exposure of the proximal stem (stage IIb) for an average length of 2.8 mm. Of the 3 cases with a circumferentially exposed stem, the length of exposed stem on any surface (cortex) averaged 2.3 mm and never exceeded 4.0 mm. The amount of bone loss resulting from stress shielding seen in these cases is unlikely to be detrimental if revision is necessary. If revision was required, the only potential consequence would be that bone loss might require conversion to a longer stem. However, in the experience of the senior author (S.O.D.), revision typically requires conversion to a longer stem in the absence of stress shielding, because of endosteal bone loss and sclerosis that precludes secure fixation of another standard-size stem. The time course of stress shielding was variable, but it was first evident between 1 and 15 months postoperatively. For patients with sufficient interval x-rays before final follow-up, stress shielding did not progress after 26 months. For patients in whom we obtained follow-up beyond 26 months (11 of 19 patients), the average follow-up time was 40 months (range, 26 –70 mo). One patient (case 17) had evidence of progression, but the maximum length of exposed stem was 3 mm in any cortex. Although information is sparse regarding the effect of single cortex loss on radial head implant stability, we think that 3 mm bone loss from a 25-mm stem (12%) would not likely compromise implant fixation. Longer follow-up will be needed to determine whether bone loss from stress shielding can progress to a degree that would be of clinical concern. This study had some limitations, including the use of plain radiographs for measuring and categorizing bone loss. Although we standardized positioning of patient limbs for these radiographs, some variability may have affected apparent changes in bone loss. For example, there may have been some overlap between cortices depending on the orientation of the x-ray beam. However, we were certain that the lateral cortex, where we noted the most bone resorption, was always clearly visualized. Although the accuracy of these measurements might improve with the use of computed tomography imaging, the expense would not be warranted based on the conclusion of this radio-
JHS 䉬 Vol A, October
2124
TABLE 1.
Prosthesis Details, Follow-up Interval, Column Involvement, and Stage of 17 Cases of Stress Shielding Exposed Stem
Case
Prosthesis Type
Cement Status
Primary/Revision
Interval (mo)
Volar (mm)
Dorsal (mm)
Lateral (mm)
Medial (mm)
Mean Exposed Stem (mm)
Stage
1
rHead (Sbi)
Cemented
Primary
38
0
0
0
0
0.0
I
2
Evolve (Wright)
Cemented
Primary
70
0
0
0
0
0.0
I
RHS (Tornier)
Uncemented
Revision
14
2
1
0
1
1.0
IIa
Anatomic (Acumed)
Uncemented
Primary
32
0
5
5
0
3.0
IIa
JHS 䉬 Vol A, October
5
Anatomic (Acumed)
Uncemented
Revision
14
2
2
4
0
2.0
IIa
6
Anatomic (Acumed)
Uncemented
Primary
39
4
2
5
0
3.0
IIa
7
Anatomic (Acumed)
Uncemented
Primary
27
0
0
1
0
0.3
IIa
8
Anatomic (Acumed)
Uncemented
Revision
29
2
4
4
0
3.0
IIa
9 10
Anatomic (Acumed)
Uncemented
Primary
26
1
0
1
0
1.0
IIa
Judet (Tornier)
Cemented
Primary
61
0
3
1
0
1.0
IIa
11
Evolve (Wright)
Cemented
Primary
28
1
0
0
0
0.3
IIa
12
Anatomic (Acumed)
Uncemented
Revision
21
4
0
0
0
1.0
IIa
13
Anatomic (Acumed)
Uncemented
Primary
20
2
1
3
0
2.0
IIa
14
Anatomic (Acumed)
Uncemented
Primary
13
0
0
3
1
3.0
IIa
15
Anatomic (Acumed)
Uncemented
Primary
13
4
1
4
2
3.0
IIb
16
rHead (Sbi)
Uncemented
Primary
48
2
2
3
2
2.0
IIb
17
rHead (Sbi)
Cemented
Revision
61
1
3
2
1
2.0
IIb
13–70
0–4
0–5
0–5
0–2
0–3
33
2
1
2
0.4
2.0
Range Average
RADIAL HEAD PROSTHESES STRESS SHIELDING
3 4
RADIAL HEAD PROSTHESES STRESS SHIELDING
TABLE 2.
Stability Progression Staging at Last Follow-up
3.
Case
Final Follow-up
1
38
I
Stable after 23 mo
2
70
I
Stable after 26 mo
5.
4
39
IIA
Stable after 13 mo
6.
6
39
IIA
Stable after 27 mo
16
48
IIB
Active
17
61
IIB
Stable after 24 mo
Progression
4.
7. 8.
9.
graphic study. Another limitation was the small number of patients and relatively short-term follow-up. Longer-term follow-up with a larger patient population will be needed to elucidate the nature and potential clinical implications, if any, of stress shielding around radial head prostheses. REFERENCES
10.
11.
12.
1. Doornberg JN, Parisien R, van Duijn PJ, Ring D. Radial head arthroplasty with a modular metal spacer to treat acute traumatic elbow instability. J Bone Joint Surg 2007;89A:1075–1080. 2. Moro JK, Werier J, MacDermid JC, Patterson SD, King GJ. Arthro-
13.
2125
plasty with a metal radial head for unreconstructible fractures of the radial head. J Bone Joint Surg 2001;83A:1201–1211. Doornberg JN, Ring D. Coronoid fracture patterns. J Hand Surg 2006;31A:45–52. Engh CA Jr, Young AM, Engh CA Sr, Hopper RH Jr. Clinical consequences of stress shielding after porous-coated total hip arthroplasty. Clin Orthop Relat Res 2003:157–163. Huiskes R, Weinans H, Grootenboer HJ, Dalstra M, Fudala B, Slooff TJ. Adaptive bone-remodeling theory applied to prosthetic-design analysis. J Biomech 1987;20:1135–1150. Huiskes R, Weinans H, van Rietbergen B. The relationship between stress shielding and bone resorption around total hip stems and the effects of flexible materials. Clin Orthop Relat Res 1992:124 –134. Roesler H. The history of some fundamental concepts in bone biomechanics. J Biomech 1987;20:1025–1034. Harris WH. Will stress shielding limit the longevity of cemented femoral components of total hip replacement? Clin Orthop Relat Res 1992:120 –123. Kilgus DJ, Shimaoka EE, Tipton JS, Eberle RW. Dual-energy X-ray absorptiometry measurement of bone mineral density around porouscoated cementless femoral implants. Methods and preliminary results. J Bone Joint Surg 1993;75B:279 –287. McCarthy CK, Steinberg GG, Agren M, Leahey D, Wyman E, Baran DT. Quantifying bone loss from the proximal femur after total hip arthroplasty. J Bone Joint Surg 1991;73B:774 –778. Skoldenberg OG, Boden HS, Salemyr MO, Ahl TE, Adolphson PY. Periprosthetic proximal bone loss after uncemented hip arthroplasty is related to stem size: DXA measurements in 138 patients followed for 2–7 years. Acta Orthop 2006;77:386 –392. Popovic N, Lemaire R, Georis P, Gillet P. Midterm results with a bipolar radial head prosthesis: radiographic evidence of loosening at the bone-cement interface. J Bone Joint Surg 2007;89A:2469 –2476. Basmajian JV. Primary anatomy. 8th ed. Baltimore: Williams and Wilkins, 1985.
JHS 䉬 Vol A, October
Stress Shielding Around Radial Head Prostheses Cholawish Chanlalit, MD, Dave R. Shukla, MB, BCh, James S. Fitzsimmons, BSc, Kai-Nan An, PhD, Shawn W. O’Driscoll, MD, PhD Purpose Stress shielding is known to occur around rigidly fixed implants. We hypothesized that stress shielding around radial head prostheses is common but nonprogressive. In this study, we present a classification scheme to support our radiographic observations. Methods We reviewed charts and radiographs of 86 cases from 79 patients with radial head implants from both primary and revision surgeries between 1999 and 2009. Exclusion criteria included infection, loosening, or follow-up of less than 12 months. We classified stress shielding as: I, cortical thinning; II, partially (IIa) or circumferentially (IIb) exposed stem; and III, impending mechanical failure. Results Of 26 well-fixed stems, 17 (63%) demonstrated stress shielding: I ⫽ 2, II ⫽ 15 (IIa ⫽ 12, IIb ⫽ 3), and III ⫽ 0. We saw stress shielding with all stem types: cemented or noncemented; long or short; and straight, curved, or tapered. The only significant difference was that stems implanted into the radial shaft had less stress shielding than stems implanted into the neck or tuberosity (P ⫽ .03). The average follow-up was 33 months (range, 13–70 mo). Stress shielding was detectable by an average of 11 months (range, 1–15 mo). The pattern of bone loss was similar in 16 of 17 cases (94%), starting on the outer periosteal cortex. The 3 cases with circumferential exposure of the stem (stage IIb) averaged 2.6 mm (range, 1– 4 mm) of exposed stem. Stress shielding never extended to the bicipital tuberosity, and there were no cases of impending mechanical failure. Conclusions Stress shielding around radial head prostheses is common, regardless of stem design. However, it is typically minor, nonprogressive, and of questionable clinical consequence. (J Hand Surg 2012;37A:2118–2125. Copyright © 2012 by the American Society for Surgery of the Hand. All rights reserved.) Type of study/level of evidence Therapeutic IV. Key words Loosening, radial head prosthesis, radiographic bone loss, stress shielding. HE USUAL INDICATIONS for radial head replacement are displaced, comminuted fractures of the radial head, especially when associated with complex elbow or forearm instability.1,2 Patients who
T
From the Biomechanics Laboratory, Department of Orthopedic Surgery, Mayo Clinic, Rochester MN; and the Department of Orthopedics, Faculty of Medicine, HRH Princess Maha Chakri Sirindhorn Medical Center, Srinakhrinwirot University, Bangkok, Thailand. Received for publication October 4, 2011; accepted in revised form June 21, 2012. ThisstudywasfundedbytheMayoFoundation.S.W.O.receivesroyaltiesfromAcumed,LLC,which is related to the subject of this work. The Mayo Foundation also receives royalties from this company. Corresponding author: Shawn W. O’Driscoll, MD, PhD, Mayo Clinic, 200 First Street, SW, Rochester, MN 55905; e-mail: [email protected]. 0363-5023/12/37A10-0024$36.00/0 http://dx.doi.org/10.1016/j.jhsa.2012.06.020
2118 䉬 © ASSH 䉬 Published by Elsevier, Inc. All rights reserved.
require treatment for this type of injury are often young and active.1–3 Thus, there are many years during which complications could develop after prosthetic replacement of the radial head. Stress shielding refers to bone loss around an implant in response to altered (diminished) mechanical stress.4 –7 This phenomenon is seen with well-fixed implants in which bone resorption occurs in regions where load transfer across the implant bypasses a portion of bone. There is no clear documentation of any statistical correlation between stress shielding and complications involving implants in the hip, where this phenomenon has been most studied. However, potential complications discussed in the literature include reduced implant longevity and increased fracture risk.4,6,8–11
RADIAL HEAD PROSTHESES STRESS SHIELDING
2119
FIGURE 1: A Anteroposterior x-ray in neutral rotation optimizing dorsal and volar cortex visualization showing no stress shielding. B Lateral view optimizing medial and lateral cortex visualization showing no stress shielding. C Stress shielding was quantitated by measuring periosteal (PL) bone loss and exposed stem (ES). Copyright © 2012 Mayo Foundation. Used with permission.
Popovic et al12 observed lucencies around 31% of bipolar radial head prosthetic replacements at midterm follow-up. However, the authors attributed bone resorption to polyethylene wear, rather than to stress shielding. We performed the present study to investigate the prevalence and radiographic characteristics of stress shielding around radial head prostheses. Our hypothesis was that stress shielding around radial head prostheses is common but nonprogressive. In this study, we present a classification scheme to support our radiographic observations. MATERIALS AND METHODS We conducted a retrospective review study evaluating 86 consecutive cases from 79 patients on whom the senior author (S.O.D.) performed insertion, removal, or revision of a radial head replacement between August 1, 1999, and March 30, 2009. After we obtained patient consent and institutional review board approval, 42 cases in 40 patients met exclusion criteria, including infection, loosening, inadequate follow-up (⬍ 12 mo), having had a custom-made prosthesis, or having an-
other risk factor for bone loss (ie, chronic regional pain syndrome, particle-induced osteolysis resulting from a disengaging bipolar implant). Indications for radial head prosthetic replacement were acute or chronic radial head fracture associated with elbow or forearm instability in 8 cases and radiocapitellar arthritis in 3 cases; revision in 11 cases for overstuffed, loose, malaligned, or disengaged radial head; prior resection in 3 cases; and osteochondromatosis in 1. An observer who was a trained orthopedic surgeon (C.C.) reviewed radiographs and records of the remaining 26 cases of radial head arthroplasty in 26 patients. Multiple variables were reviewed, including type of radial head prosthesis, operative procedure, and radiographic appearance. Types of radial head prostheses investigated included cemented and noncemented stems and long and short stems. In addition, we examined straight, curved, and tapered stems. We performed radiographic assessment immediately postoperatively and at 6 weeks, 3 months, 6 months, 12 months, and annually thereafter. We obtained anteroposterior (AP) and lateral plain film
JHS 䉬 Vol A, October
2120
RADIAL HEAD PROSTHESES STRESS SHIELDING
radiographs of the elbow in a consistent manner according to standard x-ray practices for elbow radiographs, with the patient seated on a stool and the upper extremity resting on a table. We took the AP radiograph with the shoulder flexed forward, the elbow extended as much as possible, and the forearm in supinated or neutral rotation, depending on the range of the patient’s forearm rotation. We took lateral radiographs with the forearm in neutral rotation. These positions allowed the 4 sides of the radial neck to be imaged with plain films, as shown in Figure 1. The beam was centered on the midpoint of the antecubital crease, perpendicular to the elbow. We obtained the lateral radiograph with the elbow flexed to 90° and the shoulder internally rotated while the arm and elbow were supported on the table. We documented the timing, location, and pattern of bone loss from stress shielding. The distal extent of bone loss on the periosteal and endosteal surfaces was measured from the collar of the implant. We digitized the radiographic images in Digital Imaging and Communications in Medicine format and measured them using QREADS Clinical Image Viewer (Mayo Clinic, Rochester, MN), a picture archiving and communication system. It was possible to infer a 3-dimensional image from the plain films by imagining that in each view different cortices of the proximal radius were visible. We measured 2 cortical bone lines in each image: periosteal bone loss, representing the outer cortex, and endosteal bone loss. We assessed periosteal loss by examining the existing periosteum and imagining a continuation of where that line would lie if it were intact. We measured endosteal length loss, representing the inner cortex loss, in the same manner (Fig. 1). From the data, we classified bone loss from stress shielding according to 2 variables: location and severity. We identified the cortices as lateral, medial, dorsal, and volar, according to the anatomic position.13 Overall severity of bone loss was qualified as follows: stage I: cortical bone thinning; stage IIa: partially exposed stem; stage IIb: circumferentially exposed stem; and stage III: mechanical or impending failure (ie, actual or impending loss of stem fixation or periprosthetic fracture). We defined cortical bone thinning as any visible loss of the width of the cortex but with some endosteal bone remaining around the stem. If the endosteal cortex was breached, the severity was classified as stage II (exposed stem). The labeling of a case as having mechanical or impend-
FIGURE 2: Cumulative graph displaying the sequential detection of stress shielding. We detected the first case of stress shielding at 1 month postoperatively. The latest detection, when all cases had presented, was 15 months. Copyright © 2012 Mayo Foundation. Used with permission.
ing failure would have been done retrospectively. For example, if an implant had failed as the result of stress shielding, the image taken before failure would have qualified as stage III. Statistical analysis We calculated the average lengths of exposed stem (in millimeters) and compared them by performing a nonparametric repeated-measures analysis using a Friedman test. We used a paired sign test to compare the amounts of exposed stem among the 4 cortices (ie, anterior, posterior, medial, and lateral). For both analyses, P ⱕ .05 was considered significant. RESULTS Of the 26 patients, 14 were male and 12 were female; 15 cases underwent primary arthroplasty and the other 11 were revisions. The average follow-up period was 33 months (range, 13–70 mo). Radiographic assessments Seventeen cases (63%) had evidence of stress shielding on plain films. We initially detected stress shielding an average of 7 months postoperatively and as early as 1 month postoperatively (Fig. 2). All cases of stress shielding had presented by 15 months, the progression of which is shown in Figure 3. Cortex involvement Of the 17 cases, there was lateral cortex involvement in 12 (71%), volar cortex involvement in 11 (65%), dorsal cortex involvement in 10 (59%), and medial cortex
JHS 䉬 Vol A, October
RADIAL HEAD PROSTHESES STRESS SHIELDING
2121
FIGURE 4: Average lengths and standard error bars of exposed radial head prosthetic stem by cortex. Cortices with different lowercase letters are significantly different from each other (P ⫽ .01). Copyright © 2012 Mayo Foundation. Used with permission.
involvement in 5 (29%). The amount of exposed stem varied depending on which cortex was being measured (P ⫽ .02). Paired sign test analysis revealed that the amount of exposed stem in the medial cortex was significantly less than in the lateral and volar cortex (P ⬍ .02), whereas the lateral and volar cortices were not significantly different from each other (P ⫽ .09). The length of exposed stem in the dorsal cortex was similar to that of the medial cortex (P ⫽ .18). Initiation of stress shielding In 7 patients, we observed stress shielding to affect 1 cortex in isolation before involving other cortices. In 5 patients (71%), the lateral cortex experienced bone loss first and in isolation. In 1 patient, the volar cortex was affected first and in isolation; the medial cortex was also affected first in 1 patient. For all patients, the average length of exposed stem in any cortex averaged 2 mm (range, 0 –5 mm) (Fig. 4).
FIGURE 3: Progression of stress shielding detected at 3, 6, 9, 12, and 15 months postoperatively. Copyright © 2012 Mayo Foundation. Used with permission.
Pattern and severity of stress shielding The shape pattern of stress shielding can be conveyed by measuring periosteal bone loss and endosteal bone loss. Periosteal bone loss was consistently more extensive than endosteal bone loss (Figs. 5, 6). For the 3 patients with circumferential bone resorption (stage IIb), the mean length of exposed stem averaged 2.3 mm (range, 0 – 4 mm) (Table 1). Stress shielding did not extend to the bicipital tuberosity. We noticed no common pattern of bone loss in any cases involving cemented stems. To assess for progression of bone resorp-
JHS 䉬 Vol A, October
2122
RADIAL HEAD PROSTHESES STRESS SHIELDING
FIGURE 5: Example of stress shielding beginning in the lateral cortex. A Immediately postoperatively, there was good contact between solid bone and the collar. B Stage I stress shielding was apparent 2 months postoperatively in the lateral cortex. C, D Stage IIb stress shielding 6 and 18 months postoperatively, respectively. Circumferential exposure of the proximal stem was present but was not progressive. Copyright © 2012 Mayo Foundation. Used with permission.
tion, we analyzed the 6 implants with the longest follow-up. Only 1 had evidence of progression (case 16), but the maximum length of exposed stem was 3 mm in any cortex. Stress shielding appeared to be nonprogressive in the other 2 (Table 2) and stabilized by a mean of 23 months (range, 13–27 mo).
Influence of revision Five cases that experienced stress shielding were the result of revision surgery. We statistically analyzed these cases separately. There were no significant differences in the amount of exposed stem in any cortices between the revision and primary cases (P ⫽ .11–.83).
Stress shielding by stem type We saw stress shielding with all 5 prosthetic designs used in this series. No single implant design (ie, by manufacturer) was more predisposed to developing stress shielding when all manufacturers were statistically compared (P ⫽ .08). We also saw stress shielding with cemented as well as cementless prostheses, bipolar and monopolar, regardless of stem length, shape, and surface texture. Some groups were too small to derive valid statistical conclusions, but the only statistically significant factor was the use of a long cemented stem, which had less stress shielding (P ⫽ .03). However, that stem required resection of 19 or 22 mm of head and neck to a level just proximal to the bicipital tuberosity. Therefore, we cannot directly compare it with implants with conventional stem lengths.
DISCUSSION This study demonstrated that stress shielding around well-fixed radial head prostheses is common; it occurred in 17 of 26 prostheses (65%). We noted it with all 5 stem designs studied, whether cemented or porous-coated, and regardless of stem shape, length, and surface texture. However, it was typically minor and nonprogressive, and it never extended to involve the bicipital tuberosity. In this series, no implants appeared to be at risk of failure as a result of radiographic bone loss from stress shielding. We confirmed solid fixation of the stems in a number of cases that required revision. We observed consistent patterns of bone loss. Resorption always began on the periosteal surface. Even when resorption progressed to involve the
JHS 䉬 Vol A, October
RADIAL HEAD PROSTHESES STRESS SHIELDING
FIGURE 6: Three-dimensional reconstructed computed tomography image demonstrating bone loss resulting from stress shielding (case 13). Copyright © 2012 Mayo Foundation. Used with permission.
endosteal surface, periosteal bone loss always exceeded endosteal bone loss. We consistently noted the radiographic appearance of a convex outer surface of the proximal radial neck. Consistent with Wolff’s law4 –7 of bone remodeling, load transfer across the radiocapitellar joint to the radial head prosthetic stem dissipates into the endosteal cortex. This biomechanical concept explains why the periosteal cortex resorbed before the endosteal cortex. This characteristic pattern of bone loss is valuable when distinguishing between stress shielding and bone loss that is caused by infection, loosening, or osteolysis. In our opinion, the latter 3 conditions can cause endosteal or concave erosions, which are not seen in stress shielding. We analyzed bone loss around the prostheses separately for each of the 4 quadrants represented on the AP and lateral radiographs. Cortical involvement was variable; the lateral cortex was most frequently involved, the medial cortex least frequently involved, and dorsal and volar cortical involvement fell in between. Whereas bone loss began in an asymmetrical fashion, it never in-
2123
volved 1 cortex excessively without starting to advance circumferentially until the proximal stem was exposed. We noticed this recurring pattern in 16 of the 17 cases (94%). Endosteal bone loss never exceeded 5 mm in any 1 cortex without involving at least 1 other cortex. In 3 of 17 cases (18%), bone resorption was sufficient to cause circumferential exposure of the proximal stem (stage IIb) for an average length of 2.8 mm. Of the 3 cases with a circumferentially exposed stem, the length of exposed stem on any surface (cortex) averaged 2.3 mm and never exceeded 4.0 mm. The amount of bone loss resulting from stress shielding seen in these cases is unlikely to be detrimental if revision is necessary. If revision was required, the only potential consequence would be that bone loss might require conversion to a longer stem. However, in the experience of the senior author (S.O.D.), revision typically requires conversion to a longer stem in the absence of stress shielding, because of endosteal bone loss and sclerosis that precludes secure fixation of another standard-size stem. The time course of stress shielding was variable, but it was first evident between 1 and 15 months postoperatively. For patients with sufficient interval x-rays before final follow-up, stress shielding did not progress after 26 months. For patients in whom we obtained follow-up beyond 26 months (11 of 19 patients), the average follow-up time was 40 months (range, 26 –70 mo). One patient (case 17) had evidence of progression, but the maximum length of exposed stem was 3 mm in any cortex. Although information is sparse regarding the effect of single cortex loss on radial head implant stability, we think that 3 mm bone loss from a 25-mm stem (12%) would not likely compromise implant fixation. Longer follow-up will be needed to determine whether bone loss from stress shielding can progress to a degree that would be of clinical concern. This study had some limitations, including the use of plain radiographs for measuring and categorizing bone loss. Although we standardized positioning of patient limbs for these radiographs, some variability may have affected apparent changes in bone loss. For example, there may have been some overlap between cortices depending on the orientation of the x-ray beam. However, we were certain that the lateral cortex, where we noted the most bone resorption, was always clearly visualized. Although the accuracy of these measurements might improve with the use of computed tomography imaging, the expense would not be warranted based on the conclusion of this radio-
JHS 䉬 Vol A, October
2124
TABLE 1.
Prosthesis Details, Follow-up Interval, Column Involvement, and Stage of 17 Cases of Stress Shielding Exposed Stem
Case
Prosthesis Type
Cement Status
Primary/Revision
Interval (mo)
Volar (mm)
Dorsal (mm)
Lateral (mm)
Medial (mm)
Mean Exposed Stem (mm)
Stage
1
rHead (Sbi)
Cemented
Primary
38
0
0
0
0
0.0
I
2
Evolve (Wright)
Cemented
Primary
70
0
0
0
0
0.0
I
RHS (Tornier)
Uncemented
Revision
14
2
1
0
1
1.0
IIa
Anatomic (Acumed)
Uncemented
Primary
32
0
5
5
0
3.0
IIa
JHS 䉬 Vol A, October
5
Anatomic (Acumed)
Uncemented
Revision
14
2
2
4
0
2.0
IIa
6
Anatomic (Acumed)
Uncemented
Primary
39
4
2
5
0
3.0
IIa
7
Anatomic (Acumed)
Uncemented
Primary
27
0
0
1
0
0.3
IIa
8
Anatomic (Acumed)
Uncemented
Revision
29
2
4
4
0
3.0
IIa
9 10
Anatomic (Acumed)
Uncemented
Primary
26
1
0
1
0
1.0
IIa
Judet (Tornier)
Cemented
Primary
61
0
3
1
0
1.0
IIa
11
Evolve (Wright)
Cemented
Primary
28
1
0
0
0
0.3
IIa
12
Anatomic (Acumed)
Uncemented
Revision
21
4
0
0
0
1.0
IIa
13
Anatomic (Acumed)
Uncemented
Primary
20
2
1
3
0
2.0
IIa
14
Anatomic (Acumed)
Uncemented
Primary
13
0
0
3
1
3.0
IIa
15
Anatomic (Acumed)
Uncemented
Primary
13
4
1
4
2
3.0
IIb
16
rHead (Sbi)
Uncemented
Primary
48
2
2
3
2
2.0
IIb
17
rHead (Sbi)
Cemented
Revision
61
1
3
2
1
2.0
IIb
13–70
0–4
0–5
0–5
0–2
0–3
33
2
1
2
0.4
2.0
Range Average
RADIAL HEAD PROSTHESES STRESS SHIELDING
3 4
RADIAL HEAD PROSTHESES STRESS SHIELDING
TABLE 2.
Stability Progression Staging at Last Follow-up
3.
Case
Final Follow-up
1
38
I
Stable after 23 mo
2
70
I
Stable after 26 mo
5.
4
39
IIA
Stable after 13 mo
6.
6
39
IIA
Stable after 27 mo
16
48
IIB
Active
17
61
IIB
Stable after 24 mo
Progression
4.
7. 8.
9.
graphic study. Another limitation was the small number of patients and relatively short-term follow-up. Longer-term follow-up with a larger patient population will be needed to elucidate the nature and potential clinical implications, if any, of stress shielding around radial head prostheses. REFERENCES
10.
11.
12.
1. Doornberg JN, Parisien R, van Duijn PJ, Ring D. Radial head arthroplasty with a modular metal spacer to treat acute traumatic elbow instability. J Bone Joint Surg 2007;89A:1075–1080. 2. Moro JK, Werier J, MacDermid JC, Patterson SD, King GJ. Arthro-
13.
2125
plasty with a metal radial head for unreconstructible fractures of the radial head. J Bone Joint Surg 2001;83A:1201–1211. Doornberg JN, Ring D. Coronoid fracture patterns. J Hand Surg 2006;31A:45–52. Engh CA Jr, Young AM, Engh CA Sr, Hopper RH Jr. Clinical consequences of stress shielding after porous-coated total hip arthroplasty. Clin Orthop Relat Res 2003:157–163. Huiskes R, Weinans H, Grootenboer HJ, Dalstra M, Fudala B, Slooff TJ. Adaptive bone-remodeling theory applied to prosthetic-design analysis. J Biomech 1987;20:1135–1150. Huiskes R, Weinans H, van Rietbergen B. The relationship between stress shielding and bone resorption around total hip stems and the effects of flexible materials. Clin Orthop Relat Res 1992:124 –134. Roesler H. The history of some fundamental concepts in bone biomechanics. J Biomech 1987;20:1025–1034. Harris WH. Will stress shielding limit the longevity of cemented femoral components of total hip replacement? Clin Orthop Relat Res 1992:120 –123. Kilgus DJ, Shimaoka EE, Tipton JS, Eberle RW. Dual-energy X-ray absorptiometry measurement of bone mineral density around porouscoated cementless femoral implants. Methods and preliminary results. J Bone Joint Surg 1993;75B:279 –287. McCarthy CK, Steinberg GG, Agren M, Leahey D, Wyman E, Baran DT. Quantifying bone loss from the proximal femur after total hip arthroplasty. J Bone Joint Surg 1991;73B:774 –778. Skoldenberg OG, Boden HS, Salemyr MO, Ahl TE, Adolphson PY. Periprosthetic proximal bone loss after uncemented hip arthroplasty is related to stem size: DXA measurements in 138 patients followed for 2–7 years. Acta Orthop 2006;77:386 –392. Popovic N, Lemaire R, Georis P, Gillet P. Midterm results with a bipolar radial head prosthesis: radiographic evidence of loosening at the bone-cement interface. J Bone Joint Surg 2007;89A:2469 –2476. Basmajian JV. Primary anatomy. 8th ed. Baltimore: Williams and Wilkins, 1985.
JHS 䉬 Vol A, October