Metal Ion Concentrations in Body Fluids after Implantation of Hip Replacements with Metal-on-Metal Bearing – Systematic Review of Clinical and Epidemiological Studies

Introduction The use of metal-on-metal (MoM) total hip arthroplasty (THA) increased in the last decades. A release of metal products (i.e. particles, ions, metallo-organic compounds) in these implants may cause local and/or systemic adverse reactions. Metal ion concentrations in body fluids are surrogate measures of metal exposure. Objective To systematically summarize and critically appraise published studies concerning metal ion concentrations after MoM THA. Methods Systematic review of clinical trials (RCTs) and epidemiological studies with assessment of metal ion levels (cobalt, chromium, titanium, nickel, molybdenum) in body fluids after implantation of metalliferous hip replacements. Systematic search in PubMed and Embase in January 2012 supplemented by hand search. Standardized abstraction of pre- and postoperative metal ion concentrations stratified by type of bearing (primary explanatory factor), patient characteristics as well as study quality characteristics (secondary explanatory factors). Results Overall, 104 studies (11 RCTs, 93 epidemiological studies) totaling 9.957 patients with measurement of metal ions in body fluids were identified and analyzed. Consistently, median metal ion concentrations were persistently elevated after implantation of MoM-bearings in all investigated mediums (whole blood, serum, plasma, erythrocytes, urine) irrespective of patient characteristics and study characteristics. In several studies very high serum cobalt concentrations above 50 µg/L were measured (detection limit typically 0.3 µg/L). Highest metal ion concentrations were observed after treatment with stemmed large-head MoM-implants and hip resurfacing arthroplasty. Discussion Due to the risk of local and systemic accumulation of metallic products after treatment with MoM-bearing, risk and benefits should be carefully balanced preoperatively. The authors support a proposed „time out“ for stemmed large-head MoM-THA and recommend a restricted indication for hip resurfacing arthroplasty. Patients with implanted MoM-bearing should receive regular and standardized monitoring of metal ion concentrations. Further research is indicated especially with regard to potential systemic reactions due to accumulation of metal products.


Introduction
Total hip arthroplasty (THA) for patients with osteoarthritis is one of the most successful surgical interventions in general inducing substantial improvement of health-related quality of life of affected patients [1]. Aseptic loosening is a typical long-term complication that significantly determines implant survival. Compared to regular bearings with conventional polyethylene, one advantage of metal-on-metal (MoM) bearings is that they produce less volumetric wear [2]. However, MoM hip replacements may release metallic products (i.e. particles, ions, metallo-organic compounds) due to wear and corrosion [3,4]. Metal ions from the corresponding alloying element (i.e. cobalt -Co, chromium -Cr, titanium -Ti, nickel -Ni, molybdenum -Mo) can be measured in the joint itself as well as in surrounding tissue and body fluids, and may potentially cause local and or/systemic adverse reactions [5,6,7,8]. Recently, metal-related local adverse reactions (i.e. adverse reactions to metal debris, ARMD) in patients with MoM hip replacement gained substantial scientific and public attention [9,10,11,12,13,14]. Endoprothesis registries from Australia and the UK [15,16,17] indicated increased revision rates following hip replacements with MoM-bearingespecially large-head THA and hip resurfacing arthroplasty (HRA). The risk of local adverse reactions of MoM-THA has been reported to correlate with the level of systemic metal ion concentrations [5,6,8,13,18,19,20]. However, several clinically relevant issues related to the safety of MoM-THA such as the impact of the different types of MoM implants on metal ion concentrations and related risks and the long-term course of postoperative metal ion concentrations require further investigation. Case reports [21,22] also suggest systemic risks due to metal burden after MoM-THA, but systematic research on this important issue is missing. There is substantial evidence from occupational medicine indicating that employees in metal producing and -processing plants exposured to chromium(VI) resp. cobalt compounds are at increased risk of nasal septum ulcerations, lung cancer resp. and cardiomyopathy [23,24,25]. It has to be noted, however, that toxicity of Cr(III) compounds is substantially lower than those of Cr(VI) compounds. Because the primary route of metal exposure in occupational medicine is the respiratory tract, generalizability of these findings to patients with metallic hip replacement remains unclear.
Despite the current uncertainty a systematic review on the safety of MoM-hip replacements is missing. We systematically appraised published clinical and epidemiologic studies to clarify the following issues related to the safety of MoM-hip replacement: -What are median and maximum metal ion concentrations

Methods
We undertook a systematic review to identify, summarize, and critically appraise the clinical and epidemiological evidence concerning the impact of metalliferous hip replacements on metal ion levels in body fluids. In addition to the type of bearing as the hypothesized determinant of metal ion concentrations we were particularly interested in patient characteristics as well as study quality characteristics as potential secondary determinants of metal ion concentrations after THR, and in the clinical consequences resulting from increased metal ion concentrations.

Inclusion criteria
All randomized controlled trials (RCTs) and epidemiological studies (cohort, case-control and cross-sectional studies, case series) with metal ion measurement (cobalt, chromium, titanium, nickel, molybdenum) in body fluids (full blood, serum, plasma, erythrocytes, synovia, urine) after implantation of metalliferous hip replacements in at least 20 patients were considered eligible. Studies were required to be published as an original article in English, German, or French language to be included.

Literature search
Systematic electronic literature searches were conducted in PubMed and EMBASE (until January 19,2012). Combinations of MeSH-terms were used to identify relevant trials with a high sensitivity. The exact search string used is provided in table 1. Systematic electronic search was supplemented by hand search in

Data abstraction
The following information was abstracted from the studies included using standardized and beta tested evidence tables: study characteristics (e.g. author, geographical region, study design, time points of assessment, follow-up period).
-Patient characteristics (e.g. number of patients included and followed up to each point of assessment of metal ion level, UCLA -University of California Los Angeles score [26], body mass index).
-Implant characteristics (type of bearing material and head size in three different groups of implants: small head (SH-)THA with head diameter #32 mm, stemmed large-head (LH-)THA with head diameter $36 mm, HRA).
-Implant position (inclination of acetabular cup in the frontal plane).
-Details on metal ion assessment, i.e. type of metal ions assessed (cobalt, chromium, titanium, nickel, molybdenum), the medium of assessment, method of analysis, metal ion levels (median, mean, interquartile range (IQR), outliers, definition of outliers.

Rating of methodological study quality
Standardized study quality assessment was based on the CASP und SIGN Checklists [27,28]. Based on consented and a priori defined study quality criteria the risk of bias was rated for each study with the following categories: very low risk of bias: ''++'' low risk of bias: ''+'' high risk of bias: ''2'' We considered the following criteria to increase the risk of bias: no consideration of confounding and/or explanatory factors such as other metallic implants, type, size, and position of implant.
missing information on the methods used to measure metal ion levels.

Evidence-synthesis
The qualitative evidence-synthesis included a comparison of the pre-vs. postoperative metal ion concentrations (median (alternatively: mean), IQR, maximum values) stratified by type of bearing as the primary explanatory factor. Patient characteristics (mean age, sex ratio) and implant characteristics (bearing size and position) were considered as secondary explanatory factors. Additionally, the definition of cut-off levels of metal ion concentrations in different studies was compared.
RCTs and epidemiological studies were analyzed separately. The detection limit of metal ions depends on the method and device used. The range of the detection limits reported and the handling of values below the detection rate was also part of the qualitative synthesis of the published evidence.
The course of metal ions in body fluids over time was assessed based on studies reporting baseline serum Co-values and at least 2 postoperative Co-measurements. In these studies, we investigated the course of median serum Co-concentration after implantation of hip replacements with different kinds of metal-on-metal bearings.
Metal ion concentrations in all tables and figures below generally relate to patients with unilateral THR, unless stated differently. We initially planned to conduct a quantitative summary (meta-analysis) of the results of qualitatively homogeneous studies with very low or low risk of bias.          Gathered information refer to patients with at least one metallic part of the bearing being investigated on metal ions. Studies were summarized due to multiple reporting of identical patient population.      Data on all relevant metal ions (cobalt, chromium, titanium, nickel, molybdenum) was extracted from the studies included. In the results section, we present cobalt levels as the proposed reference metal ion concentration after THR, as suggested by an international, multiprofessional expert panel [30]. Figure 1 summarizes the yield of systematic search and study selection [31]. Overall, 104 studies (11 RCTs, 14 cohort studies, 1 case-control study, 55 cross-sectional studies, 23 case series) totaling 9.957 patients with measurement of metal ions in body fluids were identified and analyzed. The majority of studies were performed in Europe (n = 71) and North America (n = 26). Table 1 summarizes the characteristics of the 11 RCTs included. RCTs were generally small and included between 13 and 117 patients. Two RCTs examined MoM vs. ceramicpolyethylene (CoP) [32,33], five studies MoM vs. metal-polyethylene (MoP) [34,35,36,37,38,39], two studies examined MoM vs. ceramic-ceramic (CoC) [26,40,41], and two MoM vs. hip resurfacing arthroplasty (HRA) [42,43,44]. LH-THA was investigated in one US-American study [35]. All other RCTs investigated MoM SH-THA.

Determination of metal ions in RCTs
As detailed in table 4, median Co-concentrations were elevated at each postoperative assessment after implantation of MoM SH-THA, MoM LH-THA, and HRA compared to THA with only one or without metallic part of the bearing. Following MoM SH-THA median serum Co-concentrations varied between 0.66-1.0 mg/L at 6 months [32,33,34,35] and 0.73-1.2 mg/L at 2 years [33,34,35,41] after surgery. After MoM LH-THA median serum Co-levels of 0.66 mg/l and 0.73 mg/L were reported at 6-month and 2-year follow-up, respectively [35]. After HRA, median whole blood Co-concentrations varied between 0.78 and 1.3 mg/L up six months und between 0.16 and 1.2 mg/L two years postoperatively [42,43,44]. The only direct comparison between MoM SH-and LH-THA revealed no relevant differences in the median serum Co-concentration [35].
Two RCTs directly compared MoM SH-THA vs. HRA and revealed qualitatively different results: While Smolders et al. [42] observed higher median and maximum Co-concentrations following HRA compared to MoM SH-THA, Vendittoli et al. [43,44] did not observe relevant differences in mean Coconcentrations following HRA and MoM SH-THA. THA with only one metallic part of the bearing resulted in not or only slightly elevated serum Co-concentrations (Table 4).  Table 5 summarizes aggregated characteristics of the epidemiological studies included. Table 6 provides details of the characteristics of each epidemiological study included. Metal ion concentrations following HRA were reported in 48 studies (52%). Information on inclination and anteversion were reported in 30 studies (32%) and in 16 (17%) studies, respectively. Information on mean age was reported in 66 studies (71%). The sex distribution of patients was reported in 64 studies (69%) and varied substantially with 19 to 90% of the study populations being female.
The vast majority of studies (n = 86; 83%) had significant methodological shortcomings such as a lack of reference group, lack of preoperative baseline assessment, lack of essential information on implant characteristics or metal ion measurement, and/or insufficient follow-up rates and were therefore considered as having a high risk of bias.

Determination of metal ions in epidemiological studies
87 and 85 epidemiological studies reported Co and Cr values following hip replacement, respectively. No study differentiated between Cr(III) and Cr(VI). Metal ion concentration was most often measured in whole blood (n = 51), serum (n = 47), and urine (n = 19). Few studies investigated ion levels in erythrocytes (n = 2), plasma (n = 5), or in synovia (n = 3). Inductively coupled plasma mass spectrometry (ICP-MS) was used in 56 studies (60%), atomic absorption spectrometry (AAS) in 33 studies (36%), and inductively coupled plasma atomic emission spectrometry (ICP-AES) resp. inductively coupled plasma optical emission spectrometry (ICP-OES) in three studies for metal ion detection. Only 24studies (26%) reported preoperative (baseline) metal ion concentrations. For more details on metal ion assessment in the epidemiological studies included please refer to table 7. Table 8 summarizes median serum Co-concentrations, 75th percentiles, and maximum values of Co-concentrations before and after hip replacement stratified by the type of intervention (MoM SH-THA, MoM LH-THA, HRA). After MoM SH-THA median Co-concentrations varied between 0.65 and 1.5 mg/L at six months and between 0.7 and 1.7 mg/L two years postoperatively [42,45,46]. After HRA, median serum Co-concentrations varied between 1.12 and 3.7 mg/L six months and between 0.54 and 4.28 mg/L two years postoperatively indicating higher Co-levels after HRA vs. MoM SH-THA [45,46,47,48,49,50]. Median Coconcentrations after MoM LH-THA varied between 0.7 and 3.26 mg/L six months and between 3.77 and 5.38 mg/L two years postoperatively [48,49,51].
Tables 9 and 10 summarize the results of comparative epidemiological studies. In accordance with these indirect comparisons from epidemiological studies, median Co-concentrations following MoM LH-THA and HRA tended to be higher compared to MoM SH-THA. Consistently, MoM LH-THA, MoM SH-THA, and HRA resulted in higher metal ion concentrations than THA with CoP, CoC, and MoP-implants.
Conclusions on the role of patient characteristics (age, sex) on metal ion concentration could not be drawn due to a lack of Table 10. Cont. Course of metal ion concentration pre-vs. postoperative Due to substantial differences in the design, interventions, methods of metal ion assessment, study populations and study reporting, we considered statistical meta-analysis not to be indicated.

Systemic clinical reactions
Five studies [61,62,70,71,72] examined possible associations between metal ion concentrations and nephrotoxicity. Daniel et al. [71] examined renal clearance and renal concentrating efficiency of cobalt. The renal efficiency, i.e. the ratio of urine cobalt concentration to plasma cobalt concentration, was 0.9 (IQR 0.7-1.6) for preoperative controls and 3.2 (IQR 1.7-5.1) in patients with MoM THA or HRA. No threshold was endorsed at which renal capacity is overextended. Corradi et al. [70] examined metal ion concentrations in whole blood and renal markers in patients with HRA and in healthy controls. The median Co-excretion in patients with HRA was 12.9 mg/24-h urine (range 6.1-71.5 mg). No elevated renal markers were found in comparison with controls. Gruebl et al. and Marker et al. [61,62] investigated serum metal ions, blood urea nitrogen, and serum creatinine in overlapping cohorts of patients with MoM THA. The median (range) serum creatinine value preoperatively and at 10 years follow-up was 0.88 mg/dL (0.63-1.21 mg/dL) and 0.86 mg/dL (0.55-1.51 mg/dL), respectively. Evidence for or against further systemic toxicity or carcinogenicity could not be revealed from the studies included.

Discussion
As highlighted in this comprehensive systematic review, there is substantial and consistent evidence that patients receiving hip replacement with a MoM-bearing are at increased risk for systemic accumulation of metallic products. In the 104 studies analyzed, median metal ion concentrations were persistently elevated after implantation of MoM-bearings in all investigated mediums (whole blood, serum, plasma, erythrocytes, urine), irrespective of patient characteristics and study characteristics.
Overall, metal ion concentrations in body fluids were assessed in 9.957 patients in the 11 RCTs and 93 epidemiological studies  Table 11. Maximal postoperative Co-concentrations after THA in RCTs and epidemiological studies.   included in this review. Despite heterogeneity in the study designs, techniques and medium of metal ion assessment, investigators consistently observed elevated median/mean metal ion concentrations after MoM THA and HRA compared to baseline, i.e. before surgery. Metal-free hip replacements did not result in increased metal ion levels. Metal ion concentrations following MoP and MoC THA were much lower compared to MoM THA or HRA. One important finding from this review is that stemmed largehead MoM-implants and HRA tended to result in higher Coconcentrations than small-head MoM-implants. In several studies very high serum cobalt concentrations above 50 mg/L were measured in patients who had received large-head MoM-implants or HRA.
These findings have significant clinical relevance, as increased metal ion concentrations translate into increased risk for the development of local adverse reactions such as ARMD. In many cases ARMD results in the indication for the revision of MoMimplants. [13] One current issue of debate is the definition of a cutoff cobalt level, above which revision should be considered.
Hart et al. [8,18,30] recommend a serum cobalt threshold level of 4.97 mg/L based on ROC-curve analyses. However, no explicit advice is given on how to treat patients above this value. Recommendations of present literature currently state Co-  Figure 3. Summarizes six studies which reported Co-concentrations in well and poorly functioning implants [55,18,19,20,53]. Cases with local metal-related adverse reactions (poorly functioning implants) had consistently higher metal ion concentrations than patients with wellfunctioning THA. doi:10.1371/journal.pone.0070359.g003 concentrations in serum or plasma greater 2 up to 7 mg/L as a predictor for a subjectively adverse outcome and an increased risk of MoM-specific complications [8,18,30,73]. While MoM SH-THA (head diameter #32 mm) seem to show similar long-term survival rates as hip replacement with other bearings [61,74,75,76,77], the implantation of stemmed LH-THA (head diameter $36 mm) is associated with significantly higher short-term revision rates in clinical studies as well as arthroplasty registries [15,16,17,44,45,48,68,78,79]. The elevated release of metal products in these stemmed LH-implants may be due to fretting corrosion at the head-taper-junction in addition to a metal particle release from bearing surfaces.
Beside local tissue damages, it is important to gain better understanding about the potential systemic adverse effects induced by metal ion accumulation, i.e. toxicity, carcinogenicity, teratogenicity. The degree to which increased metal ion concentrations after MoM THA translate into increased risk for systemic toxicity cannot be sufficiently answered based on the studies identified and analyzed in this review. Until now, epidemiological studies have not revealed clinically relevant toxic damages of the kidney, heart or nervous system after MoM THA [45,[58][59][60]. Case reports, however, indicate the possibility of metal-induced cardiomyopathy [21,22]. An elevated risk of incident cancer following hip replacements with MoM bearing could not be identified yet [80,81,82], but studies may have been underpowered.
There is substantial evidence that occupational metal exposure is related to increased cancer risk. It has to be noted, however, that bioavailability of Cr(III) compounds is substantially lower than those of Cr(VI) compounds. Cr(VI) compounds are able to infiltrate into cells due to transmembrane motion and to operate genotoxic following reduction to Cr(III). Carbid metal workers exposed to Co are at increased risk for fatal lung cancer [83]; the ''International Agency for Research on Cancer (IARC)'' classified Co to be possibly carcinogenic. Persons being occupationally exposed to Co have higher Co urine concentrations when compared to the general population. It should be noted that specific attentiveness was laid on possibly elevated Cr(VI) in body fluids due to ascertained carcinogenicity of Cr(VI) compounds. Due to considerable differences in exposure routes, the effects of increased metal ion concentrations as a consequence of MoM hip replacement cannot be directly compared with the systemic effects of occupationally acquired (mainly inhaled) metals.

Strengths and weaknesses of this review
This systematic review was conducted in accordance with the PRISMA checklist [31]. Systematic literature search and assessment of eligibility of studies identified was done independently by two reviewers. Study quality assessment was based on a priori defined criteria. Due to methodological limitations in most of the studies included and due to substantial qualitative differences in the study design, conduct, and reporting, quantitative metaanalysis was not indicated. However, as highlighted above, the qualitative results are consistent despite the heterogeneity of the studies included so that we consider the conclusions drawn to be robust and generalizable.

Implications for clinical practice and future research
After hip replacement with contemporary MoM bearings the release of metal ions is highest in stemmed implants with large heads followed by resurfacing devices and also -but on a lower level -small heads. As the deposition of metal products may not only lead to local but possibly also systemic adverse health outcomes, the conclusions of this review have high relevance not only for orthopaedic surgeons, but also for other medical disciplines.
Due to the risk of systemic accumulation of metal ions following implantation of hip replacements with MoM bearing, consideration on risks and benefits should be done carefully and individually for every patient prior to surgery.
The authors support a ''time out'' of stemmed large-head MoM-THA and recommend a restricted indication for hip resurfacing arthroplasty to patients without risk factors such as small implant size, female gender, and renal insufficiency [5,79]. Patients with status post implantation of MoM should be followed by standardized monitoring. Especially examined ions, medium and analysis technique should be standardized to allow comparability of results and further analysis. Close interdisciplinary cooperation is necessary in case of potential systemic reactions due to increased metal ion concentrations.
An approach to this unresolved difficulty was one of the main objectives of an international and interdisciplinary expert conference, which took place in our institution. In April 2012, we hosted an international multi-disciplinary expert conference endorsed by the ''European Federation of National Associations of Orthopaedics and Traumatology'' (EFORT), ''European Hip Society'' (EHS), and the ''German Osteoarthritis Society'' in order to provide clinically-relevant advice on how to treat and monitor current and future patients with MoM THR. Beside orthopaedic surgeons being experienced with MoM hip endoprosthetics, epidemiologists, toxicologists, biomechanics, and pathologists as well as a patients representative and regulatory agency representative from 7 European countries and the US participated.
The statement resulting from this consensus initiative is published in detail on web sites of European [30] and German [84,85] orthopaedic societies [86]. Beside detailed recommendations on monitoring of MoM hip replacements and metal ion measurement the statement also summarizes prioritized questions for future research. One research issue that needs to be prioritized is the investigation of potential systemic risks due to accumulation of metal ions.