A Systematic Review of Systemic Cobaltism After Wear or Corrosion of Chrome-Cobalt Hip Implants.

OBJECTIVES: We sought to synthesize data on systemic arthroprosthetic cobaltism, a recently described syndrome that results from wear or corrosion of chrome-cobalt hip components.
METHODS: We conducted a systematic literature review to identify all reported cases of systemic arthroprosthetic cobaltism. To assess the epidemiologic link between blood cobalt levels (B[Co]), we developed a symptom scoring tool that evaluated 9 different symptom categories and a category of medical utilization.
RESULTS: We identified 25 patients reported between 2001 and 2014 with a substantial increase in case reports over the past 3 years. Symptoms were diverse and involved the hip (84%), cardiovascular system (60%), audiovestibular system (52%), peripheral motor-sensory system (48%), thyroid (48%), psychological functioning (32%), visual system (32%), and the hematological, oncological, or immune system (20%). The mean latency from implantation to presentation or revision was 41 months (range, 9-99 months). The mean B[Co] was 324 μg/L and 4 patients had levels less than 20 μg/L. The B[Co] but not blood chromium level was highly associated with a quantitative measure of overall symptom severity (r, 0.81; P < 0.001). Mean B[Co] and symptom scores were substantially higher in patients with revisions of failed ceramic-on-ceramic prostheses than those with primary metal-on-metal prostheses.
CONCLUSIONS: Systemic arthroprosthetic cobaltism is an increasingly recognized complication of wear or corrosion of chrome-cobalt hip implants, may involve a large number of organ systems, and may occur with relatively low B[Co]. There is an urgent need to better define the overall scope of the problem and to develop screening and management strategies.

The ill effects of ingestion or inhalation of cobalt on the peripheral and central nervous system, heart, and thyroid were extensively catalogued in 1981.[1] Historically, iatrogenic cobaltism resulted from oral cobalt chloride treatment of anemia, vocational cobaltism from exposure to cobalt power, and “beer drinkers' ” cardiomyopathy from a cobalt foam–stabilizing additive. This formerly esoteric malady is important because millions of hips implanted over the past decade use chrome-cobalt components. It was thought that metal-on-metal articular surfaces would be protected from contact and wear by fluid film lubrication.[2] Analysis of explanted metal-on-metal hips suggests that this protective mechanism rarely occurs in vivo.[3] Rather, metal-on-metal hips can wear or corrode, resulting in periprosthetic metallosis and hypercobaltemia (blood cobalt level, B[Co], >1 μg/L). Metal-on-metal hips comprising a third of American hips implanted over the past decade use chrome-cobalt femoral and acetabular articular surfaces and may have a particularly high risk for generating chrome-cobalt metallosis.[4,5] A recent survey of 498 patients implanted with metal-on-metal hips found that one-third had B[Co] of more than 4 μg/L, with periprosthetic complications that increased as B[Co] increased.[6]

Subjects without abnormal cobalt exposure have a mean B[Co] of 0.3 μg/L, and 95% have a value of less than 0.6 μg/L.[7,8] The biologic exposure threshold for B[Co] is 1 μg/L, and an end-of-work-week value of 1 or greater indicates potentially unsafe vocational exposure.[9] Patients with well-functioning metal-on-metal hip resurfacings have a mean B[Co] of 2 μg/L, and a level greater than 3 μg/L in a patient with a unilateral metal-on-metal hip indicates a degree of periprosthetic chrome-cobalt metallosis that is likely to incite a periprosthetic adverse reaction to metallic debris.[10,11]

Neurocobaltism and cardiac cobaltism occur when cobalt impairs mitochondrial metabolism resulting in cellular dysfunction or death.[1,12,13] Studies in rabbits have found that the neurons in the cochlea and retina have the greatest susceptibility to oxidative stress and the greatest likelihood of dying after experimentally induced cobaltism.[13] Different mechanisms may explain the effects of cobalt or chromium on mood and cognition. Acute cobalt exposure increases neural activity and can result in seizures.[1,14] Because cobalt may deposit preferentially in the myocardium and pericardial fluid, cobaltism may have predominant or only cardiac symptoms.[15-18] Because hip replacement often occurs among elderly patients, age-associated decreases in renal cobalt clearance may contribute to increased cobalt levels in some patients.[19]

The published peer-reviewed literature contains an increasing number of case reports of systemic cobaltism after hip arthroplasty. We present a review of all cases resulting from metal-on-metal primary arthroplasties or revised arthroplasties with chrome-cobalt femoral heads to define symptom presentation, identify basic characteristics of affected patients, and determine if symptom severity correlates with cobalt level or measures of cobalt exposure.

METHODS

Search Strategy

We conducted a systematic search of PubMed for manuscripts published at any point in the past through June 2014 with no language or geographic restrictions and no specified start date using the following terms: (“cobalt” OR “cobaltemia” OR “cobaltic” OR “cobaltism”) AND (“hip” OR “acetabulum” OR “total hip arthroplasty” OR “case report” OR “cohort”).

This search yielded 2318 manuscripts whose titles and abstracts were screened for potential inclusion (Fig. 1). We also screened references of retrieved full-text manuscripts. As a last informal check on our search strategy, we conducted a nonsystematic review of EBSCOhost and Google Scholar using the terms “hip replacement” OR “hip arthroplasty” OR “total hip arthroplasty” OR “cobaltism” OR “cobalt” OR “case report” OR “arthroprosthetic cobaltism.” The goal of this search was to determine if any manuscripts were identified that obviously should be included. Neither of these 2 additional search strategies yielded additional manuscripts.

FIGURE 1

Flow chart for a systematic review of systemic cobaltism after wear or corrosion of chrome-cobalt hip implants.

Inclusion criteria included human studies with reported systemic illness, a surgical history of hip arthroplasty with at least one chrome-cobalt component at the bearing surface, and a reported blood or serum cobalt level greater than 7 μg/L. The selected blood cobalt level cutoff was that used by the United Kingdom Medicines and Healthcare Products Regulatory Agency as indicative of a bearing whose wear may induce periprosthetic tissue damage.[20] In addition, we only included cases with at least 2 of 3 details of the patient's clinical course: latency to illness, latency to revision, or latency to hip symptoms. Patients who had isolated periprosthetic pseudotumors, neuropathies associated with local inflammation or masses (e.g., femoral or sciatic neuropathy from pseudotumor), and elevated C-reactive protein or erythrocyte sedimentation rate in isolation were not considered to have systemic illness, and their case reports were not included.

Exclusion criteria included occupational exposure to cobalt (e.g., ingestion, inhalation, or transdermal delivery), exposure to therapeutic doses of radioactive cobalt, and multipatient studies for which measured cobalt levels could not be associated with a specific patient's clinical symptoms. In addition, cases were excluded if the patient had had a diagnosis of a clear underlying disorder that explained all of their systemic symptoms (e.g., Hashimoto disease in a case of thyroid dysregulation). Although we did not limit our search to English language manuscripts, we did not include one manuscript for which we could not obtain a suitable translation.[21] We also excluded one report of periprosthetic cobaltism during pregnancy to avoid confounding by the numerous symptoms that can complicate pregnancy.[22]

We independently reviewed all 60 potentially relevant manuscripts. Where disagreement on inclusion occurred, the reviewers met and developed a consensus opinion on inclusion. This process led to inclusion of 18 manuscripts. Two manuscripts that were initially excluded[23,24] were included after the authors were contacted, and they provided previously missing chronological details. We included 2 additional manuscripts known previously to one of the authors,[15,25] which may not have been identified in our search because of their publication in Dutch and Italian, respectively. Although the manuscript by Megaterio et al[15] did not include blood cobalt levels, we elected to include it because it was the first case report of arthroprosthetic cobaltism and the patient reportedly had significant increases in blood and urine cobalt levels. In total, we included 22 manuscripts,[14-16,23-41] one of which reported 3 patients,[37] three of which reported 2 patients,[32,38,39] and 2 occasions where descriptions of the same patient were reported twice.[29,34,40,41] This led to results for 25 unique patients.

Data Abstraction and Synthesis

We independently abstracted data using a standard data abstraction form for all 22 relevant manuscripts (Fig. 1). In the 2 instances where a case was described in more than one publication, we used all sources to create the most complete clinical history. One of the current study authors authored 2 case reports included in the current review[39,40]; clinical histories of patients described in these reports were supplemented by additional data not published in the original manuscripts.

All 25 cases were scored using a cobaltism scoring rubric developed by the authors (Table 1). This rubric was not designed for clinical management purposes but rather for the epidemiological purpose of determining if blood cobalt concentrations correlated with clinical severity. Within organ systems, medical specialists reviewed the scoring tool (for example, a cardiologist for the cardiovascular symptoms, a psychiatrist for psychiatric symptoms, etc). We independently scored each patient. Where discrepancies occurred, these were resolved through consensus. To arrive at a final summary cobaltism score, we added together the scores for each symptom category.

TABLE 1

Scoring Tool Developed to Determine Symptom Severity of Arthroprosthetic Cobaltism

Elevations of blood cobalt are likely not constant throughout the period of hip implantation. Poorly positioned metal-on-metal hips are prone to edge loading resulting in accelerating rates of wear as articular incongruity increases.[3] Because of this issue, ideally, cobalt exposure would be based on frequent monitoring of B[Co] for the duration of the prosthetic placement. Practically speaking, most patients had a single prediagnosis or peridiagnosis cobalt level available from the time of diagnosis. Consequently, we assessed 3 measures of exposure: highest B[Co], latency from prosthetic placement to revision, and the product of these 2 variables. As a control analysis, we also assessed blood chromium levels as an exposure variable.

In addition to descriptive analyses, we conducted linear regression analysis to determine the relationship between exposure measures and clinical outcomes. Evaluated clinical outcomes included overall symptom score and scores within symptom categories. An adjusted model included patients' age and sex. All analyses were conducted with SPSS (IBM, SPSS Statistics, version 19, 2010).

RESULTS

We identified 25 patients with reported systemic arthroprosthetic cobaltism after wear or corrosion of chrome-cobalt hip implants. Sixteen patients (64%) were men, and mean age was 56 years (median, 54 years; range, 39–75 years). Seven of the patients resided in the United States (5 from Alaska), 5 in Australia, 4 in Germany, 3 each in Italy and The Netherlands, and 1 each in Canada, the Czech Republic, and Japan. The number of reported cases increased over time, with 1 case reported during 2001, 1 during 2006, 2 cases during 2009, 3 during 2010, 2 during 2011, 6 each during 2012 and 2013, and 4 through the first half of 2014.

Hip symptoms were identified in 21 patients (84%). Because we only included cases with systemic manifestations, all patients also had symptoms from one of the categories in Table 1 other than hip and medical utilization, most commonly constitutional and cardiovascular symptoms (Table 2). The mean number of systemic symptom categories involved was 3.6 (median, 4; range, 1–7).

TABLE 2

Frequency of Symptoms Among 25 Patients With Reported Systemic Arthroprosthetic Cobaltism

All but one included patient (18) had at least one recorded B[Co] with the mean peak level equal to 324 μg/L (range, 10–1085 μg/L). Of the 24 patients with available peak B[Co], 4 (17%) had a level of less than 20 μg/L and 11 (46%) had a level of less than 100 μg/L. Of 21 persons with blood chromium levels (B[Cr]) available, the mean was 57 μg/L (range, 4.1–249 μg/L). Five patients (24%) had a B[Cr] level of less than 20 μg/L, and 20 (91%) had a B[Cr] level of less than 100 μg/L. Twenty-two patients had information on the latency from the original insertion of a chrome-cobalt component to revision (if this was documented) or presentation (if revision latency was not documented), with a mean of 41 months (range, 9–99 months). One patient had a latency of less than 1 year, 6 (27%) from 1 to less than 2 years, 2 (9%) from 2 to less than 3 years, 5 (23%) from 3 to less than 4 years, and 8 (36%) for 4 years or greater.

Twenty patients (80%) had documentation of hip explantation. Reporting of postexplantation symptoms was nonsystematic and incomplete, making analysis impossible. However, for 19 of the patients, the authors noted that at least some of the symptoms improved; the 20th patient died of cobalt cardiomyopathy 2 months after hip resection,[29] the only death reported among the cases we identified. Despite improvement, for all patients where this was noted, at least some symptoms attributable to cobaltism persisted after explantation of the cobalt-containing prosthesis.

During linear regression analysis, the peak B[Co] was associated with the total symptom score on unadjusted and adjusted analyses (Fig. 2 and Table 3). The symptom category most strongly associated with B[Co] was thyroid abnormality. Latency from implantation to explantation or presentation was less strongly associated with total symptom score than peak B[Co]. However, for the individual categories of hip, audiovestibular, and peripheral motor sensory, latency predicted severity better than B[Co]. B[Cr] was not associated with total symptom score and infrequently associated with individual symptom categories.

FIGURE 2

Scatter plot of the association between blood cobalt level and a summary clinical symptom score in patients with systemic cobaltism after wear or corrosion of chrome-cobalt hip implants.

TABLE 3

Unadjusted and Adjusted R2 Values for the Association Between Exposures and Clinical Severity Scores Overall and Within Organ Systems

Sixteen of the 25 patients had a metal-on-metal primary arthroplasty, whereas 9 had a chrome-cobalt revision femoral head after fracture of a primary ceramic-on-ceramic arthroplasty. The mean B[Co] among the 9 patients with revision surgery was 613 μg/L (range, 398–1085 μg/L) compared with 104 μg/L (range, 10–398 μg/L) among the remaining patients (P < 0.001; Student 2-tailed t test). Similarly, the mean symptom score was 35 (range, 22–53) among those with revision surgery and 15 (range, 3–40) among the remaining patients (P < 0.001); those with revision surgery had a symptom score significantly higher at the 95% confidence level than other patients for hip symptoms (6.9 versus 2.6; P = 0.033), audiovestibular involvement (3.3 versus 1.0; P = 0.039), peripheral motor-sensory involvement (4.8 versus 0.69; P = 0.030), and thyroid disease (2.0 versus 0.63; P = 0.003).

DISCUSSION

Our study identified 25 cases of reported arthroprosthetic cobaltism that occurred since the initial report during 2001. Case reports have accelerated recently and have occurred from 8 countries. Despite the relatively widespread distribution, all but 2 US cases have originated from Alaska, suggesting the true scope of the problem has gone largely unreported. This likely has resulted from a combination of several factors including the slow onset, the protean symptoms that may mimic many different conditions, lack of uniform screening strategies for at-risk patients, and lack of awareness of arthroprosthetic cobaltism among surgeons and primary care physicians.

An additional difficulty is that hip replacement surgery often occurs in elderly patients or patients with underlying conditions, making it difficult to associate symptoms with cobaltism. In part to address this issue in our review, we developed a symptom scoring tool, which confirmed a strong association between B[Co] but not B[Cr] and symptom severity. This association differed by individual symptom category, with the strongest association between B[Co] and thyroid abnormalities and the weakest with optic abnormalities. This finding may indicate that not all symptoms in the case reports actually resulted from arthroprosthetic cobaltism. It may also indicate that presentation in any particular individual is modified by a variety of factors including underlying illness, age, activity level, and type of physical activities in which the patient engages.

Chrome-cobalt femoral head placement after revision of a fractured ceramic head led to higher B[Co] levels and more severe symptoms than primary metal-on-metal arthroplasty. The mechanism is likely to be due to abrasion of the metal surface by retained ceramic fragments. This was suggested in one case history in which the pieces of the fractured ceramic head were reconstructed and found to be incomplete.[14] Whereas use of a chrome-cobalt head for revision of a failed ceramic-on-ceramic hip likely identifies a particularly high-risk group, the substantial symptoms among those with metal-on-metal primary arthroplasty indicates screening and management strategies need to be developed for both groups.

The American Academy of Orthopedic Surgeons upgraded monitoring guidelines for patients with metal-on-metal hips in 2014.[10] These recommendations focus primarily on local adverse reactions to metallic debris and advise B[Co] levels and cross-sectional hip imaging for patients presenting with pain, noise, or swelling at the hip. Going further is Britain's national Medicines and Healthcare Products Regulatory Agency, which in 2012 issued a medical device alert (MDA).[20] This medical device alert recommends B[Co] determination for most patients implanted with metal-on-metal hips and notes that a rising level or abnormal hip imaging should prompt consideration for revision surgery even in asymptomatic patients. Perhaps most comprehensively, Australia's independent, not-for-profit group NPS MedicineWise recommends annual B[Co] determination for all patients implanted with metal-on-metal hips (increasing to every 3 months if levels rise), surveillance for periprosthetic complications by a surgeon, and evaluation for constitutional, psychological, neurologic, and cardiovascular manifestations of cobaltism by the patient's primary provider.[42] This approach could potentially avert the substantial systemic morbidity documented in the case reports presented here as well as lead to earlier identification of the pseudotumors that can result in progressive periprosthetic tissue damage and complicate revision surgery.[43,44] In this respect, it is worth noting that 4 of the 25 patients did not have hip symptoms; and thus, using this as a sole screening criterion for further evaluation may delay identification of systemic manifestations.

Complicating matters is that the B[Co] level at which risk for adverse outcomes increases is poorly defined. Whereas the mean B[Co] level among reported cases was relatively high (324 μg/L), 17% had a level of less than 20 μg/L and almost half had a level of less than 100 μg/L. These data suggest that at a minimum, B[Co] level greater than 10 μg/L should trigger additional evaluation, but the scope of this evaluation remains undefined. For example, evaluation could include hip radiographs to assess component fixation and position and to assess for bone loss; cross-sectional hip imaging to assess the condition of periprosthetic soft tissue[45,46]; and audiograms, echocardiograms, and thyroid function studies. Additionally, patients with metal-on-metal hips might benefit from baseline assessment of various organ systems. The use of these approaches will depend on the prevalence of cobaltism in the total population of persons with metal-on-metal hips, their average symptom severity and the proportion with severe symptoms, as well as the economic benefit of routine screening at varying levels of intensity versus waiting until patients become symptomatic. Unfortunately, none of this information is known.

CONCLUSIONS

Future work should be done to define the prevalence and burden of arthroprosthetic cobaltism and the cost-effectiveness of various screening and management strategies. In addition to targeted and systematic studies, another strategy to facilitate these goals is the establishment of joint registries in the United States and other countries similar to those used in many European countries. In the meantime, surgeons and general practitioners should receive training on the potential risk for cobaltism associated with wear or corrosion of chrome-cobalt hip implants. A validated symptom scoring tool would facilitate clinical screening. Finally, it seems prudent to assess B[Co] yearly in patients with chrome-cobalt–containing hip prostheses and to develop a management plan when levels greater than 10 μg/L are identified.