Correlation of Blood Metal Concentrations with Cognitive Scores and Neuroimaging Findings in Patients with Total Joint Arthroplasty

INTRODUCTION

Total joint arthroplasty (TJA) is a common surgical procedure. In the United States, more than seven million individuals have undergone a TJA procedure and are currently living with joint implants [1]. These implants are composed of metals (i.e., titanium, cobalt, and chromium), ceramics and/or polyethylene. Over time, friction, force, and repetitive pressure on the implants lead to wear and tear as well as corrosion (i.e., tribocorrosion) of the metal and plastic components. As a result, organo-metallic particles, metal ions, metal, and plastic debris accumulate locally and are released into the systemic circulation [2].

Previous studies suggest that the debris released from metal implants may possess neurotoxic properties [3, 4], which could be relevant to Alzheimer’s disease and related dementias etiology. Moreover, TJA patients frequently present with high metal concentrations both locally and systemically, but the current treatment approaches only focus on treating the adverse local tissue reactions at the implant site without taking into consideration how other parts of the body are affected. In case reports and small case series of TJA patients, neuropsychiatric symptoms, poor concentration, memory deficits, impaired attention/executive function, sensorineural hearing loss, and numbness were reported as well as magnetic resonance imaging (MRI) findings of subtle structural changes in the visual pathways and the basal ganglia [5–13]. In a recent longitudinal study by our group [14], we did not observe a significant difference in the rate of cognitive decline in persons with and without TJA until after 80 years of age, but a slightly faster cognitive decline in older knee arthroplasty patients. Yet the difference was small and of unclear clinical significance.

None of the previous studies systematically examined the association between systemic metal concentrations and cognitive and structural neuroimaging findings. The aim of the current study was to examine the cross-sectional association between blood metal concentrations with cognitive scores and MRI findings of the brain in patients with TJA. We hypothesized that increased systemic metal concentrations would be associated with lower cognitive scores, reduced cortical thickness and hippocampal volume.

METHODS

The study comprised 113 patients with hip and/or knee TJA, presenting to the orthopedic outpatient clinic, and a history of having elevated blood metal concentrations of either titanium, cobalt, and/or chromium, as flagged in the electronic medical records (concentrations higher than 1 ng/mL for cobalt and chromium, and higher than 2 ng/mL for titanium). Study protocols were approved by the Mayo Clinic Institutional Review Board and participants provided written informed consent before participation.

A one-time blood draw of 60 mL was used to measure systemic metal concentrations of titanium, cobalt, and chromium. Metal concentrations were assayed using inductively-coupled plasma mass spectrometry (ICP-MS) within an International Organization for Standardization Class-7 cleanroom in the Metals Laboratory [15, 16]. Chromium and cobalt in EDTA whole blood was analyzed by ICP-MS using gallium as an internal standard and a salt matrix calibration. Chromium was analyzed in Dynamic Reaction Cell mode with ammonia gas to react away the polyatomic interferences while cobalt was analyzed in Kinetic Energy Discrimination mode with helium gas to attenuate the polyatomic interferences. The lower limit of quantification of both chromium and cobalt was 1 ng/mL. Titanium in serum was analyzed by triple quadrupole inductively coupled plasma mass spectrometry in mass shift mode using ammonia as the reaction gas, with gallium as the internal standard, and a salt matrix calibration. Titanium was measured as cluster ions and the three isotopes m/z 114, 131, and 132 were summed. The lower limit of quantification was 1 ng/mL. All analytes were collected in appropriate trace element free tubes to minimize any contamination.

All 113 patients underwent cognitive evaluation by a psychometrist. The cognitive tests assessed four different cognitive domains (i.e., memory, language, attention/executive, and visuospatial) using nine measures, as described previously [17] administered by a psychometrist. Participating patients were also given the opportunity to undergo MRI and 81 patients had completed MRI scans with available measures of cortical thickness and are included in the present study. Patients who had ferromagnetic objects or nerve stimulators were excluded from MRI. MRIs were acquired on 3T Siemens Prisma scanners using 64-channel receiver head coils using previously published protocols and analyses [18]. All structural scans were acquired using magnetization prepared rapid gradient echo (MPRAGE) sequence having the following parameters: TR = 2300 ms, TE = 3.14 ms, TI = 945 ms, flip angle = 9°, and isotropic resolution = 0.8 mm. We used a FreeSurfer (version 5.3)-derived temporal meta-ROI cortical thickness as the neurodegeneration MRI measure which consisted of the surface-area weighted average of the mean cortical thickness in the individual ROIs: entorhinal cortex, fusiform, inferior temporal, and middle temporal [19]. Cortical thickness is a commonly used measure in imaging studies to capture neurodegeneration and compare the extent of atrophy across participants. Reduced temporal meta-ROI cortical thickness was defined as≤2.86 mm [19, 20]. Hippocampal volume and cortical thickness were also measured. The vascular neuroimaging biomarker acquisition has been described in detail previously [18, 21]. 3D magnetization prepared rapid gradient echo (MPRAGE) and 3D T2-weighted FLAIR images were used to assess white matter hyperintensity (WMH) volume via a fully automated algorithm. Subcortical (including lacunar) and cortical infarctions were identified and classified as previously described [21]. 3D MPRAGE and FLAIR images were reviewed. Trained image analysts marked the site of possible infarcts which were reviewed independently by a vascular neurologist and a neuroradiologist. Image analysts and clinicians were blinded to participant information. T2* GRE was used in analyzing microbleeds (CMBs) [22]. For the T2* GRE images, CMBs were annotated by image analysts and were further confirmed by a radiologist or a vascular neurologist [22].

Patient characteristics were summarized using descriptive statistics (mean, standard deviation [SD], median, interquartile range [IQR], count, percentage). The raw scores for the neuropsychological tests in each cognitive domain were z-scored and averaged to create domain-specific cognitive z scores. A patient global z score for overall cognitive performance was calculated by taking the mean of the four z scores (memory, language, attention, visual/spatial) for each patient, subtracting out the mean global z score for the cohort, and dividing by the cohort global z score standard deviation. The distribution of metal concentrations was highly skewed, and the data were log transformed to facilitate analysis. Metal concentrations were below the lower limit of quantification (LLOQ) for some patients. Missing values were imputed with a fixed value just below the LLOQ to preserve continuity of the distribution after log-transformation. The associations between the metal concentrations with both the cognitive scores and the MRI findings were examined with Pearson correlation coefficients. We also used age, sex, and education (higher than high school versus not)-adjusted linear and logistic regression to calculate the parameter estimates and confidence intervals. We built linear and logistic regression models, having all three two-way metal interactions (i.e., chromium*cobalt, chromium*titanium, cobalt*titanium) as independent variables together with chromium, cobalt, titanium concentrations and adjusted also for age, sex, and education for each of the cognitive z-scores as outcomes. No two-way interaction between pairs of metals measures was found to be significantly associated with any cognitive measure, and subsequently these models were not reported. In models for WMH and hippocampal volume, we also adjusted for total intracranial volume (TIV). No multiple comparisons correction was applied [23], as we largely examined the associations to generate hypotheses for future studies.

Analyses were considered statistically significant at a p-value<0.05. Statistical analyses were performed using SAS version 9.4 (SAS Institute; Cary, NC) and R version 3.6.2 (R Foundation for Statistical Computing; Vienna, Austria).

RESULTS

The mean age (SD) of the 113 TJA patients was 67.6 (9.7) years and 48 (42.5%) were female (Table 1). All but two patients were White. Additionally, 4 patients were current and 37 were former smokers. A total of 82 (72.6%) patients had a history of THA only, 11 (9.7%) TKA only, and 20 (17.7%) patients had both surgeries. Of the 102 patients with at least one THA, 54 had a history of metal-on-metal implants, 6 had ceramic-on-poly implants, and 3 had both metal-on-metal and ceramic-on-poly implants. The average time (SD) since their first hip or knee TJA surgery was 13.8 (5.7) years. Metal concentrations were detectable in 46 patients for chromium, 53 patients for cobalt, and 91 patients for titanium. Examining only detectable metal concentrations, the median (IQR) of chromium, cobalt, and titanium metal levels were 1.9 ng/mL (1.2, 3.6), 3.7 ng/mL (1.9, 6.3), and 3 ng/mL (2, 5), respectively. After substituting the undetectable metal concentrations with 0.99, the median (IQR) metal concentrations were 0.99 (0.99, 1.6), 0.99 (0.99, 3.3), and 3 (1, 5) ng/mL, respectively.

Figure 1 shows the correlation matrix of the 3 metals and all 5 cognitive measures. Univariate distributions of each of the metals and cognitive measures are shown on the diagonal. The lower left section shows scatterplots comparing intersecting variables. For example, the bottom left-most panel shows the scatterplot between log cobalt and the visuospatial z score. The upper right sections show the Pearson correlation coefficient between intersecting measures. For example, R = 0.02 in the top right corner is the correlation between the log of cobalt and the visuospatial z score. Figure 2 shows the correlation matrix of the 3 metals and MRI measures. There were no strong correlations between metal concentrations and cognitive measurements or the MRI findings.

OPEN IN VIEWER

Fig. 1

Pearson Correlation of Log Metal Concentrations with Cognitive Scores. Distributions of each of the metals and cognitive measures are shown on the diagonal. The lower left section shows scatterplots comparing intersecting variables. For example, the bottom left-most panel shows the scatterplot between log cobalt and visuospatial z score. The upper right sections show the Pearson correlation coefficient between intersecting measures. For example, R = 0.02 in the top right corner is the correlation between the log of cobalt and the visuospatial z score. Co, cobalt; Ti, titanium, Cr, chromium; Vsp, visuospatial.

OPEN IN VIEWER

Fig. 2

Pearson Correlation of Log Metal Concentrations with MRI measurements. Distributions of each of the metals and neuroimaging measures are shown on the diagonal. The lower left section shows scatterplots comparing intersecting variables. The upper right sections show the Pearson correlation coefficient between intersecting measures. Co, cobalt; Ti, titanium, Cr, chromium; Temporal meta-ROI, temporal meta-ROI cortical thickness; WMV, white matter hyperintensity volume; HV, hippocampal volume.

In age, sex and education-adjusted linear regression analyses, regression coefficients ranged between –0.11 and 0.20 for the association with cognitive scores but were not statistically significant (Fig. 3). As chromium concentrations increased, so did temporal meta-ROI cortical thickness (Estimate 0.04, 95% CI 0.002 to 0.08, p = 0.04) adjusting for age, sex, and education. After correcting for the log transformation of the chromium measure, double the amount of chromium in the blood corresponded to 0.01 mm (95% CI 0.0006 mm to 0.03 mm) increase in temporal meta-ROI.

OPEN IN VIEWER

Fig. 3

Associations between metal concentrations with cognitive scores and MRI measurements. A) Parameter estimates and confidence intervals from age, sex, education adjusted linear regression models (and total intracranial volume for the hippocampal and white matter hyperintensity volumes). B) Odds ratio and confidence interval from age, sex, education adjusted logistic regression models. Vsp, visuospatial; Temporal meta-ROI, temporal meta-ROI cortical thickness.

Cobalt concentration was significantly associated with odds for infarction (OR 4.3, 95% CI 1.8 to 10.1, p = 0.001). A doubling in the concentration of cobalt in the blood (i.e., 100% increase) was associated with a 21% increase in the odds of infarction (OR adjusted for log transformation 1.21, 95% CI 1.08 to 1.35).

Similarly for chromium, which was also significantly associated with odds for infarction (OR 2.1, 95% CI 1.03 to 4.5, p = 0.04) after adjustment for log transformation of the chromium measure, corresponded to 10% increase in odds for infarction with a doubling of chromium in the blood (adjusted OR 1.10, 95% CI 1.00 to 1.22). Otherwise, there were no statistically significant associations with the other MRI biomarkers, adjusting for age and sex.

DISCUSSION

In this exploratory study, we examined the cross-sectional association of blood metal concentrations with cognitive scores and neuroimaging measures in a convenience sample of patients with TJA with a history of having elevated metal concentrations of either titanium, cobalt and/or chromium. We hypothesized that cognitive performance and neuroimaging measures would be significantly worse among TJA patients with high metal concentrations. We found no evidence for an association between titanium, cobalt, and chromium concentrations with the cognitive scores. Higher whole blood cobalt and chromium concentrations were associated with higher odds of infarctions. However, higher concentrations of chromium were associated with higher temporal meta-ROI cortical thickness, opposite to our hypothesis. Given the convenience and small study sample with skewed metal measurements with a high percent of samples having metal concentrations below the detectable levels (especially for cobalt and chromium), findings may also be due to chance. Most of our findings did not support our hypotheses. Nevertheless, this study addresses a limited area of research, where larger studies with longitudinal follow-up are needed to further examine these associations, especially as TJA is such a common surgical procedure.

Cobalt and chromium may cross the blood brain barrier and be deposited in the brain [6]. Research is very limited but MRI findings of subtle structural changes in the visual pathways and the basal ganglia have been reported in a small cross-sectional study comparing brain volumes, metal deposition, and gray matter attenuation in patients with metal-on-metal hip resurfacing (8 years after the surgery) vs. age-,sex-, and time since surgery– matched patients with the same underlying disease, but with conventional hip prosthesis [6]; cobalt and chromium whole blood concentrations were elevated (5–10 time higher than the comparison group) but not indicative of failing prosthesis.

We observed a positive association between chromium concentrations and temporal meta-ROI cortical thickness. However, the estimates were small with wide confidence intervals reflecting the small sample size and uncertain clinical significance. There is a knowledge gap in this area of research. While chromium exists in several valence states (e.g., trivalent and hexavalent), the methodology used is unable to distinguish the valence states and measures total chromium. Whether a specific chromium valence state is responsible for the association with cortical thickness or this finding may simply be due to chance in this convenience sample is unknown. Studies in patients with metal-on-metal implants usually identify trivalent chromium, typically as chromium phosphate. However, it is difficult to assess whether chromium released from implants forms ions of higher valency as reduction to trivalent chromium happens rapidly; and this is important as hexavalent chromium is more toxic [13]. It can induce widespread oxidative damage, but its effects have not been fully elucidated, particularly in the brain [24]. Chromium can accumulate in the brain and probably bioaccumulates over the lifespan; high levels of chromium have been observed in the pituitary, hypothalamus, cerebellum, and temporal lobe [24]. However, additional studies are needed to understand chromium deposition, valence in the brain, and symptoms resulting from chromium induced toxicity. We did not find any statistically significant associations between titanium concentrations and cognitive or neuroimaging outcomes. Titanium dioxide (a frequently used biomaterial in orthopedic and dental implants, stents) is considered an inert and benign compound. Its ability to enter the nervous system in humans is largely unknown; titanium has been detected in the brain of rats treated intravascularly with titanium dioxide nanoparticles but not in those with titanium implants [4], while both groups of rats presented memory impairments. The main limitations of the present study were that we assessed cross-sectional associations and the small, convenience sample. We measured the metal concentrations, cognitive scores and performed the MRI scans at a fixed point in time. Future directions for this research include performing a longitudinal study where patients are followed regularly over multiple years and tested for blood metal concentrations, along with cognitive tests and MRI scans. Although the study assessed cognitive performance, it did not assess additional neurological symptoms. Another limitation was that we were not able to determine whether these patients were previously exposed to these metals from environmental sources, such as diet, and/or previous workconditions.

Given the high prevalence of TJA and limited evidence on potential long-term toxicity of metal implants, more research is warranted to better understand how systemic distribution of metal ions and elevated metal concentrations may affect other organs including the brain.