Establishment and verification of chromium and cobalt concentrations in joint fluids from a reference population

Highlights

• •Joint fluid measurements provide supplemental information in special cases.
• •Reference intervals in synovial fluid allow comparison of patients to a reference population.
• •The method presented here demonstrates acceptable performance with a urine calibration curve.

Abstract

This study aims to establish joint fluid reference levels for Chromium (Cr) and Cobalt (Co) in a reference population of available fluid types. Method performance was evaluated on an existing urine matrix calibration method using inductively coupled plasma-mass spectrometry. Method performance characteristics, including intra- and inter-assay imprecision, accuracy, linearity, AMR (analytical measurement range), sensitivity, and carryover were determined in accordance with clinical laboratory standards. Additionally, analytical and clinical recoveries were assessed to investigate comparability between available joint fluid types and existing calibrators prepared in urine to demonstrate acceptability of a matrix-substitution design. 124 de-identified joint fluid samples submitted for unrelated testing were used to establish the reference levels. Reference levels were determined, by using the 97.5th percentile, to be <20.4 µg/L for Cr and <29.9 µg/L for Co. In addition, the presented method overcomes the lack of an ample volume of joint fluid to use for matrix matched calibrators by employing a urine-based calibration curve. The data demonstrate acceptable matrix comparability.

1. Introduction

Total hip arthroplasty (THA), the replacement of the hip joint with a prosthesis, is a common surgery for patients suffering from hip disorders that can dramatically improve quality of life and reduce pain [1], [2]. It is estimated that more than 500,000 primary THA surgeries will be performed annually by the beginning of 2020 [3]. Several different types of hip prostheses are used for THA including metal-on-polyethylene, metal-on-metal (MoM) and ceramic-on-ceramic. MoM surface materials are predominantly composed of chromium (Cr) and cobalt (Co) alloys. In recent years, there have been reports of increased Cr and Co concentrations in joint synovial fluid, blood, and surrounding tissues of patients with MoM hip prostheses [4].

Monitoring of metal ion levels in asymptomatic patients with MoM bearings remains controversial [4]. Differences in the testing methodologies, result interpretations, reference intervals, clinical relevance, sample type selection, and unclear relationship between the metal ion levels and adverse reactions in patients vary considerably in the published literature [4], [5], [6], [7]. Whole blood, serum, and urine are most commonly used in toxic element assessment, as there are well-established reference intervals for each matrix. Although it has been proposed that joint fluid concentrations may be informative, to our knowledge, a reference interval study has not been previously published for Cr and Co in the general population.

Body fluids in general are relatively difficult matrices to acquire in quantities sufficient for use in matrix-matched calibration and control preparation. One goal of the present study was to assess the comparability of results using ICP-MS for various fluid samples analyzed against a urine matrix calibration. A recovery study was performed to ensure that recovery and ion suppression profiles between the matrices were equivalent and the resultant concentrations were accurate. In order to establish a reference interval for Cr and Co, we characterized and analyzed 124 fluid samples that were submitted for unrelated testing. 56 were described as synovial or joint fluid. The remaining samples were labeled as being from the following joints: knee (41), hip (11), elbow (3), leg (1), shoulder (1), and hand (1); the other ten samples had no description.

2. Materials and methods

2.1. Reference population

Previously analyzed fluid samples and retrospective data were de-identified and saved for analysis in compliance with the University of Utah Institutional Review Board (IRB #00007275). The 124 de-identified fluid samples that were analyzed to establish reference levels were submitted to ARUP Laboratories (Salt Lake City, UT, USA) for other testing including uric acid, glucose, rheumatoid factor, total protein, lactate dehydrogenase, amylase, bilirubin, sodium, potassium, borrelia species (lyme disease), C-reactive protein, and ANA (Anti-nuclear antibody) IgG. The choice of patient samples for reference level determination did not contain samples submitted for trace metal testing. This biased selection strategy was expected to reduce the occurrence of inadvertently including samples with MoM artificial joint complications. No definitive information about the presence of artificial joints in the reference population, or measures to reduce contamination during the collection process, was available.

2.2. Specimens

Specimen preparation included centrifugation after aspiration to remove cellular material. Subsequently, patient samples were transferred to ARUP standard transport tubes. The minimum acceptable sample volume was 0.2 mL. All de-identified joint fluid samples were stored frozen (-20 °C) until testing. For comparison to the reference population, results for 1152 previous fluid samples ordered for Co and 1092 ordered for Cr were tabulated retrospectively. All fluid samples were shipped to ARUP under refrigeration (2–8°C) and were maintained in frozen (−20 °C) storage after initial testing.

2.3. Chemicals, reagents, and solutions

Nitric acid was purchased from VWR (Radnor, PA) and Triton X-100 from Mass Supply (Highland Ranch, CO). Clinical Laboratory Reagent Water (CLRW) was from a Barnstead Nanopure Diamond System, from Thermo Scientific (Waltham, MA). Cr, Co, gallium (Ga) and yttrium (Y) stock solutions (1000 mg/L; >98% purity) were purchased from Inorganic Ventures (Christiansburg, VA) and the target concentrations listed on the accompanying Certificate of Authenticity was used for preparation of samples where applicable. Cr and Co stocks were prepared in CLRW. Calibration standards were prepared at 1, 5, 10, and 25 µg/L for Cr and 1, 5, and 10 µg/L for Co. Quality control samples were made in-house from separate elemental stocks spiked into synthetic urine matrix. Synthetic urine was prepared in CLRW in house with the following constituents and final concentrations: 0.33 M urea (Mass Supply), 0.12 M sodium chloride (VWR), 0.016 M potassium phosphate dibasic (Mass Supply), 0.007 M creatinine (Mass Supply), and 0.004 M sodium phosphate monobasic (VWR) as previously described [8].

2.4. Sample preparation

Patient joint fluid specimens and synthetic urine matrix quality control samples (100 µL), were diluted with 100 µL 1% nitric acid and 4.8 mL of acidic diluent containing 0.5% nitric acid and 0.05% Triton-X 100, as well as Ga and Y as internal standards (1 μg/L final concentrations). Y was used as the internal standard (IS) for Co analysis, while Ga was used as the IS for the analysis of Cr. Aqueous calibrators (100 µL) were prepared with 100 µL synthetic urine matrix, and 4.8 mL acidic diluent. A matrix-only sample was also prepared with 100 µL synthetic urine matrix, 100 µL 1% nitric acid, and 4.8 mL acidic diluent, to account for any endogenous elements found in the urine matrix. Each sample was prepared in 10 mL metals-free polypropylene tubes. The mixture was capped, inverted, and vortexed before introduction to the ICP-MS. Samples were quantitated using a standard addition curve with three and four calibrators for Co and Cr, respectively.

2.5. Cr ICP-MS method

Cr analysis was performed using a previously validated assay on a Perkin Elmer DRC II ICP-MS via a CETAC ASX-500 series autosampler (Omaha, NE) equipped with a 0.5 mm ID probe. The instrument was controlled by Elan software (ver. 3.4). The integrated peristaltic pump introduced the samples via a PFA PTFE Scott spray chamber with a MicroFlow PolyPro-ST nebulizer, followed by a set of platinum interface cones, and standard quartz torch with a 2.0 mm injector. The instrument was operated in Dynamic Reaction Cell (DRC) mode, with ammonium gas flowing at 0.55 mL/min. Cr (m/z 52) was monitored, as well as Ga at mass 69, as the internal standard. The method consisted of four sweeps, three readings, and two replicates with a dwell time of 500 ms per element. Calibration was achieved with four calibrators for Cr (i.e., 1–25 µg/L) using standards prepared in synthetic urine matrix with <1 µg/L Cr for matrix-matching, as previously described. A simple linear curve fit was applied with reagent and matrix blank subtraction.

2.6. Co ICP-MS method

Co analysis was performed using a previously validated assay on either a Perkin Elmer DRC II ICP-MS or Perkin Elmer 9000 ICP-MS via a CETAC ASX-500 series autosampler (Omaha, NE) equipped with a 0.5 mm ID probe. The instrument was controlled by Elan software (ver. 3.4). The integrated peristaltic pump introduced the samples via a PFA PTFE Scott spray chamber with a MicroFlow PolyPro-ST nebulizer, followed by a set of platinum interface cones, and standard quartz torch with a 2.0 mm injector. The instrument was operated in standard mode. Co (m/z 59) was monitored, as well as Y at mass 89, as the internal standard. The method consisted of three sweeps, two readings, and three replicates with a dwell time of 400 ms per element. Calibration was achieved with three calibrators for Co (i.e., 1–10 µg/L) using standards prepared in synthetic urine matrix with <0.5 µ g/L Co for matrix-matching. A simple linear curve fit was applied with reagent and matrix blank subtraction.

2.7. Method validation

The method validation included an assessment of intra- and inter-assay imprecision, accuracy, linearity, AMR (analytical measurement range), sensitivity, and carryover.

Intra- and inter-assay imprecision were determined using fluid specimens fortified with Cr and Co. Intra-assay imprecision was calculated based on the analysis of samples fortified at two different concentrations (level 1 and level 2), prepared and injected 20 times within a single batch. For inter-assay imprecision, samples were analyzed in quadruplicate for 5 days. Accuracy was determined by analysis of fortified fluid samples ranging from 1 to 30 µg/L. A total of 40 samples for each element was analyzed.

Linearity and AMR were assessed by analyzing fortified fluid samples over a range of 0–30 µg/L for Cr and Co. Prepared samples were run in quadruplicate in a single batch. To establish the limit of quantification (LOQ) fluid was fortified with elemental stock to 1 µg/L Cr and Co and analyzed in quadruplicate over 5 days. To determine the limit of blank (LOB) a diluent only sample was also analyzed in quadruplicate over 5 days with the LOB equal to the mean plus 1.65 times the standard deviation. Calculation of the LOB in this manner provides a minimum concentration for the matrix additive (e.g., synthetic urine) to ensure the reagent blank and the matrix only sample used in the calibration do not overlap in CPS which can lead to overcorrection by the instrument software and negative patient values. A dilution strategy with nanopure water used to dilute the specimen prior to diluent addition was validated at 2×, 5×, 10×, and 100× dilutions. Target values ranged from 40 to 2000 µg/L for Cr and 10–500 µg/L for Co.

Carryover was assessed using a target of 100 µg/L for Cr and Co as elevated concentrations. Three sets of two low-concentration samples (i.e., 5 µg/L Cr and Co) proceeded by two high-concentration samples were analyzed. Injection sequence was as follows: High (H1), High 2 (H2), Low 1 (L1), Low 2 (L2) for a total of three series of injections. The High samples were fortified to 100 µg/L while the Low samples were fortified to 5 µg/L. The percent carryover was then calculated using the following equation: Percent Carryover = 100 * [(L1_AVG – L2_AVG)/H2_AVG].

2.8. Analytical and clinical recovery

Five unique patient pools in fluid and in urine were prepared for the assessment of recovery, and to ensure the matrices were analytically similar. Each sample was either fortified with Cr and Co prior to acidic dilution (Pre) or after acidic dilution (Post) with comparison to fortified diluent without matrix (Unextracted, UNX). After analysis, the results were reviewed in three sample sets, Pre/Post to assess extraction recovery, Post/UNX for assessment of ion suppression, and Pre/UNX for overall efficiency and matrix effects. Normalized recovery ratios were calculated after the counts per second (CPS) were normalized to Ga or Y internal standards, while the non-normalized recovery was determined based on absolute elemental CPS without internal standard normalization.

2.9. Reference level determination

To our knowledge, Cr and Co reference levels in joint fluid have not been previously reported. The 124 de-identified joint fluid samples were analyzed to establish the reference levels.

Of the samples analyzed, 56 were described as synovial or joint fluid. The remaining samples were described as being from the following joints: knee (41), hip (11), elbow (3), leg (1), shoulder (1), and hand (1); the other ten samples had no description.

2.10. Data analysis

Data analysis was performed using Microsoft Excel (Microsoft, Redwood CA) and EP-Evaluator (Data Innovations, Burlington, VT) where noted.

3. Results

3.1. Method validation

Intra-assay and inter-assay imprecision were assessed at 5 and 10 µg/L for each element. The CV for both elements was <6% for intra-assay imprecision and <9% for inter-assay imprecision.

Accuracy for Cr was determined through comparison of 40 fortified fluid samples. Deming regression for the entire data set indicated a slope of 0.92 and an intercept of 0.89 with an R² of 0.96. The relative bias was 0.018 μg/L for the entire data set. Accuracy for Co was determined through comparison of 40 fortified fluid samples. Deming regression indicated a slope of 1.04 and an intercept of 0.90 with an R² of 0.98. The relative bias was −0.132 μg/L for the data set.

The AMR for Cr and Co was established as 1–25 µg/L and 1–10 µg/L, respectively. The maximum dilution factor of 100× extended the reportable range up to 2500 µg/L for Cr and 1000 µg/L for Co. The LOQ for Cr and Co was established at 1 µg/L with a CV of 17.0% and 14.2%, respectively, and the LOB was <0.85 µg/L for both elements. Carryover was determined to be 0.15% and 0.28% for Cr and Co at 100 µg/L.

3.2. Analytical and clinical recovery

Five unique patient pools were fortified with Cr and Co prior to dilution (Pre, total recovery), and after dilution (Post, ion suppression) with comparison to elementally fortified diluent without matrix (UNX, unextracted). Clinical recoveries with normalization and analytical recoveries based on non-normalized elemental counts were compared. This process was performed on both the fluid patient pools and urine patient pools to ensure matrix equivalence. Fluid clinical recovery results are summarized in Table 1.

Table 1.

Non-normalized and normalized recovery data for Cr and Co in fluid and urine.

Condition	Avg. Normalized Recovery (%), 5 pools			Avg. Non-Normalized Recovery (%), 5 pools
	Pre/Post	Post/UNX	Pre/UNX	Pre/Post	Post/UNX	Pre/UNX
Cr (joint fluid)	109	73	79	108	62	67
Cr (urine)	108	80	86	109	67	73
Co (joint fluid)	97	105	102	99	87	86
Co (urine)	106	107	113	107	91	97

3.3. Reference level determination

Out of the initial 124 samples obtained for reference level determination, single outliers were observed for Cr (at approximately 3123 µg/L) and Co (at 675 µg/L) and removed from the dataset. Non-parametric reference levels for both metals were set at the 97.5th percentile of the remaining 123 samples. In addition, a retrospective analysis of over 1000 samples for both Cr and Co fluid testing was performed. The percent of retrospective samples above the 97.5th percentile of the reference level was calculated. Reference level characteristics are summarized in Table 2.

Table 2.

Reference level characteristics and Interquartile Range (IQR) for Cr and Co metal ions.

Reference population Metal ion conc. (µg/L) n = 123 (73 males, 50 females)								n = 1092 (Cr) n = 1152 (Co)
	97.5 percentile	Median	Mean	Minimum	Maximum	IQR	>97.5 percentile*
Cr	20.4	<1	4.10	<1	99	2.7	79% (862)
Co	29.9	<1	4.62	<1	252	0.5	81% (928)

^*Retrospective analysis of patient samples submitted for Cr and Co joint fluid testing.

4. Discussion

Cr and Co joint fluid ICP-MS assays were validated and shown to be accurate and robust using a synthetic urine calibration curve allowing for incorporation into an existing urine assay. Comparative recovery studies of urine and joint fluid matrices revealed matrix equivalence. This eliminated the need to obtain a matched calibration matrix for this assay that would have otherwise been a major challenge due to limited sources of joint fluid matrix without detectable metal levels. The assays showed excellent intra- and inter-assay precision, with AMRs and reportable ranges of 1–25 µg/L and 1–2500 µg/L for Cr and 1–10 µg/L and 1–1000 µg/L for Co, respectively. The 40 fortified fluid samples used to test accuracy were within the linear range and showed correlations of 0.96 and 0.98 for Cr and Co, respectively. In addition, the LOQ (1 µg/L Cr and Co), LOB (<0.85 µg/L Cr and Co), and carryover at 100 µg/L (0.15% Cr and 0.28% Co) were assessed. This assay can be used in the assessment of joint fluid collected from patients with MoM hip implants.

The reference level was set at the 97.5 percentile in accordance with the Centers for Disease Control guidelines for lead levels [9], as this method is predominantly used for the determination of values from skewed distributions in the setting of toxicity assessment. Based on these criteria, 79% (Cr) and 81% (Co) of patient joint fluid samples originally submitted for metal testing, most likely associated with MoM implant surveillance, were above the established reference level. Taken together, this portion of patients can be successfully differentiated from a reference population using the reference levels established in this study. The utility of these reference levels is further supported by a previous report [10] that used comparable methodology (i.e., ICP-MS). Their reports (N = 26) indicate median values for Cr (i.e., 179.5 μg/L) and Co (i.e., 106.25 μg/L) for patients with MoM implants, but without metallosis, are much higher than the 97.5th percentile of our established reference levels. It is important to mention that the minimum concentrations of Cr (i.e., 19 μg/L) and Co (i.e., 13 μg/L) observed by De Smet et al. in patients with MoM implants, but without metallosis, would fall below the established reference levels, but would not pose misclassification risk to these patients as having metallosis. More importantly, our established reference levels for Cr and Co are much lower than the minimum values observed in MoM implant patient with metallosis: 155 μg/L for Cr and 110 μg/L for Co, as reported by De Smet et al. In vitro studies using MLO-Y4 osteocytes indicate dose-dependant cytotoxicity of Co at concentrations greater than 0.1 mM (5.9 mg/L). No Cr toxicity was observed in vitro for osteocytes at 0.5 mM (26.0 mg/L). In addition, Co-associated morphological changes to the osteocytes have been documented to occur at concentrations of 100 μg/L [11]. Reduction in collagen type I synthesis in osteocytes has also been documented for Co and Cr at concentrations of 10 μg/L or 100 μg/L [12]. Our reference levels for Cr (i.e., <20.4 μg/L; <0.00039 mM) and Co (i.e., <29.9 μg/L; <0.00051 mM) are below what is reported as cytotoxic to osteocytes in most studies.

Despite the assumption that Cr should not be present in synovial fluid, Cr has been reported in patients with rheumatoid arthritis (N = 13; range of 6–100 µg/L), but without MoM implant [13], albeit using a different methodology (i.e., emission spectrometry). That cannot also be said about Co, as the same authors did not assess the presence of Co in these patient samples. As with all clinical tests, reference intervals provide a necessary context to test results and are required to be reported. Up to this point, joint fluid results have been difficult to interpret due to the lack of suitable reference intervals and poor correlation with serum or whole blood specimens. Davda et al. [5] reported that for a wide range of Cr and Co joint fluid levels in vivo the results correlated only moderately with whole blood analysis. However, De Smet et al. [10] demonstrated a high correlation between serum and joint fluid samples. This may be due to a lack of correlation between serum and whole blood samples and the distribution of Cr species between the cellular compartment and plasma [14], [15]. Generally, serum or blood testing is the preferred specimen for the assessment of metal toxicity from failing MoM prosthetics; however, a secondary source of metal toxicity or renal insufficiency may complicate serum result interpretation. In such instances, joint fluid aspirations can aid in identifying excessive wear of the joint and its contribution to total metal toxicity in a patient. Despite its usefulness, the difficulty of justifying synovial fluid collection in healthy donors has hindered efforts to establish reference intervals for Cr and Co in synovial fluid. Given this limitation, it is clinically valuable to establish a reference level using residual samples submitted for unrelated tests. The utility of such a reference level is further supported by the ability to distinguish the majority of tests submitted for Cr and Co testing in joint fluid in our lab, as well as the ability to distinguish from values of MoM patients with metallosis reported in the literature. It is important to acknowledge the limitations of this study, including the absence of information about the presence of artificial joints in the reference population and other routes for potential metal contamination that could have occurred during the collection process.

5. Conclusions

The method presented here for the analysis of Cr and Co in fluid addresses the lack of available joint fluid needed for matrix matched calibrators by demonstrating acceptable performance using a urine-based calibration curve. We have shown that the joint and urine matrices are comparable and validated the method quantify Cr from 1 to 25 µg/L and Co from 1 to 10 µg/L in joint fluid. Validation of sample dilutions extended the AMR of Cr to 2500 μg/L and Co to 1000 μg/L. In addition, reference levels were successfully set for Cr at <20.4 µg/L and Co at <29.9 µg/L.

Conflicts of interest

The authors included have no conflicts of interest to disclose.

Author contributions

C. J. Haglock-Adler is the primary contributor to research result acquisition and initial manuscript drafting. V. Gruzdys is responsible for critical revision of the manuscript and formatting of the results. K. Womack contributed to the identification and collection of joint fluid samples as well as their initial characterization. F. G. Strathmann is responsible for research design, data analysis and interpretation, and approval of the submitted and final versions of the manuscript. All authors have read and approved the manuscript.