*2.2. Cytokine Profile Analysis*

Snap frozen tissue samples from group AL (+), group AL (−) and the control group were mechanically disrupted and homogenized (Precellys®24 and Cryolys®—Bertin Technologies, Bie & Berntsen A/S 2730, Herlev, Denmark) at 4 ◦C for 4 × 20 s in lysis buffer containing protease inhibitor cocktail (REF 11836145001, Roche Diagnostics, Indianapolis, IN, USA). Homogenized tissue samples were then spun for 10 min at 10,000 × G at 4 ◦C (Microcentrifuge 157MP—Ole Dich Instrumentmakers ApS, Hvidovre, Denmark) and the protein concentration in the supernatant was estimated by Bradford protein assay [28] using Coomassie blue (#1610436. Bio-Rad Laboratories, Inc., Hercules, CA, USA). Prior to cytokine analysis, total protein concentrations of the samples were adjusted to 0.5 mg/mL. Cytokine analysis was performed using a validated V-PLEX electrochemiluminescence immunoassays (Meso Scale Discovery, Rockville, MD, USA). A total of 11 cytokines divided on two separate kits were analyzed. Proinflammatory Panel 1 contained; IL-1β, IL-2, IL-4, IL-6, IL-8, IL-10, IFN-γ and TNF-α (catalog # K15049D-1), and cytokine panel 1 kit contained; IL-15, IL-17A, GM-CSF (catalog # K15050D-1). Samples were analyzed in triplicates (MESO QuickPlex SQ 120—Meso Scale Discovery). Calibration curves used to calculate cytokine concentrations were established by fitting to a 4 parameters logistic model with a 1/Y<sup>2</sup> weighting. Cytokine concentrations were calculated using the Discovery workbench

4.0.12 software (Meso Scale Discovery). Serum samples were analyzed undiluted using the same cytokine kits as used for the tissue.

#### *2.3. Patch Testing*

A special patch test series, provided by Smart Practice®(Phoenix, AZ, USA), was used in this study. The patch contained prefabricated panels with metallic compounds associated with orthopedic prostheses on Scanpor tape. Standard metal allergens included; nickel (II) sulphate NiSO4 (1.0 wt.%), potassium dichromate (VI) K2Cr2O7 (0.054 wt.%) and cobalt (II) chloride CoCl2 (0.02 wt.%). In addition, a customized panel with the following metals and corresponding titrations were included; vanadium (IV) oxide sulfate hydrate VOSO4·H2O (0.36, 0.18, 0.06, 0.02 wt.%), vanadium (III) chloride VCl3 (0.24, 0.12, 0.013, 0.04 wt.%), manganese (II) chloride MnCl2·4H2O (0.24, 0.08, 0.06, 0.0057 wt.%), aluminum (III) chloride AlCl3·6H2O (0.72, 0.38, 0.039 wt.%), ammonium molybdate (VI) (NH4)6Mo7O24 4H2O (0.12, 0.013, 0.04 wt.%), titanium (IV) oxalate hydrate TiC4O8·H2O (0.32, 0.16, 0.08, 0.04 wt.%), titanium (IV) dioxide TiO2 (0.24 wt.%), potassium titanium (II) oxide oxalate C4K2O9Ti·2H2O (2.4, 1.2, 0.6 wt.%), ammonium titanium (II) lactate, solution Ti [(C3H4O3)2(NH4OH)2] (0.16, 0.08, 0.04 wt.%), ammonium titanium (IV) peroxocitrate (NH4)4[Ti2(C6H4O7)2(O2)2]·4H2O (0.32, 0.16, 0.08, 0.04 wt.%). methyl methacrylate C5H8O2 (2 wt.%), gentamycin sulfate (20 wt.%) and ferrous chloride FeCl2 (2 wt.%) were tested by manually loading of a Finn chamber on Scanpor tape. Patches were applied on the upper back and were occluded for 48 h. Readings were completed 96 h after application [29]. The patients were instructed to remove the panels after 48 h, and not to shower, scratch or expose to sunlight. Reactions were scored using the International Contact Dermatitis Research Group's (ICDRG) criteria [30]. Only definite +1, +2 and +3 reactions were regarded as positive.

#### *2.4. ICP-MS (Serum)*

Blood samples were sent to Vejle Hospital, Department of Clinical Biochemistry, Denmark, for determination of chromium and cobalt levels before the surgery. The samples were analyzed by ICP-MS instrument (iCAPq, Thermo Fisher Scientific Inc., Waltham, MA, USA). The samples were diluted with 0.5% HNO3, gallium was added as an internal standard prior to analysis. The detection limit was 10 nmol/L equivalent to 0.59 ppb (cobalt) and 0.52 ppb (chromium).

### *2.5. ICP-MS (Tissue and Serum)*

Elemental analysis of tissues and titanium (Ti) analysis of blood was performed at the National Food Institute at the Technical University of Denmark.

*Elemental analysis in tissues:* Tissue samples (0.1–0.5 g) were digested with a mixture of concentrated nitric acid (4 mL; PlasmaPure, SCPScience, Courtaboeuf, France) and hydrogenperoxide (1 mL; Merck, Darmstadt, Germany) in a microwave oven (Multiwave 3000, Anton Paar, Graz, Austria). The concentration of aluminum (Al), vanadium (V), chromium (Cr), cobalt (Co) and nickel (Ni) was determined using ICPMS (iCAPq, Thermo Fisher Scientific, Waltham, MA, USA) using rhodium as an internal standard and external calibration. The ICPMS instrument was run in the kinetic energy discrimination (KED) mode using helium as a collision cell gas. The limit of detection was estimated at 100 μg/kg for all elements.

*Determination of Ti in tissue and blood:* The acid digests of tissues were also subjected to Ti analysis. Serum subsamples (200 μL) were diluted with 4.8 mL diluent solution consisting of 0.5% Triton X-100, 10% ethanol (both Merck) and 1% nitric acid (SCPScience) prior to the analysis of the concentration of Ti using a triple quadrupole ICPMS (Agilent 8800 ICP-QQQ, Agilent Technologies, Yokogawa, Japan) and using ammonia as a cell gas with determination of Ti after MS/MS mass shift from *m*/*z* 48 ≥ *m*/*z* 150 with scandium (Sc) as internal standard and external calibration. The data quality of Ti analysis was assessed by the analysis of the reference material Seronorm (Sero, Oslo, Norway). The obtained value 7.2 μg/L was in good agreement with the reference value 6.8 μg/L. The limit of detection was estimated

at 1 μg/L in serum samples and 20 μg/kg in tissues. All calibration standards and internal standards were produced from certified single-element stock solutions (SCPScience).

### *2.6. Statistical Analysis*

For group comparison the Kruskal-Wallis test was used, and if statistically significant, the Mann-Whitney U test was used to compare between individual groups. By convention, to calculate group medians, metal concentrations below the detection limit were assigned a value of one-half the detection limit. Comparisons were made using the Mann-Whitney test. Contingency tables (patch test) were analyzed using Fisher's exact test. A significance level of *p* < 0.05 was considered statistically significant. Matlab R2014a (8.3.0.532) with statistical toolbox (MathWorks Inc. Natick, MA, USA) was used for statistical analysis. For graphical representation Prism 6.0 (GraphPad Software, San Diego, CA, USA) was used.

### **3. Results**
