Correlation of m XRF and LA-ICP-MS in the analysis of a human bone-cartilage sample †

Determining the local distribution of major, minor and trace elements in biological tissues provides fundamental information on the chemical composition and often allows conclusions on the underlying biochemical mechanisms. Within this project we aimed for the elemental characterization of a complex biological sample, a human femoral head, which includes both hard (bone and mineralized cartilage) and soft tissue (hyaline cartilage). Two elemental imaging methods were employed – microbeam X-ray ﬂ uorescence spectrometry ( m XRF) and laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS). m XRF and LA-ICP-MS are widely used for chemical analysis in various applications. While bone consists mostly of mineralized material and its elemental composition is in the focus of investigation (both in health and disease), hyaline cartilage is primarily composed of organic molecules such as glycosaminoglycans, proteoglycans, collagen, and its elemental content is not well studied. In this proof-of-principle study, m XRF was followed by LA-ICP-MS on the very same sample and region. The measurements resulted in the elemental maps of bone, as well as a cartilaginous area of the sample (including the tidemark – interface between calci ﬁ ed and hyaline cartilage). In addition to the spatial distribution of elements, we also obtained quantitative information by using matrix-matched standards.


Introduction
Investigation of biological tissues oen requires obtaining comprehensive chemical elemental information. Recent developments in elemental imaging techniques allow visualisation of elemental distributions on different levels of organizationfrom model organisms 1 to tissues 2,3 and single cells. 4 Bone, due to its inorganic componentsit contains a mineral phase (hydroxyapatite), is a tissue especially suitable for elemental analysis. Furthermore, the sample preparation and sample handling requirements are oen simpler compared to other biological tissues as there is usually no need for cryopreservation and thicker samples are self-supporting, which reduces the need for a backing material. The topics of investigation which were successfully addressed using elemental imaging in bone range from bone architecture, 5 content of specic trace elements in various conditionsi.e. osteosarcoma, 3 to investigation of novel orthopaedic materials. 6 Within joints, bone is covered by articular cartilagewhich, based on the degree of mineralization, is subdivided into mineralized and hyaline cartilage. The interface between mineralized and hyaline cartilage is called tidemark. The mineralized tissue and especially the tidemark are intensely studied. It is known that the tidemark is the site of metal accumulation, e.g. zinc and lead, and several elemental imaging experiments have been devoted to this topic on animal or human cartilage samples. [7][8][9][10] Compared to bone, hyaline cartilage is substantially less investigated in terms of elemental composition. There are some imaging approaches to visualize hyaline cartilage, such as uorescent microscopy, contrastenhanced scanning electron microscopy, computed tomography and magnetic resonance imaging, 11,12 however, the imaging experiments involving the elemental distribution are scarce.
Therefore, in our proof-of-principle experiment we tested the analytical capabilities of two elemental imaging techniques within one samplethe cuts of a femoral head which included both bone and cartilage tissues. For analysis we employed the methods of microbeam X-ray uorescence analysis (mXRF) and laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS).

mXRF
The method is based on the excitation of the sample by X-rays, resulting in the emission of characteristic radiation by each particular element, provided that the excitation energy is above the absorption edge of the element of interest (the measurable range generally spans from sodium to uranium, and can depend on the technical characteristics of the spectrometer and experimental parameters chosen). The uorescence radiation emitted from the sample is registered by an energy dispersive detectorand the information about various elements present in the sample is obtained simultaneously. The resultant characteristic spectra are deconvoluted using dedicated soware: spectral overlaps and artefacts are considered, and background is subtracted in order to extract information about elements of interest. In imaging experiments a sample is raster scanned, yielding elemental maps. The spatial resolution of such a scan/ map depends on the size of the X-ray beam (as well as on the energy of the exciting X-rays and the emitted characteristic X-ray of the sample element). In the case of mXRF, the beam is focused to a few tens of micrometres, so the resolution is in the similar range. The technique is non-destructive, and hence the sample can be further analysed. The technique is oen used for qualitative imaging; nonetheless, quantication is possible; e.g. by use of matrix-matched standards.

LA-ICP-MS
In this technique, a short-pulsed, high-power laser beam is focused onto the sample surface, leading to the ablation of a nite volume of sample material. The resulting solid aerosol is transported by a carrier gas stream into an inductively coupled plasma, where the sample is atomized and ionized. In a nal step, the formed ions are introduced to a mass spectrometer, usually equipped with a quadrupole mass analyser. LA-ICP-MS provides major, minor, and trace element information with a wide elemental coverage. It also offers the possibility to perform elemental mapping with high spatial resolution (mostly determined by the laser beam diameter) in the micrometre range. Since LA-ICP-MS suffers from sample-dependent ablation behaviour and elemental fractionation, reliable signal quantication of such imaging experiments is challenging. To overcome these issues, matrix-matched calibration standards in combination with an internal standard correction are generally used. 13 Both mXRF and LA-ICP-MS are well-established in the analysis of biological samples. [14][15][16] As both techniques provide elemental information, there exist some comparative studies. Davies et al. compared synchrotron-induced mXRF and LA-ICP-MS applied to consecutive slices of neurological tissue (mouse brain) for imaging and quantication of metals. 17 Few studies have been published on the sequential use of both techniques on the same biological sample. Gholap et al. used laboratory mXRF and then LA-ICP-MS for mapping the same thin section of Daphnia magna; 1 and Müller et al. used triple quadrupole LA-ICP-MS (LA-ICP-TQMS) preceded by mXRF on the same murine liver section. 18 In our study we applied mXRF followed by LA-ICP-MS in a partially complementary manner to obtain comprehensive information on the distribution of the elements within tissues. The objectives of the study included: characterisation of bone and cartilage tissues in terms of elemental distribution, collection of quantitative information for the detected elements in those tissues, and comparison of the two elemental imaging methods.
The project is a part of ongoing effort at TU Wien to advance correlation between imaging methods in biomedical applications. [19][20][21] Materials and methods (1) Sample preparation A femoral head biopsy sample from a female patient (93 years old), previously collected for a study investigating the relation between bone quality and homocysteine levels (high homocysteine level group) 22 was used in the present study. All studyrelated procedures were approved and supervised by the local ethical committee of the AUVA Trauma Centre Meidling, Vienna, Austria, following the Declaration of Helsinki (DoH). The patient provided written consent for anonymized outcomes to be further used in scientic examinations.
The undecalcied biopsy was xed in 70% ethanol, dehydrated through a graded ethanol series, and embedded in PMMA (polymethylmethacrylate). Thin sections were prepared out of a single PMMA block. 15 mm slice was cut using a microtome (LEICA SM2500; Leica Microsystems GmbH, Wetzlar, Germany); while 41 mm was obtained by grinding with sandpaper (rst abrasive grit size of 1200 up to the desired thickness then abrasive grit size 2400) and polishing using polycrystalline diamond spray (3 mm and 1 mm). Further details on sample preparation can be found elsewhere. 23 The samples were prepared in such a way as to include both bone and articular cartilage tissues. Fig. 1 shows the morphology of the sample. Two slices of thickness 15 and 41 mm were used for the imaging experiments.
(2) mXRF measurements, data analysis and quantication An in-house-built mXRF setup at the Atominstitut, TU Wien, was employed non-confocally (i.e. only one focusing optic, a polycapillary lensbetween X-ray tube and sample). More information on the setup can be found in 6,21,24 (see also Fig. S1 in ESI †).
For the measurements, the samples were sandwiched between acrylic glass frames and attached to a sample holder, which facilitated direct measurement on the sections and eliminated unwanted uorescent/scattering contributions of backing materials. The samples were positioned vertically in the spectrometer chamber; the standard geometry 45 between the incident radiation beam and the sample, and 90 between the incident beam and the detector was maintained. For these measurements, a Rh-anode low power X-ray tube was used; the tube voltage and current were set to 50 kV and 0.4 mA, respectively. The measurement conditions were: 50 mm step size, 60 s acquisition time per spot (real time), in vacuum (<3 mbar). The step size was chosen due to the estimated beam size of 50 mm (at 8 keV). Deconvolution of the spectra was performed using the QXAS-AXIL soware package. 25 The original elemental maps were created as text les using the soware LP-map 26 and processed (negative values zeroed for correct representation) and plotted with ImageJ (version 1.50b). 27 The XRF data were dead time corrected. The XRF maps were quantied using the sensitivity values obtained via the measurements of IAEA Animal Bone (H-5) certied reference material, the details can be found in the Master's thesis of V. Pichler. 28 The sensitivity values of the elements for which data was not availablesulphur and potassium, were derived by interpolation. The thickness of the cuts was not considered in quantication. The elemental maps, showing the whole scanned areasource images, are included in the ESI. † For easier comparison and correlation of XRF and LA-ICP-MS data, further analysis by ImageLab (v. 2.99 Epina GmbH, Pressbaum, Austria) was performed (see also below in (3) LA-ICP-MS measurements and data analysis)and the resultant elemental maps, showing the very same area scanned by both techniques are included in the current manuscript. The original mXRF maps can be found in the ESI. † Fluorescence sum spectra were plotted with Origin (version OriginPro 2015, OriginLab Corporation, Northampton, MA, USA).

(3) LA-ICP-MS measurements and data analysis
A NWR213 laser ablation system (ESI, Fremont, CA, USA) equipped with a frequency quintupled 213 nm Nd:YAG laser and a fast-washout ablation cell, always held above the actual ablation site, was used for all shown LA-ICP-MS experiments. The device was coupled to an iCAP Q quadrupole ICP-MS instrument (Thermo Fisher Scientic, Bremen, Germany) using PTFE tubing (inner diameter 2 mm). A schematic representation of a general LA-ICP-MS setup is shown on Fig. S1 in ESI. † Helium was used as carrier gas for cell washout, which was mixed with Ar make-up gas upon introduction into the plasma. For data acquisition, Qtegra soware provided by the manufacturer of the instrument was used. The tune settings of the MS instrumentation were optimized for maximum 115 In signal prior to each experiment using a NIST 612 trace metal in glass standard (National Institute of Standards and Technology, Gaithersberg, MD, USA). Additionally, the oxide ratio was monitored by the 140 Ce 16 O/ 140 Ce ratio, which was below 2.0% for all experiments. To reduce the inuence of polyatomic interferences, kinetic energy discrimination (KED) mode utilizing a collision cell containing a mixture of helium with 7% hydrogen was used. A list of potential polyatomic interferences is given in Table S2. † Two-dimensional elemental distribution maps were created using line scan ablation patterns with adjoining lines. In case of the 15 mm slice, an additional pre-ablation step using a similar pattern and reduced laser uence (4.6 J cm À2 ) was performed prior to the measurement, removing potential surface contaminations. For image processing, the soware ImageLab (v. 2.99 Epina GmbH, Pressbaum, Austria) was used. A pixel size of 25 mm was chosen for the distribution maps, which corresponds to the used laser beam diameter. For plotting the images, interpolated colour surfaces were employed. To evaluate the agreement between obtained maps with the corresponding mXRF images, a correlation analysis was performed using the correlation feature integrated in the soware. Detailed information about the used LA-ICP-MS parameters is given in Table 1.
Taking into account, that the samples include tissues with signicant variations in chemical composition (mineralized tissues and hyaline cartilage), two different standard materials were used for the quantication. While analytes predominately expected in the calcied area (Ca, Mg, Na, P, Zn) were quantied using an IAEA Animal Bone (H-5) certied reference material,  The measurement of the standards was performed directly before the imaging experiments using identical instrument and laser settings. Each standard was analysed four times using laser patterns consisting of adjacent line scans. The investigated area was 0.35 mm 2 per laser pattern. For further data processing, the mean values of the four-fold analysis were used. To remove potential surface contaminations, a pre-ablation step was performed before the measurements.

Results and discussion
Local elemental distributions were obtained from the measurement areas, which were chosen such as to include bone, articular cartilage tissue and the metabolically active area of tidemarkthe interface between calcied and non-calcied cartilage; the elemental maps of 15 mm and 41 mm cuts are shown on Fig. 2 and 3, respectively. These gures show the very same area within each slice imaged by mXRF and LA-ICP-MS (further details are provided in the text below).

mXRF
The sum spectra (generated by adding spectra from each measured point) obtained from mXRF scans of two cuts are shown in Fig. 4. High content of Ca and P corresponds to the bone and calcied cartilage area, while high S peak can be attributed to the hyaline cartilage. Zn, Pb and Sr are expected to be found in the sample (Zn in both cartilage and bone, accumulation of Zn and Pb in tidemark, Srin bone), although Pb and Sr peaks are hardly discernible, therefore, these metals were excluded from further analysis. Fe, although normally present in tissues (e.g. from blood), can also be introduced in the sample by the cutting equipment, therefore, Fe, Co, as well as Ti and Cr (only found in the 41 mm cut) were considered contaminations. Interestingly, in the sum spectrum of 41 mm cut, peak of K can be recognised on the le shoulder of Ca-Ka peak. Fig. 2 and 3 show segments of the scans performed on the 15 mm cut and 41 mm cut, respectively, to match the areas imaged by both methods. The whole mXRF scans are included in the ESI (Fig. S2 and S3 †). A detailed description of the mXRF results provided below is based on the original mXRF maps. Please note that the values discussed further in this section originate from the Tables S3 and S4 in the ESI. † mXRF results for 15 mm slice (Fig. 2, le column). Ca and P content is naturally high in the calcied tissues (bone and mineralized cartilage) with average concentrations of 120 mg g À1 and 80 mg g À1 , respectively. Ca is also present throughout the hyaline cartilage, although not visible on the map due to the colour scheme scaling. Within the cartilage, Ca content is overall rather homogeneous in the middle cartilage layer -1.5-2.5 mg g À1 , and slightly lower in supercial layer, 1-1.5 mg g À1 . In the vicinity of the tidemark Ca concentration rises sharply, and within tidemark Ca mean values are similar to bonearound 100 mg g À1 .
Sulphur is present exclusively in the hyaline cartilage with an average concentration of ca. 17 mg g À1 . It should be noted that the presence of S (values #4 mg g À1 ) within calcied tissues occurs due to a spectral artifact: the high Ca signal causes an escape peak of Ca-Kb line at 2.27 keV. This articial peak has similar energy to S (2.3 keV), therefore tting soware misrelates the signal. The line structures, which can be seen within the hyaline cartilage in microscopy picture and also in the S map correspond to the folds on the sample surface. The concentrations of S and Ca in the folds are nearly two times higher than in the surrounding tissue.
Zn is present in bone and, in higher content in the tidemark zone. The estimated mean Zn concentration is 15 mg g À1 in bone and 50 mg g À1 in tidemark.
mXRF results for 41 mm slice (Fig. 3, le column). Similar to the 15 mm slice, there are high concentrations of Ca and P in the calcied tissues quantied to 190 mg g À1 and 100 mg g À1 , respectively (average values over the calcication region). Ca is also conrmed in every scan point within hyaline cartilage, its content ranges from 2-3 mg g À1 in middle layer to up to 160 mg g À1 in the tidemark zone.
Average S content in hyaline cartilage is around 27 mg g À1 . The maps are subject to the same tting problem for S in the presence of high Ca content as described above for the 15 mm slice. However, unlike the 15 mm slice, S distribution was conrmed in the voids of trabecular bone, its origin remains unknown.
Zn is detected in bone, as well as in tidemark, where the concentration reaches 150-200 mg g À1 . When comparing Zn XRF maps of the two slices, it can be seen that overall a more distinct Zn distribution was obtained for the thicker sample. However, the differentiation between bone and tidemark in the 41 mm cut is not as clear as in 15 mm slice. This can be understood considering that the theoretical information depth for Zn is about 98 mm. Therefore, in both cuts the signal from the entire sample thickness is obtainedalbeit stronger in 41 mm slice. In the thicker sample the information depth contributes also to the smearing effect for such a thin structure as tidemark, making tidemark in 41 mm sample less pronounced than in the 15 mm cut.
K is present in the hyaline cartilage of the 41 mm slice, and there is a clear separated peak of K in corresponding spectra. The average K concentration in hyaline cartilage is 1.3 mg g À1 . The K map seems to exhibit a higher K signal in the calcied zone, however in reality we are unable to conrm this element in mineralized tissue. This effect is caused again by intense Ca signal, in this case false-positive tting of K occurs in tail of high Ca-Ka peak.
For the interpretation of the results, it is important to note, that in the mXRF maps the depth resolution is elementdependent, and (in case of thin sections discussed in present paper) oen is the representation of the signal from the entire sample thickness. In cortical bone, for example, it can be estimated as 9 mm for P; 39 mm for Ca and 98 mm for Zn, using the mass attenuation coefficients for cortical bone (m/r) 29 and the theoretical density for cortical bone of 1.92 g cm À3 . 30 This is the reason why maximum Ca value in 41 mm slice is about 50% higher than in the 15 mm slice. Further, the information depth will be different for bone and hyaline cartilage, as it depends strongly on the actual material composition. An approximation for the information depth values in cartilage can be obtained using the mass attenuation coefficients for so tissue 29 and the theoretical density for cartilage of 1.1 g cm À3 (ref. 30) as 16 mm for P, 24 mm for S, 50 mm for K, 111 mm for Ca and 872 mm for Zn. Quantication of cartilaginous zone by using IAEA Animal Bone (H-5) certied reference material (for mXRF, see Materials and methods) is, therefore, not optimal. Unfortunately, cartilage certied reference material is not commercially available, and, thus, a well-characterized homogenous standard material (optionally doped with specic elements) will be benecial in future measurements.

LA-ICP-MS
Major elements (Ca, Mg, Na, K, P, S) as well as minor/trace elements (Co, Cr, Fe, Mn, Zn) expected to be found in a sample were included in the LA-ICP-MS analysis. Although C is measurable with ICP-MS, it was excluded from the analysis since the C-rich embedding medium would falsify the analysis. For each analyte, one isotope was chosen using absence of potential (isobaric and polyatomic) interferences as well as natural abundance as selection criteria. We assume that the information depth of the LA-ICP-MS measurements is about 5 mm, it is similar for both slices independently of their thickness, and only varies among different histological areas, i.e. mineralized tissues and hyaline cartilage. 50 Cr and 55 Mn did not provide sufficient signal-to-noise ratios for the generation of elemental maps and were therefore excluded from further analysis. 56 Fe and 59 Co show signicant signal intensities in some sample areas, however, since they most likely originate from contaminations due to sample preparation they were excluded from further analysis as well. The elemental maps of the remaining analytes were quantied via external calibration using the measured IAEA Animal Bone (H-5) certied reference material as well as the self-made porcine cartilage standard. To keep matrix effects as small as possible, those analytes predominately in the calcied area ( 42 Ca, 24 Mg, 23 Na, 31 P, 64 Zn) were quantied using the bone standard, while the cartilage standard was used for those analytes predominately in the hyaline cartilage area ( 34 S, 39 K). This means that the analysis is not perfectly matrix-matched for the tissue in which the analyte shows lower concentration (e.g. Ca in hyaline cartilage tissue). However, since the presence of most elements is rather limited to one kind of tissue (e.g. Ca in mineralized tissue), the resulting error is hardly visible in the elemental maps and therefore negligible. Image quantication was performed without internal standard correction since no element suitable as internal standard (i.e. constant concentration in whole sample) was available. Finding a way for utilizing internal standardization to compensate differences in material ablation and instrumental dris could potentially further improve the accuracy of the analysis. The LA-ICP-MS elemental maps (cf. Fig. 2, 3, Tables 2 and 3) show high Ca and P concentrations (ca. 170 mg g À1 and 75 mg g À1 , respectively) as well as signicant Na, Mg, and Zn content (ca. 3 mg g À1 , 1.3 mg g À1 , and 140 mg g À1 , respectively) in the calcied zone of the sample. In case of Zn, there is a signicant enrichment in the tidemark zone (up to ca. 400 mg g À1 ) visible. In case of Na, the map of the 41 mm slice shows also a relatively high concentration in the hyaline cartilage area, which cannot be observed for the 15 mm sample. Since a preablation step was performed for the later one (for cleaning purposes; see Experimental), the Na measured in the hyaline cartilage tissue of the 41 mm slice probably originates from surface contaminations.
For the areas consisting of hyaline cartilage tissue, a relatively high S content (up to 12 mg g À1 ) can be observed. Despite S being a main constituent, the image quality is rather poor, which is caused by the relatively high ionization energy of S, the fact that the main S isotope ( 32 S) cannot be measured due to polyatomic interferences (O 2 + species), and a relatively high Sbackground caused by the applied ablation chamber and tubing system. Beside S also K can predominately be observed in the hyaline cartilage tissue (0.2-1.5 mg g À1 ). Even though the measurement of K suffers from a high background due to interferences originating from the plasma gas ( 38 Ar 1 H + ), the quality of the obtained images is sufficient.

Method correlation
Since every analytical technique has certain requirements to the samples, the rst challenge was to choose a sample preparation approach compatible with both imaging methods. In our experiments, PMMA embedding resulted in self-supporting thin sections, which allowed direct analysis with mXRF; and acrylic glass frames ensured that the sample is attened during the measurement. Aer the non-destructive mXRF measurements, an intact sample was transferred to LA-ICP-MS instrument. For LA-ICP-MS measurements sample was glued onto a plastic slide using sticky tape on the edges of the samplesimilarly, to straighten and x the sample horizontally; no other sample handling manipulations were required. While the method of PMMA embedding is routinely used in bone samples preparation, it is worth mentioning, that the preservation of hyaline cartilage in the same way might not be optimal and some elements (diffusible ions, like Na + and K + ) might have been washed out or relocated. 31 Further investigations and more rened sample preparation techniques (e.g. cryoxation/ freeze-drying) are needed to obtain the information on chemical distribution within the non-mineralized cartilage in nearnatural state. Due to the non-destructive character of mXRF, these measurements were performed rst. An advantage of this experimental order is that mXRF results provide guidance for the elements of interest in subsequent LA-ICP-MS analysis. In our experiments we rst focused on the expected major elements, which could be measured by both mXRF and LA-ICP-MS. In addition LA-ICP-MS provided the images for light metals -Na and Mg, which could not be imaged by laboratory mXRF due to limitations of the measuring system (e.g. detector window transmission).
We obtained quantitative maps with both methods, the employed quantication approach resulted in the concentrations in the same order of magnitude for the investigated elements imaged by both mXRF and LA-ICP-MS. To directly compare the elemental concentrations obtained by mXRF and LA-ICP-MS within the very same areas imaged by both techniques, we subdivided the maps shown in Fig. 2 and 3 in two subsectionsmineralized tissue (selection criteria: Ca value in mXRF map > 100 mg g À1 ) and hyaline cartilage (selection criteria: S value in mXRF map >10 mg g À1 ). The averaged values and standard deviation for the elements within these subsections are presented in Table 2 (15 mm slice) and Table 3 (41 mm  slice).
Generally for mXRF, it should be noted that the excited/ detected sample mass in 15 mm is lower than in 41 mm, which leads to different concentration values. As LA-ICP-MS is not affected by sample thickness, similar values are observed for 15 mm and 41 mm slices, except for Kwhich is notably higher in the 41 mm slice. Specically for K, it should be emphasized, that this element could be detected by mXRF only in the 41 mm slice in cartilagehere we assume the inuence of larger sampling volume in thicker sample. The mXRF mean cartilage value for potassium corresponds to the potassium content reported by LA-ICP-MS.
Since S maps are more detailed in mXRF measurements, we conclude that LA-ICP-MS is less sensitive to S, most likely due to the previously mentioned high S-background (cf. the estimated detection limits included in ESI, Table S5 †). In case of Zn, LA-ICP-MS provided more detailed distribution maps for both samplesmore features can be seen in slice 15 mm in  comparison to mXRF; in slice 41 mm the histological structures are clearer, due to higher depth resolution of LA-ICP-MS.
To further evaluate the agreement between the images obtained by mXRF and LA-ICP-MS, we calculated the correlation between the corresponding elemental maps on Fig. 2 and 3. We used the Pearson correlation coefficient for the evaluation. It assumes values in the range from À1 to +1, where 1 indicates strongest possible agreement, 0 strongest possible disagreement, and À1 perfect negative correlation. Table 4 displays the result -Pearson correlation coefficients calculated for each analyte obtained by mXRF and LA-ICP-MS. Overall, lower coef-cients were obtained for 15 mm compared to 41 mm sample. In case of Ca and P, the method shows excellent agreement with high correlation coefficients for both samples. Also for S and Zn strong correlations can be observed, especially in 41 mm slice. The only analyte showing poor correlation is K. The reason for that is imperfect mXRF spectra tting, leading to false-positive K distribution in calcied zone (see above), ultimately resulting in mismatched elemental map and, consequently, poor correlation. With the exception of K, the correlation analysis conrms that the mXRF and LA-ICP-MS images are in good agreement.
Although many studies have been published on elemental distribution analyses of human bone with XRF-based methods, we were unable to nd reports on quantitative mapping (absolute quantication). Papers on quantitative imaging with use of LA-ICP-MS exist, albeit limited to archaeological objects. 32,33 Quantitative elemental analysis of bone will be of interest for the diagnostic reasons, for example, investigation of Sr incorporation from strontium ranelate with implications for the quality of the mineral crystal; 34 quantication of trace element accumulations (Zn) in the tumorous tissue 3 or estimation of metal transfer from orthopaedic implants. 35 We have recently reported accumulations of elements-xenobiotics, e.g. lead in tidemark 8 and gadolinium in bone, 36 and quantication could help in assessing the risk of potential re-release of those elements in bloodstream.
As for rather neglected area of cartilage analysis, most of the elemental imaging experiments with LA-ICP-MS or mXRF found in literature were performed qualitatively on the animal tissues. This limits the direct comparison to our results; nonetheless, these published ndings point out other chemical elements, which should be considered in future measurements. Sussulini et al. imaged bovine cartilage by means of LA-ICP-MS, and in addition to Na, Mg, P, S, K, Ca, and Zn, also C, Cl, Ti and Cu were detected in articular cartilage tissues. 37 Whereas C and Cl are expected as intrinsic components of biological tissues, Ti is normally not present in cartilage and in our sample (41 mm) was considered as contamination. Cu, however, is an essential element and was also detected and imaged in dried droplet of synovial uid of osteoarthritic patient by Austin et al. 38 The authors also report distribution of Ca, P, Mg, Sr and Zn in a knee cartilage section of OA patient. Notably, Austin et al. used thin samples for the experimentsdried 14 mm sections, without any other treatmentwhich seemingly resulted in folding and twisting of the tissue. In our study, we observed only few cartilage folds in the thinner, 15 mm sample, otherwise PMMA embedding facilitated preserving the overall shape of the tissue.
Reinert et al. obtained distributions of P, S, Cl, K and Ca in articular cartilage of juvenile pig's knee joint using particle induced X-ray emission (PIXE). 39 They reported variations in S distribution within different histological zones, consistent with proteoglycan content, positively correlated with K, and displaying maximum concentration in hypertrophic cartilage (zone of hyaline cartilage contiguous to tidemark). Further, Reinert et al. detected calcium-phosphorus accumulation within the supercial cartilage. None of these was observed in our experiments. These dissimilarities could be attributed to the origin of the sampleanimal vs. human; or the sample preparationfreeze-drying by Reinert et al. and PMMA embedding in our case.
In future, adding histological analysis for detailed assignment of the cartilaginous zones and coupling elemental imaging with molecular (e.g. MALDI) could be benecial. Human bone sample preparation allowing preservation of biomolecules (lipids) followed by consecutive time-of-ight secondary ion mass spectrometry (TOF-SIMS) and atmospheric pressure scanning microprobe matrix assisted laser desorption/ionization (AP-SMALDI) molecular imaging have been recently presented by Schaepe et al. 40 Knowledge on distributions and content of certain elements in human tissues will further histopathological studies. For instance, the elemental imaging of bone and cartilage will be of use for investigation of Ca deposits and crystals in cartilage, as well as osteoarthritic lesions. 41,42 The presented spatial distribution of elements accompanied by absolute quantitative data is an exceptional combination in human bone and cartilage analysis, and we hope that our research will serve as a base for future studies in this area.

Conclusions
The current paper presents the analysis by two elemental imaging techniques -mXRF and LA-ICP-MSof a complex human bone-cartilage sample (femoral head). The measurements were performed in consecutive manner on the very same sample surfaces of two slices. Quantitative distribution mapping for both mXRF and LA-IPC-MS was performed with the use of matrix-matched standards, and the concentrations obtained by both methods are either similar or at least in the same order of magnitude. We report concentrations for various elements within different histological zonese.g. mineralized tissues, tidemark, hyaline cartilage. Further, we compared LA-ICP-MS and mXRF by means of correlation analysis applied to elemental maps, and observed strong (Ca, P in both samples, and S, Zn in 41 mm slice) or good (S and Zn in 15 mm slice) correlation of the methods. We believe that this is the rst holistic descriptionimaging of elemental distribution supported by quantitative dataof bone and cartilage by either of the methods. Combined sequential application of mXRF and LA-ICP-MS within one sample offers additional benets, and should be further applied in the biomedical research.

Conflicts of interest
There are no conicts of interest to declare