Silicon balance in human volunteers; a pilot study to establish the variance in silicon excretion versus intake
© Pruksa et al.; licensee BioMed Central Ltd. 2014
Received: 27 September 2013
Accepted: 27 December 2013
Published: 9 January 2014
Accumulating evidence suggests a role for silicon in optimal connective tissue health. Further proof of its importance/essentiality may be provided by studies involving imposed depletion followed by 29Si challenge to estimate metabolic balance. Prior to conducting these expensive studies, we first established the variance of estimating normal Si excretion versus intake using a single oral dose of typical dietary Si, orthosilicic acid.
Healthy volunteers were recruited from Loei Rajabhat University, separated into two matched groups (three males and three females/group) and maintained on a standardized diet for the three study days. One group ingested 500 ml water containing orthosilicic acid (28.9 mg Si) and the other group received 500 ml water alone, all on a fasted stomach. Blood samples and total urine and faeces were collected over the 48 h post-dose period and 24 h before-hand (baseline) and analysed for silicon by inductively coupled plasma optical emission spectrometry.
Serum Si analysis confirmed the ready absorption of silicon from the orthosilicic acid solution. Mean total urinary and faecal Si excretions over the 24 h post-dose period accounted for 57 ± 9.5% and 39 ± 9.4% of the ingested dose, respectively. Thus in total 96.3 ± 5.8% of the ingested dose was recovered in faecal plus urinary excretions over the 24 h post-dose period.
We report that in healthy subjects (presumably in Si balance), the ingestion of a soluble dose of dietary Si results in the same quantity (within analytical error) being excreted within 24 h. It is currently not known if this all originated from the dose solution or if there was some exchange with the body Si pool but, given the low variance in these silicon balance data, isotopic studies are now merited.
KeywordsSilicon Orthosilicic acid Absorption Balance studies Urine Faeces
Silicon is a critical element in the biology and/or survival of a number of lower life forms, including diatoms, certain sponges and many plants [1–4]. In humans and other mammals its role (if any) is less well defined despite being a common dietary trace element (20–50 mg/day is ingested by adults in western populations [5–8]). Indeed, the environmental ubiquity and limited (bio) chemistry of silicon have led to claims that its ingestion, ready absorption and excretion by mammals are all simply inevitable consequences of oral exposure to a small soluble molecule (orthosilicic acid, Si(OH)4) that ‘washes through’ the system and has no biological function [9, 10]. Against this, evidence is accumulating to suggest that, in mammals, silicon plays an important role in optimal connective tissue health [11–14]. Its exact role/function remains unestablished, but there is evidence to suggest it’s involved in the synthesis and/or stabilisation of extracellular matrix components, namely collagen, and in the proliferation of connective tissue cells [9, 14]. There is also evidence to suggest that silicon is carefully conserved when dietary deficiency is imposed [15, 16]. To translate these findings to humans, and provide more evidence for its essentiality, balance studies using Si isotope(s) following a low silicon diet may demonstrate (i) retention of silicon following ingestion and (ii) whether ingested and absorbed silicon displaces some endogenous silicon or is simply washed through.
Despite the simple form of balance studies, where quantitative faecal and urinary excretion of a substance (and/or metabolites) are compared to the amount of substance ingested, they are fraught with difficulties . Faecal collection and analysis is especially demanding: gastrointestinal transit times vary between individuals and temporary mucosal retention of a substance may also occur, adding further variability. Volunteers must provide complete collections and analysis of different fractions and sample types causes inevitable compound error.
For silicon, a basic human balance study, no matter how precise or accurate, would tell us little about the homeostasis of the element. If silicic acid is not utilised/metabolised, but is washed through the system, then a perfect study would recover 100% of the ingested dose. On the other hand, if it were utilised and/or metabolised, a 100% recovery would still be expected assuming that volunteers are themselves in balance (i.e. not deficient). Nonetheless, the value of such a study would be to determine what sort of variance one might expect in balance if this were to be attempted in subsequent isotope and/or depletion studies (i.e. are these expensive and time consuming studies worth doing?). It would also have the added value of confirming whether urinary silicon, which is typically used to estimate silicon absorption in humans, is a valid measure for this purpose.
Here we sought to determine the balance in excretion of silicon (faecal and urinary) versus intake, using a single oral dose of silicic acid (28.9 mg Si) in human volunteers on a standardized diet.
Subjects and methods
Characteristics of the study volunteers
Control group (3 M & 3 F)
Si-supplemented group (3 M & 3 F)
Mean ± SD
Mean ± SD
21.2 ± 1.9
21.0 ± 1.3
55.1 ± 7.3
56.4 ± 6.6
163 ± 10
165 ± 8
20.6 ± 1.7
20.8 ± 2.5
Serum creatinine 1 (mg/dL)
0.92 ± 0.15
0.88 ± 0.15
Baseline 24 h Si excretion:
Urine (mg/24 h)
12.48 ± 2.26
12.93 ± 2.85
Faeces (mg/24 h)
9.50 ± 1.41
9.35 ± 2.01
Glassware was avoided throughout the study to prevent Si contamination. Ultra high purity (UHP) water was from a water purifier (Labscan Asia Co Limited, Bangkok, Thailand). The stock basic sodium silicate solution was from Lakehead University, Canada (Professor Stephen Kinrade). The stock silicon ICP standard solution (1,000 mg/L Si) was from Merck Ltd (Poole, UK). Nitric acid (65% (w/v) HNO3) and hydrochloric acid (37%) were high purity from RCI Labscan Limited (Bangkok, Thailand). Polypropylene tubes (15 and 50 mL) were from Elkay Laboratory Products UK Ltd (Basingstoke, UK). Polypropylene bottles (30 and 2,000 mL) were from VWR International (Poole, UK). All intravenous catheters (1.2 × 45 mm) and plastic syringes were from Nipro Ltd (Pranakhonsriayuthaya, Thailand). Pasteur pipettes (3.5 mL), used for sample transfer, were from Greiner Bio-One Limited (Stonehouse, UK). Pipette tips (100–1,000 μL) were from Hycon (Biomed C. Ltd., Bangkok, Thailand).
Preparation of Si supplement
The Si supplement (orthosilicic acid solution, OSA) was prepared fresh, just prior to ingestion, by dilution of the stock basic sodium silicate solution (1.58 mol Si/L or 45.72 g Si/L) into UHP water and pH neutralization to 7.2 with HCl. The Si concentration in the test solution (2.06 mmol/L or 57.78 mg/L) was confirmed by inductively coupled plasma – optical emission spectrometry (ICP-OES; Perkin Optima, model 2100 DV).
Estimated silicon content 1 of the meals (mg Si/meal) provided to all participants during the study period
Day 0 3 (mg Si)
Day 1 (mg Si)
Day 2 (mg Si)
Egg fried rice with pork/chicken, orange juice, drinking water 2
Rice with chicken, drinking water 2 , yogurt, sponge cake
Rice with stir-fried pork/chicken with kale, melon, drinking water 2
Total Si intake per day
Blood samples were collected in 15 mL polypropylene tubes and left to stand for at least 1 h at room temperature to clot. The clotted blood samples were then centrifuged (Hettich Zentrifugen, Tuttlingen, Germany) at 3,000 rpm for 10 min at room temperature. The separated serum fractions were collected into new 15 mL polypropylene transport tubes and stored at -20°C until elemental analysis. Prior to analysis, the serum samples were thawed at room temperature (23°C) and then diluted 1 + 4 with 0.25% (v/v) high purity HNO3.
Urine collections were weighed and volumes calculated assuming a density of 1. After thorough mixing, a 10 mL homogenous sample was collected into a 30 mL polypropylene bottle and diluted with equi-volume 0.7% (v/v) high purity HNO3 (i.e. 1 + 1 dilution) to reduce any precipitation during storage . The diluted samples were stored at 4°C until elemental analysis. Prior to analysis, the diluted samples were incubated overnight in their closed containers at 40°C in an oven to dissolve any precipitates that may have formed during storage . Samples were cooled to room temperature prior to total elemental analysis for Si.
Faecal collections were weighed and after thorough manual mixing with a disposable wooden spatula, a homogenous sample (from each collection) was collected into a 50 mL polypropylene bottle and stored at -20°C. Prior to analysis, approximately 0.25-0.5 g of the faecal samples was digested with an equi-volume (5 mL) of concentrated (65% (w/v)) HNO3 and hydrogen peroxide (30-40%) at room temperature for 24 h. These were incubated at 40°C until total digestion was obtained. Sample (acid) blanks were similarly prepared. An aliquot (1 mL) of the digested samples and sample blanks were diluted with 5 mL UHP water prior to total element analysis for Si.
Total elemental analysis
Total elemental analysis for Si was carried out (at 251.611 nm) by ICP-OES; Perkin Elmer Optima model 2100 DV, equipped with a Cross flow nebuliser and Cyclonic spray chamber. Nebulizer flow rate was 0.8 L/min. Peak area was 7.7 points and integration times were 20 seconds/analysis/element. Sample flow rate was 2 mL/min. Matrix matched standards, sample blanks, diluents and quality control samples were run alongside the samples.
Serum, urine and faecal samples
The diluted serum, urine, and faecal samples from the same subject were analyzed together in the same batch. Sample-based standards were prepared in the pooled diluted sera, baseline urine, or baseline faecal samples using the 1,000 mg/L Si ICP standard solution.
The Si supplement solution, UHP water and drinking water (part of standardized meals) were also analyzed for Si by ICP-OES using appropriate standards.
As the serum and urine samples were diluted with 0.25-0.7% HNO3, Si content of the acid diluents was also measured by ICP-OES. Although minor, this contaminant Si was subtracted from each of the sample data.
Sample size (power) calculation was based on the available data on urinary Si excretion . No previous data exist for faecal Si excretion. A relative standard deviation (σ) of 9.4%  was estimated for the variance in urinary Si and a potential difference of 20% for the excretion of urinary Si between the Si supplement and water test solutions was assumed, with 90% power at a 5% significance level. Sample size formula for the difference of two independent means was used for the calculation and six completed subjects were the minimum required for each test solution.
Area under the curve (AUC) of serum Si was calculated using the linear trapezoidal rule . Due to a small number of subjects in each group, differences in serum AUC, and in urinary and faecal excretions of Si, between the two groups (Si vs. control), were analysed non-parametrically using the Mann–Whitney Rank test. Statistical analyses were two sided and a P value ≤ 0.05 was considered significant. SPSS for Windows version 13.0 (SPSS Inc., Chicago, Illinois, USA) was used for all statistical analyses.
Gender specific analysis showed no significant difference in silicon absorption, excretion (urinary and fecal) and balance between male and female subjects and so the combined (pooled) dataset is shown for clarity and because of the small number of subjects.
Serum Si absorption
Urinary Si excretion
Mean total baseline 24 h urinary Si excretion was 12.48 ± 2.26 (range 9.72 – 15.31) and 12.93 ± 2.85 (range 9.11 – 16.34) mg, respectively, in the control and Si-supplemented groups and did not change markedly in the control group following ingestion of UHP water (13.34 ± 2.48 mg; range 11.36 – 18.21 mg). In contrast, in the Si-supplemented group, ingestion of the orthosilicic acid solution (28.9 mg Si in UHP water) led to a marked increase in urinary excretion of Si in the 0–12 h post-dose collection (P = 0.002; Figure 2B). A more detailed analysis of the 0–12 h collection in the Si-supplemented group showed that the peak increase in Si output was at 0–3 h post-dose (Figure 2C). Again, the ingestion of meals at 4.5 h and 8.5 h post-dose maintained urinary Si output above baseline in the 6–9 h and 9–12 h post-dose collections, as clearly mirrored in the control group (Figure 1C). Urinary Si output in the remaining collections (12–24 h, 24–36 h and 36–48 h), were comparable to baseline levels and similar between the two groups.
The increase in urinary Si output in the Si-supplemented group over the 24 h post-dose period following ingestion of the orthosilicic acid solution (i.e. over and above baseline urinary Si excretion) was 16.5 ± 2.7 mg (range 14.4 – 20.5 mg) and this accounted for 57.0 ± 9.5% (range 49.8 – 71.0%) of the supplemental Si dose (28.9 mg) ingested.
Faecal Si excretion
Silicon intake, excretion, and balance over 24 h period (study day 1)
Control group (n = 6)
Si-supplemented group (n = 6)
Mean ± SD
Mean ± SD
Si supplement (mg)
Dietary Si intake (mg)
Total Si intake (mg/24 h)
Urinary Si (mg/24 h)
13.34 ± 2.48
29.41 ± 3.55
Faecal Si (mg/24 h)
10.74 ± 1.41
19.96 ± 4.58
Total excretion (mg/24 h)
23.77 ± 1.94
50.02 ± 4.98
Silicon balance (mg)
0.09 ± 1.94
2.73 ± 4.98
The present study investigated the balance in excretion of silicon versus its intake, using a single dose of typical dietary silicon (28.9 mg) in healthy human volunteers on a standardized diet. Our results show that, within error, all (96 ± 6%) of the ingested dose was excreted in urine and faeces over the 24 h post-dose period. Whether it is, fully, the same Si being excreted that is being absorbed will need to be addressed with a different study design. However, to get within 5% of intake with variance of ~ 6% is better than may have been expected, especially with the complexities of faecal and urine collection and analysis [17, 22]. Typical recovery from such studies, even with radiolabelled compounds, can be 80% or less, much less than from animal studies [23, 24]. The high renal clearance of Si and the lack of interaction with serum proteins probably aids recovery [25, 26].
As mentioned previously, 100% recovery is expected if (a) Si metabolism is regulated but the subjects are in Si balance (i.e. are Si replete) or (b) if Si has no active biological function and thus Si metabolism is not regulated at all. This study cannot prove which is true but we believe that the former is more likely based on previous murine data where urinary Si output was found to be conserved in Si-deprived animals to maintain tissue Si levels . To now show this in humans we will need to repeat the study with subjects who are in negative Si balance (i.e. Si deplete at the start of the supplementation period by prior dietary Si deprivation for a week or so). Supplementation with the Si dose should then result in less Si being excreted, as more is retained to replenish the depleted body Si pool, compared to a Si-replete group.
Secondly, from this current work, we cannot be certain that the Si excreted in urine and faeces all originated from the ingested Si dose and that there was not some exchange with the body Si pool- as for example occurs with dietary phosphate . This can only be answered with an isotope label study, where isotopic Si is used for the dose solution to discriminate it from Si of the body pool and from dietary sources (i.e. the meals ingested). However this is also not straight forward. 31Si and 32Si are both radioactive and would result in exposure to radiation (beta decay) with short (t½ = 157 min) or long (t½ = 153 years) half-lives, respectively. Using a stable isotope such as 29Si would avoid radioactive exposure but it has a high natural abundance (ca. 5% of all endogenous Si(OH)4). Hence, a relatively accurate balance, as now proven is possible in this study, will be key to the success of the follow on stable isotope work. Moreover, with recent developments in inductively coupled plasma – mass spectrometry methods, to measure 28Si and 29Si in biological samples , we are confident that it will now be possible to discriminate the source of excreted Si (i.e. all ‘washed through’ following ingestion or some from the body pool following exchange with absorbed silicon). Both this question and that of Si retention following oral Si challenge to Si-depleted volunteers are big questions in human Si metabolism and this study proves that they may now be answered with carefully designed isotopic balance studies in Si replete and deplete individuals. In the work presented herein subjects were carefully matched to reduce variability between the two groups, however a cross-over design is undoubtedly more robust to really minimise inter-individual variation in silicon handling [29, 30]. Thus, although more burdensome to the subjects, for the future work a cross-over study design will be seriously considered.
Finally, measurement of faecal Si excretion for the first time in a human study, as we report here, allowed the absolute absorption of Si from the orthosilicic acid dose solution to be estimated which, at 61 ± 9% of the ingested dose, is similar to the estimate from total urinary Si output over the 24 h collection period (57 ± 10%). These estimates are comparable with previous data (absorption being stated as ~ 50–60% of the ingested dose) from shorter urine collections, 0–6 or 0–8 h post-dose [20, 25, 29–34]. Hence, we can conclude that (a) urinary silicon does measure silicon absorption following oral Si challenge and (b) in general 0–6 or 0–8 h urinary collections are adequate to estimate absorption/bioavailability of Si from readily absorbed dietary sources, supplements and test solutions (Si materials requiring prolonged digestion prior to absorption may differ in this respect as previously noted .
In conclusion the present study reports that urinary and faecal Si excretion can be measured with high precision (with inter-subject variance ~ 6%) and that in normal healthy subjects who are presumed to be in Si balance, ingestion of a soluble dose of Si results in an equivalent quantity being excreted within 24 h. We confirm that urinary silicon may be used as an accurate measure of silicon absorption, assuming robust study design. This work also provides clear evidence that a proper isotope-based balance study can now be undertaken in silicon replete and depleted volunteers to inform on homeostasis of silicon and thus provide strong evidence (or not) for beneficial utilisation of the element by humans.
Area under the (serum) curve
Inductively coupled plasma-optical emission spectrometry
This study was supported by the Office of Higher Education Commission, Thailand Research Fund and core institutional funds: Medical Research Council (grant number MC_US_A090_0008/Unit Programme number U1059).
- Hildebrand M, Volcani BE, Gassman W, Schroeder JI: A gene family of silicon transporters. Nature. 1997, 385: 688-689.View Article
- Uriz MJ, Turon X, Becerro MA: Silica deposition in Demosponges: spiculogenesis in Crambecrambe. Cell Tissue Res. 2000, 301: 299-309.View Article
- Epstein E: Silicon. Annu Rev Plant Physiol Plant Mol Biol. 1999, 50: 641-664.View Article
- Ma JF, Yamaji N, Mitani-Ueno N: Transport of silicon from roots to panicles in plants. Proc Jpn Acad Ser B Phys Sci. 2011, 87: 377-385.View Article
- Pennington JA: Silicon in foods and diets. Food Addit Contam. 1991, 8: 97-118.View Article
- Van Dyck K, Van Cauwenbergh R, Robberecht H, Deelstra H: Bioavailability of silicon from food and food supplements. Fresenius J Anal Chem. 1999, 363: 541-544.View Article
- Jugdaohsingh R, Anderson SH, Tucker KL, Elliott H, Kiel DP, Thompson RP, Powell JJ: Dietary silicon intake and absorption. Am J Clin Nutr. 2002, 75: 887-893.
- McNaughton SA, Bolton-Smith C, Mishra GD, Jugdaohsingh R, Powell JJ: Dietary silicon intake in post-menopausal women. Br J Nutr. 2005, 94: 813-817.View Article
- Jugdaohsingh R, Kinrade SD, Powell JJ: Is There a Biochemical Role for Silicon. Metal Ions in Biology and Medicine, vol X. Edited by: Collery P, Brätter P, Negretti De Brätter V, Khassanova L, Etienne JC. 2008, Paris: John Libbey Euro Text, 45-55.
- Nielsen FH: Nutritional requirements for boron, silicon, vanadium, nickel, and arsenic: current knowledge and speculation. FASEB J. 1991, 5: 2661-2667.
- Carlisle EM: Silicon: an essential element for the chick. Science. 1972, 178: 619-621.View Article
- Schwarz K, Milne DB: Growth-promoting effects of silicon in rats. Nature. 1972, 239: 333-334.View Article
- Jugdaohsingh R, Tucker KL, Qiao N, Cupples LA, Kiel DP, Powell JJ: Dietary silicon intake is positively associated with bone mineral density in men and premenopausal women of the Framingham Offspring cohort. J Bone Miner Res. 2004, 19: 297-307.View Article
- Jugdaohsingh R: Silicon and bone health. J Nutr Health Aging. 2007, 11: 99-110.
- Jugdaohsingh R, Calomme MR, Robinson K, Nielsen F, Anderson SHC, D’Hease P, Geutsens P, Loveridge N, Thompson RPH, Powell JJ: Increased longitudinal growth in rats on a silicon-depleted diet. Bone. 2008, 43: 596-606.View Article
- Jugdaohsingh R, Calomme MR, Robinson K, Nielsen F, Anderson SHC, D’Hease P, Geutsens P, Loveridge N, Thompson RPH, Powell JJ: Letter to the Editor: reply to Dr Exley comments on: Jugdaohsingh R et al., increased longitudinal growth in rats on a silicon depleted diet. Bone. 2009, 44: 1019-1020.View Article
- Mertz W: Use and misuse of balance studies. J Nutr. 1987, 117: 1811-1813.
- Powell JJ, McNaughton SA, Jugdaohsingh R, Anderson SH, Dear J, Khot F, Mowatt L, Gleason KL, Sykes M, Thompson RPH, Bolton-Smith C, Hodson MJ: A provisional database for the silicon content of foods in the United Kingdom. Br J Nutr. 2005, 94: 804-812.View Article
- Burden TJ, Powell JJ, Taylor PD, Thompson RP: Optical accuracy, precision and sensitivity of inductively coupled plasma optical emission spectrometry: bioanalysis of aluminium. J Anal At Spectrom. 1995, 10: 259-266.View Article
- Sripanyakorn S, Jugdaohsingh R, Elliott H, Walker C, Mehta P, Shoukru S, Thompson RP, Powell JJ: The silicon content of beer and its bioavailability in healthy volunteers. Br J Nutr. 2004, 91: 403-409.View Article
- Abramowitz M, Stegun IA: Handbook of mathematical functions. 1972, New York: Dover Publications
- Jackson MJ, Jones DA, Edwards RH, Swainbank IG, Coleman ML: Zinc homeostasis in man: studies using a new stable isotope-dilution technique. Br J Nutr. 1984, 51: 199-208.View Article
- Roffey SJ, Obach RS, Gedge JI, Smith DA: What is the objective of the mass balance study? A retrospective analysis of data in animal and human excretion studies employing radiolabeled drugs. Drug Metab Rev. 2007, 39: 17-43.View Article
- Palacios C, Wigertz K, Braun M, Martin BR, McCabe GP, McCabe L, Pratt JH, Peacock M, Weaver CM: Magnesium retention from metabolic-balance studies in female adolescents: impact or race, dietary salt, and calcium. Am J Clin Nutr. 2013, 97: 1014-1019.View Article
- Reffitt DM, Jugdaohsingh R, Thompson RP, Powell JJ: Silicic acid: its gastrointestinal uptake and urinary excretion in man and effects on aluminium excretion. J Inorg Biochem. 1999, 76: 141-147.View Article
- D’Haese PC, Shaheen FA, Huraid SO, Djukanovic L, Polenakovic MH, Spasovski G, Shikole A, Schurgers ML, Daneels RF, Lamberts LV, Van Landeghem GF, De Broe ME: Increased silicon levels in dialysis patients due to high silicon content in the drinking water, inadequate water treatment procedures, and concentrate contamination: a multicentre study. Nephrol Dial Transplant. 1995, 10: 1838-1844.
- Berndt T, Kumar R: Phosphatonins and the regulation of phosphate homeostasis. Annu Rev Physiol. 2007, 69: 341-359.View Article
- Koller D, Ratcliffe S, Jugdaohsingh R, Powell JJ, Bruggraber SF: Optimisation of Operative Conditions for the Determination of Silicon Concentrations in in Vitro and ex Vivo Biological Matrices by DRC-ICP-MS. 2013, Krakow, Poland: Winter Conference on Plasma Spectrochemistry
- Jugdaohsingh R, Reffitt DM, Oldham C, Day JP, Fifield LK, Thompson RPH, Powell JJ: Oligomeric but not monomeric silica prevents aluminium absorption in human. Amer J Clin Nutr. 2000, 71: 944-949.
- Jugdaohsingh R, Hui M, Anderson SH, Kinrade SD, Powell JJ: The silicon supplement ‘Monomethylsilanetriol’ is safe and increases the body pool of silicon in healthy pre-menopausal women. Nutr Metab (Lond). 2013, 10: 37-View Article
- Jugdaohsingh R, Sripanyakorn S, Powell JJ: Silicon absorption and excretion is independent of age and sex in adults. Br J Nutr. 2013, 25: 1-7. [Epub ahead of print]
- Sripanyakorn S, Jugdaohsingh R, Dissayabutr W, Anderson SHC, Thompson RPH, Powell JJ: The comparative absorption of silicon from different foods and food supplements. Br J Nutr. 2009, 102: 825-834.View Article
- Bellia JP, Birchall JD, Roberts NB: The role of silicic acid in the renal excretion of aluminium. Ann Clin Lab Sci. 1996, 26: 227-233.
- Calomme MR, Cos P, D’Haese PC, Vingerhoets R, Lamberts LV, De Broe ME, Van Hoorebeke C, VandenBerghe DA: Absorption of Silicon in Healthy Subjects. Metal Ions in Biology and Medicine, vol V. Edited by: Collery P, Brätter P, Negretti De Brätter V, Khassanova L, Etienne JC. 1998, Paris: John Libbey Euro Text, 228-232.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.