DNA from Nails for Genetic Analyses in Large-Scale Epidemiologic Studies

Nails contain genomic DNA that can be used for genetic analyses, which is important for large epidemiologic studies that have collected nail clippings at baseline and for future epidemiologic studies that consider collecting nails as a DNA source for genetic analyses. The DNA in nail clippings (keratinous tissue) originates from germinal matrix cells in the nail root. During nail growth, these cells differentiate into nail plate and are filled with keratin. In this keratinization phase, the cells undergo programmed cell death, which results in considerable DNA fragmentation (1). Once the nail root cells are keratinized (and the DNA fragmented) during nail growth, the keratin tissue probably protects the DNA from further damage because keratinous tissue makes the DNA less accessible to oxidants and does not contain water. Water in samples leads to DNA damage through hydrolytic deamination of cytosine (1).

Nails are easy and cheap to obtain and store compared with other tissues such as blood or saliva samples. Moreover, study participants are likely to be more willing to participate in a study in which they are asked to provide nail samples than when they are asked to provide a blood sample.

In a previous publication, we described the use of toenails as a source of genomic DNA for genotyping a limited set of 10 single-nucleotide polymorphisms (SNP; ref. 2). We concluded that DNA could be successfully isolated from 20-year-old toenail material from the Netherlands Cohort Study on diet and cancer (NLCS) at a higher success rate compared with buccal swabs, and that we could successfully genotype 10 SNPs simultaneously, by means of multiplex PCR amplification and single base extension. Importantly, 20-year-old toenail DNA performed just as well as DNA from freshly clipped nails, as expected because of the keratinous tissue protecting the DNA from further damage.

Since our previous article on DNA from toenails, we have explored the further possibilities for using toenail DNA for other genetic analyses. In the era of genome-wide association studies, 10 SNPs are a limited number and it is important to know how well toenail DNA performs in arrays focused on determining many genetic markers at once.

In the present article, we report on the yield and purity of DNA from nails, the application of toenail-derived DNA in the Sequenom MassARRAY iPLEX for the genotyping of SNPs (3), and the application of toenail-derived DNA in the HumanCytoSNP_FFPE-12 DNA Array (>262,000 genetic markers) after restoration with the Infinium HD FFPE DNA Restore Kit (4). We discuss our findings together with findings from other studies on nail DNA and we compare nails as a source of DNA to other frequently used sources of DNA.

The toenails used for the studies described in this article originate from NLCS. The NLCS is a prospective cohort study, which started in 1986 and includes 120,852 subjects ages 55 to 69 years (5). At baseline (1986), the cohort members completed a self-administered food-frequency questionnaire on dietary habits and other risk factors for cancer and approximately 90,000 (∼75%) participants provided toenail clippings in a small paper envelope, which were collected initially with the purpose of determining selenium status. The envelopes with the clippings have since been stored (>25 years) in a basement with no climate control, without any further treatment of the nails.

For the isolation of the DNA used in all the pilot studies described in this article, the DNA extraction protocol of Cline and colleagues (6) was chosen as the standard protocol, with some adjustments (2). To remove possible nail polish, nails (15 mg) were twice soaked in acetone for 10 minutes and dried. Exogenous material was removed by incubating the nails in 200 μL 1% SDS/25 mmol/L EDTA and 1 μL 20 mg/mL proteinase K for 1 hour. The nails were then rinsed three times in MilliQ and incubated in 200 μL of 2 mol/L NaOH on an automated vortex overnight to dissolve the nail material. The solution was neutralized by adding 100 μL of 200 mmol/L Tris (pH, 7–8) and 34.5 μL of 11.6 mol/L HCl. Diluted NaOH was added to redissolve the precipitates. Next, an equal volume of phenol/chloroform was added to the neutralized sample, mixed for 10 minutes by inversion, and centrifuged at 19,000 ×g for 15 minutes at 4°C. The aqueous top layer was transferred to a clean tube and DNA was precipitated by adding 1/10 volume (35 μL) of 3 mol/L sodium acetate (pH, 5.2), 2 volumes (700 μL) of 95% ethanol and 1 μL 20 mg/mL glycogen overnight at −20°C. DNA was pelleted by centrifugation at 19,000 ×g for 30 minutes at 4°C, washed once in 700 μL 75% ethanol, centrifuged at 19,000 ×g for 15 minutes at 4°C, after which the supernatant was removed and the pellet dried and resuspended in 100 μL 10 mmol/L Tris–HCl (pH, 8).

Currently, we have isolated DNA from toenail clippings of more than 21,000 NLCS study participants, among which are samples of controls (n = 3,856) and of various cancer case groups with 20.3 years of follow-up: head and neck (n = 507), esophageal (n = 359), stomach (n = 795), colorectal (n = 3,560), pancreatic (n = 544), lung (n = 2,593), melanoma (isolation in progress), breast (n women = 2,512; n men = 31), endometrial (n = 434), ovarian (n = 400), advanced prostate (n = 1,016), invasive bladder (n = 845), renal cell (n = 585), and brain (isolation in progress) cancers, and lymphatic malignancies (isolation in progress).

DNA yield and purity of the toenail samples was assessed using the NanoDrop 1000 Spectrophotometer (Thermo Fisher Scientific). The average yield of DNA obtained from the toenails using the protocol described above was 393 ± 873 ng/mg toenail (average ± SD). It should be noted that the spectrophotometrically assessed amount of DNA is not only human DNA, because bacterial and fungal DNA may be present in the nail material. In addition, phenol used for isolation of the DNA has its absorption peak at 270 nm, which, to some extent, carries over to absorbance at 260 nm, and also the absorption peak of proteins at 280 nm overlaps with absorption at 260 nm. Thus, the amount of true human DNA in the toenail DNA samples may be overestimated to varying degrees when assessed with the NanoDrop.

The average A260/A280 ratio of the toenail DNA samples was 1.5 ± 0.2, which is lower than the ratio of pure double-stranded DNA (i.e., ∼1.8). This could be due to contamination of the samples with proteins (e.g., keratin) and phenol.

The average A260/A230 ratio of the samples was 0.7 ± 0.5, substantially lower than the ratio of pure double-stranded DNA, which is ∼2.0. This could indicate that the toenail DNA samples are contaminated with organic compounds and salts used for DNA isolation such as EDTA, sodium acetate, ethanol, and glycogen. We have tried to purify the isolated toenail DNA samples using GlycoBlue Coprecipitant, but this led to considerable losses of DNA. As shown later in this article, toenail DNA rendered good results in the genotyping assays that we have applied so far and, therefore, we have not yet studied other ways to purify the toenail DNA with high DNA recovery.

Other studies have also determined DNA yields from nails, but their protocols for DNA extraction and techniques to quantify DNA yield were different from those that we used. Klassen and colleagues (7) obtained a yield of 9.9 ± 1.7 ng/mg fingernail DNA, using a fluorescence method to quantify DNA, which specifically assesses double-stranded DNA. Using a commercial DNA extraction kit, Park and colleagues (8) obtained a yield of 68.7 ng/mg finger or toenail using NanoDrop to quantify DNA, but only 2.9 ng/mg nail when using PicoGreen. When they used NanoDrop, fingernails had almost two times more DNA yield than toenails, but with PicoGreen, the yield of toenails was slightly higher than that of fingernails. Also using a commercial DNA isolation kit, Tanigawara and colleagues (9) obtained 25 to 50 ng/mg fingernail as assessed spectrophotometrically. Using a protease derived from musk melons, Yoshida-Yamamoto and colleagues (10) achieved yields up to 250 ng/mg fingernail using phenol–chloroform extraction and NanoDrop to quantify DNA, which is comparable with what we observed. The average A260/A280 ratio of their melon-based protease-treated DNA samples was slightly better than the ratio of our samples, namely 1.7. This can be explained by the fact that the protease from the musk melon has a higher keratinolytic activity than proteinase K (10). When these authors used proteinase K to digest protein, they achieved a DNA yield of 60 ng/mg nail, which corresponds exactly to the yield that Kaneshige and colleagues (11) had for fingernails also using proteinase K digestion. Nakashima and colleagues (12) obtained 100 ng DNA per mg of nail using crushed fingernail clippings. Matsuzawa and colleagues (13) obtained enough DNA from a 1 × 10 mm piece of fingernail to do at least 500 PCR reactions requiring 5 ng each using finely crushed fingernail clippings and a commercial DNA extraction kit.

In conclusion, the yield of DNA from nails seems to vary according to the protocol by which DNA was isolated. The method that we have used, as described in this article, seems to have the highest DNA yield. In addition, DNA yield varies according to the method by which it was quantified. Most studies have used NanoDrop and the yield across the studies varied from 60 to 393 ng/mg nail. PicoGreen only measures double-stranded DNA, whereas single-stranded DNA is also suitable for PCR and genotyping. Both NanoDrop and PicoGreen are not specific for human DNA, and NanoDrop in addition measures RNA, nucleotides, and protein at 260 nm.

Therefore, we recently tested the amplification of a 596-bp sequence in the cytochrome P4501A2 (CYP1A2) gene and a 197-bp sequence in the myeloperoxidase (MPO) gene in a qPCR analysis in 30 toenail samples. Input toenail DNA was 2 ng as determined by NanoDrop. Human high-molecular weight DNA was obtained from A549 cells and a calibration curve was obtained by 10-fold dilutions of 2 ng to 0.002 ng DNA. We determined threshold cycle numbers of the toenail DNA samples, and the amount of amplifiable DNA in toenail DNA was then determined by using the calibration curve. The starting toenail DNA amount for the PCR of MPO as derived from the calibration curve was only 0.021 ± 0.006 ng, which equals approximately 1% of the spectrophotometrically determined input DNA. Because toenail DNA is highly fragmented, we expected the production of the CYP1A2 amplicon to be less effective compared with the MPO amplicon. Indeed, amplification of the 600-bp product resulted in 0.006 ± 0.001 ng input DNA, which is only approximately 0.3% of the amount that was expected spectrophotometrically. These results are in line with data from Park and colleagues (8), who measured only a fraction (∼3%) of the DNA yield that they measured with NanoDrop when they used PicoGreen. Thus, when using nail DNA for genetic analyses, it is important to take into account that NanoDrop overestimates the amount of amplifiable DNA (>200 bp) in the samples and that more DNA should be used than what is required for a certain assay when DNA concentration is quantified with NanoDrop. However, using the PicoGreen method to assess the amount of sample needed to meet input DNA requirements for a genotyping assay probably leads to an overestimation of the amount of sample needed because single-stranded DNA is also useful for genotyping. Thus, when setting up a study using nails as DNA source for genotyping, it is best to first test what the optimal amount of input nail DNA for that specific assay is.

Toenail DNA is highly fragmented. We investigated DNA fragment length of our toenail samples with agarose gel electrophoresis. Most of the fragments appeared to be <200 bp, as shown in Fig. 1, but larger fragments were present.

Over the course of DNA isolation of the >21,000 NLCS toenail samples, we have performed a PCR analysis on 618 random samples (1–2 samples per 48 isolated samples to monitor isolation quality) to determine the presence of the earlier mentioned 596-bp and 197-bp CYP1A2 and MPO amplicons. Amplification of the 596-bp fragment was successful for 85% of the samples and amplification of the 197-bp fragment for 80% of the samples.

There are few observations from other studies on DNA fragment sizes in nails. Park and colleagues (8) were able to produce amplicons of 100-, 200- and 400-bp sequences in the nuclear b-actin gene for 62.5%, 22.2%, and 16.7% of the samples, respectively. These seem low percentages compared with our results. This difference may be explained by differences in washing and extraction methods used, resulting perhaps in increased DNA fragmentation or increased presence of PCR inhibitors in the DNA samples of Park and colleagues, or by the fact that analyses were done on different DNA sequences. Klassen and colleagues were able to generate a 456-bp amplicon of RYR2 gene exon 97 in 66% of the fingernail DNA samples, more than twice as often as for formalin-fixed paraffin-embedded (FFPE) tissue DNA samples. Amplification of 911- and 969-bp PCR products in fingernail DNA samples was unsuccessful. In the majority of fingernail samples they tested, Yoshida-Yamamoto and colleagues (10) were able to amplify a 286-bp fragment of the ESRX gene.

In conclusion, our study and other studies have shown that toenail DNA is considerably fragmented to sizes generally <200 bp, but larger fragments are present and can be amplified in part of the samples, probably depending on the protocols used to isolate the DNA from the nails and the specific genomic regions of interest to be amplified.

For this pilot study on using Sequenom MassARRAY iPLEX technology for SNP genotyping, 82 DNA samples from toenails of prostate cancer cases from the NLCS were used. Thirty-three SNPs (as listed in Table 1) in the insulin-like growth factor pathway genes were determined. SNP genotyping analysis on 100 ng of toenail DNA (as measured with NanoDrop) was carried out by Sequenom in Hamburg using the MassARRAY iPLEX SNP genotyping platform (Sequenom; ref. 3). Four aliquots from each of the 82 DNA samples (100 ng each, as measured by NanoDrop) were taken and genotyped.

Genotyping of two SNPs (rs2471551 and rs2854746) failed because primers were not adequate. In addition, we excluded the results of one additional SNP (rs8063461) because this SNP was frequently called in water samples, which was probably caused by self-extension due to formation of a hairpin or extension due to a primer-dimer formation with another primer (please note that the analysis is based on a multiplex PCR reaction).

Table 1 shows the average and range of the sample call rates for each quadruplicate analysis and the average call rate across the quadruplicate analyses. The sample call rate was high, with averages ranging from 97.7% to 99.5%.

SNP call rates fluctuated slightly across quadruplicate analyses and ranged from 93.9% to 100% (Table 1).

Table 2 shows the reproducibility of the SNP call rates across the four repeated analyses, which ranged from 98.8% to 100%. On 11 occasions, there were differences in alleles (heterozygotes called as homozygotes) between quadruplicate analyses residing in three samples in total. After exclusion of samples below a sample call rate threshold of 95%, only two discrepancies (each in a different sample) remained out of a total of 9,643 calls that were made (∼0.02%).

In conclusion, DNA from toenails enables genotyping using Sequenom's MassARRAY technique with good sensitivity and reproducibility.

The Sequenom MassARRAY iPLEX platform has so far been used in three large NLCS genotyping projects on gene–environment interactions (containing sets of 24, 20, and 30 SNPs (all with minor allele frequency > 10%) and 6,230, 2,872, and 4,238 samples, respectively) and three more such genotyping projects are ongoing. For these three projects, Sequenom genotyped 7,635 unique toenail DNA samples on the MassARRAY iPLEX platform. Articles describing the results are published (14) and in preparation.

In each of these projects, the genotyping was successful: average sample call rates were higher than 95% (overall, 97.1%). Of note, 95.5% of the samples had a call rate >90% and 92.3% had a call rate >95%. Seventy-one of the 74 SNPs that were determined across the three projects adhered to Hardy–Weinberg equilibrium. In one of the projects, 23 samples of toenail DNA were compared with 23 samples of paraffin-embedded normal tissue DNA samples from the same NLCS participants. The genotype concordance (of 30 SNPs) between the two types of samples was 99.1%. Using data from the three projects, we have investigated whether the A280/A260 and A230/A260 ratios affected the sample call rates. Table 3 illustrates the average call rates for strata of the A260/A280 and A260/A230 ratios and the percentage of samples that passed sample call rate thresholds of 95% and 90%. Average sample call rates and the percentage of samples passing sample call rate thresholds were slightly lower for lower 260/A280 and A260/A230 ratios. However, despite the relatively low purity ratios, toenail DNA renders good results with the Sequenom MassARRAY technique. For other genetic analyses, the purity of the toenail DNA may be more crucial. Therefore, it would be worthwhile to test DNA purification methods with a high DNA recovery.

The Infinium HD FFPE DNA Restore Kit was originally developed to restore degraded DNA in FFPE tissues in preparation for use with the Infinium HD FFPE Assay (4). The Infinium HD FFPE Assay gives results for 262,739 genetic markers: selected tag SNPs that provide comprehensive coverage of the genome. For Infinium platforms, Illumina recommends fragment sizes of >2 kb. To test if samples are eligible for restoration, the Infinium HD FFPE Quality Control (QC) Kit is used. The QC Kit provides primers and DNA template for a real-time PCR using standard instrumentation and reagents. Extracted FFPE samples that pass the QC test (i.e., do not cross a certain PCR cycle threshold) are suitable for restoration using the restore kit.

We tested whether toenail DNA could be restored with this restore kit as well. We selected 29 prostate cancer samples from the NLCS study, from which 24 samples (83%) were eligible for restoration based on the QC test.

After restoring, we prepared the 24 toenail DNA samples for genotyping with the Illumina HumanCytoSNP_FFPE-12 DNA Analysis Kit, through which DNA is purified and whole-genome amplified. The chips were prepared and analyzed at the Genomics Core at Leuven University using the Illumina iScan reader and the iScan Array Scanner.

The average sample call rate of the 24 toenail DNA samples was 93.8%. Nineteen of the 24 samples (79%) had a sample call rate greater than 90%, whereas two other samples had a call rate very close to 90% (Fig. 2). This percentage is comparable with the percentage of FFPE samples that passes the threshold of >90% sample call rate, as reported in the Illumina product description leaflet of the HumanCytoSNP_FFPE DNA Restore Kit (4). The average SNP call rate was 93.8% across the 24 samples, with a SD of 9.8%. Of the total of 262,739 SNPs, 0.01% (n = 28) could not be called at all. Forty-two percent of the SNPs called 100%, and 69%, 82%, and 96% of the SNPs had a call rate greater than 95%, 90%, and 75%, respectively. It should be noted that we have overestimated the amount of input DNA that was used in this assay, because NanoDrop was used to quantify DNA and we did not yet have the results of the qPCR experiment that showed that only a fraction of spectrophotometrically determined DNA in toenails is amplifiable (>200 bp) human DNA. The results on the Illumina HumanCytoSNP_FFPE-12 DNA chip might have been even better if we had used more input DNA.

This pilot study shows that the HumanCytoSNP_FFPE array works for toenail DNA samples. However, in a more elaborate study, the reproducibility and accuracy of the genotyping of nail samples through this array should be investigated. In this more elaborate study, more input DNA should be used and it may be preferable to call the samples based on internal clustering, so on the clusters of the nail samples. In the pilot described in this article, calling was done based on a standard clustering of FFPE samples provided by Illumina.

For large-scale genetic epidemiologic studies, it is of interest to know how different tissues as a source of DNA compare with regard to various aspects. Not only the quantity and quality of DNA is important, but also the ease of obtaining and storing samples, and participation rates of study participants are variables to take into consideration when choosing one tissue or the other. Table 4 shows various characteristics of different sources of DNA. This table is not an exhaustive summary of the literature.

In terms of DNA quantity and quality, blood is the optimal source of DNA. However, drawing blood is invasive and cannot be performed by study participants themselves. Thus, asking to provide a blood sample will reduce participation rates and it will make large-scale studies, especially of a geographically disperse population, difficult, if not impossible, in terms of logistics and associated costs. DNA yields from non-blood sources of DNA (buccal cells, hair, and nail) vary greatly between sources, depending on collection methods, storage conditions before DNA isolation, and DNA extraction methods. Nevertheless, of the non-blood DNA sources, mouth wash or saliva samples probably render the highest yield of human DNA of the highest quality, both in terms of DNA integrity and purity. Disadvantages of mouth wash samples are the relatively high costs associated with obtaining the samples (transport costs) and the problem of obtaining mouth wash samples in pediatric populations. Compared with the other sources of DNA, obtaining nails is least invasive and most suitable for use in pediatric studies, and together with hair samples, nail samples entail the lowest costs for obtaining and storing. On the other hand, isolating DNA from nails is more laborious than isolating DNA from blood samples or buccal samples, due to the washing steps for removing possible nail polish and other exogenous material and the incubation step to dissolve the nail material.

The pilot studies described in this article add to our previous findings (2) that nail clippings are a good source of DNA for genetic analyses, despite the fragmented nature of the DNA (mostly <200 bp) and the relatively low purity. Toenail DNA samples proved well suited for genotyping of a limited set of SNPs on the Sequenom MassARRAY iPLEX with good call rates and reproducibility. After restoration with the Infinium HD FFPE DNA Restore Kit, toenail DNA showed satisfactory call rates on the Illumina HumanCytoSNP_FFPE-12 DNA high-density array (>262,000 markers). Good results using nail DNA for high-density (HD) genotyping were also obtained by others. Nakashima and colleagues (12) have genotyped five fingernail samples on the Affymetrix GeneChip Human mapping 250K Array, obtaining an average sample call rate of 94.8%. When they compared nail samples with blood samples, the concordance rates for homozygotes and heterozygotes were 99.8% and 98.8%, respectively. They did not observe differences between fresh clippings and >5 year old clippings stored at room temperature, which once again shows that nail DNA does not degrade when nail samples are stored at room temperature for extended periods (2). Oikawa and colleagues (15) have genotyped 11 fingernail DNA samples with the Affymetrix GeneChip mapping 10K 2.0 array optimized for FFPE tissue samples. The average sample call rate was 92.5%, which was comparable with the average SNP call rate (93.8%) in our pilot study using the Illumina HumanCytoSNP_FFPE-12 DNA Array.

Nakashima and colleagues and Oikawa and colleagues isolated DNA from frozen and crushed nails, whereas we lysed nails with NaOH before DNA isolation. We have compared DNA yield and fragment sizes of both preprocessing methods (10 samples) and we obtained better results for lysed nail method than for the freeze/crush method (results not shown). Using the latter method, we did not observe DNA fragments with a size of >200 bp, which we did observe for the lysed nail method, whereas the DNA yield was not meaningfully different between the two methods. Therefore, we speculate that nail DNA obtained through lysis of nails might lead to even better results using Affymetrix GeneChip mapping arrays than nail DNA obtained with the freeze/crush method.

Our toenail DNA was also tested on the Illumina VeraCode platform combined with Illumina GoldenGate genotyping, but the call rates were too low for meaningful analyses (results not shown). Possibly, toenail DNA does not contain enough DNA fragments >200 bp, as required for this technique, this technique is relatively sensitive to the low purity of the toenail DNA samples or not enough input DNA was used. Illumina advises to quantitate DNA by PicoGreen, but we used NanoDrop to quantify DNA.

In the near future, genotyping is likely to be performed through next-generation sequencing (NGS), and then toenails are expected to be a very suitable source of DNA for large-scale epidemiologic studies because NGS does not require high-integrity DNA, contrary to some HD genotyping platforms. NGS has been successfully applied in the field of paleontology, which deals with highly degraded ancient DNA with low quantity (16). We have genotyped 96 SNPs in two toenail DNA samples through the molecular inversion probe technique followed by NGS (Ion PGM sequencer) and this rendered equally good results as blood samples (preliminary results, not shown). Currently, we are performing a more elaborate study using these techniques on more toenail DNA samples.

In conclusion, nails are a cheap and easily obtainable and storable source of DNA suitable for genotyping of a limited set of candidate gene SNPs and HD genotyping in GWAS studies, despite its fragmented nature and relatively high pollution level. For HD genotyping, however, methods are needed (e.g., restore kit, PCR conditions) that are suitable or optimized for fragmented DNA.

No potential conflicts of interest were disclosed.

Conception and design: J.G.F. Hogervorst, P.A. van den Brandt, M.P. Weijenberg, J.R. Vermeesch, F.J. van Schooten, L.J. Schouten

Development of methodology: R.W.L. Godschalk, P.A. van den Brandt, M.P. Weijenberg, J.R. Vermeesch, F.J. van Schooten, L.J. Schouten

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R.W.L. Godschalk, P.A. van den Brandt, L. Jonkers, J. Goessens, J.R. Vermeesch, F.J. van Schooten, L.J. Schouten

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J.G.F. Hogervorst, P.A. van den Brandt, L.J. Schouten

Writing, review, and/or revision of the manuscript: J.G.F. Hogervorst, R.W.L. Godschalk, P.A. van den Brandt, M.P. Weijenberg, B.A.J. Verhage, L. Jonkers, J. Goessens, C.C.J.M. Simons, F.J. van Schooten, L.J. Schouten

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): B.A.J. Verhage, L. Jonkers, J. Goessens

Study supervision: J.G.F. Hogervorst, P.A. van den Brandt, F.J. van Schooten, L.J. Schouten

The authors thank Simone van Breda, Stijn Lumeij, and Kristien Lemmens for their help with the pilot work on DNA isolation from toenails and the first phases of DNA isolation from the cohort toenail samples. Furthermore, the authors are indebted to Harry van Montfort for designing the databases used for storing and managing toenail DNA data; Sacha van de Crommert and Conny de Zwart for data management; Ivette Deckers and Milan Geybels for providing data; and Lara Oheimer for collecting toenail samples in the basement and weighing them. In addition, the authors thank the Biobank Maastricht UMC+ (Berry Soute and Claudia Bosma) for storing and retrieving our toenail DNA samples.