Introduction

The gene tyrosine hydroxylase 1 (TH01) has been suggested as a candidate for human longevity. Given its role in stress response, it is conceivable that allelic variation in this gene influences survival into old age. A previous study investigated a possible association of the short tandem repeat (STR) polymorphism in intron 1 of TH01 (localization: 11p15.5–p15, motif: TCAT) with longevity in Italians.1 The authors reported a statistically significant reduction in the frequency of homozygotes of long TH01 tetranucleotide repeats (≥9) in male centenarians compared with younger men. There was no association in females. Subsequent TH01–longevity studies, which differed from the original investigation1 in terms of study design, the analysis approaches and examined markers, described contradictory findings.2, 3, 4 For instance, Tan et al3 observed an increase in carriers of the long allele 9.3 in Italian centenarians versus younger individuals. In summary, the involvement of TH01 variation in human longevity still remains unresolved. However, if this association could be confirmed, this would have important ramifications for the use of this locus in forensic genetics.

The TH01-STR is one of the most widely used markers in forensic casework. It is a mandatory locus in many forensic databases (eg, CODIS in the US, DAD in Germany, UK Forensic Database in Great Britain) and has therefore been included in most commercially available STR multiplex PCR kits.5 However, in recent years, TH01 has not only been linked to human longevity but was also described to be associated with psychiatric disorders6 and sudden infant death syndrome.7 Other studies investigated a possible role of TH01 in arterial hypertension8, 9 or smoking,10, 11, 12, 13 albeit with conflicting results. As a forensically employed genetic polymorphism – at least in Germany – must not reveal information about phenotype, personality or disease susceptibility of the tested individual,14 a confirmed association between TH01 alleles and longevity could result in the exclusion of this marker from forensic investigations.

In view of the important role of the TH01 locus in legal medicine and the implications a genuine association would have, we sought to replicate, as primary end point, the initial finding in Italians1 in our collection of 471 long-lived individuals (LLI) and 462 younger controls from Germany. Furthermore, we investigated whether allele 9.3 exerts a role as a modulator of longevity as suggested by Tan et al.3

Materials and methods

Study population

Four hundred seventy-one unrelated German LLI between 96 and 110 years of age at recruitment (median age: 101 years) were studied. Three hundred fifty-four (75%) of the LLI were centenarians. The sample of younger German controls was obtained from the popgen biobank15 and comprised 462 living subjects aged between 19 and 75 years (median age: 59 years). The controls were matched to the LLI in terms of ancestry and gender, thus minimizing any systematic genetic differences between these samples that might arise due to very low levels of undetected population structure. Based on current predictions only 1.5% of all 60-year-old and 1.8% of all 75-year-old German females will live up to 100 years. For males, the probability is even lower (Human Mortality Database; http://www.demogr.mpg.de). Thus, fewer than eight of the younger controls are estimated to become centenarians themselves, a negligible proportion that does not affect power. The recruitment procedures for LLI and controls have been described in detail elsewhere.16, 17 Most of the participants provided, as part of a health and family history questionnaire, information about their nicotine consumption habits and arterial hypertension status. The study was approved by the Ethics Committee of the University Hospital Schleswig-Holstein in Kiel.

TH01 typing

Five nanograms of genomic DNA from each individual, extracted from whole-blood samples, was subjected to a duplex PCR amplifying a TH01-specific STR fragment and the gender-determining amelogenin locus. Primers and PCR conditions were used as described previously.18 Amplification products were separated and detected on an ABIPrism3130 Genetic Analyzer with ROX500 standard fragments (all Applied Biosystems, Weiterstadt, Germany) in comparison with a self-made allelic ladder. Electrophoresis results were analyzed using the GeneMapper ID Software v3.2 with the respective self-designed bin set. Allele peaks were scored when the relative fluorescence units were between 50 and 3000.

Statistical analysis

The primary aim of the study was to try to validate the result of De Benedictis et al.1 Secondary end points of the study were other potential differences in the TH01 allele distribution between cases and controls, with a special focus on allele 9.3. All P values were two-sided and were considered significant if smaller than 0.05. For the primary end point, only one test was performed. Owing to the exploratory nature of the study with respect to the secondary end points, no adjustment for multiple testing was conducted.

Alleles were grouped according to the previous study of De Benedictis et al:1 small (S) – 5, 6, 7, 8; large (L) – 9, 9.3, 10. The alleles ‘S’ and ‘L’ are designated in the following as grouped alleles. The allele ‘10–1’ of De Benedictis et al,1 also called ‘10*’ by Tan et al,3 is referred to as ‘9.3’ in the forensic literature19 and in this study. Allele and genotype distributions between cases and controls were compared by applying Fisher's exact test. Independence of the occurrence of allele 9.3 and other alleles in the same genotype was tested with a χ2 goodness-of-fit test. This coincided with a test for Hardy–Weinberg equilibrium. Logistic regression was performed with a carrier status of allele 9.3 as outcome and age as continuous influence variable. The distribution of 9.3 carriers was also studied in four different age groups. Sample size and power calculations were done with BIAS for Windows 8.03. For all other statistical calculations the statistics program R was used (version 2.5.120). For tests for Hardy–Weinberg equilibrium the package ‘genetics’ was applied: HWE.exact for two alleles and HWE.chisq for more than two alleles (R package version 1.3.0, 2007).

Smoking and high blood pressure both affect mortality and ageing and may be associated with variation in the TH01 gene.7, 12, 13 To investigate the robustness of our results and to prevent confounding, we repeated the analysis considering only non-smokers (n=470) and individuals free of hypertension (n=550).

Results

In the German samples, seven alleles were observed at the TH01 locus in both LLI and controls (Figure 1). No significant deviation from Hardy–Weinberg equilibrium was found in either group (P>0.05).

Figure 1
figure 1

Relative frequencies of ungrouped alleles of the TH01 STR in German LLI and younger controls.

Full size image

For our primary end point, we applied the same study design as described in De Benedictis et al1 and grouped the alleles as small (S) or large (L). In De Benedictis et al1 it was shown that the LL genotype was significantly less frequent in male centenarians when compared with the other two genotypes, SL and SS. For a significance level of 0.05, our samples had a power of over 99 or 89%, respectively, to replicate the previous results observed for South Italian males and females.1 Despite sufficient power, the association between TH01 and longevity was not replicated in the studied German samples (P values for comparison of LL versus SL/SS: 0.25 for males, 0.91 for females; Table 1). Furthermore, we repeated our analysis for the subgroup of centenarian male cases (the German female cases were all centenarians). The result was again non-significant (relative LL frequency in LLI: 0.30, in controls: 0.21, P-value for comparison: 0.067). The frequency of the LL genotype was higher in German LLI than in controls, in contrast to the result in Italians.1

Table 1 Genotype distribution of grouped TH01 STR alleles in German long-lived individuals and younger controls
Full size table

In our sample, the distribution of the ungrouped allele and genotype frequencies in cases and younger controls was very similar (Figure 1). An overall comparison of allele or genotype frequencies showed no significant difference between the two groups (alleles: P=0.48, genotypes: P=0.71).

Allele 9.3 was the most common one in Germans (Figure 1). Tan et al3 reported an increase in carriers of allele 9.3 in Italian centenarians versus younger controls. When 9.3 was analyzed separately against all other alleles in the German sample, no significant deviation between cases and controls was found (P=0.24). In addition, there was no significant result of the increased presence of 9.3 with any of the other alleles in a heterozygote genotype (P>0.05 for all other alleles). Furthermore, logistic regression analysis, in which the 9.3 carrier status was considered as outcome and age as continuous influence variable, showed no significant result (P=0.73). Also, the frequency distribution of 9.3 carriers in the four examined age groups was very similar and appeared to fluctuate randomly (P=0.16; Table 2). The frequency of carriers in the centenarian group was lower than in the younger age groups, in contrast to previous findings.3

Table 2 Frequency distribution of carriers of the TH01 STR allele 9.3 in four age groups
Full size table

When the data were analyzed excluding smokers and individuals with high blood pressure, the results for the primary and secondary end points of the study were similar to those obtained considering all individuals. The results for the primary end point are shown in Table 3 and for detailed subgroup analyses in the Supplementary Material.

Table 3 Results for comparison of grouped TH01 allele frequencies in subsamples of LLI and controls
Full size table

The TH01 allele distribution differed considerably between Germans and Italians (Table 4), with the former exhibiting higher frequencies of alleles 9.3 and 10 while in the latter allele 6 is more common. However, our dataset is consistent with that obtained from a large representative sample from Germany and the allele distribution described by Tan et al3 is in accordance with other data derived from Italy (Table 4). These findings suggest population-specific differences and ascertain the accuracy of the obtained data in both our and the Italian study.3

Table 4 Frequency distribution of TH01 alleles in Germans and Italians
Full size table

Discussion

Data from our samples of centenarians, nonagenarians and younger controls suggest that the TH01 STR locus exhibits no significant influence on the ability of attaining exceptional old age in Germans. In a previous study, TH01 was reported to be associated with longevity in Italian men.1 A statistically significant decrease in the frequency of long alleles (≥9) was observed in centenarians compared with younger males from the northern (26 centenarians versus 49 controls) and southern (27 centenarians versus 113 controls) parts of the country. This finding would indicate a potential survival disadvantage for LL carriers. On the other hand, a later study on an extended and updated version of the original De Benedictis’ data described an increase in carriers of allele 9.3 in centenarians (n=210) versus younger individuals (n=755).3 This result suggests a beneficial effect of 9.3, a long allele, which seems to contradict the earlier finding.1 However, in our collections, there was no evidence for an association between long alleles of TH01, or specifically allele 9.3, and old age (Tables 1 and 2). The first association reported in Italians might reflect a potentially false-positive result caused by the very small sample sizes examined.1 The discrepant finding between Italians as reported by Tan et al3 and Germans (our study) could be due to a population-specific effect. Allele 9.3 is much less common in Italians (22%, Table 4) than in Germans (34%, Table 4). This remarkable frequency difference is consistent with other published data from the same populations (Table 4). The distribution of allele 9.3 differs remarkably among European populations, displaying a geographic gradient of decreasing frequency from West to East and North to South (Figure 2). If the TH01 STR is selectively neutral, the observed cline could be a consequence of past migrations or random genetic drift. But since the STR may regulate gene transcription in an allele-specific manner21 and may therefore be phenotypically relevant, the geographic allele distribution might very well reflect the adaptation to different environments and selective pressures. More detailed functional studies together with population genetic analyses are necessary to clarify this question.

Figure 2
figure 2

Map showing the frequency distribution of alleles 9.3 and 10 across Europe. For the comparison, allele 9.3 was considered together with the rare allele 10 as for many of the populations studied here no data are available for 9.3 separately. Data are from Huckenbeck et al.22

Full size image

Taken together, our results do not provide evidence for the involvement of the TH01 locus in the modulation of human longevity in Germans. However, this finding does not rule out the intriguing possibility that such an association may exist in southern and/or eastern European populations. Furthermore, it highlights the need to take into consideration population genetic data when dealing with association studies. The lack of the TH01–longevity association in the investigated German sample means that – at least based on our current knowledge – the forensic application of this robust and widely used marker is still legitimate.