Telomeres, Methylation Drift and Senescent Cells at Ages 24–400 Years: 1,584 Assessments of 389 Pan-Humans in the Faculty Longevity Panel of GSV Terms And Conditions Apply, 1550–1849 CE

1. Introduction

A pan-human of the Culture is shaped long before the first birthday. The genome carries the modifications of the Culture's usual upbringing, glands that can secrete a wide range of compounds at will, and a body that heals and regulates itself far better than the unmodified form. Most such citizens live for 350 to 400 years. What happens inside their cells across that span has been described mainly from cross-sectional surveys, in which people of different ages are measured once and compared.

Cross-sectional surveys cannot separate two things. The first is the change a given person undergoes with the passing of years. The second is any difference between people who happen to be of different ages, whether in the modifications they were born with or in the kind of person who attends a clinic at a given age. An early published telomere survey of adults found shorter telomeres in older citizens and did not claim more (Ventrelline, 1688 CE). Work on gland-supported cell maintenance has since shown that citizens may alter their own cellular turnover (Ostreck, 1744 CE), which makes the comparison harder still.

Repeated measurement of the same people answers some of this. The Pan-Human Genetics Faculty of GSV Terms And Conditions Apply has kept a longevity panel since 1550 CE, open to any citizen the Faculty invites and who chooses to take part, and has described its design and recruitment elsewhere (Crazdet & Dessavoy, 1846 CE). This paper is the first analysis of the panel as a whole. We write in 1850 CE, and every assessment analysed here was made before the end of 1849 CE.

We ask how three markers of cellular ageing change with age across the span of a life, whether the pace changes within it, and how the markers relate to the amount of elective modification a citizen has chosen since birth. We did not fix the form of the age curve beforehand. We say below where we selected it from the data.

2. Methods

Participation was by invitation and by choice. The Faculty invited citizens in four age bands at enrolment (20–99, 100–199, 200–299 and 300 years and over), taking those who had agreed to be contacted, and offered a visit with a blood sample at intervals the participant set. Nobody was invited beyond 400 years of age. The panel therefore says nothing about lives longer than that. Of 520 citizens invited, 108 declined (20.8%) and 412 consented (79.2%). Of those who consented, 23 (5.6%) gave no usable sample, usually because they chose not to attend a first visit, and 389 remain. Those 389 contributed 1,584 assessments between 1550 and 1849 CE, a mean of 4.07 each (range 1–9). Twenty-three were assessed once and 366 more than once. Among those 366, follow-up ran a median of 41 years (interquartile range 27–60; longest 134).

Ages at assessment ran from 24 to 400 years. At enrolment, 67 participants were in the youngest band (17.2%), 113 in the second (29.0%), 114 in the third (29.3%) and 95 in the oldest (24.4%). In all, 141 participants were assessed at 300 years or more, and 223 assessments, from 95 participants, were made at 350 years or more. The Faculty also kept a skin biopsy for any visit at which the participant agreed to one. Biopsies were taken at 409 assessments (25.8%) from 155 participants (39.8%).

Three markers were measured. Leukocyte telomere length was measured in kilobases (kb) from the blood sample. The methylation drift index is the Faculty laboratory's own scale, set from methylation at a fixed panel of sites. Higher values mean greater distance from the profile of the Faculty's reference adults aged 20 to 29. The index has no units, and its zero is not the reference adults' mean: the panel's 11 assessments at ages under 30 averaged 15.6, and values below that occur in individuals. Senescent cell burden was the percentage of cells in a skin section that stained positive for a senescence marker, which we analysed on the log scale. All three scales and the reference sample are this paper's own. Technical reliability was checked on 120 duplicate samples, with intraclass correlations of 0.98 for telomere length and 0.93 for the drift index.

Modification history was coded by two readers from the Faculty's records of each participant. The Faculty's records use "course of elective modification" for an adjustment to the genome, made after birth at the citizen's own request, that the Faculty carried out or logged. We grouped participants at each assessment by the number of courses completed to that date: none, one or two, and three or more. Birth modification, common to all participants, was not coded. A second variable recorded whether the participant had ever completed a voluntary sex change, which the Faculty also records by invitation (see the companion study of voluntary sex change). The two readers agreed on the group for 142 of 150 records examined by both (94.7%, κ = 0.91).

The primary model was a linear regression of each marker on age, with a second slope for years beyond a knot, group of elective modification and sex-change history. Standard errors were clustered by participant, which allows for repeated assessments of the same person, and intervals are 95% Wald intervals. We selected the knot from the candidates 200, 250, 300 and 350 years by the smallest residual sum of squares, for telomere length and independently for the drift index. Both chose 300. All p-values for the age slopes are therefore conditional on a break selected from the data, and we treat them as descriptive. A second model replaced age with age at enrolment and years of follow-up, so that differences between people of different ages could be compared with change within people. Biopsy burden was analysed on the log scale with the same covariates and the knot carried over from the blood markers, and we report ratios. We made no correction for the number of comparisons.

3. Results

Both blood markers moved with age, and moved faster in the fourth century. Table 1 gives the means by age at assessment. Mean telomere length fell from 8.92 kb in the youngest band to 7.64 kb in the oldest, a difference of 1.28 kb, while the drift index rose from 15.23 to 22.27. Most of each change fell in the last band.

[object Object]

The regression puts numbers on the break. Telomere length fell by 0.31 kb per century of age up to 300 years (95% CI 0.22 to 0.40), and by a further 0.87 kb per century beyond it (0.58 to 1.15), so that after 300 years the slope was −1.18 kb per century (−1.41 to −0.94). The drift index rose by 1.85 points per century up to 300 years (1.19 to 2.51) and by 4.90 more beyond it (2.80 to 7.00), a combined slope of 6.74 (5.03 to 8.45). Table 2 gives the full models. The candidate knots at 200, 250 and 350 years fitted worse for both markers, and the 300-year knot did not arise from any prior reason, so we regard its location as approximate.

[object Object]

Elective modification was associated with small differences. At the same age, participants with three or more courses had telomeres 0.20 kb longer than those with none (95% CI 0.02 to 0.37, p = .026), and those with one or two had telomeres 0.15 kb longer (0.01 to 0.30, p = .042). For the drift index, three or more courses went with a value 1.71 points lower (−2.84 to −0.58, p = .003), but one or two courses did not differ detectably from none (−0.47, −1.43 to 0.49, p = .34). Sex-change history showed no detectable association with either marker. Of the 389 participants, 105 (27.0%) had completed one by their last assessment. At their last assessment, 49 participants had no courses (12.6%), 140 had one or two (36.0%) and 200 had three or more (51.4%).

The second model separated people from change within people. For telomere length, the difference between participants who were a century apart at enrolment was −0.49 kb (−0.57 to −0.42), and the change within participants over a century of follow-up was −0.42 kb (−0.59 to −0.26). For the drift index the corresponding figures were 2.89 points (2.40 to 3.37) between people and 1.85 points (0.50 to 3.20) within them. We did not test whether the two slopes differ. Follow-up is short against the span of a life, so the within-person slopes rest on stretches of a few decades.

Biopsy burden rose with age in the 155 participants with biopsies. Median burden was 2.46% of cells in the youngest band and 4.11% in the oldest. Burden rose by a factor of 1.22 per century up to 300 years (95% CI 1.13 to 1.31). The additional factor beyond 300 years was 1.10 (0.88 to 1.37), and the combined factor after 300 years was 1.34 per century (1.11 to 1.61). Participants with three or more courses had a burden 0.88 times that of participants with none (0.77 to 1.00, p = .054), and those with one or two courses did not differ (1.00, 0.90 to 1.11).

4. Discussion

In this panel, telomere length and methylation drift changed slowly through the first three centuries of life and faster in the fourth. The change was not a straight line. The slope of telomere loss was nearly four times as steep after 300 years as before, and the drift index rose at more than three times the earlier pace. The biopsy markers point the same way. The additional slope beyond 300 years was not itself detectable for biopsy burden, but the combined slope was, and the sample at that age is smaller.

The panel shows the pace of change in 389 people who chose to take part. It does not show when any individual's pace changes. Our break at 300 years was the best of four candidates, not a threshold we have shown to exist in anyone's biology, and the p-values attached to it are conditional on the choice. Different candidates, or a smooth curve, might locate the change differently. The most we claim is that the average pace in this panel is higher after about 300 years than before.

Elective modification relates to the markers less strongly and less evenly than age does. Three or more courses went with longer telomeres and lower drift at the same age, and the telomere differences are small beside the 1.28 kb that separates the youngest and oldest bands. Their intervals come close to zero, one or two courses did not move the drift index, and the biopsy ratio of 0.88 had an interval that reached 1.00. We cannot call these effects. Citizens chose their courses. Those who take many may differ from those who take none in what they were born with, in how they look after themselves, in the glands they use, or in their curiosity about their own bodies, and the Faculty's records say little of these (Castravey, 1831 CE).

Between-person and within-person slopes differ in size for both blood markers, and for the drift index the between-person slope is the larger. We did not test the difference, so we make no claim about it. If it holds, part of what appears in a cross-sectional survey would arise from who is in the sample at each age. That is what the earlier surveys could not exclude, and the present panel cannot exclude it either.

5. Limitations

The panel is a volunteer series from one General Systems Vehicle, and 20.8% of those invited declined. We cannot compare decliners with participants, and in other volunteer panels participation has depended on age and on disposition (Fennitor, 1802 CE). In another panel kept by an Orbital Hub, consent fell among the oldest residents (see the companion time-use diary study). The oldest band here is the one where such selection would matter most.

Follow-up is short. The median of 41 years, and a longest of 134, cover a small part of a life of three and a half centuries or more. The panel's three centuries are built from overlapping stretches of different people, and the age curve is joined from them. Participants who left the panel, by choice or because they moved beyond the Faculty's reach, are not analysed, and we do not know whether they differed from those who remained.

The markers are the Faculty's own. Laboratory methods changed several times in three centuries, and while the Faculty re-measured its reference store at each change to keep the scales comparable, we did not model an assay period, and a shift at a change would bias the slopes from follow-up (Dunmerrow & Plaskett, 1838 CE). Methylation indices of this kind have been tied to the reference ranges of the laboratory that produced them and not to any common scale (Okhumsett, 1817 CE). Telomere length and the drift index in blood do not stand for the state of other tissues, and the biopsy subset was self-selected.

Modification history came from records that begin with a citizen's request. Courses that the Faculty neither carried out nor logged are missing, and the grouping by number of courses ignores which adjustments they were. Every participant carried the modifications of ordinary upbringing, so the panel cannot show what ageing looks like without them. Finally, the analysis is observational and unadjusted for many things a citizen chooses, and the 300-year knot was selected from the data.