Subtraction gives the numbers the totals hide: 25 genes (5,061 minus 5,036) and 136 metabolites (181 minus 45) did not move in one group-wide direction [1][2]. As shares of what changed, that is half a percent of the genes against three quarters of the metabolites, roughly a 150-fold difference in how personal the trajectory is [3]. Both layers were mostly quiet, too: fewer than a third of the genes examined and about a fifth of the metabolites changed measurably at all across the eight years [4].
The design is why any of this is visible. Co-author Julia El-Sayed Moustafa of King's College London said most aging studies take only a snapshot, comparing different people at a single point in time, whereas here each woman was measured against her own earlier samples [15]. Blood gene expression is read on a mixture of cells, and the paper reports that different immune cell types carried distinct age-related expression patterns [12], so part of a shared group-level direction can be shared drift in which cells are present. The divergence that does appear is not just noise: identical twins' expression patterns were more similar than fraternal twins', which the authors read as a strong genetic contribution [11]. Metabolites sit further downstream, which is the same reason the seasonal and daily sensitivities land hardest on them [10].
The paper contains a cleaner demonstration of the underlying problem in an unexpected result: blood levels of PFOA and PFOS fell over the study window, a decline the researchers attribute to restrictions on the chemicals rather than to anything about growing older [13]. Eight years of within-person change carries the era along with the aging.
The thing this doesn't tell you is whether any particular clock breaks. As reported, the Sept. 3 Science paper measured gene activity and metabolite levels at each visit [1][4], while the aging clocks Live Science describes read molecular markers such as chemical tags on DNA [16]; no clock output was scored against these trajectories. The cohort is 335 women aged 32 to 80 [2], and each person's direction rests on at least three visits [3], enough to see someone depart from the group average and thin for describing that person's curve. The material also does not document who is selling biological-age scores or which trials are using them as endpoints.
Raghav Sehgal of the Yale School of Medicine, who was not involved in the work, told Live Science that biological aging is "less like a fixed score and more like a changing set of molecular trajectories" [17]. For the metabolites, that is what these data show, and the paper adds that each woman's overall metabolite profile drifted away from her own earlier profile [9]. For the blood transcriptome measured here, near-unanimity was the rule [6], and near-unanimity is the regularity a group-fitted clock lives on. The defensible claim is narrower than the general one: a single aging number is under-specified until it says which molecular layer it reads and when the blood was drawn.
