Neurofilament light chain in blood may indicate aging and predict lifespan across species

Neurofilament light chain (NfL) is expressed in neurons of the central and peripheral nervous system, and elevated levels have been associated with neuronal injury and neurodegenerative diseases [1]. However, NfL also increases with aging in people without known neurological disorders [2]. NfL levels were first associated with an increased risk of all-cause mortality in centenarians and non-centenarians [3] and have subsequently been proposed as a prognostic marker for all-cause mortality in population-based studies with younger cohorts (mean ages of 45–73 years) [4–7]. To investigate the generality of the age-related increase in NfL to other species, we examined NfL levels in the blood of a variety of animal species using the “Single Molecule Array” (Simoa) technology. Furthermore, we employed longitudinal blood sampling in aging mice to examine NfL predictability of death in a second species beyond those reported in humans.

Blood samples from a total of n = 862 animals of 57 species were obtained from a commercial and an academic veterinary diagnostic laboratory, a local zoo, and the Hertie Institute for Clinical Brain Research. Blood samples with known evidence of neural injury were excluded before measurement. In addition, NfL was measured in longitudinal blood samples (n = 157 observations across 44 mice) from aging mice housed at the Hertie Institute for Clinical Brain Research. For comparison, human blood NfL measurements from a recently published human age cohort (n = 438) were included in the present analysis (see Methods for details).

To determine the relationship between NfL levels and age across species, we first categorized animal (mouse, cat, dog, horse) and human samples into three key life stages—young, adult, and aged—based on the known mean life span of the species. The threshold for the aged group was set at the mean life span itself, including all samples from individuals who had exceeded that age. The adult group was classified as those between half the mean life span and the mean life span. The young group comprised samples from individuals up to half of the mean life span (S1 Table). Samples from very young animals (first 5% of their mean life span, see Methods for details) were excluded due to reports from humans and calves that NfL is elevated in early postnatal stages [8,9]. Results revealed across all species age-related increases in NfL, with an observed 3-fold (horse), 7-fold (human), about 14-fold (mouse and dog), and 36-fold (cat) increase with age (Fig 1a). Sex differences (as far as this information was available, see S1 Table) were not significant.

To examine the age-related trajectories of NfL across species, baseline NfL measurements were modeled as a function of aging. In all four animal species, we observed non-linear associations and observed age trajectories similar to those in humans. Trajectories were characterized by a slow increase during adulthood and a sharp rise towards the end of life (Fig 1b and 1c).

While longitudinal blood measurements and tracking of the animals until death were not possible for most species, we followed an independent cohort of C57BL/6J mice under laboratory conditions. This allowed us to assess within-individual changes in NfL as a function of aging. Blood was collected at 19−20 months of age and again one month later. This procedure was repeated three months later (at 23−24 months) and once more one month later, yielding a total of four longitudinal blood samples (Fig 2a; see Methods for details). The rate of change in NfL concentration (slope; ΔNfL) was calculated from a linear mixed-effects model (see Methods for details) and was inversely correlated with remaining life span (p = 0.0048; Fig 2b and 2c). Dividing the mice into two groups based on whether their NfL slope was higher or lower than the mean (62.587 pg·ml⁻¹·month⁻¹) and using Kaplan–Meier survival analysis, revealed a robust separation between the groups with mice with a lower slope surviving longer than mice with a higher slope (Fig 2d). Subsequent Cox regression analyses indicated that each standard deviation increase in NfL slope was associated with a 73.5% (or 69.1% including sex) higher mortality risk (Fig 2e).

Next, we extended our analyses of NfL to a larger set of species. While blood samples from aged animals were relatively rare, NfL measurements for young animals with blood from at least 5 animals were available for 12 species in addition to humans (Fig 3a; S2 Table). Robust differences in baseline NfL levels were found between the species, with long-lived species tending to have lower levels (Fig 3a), although this association was not significant (Fig 3b).

Since body weight has been shown to be associated with NfL levels in humans [10], NfL levels were compared to the mean body weight within species. Results revealed again an inverse relationship between baseline NfL levels and body weight (Fig 3c). To account for differences in body weight in the correlation analysis between NfL levels and life span, we tested two models according to a previous publication [11]: NfL/body weight ~ maximum life span (r = −0.7253, p = 0.0067**) and NfL ~ maximum life span + body weight (r = −0.467, p = 0.1103). However, the high correlation between body weight and life span in the present dataset (r = 0.7637, p = 0.0034**) likely explains the strong correlation in the first model.

Blood NfL could be measured in a further 39 mammalian species (Fig 4a). Consistently, the amino acid sequence of the NfL epitope critical for the present immunoassay [12–14] revealed a very high degree of conservation for all species for which the NfL sequence was publicly available (Fig 4b). While NfL could be detected in every mammalian blood sample examined, NfL could only be detected in 5 out of 48 (5/48) tortoises or turtles (Testudinidae, Geoemydidae, and Emydidae), in 1/6 snakes (royal python, boa constrictor, corn snake), in 1/1 crocodile, in 3/19 lizards (iguana, bearded dragon, scorpion bearded lizard), and in 1/4 birds (i.e., in a 29-year-old Amazon parrot, but not in any chicken or cockatiel. The reptiles and bird measurements may be false positives, but it is equally likely that healthy reptiles and birds have NfL levels below the detection limit of the current immunoassay.

We found significant age-related increases in NfL for all species for which blood samples were available across the entire age range. Age-related increases have previously been noted for dogs [15,16] and among adult cattle [8]. Increased levels of phosphorylated neurofilament heavy chain have been reported in horses [17]. Our data also suggest that NfL trajectories in aged mice predict mortality, as has been reported in aged humans [3–7].

Due to the nature of samples sent to veterinary diagnostic laboratories, many blood samples from this source are likely to come from animals with unknown health problems, which in turn may affect NfL levels. However, unknown health problems are expected to occur equally in samples from all age groups sent to a diagnostic laboratory. Moreover, for the analysis of aging and baseline NfL, the median was used to smooth out differences coming from individual health problems.

In humans, the age-related increase in NfL predicts the progression of neurodegenerative diseases [18] but is also associated with all-cause mortality in aging cohorts [3–7]. Consistently, large blood proteomic studies have identified NfL as one of the top proteins predictive for/associated with mortality [19,20], suggesting a link between mortality and the aging nervous system. While it has been suggested that underlying subclinical neurodegenerative diseases might drive the association between NfL levels and mortality in humans, neurodegenerative diseases as known in humans are rare in animals. While in aged cats, dogs and horses, aspects of Alzheimer pathology (Aβ plaques or Tau tangles) have been reported, importantly, C57BL/6J mice (as used in the present study) do not exhibit any signs or core pathological features of human neurodegenerative diseases [21]. This suggests that a faster NfL rate of change predicts death independently of underlying subclinical neurodegenerative diseases. A limitation of the experimental longitudinal mouse cohort design used here, with the final blood sample taken around 24 months of age, is that mortality was assessed in ‘aged’ mice, preventing conclusions whether changes in NfL at younger age would also be predictive of mortality. Additionally, the statistical basis for the mortality analysis rests on the critical assumption that the timing of the pre-defined humane endpoint is an accurate and unbiased proxy for the timing of natural death.

Variation in baseline NfL levels in the blood of mammals showed an inverse relationship with body size and life span, but the collinearity between these two factors makes the interpretation difficult. Differences in baseline NfL levels among species may reflect the brain/body mass ratio - larger brains have more NfL, and brain-derived NfL is more diluted in larger bodies. However, despite similarities in brain-to-body mass ratios between mice and humans [22], we observed a 7.4-fold difference in baseline (median) NfL concentration. NfL is also a component of the peripheral nervous system [1,23] and its scaling relative to body size is unclear.

An alternative explanation could be that baseline NfL levels align more closely with species-specific metabolic differences [24,25]. Smaller animals generally have higher metabolic rates and more rapid neuronal turnover, which may contribute to increased calpain-mediated neuronal NfL release [13]. This interpretation aligns with our previous finding that diet-restricted animals have lower NfL levels [3], which is consistent with reduced metabolic rates in diet-restricted animals and humans [26,27]. This could also explain why NfL levels are detectable in only a few reptile species, despite overall homology in the blood NfL fragment sequence. One possibility to address such mechanistic questions would be to assess NfL levels in species (e.g., naked mole rats, bats) that deviate from the typical body weight-life span correlation observed in mammals [11].

The sequence conservation of the blood NfL fragment among species suggests an evolutionarily conserved process of neuronal NfL release into the blood. For mice, a calpain-mediated release of NfL has been suggested with an immune (microglia) activation of the secreted NfL fragments [13] and microglia activation with aging is commonly observed in many species [28,29]. A conserved NfL-mediated immune mechanism is an exciting possibility and would advance NfL as a cross-species biomarker of neural aging [30].

The current findings open the door to broader applications of NfL measurements beyond human medicine. They demonstrate the feasibility of using NfL as a biomarker to track neurological health and assess potential aging interventions in non-human species with potential applications in veterinary medicine.