The biological scaling of Alzheimer’s disease neuropathological changes across primate species

Abstract

Background

Studying the spontaneous emergence of biological anomalies within the animal kingdom can provide insights into the causes of diseases. It is often assumed that Alzheimer’s Disease (AD), like other neurodegenerative diseases, is specific to humans. However, the age-related occurrence of AD neuropathological changes (ADNC) in non-human primates (NHPs) and their comparison with humans has not been formally studied. Moreover, a conceptual framework for interpreting the spontaneous occurrence of ADNC in NHPs has yet to be established.

Methods

We conducted a systematic review of the available data describing spontaneous ADNC in various NHP species. To study the biological scaling of ADNC, we used logistic regression models to compare NHP and human findings, based on both chronological age and age standardized to each species’ maximum longevity.

Results

Amyloid plaques appear in all primate species according to the same temporal dynamics once the theoretical maximum age is considered, and are significantly more frequent in NHPs than in humans. In contrast, tau neurofibrillary tangles are rare in NHPs and only appear at the limit of their life expectancy.

Conclusion

The biological scaling of amyloid plaque development follows an isometric model (proportional to lifespan), whereas tau tangles emerge at a similar temporal horizon across primate species, regardless of lifespan (a chronometric model). This temporal decoupling challenges the amyloid cascade hypothesis as a universal, cross-species biological mechanism in late-onset sporadic AD. The occurrence of full-blown ADNC may depend on the phylogenetic temporal coupling of these two biological processes.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13195-025-01889-2.

Background

The concept of allometry is classically applied for interpreting the spontaneous occurrence of biological anomalies across different animal species. In comparative anatomy, allometry refers to the co-variation of traits across species, where proportions, shapes and functions of organs change with body size as a whole [1]. For instance, when comparing small and big quadrupeds, the length and thickness of bones cannot remain proportional, leading to allometric change of shape. Developmental biology generally explains allometry by different durations or rates of developmental processes. When it comes to stages of life history, common sense often assumes strict conservation of proportions, i.e., isometry, as exemplified by the widespread belief that the age of a cat should be multiplied by 7 to determine its “human-equivalent” age.

It will soon be 120 years since Aloïs Alzheimer described the characteristic lesions of the disease that now bears his name, linking them to the presence of early-onset dementia [2]. The mechanisms behind the emergence of AD neuropathological changes (ADNC) - now identified as amyloid plaques, neuritic plaques, and tau neurofibrillary tangles (NFTs) - remain poorly understood in sporadic AD, but the stronger determinant is age [3]. Thus, we hypothesized that allometry can be applied to pathological processes such as ADNC as well as to physiological processes, and that evaluating the allometric scaling of amyloid plaques, neuritic plaques and NFTs in non-human primates (NHPs) may provide valuable insights into the biological mechanisms underlying sporadic AD.

Methods

Search strategy

A literature search in PubMed and Web of Science was performed using the following search string: for each of 56 NHP species: (scientific name OR vernacular name) AND (”amyloid” OR ”A-beta” OR ”tau” OR ”tauopathy” OR “NFT” OR “neurofibrillary tangles” OR “amyloidopathy” OR ”Alzheimer” OR “dementia”). The search was completed with a generic search for the same keywords associated with “non-human primate” OR ”ape”. Articles published between 1990 (the widespread use of anti-tau and anti-amyloid antibodies) and July 2024 were considered. The flow chart outlining this selection is shown in Fig. 1.

Fig. 1.

Flow chart of the study selection for systematic review. A literature search in PubMed and Web of Science was performed for 56 NHP species. The search was completed with a generic search for the same keywords associated with “non-human primate” OR ”ape”. Titles and abstracts were reviewed to select observational studies reporting the search for both naturally occurring amyloid and tau lesions. The articles reporting experimentally induced lesions were excluded. We then removed articles that did not provide the age at death and articles that described previously published observations of the same animals. Final analyses were based on a selection of species for which we had at least n ≥ 3 brains by decades. NFT: Neurofibrillary Tangles; NHP: Non-Human Primate

The maximum longevity for each species is adapted from The Animal Ageing and Longevity Database (https://genomics.senescence.info/species/index.html).

The data collected in NHPs were compared with human neuropathological data reported as a function of age by Spires-Jones and colleagues [4]. Since NHP studies generally provide only binary presence/absence information on lesions, without staging, we pooled the human data across Braak/Thal stages and extracted overall frequencies.

Statistical analyses

Statistical analyses were performed using R (version 4.2.2). Logistic regression was performed to assess the effects of species and age on the presence of amyloid plaques and NFTs, before and after time standardisation according to the maximum longevity of each species (age/maximum lifespan). Generalised linear model was preferred to a traditional linear model to better model a binary response (presence/absence of lesions), by specifying a family of binomial distributions with a logit function. Two models were compared for both amyloid plaques and NFTs :

1

2

Results

Study sample

We conducted a systematic review of the literature reporting spontaneous ADNC in NHPs between January 1990 and July 2024 (Fig. 1). We considered the presence or absence of any amyloid plaques, whatever their density, and NFTs (or ‘tangle-like’) when they were explicitly mentioned in articles and did not include for instance individuals with isolated cerebral amyloid angiopathy or diffuse phospho-tau staining. Too few articles distinguished between the definition of diffuse plaques and neuritic plaques for our analyses to systematically take this difference into account. We selected species for which we had at least n ≥ 3 brains assessed for amyloid and tau pathology in each age category (decades). It resulted in a selection of 38 peer-reviewed articles describing the occurrence of amyloid and tau pathology in 543 NHP brains (see the complete list in Supplemental Information).

The 4 NHP species available for final analyses covered a broad range in the evolutionary tree of NHPs, including the prosimians (lemurs; Microcebus murinus; maximum reported longevity = 18 years; n = 101), old world monkeys (cynomolgus monkeys; Macaca fascicularis; maximum reported longevity = 39 years; n = 193 & rhesus monkeys; Macaca mulatta; maximum reported longevity = 40 years; n = 210) and great apes (chimpanzees; Pan troglodytes; maximum reported longevity = 68 years; n = 39).

Temporal progression of amyloid plaques and age-related occurrence of NFTs

To assess whether allometric scaling can be applied to the development of ADNC across primate species, we analysed the frequency of amyloid plaques and NFTs in relation to both chronological age and relative age (expressed as a percentage of maximum longevity). For example, a 30-year-old rhesus monkey has reached about 75% of its maximum lifespan. We included all reports of amyloid plaques and NFTs in NHP, regardless of their anatomical localization or pathological stage, as long as the authors described their presence. We then compared the NHP data to the corresponding pathologic process by age categories in humans, as previously described [4, 5]. Figure 2 shows the chronological and age-related frequencies of amyloid plaques and NFTs in lemurs, cynomolgus monkeys, rhesus monkeys, chimpanzees and humans.

Fig. 2.

Chronological (age in years) and standardized (relative to the maximum longevity of each species) frequencies of amyloid plaques and NFTs in lemurs, cynomolgus monkeys, rhesus monkeys, chimpanzees and humans. The number of individuals (n) studied for each species and each age group is indicated below each diagram

On a chronological scale, the first amyloid plaques appeared earlier in all NHPs species compared to humans. The frequency of amyloid plaques in the oldest individuals reaches at least 70% in all primate species. Notably, cynomolgus monkeys, rhesus monkeys, and chimpanzees exhibited particularly high frequency, reaching 100% in the case of cynomolgus monkeys. In contrast, the occurrence of NFTs is lower in older NHPs (around 50% in rhesus monkeys, 30% in chimpanzees, and non-existent in lemurs), whereas they are found in humans at advanced ages almost universally, across all Braak stages.

On a scale normalized to the maximum lifespan of each species, the age-related frequency of amyloid plaques appeared remarkably similar across primate species. In the last decile of lifespan, the presence is homogeneous and high across species (70–100%). On this allometric scale, NFTs appear prematurely in humans (at around 20% of the maximum lifespan), with a frequency that gradually increases to reach almost 100% at 80% of the maximum lifespan. In contrast, in NHPs, NFTs only appear at the upper end of the lifespan and with a much lower frequency (20–50%).

In statistical analyses, the logistic regression of the frequency of amyloid plaques over time and species shows that these lesions are significantly more frequent in NHPs than in humans, with higher Log Odds Ratios (Log(OR)s) (Table 1). Compared to humans as a reference, when age is normalized to the maximum lifespan, the Log(OR)s all approach zero, showing that the effect of species on the frequency of lesions becomes negligible, and even non statistically significant for chimpanzees. Although amyloid plaques appear earlier in chronological age in NHPs, our analysis shows that this effect is almost entirely proportional to the maximum longevity in each species, suggesting an isometric scaling. In contrast, the presence of NFTs is significantly less frequent in NHPs than in humans and even absent in lemurs. The Log(OR)s for the presence of NFTs deviates even further from zero when age is normalized to the species’ maximum lifespan. This finding rules out allometric scaling in the development of NFTs.

Table 1.

Logistic regression of the frequency of amyloid plaques and neurofibrillary tangles across time and species. Two models were used for both amyloid plaques and neurofibrillary tangles, considering either chronological age (Model 1: Time) or age normalized to the species’ maximum lifespan (Model 2: Time/Maximum Lifespan)

	Amyloid plaques Model 1 (Time)	Amyloid plaques Model 2 (Time/Maximum lifespan)	Neurofibrillary tangles Model 1 (Time)	Neurofibrillary tangles Model 2 (Time/Maximum lifespan)
Humans	Ref	Ref	Ref	Ref
Chimpanzees Log(OR) [95% CI]	2.6 [1.7; 3.5]	0.7 [-0.13; 1.5]	-3.0 [-4.8; -1.7]	-7.0 0 [-9.5; -5.2]
Rhesus monkeys Log(OR) [95% CI]	6.0 [4.7; 7.5]	1.3 [0.9; 1.7]	-1.8 [-3.3; -0.8]	-7.6 0 [-9.8; -6.0]
Cynomolgus monkeys Log(OR) [95% CI]	5.9 [4.7; 7.5]	1.4 [1.0; 1.7]	-0.7 [-1.8; 0.2]	-5.0 0 [-6.2; -4.1]
Lemurs Log(OR) [95% CI]	8.7 [7.2; 11.0]	2.0 [1.5; 2.5]	-19 [-1439; 202]	-22 0 [-778; 104]

Maximum ADNC level according to species evolution

Our frequency analyses did not systematically account for the staging of lesions because data were available in only a few studies. Thus, we reported in Fig. 3 the highest levels of lesions observed (at least once) in each species, based on international criteria for human ADNC (NIA-AA 2012) [6]. These criteria include the staging of amyloid plaques according to Thal phases, neuritic plaques based on the CERAD classification, and NFTs according to Braak staging. A systematic analysis using these three classifications in NHPs was only available for chimpanzees [7]. A 57-year-old male animal was reported with Thal 4, CERAD moderate and Braak V, corresponding to a high ADNC level. The information is more fragmented for other NHP species, but it appears that the closer a species is to humans on the evolutionary tree, the higher the possible maximum ADNC level. These results, based on a small number of animals in convenience samples, should nevertheless be interpreted with caution and need to be confirmed in more systematic studies.

Fig. 3.

Highest level of Alzheimer Disease Neuropathological Changes (ADNC) lesion reported at least once for each species, according to international neuropathological criteria (NIA-AA 2012). T = Thal phase (0–5); C = CERAD score (0–3); B = Braak stage (0–6). * Unlike studies in humans and a case report in chimpanzees, no study systematically assesses all three pathological scores in macaques; this maximal theoretical ADNC level therefore corresponds to observations from different animals

Discussion

We conducted the first systematic review of the presence of amyloid plaques and NFTs in NHPs, combining a comprehensive synthesis of existing knowledge with a rigorous statistical approach. This work represents the first quantitative analysis of the occurrence of ADNC in relation to both chronological age and relative age (i.e., age as a proportion of maximum lifespan), enabling comparative trends to be established across primate species. It offers a novel framework for understanding the involvement of amyloid and tau in neurodegenerative processes. We found that amyloid plaques appear in all species according to the same temporal dynamics once the theoretical maximum age is taken into account. Over equivalent portions of their respective lifespans, we even observed a higher frequency of amyloid plaques in NHPs compared to humans. This finding indicates that NHPs may be more predisposed to developing amyloid plaques than humans, which is in line with some previous reports [8, 9]. This is also consistent with the fact that all NHPs are ApoE4 homozygous according to human nomenclature (the main genetic risk factor for developing AD) [10]. We did not observe a similar temporal progression of NFTs frequency across species. The occurrence of NFTs aligns more closely with a pattern of lifespan-independent progression. As a result, NFTs can be very common in humans, who are the longest-lived.

Beyond the primate species analyzed in this systematic review, frequent deposits of amyloid plaques have also been reported in marmosets, tamarins, and squirrel monkeys (usually associated with cerebral amyloid angiopathy) [11–13]. These studies do not describe mature NFTs in these NHP species, but only occasionally the presence of non-pathological hyperphosphorylated tau, as in the marmoset, for example. Beyond primates, it appears that the frequent presence of amyloid plaques, which seem to accumulate proportionally to lifespan, is common across many mammalian species, both terrestrial and marine, such as dogs [14] and dolphins [15]. Again, NFTs have not been reported at comparable ages, although hyperphosphorylated tau lesions are sometimes observed.

When the duration of a process scales strictly in proportion to lifespan across different species, it is called isometric. When there is a corrective factor causing the scaling to deviate from strict proportionality, it is called allometric. An example of an isometric trait is the proportion of adult life spent in the post-reproductive phase among wild primates [16], while the rate of telomere shortening appears to be allometric [17]. However, not all processes follow isometric or allometric patterns across species. For instance, the turnover rate of enterocytes in humans, mice, and rats seems to be independent of their lifespan [18]. Some processes may even be invariable, such as the timing of menopause, as observed in comparisons between chimpanzees and humans [16]. To our knowledge, there is no established term for biological processes whose duration does not co-vary with other biological traits. We propose the term chronometric to describe processes that maintain a similar duration across species regardless of differences in lifespan.

In our statistical analyses, the Log(OR)s of amyloid plaque frequency over time and across species approach zero in the model that includes age adjusted for the maximum lifespan of each species. Furthermore, the temporal dynamics of amyloidosis occurrence were consistent across species, particularly in the later stages of life. This supports the idea that the process of amyloid plaque occurrence may follow an isometric pattern, meaning the process is faster in short-lived species and slower in long-lived species, scaling strictly in proportion to lifespan. NFTs’ occurrence, however, does not follow the same pattern. The significant negative Log(OR) suggests a form of protection against the development of these lesions in NHPs compared to humans, as NFTs were less frequently observed throughout their lifespan. NFTs have only been reported in NHPs at extreme ages of life, between 30 and 50 years old in macaques and chimpanzees, which corresponds to the same chronological age in humans for the early NFTs Braak stages. Thus, NFT progression is not allometric and may be chronometric, progressing independently of lifespan. Regardless of the biological environment and species, it takes 30 to 50 years for the first lesions to develop. This chronometric characteristic is corroborated by the chemical kinetics properties of tau aggregates. Indeed, it has recently been shown, both in vivo through tau-PET imaging and postmortem using seeding amplification assay, that the local replication of aggregates controls the rate of accumulation of tau pathology, and that the number of aggregates can at most double every five years [19].

Although initially based on the observation of autosomal dominant genetic forms of AD [20], the amyloid cascade hypothesis is the most widely accepted pathophysiological model for explaining also the sporadic forms of the disease [21]. This theory postulates a biological continuum starting with brain amyloidosis, followed by tau pathology, neurodegeneration, and ultimately the appearance of symptoms. Thus, according to this theory, the biological scaling of amyloid and tau pathologies should be considered the same across species unless evidence indicates biologically relevant variations. This is not what we observe in our analyses: first amyloid depositions follow an isometric pattern across species, whereas NFTs appear according to a chronometric pattern. Our observations suggest that the two processes do not progress simultaneously, supporting an alternative hypothesis in which these two proteinopathies may develop independently, each following its own timeline, before potentially converging in the brain of the same individual. The importance of age and of the temporal coupling between the emergence of amyloid and tau lesions, enabling their synergy, is discussed even within the human species. Indeed, neuropathological community-based autopsy studies have shown that the association between ADNC and dementia is stronger in younger old persons than in older old persons [22–24].

The synergistic effect of the co-occurrence of these amyloid and tau lesions could then lead to the development of the neurodegenerative process that results in symptoms [25]. This is consistent with our previous experimental work in macaques, showing that following intracerebral injections of tau seeds extracted from AD brains, tau lesions can propagate on their own in the brain, with Aβ oligomers mainly playing a role in lesion maturation [26]. This also aligns with observations in humans showing that a patient with an isolated proteinopathy can remain asymptomatic throughout his life [27], and that the mechanisms driving the transition to a symptomatic neurodegenerative disease are still poorly understood. This suggests that asymptomatic proteinopathies in general (and brain amyloidosis in NHPs in particular) should be considered as discrete nosological entities rather than diseases following a deterministic biological continuum [28].

This phylogenetic temporal uncoupling between brain amyloidosis and tauopathy would explain a potential human specificity of AD despite the fact that amyloid plaques are more frequent in NHPs. The occurrence of full-blown AD in each species thus depends on the phylogenetic temporal coupling of these two biological processes. This coupling is therefore theoretically impossible in NHPs due to the chronometric development of NFTs over 30 to 50 years, except in exceptional cases such as one reported instance in chimpanzees where a level of ADNC equivalent to that seen in human dementia was observed [7]. Alternative hypotheses may account for this decoupling in NHPs. One possibility is that a vulnerability factor, specific to the human species (and to certain individuals in particular), represents the essential biological link between amyloid pathology and the development of a clinically relevant tauopathy. Such a factor within the cascade would thus emerge as a promising target for therapeutic intervention. Another hypothesis is that the amyloid plaques forming in NHPs do not have the same biological characteristics as those observed in humans, limiting the initiation of the pathophysiological cascade. Along these lines, a study comparing amyloid plaques in squirrel monkeys and humans showed largely identical biochemical profiles in mass spectrometry analysis but also showed differences in the conformation of the aggregates and in certain post-translational modifications [8]. Finally, beyond amyloid peptide and tau protein, it is important to emphasize the impact of co-pathologies on the emergence of dementia in humans, particularly at advanced ages [22, 24]. Some of these co-pathologies, such as TDP-43 protein inclusions, have never been described in NHPs and could therefore play a role in explaining human-specific features [29, 30].

Our findings call for caution when reading and understanding the results of transgenic rodent studies supporting the amyloid cascade, which are de facto affected by genetic differences and short lifespans (3–3.5 years maximum for mice and rats respectively). In transgenic mouse models overexpressing the APP protein, for instance, it is noteworthy that animals develop abundant amyloid plaques but no NFTs. This further suggests the existence of a temporal dissociation that prevents the concomitant occurrence of amyloid and tau pathology in short-lived species. NFTs are observed only in mice that additionally overexpress a mutated form of tau, imposing experimentally the biological scaling of the two proteinopathies [31].

From our perspective, it is important to emphasize the broad relevance of applying allometric approaches to the study of aging-related pathologies. These frameworks provide valuable insights into how pathological processes scale with lifespan and offer a novel way to interpret the progression of aging-related diseases across species. We envision that methodologies similar to ours could, in the future, be applied to diseases beyond AD.

Our study has several limitations. First, the sample sizes of NHPs in each age group are sometimes limited, likely making some frequency estimates uncertain. This also prevented us from studying other NHP species for which we did not have information across all age groups (we included in our analyses only the species for which we had at least n = 3 brains examined per decade, see methods). Second, although we considered as ‘NFT’ any lesion described as such (or as ‘tangle-like’) in the articles included in our literature review, it is important to note that the definition of NFT and their degree of maturity vary across laboratories, depending on the histological techniques, antibodies, and dyes used [32]. For example, a recent article (published after the cutoff date of our systematic review) showed in a cohort of 32 rhesus macaques that, while immunohistochemical methods allowed the identification of phospho-tau (AT8) positive lesions in some animals, silver staining or the use of Thioflavin S (a marker of mature tangles) did not reveal any lesions [33]. Indeed, while phospho-tau labelling may represent a precursor of NFTs (a pre-tangle), it can also be reversible. Thus, the occurrence of NFT reported in our manuscript in NHPs, although already rare, may in fact be overestimated. Third, we only considered the occurrence of lesions (amyloid plaques and NFTs), and not their location, spread, or density and we were unable to analyze the distinction between diffuse and neuritic plaques due to insufficient data. Such information is unfortunately rarely reported in animal studies, in contrast to human neuropathological studies. However, we were able to report in the Fig. 3 the highest lesion level documented at least once in each species, based on international neuropathological criteria (NIA-AA) [6]. Notably, no Thal phase 5 or Braak stage 6 has ever been reported in NHPs (Fig. 3). These data confirm the dissociation between amyloid and tau observed in NHPs that are phylogenetically distant from humans. Fourth, we generally lack behavioural information associated with the neuropathological status of the animals studied, which prevents us from establishing the clinico-biological link that can be made in humans. Although age-related cognitive decline has now been well documented in many monkey species [30], the notion of dementia is difficult to apply to any species other than humans, as it is based on the impairment of instrumental activities of daily human life.

Conclusion

Our study is the first to systematically review the literature reporting amyloid and tau pathology across different NHPs species and to compare these findings with available human data. We analysed these data using a novel conceptual approach in biology of aging, through the lens of biological scaling, a framework more commonly applied to developmental biology. We identified a temporal uncoupling across primate species in the emergence of amyloid plaques (isometric) and NFTs (chronometric). This challenges the amyloid cascade hypothesis as a universal and cross-species applicable biological mechanism.

Authors’ contributions

Investigation, Analysis, Visualization : C.T. Supervision, conceptualization, and methodology : M.L and V.P. Writing original draft : C.T. M.L and V.P. Discussion of the findings and revision for important intellectual content : E.B.

Funding

No Funding source.

Data availability

No new materials were generated for this study.

Declarations

Ethics approval and consent to participate

Not applicable.

Competing interests

E.B. is a director and shareholder of Motac Neuroscience Ltd. During the past 3 years, V.P. was a local unpaid investigator or sub-investigator for clinical trials granted by Novo Nordisk, Biogen, Janssen, and Alector. V.P. served as a consultant for Motac Neuroscience Ltd, outside the submitted work. Other authors declare no competing interests.