Plasma neurofilament light protein is differentially associated with age in individuals with treatment-resistant schizophrenia and bipolar affective disorder compared to controls

Accelerated brain ageing has been observed in multiple psychiatric disorders. This study examined whether relationships between age and plasma neurofilament light (NfL) protein differed in individuals with psychiatric disorders (major depressive disorder ( n = 42), bipolar affective disorder ( n = 121), treatment-resistant schizophrenia (TRS, n = 82)) compared to two healthy control (HC) groups ( n = 1,926 and n = 59). Compared to two independent HC samples, individuals with TRS demonstrated a stronger positive relationship between age and NfL levels. Individuals with BPAD had a stronger negative relationship between age and NfL levels compared to the large normative HC cohort, but not locally-acquired HCs. These findings show that plasma NfL levels are differentially associated with age in individuals with TRS and BPAD compared to healthy individuals.


Introduction
Structural and functional brain abnormalities have consistently been observed in psychiatric disorders such as depression, anxiety, bipolar disorder and schizophrenia (Bethlehem et al., 2022;Opel et al., 2020).Recent studies indicate that subtle neuroprogressive brain changes occur across the course of these illnesses, with one possible explanation for these findings being accelerated ageing (Koutsouleris et al., 2014).
Increased brain age gaps, a marker of accelerated brain ageing, have been observed in individuals with schizophrenia, major depression, borderline personality disorder, bipolar disorder, and in those at ultra high risk for psychosis (Koutsouleris et al., 2014;Tian et al., 2023).Across these studies, accelerated brain ageing was greatest in individuals with schizophrenia.These cross-sectional findings are consistent with a longitudinal study which found that individuals with schizophrenia not only had increased brain age at baseline, but also that the brain age gap increased over a 4-year follow-up period (Schnack et al., 2016).Together, these findings support the notion that individuals with serious mental illness experience accelerated brain ageing, with the greatest impact seen in those with schizophrenia.
Given evidence for accelerated brain ageing in psychiatric disorders, numerous studies have sought to determine the potential mechanisms and consequent biomarkers underlying this process.One possible biomarker is neurofilament light (NfL).NfL forms an essential component of the neuronal cytoskeleton particularly important for growth and stability of axons (Yuan et al., 2017).It also is a marker of potential neuronal injury.While some studies have observed increased NfL in schizophrenia and depression (Bavato et al., 2021;Chen et al., 2022), recent studies from our group using ultrasensitive technology to measure plasma NfL levels found no evidence of increased NfL in treatment-resistant schizophrenia (TRS) or major depressive disorder (MDD), but a slight increase in individuals with bipolar affective disorder (BPAD) (Eratne et al., 2023(Eratne et al., , 2022;;Jakobsson et al., 2014).However, only one study has examined whether age trajectories of NfL in psychiatric disorders differ from those seen in healthy individuals, where higher NfL levels have been associated with increasing age (Khalil et al., 2020).This study observed a positive relationship between age and NfL in depressed BPAD individuals (mean age 48 years) but not controls.Given findings of accelerated brain ageing in schizophrenia, BPAD, and MDD, it is possible that NfL levels would increase at a faster rate in individuals with psychiatric disorders compared to healthy individuals.If this is the case, NfL could potentially represent a theragnostic, stage, or prognostic marker (Davis et al., 2015) in psychiatry where there is a dearth of valid single marker assays.The aim of the current study was therefore to examine whether relationships between age and plasma NfL were increased in individuals with primary psychiatric disorders compared to healthy individuals.

Participants
The current study constitutes a secondary analysis of three patient cohorts and two control groups in which we have previously examined NfL levels.Eratne et al. (2023), Eratne, Janelidze and Malpas (2022) Participant samples and data have been described in detail previously.Berk et al. (2019), Dean et al. (2014), Eratne et al. (2023), Simrén et al. (2022), Wannan et al. (2019) All studies that contributed data and samples to this study, had ethical approval at the relevant Human Research Ethics Committees.All participants provided written informed consent prior to participation (Berk et al., 2019;Dean et al., 2014;Eratne et al., 2023;Simrén et al., 2022;Wannan et al., 2019).Patient cohorts included individuals from Australia with BPAD (n = 121), MDD; n = 42, and TRS, n = 82.Healthy control (HC) groups included a large reference normative control group derived from several European cohorts (Eratne et al., 2023;Simrén et al., 2022Simrén et al., ) (n = 1926) ) and a local sample from Melbourne, Australia (n = 59) with no current or past psychiatric or neurological illness.

Medication
In the TRS sample only, data relating to medication (clozapine dose, chlorpromazine equivalent dose), and serum clozapine levels were available.

Sample analysis
Plasma aliquots from all samples were stored at − 80 • C. Samples were randomised before analysis, and analyses blinded to diagnosis.Plasma NfL levels were measured on Quanterix SR-X and HD-X analysers using Simoa NF-Light kits, according to the manufacturer's recommendations.All patient samples and the local control group were analysed in Lund, Sweden.The large normative control group samples were analysed in Gothenburg, Sweden.

Data analysis
Data were analysed using R version 4.2.2.

Calculation of z-scores
NfL Z-scores were calculated from age-adjusted percentiles from the large reference cohort, derived using generalised additive models for location, scale, and shape (Eratne et al., 2023).

Examination of the impact of diagnosis on relationships between age and NfL
Primary analyses were conducted using the large reference normative control group.A general linear model (GLM) was used to examine diagnosis by age interactions on NfL z-scores, with HCs as the reference group and sex added as a covariate.Where significant diagnosis by age interactions were observed, two post-hoc analyses were performed.First, relationships between age and NfL levels were examined separately in the relevant diagnostic groups.Second, to determine the ages at which patient groups differed from HCs, we re-fit the GLMs using only the relevant patient and HC group, and estimated between-group contrasts at each age using the emmeans package (Lenth, 2022).
We also examined relationships between NfL levels and polynomial contrasts for age (up to the fourth order) to examine non-linear relationships, however, relationships were strongest and Bayesian information criteria lowest for raw age values and we therefore present these findings.
For all analyses, 95 % confidence intervals were computed (nonparametric bootstrapping, 1000 replicates), with statistical significance defined as any confidence interval not including the null (at 95 % level).
Secondary analyses were conducted using the local healthy control group in order to (1) examine the robustness of our findings using a distinct HC sample, and (2) determine the impact of body weight, which was not available in the large reference group, on the relationship between age and NfL, and diagnosis by age interactions.

Supplementary analyses
Relationships between NfL z-scores and in TRS individuals were further interrogated while adjusting for clozapine dose, chlorpromazine equivalent dose, and serum clozapine levels, adjusting for age and sex, in order to determine whether antipsychotic medication impacted on relationships between age and NfL levels.
Post hoc analyses revealed a positive correlation between NfL levels and age among individuals with TRS (β = 0.05 [0.01, 0.08], p = .005),reflecting an accelerated increase.Conversely, a negative correlation was found among individuals with BPAD (β = − 0.02 [− 0.04, − 0.01], p = .008),indicative of a slower increase in NfL levels with increasing age.After re-fitting the GLM in TRS individuals and HCs only, we observed that TRS individuals aged ≤36 had lower NfL levels compared to sameage HCs, whereas TRS individuals aged ≥50 had higher NfL levels compared to same-age HCs (Fig. 2A).After re-fitting the GLM in BPAD individuals and HCs, we found that BPAD individuals aged ≤53 had higher NfL levels compared to same-age HCs (Fig. 2B).

Discussion
Compared to two samples of healthy individuals, individuals with TRS demonstrated a stronger positive relationship between age and NfL levels.Younger TRS individuals (age ≤ 36) had lower NfL levels than HCs, whereas older TRS individuals (age ≥ 50) had higher NfL levels than controls.Higher than expected NfL levels in older individuals with TRS may provide further evidence for accelerated brain ageing in older individuals with treatment-resistant schizophrenia, consistent with a previous study showing accelerated ageing of white matter, the brain measure most commonly associated with NfL, only in older individuals with schizophrenia.The potential mechanisms driving lower NfL levels in younger TRS individuals compared to controls are unclear, particularly given the lack of studies examining NfL in younger populations and those without neurodegenerative disorders.However, it is possible that different mechanisms are impacting on NfL levels in the earlier vs later stages of illness: previous studies have found evidence for differential brain structural changes (Dietsche et al., 2017), inflammatory markers (Dickerson et al., 2016;Laskaris et al., 2016), and white matter microstructure (Cetin-Karayumak et al., 2020) in early vs late schizophrenia.Thus, while NfL increases in older TRS individuals may reflect accelerated ageing, consistent with previous neuroimaging studies, distinct neuropathological processes may be driving reductions in NfL in younger TRS individuals.
Compared to a large normative reference sample, individuals with BPAD demonstrated a stronger negative relationship between age and NfL levels, with increased NfL levels observed in BPAD individuals aged ≤53 relative to HCs.However, this attenuated age-NfL relationship was not replicated using a locally-acquired healthy control sample.Despite confirming increases in NfL in BPAD individuals that have been observed in previous studies (Eratne et al., 2023(Eratne et al., , 2022;;Jakobsson et al., 2014), our finding of an attenuated increase in NfL with age in this population are inconsistent with a previous study of individuals with depressed BPAD, which identified a positive relationship between age and NfL in patients but not controls (Aggio et al., 2022), and with previous findings of increased NfL levels only in BPAD individuals with an illness duration greater than 3 years (Queissner et al., 2024).Given these inconsistencies, it is clear that further research is needed to fully parse the role of NfL in BPAD.
A limitation of the current study is its cross-sectional design: although we can posit that stronger relationships between age and NfL are indicative of accelerated brain ageing in TRS, further longitudinal studies are needed to fully interrogate this hypothesis.Additionally, NfL values from our large normative healthy sample were obtained from a different laboratory than our psychiatric and locally-acquired healthy control groups, which may lead to site-related differences.However, our chosen assay has previously been shown to perform well across 17 sites, with intra-and inter-assay coefficients varying less than 10 % (Wilson et al., 2024).Validation of our TRS finding in our locally-acquired HCs also suggests that these findings are robust to any potential site differences.
In summary, this study found that differential relationships between age and NfL levels in both TRS and BPAD individuals compared to healthy controls.While higher NfL levels in older TRS individuals may support the notion of accelerated brain ageing as part of neuroprogressive processes in schizophrenia-spectrum disorders (Davis et al., 2014), further research is required to fully understand relationships

Declaration of competing interest
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors acknowledge the financial support of the CRC for Mental Health.The Cooperative Research Centre (CRC) programme is an Australian Government Initiative.The authors acknowledge the CRC Scientific Advisory Committee, in addition to the contributions of study participants, clinicians at recruitment services, staff at the Murdoch Children's Research Institute, staff at the Australian Imaging, Biomarkers and Lifestyle Flagship Study of Aging, and research staff at the Melbourne Neuropsychiatry Centre, including coordinators Merritt, A., Phassouliotis, C., and research assistants, Burnside, A., Cross, H., Gale, S., and Tahtalian, S. Participants for this study were sourced, in part, through the Australian Schizophrenia Research Bank (ASRB), which is supported by the National Health and Medical Research Council of Australia (Enabling Grant N. 386500), the Pratt Foundation, Ramsay Health Care, the Viertel Charitable Foundation and the Schizophrenia Research Institute.We thank the Chief Investigators and ASRB Manager: Carr, V., Schall, U., Scott, R., Jablensky, A., Mowry, B., Michie, P., Catts, S., Henskens, F., Pantelis, C., Loughland, C. We acknowledge the help of Jason Bridge for ASRB database queries.The authors are grateful for assistance from Brett Trounson and Dr Christopher Fowler and the team at The Florey Oak St Biobank.

Fig. 2 .
Fig. 2. Group differences in NfL at each age between (A) treatment-resistant schizophrenia and healthy controls and (B) bipolar affective disorder and healthy controls based on estimated marginal means.Blue dots indicate significant differences.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)