Received: 4 June 2020 | Revised: 21 August 2020 | Accepted: 2 September 2020

DOI: 10.1111/jnc.15183

ORIGINAL ARTICLE

Plasma isoprostanoids assessment as Alzheimer’s disease

progression biomarkers

Carmen Peña-Bautista1 | Lourdes Álvarez2 | Miguel Baquero2 | Inés Ferrer2 |

Lorena García2 | David Hervás-Marín3 | Consuelo Cháfer-Pericás1

1
Neonatal Research Unit, Health Research
Institute La Fe, Valencia, Spain Abstract
2
Neurology Unit, University and Polytechnic Alzheimer's Disease (AD) is the most common neurodegenerative disease worldwide.
Hospital La Fe, Valencia, Spain
So, there is a need to identify AD early diagnosis and monitoring biomarkers in blood
3
Biostatistics Unit, Health Research Institute
La Fe, Valencia, Spain
samples. The aim of this study was to analyse the utility of lipid peroxidation bio-
markers in AD progression evaluation. Participants (n = 19) were diagnosed with AD
Correspondence
Consuelo Cháfer-Pericás, PhD Health
at early stages (Time 0, T0), and they were re-evaluated 2 years later (Time 1, T1).
Research Institute La Fe, Avenida Fernando Plasma biomarkers from AD patients were determined at both times. Some analytes,
Abril Martorell 106, Valencia E46026, Spain.
Email: [Link]@[Link]
such as dihomo-isoprostanes (17-epi-17-F2t-dihomo-IsoP, 17-F2t-dihomo-IsoP, Ent-
7(RS)-7-F2t-dihomo-IsoP), and neuroprostanes (10-epi-10-F4t-NeuroP) showed very
FUNDING INFORMATION
This work was supported by the Instituto
high probability of showing an increasing trend over time. Baseline values allowed to
de Salud Carlos III (Miguel Servet I Project develop an affordable preliminary regression model to predict long-term cognitive
(CP16/00082) (Spanish Ministry of Economy
and Competitiveness).
status. So, some lipid peroxidation biomarkers would deserve consideration as useful
progression AD biomarkers. The developed prediction model would constitute an
important minimally invasive approach in AD personalized prognosis and perhaps
could have some interest also in experimental treatments evaluation.

KEYWORDS

Alzheimer's disease, biomarker, isoprostane, oxidative stress, plasma

1 | I NTRO D U C TI O N (Guzman-Martinez et al., 2019). In this sense, follow-up studies have

been carried out looking for reliable prognosis biomarkers. As found
Alzheimer's Disease (AD) is the most common cause of dementia in literature, most studies are focused on CSF biomarkers, whereas a
worldwide. Nevertheless, effective treatments and minimally invasive few of them assayed blood biomarkers reflecting some advantages in
early diagnosis are not available, yet (Arvanitakis et al., 2019). Current minimally invasive samples (Seppala et al., 2010; Younes et al., 2019).
diagnosis is based on neuroimaging, neuropsychological evaluation Among potential prognosis biomarkers, amyloid peptides and
and cerebrospinal fluid (CSF) biomarkers. So, it can be considered tau protein are the most studied compounds, but also lipids or in-
expensive and invasive (Jack et al., 2018). In addition, there is a lack flammation biomarkers have been evaluated as potential progno-
of AD monitoring biomarkers, which would be necessary in clinical sis biomarkers (Dorninger et al., 2018; Pedrini et al., 2017; Toledo
prognosis evaluation, as well as for further treatments assessment. et al., 2013). Regarding CSF biomarkers, some contradictory results
Recent studies in AD biomarkers are focused in early diagnosis, were observed. For instance, tau increment was observed at 2-year
but more attention is required on physiopathological progression AD development, whereas the levels for β-amyloid did not show

Abbreviations: AD, Alzheimer's disease; CDR, clinical dementia rating; CEIC, ethics committee; CMGA, chromogranin A; CSF, cerebrospinal fluid; DHA, docosahexaenoic acid; EDTA,
ethylenediaminetetraacetic acid; MCI-AD, mild cognitive-impairment due to AD; MMSE, mini-mental state examination; NIA-AA, national institute on aging-Alzheimer's association;
NPTXR, neuronal pentraxin receptor; NrCAM, neuronal cell adhesión molecule; PET, positron emission tomography; [Link], Repeatable battery for the assessment of
neuropsychological status-delayed memory.

Journal of Neurochemistry. 2020;00:1–8. [Link]/journal/jnc

© 2020 International Society for Neurochemistry | 1
2 | PEÑA-BAUTISTA et al.

significant differences along this period (Buchhave et al., 2009). In Memory ([Link]) ≤85) following the National Institute on
addition, amyloid precursor protein peptide A4-117, neuronal pen- Aging-Alzheimer's Association (NIA-AA) criteria (Albert et al., 2011;
traxin receptor (NPTXR), chromogranin A, and neuronal cell adhe- McKhann et al., 2011). Some of these participants maintain an ac-
sion molecule (NrCAM) showed some decrement over time in AD ceptable function on daily living activities with CDR ≤1. So, they
(Wildsmith et al., 2014). Similarly, CSF tau, p-tau and β-amyloid were form a suitable subgroup of patients to be included in the 2-year
determined in several studies showing their levels impaired along follow-up study. Regarding exclusion criteria, patients with comor-
10–30 years before clinical symptoms (Younes et al., 2019). bidity (tumour, stroke, major head trauma, epilepsy, major psychi-
As regards minimally invasive samples (e.g. plasma, serum, etc.), atric disorders) were excluded, as well as patients with CDR >1 at
few studies looking for prognosis biomarkers can be found in lit- the diagnosis time, since they would not able to undergo neuropsy-
erature. In this sense, Pedrini et al found that levels of Eotaxin-3, chological evaluations at 2-year period. Also, some participants were
and peptides TARC and YY showed an increase tendency along AD excluded at the second evaluation time (2 years after) since their
development, specifically between 18–54 months from diagnosis clinical symptoms avoid the neuropsychological evaluation (CDR >2).
(Pedrini et al., 2017). Finally, depending on the degree of cognitive symptoms, partici-
Nowadays, there is an increasing need for AD progression bio- pants included in this study (n = 19) were patients initially diagnosed
markers in minimally invasive samples, since it is required to evaluate with mild cognitive-impairment due to AD (MCI-AD) (n = 13), they
efficient and personalized treatments. Our previous study in plasma showed cognitive complaints without daily living activities impair-
samples revealed the potential utility of isoprostanoids for early ment (CDR 0–0.5); as well as patients diagnosed with mild demen-
AD diagnosis (Peña-Bautista et al., 2018). In fact, these compounds tia-AD (n = 6), they showed minor daily living activities impairment
could play an important role in the development of neurodegener- (CDR 0.5–1). All of them were able to carry out the neuropsychologi-
ation (Peña-Bautista et al., 2019). It would be explained by the high cal evaluation at 2 years (see scheme S1 in Supplementary Material).
lipid composition of the brain, as well as its high oxygen consumption Plasma samples were obtained from all the participants. For this,
(Sultana et al., 2013). The analytes studied here were chosen be- whole-blood samples were collected in cryotubes with ethylenedi-
cause they had been previously shown to be quantitatively different aminetetraacetic acid at the diagnosis time (time 0, T0) and 2 years
in AD patients compared with controls (Peña-Bautista et al., 2018; later (time 1, T1). Blood samples were centrifuged obtaining plasma
Peña-Bautista, et al., 2019), allowing the development of a potential samples that were stored, with butyl hydroxytoluene as antioxidant,
diagnosis approach. However, the corresponding progression utility at −80ºC until the analysis.
and stability along time of these isoprostanoids biomarkers have not
been established. The aim of this study was to evaluate lipid peroxi-
dation biomarkers evolution, in plasma samples from accurately and 2.2 | Plasma isoprostanoids determination
early diagnosed AD patients, along a 2-year follow-up period.
Isoprostanoids from plasma samples were determined by liquid
chromatography coupled to tandem mass spectrometry (MS), as
2 | M ATE R I A L A N D M E TH O DS it was previously described by Peña-Bautista et al. (Peña-Bautista
et al., 2018). The targeted metabolites were quantified using frag-
2.1 | Participants recruitment and samples mentation and multiple reaction monitoring, and referencing each
collection metabolite to true standards at known concentrations. The data
were acquired and processed using the MassLynx 4.1 and QuanLynx
The study protocol was approved by the Ethics Committee (CEIC) 4.1 (Waters) softwares respectively. Peak area integration was used
from Health Research Institute La Fe (Valencia, Spain) (refer- to quantify the analytes. Regarding sample treatment, hydrolysis,
ence number 2016/0257). The study was not pre-registered. The pre-concentration and purification by solid phase extraction were
methods were carried out in accordance with relevant guidelines carried out. The experimenter was unaware of the type of partici-
and regulations, and informed consent was obtained from all par- pant during experimentation.
ticipants. No randomization was performed to allocate subjects in
the study. No sample calculation was performed. The enrolment
period was 1 year, participants were between 50 and 80 years old, 2.3 | Statistical analysis
they manifested memory impairment, and they were subjected to
standard AD diagnosis tests (n = 130). Some of these participants Data were summarized using mean (standard deviation) and median
showed positive results for amyloid PET or CSF biomarkers (amyloid (1st, 3rd quartile) in the case of continuous variables and relative and
β < 700, tau >350, or tau/amyloid β > 0,51) as first inclusion criteria, absolute frequencies in the case of categorical variables. To assess
so they were diagnosed with AD (n = 74), and they were initially the trend of the different isoprostanoids over time Bayesian ordinal
included in the study. In addition, they showed altered neuropsy- models were adjusted. These models included a random intercept
chological evaluation (clinical dementia rating (CDR) ≥0; Repeatable for each individual to control dependency among observations. For
Battery for the Assessment of Neuropsychological Status- Delayed each model the null hypothesis was assessed by computing a Bayes
PEÑA-BAUTISTA et al. | 3

factor between the hypothesis and its alternative via the Savage- TA B L E 1 Clinical and demographic variables for the participants
Dickey density ratio method (Verdinelli & Wasserman, 1995).
Participants
A prediction model for MMSE values at T1 based on MMSE values Variables group (n = 19)
and isoprostanoids concentrations at T0 was developed by perform-
Age (years, median (IQR)) 70 (69, 74)
ing an elastic net penalized ordinal regression model. Parameters
Gender (female, n (%)) 10 (53)
of the model were selected using 500 repetitions of 10-fold cross
Initial diagnosis MCI-AD 13 (68)
validation. The variables selected by the elastic net were then in-
(n (%)) Dementia-AD 6 (32)
troduced in a non-penalized multivariable regression model which
Smoking1 (n (%)) 9 (47)
included also baseline MMSE values as a covariate. Performance
of this model was assessed by estimating the R-squared value. For Alcohol consumption (n (%)) 1 (5)

all estimates a 95% credibility interval was computed. All statisti- Genotype (n (%)) ε3/ε3 7 (37)

cal analyses were performed using R (version 3.6.3) and R packages ε3/ε4 10 (53)
brms (version 2.12.2), rms (version 5.1-4) and clickR (version 0.4.47). ε4/ε4 2 (10)
Treatments (n (%)) Statins 10 (53)
Benzodiazepines 3 (16)
3 | R E S U LT S Antihipertensive 10 (53)
Corticoids 1 (5)
3.1 | Demographic and clinic participants
Comorbidity (n (%)) Dyslipidaemia 10 (53)
description
Diabetes 2 (11)
Hypertension 9 (47)
Participants’ demographic and clinical data are shown in Table 1.
Heart Disease 1 (5)
Participants showed a median age of 70, and about half of partici-
pants were females (n = 10). Regarding medication, statins and an- Education level (n (%)) Primary 9 (53)

tihypertensive were used for almost a half of participants at T0, and Secondary 2 (12)

their use was equivalent in both groups. In fact, they were consid- University 6 (35)
ered chronic treatments for them. However, few participants took β-Amyloid (pg/mL, median (IQR)) 600 (516,
benzodiazepines and corticosteroids at T0, and also few participants 696))

initiated the corresponding chronic treatments along the 2-year t-Tau (pg/mL, median (IQR)) 541 (401, 621)
period. p-Tau (pg/mL, median (IQR)) 81 (72, 91)
Plasma samples were obtained from all participants at the diagno- CDR (median (IQR)) 0.5 (0.5, 0.5)
sis time (T0) and at 2.04 (±0.14) years (T1). At T0, samples were ob- MMSE (median (IQR)) 25 (23, 27)
tained from participants initially diagnosed with dementia-AD (n = 6) [Link] (median (IQR)) 48 (40, 64)
and with MCI-AD (n = 13). During the 2 years between T0 and T1, AD 1
Yes or along more than 10 years.
patients suffer from some progression of the disease symptoms. This
progression was evaluated from neuropsychological tests, mainly with
CDR scale. At T1 some participants from the larger group (MCI-AD) ordinal regression models were used comparing T0 and T1 concen-
remained at CDR 0.5 (n = 7), and some showed progression to CDR trations for each compound. The probability for the hypothesis of
1 (n = 5) and CDR 2 (n = 1); from the CDR 0 group both participants progression was estimated in each case, along with the evidence
showed progression to CDR 0.5 (n = 2); from the CDR 1 group some ratio and the estimate with its 95% credible interval. All results
participants remained (n = 3) and one participant showed progression are presented in Table 3. To visualize the results, a representation
to CDR 2. In addition, Mini Mental State Examination (MMSE) scores of the estimates for the difference between T0 and T1 is provided
reflect approximately this evolution, at T0 they showed median levels (Figure 1). Positive values corresponded to an increase along time,
of 25 (23, 27) and at T1 23 (17, 26). Regarding RBANS test, which is the whereas negative values corresponded to a decrease. Additionally,
most indicative test of the clinical progression, some patients were not the progression of each isoprostanoid for each participant has been
able to be explored with this tool at T1. represented (Figure S2). In general, it was observed a slight increase
in isoprostanoids levels in plasma from AD diagnosed patients in a
2-year period. Specifically, the dihomo-isoprostanes (17-epi-17-F2t-
3.2 | Isoprostanoids stability dihomo-IsoP, 17-F2t-dihomo-IsoP, Ent-7(RS)-7-F2t-dihomo-IsoP),
the prostaglandins (PGE2, PGF2α) and the isoprostane (5-F2t-IsoP)
The isoprostanoids stability or progression along AD development showed a very high probability of showing an increasing trend
were evaluated for each compound. For this, the baseline concentra- over time; also some neuroprostanes showed very high probability,
tion levels of the different isoprostanoids are presented in Table 2. To 10-epi-10-F4t-NeuroP, whereas 14(RS)-14-F4t-NeuroP decreased
assess the stability or progression along AD development, Bayesian along time (Figure 1). However, other compounds showed quite
4 | PEÑA-BAUTISTA et al.

TA B L E 2 Baseline analytes
Time 0 (T0) Time 1 (T1)
concentrations in plasma samples from
Median (IQR) participants
Analytes Median (IQR) (nmol/L) (nmol/L)

15(R)−15-F2t-IsoP 0.30 (0.25, 0.35) 0.28 (0.18, 0.64)

PGE2 0.05 (0, 0.1) 0.38 (0, 0.54)
2,3-dinor−15-epi−15-F2t-IsoP 0 (0,0.02) 0 (0,0)
15-keto−15-E2t-IsoP 0.2 (0.1, 0.36) 0 (0, 0.27)
15-keto−15-F2t-IsoP 0.22 (0.05, 0.3) 0.31 (0.15, 0.96)
15-E2t-IsoP 0.3 (0.11, 0.41) 0.39 (0.19, 0.98)
5-F2t-IsoP 0.55 (0.35, 1.04) 1.44 (1.05, 3.43)
PGF2α 0.48 (0.26, 0.7) 0.81 (0.6, 2.01)
4(RS)-F4t-NeuroP 1.1 (1.02, 1.25) 0 (0, 3.68)
10-epi−10-F4t-NeuroP 0.08 (0.02, 0.12) 0.55 (0.46, 0.63)
14(RS)−14-F4t-NeuroP 0.55 (0.25, 1.01) 0 (0, 0)
Isoprostanes1 0.30 (0.21, 0.35) 0.22 (0.19, 0.55)
Ent−7(RS)−7-F2t-dihomo-IsoP 0.08 (0.08, 0.12) 0.89 (0.83, 0.97)
17-F2t-dihomo-IsoP 0 (0,0) 0.3 (0.23, 0.54)
17-epi−17-F2t-dihomo-IsoP 0.02 (0, 0.02) 0.21 (0.04, 0.49)
17(RS)−10-epi-SC-Δ1 −11-dihomo-IsoF 0 (0,0) 0 (0, 0)
8
7(RS)-ST-Δ −11-dihomo-IsoF 0.02 (0.01, 0.06) 0 (0, 0.58)
Isofurans1 0.08 (0.07, 0.11) 0.08 (0.06, 0.22)

Note: IQR, inter-quartile range.

1
Total parameters, arbitrary units: (intensity of signal units × (internal standard concentration,
nmol/L).

TA B L E 3 Estimates, evidence ratios

EV for PP for
and posterior probabilities for the
Isoprostanoid Estimate CI (95%) HA HA
hypotheses that isoprostanoids have a
14(RS)-14-F4t-NeuroP −8.40 −12.32 −4.91 >1,000 100% trend over time (HA)
15
17(RS)-10-epi-SC-Δ -11-dihomo-IsoF −3.14 −7.16 0.55 2.3 69.3%
2,3-dinor-15-epi-15-F2t-IsoP −2.60 −5.29 −0.40 5 83.2%
15-keto−15-E2t-IsoP −1.64 −3.00 −0.35 5.2 83.8%
4(RS)-F4t-NeuroP −1.42 −2.90 −0.01 1.4 57.7%
Isoprostanes −1.35 −2.71 0.00 1.4 57.9%
7(RS)-ST-Δ8−11-dihomo-IsoF −1.12 −2.51 0.23 0.85 46.1%
15(R)−15-F2t-IsoP −0.39 −1.56 0.78 0.26 20.4%
Isofurans 0.07 −1.18 1.36 0.22 17.9%
15-E2t-IsoP 1.24 0.00 2.49 1.3 57.1%
15-keto−15-F2t-IsoP 1.48 0.15 2.80 2.4 70.7%
Ent−7(RS)−7-F2t-dihomo-IsoP 3.59 1.91 5.40 >1,000 100%
PGE2 4.29 2.35 6.40 >1,000 100%
PGF2α 4.61 2.99 6.36 >1,000 100%
17-epi−17-F2t-dihomo-IsoP 4.91 2.87 7.28 >1,000 100%
17-F2t-dihomo-IsoP 5.82 3.57 8.72 >1,000 100%
5-F2t-IsoP 6.10 4.27 8.03 >1,000 100%
10-epi−10-F4t-NeuroP 10.41 7.81 13.25 >1,000 100%

Note: CI, credible interval; EV, evidence ratio; PP, posterior probability.
PEÑA-BAUTISTA et al. | 5

10-epi-10-F4t-NeuroP
5-F2t-IsoP
17-F2t-dihomo-IsoP
17-epi-17-F2t-dihomo-IsoP
PGF2α
PGE2
Ent-7(RS)-7-F2t-dihomo-IsoP
15-keto-15-F2t-IsoP
15-E2t-IsoP
Isofurans
15(R)-15-F2t-IsoP
7(RS)-ST-Δ8-11-dihomo-IsoF
Isoprostanes
4(RS)-F4t-NeuroP
15-keto-15-E2t-IsoP
2,3-dinor-15-epi-15-F2t-IsoP
17(RS)-10-epi-SC-Δ15-11-dihomo-IsoF
14(RS)-14-F4t-NeuroP

F I G U R E 1 Estimates and corresponding 95% credible intervals for the difference between T0 and T1 for each isoprostanoid. Positive
values correspond to an increase along time, whereas negative values correspond to a decrease

stability in a 2-year period of time (15(R)-15-F2t-IsoP, 7(RS)-ST-Δ8- ratio test (p = .019). The validation of the model yielded an optimism
11-dihomo-IsoF, Isofurans, 15-E2t-IsoP, Isoprostanes). In addition, a corrected R-squared of 0.5.
great variability among participants was observed for the levels of The equation for predicting MMSE at T1 with baseline MMSE
some compounds such as 15-keto-15-E2t-IsoP, 15-keto-15-F2t-IsoP and 10-epi-10-F4t-NeuroP values is the following:
and 4(RS)-F4t-NeuroP. Moreover, the tendency in biomarkers levels
during the 2-year period did not show differences between MCI-AD MMSE (T1) = 2.25 + 1.03 ∗ MMSE (T0) + 2.29 ∗ log(10 − epi − 10 − F4t − NeuroP (T0) )

and dementia patients (Figure S2). Also, the correlation between lev-
els of each biomarker at T0 and the 2-year variation in the cognitive
impairment (CDR) was not statistically significant in any case. 4 | D I S CU S S I O N

This study is focused on the identification of potential progression

3.3 | Isoprostanoids and Alzheimer's variables in AD different from the previously known and intuitive
disease evolution variables with prognostic value. In general, known prognostic fac-
tors (e.g. age, neuropsychology, neuroimaging, glucose PET) could
Two years after the diagnosis (T1), participants were re-evaluated be described in a directly intuitive way, ‘as worse findings worse
by the clinicians to determine if the pathology was stable or with prognosis’. Moreover, other variables, such as t-Tau and p-Tau in CSF
progression in patients’ cognitive performance. For this, participants have been related to prognosis, and possibly they could be deter-
were evaluated by the clinicians at T1 by means of CDR and MMSE mined in blood samples with high reliability and similar utility. In this
tests. Association between neuropsychological tests progression sense, our work yields novel biomarkers with added prognostic value
and biomarkers was assessed using linear models including base- without previous intuitive appearance.
line MMSE values as covariate. From this, a predictive model with Participants in this study were accurately diagnosed with AD from
10-epi-10-F4t-NeuroP was adjusted, achieving an R-squared value of CSF biomarkers. They showed impaired CDR, RBANS and MMSE
0.6 (Table 4). The improvement of this model over a simpler model scores at T0. These values evolved along time, reflecting deterioration
including only baseline MMSE values was assessed using a likelihood of cognitive status 2 years later. This fact would explain the difficulty
6 | PEÑA-BAUTISTA et al.

TA B L E 4 Predictive model for MMSE

Variables Estimate SE Lower.95. Upper.95. p value
at T1 based on the baseline MMSE and
(Intercept) 2.25 6.8 −12.1 16.6 0.74 10-epi-10-F4t-Neuro p values
MMSE (T0) 1.03 0.25 0.5 1.56 <0.001
log(10-epi-10-F4t-NeuroP) 2.29 0.88 0.44 4.15 0.019
(T0)
R2 0.60
Adj.R Squared 0.55

associated to neuropsychological evaluation in a longitudinal study. et al., 2009). In addition, in a 7-year assessment period there was
Therefore, the evaluation at T1 was limited to CDR and MMSE in order not observed a clear trend, so the course of the disease could not be
to reduce considerably the neuropsychological evaluation time. predicted with the CSF biomarkers (Wang, et al., 2018). However,
A 2-year follow-up study was carried out, evaluating isopros- a few compounds, not showing diagnosis ability in previous work
tanoids stability and long-term potential clinical prediction. Some (Peña-Bautista, et al., 2019), did show a progression during the
isoprostanoids determined in plasma samples from MCI-AD and evaluated 2-year period in this study. In fact, potential AD progres-
dementia-AD participants showed an increase in their levels along sion biomarkers could be PGE2, 5-F2t-IsoP, 10-epi-10-F4t-NeuroP,
a 2-year assessment period. Moreover, one of these isoprostanoids 17-epi-17-F2t-dihomo-IsoP, 17-F2t-dihomo-IsoP, Ent-7(RS)-7-F2t-
together with baseline cognitive status showed some clinical predic- dihomo-IsoP, PGF2α, 14(RS)-14-F4t-NeuroP. In this sense, Baldeiras
tion capacity. et al. found different levels of oxidative stress biomarkers, such as
In a previous work it was demonstrated the capacity of lipid per- antioxidants and lipid and protein oxidation biomarkers in MCI pa-
oxidation biomarkers in order to distinguish between early AD pa- tients that progressed in a 2-year period to AD from those patients
tients and controls from plasma samples (Peña-Bautista et al., 2018). that did not progress (Baldeiras et al., 2008). Probably, the different
However, it was not clear the long-term stability for these com- results found in literature could be explained by the different diag-
pounds, as well as their potential utility as AD progression or prog- nosis criteria used in each study. Most of works in literature iden-
nosis biomarkers. In this sense, a longitudinal study was initiated tified AD patients using different neuropsychological scales, and
from some patients diagnosed with AD 2 years before (MCI or mild few of them used CSF biomarkers. This study shows the strength
dementia). For this, the previously validated analytical method was of biological definition (CSF biomarkers) to identify accurately the
applied (García-Blanco et al., 2018; Peña-Bautista, et al., 2019), and AD patients.
the lipid peroxidation compounds were determined with high sensi- Regarding long-term cognitive status prediction, a prelimi-
tivity and specificity from plasma samples collected at two different nary regression model was developed from 10-epi-10-F4t-NeuroP
points along the AD development in accurately selected partici- and MMSE values at diagnosis time, reflecting some relationship.
pants. From literature, few studies have focused on AD progression Nevertheless, a previous study did not found correlations between
assessment, and most of them employed CSF samples (Hendrickson MMSE score variations and oxidative stress biomarkers (Baldeiras
et al., 2015; Wang, et al., 2018; Wildsmith et al., 2014), whereas et al., 2008). In fact, they only found slight correlations between
few of them focused on blood samples (Faura et al., 2020; Seppala neuropsychological evaluation and antioxidants plasma levels. In
et al., 2010). In addition, most of them focused on biomarkers such spite of the CDR utility in the function evaluation depending on
as, amyloid peptides, tau proteins and their derivatives (Lauridsen the cognitive impairment, it was not useful in the present 2-year
et al., 2016; Younes et al., 2019). However, there is an increasing re- term evaluation since it shows high stability along 2 years. Other
search postulating oxidative stress and lipid biomarkers as potential kind of biomarkers, such as miRNAs, showed correlation with neu-
biomarkers for AD (Bradley et al., 2012; Weng et al., 2019). ropsychological evaluation but these differences were found in
According to the evaluated compounds that showed high di- a sectional study (Wang et al., 2020). Regarding predictive mod-
agnosis capacity at early AD stages, such as 15(R)-15-E2t-IsoP, elling of AD progression, Wang et al. approach was based on the
15-E2t-IsoP and 4(RS)-F4t-NeuroP (Peña-Bautista et al., 2018), they inherent temporal and medical patterns derived from patients' his-
did not represent potential AD progression biomarkers, since their torical visits using recurrent neural networks (Wang et al., 2018).
values were quite stable or they did not evolve homogeneously for Other models in literature were based on both cognitive scores
all the participants at similar clinical point. Similarly, as Cervellati and neuroimaging measures to predict MCI conversion to AD
et al described, oxidative stress biomarkers such as hydroperoxides (Minhas et al., 2018), or long-term clinical scores (Yang et al., 2019).
and antioxidant capacity were good in early AD diagnosis when pa- However, to our knowledge this is the first study evaluating AD
tients were in MCI stage, but they could not discriminate between progression from early stages using one plasma biomarker and one
patients progressing or not to AD (Cervellati et al., 2014). Standard neuropsychological test. Finally, this study showed some limita-
biomarkers in AD diagnosis (amyloid, tau and p-tau) do also main- tions, such as, the small number of participants. However, only 30%
tain their levels relatively stable in a 2-year period time (Buchhave of the participants in this follow-up study were able to carry out the
PEÑA-BAUTISTA et al. | 7

neuropsychological evaluation at 2-year term. Another limitation is Alzheimer’s disease. Alzheimer's & Dementia, 7, 270–279. [Link]
org/10.1016/[Link].2011.03.008
the fact that subjects could receive treatments and antioxidants
Arvanitakis, Z., Shah, R. C., & Bennett, D. A. (2019). Diagnosis and
supplements (e.g. docosahexaenoic acid) along the 2-year assess- management of dementia: Review. JAMA, 322, 1589. [Link]
ment period, influencing the levels of isoprostanes determined in org/10.1001/jama.2019.4782
plasma. Finally, some of the evaluated compounds (prostaglandins) Baldeiras, I., Santana, I., Proença, M. T., Garrucho, M. H., Pascoal, R.,
Rodrigues, A., Duro, D., & Oliveira, C. R. (2008). Peripheral oxidative
are inflammation biomarkers, so further studies evaluating their in-
damage in mild cognitive impairment and mild Alzheimer’s disease.
crease in some patients with potential worse AD prognosis should Journal of Alzheimer's Disease, 15, 117–128. [Link]
be carried out. JAD-2008-15110
In conclusion, Alzheimer's disease is a pathology with a long pro- Bradley, M. A., Xiong-Fister, S., Markesbery, W. R., & Lovell, M. A. (2012).
gression time, maintaining its clinical feature quite stable for several Elevated 4-hydroxyhexenal in Alzheimer’s disease (AD) progression.
Neurobiology of Aging, 33, 1034–1044.
years. In this sense, some lipid peroxidation compounds are quite sta-
Buchhave, P., Blennow, K., Zetterberg, H., Stomrud, E., Londos, E.,
ble and do not show a clear tendency in a 2-year period. In general, Andreasen, N., Minthon, L., & Hansson, O. (2009). Longitudinal study
only a few plasma isoprostanoids showed some tendency in that pe- of CSF biomarkers in patients with Alzheimer’s disease. PLoS One, 4,
riod, generally increasing, so they could be considered progression e6294. [Link]
Cervellati, C., Romani, A., Seripa, D., Cremonini, E., Bosi, C., Magon,
AD biomarkers. From this, and considering baseline cognitive status,
S., Bergamini, C. M., Valacchi, G., Pilotto, A., & Zuliani, G. (2014).
a potential long-term prediction model was developed. It could con- Systemic oxidative stress and conversion to dementia of elderly pa-
stitute an important minimally invasive approach in AD personalized tients with mild cognitive impairment. BioMed Research International,
prognosis, as well as in future therapeutic treatment evaluation or 2014, 1–7.
Dorninger, F., Moser, A. B., Kou, J., Wiesinger, C., Forss-Petter, S., Gleiss,
monitorization. Nevertheless, further research about lipid peroxi-
A., Hinterberger, M., Jungwirth, S., Fischer, P., & Berger, J. (2018).
dation biomarkers and AD progression evaluation should be carried Alterations in the plasma levels of specific choline phospholipids in
out, including a high number of participants accurately identified Alzheimer’s disease mimic accelerated aging. J. Alzheimer’s Dis., 62,
(e.g. PET amyloid, CSF biomarkers) as AD patients. Specifically, it 841–854. [Link]
Faura, J., Bustamante, A., Penalba, A., Giralt, D., Simats, A., Martínez-
would be required to carry out validation studies, which would allow
Sáez, E., Alcolea, D., Fortea, J., Lleó, A., Teunissen, C. E., van der Flier,
to corroborate the present preliminary results. W. M., Ibañez, L., Harari, O., Cruchaga, C., Hernández-Guillamón,
M., Delgado, P., & Montaner, J. (2020). CCL23: A chemokine associ-
AC K N OW L E D G E M E N T S ated with progression from mild cognitive impairment to Alzheimer’s
disease. J. Alzheimer’s Dis., 73, 1585–1595. [Link]
CC-P acknowledges a postdoctoral “Miguel Servet” grant
JAD-190753
CP16/00082 from the Health Research Institute Carlos III (Spanish García-Blanco, A., Peña-Bautista, C., Oger, C., Vigor, C., Galano, J.-M.,
Ministry of Economy, Industry and Innovation). CP-B acknowledges Durand, T., Martín-Ibáñez, N., Baquero, M., Vento, M., & Cháfer-
a predoctoral contract associated to the Miguel Servet project Pericás, C. (2018). Reliable determination of new lipid peroxidation
CP16/00082, from the Health Research Institute Carlos III (Spanish compounds as potential early Alzheimer Disease biomarkers. Talanta,
184, 193–201. [Link] a.2018.03.002
Ministry of Economy, Industry and Innovation). The authors are
Guzman-Martinez, L., Maccioni, R. B., Farías, G. A., Fuentes, P., &
grateful for the synthesis of the lipid peroxidation compounds by Navarrete, L. P. (2019). Biomarkers for Alzheimer’s disease. Current
Professor Durand's team at the Institute des Biomolécules Max Alzheimer Research, 16, 518–528. [Link]
Mousseron (IBMM) (Montpellier, France). 05016​66619​05171​21140
Hendrickson, R. C., Lee, A. Y. H., Song, Q., Liaw, A., Wiener, M., Paweletz,
All experiments were conducted in compliance with the ARRIVE
C. P., Seeburger, J. L., Li, J., Meng, F., Deyanova, E. G., Mazur, M. T.,
guidelines. Settlage, R. E., Zhao, X., Southwick, K., Du, Y. I., Holder, D., Sachs, J.
R., Laterza, O. F., Dallob, A., & Yates, N. A. (2015). High resolution
D I S C LO S U R E discovery proteomics reveals candidate disease progression markers
of Alzheimer’s disease in human cerebrospinal fluid. PLoS One, 10,
The authors declare not having anything to disclose in relation with
e0135365. [Link]
the present manuscript. Jack, C. R., Bennett, D. A., Blennow, K., Carrillo, M. C., Dunn, B.,
Haeberlein, S. B., Holtzman, D. M., Jagust, W., Jessen, F., Karlawish,
ORCID J., Liu, E., Molinuevo, J. L., Montine, T., Phelps, C., Rankin, K. P.,
Rowe, C. C., Scheltens, P., Siemers, E., Snyder, H. M., & Silverberg, N.
Miguel Baquero [Link]
(2018). NIA-AA research framework: Toward a biological definition
Consuelo Cháfer-Pericás [Link] of Alzheimer’s disease. Alzheimer's & Dementia, 14, 535–562. https://
org/0000-0002-0711-5877 [Link]/10.1016/[Link].2018.02.018
Lauridsen, C., Sando, S. B., Shabnam, A., Møller, I., Berge, G., Grøntvedt,
G. R., Bakken, I. J., Salvesen, Ø., Bråthen, G., & White, L. R. (2016).
REFERENCES
Cerebrospinal fluid levels of amyloid beta 1–43 in patients with
Albert, M. S., DeKosky, S. T., Dickson, D., Dubois, B., Feldman, H. H.,
amnestic mild cognitive impairment or early Alzheimer’s disease: A
Fox, N. C., Gamst, A., Holtzman, D. M., Jagust, W. J., Petersen, R.
2-year follow-up study. Frontiers in Aging Neuroscience, 8, 1–43.
C., Snyder, P. J., Carrillo, M. C., Thies, B., & Phelps, C. H. (2011).
McKhann, G. M., Knopman, D. S., Chertkow, H., Hyman, B. T., Jack, C.
The diagnosis of mild cognitive impairment due to Alzheimer’s
R., Kawas, C. H., Klunk, W. E., Koroshetz, W. J., Manly, J. J., Mayeux,
disease: Recommendations from the National Institute on Aging-
R., Mohs, R. C., Morris, J. C., Rossor, M. N., Scheltens, P., Carrillo,
Alzheimer’s Association workgroups on diagnostic guidelines for
8 | PEÑA-BAUTISTA et al.

M. C., Thies, B., Weintraub, S., & Phelps, C. H. (2011). The diagnosis Verdinelli, I., & Wasserman, L. (1995). Computing bayes factors using a
of dementia due to Alzheimer’s disease: Recommendations from the generalization of the savage-dickey density ratio. Journal of American
National Institute on Aging-Alzheimer’s Association workgroups on Statistical Association, 90, 614.
diagnostic guidelines for Alzheimer’s disease. Alzheimer's & Dementia, Wang, H., Stewart, T., Toledo, J. B., Ginghina, C., Tang, L. U., Atik, A., Aro,
7, 263–269. [Link] P., Shaw, L. M., Trojanowski, J. Q., Galasko, D. R., Edland, S., Jensen,
Minhas, S., Khanum, A., Riaz, F., Khan, S. A., & Alvi, A. (2018). Predicting P. H., Shi, M., & Zhang, J. (2018). A longitudinal study of total and
progression from mild cognitive impairment to Alzheimer’s disease phosphorylated α-synuclein with other biomarkers in cerebrospinal
using autoregressive modelling of longitudinal and multimodal fluid of Alzheimer’s disease and mild cognitive impairment. Journal
biomarkers. IEEE Journal of Biomedical and Health Informatics, 22, of Alzheimer's Disease, 61, 1541–1553. [Link]
818–825. JAD-171013
Pedrini, S., Gupta, V. B., Hone, E., Doecke, J., O’Bryant, S., James, I., Wang, J., Chen, C., & Zhang, Y. (2020). An investigation of microRNA-103
Bush, A. I., Rowe, C. C., Villemagne, V. L., Ames, D., Masters, C. L., and microRNA-107 as potential blood-based biomarkers for dis-
& Martins, R. N. (2017). A blood-based biomarker panel indicates ease risk and progression of Alzheimer's disease. Journal of Clinical
IL-10 and IL-12/23p40 are jointly associated as predictors of β-am- Laboratory Analysis, 34(1), e23006.
yloid load in an AD cohort. Scientific Reports, 7, 14057. [Link] Wang, T., Qiu, R. G., & Yu, M. (2018). Predictive modeling of the progres-
org/10.1038/s4159​8-017-14020​-9 sion of Alzheimer’s disease with recurrent neural networks. Scientific
Peña-Bautista, C., Baquero, M., Vento, M., & Cháfer-Pericás, C. (2019). Reports, 8, 9161. [Link]
Free radicals in Alzheimer’s disease: Lipid peroxidation biomark- Weng, W.-C., Huang, W.-Y., Tang, H.-Y., Cheng, M.-L., & Chen, K.-H.
ers. Clinica Chimica Acta, 491, 85–90. [Link] (2019). The differences of serum metabolites between patients
cca.2019.01.021 with early-stage Alzheimer’s disease and mild cognitive impairment.
Peña-Bautista, C., Carrascosa-Marco, P., Oger, C., Vigor, C., Galano, Frontiers in Neurology 10, 1223.
J.-M., Durand, T., Baquero, M., López-Nogueroles, M., Vento, M., Wildsmith, K. R., Schauer, S. P., Smith, A. M., Arnott, D., Zhu, Y., Haznedar,
García-Blanco, A., & Cháfer-Pericás, C. (2019). Validated analytical J., Kaur, S., Mathews, W., & Honigberg, L. A. (2014). Identification of
method to determine new salivary lipid peroxidation compounds as longitudinally dynamic biomarkers in Alzheimer’s disease cerebrospi-
potential neurodegenerative biomarkers. Journal of Pharmaceutical nal fluid by targeted proteomics. Molecular Neurodegeneration, 9, 22.
and Biomedical Analysis, 164, 742–749. [Link] [Link]
jpba.2018.11.043 Yang, M., Elazab, A., Yang, P., Xia, Z., Wang, T., & Lei, B. (2019) Joint and
Peña-Bautista, C., Vigor, C., Galano, J.-M., Oger, C., Durand, T., Ferrer, I., long short-term memory regression of clinical scores for Alzheimer’s
Cuevas, A. (2018). Plasma lipid peroxidation biomarkers for early and disease using longitudinal data, in 2019 41st Annu. Int. Conf. IEEE
non-invasive Alzheimer Disease detection. Free Radical Biology and Engineering in Medicine and Biology Society, 1, 281–284.
Medicine, 124, 388–394. Younes, L., Albert, M., Moghekar, A., Soldan, A., Pettigrew, C., & Miller,
Peña-Bautista, C., Vigor, C., Galano, J.-M., Oger, C., Durand, T., Ferrer, I., M. I. (2019). Identifying changepoints in biomarkers during the pre-
Cuevas, A., López-Cuevas, R., Baquero, M., López-Nogueroles, M., clinical phase of Alzheimer’s disease. Frontiers in Aging Neuroscience,
Vento, M., Hervás-Marín, D., García-Blanco, A., & Cháfer-Pericás, C. 11, 1–11.
(2019). New screening approach for Alzheimer’s disease risk assess-
ment from urine lipid peroxidation compounds. Scientific Reports, 9,
14244. [Link] S U P P O R T I N G I N FO R M AT I O N
Seppala, T. T., Herukka, S.-K., Hanninen, T., Tervo, S., Hallikainen, M., Additional supporting information may be found online in the
Soininen, H., & Pirttila, T. (2010). Plasma A 42 and A 40 as markers
Supporting Information section.
of cognitive change in follow-up: A prospective, longitudinal, pop-
ulation-based cohort study. Journal of Neurology, Neurosurgery and
Psychiatry, 81, 1123–1127.
Sultana, R., Perluigi, M., & Butterfield, D. A. (2013). Lipid peroxida- How to cite this article: Peña-Bautista C, Álvarez L, Baquero
tion triggers neurodegeneration: A redox proteomics view into M, et al. Plasma isoprostanoids assessment as Alzheimer’s
the Alzheimer disease brain. Free Radical Biology and Medicine, 62,
disease progression biomarkers. J Neurochem. 2020;00:1–8.
157–169.
Toledo, J. B., Xie, S. X., Trojanowski, J. Q., & Shaw, L. M. (2013). [Link]
Longitudinal change in CSF Tau and Aβ biomarkers for up to 48
months in ADNI. Acta Neuropathologica, 126, 659–670.