Journal of Alzheimer’s Disease xx (20xx) x–xx 1

DOI 10.3233/JAD-180025
IOS Press

1 Association of Plasma Neurofilament Light

2 Chain with Neocortical Amyloid-␤ Load
and Cognitive Performance in Cognitively

f
3

roo
4 Normal Elderly Participants

rP
5 Pratishtha Chatterjeea,b,c,1 , Kathryn Goozeea,b,c,d,e,f,g,1 , Hamid R. Sohrabia,b,e,f , Kaikai Shenh ,
6 Tejal Shaha,b,f , Prita R. Asihc,i , Preeti Davea,d , Candice ManYand , Kevin Taddeib,f , Roger Chunga ,
7 Henrik Zetterbergj,k,l,m , Kaj Blennowj,k and Ralph N. Martins,a,b,c,e,f,g,∗
8
a Department of Biomedical Sciences, Macquarie University, North Ryde, NSW, Australia
b School

tho
9 of Medical Health and Sciences, Edith Cowan University, Joondalup, WA, Australia
10
c KaRa Institute of Neurological Disease, Sydney, Macquarie Park, Australia

11
d Department of Clinical Research, Anglicare, Sydney, Castle Hill, NSW, Australia

12
e School of Psychiatry and Clinical Neurosciences, University of Western Australia, Crawley, WA, Australia

13
f Australian Alzheimer Research Foundation, Nedlands, WA, Australia
Au
g The Cooperative Research Centre for Mental Health, Carlton South, Australia
14

15
h Australian eHealth Research Centre, CSIRO, Floreat, Australia

16
i School of Medical Sciences, University of New South Wales, Kensington, NSW, Australia

17
j Department of Psychiatry and Neurochemistry, Institute of Neuroscience and Physiology, University of Gothen-

18 burg, Mölndal, Sweden

k Clinical Neurochemistry Laboratory, Sahlgrenska University Hospital, Mölndal, Sweden
d

19

20
l Department of Molecular Neuroscience, UCL Institute of Neurology, Queen Square, London, UK
m UK Dementia Research Institute at UCL, London, UK
cte

21

Accepted 9 February 2018

rre

22 Abstract.
23 Background: The disruption of neurofilament, an axonal cytoskeletal protein, in neurodegenerative conditions may result
24 in neuronal damage and its release into the cerebrospinal fluid and blood. In Alzheimer’s disease (AD), neurofilament light
25 chain (NFL), a neurofilament subunit, is elevated in the cerebrospinal fluid and blood.
co

26 Objective: Investigate the association of plasma NFL with preclinical-AD features, such as high neocortical amyloid-␤ load
27 (NAL) and subjective memory complaints, and cognitive performance in cognitively normal older adults.
28 Methods: Plasma NFL concentrations were measured employing the single molecule array platform in participants from the
29 Kerr Anglican Retirement Village Initiative in Ageing Health cohort, aged 65–90 years. Participants underwent a battery of
Un

30 neuropsychological testing to evaluate cognitive performance and were categorized as low NAL (NAL-, n = 65) and high NAL
31 (NAL+, n = 35) assessed via PET, and further stratified into subjective memory complainers (SMC; nNAL- = 51, nNAL+ = 25)
32 and non-SMC (nNAL- = 14, nNAL+ = 10) based on the Memory Assessment Clinic–Questionnaire.
33 Results: Plasma NFL inversely correlated with cognitive performance. No significant difference in NFL was observed
34 between NAL+ and NAL- participants; however, within APOE ␧4 non-carriers, higher NAL was observed in individuals
35

1 These authors contributed equally to this work.

∗ Correspondence to: Ralph N. Martins, PhD, School of Medical WA 6027, Australia. Tel.: +61 8 6304 5456; Fax: +61 8 6304 5851;
Science, Edith Cowan University, 270 Joondalup Drive, Joondalup, E-mail: r.martins@ecu.edu.au.

ISSN 1387-2877/18/$35.00 © 2018 – IOS Press and the authors. All rights reserved
 2 P. Chatterjee et al. / Plasma Neurofilament Light Chain and Cognition

36 with NFL concentrations within quartiles 3 and 4 (versus quartile 1). Additionally, within the NAL+ participants, SMC had
37 a trend of higher NFL compared to non-SMC.
38 Conclusion: Plasma NFL is inversely associated with cognitive performance in elderly individuals. While plasma NFL may
39 not reflect NAL in individuals with normal global cognition, the current observations indicate that onset of axonal injury,
40 reflected by increased plasma NFL, within the preclinical phase of AD may contribute to the pathogenesis of AD.

41 Keywords: Alzheimer’s disease, blood, neurofilaments, positron emission tomography, cognitive function, episodic memory,
42 executive function, verbal memory, visual memory

f
roo
36 INTRODUCTION NFL and cognitive performance in cognitively nor- 78

mal elderly individuals has also not been examined. 79

37 Neurofilament (NF) is a crucial axonal cytoskeletal Therefore, the current study investigated the 80

38 component comprising three subunits, namely, neu- association between plasma NFL and NAL, employ- 81

39 rofilament light chain (NFL), neurofilament medium ing a standard uptake value ratio (SUVR) cut-off 82

rP
40 chain (NFM), and neurofilament heavy chain (NFH) value of 1.35, categorizing study participants as 83

41 [1]. The disruption of NF in neuronal damage occur- high NAL (SUVR≥1.35; NAL+) and low NAL 84

42 ring within neurodegenerative conditions, results in (SUVR<1.35; NAL-), given that the aberrant build- 85

43 the release of NF into the cerebrospinal fluid (CSF) up of NAL begins as early as two decades prior 86

[2]; consequently giving rise to elevated NFL concen- to the clinical manifestation of AD [13, 14]. Based

tho
44 87

45 trations in the CSF and blood, as has been reported on the Memory Assessment Clinic–Questionnaire 88

46 in Alzheimer’s disease (AD) [3] and other neurode- (MAC-Q) [15], NAL+ participants were further cat- 89

47 generative diseases [4–7]. egorized into subjective memory complainers (SMC, 90

48 With the invasive nature of CSF collection via MAC-Q = 25–35) and non-complainers (non-SMC, 91
Au
49 lumbar puncture, the potential of employing NFL MAC-Q≤24) to examine NFL alterations within the 92

50 as a biomarker for disease diagnosis or progres- preclinical cohort subset at highest risk of AD (NAL+ 93

51 sion, and possibly for screening, further drove the SMC). The current study also examined whether 94

52 investigation of blood NFL alterations, in disease plasma NFL levels inversely correlated with cogni- 95

53 [2, 8–10]. Employing the Alzheimer’s Disease Neu- tive performance within cognitively normal elderly 96

rodegenerative Initiative (ADNI) cohort, Mattsson individuals. Our investigations were therefore aimed
d

54 97

55 and colleagues reported significantly higher plasma at gaining better insight into whether axonal injury, 98

NFL in AD and mild cognitively impaired (MCI) reflected by increased plasma NFL, is present in the
cte

56 99

57 patients compared to controls [11]. Additionally, early pathogenesis of AD. 100

58 higher plasma NFL has been associated with com-

59 promised cognition and hippocampal atrophy both METHODS 101

60 cross-sectionally and longitudinally. Furthermore,

rre

61 plasma NFL was inversely associated with brain glu- Participants 102

62 cose metabolism, longitudinally in the same highly

63 characterized ADNI cohort [11]. More recently, a Study participants belonged to the Kerr Angli- 103

64 study conducted on autosomal dominant Alzheimer’s can Retirement Village Initiative in Ageing Health 104
co

65 disease (ADAD) families found that plasma NFL (KARVIAH) cohort, at baseline. All participants 105

66 was higher in the asymptomatic mutation carriers were residents of Anglicare, New South Wales, Aus- 106

67 and symptomatic mutation carriers compared to their tralia. 107

68 non-carrier relatives, wherein plasma NFL concen- Cohort volunteers (N = 206) were required to meet 108
Un

69 trations were reported to be about 1.3-fold higher in the set screening inclusion and exclusion criteria to 109

70 asymptomatic mutation carriers and 3.6-fold higher be eligible for the KARVIAH cohort. Briefly, the 110

71 in symptomatic mutation carriers [12]. inclusion criteria for the KARVIAH cohort com- 111

72 However, plasma NFL in individuals with preclin- prised an age range of 65–90 years, good general 112

73 ical AD prior to cognitive impairment, characterized health, no known significant cerebral vascular dis- 113

74 by neocortical amyloid-␤ load (NAL) measured via ease, fluent in English, adequate/corrected vision 114

75 positron emission tomography (PET) and subjective and hearing to enable testing, no objective cogni- 115

76 memory complaints, has not been investigated previ- tive impairment as screened by a Montreal Cognitive 116

77 ously. Furthermore, the association between plasma Assessment (MoCA) score ≥26. MoCA scores lying 117
 P. Chatterjee et al. / Plasma Neurofilament Light Chain and Cognition 3

118 between 18–25 were assessed on a case by case Blood collection, APOE genotyping, 167

119 basis by the study neuropsychologist following strat- measurement of plasma NFL 168

120 ification of scores according to age and education

121 [16]. The exclusion criteria comprised, the diagno- All study participants fasted for a minimum of 169

122 sis of dementia based on the revised criteria from the 10 h overnight prior to blood withdraw employing 170

123 National Institute on Aging - Alzheimer’s Associ- standard serological methods and processing [21]. 171

124 ation [17], presence of acute functional psychiatric Apolipoprotein E (APOE) genotype was determined 172

125 disorder (including lifetime history of schizophre- from purified genomic DNA extracted from 0.5 ml 173

f
126 nia or bipolar disorder), history of stroke, severe or whole blood as previously described [21]. 174

roo
127 extremely severe depression (based on the depres- Plasma NFL concentrations were measured 175

128 sion, anxiety, stress scales; DASS) and uncontrolled employing the ultra-sensitive single-molecule array 176

129 hypertension (systolic BP > 170 mm Hg or diastolic (Simoa) platform [10, 11]. Calibrators were run in 177

130 BP > 100 mm Hg). duplicates and samples were run in singlicates with 178

131 One hundred and five participants out of the 134 a 4-fold dilution. Two quality control (QC) lev- 179

rP
132 volunteers meeting the inclusion/exclusion criteria, els were run in duplicates at the beginning and 180

133 underwent neuroimaging, neuropsychometric evalu- the end of each plate. For QC with concentration 181

134 ation and blood collection, as the remaining either 12.1 pg/mL, repeatability and intermediate preci- 182

135 declined undergoing neuroimaging or withdrew from sion were 20.2%, while for QC with concentration 183

155.8 pg/mL, repeatability was 14.6% and inter-

tho
136 the study. Within these 105 participants, 100 par- 184

137 ticipants were considered to have normal global mediate precision was 14.9%. The lowest limit of 185

138 cognition based on their Mini-Mental State Exami- quantification was 6 pg/mL. 186

139 nation score [18] (MMSE ≥26), and were included in

140 the current study. Plasma NFL concentrations were Neuropsychological tests 187
Au
141 reported in all 100 participants considered to have
142 normal global cognition. All volunteers provided All study participants underwent a comprehensive 188

143 written informed consent prior to participation, and battery of neuropsychological testing. The full bat- 189

144 the Bellberry Human Research Ethics Committee, tery comprised the MoCA [16], MMSE [18], MAC-Q 190

145 Australia, provided approval for the study. [15], Rey Auditory Verbal Learning Test (RAVLT) 191

[22], Logical Memory (LM) I and II (WMS-III; Story

d

192

A only) [23], Rey Complex Figure Test (RCFT)[24], 193

cte

146 Neuroimaging Wechsler Adult Intelligence Scale – Third edition 194

(WAIS-III) Digit Span [25], WAIS–III Digit Sym- 195

147 All study participants were imaged within three bol Substitution Test (DSST) [26], D-KEFS Category 196

148 months of blood collection. Participants underwent Fluency (Boys Names) and Switching (Fruits and 197

149 PET using ligand 18 F-Florbetaben (FBB) at Mac- Furniture) Tasks [27], Controlled Oral Word Asso- 198
rre

150 quarie Medical Imaging in Sydney. Participants were ciation Test [28], Stroop Test (Victoria version) [29], 199

151 administered an intravenous bolus of FBB slowly the Boston Naming Test [30], Wechsler Test of Adult 200

152 over 30 s, while in a rested position. Images were Reading [31] and the DASS [32]. Composite scores 201

153 acquired over a 20 min scan, in 5 min acquisi- were generated for verbal and visual episodic mem- 202
co

154 tions, beginning 50 min post injection. Neocortical ory and for working memory and executive function. 203

155 amyloid-␤ load was calculated as the mean SUVR of The verbal and visual episodic memory composite 204

156 the frontal, superior parietal, lateral temporal, lateral score was created from the mean of the z-scores of 205

157 occipital, and anterior and posterior cingulate regions RAVLT List A, RAVLT short delay, RAVLT long 206
Un

158 using image processing software, CapAIBL [19, 20]. delay, LM I, LMII, RCFT 3 min and RCFT 30 min 207

159 Ninety-six of the 100 participants within the while the working memory and executive function 208

160 current study passed all standard MRI inclu- composite score was generated from the mean of the 209

161 sion/exclusion criteria, and underwent MRI as z-scores of Digit Span backward, DSST, D-KEFS 210

162 described previously using a General Electric (GE) Boys names and Fruits and Furniture Switching tasks. 211

163 3 Tesla scanner (Model 750 W) [13]. Hippocampal The global composite score was constructed from 212

164 volume calculated from the images acquired was nor- the mean of the z-score measures of RAVLT List A, 213

165 malized with the total intracranial volume comprising RAVLT short delay, RAVLT long delay, LM I, LM 214

166 the cerebrospinal fluid, grey matter, and white matter. II, RCFT 3 min, RCFT 30 min, Digit Span backward, 215
 4 P. Chatterjee et al. / Plasma Neurofilament Light Chain and Cognition

216 DSST, D-KEFS Boys names and Fruits and Furniture observation of a significantly higher APOE ␧4 carrier 246

217 Switching tasks and MMSE. frequency in the NAL+ group. 247

Plasma NFL was observed to correlate with 248

218 Statistical analyses age (r = 0.533, p < 0.0001) while no signif- 249

icant association between plasma NFL was 250

219 Descriptive statistics including means and stan- observed with education (r = –0.004, p = 0.965), 251

220 dard deviations were calculated for NAL+ and NAL- gender (mean ± SD: males, 39.26 ± 20.55 pg/mL; 252

221 groups. Chi-square tests were employed to compare females, 35.02 ± 16.30 pg/mL; p = 0.268) or 253

f
222 the frequency of gender, APOE ␧4 carrier status and APOE ␧4 carriage (mean ± SD: non-carriers, 37.49 254

roo
223 SMC between NAL+ and NAL- groups. Additionally, ± 18.29 pg/mL; carriers, 32.18 ± 15.39 pg/mL; 255

224 linear models were employed to compare continu- p = 0.226). 256

225 ous variables examined in the study between groups Plasma NFL was not observed to be significantly 257

226 of interest (e.g., NAL- versus NAL+, NAL+/non- elevated in NAL+ participants (or preclinical AD) 258

227 SMC versus NAL+/SMC, NFL quartiles Q1 versus compared to NAL- participants (mean ± SD: NAL-, 259

rP
228 Q2, Q3, Q4, etc.) with and without adjusting for 35.23 ± 17.27 pg/mL; NAL+, 38.50 ± 18.75 pg/mL) 260

229 covariates age, gender and APOE ␧4 carrier sta- with (p = 0.563) and without (p = 0.384) adjust- 261

230 tus. Continuous response variables were tested for ing for age, gender, and APOE ␧4; however, 262

231 approximate normality and variance homogeneity, a trend of elevated plasma NFL concentrations 263

and log transformed when required to satisfy test cri-

tho
232 was observed in NAL+ SMC compared to NAL+ 264

233 teria. Pearson’s correlation coefficient was employed non-SMC, adjusting for age, gender, and APOE 265

234 to investigate correlations between NFL and other ␧4 carrier status (mean ± SD, NAL+/non-SMC 266

235 continuous variables of interest, except for the corre- (n = 10): 31.50 ± 12.57 pg/mL, NAL+/SMC (n = 25): 267

236 lation between NFL and MMSE where Spearman’s 41.30 ± 20.26 pg/mL; p = 0.069). 268
Au
237 correlation was used. Partial correlations were used Additionally, after stratifying the cohort into 269

238 when associations investigated were adjusted for age. APOE ␧4 carriers (n = 21) and non-carriers (n = 79), 270

239 All analyses were carried out using IBM® SPSS® within the APOE ␧4 non-carrier subset, participants 271

240 Version 23. carrying NFL concentrations lying within quar- 272

tile 1 were observed to have significantly lower 273

NAL (1.15 ± 0.13) compared to those in quar-

d

241 RESULTS 274

tiles 3 (1.32 ± 0.29, p = 0.031) and 4 (1.38 ± 0.32, 275

cte

242 Demographic characteristics of study participants p = 0.006, Fig. 1). However, after adjusting for age, 276

243 have been represented in Table 1, wherein no sig- significance disappeared between quartile 1 and 3, 277

244 nificant differences were observed between NAL- and only a trend remained between quartile 1 and 278

245 and NAL+ participants, except for the expected 4 (p = 0.068), likely due to the strong correlation 279
rre

Table 1
Demographic characteristics of study participants
NAL- NAL+ p
Gender (M/F) 19/46 13/22 0.419
Age (years, mean ± SD) 77.61 ± 5.55 79.22 ± 5.38 0.165
co

BMI (mean ± SD) 27.38 ± 4.47 28.05 ± 4.73 0.486

Education (years, mean ± SD) 14.84 ± 3.37 13.64 ± 2.91 0.078
n APOE ␧4 carriers (%) 5 (7.69) 16 (45.71) 0.00000848
MMSE (mean ± SD) 28.50 ± 1.16 28.80 ± 1.10 0.225
Un

n SMC (%) 51 (78.46) 25 (71.42) 0.432

HV% (left; right lobes, mean ± SD) 0.195 ± 0.020; 0.194 ± 0.019; 0.805;
0.199 ± 0.021 0.199 ± 0.018 0.890
FBB-PET SUVR (mean ± SD) 1.15 ± 0.08 1.71 ± 0.26 –
Baseline characteristics including gender, age, body mass index (BMI), education, APOE ␧4 status, Mini-Mental
State Examination (MMSE) scores, number of subjective memory complainers (SMC) based on the Memory
Assessment Clinic-Questionnaire (MAC-Q), hippocampal volume (HV) normalized by the intracranial volume
and neocortical amyloid-␤ load (NAL) represented by the standard uptake value ratio (SUVR) of ligand 18 F-
Florbetaben (FBB) in the neocortical region normalized with that in the cerebellum, have been compared between
NAL- (SUVR<1.35) and NAL+ (SUVR≥1.35) study participants. Chi-square test or linear models were employed
as appropriate.
 P. Chatterjee et al. / Plasma Neurofilament Light Chain and Cognition 5

n = 41) cognitively normal participants [11]. Further- 307

more, associations between brain atrophy and plasma 308

(and CSF) NFL concentrations reported previously in 309

MCI, AD (and cognitively normal) individuals with 310

both high and low CSF A␤ load [33], indicate that 311

NFL is a general neurodegeneration biomarker, with 312

high levels in many disorders such as frontotempo- 313

ral dementia and progressive supranuclear palsy [10, 314

f
34], which also is consistent with the observations 315

roo
of NFL between NAL+ and NAL- participants in the 316

current study. However, we observed a trend of ele- 317

Fig. 1. Comparison of NAL between NFL quartiles in APOE vated plasma NFL in SMC compared to non-memory 318

non-␧4 carriers. Neocortical amyloid-␤ load (NAL) assessed via complainers within the NAL+ subset employed in 319

positron emission tomography using ligand 18 F-florbetaben was the current study—the cohort subset likely to be the 320

rP
observed to be elevated in plasma neurofilament light chain (NFL) most advanced in the preclinical AD pathogenesis 321
quartile (Q) 3 and Q4 compared to Q1, using linear models within
APOE non-␧4 carriers (n = 79); however, this was not the case in trajectory. 322

APOE ␧4 carriers (n = 21, Supplementary Figure 1). NFL quartile Interestingly, Mattsson and colleagues, also 323

cut-offs were 23.33 pg/ml, 34.05 pg/ml, and 45.41 pg/ml for Q1, observed that MCI and AD APOE ␧4 non-carriers 324
Q2, and Q3 respectively, in the APOE non-␧4 carriers. ∗ p < 0.05,

tho
∗∗ p < 0.01, error bars represent SE not adjusted for covariates age had significantly higher plasma NFL concentrations 325

and gender. Corresponding analysis adjusted for covariates is avail- compared to ␧4 carriers. Within the current study 326

able in the text. Percentages of NAL+ individuals in each quartile as well, APOE ␧4 non-carriers had a higher mean 327
have been shown within the bar graphs. plasma NFL (37.49 pg/ml) compared to APOE ␧4 328

carriers (32.18 pg/ml), although it did not reach sig- 329

Au
280 between NFL and age, confounding associations of nificance presumably due to the relatively small 330

281 NFL with other age-related outcomes within the numbers in the latter group. However, within the 331

282 scope of this modest sample size. APOE ␧4 non-carrier cohort subset, a trend of 332

283 Plasma NFL correlated inversely with verbal and higher NAL was observed in participants with plasma 333

284 visual episodic memory (r = –0.305, p = 0.002) and NFL lying within Q3 and Q4 compared to Q1, 334

working memory and executive function (r = –0.357, suggesting that the onset of axonal cytoskeletal dis-
d

285 335

286 p = 0.0002) in all participants (Fig. 2). Plasma ruption may commence as early as this preclinical 336

NFL was also observed to correlate inversely with

cte

287 phase of AD. 337

288 global cognition assessed via MMSE (r = –0.326, Within the current study we did not observe a 338

289 p = 0.001) and the global composite score (r = –0.407, significant correlation between plasma NFL and 339

290 p < 0.0001) (Fig. 2). On adjusting for age, plasma hippocampal atrophy: an observation in line with 340

291 NFL continued to significantly correlate inversely findings reported by Pereira and colleagues, wherein 341
rre

292 with working memory and executive function plasma NFL was associated with brain atrophy only 342

293 (r = –0.200, p = 0.047) and the global composite score in symptomatic cases, while CSF NFL concentra- 343

294 (r = –0.223, p = 0.026). tions were associated with brain atrophy in AD, MCI, 344

295 No significant association was observed between and cognitively normal subjects [33]. Similar obser- 345
co

296 plasma NFL and hippocampal volume within the vations have also been reported in the autosomal 346

297 current study (left hemisphere, r = –0.095, p = 0.356; dominant form of AD by Weston and colleagues [12]. 347

298 right hemisphere, r = –0.096, p = 0.352). No association between plasma NFL and neocortical 348

glucose metabolism (data not shown) was observed 349

Un

299 DISCUSSION within the current cross-sectional study, which is in 350

agreement with the findings reported by Mattsson and 351

300 Plasma NFL concentrations were not significantly colleagues [11]. 352

301 elevated in cognitively normal elderly NAL+ ver- Plasma NFL was observed to correlate strongly 353

302 sus NAL- participants. Our findings are in line with with age in the present study. This correlation is 354

303 those reported by Mattsson and colleagues wherein consistent with previous studies reporting significant 355

304 no significant difference in plasma NFL was observed correlations between age and NFL in both, CSF and 356

305 between NAL- (based on CSF A␤42 ≥192 ng/L, plasma (or serum) from healthy controls, individuals 357

306 n = 71) and NAL+ (based on CSF A␤42 <192 ng/L, with pre-symptomatic neurodegenerative disease and 358
 6 P. Chatterjee et al. / Plasma Neurofilament Light Chain and Cognition

f
roo
rP
tho
Fig. 2. Association between plasma NFL and cognition in cognitively normal elderly participants. Plasma neurofilament light chain (NFL)
Au
concentrations (pg/ml) inversely correlated with verbal and visual episodic memory (A), working memory and executive function (B) and
global cognition (C) using Pearson’s correlation coefficient. Z-scores were calculated for Rey Auditory Verbal Learning Test (RAVLT) List
A, RAVLT short delay, RAVLT long delay, Logical Memory (LM) I, LMII, Rey Complex Figure Test (RCFT) 3 min, RCFT 30 min, Digit
Span backward, Digit Symbol Substitution Test (DSST), D-KEFS Category Fluency Task (Boys names), D-KEFS Switching Task (Fruit
and Furniture) and MMSE. Composite scores for Verbal and Visual Episodic Memory were calculated by summation of z-scores of RAVLT
List A, RAVLT short delay, RAVLT long delay, LM I, LMII, RCFT 3 min, RCFT 30 min while composite scores for working memory and
executive function were generated by summation of z-scores of Digit span backward, DSST, and D-KEFS Category Fluency and Switching
d

Tasks. The global composite score was constructed by calculating the mean of z-score measures of RAVLT List A, RAVLT short delay,
RAVLT long delay, LM I, LMII, RCFT 3 min, RCFT 30 min, Digit Span backward, DSST, D-KEFS tasks, and MMSE.
cte

359 within cohorts comprising healthy controls, MCI and normal cohort with a representative proportion of 379

360 AD patients [3, 7, 9, 11, 35, 36]. preclinical AD individuals, in agreement with other 380

361 Additionally, plasma and CSF NFL concentrations established cohorts [38, 39], employing PET for NAL 381
rre

362 have previously been reported to be inversely associ- measurement, a stronger marker of AD neuropathol- 382

363 ated with MMSE, the AD assessment scale–cognitive ogy compared to CSF A␤, as employed previously 383

364 subscale, Clinical Dementia Rating Scale and the [11]. Additionally, the study also incorporates a com- 384

365 Trail Making Test-B scores in participants with prehensive battery of neuropsychological tests and 385
co

366 MCI, sporadic AD, ADAD and bipolar disorder [3, most importantly, a highly sensitive assay to measure 386

367 11, 12, 37]. The current study also observed that plasma NFL; however, the current findings need to be 387

368 plasma NFL correlated inversely with cognitive per- replicated in other cohorts, also employing alternative 388

369 formance, particularly, verbal and visual episodic methods. 389

Un

370 memory, executive function and working memory To summarize, plasma NFL was not significantly 390

371 and global cognition suggesting that higher plasma higher in NAL+ versus NAL- cognitively normal 391

372 NFL concentrations are associated with inferior cog- elderly individuals, however a trend of elevated 392

373 nitive performance in elderly individuals with normal plasma NFL was observed within the NAL+ SMC 393

374 global cognition as well. (the cohort subset likely to be the farthest in the 394

375 We acknowledge that the current study has limita- preclinical AD pathogenesis trajectory), indicating 395

376 tions with regard to its relatively modest sample size. onset of axonal injury occurs well before the onset 396

377 However, the present study also has its strengths given of clinical AD symptoms. Additionally, significant 397

378 that it utilizes a highly characterized, cognitively associations were observed between plasma NFL and 398
 P. Chatterjee et al. / Plasma Neurofilament Light Chain and Cognition 7

399 neuropsychometric parameters representing visual SUPPLEMENTARY MATERIAL 448

400 and verbal memory, executive function and working

401 memory and global cognition, in the current study. The supplementary material is available in the 449

402 Our current plasma NFL observations along with electronic version of this article: http://dx.doi.org/ 450

403 those previously published indicate that plasma NFL 10.3233/JAD-180025. 451

404 alterations begin to manifest within the end stage

405 of preclinical AD, and while NFL is not a suffi- REFERENCES 452
406 cient stand-alone marker for preclinical and clinical

f
407 AD, it has the potential to serve as an early marker [1] Petzold A (2005) Neurofilament phosphoforms: Surrogate

roo
453
408 of neurodegeneration. Moreover, given that plasma markers for axonal injury, degeneration and loss. J Neurol 454

409 NFL correlated with cognition, it is a promising Sci 233, 183-198. 455
[2] Gaiottino J, Norgren N, Dobson R, Topping J, Nissim A, 456
410 biomarker for disease progression and for monitoring Malaspina A, Bestwick JP, Monsch AU, Regeniter A, Lind- 457
411 disease modifying therapies, reaffirming the potential berg RL, Kappos L, Leppert D, Petzold A, Giovannoni G, 458
412 of elevated plasma NFL as a marker of progres- Kuhle J (2013) Increased neurofilament light chain blood 459

rP
413 sive neurodegeneration. The current findings warrant levels in neurodegenerative neurological diseases. PLoS 460
One 8, e75091. 461
414 investigation of plasma NFL levels and cognitive [3] Zetterberg H, Skillback T, Mattsson N, Trojanowski 462
415 decline longitudinally, providing further insight on JQ, Portelius E, Shaw LM, Weiner MW, Blennow 463

416 neuronal damage (characterized by plasma NFL) K, Alzheimer’s Disease Neuroimaging Initiative (2016) 464
Association of cerebrospinal fluid neurofilament light con-

tho
417 reflecting cognitive decline. 465
centration with Alzheimer disease progression. JAMA 466
Neurol 73, 60-67. 467
[4] Backstrom DC, Eriksson Domellof M, Linder J, Olsson B, 468
418 ACKNOWLEDGMENTS Ohrfelt A, Trupp M, Zetterberg H, Blennow K, Forsgren 469
L (2015) Cerebrospinal fluid patterns and the risk of future 470
dementia in early, incident Parkinson disease. JAMA Neurol 471
419 This study was funded by the Foundation for Aged
Au
72, 1175-1182. 472
420 Care, Anglicare, Sydney, the Australian Alzheimer [5] Zetterberg H, Jacobsson J, Rosengren L, Blennow K, Ander- 473
421 Research Foundation (AARF), Perth, the Swedish sen PM (2007) Cerebrospinal fluid neurofilament light 474

422 Research Council (grant #2017-00915 and #2013- levels in amyotrophic lateral sclerosis: Impact of SOD1 475
genotype. Eur J Neurol 14, 1329-1333. 476
423 2546), the Swedish Alzheimer Foundation (grant
[6] Scherling CS, Hall T, Berisha F, Klepac K, Karydas A, Cop- 477
#AF-553101), the Torsten Söderberg Professorship
d

424
pola G, Kramer JH, Rabinovici G, Ahlijanian M, Miller 478
425 in Medicine at the Royal Swedish Academy of Sci- BL, Seeley W, Grinberg LT, Rosen H, Meredith Jr J, Boxer 479

ences, the Knut and Alice Wallenberg Foundation and AL (2014) Cerebrospinal fluid neurofilament concentra- 480
cte

426
tion reflects disease severity in frontotemporal degeneration. 481
427 LUA/ALF VGR project (grant #ALFGBG-715986 Ann Neurol 75, 116-126. 482
428 and 720931), and the KaRa Institute of Neurologi- [7] van Eijk JJ, van Everbroeck B, Abdo WF, Kremer BP, Ver- 483
429 cal Diseases (KaRa MINDS), Sydney. We thank the beek MM (2010) CSF neurofilament proteins levels are 484

430 participants and their families for their participation elevated in sporadic Creutzfeldt-Jakob disease. J Alzheimers 485
Dis 21, 569-576.
rre

486
431 and cooperation, and Anglicare, KaRa MINDS and [8] Zhou W, Zhang J, Ye F, Xu G, Su H, Su Y, Zhang 487
432 AARF research and support staff for their contri- X, Alzheimer’s Disease Neuroimaging Initiative (2017) 488

433 butions to this study. We specially thank Professor Plasma neurofilament light chain levels in Alzheimer’s dis- 489

434 Ian James for his contributions to the study. We ease. Neurosci Lett 650, 60-64. 490
[9] Meeter LH, Dopper EG, Jiskoot LC, Sanchez-Valle R, Graff 491
thank Ms. Emma Toovey, Ms. Kate Fredericks, Ms.
co

435
C, Benussi L, Ghidoni R, Pijnenburg YA, Borroni B, Galim- 492
436 Bethany Ball, and Ms. Catherine Brown for their con- berti D, Laforce RJ, Masellis M, Vandenberghe R, Ber IL, 493

437 tributions to the study. We also thank the staff of Otto M, van Minkelen R, Papma JM, Rombouts SA, Bal- 494

438 the Macquarie Medical Imaging centre in Macquarie asa M, Oijerstedt L, Jelic V, Dick KM, Cash DM, Harding 495
SR, Jorge Cardoso M, Ourselin S, Rossor MN, Padovani A,
Un

496
439 University Hospital, Sydney, for their contributions. Scarpini E, Fenoglio C, Tartaglia MC, Lamari F, Barro C, 497
440 KG is a recipient of the Cooperative Research Cen- Kuhle J, Rohrer JD, Teunissen CE, van Swieten JC (2016) 498

441 tre for Mental Health top-up scholarship. Florbetaben Neurofilament light chain: A biomarker for genetic fron- 499
totemporal dementia. Ann Clin Transl Neurol 3, 623-636. 500
442 is a proprietary PET radiopharmaceutical owned by
[10] Rohrer JD, Woollacott IO, Dick KM, Brotherhood E, Gor- 501
443 Piramal Imaging SA. For this study, Florbetaben was don E, Fellows A, Toombs J, Druyeh R, Cardoso MJ, 502
444 manufactured and supplied under GMP conditions by Ourselin S, Nicholas JM, Norgren N, Mead S, Andreasson 503

445 Cyclotek (Aust) Pty Ltd. U, Blennow K, Schott JM, Fox NC, Warren JD, Zetterberg H 504
(2016) Serum neurofilament light chain protein is a measure 505
446 Authors’ disclosures available online (https:// of disease intensity in frontotemporal dementia. Neurology 506
447 www.j-alz.com/manuscript-disclosures/18-0025r1). 87, 1329-1336. 507
 8 P. Chatterjee et al. / Plasma Neurofilament Light Chain and Cognition

508 [11] Mattsson N, Andreasson U, Zetterberg H, Blennow [22] Estevez-Gonzalez A, Kulisevsky J, Boltes A, Otermin P, 573
509 K, Alzheimer’s Disease Neuroimaging Initiative (2017) Garcia-Sanchez C (2003) Rey verbal learning test is a use- 574
510 Association of plasma neurofilament light with neurodegen- ful tool for differential diagnosis in the preclinical phase 575
511 eration in patients with Alzheimer disease. JAMA Neurol 74, of Alzheimer’s disease: Comparison with mild cognitive 576
512 557-566. impairment and normal aging. Int J Geriatr Psychiatry 18, 577
513 [12] Weston PSJ, Poole T, Ryan NS, Nair A, Liang Y, Macpher- 1021-1028. 578
514 son K, Druyeh R, Malone IB, Ahsan RL, Pemberton H, [23] Elwood RW (1991) The Wechsler Memory Scale-Revised: 579
515 Klimova J, Mead S, Blennow K, Rossor MN, Schott JM, Psychometric characteristics and clinical application. Neu- 580
516 Zetterberg H, Fox NC (2017) Serum neurofilament light in ropsychol Rev 2, 179-201. 581
517 familial Alzheimer disease: A marker of early neurodegen- [24] Meyers J, Meyers K (1995) Rey Complex FigureTest 582

f
518 eration. Neurology 89, 2167-2175. and Recognition Trial. Professional Manual. Psychological 583

roo
519 [13] Goozee K, Chatterjee P, James I, Shen K, Sohrabi HR, Asih Assessment Resource, Inc. 584
520 PR, Dave P, Ball B, ManYan C, Taddei K, Chung R, Garg [25] Wechsler DA (1987) Wechsler memory scale-revised. Psy- 585
521 ML, Martins RN (2017) Alterations in erythrocyte fatty acid chological Corporation, New York. 586
522 composition in preclinical Alzheimer’s disease. Sci Rep 7, [26] Wechsler D (1997) Wechsler Adult Intelligence Scale – III 587
523 676. (WAIS-III) Administration and Scoring Manual. The Psy- 588
524 [14] Dubois B, Hampel H, Feldman HH, Scheltens P, Aisen P, chological Corporation, San Antonio, TX. 589

rP
525 Andrieu S, Bakardjian H, Benali H, Bertram L, Blennow [27] Delis D, Kaplan E, Kramer J (2001) Delis-Kaplan Executive 590
526 K, Broich K, Cavedo E, Crutch S, Dartigues JF, Duyck- Function System (D-KEFS). The Psychological Corpora- 591
527 aerts C, Epelbaum S, Frisoni GB, Gauthier S, Genthon R, tion, San Antonio, TX. 592
528 Gouw AA, Habert MO, Holtzman DM, Kivipelto M, Lista [28] Patterson J (2011) Controlled oral word association test. 593
529 S, Molinuevo JL, O’Bryant SE, Rabinovici GD, Rowe C, In Encyclopedia of Clinical Neuropsychology, Kreutzer JS, 594
530 Salloway S, Schneider LS, Sperling R, Teichmann M, Car- DeLuca J, Caplan B, eds. Springer, New York, pp. 703-706. 595

tho
531 rillo MC, Cummings J, Jack Jr CR, Proceedings of the [29] Strauss E, Sherman, Spreen O (2006) A Compendium 596
532 Meeting of the International Working Group, the American of Neuropsychological Tests: Administration, Norms, and 597
533 Alzheimer’s Association on “The Preclinical State of AD, Commentary (3rd edn). Oxford University Press, New York. 598
534 July, Washington DC USA (2016) Preclinical Alzheimer’s [30] Saxton J, Ratcliff G, Munro CA, Coffey EC, Becker JT, 599
535 disease: Definition, natural history, and diagnostic criteria. Fried L, Kuller L (2000) Normative data on the Boston 600
536 Alzheimers Dement 12, 292-323. Naming Test and two equivalent 30-item short forms. Clin 601
Au
537 [15] Crook 3rd TH, Feher EP, Larrabee GJ (1992) Assessment of Neuropsychol 14, 526-534. 602
538 memory complaint in age-associated memory impairment: [31] Wechsler D (2001) Wechsler Test of Adult Reading: 603
539 The MAC-Q. Int Psychogeriatr 4, 165-176. Examiner’s Manual. The Psychological Corporation, San 604
540 [16] Rossetti HC, Lacritz LH, Cullum CM, Weiner MF (2011) Antonio, TX. 605
541 Normative data for the Montreal Cognitive Assessment [32] Lovibond SH, Lovibond, PF (1995) Manual for the 606
542 (MoCA) in a population-based sample. Neurology 77, 1272- Depression Anxiety Stress Scales. (2nd. Ed.) Psychology 607
543 1275. Foundation, Sydney. 608
d

544 [17] McKhann GM, Knopman DS, Chertkow H, Hyman BT, [33] Pereira JB, Westman E, Hansson O, Alzheimer’s Disease 609
545 Jack Jr CR, Kawas CH, Klunk WE, Koroshetz WJ, Manly Neuroimaging Initiative (2017) Association between cere- 610
cte

546 JJ, Mayeux R, Mohs RC, Morris JC, Rossor MN, Schel- brospinal fluid and plasma neurodegeneration biomarkers 611
547 tens P, Carrillo MC, Thies B, Weintraub S, Phelps CH with brain atrophy in Alzheimer’s disease. Neurobiol Aging 612
548 (2011) The diagnosis of dementia due to Alzheimer’s dis- 58, 14-29. 613
549 ease: Recommendations from the National Institute on [34] Hansson O, Janelidze S, Hall S, Magdalinou N, Lees AJ, 614
550 Aging-Alzheimer’s Association workgroups on diagnostic Andreasson U, Norgren N, Linder J, Forsgren L, Constan- 615
551 guidelines for Alzheimer’s disease. Alzheimers Dement 7, tinescu R, Zetterberg H, Blennow K, edish BioFINDER 616
rre

552 263-269. study (2017) Blood-based NfL: A biomarker for differential 617
553 [18] Folstein MF, Folstein SE, McHugh PR (1975) “Mini- diagnosis of parkinsonian disorder. Neurology 88, 930-937. 618
554 mental state”. A practical method for grading the cognitive [35] Jakobsson J, Bjerke M, Ekman CJ, Sellgren C, Johansson 619
555 state of patients for the clinician. J Psychiatr Res 12, AG, Zetterberg H, Blennow K, Landen M (2014) Ele- 620
556 189-198. vated concentrations of neurofilament light chain in the 621
co

557 [19] Zhou L, Salvado O, Dore V, Bourgeat P, Raniga P, Macaulay cerebrospinal fluid of bipolar disorder patients. Neuropsy- 622
558 SL, Ames D, Masters CL, Ellis KA, Villemagne VL, Rowe chopharmacology 39, 2349-2356. 623
559 CC, Fripp J, AIBL Research Group (2014) MR-less surface- [36] Gisslen M, Price RW, Andreasson U, Norgren N, Nilsson S, 624
560 based amyloid assessment based on 11C PiB PET. PLoS One Hagberg L, Fuchs D, Spudich S, Blennow K, Zetterberg H 625
561 9, e84777. (2016) Plasma concentration of the neurofilament light pro- 626
Un

562 [20] Bourgeat P, Villemagne VL, Dore V, Brown B, Macaulay tein (NFL) is a biomarker of CNS injury in HIV infection: 627
563 SL, Martins R, Masters CL, Ames D, Ellis K, Rowe CC, Sal- A cross-sectional study. EBioMedicine 3, 135-140. 628
564 vado O, Fripp J, AIBL Research Group (2015) Comparison [37] Rolstad S, Jakobsson J, Sellgren C, Ekman CJ, Blennow 629
565 of MR-less PiB SUVR quantification methods. Neurobiol K, Zetterberg H, Palsson E, Landen M (2015) Cognitive 630
566 Aging 36(Suppl 1), S159-166. performance and cerebrospinal fluid biomarkers of neurode- 631
567 [21] Goozee K, Chatterjee P, James I, Shen K, Sohrabi HR, generation: A study of patients with bipolar disorder and 632
568 Asih PR, Dave P, ManYan C, Taddei K, Ayton SJ, Garg healthy controls. PLoS One 10, e0127100. 633
569 ML, Kwok JB, Bush AI, Chung R, Magnussen JS, Martins [38] Ellis KA, Bush AI, Darby D, De Fazio D, Foster J, Hud- 634
570 RN (2017) Elevated plasma ferritin in elderly individuals son P, Lautenschlager NT, Lenzo N, Martins RN, Maruff P, 635
571 with high neocortical amyloid-␤ load. Mol Psychiatry. doi: Masters C, Milner A, Pike K, Rowe C, Savage G, Szoeke 636
572 10.1038/mp.2017.146 C, Taddei K, Villemagne V, Woodward M, Ames D (2009) 637
 P. Chatterjee et al. / Plasma Neurofilament Light Chain and Cognition 9

638 The Australian Imaging, Biomarkers and Lifestyle (AIBL) [39] Mueller SG, Weiner MW, Thal LJ, Petersen RC, Jack C, 642
639 study of aging: Methodology and baseline characteristics Jagust W, Trojanowski JQ, Toga AW, Beckett L (2005) The 643
640 of 1112 individuals recruited for a longitudinal study of Alzheimer’s disease neuroimaging initiative. Neuroimaging 644
641 Alzheimer’s disease. Int Psychogeriatr 21, 672-687. Clin N Am 15, 869-877, xi-xii. 645

f
roo
rP
tho
d Au
cte
rre
co
Un