Intersite variability of CSF Alzheimer’s disease biomarkers in clinical setting

Alzheimer’s & Dementia 9 (2013) 406–413

Intersite variability of CSF Alzheimer’s disease biomarkers in clinical setting Julien Dumurgiera,*, Olivier Vercruysseb,c, Claire Paqueta,d, St
ephanie Bomboisb, Chlo
e Chaulete, Jean-Louis Laplanchef, Katell Peoc’hf, Susanna Schraenc, Florence Pasquierb, Jacques Touchone, Jacques Hugona,d, Sylvain Lehmanng, Audrey Gabellee,g a

Centre Memoire Ressources Recherche Paris Nord Ile-de-France, AP-HP, Saint Louis - Lariboisiere - Fernand Widal Hospital, Univ Paris Diderot, Sorbonne Paris Cit
e, Paris, France b Centre Memoire Ressources Recherche Lille, Lille University Hospital, Lille, France c INSERM U837, University Lille-Nord de France, Centre of Biology and Pathology, Jean-Pierre Aubert Research Center, Lille, France d INSERM U839, Institut du Fer a Moulin, Paris, France e Centre Memoire Ressources Recherche Montpellier, University of Montpellier, Montpellier, France f Service de Biochimie et Biologie Mol
eculaire, AP-HP, Saint Louis - Lariboisiere - Fernand Widal Hospital, Univ Paris Diderot, Sorbonne Paris Cit
e, Paris, France g Biochimie-Prot
eomique Clinique - IRB - CCBHM - INSERM U1040, University of Montpellier, Montpellier, France

Abstract

Background: The assessment of total tau, phosphorylated tau (pTau-181) and amyloid beta (Ab 1–42) concentrations in the cerebrospinal fluid (CSF) of subjects has been validated for the diagnosis of Alzheimer’s disease (AD). Although these measurements have shown some variability, little is known about their intersite variability in clinical settings. Methods: A total of 880 subjects (AD, n 5 515; non-AD, n 5 365) from three French memory centers were included. Receiver–operating characteristic analyses were performed to computerized area under curves (AUCs) and optimal thresholds for each biomarker in the three centers. A test–retest study was performed in a group of 32 CSF samples by repeated blind analysis of the three biomarkers using the same immunoassay batches in the three centers. Results: In the three centers, tau (AUC, 0.82–0.88) and pTau-181 (AUC, 0.83–0.89) outperformed Ab 1–42 (AUC, 0.70 –0.73) to discriminate subjects with AD from those without AD. An intersite variation of mean levels and cutoffs was observed for the three biomarkers. This variation was higher for Ab 1–42 (range of cutoff, 368–582 pg/mL) than for tau (range of cutoff, 289–353 pg/mL). In a test–retest study, the mean interlaboratory coefficients of variation were 12.2% for Ab 1–42, 11.3% for tau, and 11.5% for pTau-181. Conclusion: Intercenter variability of CSF biomarkers has been confirmed in a multisite cohort of subjects and can be improved in clinical settings. Efforts on harmonization of procedures should be encouraged to optimize the accuracy of CSF biomarkers in AD. Ó 2013 The Alzheimer’s Association. All rights reserved.

Keywords:

Alzheimer’s disease; Amyloid-b; Biomarkers; Cerebrospinal fluid; Tau; Variability

1. Introduction Alzheimer’s disease (AD) is the most common cause of dementia in elderly populations, affecting millions of people J.D., O.V., and C.P. contributed equally to the manuscript. *Corresponding author. Tel.: 133 1 40 05 49 54; Fax: 133 1 40 05 43 39. E-mail address: [email protected]

worldwide [1]. The diagnosis of AD, according to the usual National Institute of Neurological and Communicative Diseases and Stroke/Alzheimer’s Disease and Related Disorders Association (NINCDS-ADRDA) criteria, remains based on clinical exclusion criteria leading to the diagnosis of probable or possible AD [2]. Based on these criteria, the diagnosis of probable AD has a mean specificity set at around 70% [3], meaning that some people who met the

1552-5260/$ - see front matter Ó 2013 The Alzheimer’s Association. All rights reserved. http://dx.doi.org/10.1016/j.jalz.2012.06.006

J. Dumurgier et al. / Alzheimer’s & Dementia 9 (2013) 406–413

clinical criteria for AD do not have the corresponding neuropathological lesions. New disease-modifying drugs will probably be more effective in subjects with early AD prior to the occurrence of diffuse amyloid plaques and neurofibrillary tangles, and before neuronal losses become too severe [4]. Consequently, with the challenge of new drugs, it is essential to provide early and better AD diagnosis of subjects [5]. In this context, a revision of the NINCDS-ADRDA criteria has been proposed recently, and the contribution of new disease biomarkers to the diagnosis (i.e., cerebrospinal fluid [CSF] biomarkers, positron emission tomographic imaging, structural brain magnetic resonance imaging) have been discussed leading to the stage of “probable AD dementia with evidence of the AD pathophysiological process” [6]. During the past decade, the assessment of total tau, phosphorylated tau 181 (pTau-181) levels, and amyloid beta (Ab 1–42) concentration detected in the CSF of patients has been validated in vivo for AD diagnosis [7,8]. Furthermore, CSF biomarkers were correlated to the intensity of neuropathological lesions [9–11] and demonstrated good sensitivities and specificities for clinically diagnosed AD versus controls in several single-center cohorts [12,13]. However, in multicenter studies, a large variation of biomarker levels among laboratories has been reported, even in defined studies that have used the same biochemical assay [14,15], which may generate some concern about their routine clinical use [16]. Several quality control studies have found marked interlaboratory variability for CSF biomarker measurements [17–19]. A worldwide, multicenter study in 13 laboratories has reported intersite coefficients of variation (CV) of 37% for Ab 1–42 and 16% for tau, corresponding to high variation levels [19]. This intercenter variability has led to the use of weighted values of biomarkers levels, for example by linear regression [14] or by data normalization [20], in multicenter studies to compensate for these differences [21]. Many reasons may explain intersite variability in the assessment of CSF biomarkers [18,22]. First, centers may differ from each other by demographic characteristics of the patients, the origin of their recruitment, the severity of the diseases, subject selection, or the diagnostic criteria used by physicians. Second, preanalytical factors may be involved, such as factors associated with the whole process of CSF evaluation, from sample collection to transport, temperature and management before analysis, the nature of polypropylene tubes, and differences in sample handling and in conditions of storage. It is known that the type of collection tube has a strong influence on biomarker levels [23]. Third, analytical factors may be linked to batch-to-batch variations of the enzyme-linked immunosorbent assay (ELISA) kits, laboratory equipment, differences in laboratory procedures, temperature, spectrophotometer measure, and the material used that is not supplied by the manufacturer. These preanalytical or analytical differences seem to be important factors in these interlaboratory variations [18,24,25], and recent recommendations have been proposed to improve these issues [26].

407

Despite this growing concern about CSF biomarker variability in experimental studies, little is known about the use of CSF biomarkers in routine clinical practice [27], and scarce data regarding optimum cutoffs are available. Therefore, we investigated the diagnostic accuracy and optimum threshold values of CSF biomarkers and intercenter variability in a large cohort of subjects from three French memory centers in which the use of lumbar punctures is frequent in the clinical setting to explore cognitive disorders. Furthermore, to evaluate the variability linked to preanalytical and analytical procedures, we also conducted a test–retest study, with the repeated analysis of common CSF samples in each of the three centers using the same ELISA immunoassay batch for Ab 1–42, tau, and pTau-181. 2. Methods 2.1. Study participants Subjects were recruited from January 1, 2008, to and December 31, 2010, from three French clinical and research memory centers that specialize in the care and management of patients with cognitive disorders. These centers are based in Paris, Lille, and Montpellier. We included in this study all subjects from these centers for whom a lumbar puncture had been performed between January 1, 2008, and December 31, 2010, to investigate potential cognitive disorders. Subjects with unknown clinical diagnoses were excluded. Because the aim of this study was to compare CSF biomarkers of subjects with AD with subjects without AD, subjects with mild cognitive impairment (MCI) were excluded. No individual normal control subjects were included. Subjects with psychiatric disorders without dementia were also excluded. All subjects underwent an extensive examination, including clinical and neuropsychological evaluations, biological measurements, and brain imaging. Based on all available information, subjects were classified into one of two groups—AD and non-AD—without taking into account CSF test results. Complex or unclear cases were discussed, and diagnoses were then established by a multidisciplinary team of neurologists, geriatricians, and neuropsychologists. Subjects with AD had to fit criteria for probable AD as defined by NINCDS-ADRDA [2], and subjects with MCI had to meet the usual criteria established by Petersen et al. [28]. Subjects without AD also had other causes of dementia, including frontotemporal dementia, vascular dementia, Lewy body disease, alcoholic dementia, Creutzfeldt-Jakob disease, and other of dementia (Supplementary Table 1). All subjects or caregivers signed a written informed consent for CSF assessments and analysis. This research study was approved by the local ethics committee of each hospital. 2.2. CSF analysis Lumbar punctures were performed on fasting subjects during the month following the clinical diagnosis. CSF was

408

J. Dumurgier et al. / Alzheimer’s & Dementia 9 (2013) 406–413

collected in various polypropylene tubes (model TC10PCS, CML, Paris; model 352095, BD Montpellier; and model 352097, BD, Lille), with standardized conditions preferably between 11 a.m. and 1 p.m. to minimize the influence of diurnal variation of CSF Ab levels [29]. Each CSF sample was transferred to each local laboratory less than 4 hours after collection and was centrifuged at 1000g for 10 minutes at 4
C. A small amount of CSF was used for routine analyses, including total cell count, bacteriological examination, and total protein and glucose levels. CSF was aliquoted in polypropylene tubes of 1.5 mL and stored at –80
C until further analysis. CSF Ab 1–42, tau, and pTau-181 were measured using standardized commercially available tests (INNOTEST sandwich ELISA) according to the manufacturer’s procedures (Innogenetics, Ghent, Belgium). The biological teams involved in CSF analysis were unaware of the clinical diagnosis.

2.3. Test–retest study A group of 32 CSF samples from subjects with and without AD supplied by the three memory centers was sent to the three local laboratories for repeated blind analysis of the biomarkers. Consequently, each sample was analyzed three time (one per local laboratory), using the same ELISA immunoassay batch for CSF Ab 1–42, tau, and pTau-181 in the three laboratories. 2.4. Statistical analyses Participant characteristics were presented overall by center and by diagnosis (AD, non-AD). Subject characteristics were compared using analysis of variance for continuous variables and the c2 test for categorical variables. We computed receiver–operating characteristic (ROC) curves to evaluate the capacity of each CSF biomarker to

Table 1 Characteristics of the studied population Characteristics Overall (n 5 933) Age, y, mean (SD) Women, n (%) MMSE, mean (SD) AD, n (%) Ab 1–42, pg/mL, mean (SD) Tau, pg/mL, mean (SD) pTau-181, pg/mL, mean (SD) Ab/tau Ab/p-Tau AD (n 5 515) Age, y, mean (SD) Women, n (%) MMSE, mean (SD) Ab 1–42, pg/mL, mean (SD) Tau, pg/mL, mean (SD) pTau 181, pg/mL, mean (SD) Ab/tau Ab/pTau Non-AD (n 5 365) Age, y, mean (SD) Women, n (%) MMSE, mean (SD) Ab 1–42, pg/mL, mean (SD) Tau, pg/mL, mean (SD) p-Tau 181, pg/mL, mean (SD) Ab/tau Ab/pTau

Overall (n 5 880)

AD (n 5 515) 71.5 (9.5) 301 (58.5) 18.8 (6.2)

Non-AD (n 5 365)

P value*

Paris (n 5 190)

Lille (n 5 350)

Montpellier (n 5 340)

P value*

,.001 ,.001 ,.001 – ,.001

70.5 (11.0) 106 (55.8) 20.5 (5.8) 136 (71.6) 514.1 (235.6)

70.5 (9.7) 199 (56.9) 19.3 (6.3) 222 (63.4) 392.1 (183.3)

67.9 (11.0) 168 (49.4) 20.2 (6.9) 157 (46.2) 606.0 (254.0)

.002 .12 .09 ,.001 ,.001

69.5 (10.6) 473 (53.8) 19.9 (6.4) 51.5 (58.5) 501.1 (243.3)

– 426.8 (199.5)

66.7 (11.4) 172 (47.1) 21.6 (6.3) – 605.9 (260.6)

478.8 (324.9)

599.9 (319.1)

308.1 (247.6)

,.001

495.8 (321.8)

474.9 (316.8)

473.5 (335.4)

.72

74.5 (42.3)

92.1 (41.9)

49.6 (28.0)

,.001

84.0 (42.9)

80.1 (43.8)

63.4 (38.0)

,.001

1.8 (1.8) 9.7 (8.2)

1.0 (1.0) 5.9 (4.8)

3.0 (2.1) 15.1 (8.9)

,.001 ,.001

1.8 (1.8) 8.4 (6.5)

1.3 (1.2) 6.6 (4.9)

2.3 (2.2) 13.6 (9.9)

,.001 ,.001

– – – –

– – – –

– – – –

– – – –

73.2 (8.8) 81 (59.6) 20.0 (5.6) 464.7 (211.4)

71.5 (9.2) 137 (61.7) 18.3 (6.4) 341.3 (156.1)

70.2 (10.2) 83 (52.9) 18.2 (6.4) 514.8 (196.3)

.03 .22 .02 ,.001

–

–

–

–

588.5 (303.5)

574.1 (322.8)

646.3 (323.8)

.08

–

–

–

–

97.6 (40.9)

94.1 (43.5)

84.6 (39.8)

.02

– –

– –

– –

– –

1.1 (0.9) 5.8 (4.0)

0.9 (0.8) 4.5 (3.1)

1.2 (1.3) 7.9 (6.4)

.01 ,.001

– – – –

– – – –

– – – –

– – – –

63.7 (13.0) 25 (46.3) 21.9 (6.0) 638.5 (249.3)

68.8 (10.3) 62 (48.4) 21.2 (5.7) 480.1 (194.1)

66.0 (11.3) 85 (46.5) 22.0 (6.9) 684.2 (271.7)

.01 .93 .57 ,.001

–

–

–

–

262.3 (239.1)

302.8 (217.9)

325.3 (268.0)

.25

–

–

–

–

50.0 (25.4)

55.8 (32.1)

45.2 (24.7)

.004

– –

– –

– –

– –

3.7 (2.3) 15.0 (7.1)

2.1 (1.2) 10.3 (5.2)

3.3 (2.3) 18.4 (9.9)

,.001 ,.001

Abbreviations: Ab, amyloid beta; AD, Alzheimer’s disease; MMSE, Mini-Mental State Examination; pTau, phosphorylated tau; SD, standard deviation. *c2 test or analysis of variance.

J. Dumurgier et al. / Alzheimer’s & Dementia 9 (2013) 406–413

409

discriminate subjects with AD from those without AD. Analyses were stratified by centers. Areas under the curve (AUCs) for the various biomarkers were compared using nonparametric tests and the %ROC macro in SAS software. Optimum cutoffs for each biomarker to discriminate in the best possible manner subjects with AD from subjects without AD were defined using the highest Youden index, thus maximizing sensitivity and specificity based on ROC curve analyses. For the test–retest study, we first computerized the means of the 32 CVs (i.e., the standard deviation divided by the mean of the repeated three measurements) for each biomarker. Then, to analyze intersite variability, we compared each center with the other two using the methods outlined by Bland and Altman [30], which consisted of calculating both the average between each pair of measurements (reflecting systematic bias) and the standard deviation (SD) of the differences (reflecting precision of the measurement). Systematic bias linked to an effect of the center was tested comparing the mean of the difference with zero (paired t-test). Precision of the repeated measurement was assessed using the coefficient of repeatability (1.96 ! SD of the difference) expressed as a percentage of the mean measurement of the biomarker. All resulting P values were two-tailed, and P  .05 was considered statistically significant. Statistical analyses were performed using SAS version 9.2 (SAS Institute, Cary, NC, USA). 3. Results 3.1. Cohort creation Between January 1, 2008, and December 31, 2010, 1127 subjects had a lumbar puncture in one of the three study centers to investigate their cognitive disorders. Among them, 65 (6%) had no clear diagnosis (unclear even for dementia experts) and were thus excluded. One hundred twenty-nine subjects with a diagnosis of MCI and 53 subjects with a diagnosis of psychiatric disorders without dementia were also excluded. A total of 880 subjects were included in the analytic study (Paris, n 5 190; Lille, n 5 350; Montpellier, n 5 340) and their characteristics are summarized in Table 1. Compared with subjects without AD, subjects with AD were older, more often women, had lower Mini-Mental State Examination scores, and had more pronounced CSF abnormalities (lower Ab 1–42, and higher tau and pTau-181). 3.2. Intersite differences The proportion of subjects with AD was different in the three sites, ranging from 44% to 68%. Considering the various sites, subjects with AD varied by mean age (range, 70.2–73.2 years; P 5.03), their mean Mini-Mental State Examination score (range, 18.2–20.0; P 5 .02), and their mean levels of CSF Ab 1–42 (range, 341.3–514.8 pg/mL) and

Fig. 1. Receiver–operator characteristics. (A–C) Cerebrospinal fluid (CSF) biomarkers of subjects with Alzheimer’s disease (AD; n 5 515) were compared with a group of subjects without AD (n 5 365) in the three centers. Abeta 1-42, amyloid beta 1–42; AUC, area under the curve; pTau, phosphorylated tau.

410

J. Dumurgier et al. / Alzheimer’s & Dementia 9 (2013) 406–413

pTau-181 (range, 84.6–97.6 pg/mL). Mean tau levels did not differ significantly among sites (range, 574.1–646.3 pg/mL). 3.3. ROC curves analyses ROC curves comparing subjects with AD with subjects without AD (Fig. 1 and Table 2) show the values of AUC for the various biomarkers and their optimum values to discriminate at best subjects with AD from subjects without AD. In all three centers, tau (AUC range, 0.80–0.88) and pTau-181 (range, 0.83–0.89) outperformed Ab 1–42 (range, 0.70–0.73) for AD diagnosis (comparison of AUC among the three centers, P ,.001). There were no significant differences among AUC of tau and pTau-181 and the ratios of Ab to tau and Ab to pTau-181. To discriminate best subjects with AD from subjects without AD, optimum cutoffs were determined for each site. The cutoff ranges were as follows: for Ab 1–42, 368–582 pg/mL; for tau, 289–353 pg/mL; and for pTau-181, 54–65 pg/mL. 3.4. Test–retest study A test–retest study was performed for the 32 CSF samples that were analyzed three times, one time in each of the three local laboratories (total: 96 measures), using the same ELISA batch for CSF Ab 1–42, for tau, and for pTau-181 in the three laboratories. Mean values of CSF Ab 1–42 ranged from 898.8 to 958.8 pg/mL, from 249.9 to 250.6 pg/mL for Tau, and from 53.6 to 60.3 pg/mL for pTau181. The mean (SD) intersite coefficients of variation were 12.2 (8.7)% for Ab 1–42, 11.3 (7.0)% for tau, and 11.5 (5.8)% for pTau-181. Fig. 2 shows the Bland–Altman plots for the repeated measures of the biomarkers in the three centers, and a detailed analysis is displayed in Table 3. The means of the intersite differences were not significantly different from zero

for Ab 1–42 and for tau in the three centers, and were significantly differed from zero for pTau-181. The precision of intersite assessment for pTau-181 was higher (coefficient of repeatability [CR] ranged from 20%–24%) than the precisions for Ab 1–42 (CR, 35%–46%) or tau (CR, 38%–63%). 4. Discussion In this clinically based study, our results revealed clear intersite variation of levels and cutoffs for the three CSF biomarkers Ab 1–42, tau, and pTau-181 used to discriminate subjects with AD from subjects without AD in routine clinical use. These variations were more pronounced for Ab 1–42 than for tau. Despite this interlaboratory variability, the CSF biomarkers had comparable power to discriminate subjects with AD from subjects without AD at each site, and we report that tau and pTau-181 outperformed Ab 1–42 in discriminating patients. Although many previous studies have investigated the findings in subjects with AD or MCI versus healthy control subjects, our results illustrate an interest in CSF biomarkers in discriminating AD from other dementias in a clinical setting, reflecting real-life practice in specialized memory clinics. Several studies [15,31,32] have emphasized previously that, concerning CSF evaluations, multicenter studies experience variations in processes, and authors conclude that results from multisite studies have lower diagnostic accuracy than results from homogenous single-center evaluations. A recent single-center study confirmed that careful procedures in CSF handling may improve the diagnostic accuracy of CSF biomarkers [12]. Unfortunately, site-specific habits in managing lumbar punctures and CSF procedures are different all over the world, but an effort to improve these procedures is on the way, as seen in the Alzheimer’s Association’s external quality control program for CSF biomarkers [18,33].

Table 2 Area under receiver–operating characteristic curves and optimal values to discriminate subjects with Alzheimer’s disease (AD) from those without CSF biomarkers, pg/mL Paris Ab 1–42 Tau pTau-181 Ab/tau Ab/pTau Lille Ab 1–42 Tau pTau-181 Ab/tau Ab/pTau Montpellier Ab 1–42 Tau pTau-181 Ab/tau Ab/pTau

AUC (SE)

Optimal value

Sensitivity (95% CI)

Specificity (95% CI)

Youden index

0.72 (0.04) 0.88 (0.03) 0.89 (0.03) 0.87 (0.03) 0.87 (0.03)

515 289 65 1.4 6.8

0.76 (0.69–0.83) 0.90 (0.85–0.95) 0.85 (0.78–0.91) 0.80 (0.71–0.91) 0.76 (0.69–0.83)

0.65 (0.52–0.78) 0.83 (0.73–0.93) 0.89 (0.81–0.97) 0.81 (0.74–0.88) 0.87 (0.78–0.97)

1.41 1.74 1.74 1.62 1.64

0.73 (0.03) 0.80 (0.03) 0.83 (0.02) 0.83 (0.02) 0.84 (0.02)

368 353 59.2 1.1 6.3

0.71 (0.65–0.77) 0.70 (0.64–0.76) 0.82 (0.76–0.86) 0.79 (0.73–0.84) 0.82 (0.77–0.87)

0.70 (0.62–0.78) 0.78 (0.71–0.85) 0.77 (0.70–0.85) 0.77 (0.69–0.85) 0.82 (0.69–0.84)

1.41 1.48 1.59 1.56 1.59

0.70 (0.03) 0.82 (0.02) 0.83 (0.02) 0.83 (0.02) 0.83 (0.02)

582 322 54 1.6 12.2

0.73 (0.66–0.80) 0.85 (0.79–0.91) 0.83 (0.77–0.89) 0.86 (0.81–0.92) 0.86 (0.81–0.91)

0.63 (0.56–0.70) 0.69 (0.63–0.76) 0.76 (0.69–0.82) 0.73 (0.67–0.80) 0.72 (0.66–0.79)

1.37 1.54 1.59 1.59 1.58

Abbreviations: Ab, amyloid beta; AUC, area under the curve; CI, confidence interval; CSF, cerebrospinal fluid; pTau, phosphorylated tau; SE, standard error.

J. Dumurgier et al. / Alzheimer’s & Dementia 9 (2013) 406–413

411

Fig. 2. For the test–retest study, 32 cerebrospinal fluid (CSF) samples were measured repeatedly for biomarkers in each of the three centers using the same enzyme-linked immunosorbent assay batch for amyloid beta (Ab) 1–42, tau, and phosphorylated tau (pTau-181), measured in pictograms per milliliter. The Bland–Altman plots show the differences in measures (ordinate) against their mean value (ordinate). Ninety-five percent limits of agreement are represented in ordinate (gray area).

In the test–retest study, after controlling intersite preanalytical variability (using similar samples in the three centers) and decreasing assay variability (using the same ELISA batch in the three laboratories), we found that the mean intersite CV of biomarkers ranged from 11.3% to 12.2%. These coefficients are lower than those previously reported in other external quality controls studies, in which CV ranged from 15% to 35% [17–19]. This result suggests that intersite variability of biomarkers may be explained, in part, by the variability of immunoassay batches used for the measurements. In our Bland-Altman analysis, we found no significant effect for the center for the repeated assessments of Ab 1–42 and tau, although there was a significant effect for the center for pTau-181. In contrast, the precision of the repeated measurements was greater for pTau-181 than for tau and Ab 1–42. The current study has several strengths—its large size, multicenter design, inclusion of patients supervised by physicians with real clinical practice from memory centers that

routinely propose a lumbar puncture to investigate cognitive disorders, and the use of a test–retest study to attempt to decrease preanalytical and analytical intersite factors. However, some limitations must be considered. First, to avoid bias, diagnoses were assessed clinically by teams of trained experts before they were made aware of CSF biomarker results. However, neuropathological validations of AD diagnosis were performed in only 22 subjects. Furthermore, we did not assess other biomarkers, such as CSF Ab1–40, which may be of interest in discriminating subjects with AD from subjects without AD [34]. This could also apply to new CSF biomarkers, as recently proposed [35]. In addition, imaging analysis of amyloid plaques was not performed during this study based on routine procedures; but, in the future, this new assessment will certainly also improve the accuracy of AD diagnosis. In conclusion, intersite variability of CSF biomarkers has been confirmed in clinical settings in three French memory

412

J. Dumurgier et al. / Alzheimer’s & Dementia 9 (2013) 406–413

Table 3 During the test–retest study, 32 cerebrospinal fluid (CSF) samples were measured repeatedly for biomarkers in each of the three centers using the same enzymelinked immunosorbent assay batch for amyloid beta (Ab) 1–42, tau, and phosphorylated tau (pTau-181), Bland-Altman analysis CSF biomarkers Ab 1–42 P vs. L P vs. M L vs. M Tau P vs. L P vs. M L vs. M pTau-181 P vs. L P vs. M L vs. M

CRy

Mean (SD) of 2 sets of measurements

CR as percentage of mean

.85 .16 .15

417.5 395.7 326.9

911.2 (374.6) 933.2 (381.9) 936.8 (374.4)

45.8 42.4 34.9

7.6 (49.6) 7.0 (63.8) 20.6 (80.5)

.39 .54 .97

97.2 125.0 157.8

253.8 (187.0) 254.1 (157.6) 250.3 (165.2)

38.3 49.2 63.0

29.9 (6.7) 23.1 (5.4) 6.8 (5.9)

,.001 .003 ,.001

13.1 10.6 11.6

55.4 (27.1) 52.0 (27.0) 57.0 (28.3)

23.6 20.4 20.4

95% Limits of agreement, pg/mL

Mean (SD) of the difference, pg/mL

2424.8, 410.2 2447.0, 334.5 2370.8, 282.9

27.3 (213.0) 251.3 (201.9) 243.9 (166.8)

289.7, 104.9 2117.9, 131.9 2158.3, 157.1 223.0, 3.3 213.8, 7.6 24.7, 18.3

P value *

Abbreviations: CR, coefficient of repeatability; L, Lille; M, Montpellier; P, Paris; SD, standard deviation. *Difference to zero of the mean, paired t-test. y 1.96 ! SD of the mean of the difference.

centers. Interventional efforts on harmonization of preanalytical and analytical procedures should be put forward to continue optimizing the accuracy and clinical use of CSF biomarkers in AD. Acknowledgments The authors thank the other members of the teams for their contribution to this study: Drs. Florence Lebert, Marie-Anne Mackowiak, Vincent Deramecourt, Adeline Rollin-Sillaire, Pascaline Cassagnaud, Karim Bennys, Christian Geny, Eric Thouvenot, Sarah Benisty, and Claire Kiffel, and all the neuropsychologists. The authors warmly thank professors Jo€el M
enard, Herv
e Maisonneuve, Jean Marie Goehrs, and Patrick Trunet, and Mme. Annick Alp
erovitch for their very helpful advice and for revising this manuscript. They also thank Mme. Sylvie Ledoux and the Fondation Plan Alzheimer for logistic support. References [1] Ferri CP, Prince M, Brayne C, Brodaty H, Fratiglioni L, Ganguli M, et al. Global prevalence of dementia: a Delphi consensus study. Lancet 2005;366:2112–7. [2] McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan EM. Clinical diagnosis of Alzheimer’s disease: report of the NINCDS-ADRDA Work Group under the auspices of the Department of Health and Human Services Task Force on Alzheimer’s Disease. Neurology 1984;34:939–44. [3] Knopman DS, DeKosky ST, Cummings JL, Chui H, Corey-Bloom J, Relkin N, et al. Practice parameter: diagnosis of dementia (an evidence-based review): report of the Quality Standards Subcommittee of the American Academy of Neurology. Neurology 2001; 56:1143–53. [4] Shaw LM, Korecka M, Clark CM, Lee VM, Trojanowski JQ. Biomarkers of neurodegeneration for diagnosis and monitoring therapeutics. Nat Rev 2007;6:295–303. [5] Petersen RC. Early diagnosis of Alzheimer’s disease: Is MCI too late? Current Alzheimer Res 2009;6:324–30.

[6] McKhann GM, Knopman DS, Chertkow H, Hyman BT, Jack CR Jr, Kawas CH, et al. The diagnosis of dementia due to Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 2011;7:263–9. [7] Blennow K, Hampel H, Weiner M, Zetterberg H. Cerebrospinal fluid and plasma biomarkers in Alzheimer disease. Nat Rev Neurol 2010; 6:131–44. [8] Blennow K, Zetterberg H. Cerebrospinal fluid biomarkers for Alzheimer’s disease. J Alzheimers Dis 2009;18:413–7. [9] Tapiola T, Alafuzoff I, Herukka SK, Parkkinen L, Hartikainen P, Soininen H, et al. Cerebrospinal fluid {beta}-amyloid 42 and tau proteins as biomarkers of Alzheimer–type pathologic changes in the brain. Arch Neurol 2009;66:382–9. [10] Dumurgier J, Paquet C, Benisty S, Kiffel C, Lidy C, Mouton-Liger F, et al. Inverse association between CSF Abeta 42 levels and years of education in mild form of Alzheimer’s disease: the cognitive reserve theory. Neurobiol Dis 2010;40:456–9. [11] Dumurgier J, Paquet C, Peoc’h K, Lapalus P, Mouton-Liger F, Benisty S, et al. CSF Abeta levels and glucose metabolism in Alzheimer’s disease. J Alzheimers Dis 2011;27:845–51. [12] Johansson P, Mattsson N, Hansson O, Wallin A, Johansson JO, Andreasson U, et al. Cerebrospinal fluid biomarkers for Alzheimer’s disease: diagnostic performance in a homogeneous mono-center population. J Alzheimers Dis 2011;24:537–46. [13] Blennow K. Cerebrospinal fluid protein biomarkers for Alzheimer’s disease. NeuroRx 2004;1:213–25. [14] Mattsson N, Zetterberg H, Hansson O, Andreasen N, Parnetti L, Jonsson M, et al. CSF biomarkers and incipient Alzheimer disease in patients with mild cognitive impairment. JAMA 2009;302:385–93. [15] Visser PJ, Verhey F, Knol DL, Scheltens P, Wahlund LO, FreundLevi Y, et al. Prevalence and prognostic value of CSF markers of Alzheimer’s disease pathology in patients with subjective cognitive impairment or mild cognitive impairment in the DESCRIPA study: a prospective cohort study. Lancet Neurol 2009;8:619–27. [16] Parnetti L, Chiasserini D. Role of CSF biomarkers in the diagnosis of prodromal Alzheimer’s disease. Biomarkers Med 2011;5:479–84. [17] Lewczuk P, Beck G, Ganslandt O, Esselmann H, Deisenhammer F, Regeniter A, et al. International quality control survey of neurochemical dementia diagnostics. Neurosci Lett 2006;409:1–4. [18] Mattsson N, Andreasson U, Persson S, Arai H, Batish SD, Bernardini S, et al. The Alzheimer’s Association external quality control program for cerebrospinal fluid biomarkers. Alzheimers Dement 2011;7:386–95.

J. Dumurgier et al. / Alzheimer’s & Dementia 9 (2013) 406–413 [19] Verwey NA, van der Flier WM, Blennow K, Clark C, Sokolow S, De Deyn PP, et al. A worldwide multicentre comparison of assays for cerebrospinal fluid biomarkers in Alzheimer’s disease. Ann Clin Biochem 2009;46:235–40. [20] Ewers M, Schmitz S, Hansson O, Walsh C, Fitzpatrick A, Bennett D, et al. Body mass index is associated with biological CSF markers of core brain pathology of Alzheimer’s disease. Neurobiol Aging 2012;33:e1–2. [21] Petersen RC, Trojanowski JQ. Use of Alzheimer disease biomarkers: potentially yes for clinical trials but not yet for clinical practice. JAMA 2009;302:436–7. [22] Mattsson N, Zetterberg H, Blennow K. Lessons from multicenter studies on CSF biomarkers for Alzheimer’s disease. Int J Alzheimers Dis 2010. [23] Perret-Liaudet A, Pelpel M, Tholance Y, Dumont B, Vanderstichele H, Zorzi W, et al. Cerebrospinal fluid collection tubes: a critical issue for Alzheimer disease diagnosis. Clin Chem 2012; 58:787–9. [24] Bateman RJ, Wen G, Morris JC, Holtzman DM. Fluctuations of CSF amyloid-beta levels: implications for a diagnostic and therapeutic biomarker. Neurology 2007;68:666–9. [25] Shaw LM, Vanderstichele H, Knapik-Czajka M, Figurski M, Coart E, Blennow K, et al. Qualification of the analytical and clinical performance of CSF biomarker analyses in ADNI. Acta Neuropathol 2011;121:597–609. [26] Vanderstichele H, Bibl M, Engelborghs S, Le Bastard N, Lewczuk P, Molinuevo JL, et al. Standardization of preanalytical aspects of cerebrospinal fluid biomarker testing for Alzheimer’s disease diagnosis: a consensus paper from the Alzheimer’s Biomarkers Standardization Initiative. Alzheimers Dement 2012;8:65–73. [27] Kester MI, Boelaarts L, Bouwman FH, Vogels RL, Groot ER, van Elk EJ, et al. Diagnostic impact of CSF biomarkers in a local

[28]

[29]

[30] [31]

[32]

[33]

[34]

[35]

413

hospital memory clinic. Dement Geriatr Cogn Disord 2010; 29:491–7. Petersen RC, Smith GE, Waring SC, Ivnik RJ, Tangalos EG, Kokmen E. Mild cognitive impairment: clinical characterization and outcome. Arch Neurol 1999;56:303–8. Slats D, Claassen JA, Spies PE, Borm G, Besse KT, Aalst WV, et al. Hourly variability of cerebrospinal fluid biomarkers in Alzheimer’s disease subjects and healthy older volunteers. Neurobiol Aging 2012;33:e831–9. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307–10. Shaw LM, Vanderstichele H, Knapik-Czajka M, Clark CM, Aisen PS, Petersen RC, et al. Cerebrospinal fluid biomarker signature in Alzheimer’s disease neuroimaging initiative subjects. Ann Neurol 2009;65:403–13. Buerger K, Frisoni G, Uspenskaya O, Ewers M, Zetterberg H, Geroldi C, et al. Validation of Alzheimer’s disease CSF and plasma biological markers: the multicentre reliability study of the pilot European Alzheimer’s Disease Neuroimaging Initiative (E-ADNI). Exp Gerontol 2009;44:579–85. Mattsson N, Blennow K, Zetterberg H. Inter-laboratory variation in cerebrospinal fluid biomarkers for Alzheimer’s disease: united we stand, divided we fall. Clin Chem Lab Med 2010;48:603–7. Spies PE, Slats D, Sjogren JM, Kremer BP, Verhey FR, Rikkert MG, et al. The cerebrospinal fluid amyloid beta42/40 ratio in the differentiation of Alzheimer’s disease from non-Alzheimer’s dementia. Curr Alzheimer Res 2010;7:470–6. Mouton-Liger F, Paquet C, Dumurgier J, Lapalus P, Gray F, Laplanche JL, et al. Increased cerebrospinal fluid levels of doublestranded RNA-dependant protein kinase in Alzheimer’s disease. Biol Psychiatry 2012;71:829–35.