Study reference

Study characteristics

Patient characteristics

Index test

(test of interest)

Reference test

Follow-up

Outcome measures and effect size

Comments

Ritchie, 2014

Study characteristics and results for individual studies are extracted from the SR Ritchie, 2014 (unless stated otherwise)

SR

Literature search up to December 2012

A: Bjerke 2009

B: Blom 2009

C: Chiasserini 2010

D: Galluzzi 2010

E: Hampel 2004

F: Hansson 2007

G: Hertze 2010

H: Kester 2011

I: Papaliagkas 2009

J: Monge-Argiles 2011

K: Parnetti 2006

L: Shaw 2009

M: Vos 2013

N: Zetterberg 2003

Study design: prospective cohort

Setting and Country: Participants were recruited from i) secondary care – outpatient clinic (n = 12); ii) secondary care – inpatients (n = 2); iii) community care (n = 2) and mixed setting (n = 1).

Source of funding and conflicts of interest: Not reported

Inclusion criteria SR:

1) prospectively well-defined cohorts

2) any accepted definition of MCI but no dementia

3) baseline CSF or plasma Aß levels, or both, documented at or around the time the MCI diagnosis was made

4) reference standard for Alzheimer’s dementia diagnosis based on the National Institute for National Institute of Neurological and Communicative Diseases and Stroke/ Alzheimer's Disease and Related Disorders Association (NINCDS- ADRDA) or the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) criteria

5) sufficient data to construct two by two tables expressing biomarker results by disease status.

Exclusion criteria SR: Studies were excluded if they included patients with possible other causes for dementia: psychiatric, neurological, metabolic, immunological, hormonal or cerebrovascular disorders, genetic cause or early onset ADD (if studies included patients below the age of 50 years, they were excluded).

14 studies included

Important patient characteristics:

Number of patients; characteristics important to the research question; for example, age, sex, bmi, ...

N, mean age

A: 162 patients, 63.8 yrs

B: 28 patients, 63.7 yrs

C: 41 patients, 66.0 yrs

D: 64 patients, 71.6 yrs

E: 52 patients, 72.5 yrs

F: 134 patients, 70.8

G: 159 patients, 72 years

H: 100 patients, 67.8 yrs

I: 53 patients, 68.7 yrs

J: 37 patients, 73.4 yrs

K: 55 patients, age NR

L: 196 patients, 75.2 yrs

M: 214 patients, 70.7 years

N: 53 patients; mean age NR

Sex:

A: 68/162 (42.0% male)

B: 28/58 (46.4% male)

C: 20/41 (48.9% male)

D: 37/90 (41.0% male)

E: 24/52 (46.2% male)

F: 58/131 (32.8% male)

G: 65/159 (40.8% male)

H: 59/100 (59.0% male)

I: 22/53 (41.5% male)

J: 13/37 (35.1% male)

K: gender NR

L: 131/196 (66.8% male)

M: 270/625 (43.2% male)

N: gender NR

Describe index and comparator tests* and

cut-off point(s):

A: CSF Abeta42; 512 ng/l

B: CSF Abeta42; 82 pM

C: CSF Abeta42; 741 pg/ml

D: CSF Abeta42; 500 pg/ml

E: CSF Abeta 1-42; 679 ng/l

F-1: CSF Abeta42; 0.64 ng/ml

F-2: Abeta42/Abeta40 ratio; 0.95 ng/l

G: CSF Abeta42; 209 pg/ml

H: CSF Abeta42; 495 pg/ml

I: CSF Abeta42; 472 pg/ml

J: CSF Abeta42; 320 pg/ml

K: CSF Abeta42; 500 pg/l

L: CSF Abeta42; 192 pg/ml

M: CSF Abeta1-42; 500 pg/ml

N: CSF Abeta42; 532 ng/l

.....

[use the format ‘character+number’ if 2 or more index tests are being compared, e.g. A-1, A-2, etc]

Describe reference test and cut-off point(s):

NB: Cut-off NA for all reference tests

A: NINDS-ADRDA and ICD-10 criteria

B: NINCDS-ADRDA

C: NINCDS-ADRDA

D: NINCDS-ADRDA

E: NINCDS-ADRDA, DSM-IV criteria

F: NINCDS-ADRDA, DSM-III-R

G: NINCDS-ADRDA, DSM-II-R criteria

H: NINCDS-ADRDA

I: the criteria used to define progression to ADD were not clearly defined

J: NINCDS-ADRDA

K: NINCDS-ADRDA

L: NINCDS-ADRDA

M: NINCDS-ADRDA, DSM-IV criteria

N: DSM-IV

Prevalence (%)

[based on reference test at specified cut-off point]

A: 20/174 (11.5%)

B: 14/28 (50%)

C: 23/41 (56.1%)

D: 39/90 (43.3%)

E: 29/52 (55.8%)

F: 57/134 (42.5%)

G: 55/166 (33.1%)

H: 42/107 (39.3%)

I: 5/53 (9.43%)

J: 11/37 (29.7%)

K: 11/55 (20%)

L: 37/196 (18.9%)

M: 91/214 (42.5%)

N: 22/53 (41.5%)

For how many participants were no complete outcome data available?

N (%)

A: 12/162 (7.41%)

B: 30/58 (51.7%)

C: 0 (0%)

D: 44/64 (68.8%)

E: 0 (0%)

F: 3/134 (2.24%)

G: 7/160 (4.38%)

H: 53/100 (53.0%)

I: 0 (0%)

J: 0 (0%)

K: 0 (0%)

L: 4/196 (2.04%)

M: 17/214 (7.94%)

N: 0 (0%)

Reasons for incomplete outcome data described?

A: 8 patients had improved their condition at follow-up, 3 patients converted to a dementia disorder where no etiology could be clinically established, 1 patient converted to primary progressive aphasia

B: reasons NR

C: NA, because everyone had outcome data

D: 16 refused follow-up assessment, 2 did not follow-up due to logistic problems, 24 refused LP, 2 failure to reach the arachnoid space due to osteoarthrosis

E: NA, because everyone had outcome data

F: 3 participants died before completion of 4 years follow-up and were excluded from the analyses because their cognitive ability was uncertain

G: xMAP analysis resulted in technical errors

H: 7 MCI participants who converted to other dementias were excluded from the analysis; 46 participants did not have follow-up data; no further information

I: NA, because everyone had outcome data

J: NA, because everyone had outcome data

K: NA, because everyone had outcome data

L: reasons NR

M: some participants refused to participate or were untraceable or died before follow-up

N: NA, because everyone had outcome data

Endpoint of follow-up:

Outcomes were assessed

A: max 4 years

B: mean 3.4 years; max 8.0 years

C: max 4 years

D: mean 2 years

E: mean 0.7 years

F: mean 5.2 years; max 6.8 years

G: mean 4.7 years; max 7.2 years)

H: mean 1.5 years; max 2.0 years

I: mean 0.9 years; max 1.8 years

J: 0.5 years

K: max 1.0 years

L: max 1.0 years

M: mean 2.5 years (max 5 years)

N: mean 1.67 years

Outcome measures and effect size (include 95%CI and p-value if available):

Outcome measure-1/2 for Abeta42

e.g. sensitivity / specificity (%)], mean (95% CI)

A: 90 [68, 99]% / 70 [61, 77]%

B: 64 [35, 87]% / 36 [13, 65]%

C: 74 [52, 90]% / 89 [65, 99]%

D: 82 [60, 95]% / 60 [43, 74]%

E: 83 [64, 94]% / 57 [34, 77]%

F: 93 [83, 98]% / 53 [42, 65]%

G: 90 [79, 97]% / 69 [60, 78]%

H: 76 [61, 88]% / 72 [59, 83]%

I: 100 [48, 100]% / 77 [63, 88%]

J: 82[48, 98]% / 62 [41, 80]%

K: 36 [11, 69]% / 91 [78, 97]%

L: 86 [71, 95]% / 29 [22, 37]%

M: 68 [58, 78]% / 67 [58, 75]%

N: 68 [45, 86]% / 61 [42, 78]%

Pooled characteristics:

Because of the variation in assay thresholds, the authors did not estimate summary sensitivity and specificity. Instead they derived estimates of sensitivity at fixed values of specificity from the hierarchical summary receiver operating characteristic (HSROC) model used to produce the ROC curve. At the median specificity of 64%, the sensitivity was 81% (95% CI 72 to 87). This equated to a positive likelihood ratio (LR+) of 2.22 (95% CI 2.00 to 2.47) and a negative likelihood ratio (LR–) of 0.31 (95% CI 0.21 to 0.48).

Study quality (ROB): method used and results per individual study

Methods used: QUADAS-2 tool

Patient selection domain

Studies C, F, G, I, and M scored unclear risk of bias, mainly because of poor reporting.

Studies E, H, L scored high risk of bias, mainly because the participants were not consecutively or randomly enrolled.

Index test domain

Studies A, B, C, E, F, I, J, and N scored high risk of bias, mainly because the threshold used was not pre-specified and the optimal cut-off level was determined from ROC analyses; therefore, the accuracy of the plasma and CSF Aß biomarkers reported in these studies appeared to be overestimated.

Reference standard domain

Studies A, B, C, D, E, F, H, J, I, K, and L scored unclear risk of bias, mainly because of poor reporting

Flow and timing domain

Studies B, D, E, H, J, I, K, L, and N scored high risk of bias,

because not all patients were accounted for in the analysis or the time interval between the index test and reference standard was not appropriate (duration of follow-up was shorter than one year).

Place of the index test in the clinical pathway: add-on

Choice of cut-off point: The studies defined their own cut-off points. Thus no overall choice was made and this could be a source for heterogeneity.

Author’s conclusion. We conclude that when applied to a population of patients with MCI, CSF Aß levels cannot be recommended as an accurate test for Alzheimer's disease.

Sensitivity analyses:

×Exclusion of one study (Kester 2011) – main results were robust

×Exclusion of 8 studies at high or unclear risk of bias for patient selection domain – main results were robust

×Exclusion of 8 studies at high risk of bias for Index test domain – main results were robust

Ritchie, 2017

Study characteristics and results for individual studies are extracted from the SR Ritchie, 2017 (unless stated otherwise)

SR

Literature search up to January 2013

A: Amlien 2013

B: Hampel 2004

C: Hansson 2006

D: Kester 2011

E: Monge-Argiles 2011

F: Palmqvist 2012

G: Vos 2013

H: Fellgiebel 2007

I: Koivunen 2008

J: Visser 2009

K: Parnetti 2012

Study design: prospective cohort

Setting and Country: Participants were recruited from i) secondary care – outpatient clinic (n = 12); ii) secondary care – inpatients (n = 2); iii) community care (n = 2) and mixed setting (n = 1).

Source of funding and conflicts of interest: Not reported

Inclusion criteria SR: 1) prospectively well-defined cohorts, 2) any accepted definition of MCI but no dementia, 3) CSF t-tau or p-tau and CSF tau (t-tau or p-tau)/Aβ ratio documented at or around the time the MCI diagnosis was made, 4) reference standard for Alzheimer’s dementia diagnosis based on the NINCDS- ADRDA or DSM-IV criteria, 5) sufficient data to construct two by two tables expressing biomarker results by disease status.

Exclusion criteria SR: We excluded those studies that included people with MCI possibly caused by: i) a current or history of alcohol/drug abuse; ii) central nervous system (CNS) trauma (e.g. subdural haematoma), tumour, or infection; iii) other neurological conditions, e.g. Parkinson’s or Huntington’s diseases.

11 studies included

Important patient characteristics:

Number of patients; characteristics important to the research question; for example, age, sex, bmi, ...

N, mean age

A: 49 patients, MCI with abnormal CSF t-tau level: 64 (range 45 to 76); MC with normal CSF t-tau level: 58.5 (45 to 77) yrs

B: 52 patients, mean 72.6 years

C: 137 patients MCI-MCI (stable): 67 (50 to 86) yrs; MCI-AD 75 (59 to 85) yrs; MCI-other dementia 76 (54 to 82) yrs

D: 153 patients, mean 67 ± 9 yrs 9 ± 7 yrs MCI-AD

E: 37 patients, mean 73.43 ± 6.63 years

F: 133 patients, MCI-MCI: mean 69.8 (range 55 to 85) yrs; MCI-AD: 75.3 (range 55 to 87) yrs; MCI-other dementias: 71.2 (59 to 83) yrs

G: 231 patients, 70.7 ± 7.6 yrs naMCI; 70.7 yrs ± 7.8 aMC

H: 16 patients, total sample: mean age 68.6 ± 7.9 yrs; MCI-MCI: 68.8 ± 10.0 yrs; MCI-progressive: 68.5 ± 5.9 yrs (4/8 MCI-AD: 69.5 ± 7.9 yrs)

I: 15 patients, mean age 71.1 ± 7.2 yrs

J: 168 patients, 70.0 ± 7.7 yrs naMCI; 70.0 ± 7.7 yrs aMCI; 66.0 ± 7.9 yrs SCI

K: 90 patients, MCI-MCI: mean 66.35 ± 8.22 yrs; MCI-AD: mean 67.23 ± 9.04 yrs

Sex:

A: 20/39 (51.3% male)

B: 24/52 (46.2% male)

C: 60/133 (45.1% male)

D: 59/100 (59% male)

E: 13/37 (35.1% male)

F: MCI-AD: 16/52 (30.8% male)

G: 270/605 (44.6% male)

H: 9/16 (56.3% male)

I: 9/15 (60% male)

J: 88/168 (52.4% male)

K: 66/100 (66% male)

Describe index and comparator tests* and

cut-off point(s):

A: CSF t-tau; ≥ 300 ng/L for age younger than 50 years; ≥ 450 ng/L for age 50 to 69 years; ≥ 500 ng/ L for age older than 70 years (Sjogren 2001)

B: CSF t-tau; ≥ 479 ng/L

C-1: CSF t-tau; > 350 ng/L

C-2: CSF p-tau; ≥ 60 ng/L

C-3: CSF p-tau/ABeta ratio; ˂ 6.5 ng/L

D: CSF t-tau; > 356 pg/mL abnormal level;

E-1: CSF t-tau; ≥77.5 pg/mL

E-2: CSF p-tau; ≥ 54.5 pg/mL

E-3: CSF t-tau/ABeta ratio; 0.18

E-4: CSF p-tau/ABeta ratio; 0.18

F-1: CSF t-tau; > 87 pg/mL

F-2: CSF p-tau; >39 pg/mL

F-3: CSF t-tau/ABeta ratio; cut-off NR

G-1: CSF t-tau; > 450 pg/mL for age less than 70 years; > 500 pg/mL for age older than 70 years

G-2: CSF t-tau/ABeta ratio; ABeta1–42/(240 1 [1.183 t-tau]) ˂ 1.0

H: CSF p-tau; ≥ 50 pg/mL

I-1: CSF p-tau; ≥ 70 pg/mL

I-2: CSF p-tau / ABeta ratio; <6.5 pg/mL

J-1: CSF p-tau, used in clinical practice; ≥51 pg/mL

J-2: CSF p-tau >90 percentile of controls after correction of age; ≥85pg/mL

J-3: CSF p-tau/ABeta ratio; <9.92 (<10^th percentile of reference group after correction for age)

K: CSF p-tau / ABeta ratio; 1074.0

Describe reference test and cut-off point(s):

NB: cut-off points NR

A: The Global Deterioration Scale (Reisberg 1982) in combination with the research criteria for the diagnosis of Alzheimer's disease, International Working group (Dubois 2007)

B: NINCDS-ADRDA criteria; DSM-IV criteria

C: NINCDS-ADRDA and DSM-III-R for Alzheimer's disease dementia

D: NINCDS-ADRDA criteria for Alzheimer's disease dementia

E: NINCDS-ADRDA criteria

F: NINCDS-ADRDA for AD;

G: NINCDS-ADRDA criteria; DSM-IV criteria

H: progression to Alzheimer's disease dementia was assumed if CDR reached 1

I: NINCS-ADRDA criteria; DSM-IV criteria

J: NINCS-ADRDA criteria; DSM-IV criteria

K: not specified. MCI participants were clinically evaluated at least once a year during

4-year follow-up period. However, it was reported that the NINCDS-ADRDA criteria were used at

baseline to identify AD diagnostic group.

Prevalence (%)

[based on reference test at specified cut-off point]

A: 9/39 (23%)

B: 29/52 (56%)

C: 57/134 (42%)

D: 42/100 (42%)

E: 11/37 (28%)

F: 52/133 (39%)

G: 91/214 (42%)

H: 4/16 (25%)

I: 5/14 (36%)

J: 35/158 (22%)

K: 32/90 (35%)

For how many participants were no complete outcome data available?

N (%)

A: 17/39 (43.6)%

B: 0 (0%)

C: 43/180 (23.9%)

D: 7/107 (6.54%)

E: 0 (0%)

F: 0 (0%)

G: 17/231 (7.36%)

H: 1/16 (..)

I:1/15 (6.67%)

J: 10/168 (5.95%)

K: 0 (0%)

Reasons for incomplete outcome data described?

A: 4 participants with MCI objected to re-examination, 1 died of unrelated causes, and 5 were excluded because of definite other diagnoses. 7 control subjects objected to re-examination.

B: NA, because everyone had outcome data

C: initially identified 180 consecutive participants with MCI; 43/180 not included in the study:

32 refused lumbar puncture and 11 non-usable CSF samples; 3 participants died before 4 years of follow-up

(not included in the analysis)

D: 7 MCI participants who converted to other forms of dementia were excluded from the analysis.

E: NA, because everyone had outcome data

F: NA, because everyone had outcome data

G: some refused to participate or were

untraceable or died before follow-up

H: that participant was replaced by an additional, consecutively recruited patient from the memory clinic

I:reasons for loss-to-follow-up NR

J: CSF follow-up data were not available; the reason not given

K: NA, because everyone had outcome data

Endpoint of follow-up:

A: mean 2.6 ± 0.5 years (range 1.6 to 4 years)

B: mean 0.7 ± 0.43 years (range 0.17 to 2 years)

C: median 5.2 years (range 4.0 to 6.8 years)

D: median 1.5 years (IQR 1.08 – 2 years)

E: 0.5 years

F: mean 5.9 years (range 3.2 to 8.8 years)

G: mean 2.5 ± 1.0 years

H: mean 1.63 ± 0.75 years

I: 2 years

J: range 1 to 3 years for MCI

K: maximum: 4 years

mean 3.40 ± 1.01 years

Outcome measures and effect size (include 95%CI and p-value if available):

Outcome measure-1/2

[e.g. sensitivity / specificity (%)]

A: CSF t-tau: 56 [21, 86]% / 87 [69, 96]%

B: CSF t-tau: 90 [73, 98]% / 48 [27, 69]%

C-1: CSF t-tau: 51 [37, 64]% / 88 [79, 95]%

C-2: CSF p-tau: 68 [55, 80]% / 86 [76, 93]%

C-3: CSF p-tau/ABeta ratio: 96 [88, 100]% / 75 [64, 84]%

D: CSF t-tau: 83 [69, 93]% / 50 [37, 63]%

E-1: CSF t-tau: 73 [39, 94]% / 69 [48, 86]%

E-2: CSF p-tau: 82 [48, 98]% / 58 [37, 77]%

E-3: CSF p-tau/ABeta ratio: 82 [48, 98]% / 65 [44, 83]%

F-1: CSF t-tau: 81 [67, 90]% / 72 [60, 81]%

F-2: CSF p-tau: 67 [53, 80]% / 86 [77, 93]%

G: CSF t-tau: 71 [61, 80]% / 77 [69, 84]%

H: CSF p-tau: 100 [40, 100]% / 33 [10, 65]%

I-1: CSF p-tau: 40 [5, 85]% / 22 [3, 60]%

I-2: CSF p-tau/ABeta ratio: 80 [28, 99]% / 33 [7, 70]%

J-1: CSF p-tau: 89 [73, 97]% / 37 [29, 47]%

J-2: CSF p-tau/ABeta ratio: 80 [63, 92]% / 60 [51, 69]%

K: CSF p-tau/ABeta ratio: 81 [64, 93]% / 95 [86, 99]%

Pooled characteristics:

The authors did not meta-analyse pairs of sensitivity and specificity using the bivariate model, because the results were not clinically interpretable due to the mixed thresholds used by studies. Instead, the authors fitted an HSROC meta-analysis models to estimate summary ROC curves and derived estimates of sensitivity and likelihood ratios at a fixed value of specificity (chosen a priori as the median specificity for the studies that were analysed when fitting the model).

Index test-1 CSF t-tau for diagnosing ADD

77 [95% CI 67 to 85] (sens, with fixed value of spec of 72)

Index test-2 CSF p-tau for diagnosing ADD

81 [64 to 91.5] (sens, with fixed value of spec of 47.5)

Index test-3 CSF p-tau/Abeta ratio for diagnosing ADD no meta-analysis because the studies were few and small.

Study quality (ROB): method used and results per individual study

Methods used: QUADAS-2 tool

Patient selection domain

Studies B, D, and I scored high risk of bias because the participants were not consecutively or randomly enrolled or both the sampling procedure and exclusion criteria were not described. We stated that all included studies avoided a case-control design because we only considered data on performance of the index test to discriminate between people with MCI who converted to dementia and those who remained stable.

Studies H, E, F, K, J, and G scored unclear risk of bias, due to poor reporting on sampling procedure or exclusion criteria.

Index test domain

Studies H, C, E, F, and K scored high risk of bias, due to poor reporting because the threshold used was not prespecified and the optimal cutoff level was determined from ROC analyses; therefore, the accuracy of the CSF biomarkers reported in these

studies appeared to be overestimated.

Study A scored unclear risk of bias, due to poor reporting.

Reference standard domain

Studies A, H, B, D, I, E, and K scored unclear risk of bias, mainly because it was not reported whether clinicians conducting follow-up were aware of initial CSF biomarker analysis results. Some studies did not clearly report the reference standards used for diagnosing Alzheimer’s disease dementia. We were not able to obtain the information about how the reference standard was obtained and by whom, due to poor reporting.

Flow and timing domain

Studies A, H, I, E, K, J, and G scored unclear risk of bias, because not all participants were included in the analysis and/or the follow-up period was shorter than one year and/or reporting was poor.

Studies C and D scored high risk of bias because a large number of participants with non-Alzheimer's disease dementia were excluded from the analysis.

Place of the index test in the clinical pathway: add-on

Choice of cut-off point: The studies defined their own cut-off points. Thus no overall choice was made and this could be a source for heterogeneity.

Author’s conclusion. Given the insufficient evidence to evaluate the diagnostic value in MCI of CSF t-tau, CSF p-tau, CSF t-tau/ABeta ratio and CSF p-tau/ABeta ratio for Alzheimer's

disease dementia and other forms of dementias examined in this review, particular attention should be paid to the risk of misdiagnosis and overdiagnosis of dementia (and

therefore overtreatment) in clinical practice.

Sensitivity analyses: not performed, due to the limited number of studies.

Heterogeneity: Due to insufficient studies, meta-regression was not performed. So, sources of heterogeneity could not be examined.

Kokkinou, 2021

Study characteristics and results for individual studies are extracted from the SR Kokkinou, 2021

(unless stated otherwise)

SR

Literature search up to February 2020

A: Brettschneider 2006

B: Kapaki 2003

C: Knapskog 2018

D: Lewczuk 2004

E: Lombardi 2018

F: Maddalena 2003

G: Montine 2001

H: Perani 2016

I: Smach 2008

J: Rosler 2001

K: Spies 2010

L: Tapiola 2000

M: Tariciotti 2018

Study design: cross-sectional case-control (ADD vs non-ADD dementia)

Setting and Country: Participants were recruited from i) secondary care – outpatient clinic (n = 12); ii) secondary care – inpatients (n = 2); iii) community care (n = 2) and mixed setting (n = 1).

Source of funding and conflicts of interest: No known conflicts of interest. The National Institute for Health Research (NIHR), via Cochrane Infrastructure funding to the Cochrane Dementia and Cognitive Improvement group. The views and opinions expressed therein are those of the authors and do not necessarily reflect those of the Systematic Reviews Programme, NIHR, National Health Service or the Department of Health

Inclusion criteria SR: Cross-sectional studies that differentiated people with ADD from other dementia subtypes. Eligible studies required measurement of participant plasma or CSF ABeta42 levels and clinical assessment for dementia subtype. Delayed verification studies

Exclusion criteria SR: NR

13 studies included

Important patient characteristics:

Number of patients; characteristics important to the research question; for example, age, sex, bmi, ...

N, mean age

A: 109 AD patients, 71 (43-88) years for AD

B: 49 AD patients, 67.6 ± 9.3 years for AD

C: 205 patients, 84.8 ± 8.8 years for the total sample

D: 22 AD patients, 68 (62–77) years for AD

E: 32 AD patients, 66.5 ± 9.9 years for ADD;

F: 81 patients (n=51 AD), 70.1±8.7 years for AD

G: 19 AD patients, 65.3±8.7 years for AD

H: 47 AD patients, 66±6.8 years for AD

I: 73 AD patients, 73 (48–85) years for AD

J: 27 probable AD patients, mean age NR

K: 69 AD patients, 69±8 years for AD

L: 80 probable AD, 71±8 years for probable AD

M: 264 AD patients, 67.7 ± 10.4 years for ADD

….

Sex:

A: 39/109 AD (35.8% male)

B: 31/49 AD (63.3% male)

C: 53.7% male (total sample)

D: 6/22 AD (27.3% male)

E: 19/32 AD (59.4% male)

F: 54/100 (54% male)

G: NR

H: 26/47 AD (55.3% male)

I: 49/73 (67.1% male)

J: 9/27 probable AD (33.3% male)

K: 34/69 AD (49.3% male)

L: 34/80 probable AD (42.5% male)

M: 106/264 AD (40.2% male)

Describe index and comparator tests* and

cut-off point(s):

A: ABeta42; 612 pg/ml

B: ABeta42; 435 pg/ml

C: ABeta42; 550 pg/ml

D: ABeta42; 500 pg/ml

E: ABeta42; 650 pg/ml

F: ABeta42; 490 pg/ml

G: ABeta42; 1125 pg/ml

H: ABeta42; 500 pg/ml

I: ABeta42; 505 pg/ml

J: ABeta42; 375 pg/ml

K: NR

L: ABeta42; 340 pg/ml

M: ABeta42; 500 pg/ml

[use the format ‘character+number’ if 2 or more index tests are being compared, e.g. A-1, A-2, etc]

Describe reference test and cut-off point(s):

A: NINCDS-ADRDA criteria

B: NINCDS-ADRDA criteria

C: no diagnostic criteria specified; by consensus between two experienced physicians.

D: NINCDS-ADRDA criteria

E: NIA-AA criteria

F: NINCDS-ADRDA

G: NINCDS-ADRDA

H: NINCDS-ADRDA

I: NINCDS-ADRDA

J: NINCDS-ADRDA

K: NINCDS-ADRDA

L: NINCDS-ADRDA

M: NINCDS-ADRDA

Prevalence (%)

[based on reference test at specified cut-off point]

A: 109/165 (66.1%)

B: 49/70 (70%)

C: 138/205 (67.3%)

D: 22/33 (66.7%)

E: 32/45 (71.1%)

F: 51/81 (63%)

G: 19/27 (70.3%)

H: 47/75 (62.7%)

I: 73/108 (67.6%)

J: 27/51 (52.9%)

K: 69/138 (50%)

L: 80/107 (74.8%)

M: 264/1137 (23.2%)

For how many participants were no complete outcome data available?

N (%)

A: NR

B: NR

C: there is missing data, but for how many patients is NR

D: NR

E: 0 (0%)

F: NR

G: NR

H: NR

I: 20/73 (27.4%)

J: NR

K: NR

L: NR

M: 121 (10.7%).

Reasons for incomplete outcome data described?

A: NR

B: NR

C: reasons of missing data NR

D: NR

E: NA, because everyone had outcome data

F: NR

G: NR

H: NR

I: no adequate CSF samples obtained

J: NR

K: NR

L: NR
M: incomplete or uncertain diagnosis

Endpoint of follow-up:

This is NA for all individual studies, because only cross-sectional studies were included in this SR.

Outcome measures and effect size (include 95%CI and p-value if available):

Outcome measure-1/2

[e.g. sensitivity / specificity (%)]

A: 82 [73, 88]% / 46 [33, 60]%

B: 71 [57, 83]% / 80 [52, 96]%

C: 42 [30, 56]% / 33 [10, 65)%

D: 86 [65, 97]% / 82 [48, 98]%

E: 88 [71, 96]% / 67 [9, 99]%

F: 78 [65, 89]% / 70 [51, 85]%

G: 100 [82, 100]% / 25 [3, 65]%

H: 85 [72, 94]% / 46 [28, 66]%

I: 82 [71, 90]% / 71 [54, 85]%

J: 78 [58, 91]% / 58 [37, 78]%

K: 83 [72, 91]% / 74 [62, 84]%

L 69 [57, 79]% / 59 [39, 78]%

M 81 [76, 86]% / 54 [49, 58]%

Pooled characteristic

Bivariate analysis, CSF ABeta42 [pooled cut-off at all thresholds]:

Pooled sensitivity 79 [73, 85]%

Bivariate analysis, CSF ABeta42 [pooled cut-off at all thresholds]:

Pooled specificity 60 [52, 67]%

Study quality (ROB): method used and results per individual study

Methods used: QUADAS-2 tool

Patient selection domain

Studies A, B, D, E, F, G, I, K, L, M scored high risk of bias because of selective inclusion or enriching the population with a certain dementia type.

Studies C and H scored unclear risk of bias, because of poor reporting.

Index test domain

Studies A, B, D, F, I, K, and L scored high risk of bias, because the ABeta42 threshold used was not pre-specified.

Studies C, E, G, and M scored unclear risk of bias, because they did not report whether investigators interpreted the ABeta42 data without knowledge of the dementia classification.

Reference Standard domain

Study E scored high risk of bias, because investigators made the dementia assessment with the knowledge of the ABeta42 result.

Study M scored unclear risk of bias, because they did not report whether the investigator, who interpreted the results of reference standard, conducted the assessment without the knowledge of the ABeta42 data.

Flow and timing domain

Studies I and M scored high risk of bias because the final clinical diagnosis was established (reassessed) 12 months or longer after CSF sampling.

Studies C, E, H, and K scored unclear risk of bias because not all patients were included in the analysis and/or studies did not report the interval between index test and reference standard.

Place of the index test in the clinical pathway: add-on

Choice of cut-off point: The studies defined their own cut-off points. Thus no overall choice was made and this could be a source for heterogeneity.

Sensitivity analyses:

·Effect of age: removal of studies C, G and J did not substantially alter pooled estimates of sensitivity or specificity.

·Effect of studies without a pre-specified test threshold: removal of eight studies did not alter pooled estimates of sensitivity or specificity.

Heterogeneity: Sources of heterogeneity were: 1) population and dementia subtype enrolled; 2) test threshold used, and 3) the diagnostic criteria and definition of ADD and dementia subtypes.

Struyfs 2015

Type of study[1]: postmortem case-control in biobank samples.

Setting and country: Biobank, België

Funding and conflicts of interest: Several sources of funding reported, all Belgium government and European research funding grants.

The authors declare no conflicts of interest.

Inclusion criteria: Pathologically confirmed diagnoses of AD and non-AD, CSF samples available in biobank.

Exclusion criteria: None described

N=140 AD, 77 non-AD.

Below characteristics as reported by the authors in table 1.

All data are median values with 25th and 75th quartiles between brackets, except for N. To compare gender distribution between the groups, a Chi-square test was performed, while Mann–Whitney U tests were used to compare clinical and biomarker data between the groups.

AD, Alzheimer’s disease; non-AD, dementia not due to Alzheimer’s disease; MMSE, mini-mental state examination; Aβ_1–42, amyloid-β peptide of 42 amino acids; T-tau, total tau-protein; P-tau_181P, tau phosphorylated at threonine 181.

Bold values show statistically significant P-values (P < 0.05).

Describe index test: Lumbar puncture (LP) to collect CSF between the intervertebral space L3/L4 or L4/L5.

Concentrations of Aβ1–42, T-tau, and P-tau181P were determined with commercially available single analyte ELISA- kits (respectively, INNOTEST® β-AMYLOID(1–42), INNOTEST® hTAU-Ag, and INNOTEST® PHOSPHO-TAU(181P); Fujirebio, Ghent, Belgium).

Cut-off point(s):

No predefined cut-offs were specified. By maximizing the Youden index from the ROC curve, an optimal cut-off point was derived.

These were:

Aβ1–42= 500,27

T-tau = 472,35

P-tau181P = 50,35

Aβ1–42/ T-tau = 1,08

Aβ1–42/ P-tau181P = 9,11

Describe reference test[2]:

Pathological diagnoses according to standard neuropathological criteria.

Cut-off point(s): For ADD, vascular dementia and dementia with Lewy bodies the Montine criteria were applied, frontotemporal dementia according to Cairns or Mackenzie and Creutzfeld Jacob disease according to Markesbery. The ADD group also contained mixed dementia diagnoses where patients fulfilled the criteria for ADD in combination with minor pathology for other diseases such as cerebrovascular or Parkinson’s disease.

Time between the index test en reference test:

Unclear, years between sampling and death is reported but CSF test were performed later.

Years between sampling and death were:

AD = median 0 year (IQR 0-2,5 years)

Non-AD = median 0 year (IQR 0-2 years).

For how many participants were no complete outcome data available?

There was complete data for all included patients.

Outcome measures and effect size (include 95%CI and p-value if available)⁴:

Amyloidß42: sensitivity 79,3% and specificity 53,2% with an area under the curve (AUC) of 0,677 (95%CI 0,597-0,757) and an amyloidß42 cut-off of 500,27.

T-tau: sensitivity 62,1% and specificity 63,6% with an area under the curve (AUC) of 0,592 (95%CI 0,508-0,675) and a t-tau cut-off of 472,35.

P-tau: sensitivity 77,9% and specificity 61,0% with an area under the curve (AUC) of 0,720 (95%CI 0,648-0,792) and a p-tau cut-off of 50,35.

Amyloidß42/ t-tau: sensitivity 75,0% and specificity 57,1% with an area under the curve (AUC) of 0,678 (95%CI 0,601-0,755) and a amyloidß42/ t-tau ratio cut-off of 1,08.

Amyloidß42/ t-tau: sensitivity 82,9% and specificity 59,7% with an area under the curve (AUC) of 0,770 (95%CI 0,703-0,837) and a amyloidß42/ p-tau ratio cut-off of 9,11.

None

Ritchie, 2014

Yes

Yes

Yes

Yes

Yes

Yes

No

No

Ritchie, 2017

Yes

Yes

Yes

Yes

Yes

Yes

No

No

Kokkinou, 2021

Yes

Yes

Yes

Yes

Yes

Yes

No

No, only reported for systematic review.

Struyfs 2015

Was a consecutive or random sample of patients enrolled?

Unclear.

Was a case-control design avoided?

No

Did the study avoid inappropriate exclusions?

Yes

Were the index test results interpreted without knowledge of the results of the reference standard?

Unclear

If a threshold was used, was it pre-specified?

No

Is the reference standard likely to correctly classify the target condition?

Yes

Were the reference standard results interpreted without knowledge of the results of the index test?

Yes, although the pathologist had access to clinical files and neuroimaging data.

Was there an appropriate interval between index test(s) and reference standard?

Unclear, not reported. Although years between sampling and death was 0 years in both the AD and non-AD group, the interquartile ranges were 0-2, 5 years for AD and 0-2 years for non-AD.

Did all patients receive a reference standard?

Yes

Did patients receive the same reference standard?

Yes

Were all patients included in the analysis?

Yes

Are there concerns that the included patients do not match the review question?

No

Are there concerns that the index test, its conduct, or interpretation differ from the review question?

No

Are there concerns that the target condition as defined by the reference standard does not match the review question?

No

CONCLUSION:

Could the selection of patients have introduced bias?

RISK: UNCLEAR

CONCLUSION:

Could the conduct or interpretation of the index test have introduced bias?

RISK: HIGH

CONCLUSION:

Could the reference standard, its conduct, or its interpretation have introduced bias?

RISK: LOW

CONCLUSION

Could the patient flow have introduced bias?

RISK: UNCLEAR

Reference

Reason for exclusion

Hazan J, Wing M, Liu KY, Reeves S, Howard R. Clinical utility of cerebrospinal fluid biomarkers in the evaluation of cognitive impairment: a systematic review and meta-analysis. J Neurol Neurosurg Psychiatry. 2023 Feb;94(2):113-120. doi: 10.1136/jnnp-2022-329530. Epub 2022 Sep 12. PMID: 36096664.

Excluded diagnostic accuracy: unclear reference, wrong population (not only suspected AD, but mixed with depression/parkinson and other diagnosis) and wrong outcome (change in diagnosis)

Ma Y, Brettschneider J, Collingwood JF. A Systematic Review and Meta-Analysis of Cerebrospinal Fluid Amyloid and Tau Levels Identifies Mild Cognitive Impairment Patients Progressing to Alzheimer's Disease. Biomedicines. 2022 Jul 15;10(7):1713. doi: 10.3390/biomedicines10071713. PMID: 35885018; PMCID: PMC9313367.

wrong outcome (standardized mean differences in metabolite levels)

Fink HA, Linskens EJ, Silverman PC, McCarten JR, Hemmy LS, Ouellette JM, Greer NL, Wilt TJ, Butler M. Accuracy of Biomarker Testing for Neuropathologically Defined Alzheimer Disease in Older Adults With Dementia. Ann Intern Med. 2020 May 19;172(10):669-677. doi: 10.7326/M19-3888. Epub 2020 Apr 28. PMID: 32340038.

wrong outcome (reported only median values for all studies, not per study results); wrong design (does not report study characteristics in enough detail to determine if studies are conform PICO );

Rivero-Santana A, Ferreira D, Perestelo-Pérez L, Westman E, Wahlund LO, Sarría A, Serrano-Aguilar P. Cerebrospinal Fluid Biomarkers for the Differential Diagnosis between Alzheimer's Disease and Frontotemporal Lobar Degeneration: Systematic Review, HSROC Analysis, and Confounding Factors. J Alzheimers Dis. 2017;55(2):625-644. doi: 10.3233/JAD-160366. PMID: 27716663.

Wrong design (set up to discriminate AD from FTD and reports diagnostic accuracy for this comparison; unable to extract details for individuel studies, only reported point estimates for sensitivity and specificity)

Olsson B, Lautner R, Andreasson U, Öhrfelt A, Portelius E, Bjerke M, Hölttä M, Rosén C, Olsson C, Strobel G, Wu E, Dakin K, Petzold M, Blennow K, Zetterberg H. CSF and blood biomarkers for the diagnosis of Alzheimer's disease: a systematic review and meta-analysis. Lancet Neurol. 2016 Jun;15(7):673-684. doi: 10.1016/S1474-4422(16)00070-3. Epub 2016 Apr 8. PMID: 27068280.

Wrong outcome (reported ratio between Alzheimer’s disease and control cohorts or mild cognitive impairment due to Alzheimer’s disease and stable mild cognitive impairment cohorts with 95% CIs, not diagnostic accuracy measures)

Mo JA, Lim JH, Sul AR, Lee M, Youn YC, Kim HJ. Cerebrospinal fluid β-amyloid1-42 levels in the differential diagnosis of Alzheimer's disease--systematic review and meta-analysis. PLoS One. 2015 Feb 24;10(2):e0116802. doi: 10.1371/journal.pone.0116802. PMID: 25710596; PMCID: PMC4339391.

Unclear reporting (unable to extract details for bias evaluation and assessment diagnostic accuracy)

Dumurgier J, Schraen S, Gabelle A, Vercruysse O, Bombois S, Laplanche JL, Peoc'h K, Sablonnière B, Kastanenka KV, Delaby C, Pasquier F, Touchon J, Hugon J, Paquet C, Lehmann S. Cerebrospinal fluid amyloid-β 42/40 ratio in clinical setting of memory centers: a multicentric study. Alzheimers Res Ther. 2015 Jun 1;7(1):30. doi: 10.1186/s13195-015-0114-5. PMID: 26034513; PMCID: PMC4450486.

Wrong setting (longitudinal studies for conversion MCI to AD or other forms, not differential diagnosis)

Ewers M, Mattsson N, Minthon L, Molinuevo JL, Antonell A, Popp J, Jessen F, Herukka SK, Soininen H, Maetzler W, Leyhe T, Bürger K, Taniguchi M, Urakami K, Lista S, Dubois B, Blennow K, Hampel H. CSF biomarkers for the differential diagnosis of Alzheimer's disease: A large-scale international multicenter study. Alzheimers Dement. 2015 Nov;11(11):1306-15. doi: 10.1016/j.jalz.2014.12.006. Epub 2015 Mar 21. PMID: 25804998.

Wrong design (set up to distinguish AD from other dementia types, not about diagnostic accuracy for diagnosing AD)

Rosa MI, Perucchi J, Medeiros LR, Fernandes B, Fernandes Dos Reis ME, Silva BR. Accuracy of cerebrospinal fluid Aβ(1-42) for Alzheimer's disease diagnosis: a systematic review and meta-analysis. J Alzheimers Dis. 2014;40(2):443-54. doi: 10.3233/JAD-132264. PMID: 24448789.

wrong design (AD vs control)

Tang W, Huang Q, Wang Y, Wang ZY, Yao YY. Assessment of CSF Aβ42 as an aid to discriminating Alzheimer's disease from other dementias and mild cognitive impairment: a meta-analysis of 50 studies. J Neurol Sci. 2014 Oct 15;345(1-2):26-36. doi: 10.1016/j.jns.2014.07.015. Epub 2014 Jul 15. PMID: 25086857.

wrong outcome (standardized mean differences in metabolite levels)

Tang W, Huang Q, Yao YY, Wang Y, Wu YL, Wang ZY. Does CSF p-tau181 help to discriminate Alzheimer's disease from other dementias and mild cognitive impairment? A meta-analysis of the literature. J Neural Transm (Vienna). 2014 Dec;121(12):1541-53. doi: 10.1007/s00702-014-1226-y. Epub 2014 May 10. PMID: 24817210.

wrong outcome (standardized mean differences in metabolite levels)

Mattsson N, Rosén E, Hansson O, Andreasen N, Parnetti L, Jonsson M, Herukka SK, van der Flier WM, Blankenstein MA, Ewers M, Rich K, Kaiser E, Verbeek MM, Olde Rikkert M, Tsolaki M, Mulugeta E, Aarsland D, Visser PJ, Schröder J, Marcusson J, de Leon M, Hampel H, Scheltens P, Wallin A, Eriksdotter-Jönhagen M, Minthon L, Winblad B, Blennow K, Zetterberg H. Age and diagnostic performance of Alzheimer disease CSF biomarkers. Neurology. 2012 Feb 14;78(7):468-76. doi: 10.1212/WNL.0b013e3182477eed. Epub 2012 Feb 1. PMID: 22302554; PMCID: PMC3280049.

Wrong design (set up to compare age effects on diagnostic test accuracy)