Cerebrospinal Fluid Biomarkers for the Differential Diagnosis between Alzheimer’s Disease and Frontotemporal Lobar Degeneration: Systematic Review,HSROC Analysis,and Confounding Factors

Abstract

Background: Differential diagnosis in dementia is at present one of the main challenges both in clinical practice and research. Cerebrospinal fluid (CSF) biomarkers are included in the current diagnostic criteria of Alzheimer’s disease (AD) but their clinical utility is still unclear.

Objective: We performed a systematic review of studies analyzing the diagnostic performance of CSF Aβ₄₂, total tau (t-tau), and phosphorylated tau (p-tau) in the discrimination between AD and frontotemporal lobar degeneration (FTLD) dementias.

Methods: The following electronic databases were consulted until May 2016: Medline and PreMedline, EMBASE, PsycInfo, CINAHL, Cochrane Library, and CRD. For the first-time in the field, a Hierarchical Summary Receiver Operating Characteristic (HRSOC) model was applied, which avoids methodological problems of meta-analyses based on summary points of sensitivity and specificity values. We also investigated relevant confounders of CSF biomarkers’ diagnostic performance such as age, disease duration, and global cognitive impairment.

Results: The p-tau/Aβ₄₂ ratio showed the best diagnostic performance. No statistically significant effects of the confounders were observed. Nonetheless, the p-tau/Aβ₄₂ ratio may be especially indicated for younger patients. P-tau may be preferable for less cognitively impaired patients (high MMSE scores) and the t-tau/Aβ₄₂ ratio for more cognitively impaired patients (low MMSE scores).

Conclusion: The p-tau/Aβ₄₂ ratio has potential for being implemented in the clinical routine for the differential diagnosis between AD and FTLD. It is of utmost importance that future studies report information on confounders such as age, disease duration, and cognitive impairment, which should also stimulate understanding of the role of these factors in disease mechanisms and pathophysiology.

Keywords

Age Alzheimer’s disease cerebrospinal fluid markers confounding factor disease duration frontotemporal lobar degeneration HSROC analysis Mini-Mental State Examination systematic review

INTRODUCTION

Differential diagnosis in dementia is at present one of the main challenges in clinical practice and research due to heterogeneity and overlap between neurodegenerative diseases. Alzheimer’s disease (AD) is the most frequent neurodegenerative disease and the most common cause of dementia [1, 2]. AD is characterized by extracellular deposits of amyloid-β (Aβ) peptides in senile plaques and intracellular aggregates of tau protein in neurofibrillary tangles [3, 4]. The typical clinical presentation of AD includes impairment in episodic memory with cognitive decline involving other functions as well as changes in mood and behavior as the disease progresses [3]. With the emergence of new treatments aimed to modify the course of the disease, an accurate differential diagnosis of AD becomes crucial.

Frontotemporal lobar degeneration (FTLD) is the most common cause of dementia under the age of 60 [5]. FTLD is the umbrella term encompassing a heterogeneous group of well-defined clinical syndromes, including (1) three frontotemporal dementia (FTD) variants: behavioral variant frontotemporal dementia (bvFTD), progressive non-fluent aphasia, and semantic dementia; (2) progressive supranuclear palsy syndrome; and (3) corticobasal syndrome [6, 7]. The Lund-Manchester groups have actively guided the diagnostic criteria since 1994 [8 –10]. These criteria have been widely used although alternative criteria also exist [11], as well as recent refinements [12 –15]. Despite well-articulated clinical diagnostic criteria for FTLD, differential diagnosis between FTLD and AD is sometimes difficult. Pathological studies show incorrect FTLD diagnosis in up to 28% of the cases [16, 17]. This occurs especially in the early phase of both diseases as well as in atypical forms (e.g., AD with significant frontal involvement, or FTLD with significant memory deficit) [18, 19].

AD pathophysiology can now be studied in vivo thanks to certain biomarkers in the cerebrospinal fluid (CSF). Currently validated core CSF biomarkers of AD are Aβ₄₂, total tau (t-tau), and phosphorylated tau (p-tau) proteins [20]. A large body of evidence shows that AD patients present decreased levels of Aβ₄₂ and increased levels of t-tau and p-tau compared to healthy controls [21 –23]. This has led to their inclusion in the revised diagnostic criteria for AD as supportive feature of the clinical diagnosis [20]. However, findings on the potential of the CSF biomarkers to discriminate AD from other dementias are less conclusive. Although recent studies comparing AD and FTLD suggest increased t-tau and p-tau concentrations and reduced Aβ₄₂ levels in AD, results show substantial heterogeneity across studies and inconsistent findings [24 –26].

To our knowledge, three meta-analysis has previously evaluated the role of CSF tau and Aβ₄₂ in the differential diagnosis between AD and FTLD. Tang et al. [25, 26] analyzed mean differences in the concentrations of p-tau and Aβ₄₂, but not their diagnostic performance. Van Harten et al. [24] only analyzed t-tau and p-tau, obtaining 74% sensitivity and 74% specificity for t-tau, and 79% sensitivity and 83% specificity for p-tau. However, these results were obtained by calculating summary points from separate meta-analyses of sensitivity and specificity, a statistical approach that does not account for the covariation between them, or for the variability in the diagnostic cut-off values across studies [27, 28]. Furthermore, new articles have been published since then. The usefulness of CSF biomarkers for the differential diagnosis between AD and FTLD is still controversial and a topic of great interest at present. Therefore, the objective of this study was to perform an updated systematic review on the current evidence on the diagnostic performance of CSF biomarkers t-tau, p-tau, and Aβ₄₂ (as well as their ratios and combinations), in the discrimination between AD and FTLD patients. We used a more theoretically motivated statistical approach, the Hierarchical Summary Receiver Operating Characteristic (HSROC) model [29], that is deemed more appropriate when considerable variation is present in cut-offs for the discrimination between diagnostic groups. We also aimed to explore relevant confounding factors of CSF biomarkers’ diagnostic performance.

MATERIAL AND METHODS

Data source and study selection

As part of a wider search including systematic reviews and primary studies on the diagnosticperformance of CSF t-tau, p-tau and Aβ₄₂ in AD, the following electronic databases were consulted until September 2013: Medline and PreMedline, EMBASE, PsycInfo, CINAHL, Cochrane Library and Centre for Reviews & Dissemination (see Supplementary Table 1). This search has led to the publication of three previous studies: meta-review of CSF biomarkers in AD [30], meta-analysis of CSF biomarkers in mild cognitive impairment (MCI) [31], and cost-effectiveness of CSF biomarkers for AD [32]. The search was updated in the same databases until May 2016, but only for primary studies. For the aim of the present work, studies were included if: 1) recruited patients with AD and FTLD diagnosed according to well-established criteria (e.g., McKhan et al. [33], Neary et al. [8, 12], etc.); 2) evaluated CSF t-tau, p-tau, or Aβ₄₂; and 3) reported sensitivity and specificity values, or data which enabled their calculation. We excluded studies with less than five patients in the AD or FTLD groups. Two researchers (AR, LP 1 ) carried out the first selection of references by title and abstract, and another two (AR, DF2 2 ) reviewed the articles and made the final selection. Reference lists of the included studies were also reviewed to detect articles not included in the electronic search.

Data on the biomarkers evaluated, their diagnostic performance, and a pre-specified list of confounders were extracted if available. The list of confounders was initially based on a list used elsewhere [31], adjusted to the purpose of the current study. Subsequently, new factors of interest where added as a result of our systematic review and scrutiny of the selected articles. This pre-specified list included age, years of education, disease duration, global cognitive impairment (i.e., Mini-Mental State Examination (MMSE) score), neuropathological diagnosis, clinical diagnostic criteria used, FTLD variant, type of technique used to measure CSF biomarker levels, and cut-offs values for discriminating between diagnostic groups. Special attention was paid to detect overlapped samples between articles reporting diagnostic data on the same biomarkers. When this was the case, the larger sample was included.

Data collection, risk of bias, and quality assessment

Several strategies were followed in order to reduce the risk of bias related to publication, data availability, and reviewer selection (see Supplementary Table 2). Methodological quality of the included studies was independently assessed by two researchers (AR, DF) with the QUADAS-2 scale [34]. Moreover, this study was performed in accordance with the PRISMA statement [35, 36], which provides a detailed guideline of preferred reporting style for systematic reviews and meta-analyses.

Statistical analysis

HSROC models [29, 37] were preferred instead of bivariate summary points for sensitivity and specificity because of the great variability reported in the literature in cut-off values for the discrimination between diagnostic groups. The metandi package [38] of STATA 12.0 software was used. HSROC is a two-level model that uses the sensitivity and specificity values of each study to construct a summary ROC curve, accounting for the covariation between sensitivity and specificity. Five parameters are estimated in the model: diagnostic accuracy (Lambda ∧, the natural logarithm of the diagnostic odds ratio, DOR), the positivity threshold (Theta Θ, the mean of the log of sensitivity and the log of 1-specificity), their two variances (^б2_Λ and ^б2_Θ), and the parameter that defines the shape of the summary curve (Beta β). Accuracy and threshold are considered to randomly vary between studies, whereas Beta represents a fixed effect. A Beta value significantly different from zero indicates that the curve is asymmetric. This means that accuracy significantly varies with threshold, and therefore it is not appropriate to calculate a pooled value for accuracy. A Beta value equal to zero indicates that the curve is symmetric, so that Lambda appropriately represents the overall accuracy of the test.

HSROC curves were constructed for each biomarker, t-tau/Aβ₄₂ and p-tau/Aβ₄₂ ratios, and other combinations of biomarkers when more than three studies were available. Instead of calculating summary points for sensitivity and specificity, the expected sensitivity was calculated from each model at the observed median value of specificity. Optimal diagnostic performance was considered when sensitivity and specificity values were≥80%, following the international consensus [39]. From these sensitivity and specificity values, positive and negative likelihood ratios (LR+ and LR–) were calculated and interpreted following established guidelines [40] (see footnotes in Table 3). Exploratory subgroup analyses were also performed in two steps by applying HSROC models to subgroups of studies defined by the different pre-specified confounders. First, Beta values were compared between subgroups using t-test in order to assess whether the underlying curves had the same shape and were thus comparable (relationship between sensitivity and specificity is comparable between subgroups). Second, if Beta values were statistically comparable (no significant differences), Lambda values were then compared between subgroups using t-test in order to evaluate differences in diagnostic performance.

RESULTS

The selection flow is shown in Fig. 1. Key characteristics and sensitivity-specificity results of the 30 selected studies [41 –70] are displayed in Table 1, and methodological quality in Table 2. Eleven studies specified the FTLD variant included, but only one reported biomarkers’ diagnostic performance separately for each variant [51], and other two included only bvFTD patients [45, 50]. Subgroup analyses could only be performed for age, disease duration, global cognitive impairment (i.e., MMSE), postmortem diagnostic confirmation, and cut-off value (internal versus external). For continuous variables, subgroups were defined by the median value of the specific confounder (see Table 3). For FTLD diagnosis, a sensitivity analysis was performed excluding studies using other criteria than Neary et al. [8]. Table 3 shows mean values for the five parameters of the HSROC models and their 95% confidence intervals (CI).

Total tau (t-tau)

Twenty two studies were included [41 , 59–68]. The study by Bian et al. [42] was excluded from the analysis due to sample overlap with Grossman et al. [53]. Figure 2A and Table 3 show the HSROC curve and the estimated parameters. Accuracy (Lambda) was 2.14 (95% CI: 1.90–2.39), which represents a DOR of 8.49. At a median specificity value of 75%, the model yielded a sensitivity of 74% (95% CI: 68% – 79%). This represents a LR+ of 2.96 and a LR–of 0.35. Pijnenburg et al. [61] resulted as an outlier due to low sensitivity. Excluding this study from the analysis reduced the variance of Lambda and slightly increased the estimated accuracy (data not shown).

No significant differences were observed between subgroups in terms of age, disease duration, MMSE, or cut-offs values. The three studies that analyzed complete [46, 48] or almost complete [64] postmortem samples obtained sensitivity values of 63%, 72%, and 73%, and specificity values of 75%, 69%, and 94%, respectively.

Phosphorylated tau (p-tau)

Sixteen studies were included [41 , 66–68]. In order to maximize the number of studies, Buerger et al. [47] was included in the analysis despite 22 AD and 6 FTLD patients (26% of the sample, n = 106) were also included in Hampel et al. (n = 132) [54]. Most of the studies analyzed the p-tau₁₈₁ epitope but three studies reported p-tau₂₃₁ [45 , 54]. The only study that investigated both p-tau₁₈₁ and p-tau₂₃₁ (but reported data only for the last) did not find significant differences in diagnostic accuracy (p = 0.39) [54]. Accuracy (Lambda) was 2.89 (95% CI: 2.47–3.31), representing a pooled DOR of 18.0 (Fig. 2B and Table 3). At a median specificity value of 78%, the model yielded a sensitivity value of 84% (95% CI: 0.77–0.89), which represents a LR+ of 3.82 and a LR–of 0.21.

No significant differences were observed between subgroups in terms of age, disease duration, MMSE, or cut-off values. Nonetheless, p-tau showed 84% sensitivity and 79% specificity in studies including patients with greater global cognitive impairment (lower MMSE scores), with a LR–of 0.20. This LR–value corresponds with moderate increase in probability that AD is not present. Two studies with postmortem diagnosed patients showed 91% and 90% sensitivity, and 79% and 60% specificity, respectively [58, 64].

Amyloid-beta 1-42 (Aβ₄₂)

Sixteen studies were included [41 , 68]. Accuracy (Lambda) was 2.53 (95% CI: 1.87–3.19), representing a pooled DOR of 12.6. At a median specificity value of 70%, the HSROC model yielded a sensitivity value of 82% (95% CI: 72–88%) (Fig. 2C and Table 3). This represents a LR+ of 2.73 and a LR- of 0.26.

No significant differences were observed between subgroups in terms of age, disease duration or MMSE. Brunnström et al. [46] analyzed a small sample with postmortem diagnosis providing 50% sensitivity and 83% specificity. Seeburger et al. [64] analyzed a sample where 98% of the patients had postmortem diagnostic confirmation. Their results achieved 84% sensitivity and 100% specificity.

T-tau/Aβ₄₂ and p-tau/Aβ₄₂ ratios

Ten [41 , 68] and nine studies [41 , 68] reported data on the t-tau/Aβ₄₂ and p-tau/Aβ₄₂ ratios, respectively. HSROC results were similar for both ratios. For t-tau/Aβ₄₂, an expected sensitivity of 89% was obtained at a fixed median specificity value of 79%. LR+ was 4.24 and LR–was 0.14, corresponding to moderate increase in probability that AD is not present.

For p-tau/Aβ₄₂, sensitivity was 87% at a fixed median specificity value of 80%. LR+ was 4.35 and LR–was 0.16, also corresponding to moderate increase in probability that AD is not present.

No significant differences were observed between subgroups in terms of age, MMSE, or disease duration. Nonetheless, p-tau/Aβ₄₂ showed acceptable sensitivity and specificity values (83% and 80%, respectively) in studies including younger patients.T-tau/Aβ₄₂ showed 89% sensitivity and 79% specificity in studies including patients with less global cognitive impairment, with a LR–of 0.14. This LR–value corresponds with moderate increase in probability that AD is not present. A meta-analysis including the only four studies with postmortem diagnostic confirmation [42 , 64] yielded an expected sensitivity of 88% at a fixed median specificity value of 95%. LR+ was 17.6 and LR–was 0.13, corresponding to conclusive and moderate increase in the pretest probability, respectively.

Other combinations of CSF biomarkers

Few studies reported results for other sorts of combination (see Table 1). Herbert et al. [55] combined the three biomarkers using logistic regression and achieved 77% sensitivity and 60% specificity. Four studies used logistic regression or a discriminant line for the combination between p-tau and Aβ₄₂ [49 , 70]. Sensitivity values ranged from 85% to 100% and specificity values from 75% to 88%. The combination of t-tau and Aβ₄₂ also yielded good results in Toledo et al. [69] (sample with postmortem diagnostic confirmation, 90% sensitivity, 82% specificity) and Riemenschneider et al. [62] (85% sensitivity, 85% specificity). Baldeiras et al. [41] obtained lower values (88% sensitivity and 77% specificity). Finally, de Souza et al. [51] obtained sensitivity and specificity values of 92% and 91% respectively for the presence of abnormal values in both ratios (t-tau/Aβ₄₂ and p-tau/Aβ₄₂).

DISCUSSION

In this study we provide an updated and comprehensive meta-analysis on the diagnostic performance of core CSF biomarkers of AD for the differential diagnosis between AD and FTLD. Our results on t-tau and p-tau are almost identical to those obtained in a previous meta-analysis [24]. Results replication is enhanced by the fact that we followed a more appropriate statistical approach (HSROC model) and included several studies that have been recently published. Moreover, we extended this previous study by analyzing Aβ₄₂, a relevant CSF biomarker in AD. By doing so, our findings revealed that only by combining p-tau and Aβ₄₂ results achieved clinical usefulness. Another contribution is that, since the performance of CSF biomarkers’ is influenced by different confounding factors such as age [31], several subgroup meta-analyses were conducted in order to identify for which specific kind of patients these CSF biomarkers might be useful in clinicalpractice.

When using the CSF biomarkers separately, none of them achieved optimal diagnostic performance, as defined by sensitivity and specificity values ≥80% [39]. P-tau showed the best performance reaching 84% sensitivity and 78% specificity. The fact that p-tau arose as the biomarker with best diagnostic performance could be explained by the fact that p-tau is not only a marker of axonal damage and neuronal degeneration, as t-tau, but it is more closely related to AD pathophysiology and the formation of neurofibrillary tangles [71, 72]. Moreover, CSF concentrations of p-tau in FTLD are more comparable to concentrations in controls than to concentrations in AD patients [24, 26]. Combinations were needed to achieve optimal diagnostic performance. The p-tau/Aβ₄₂ ratio showed optimal results, with 87% sensitivity and 80% specificity. Negative likelihood ratios indicated that normal values in the p-tau/Aβ₄₂ ratio may help to rule-out AD. This better performance of the ratio in comparison with the CSF biomarkers alone is likely explained by the fact that two aspects of the AD pathology, i.e., plaques (Aβ₄₂), and neurodegeneration (tau), are captured by the ratio.

A novelty in this study is the evaluation of different confounders of CSF biomarkers’ diagnostic performance. To our knowledge, no previous meta-analyses have performed such kind of analyses in AD versus FTLD, despite the well-known influence of external factors on biomarkers’ performance [31]. We a priori defined a comprehensive list of confounders. However, due to the limited data available, subgroup analyses could only be performed for age, disease duration, global cognitive impairment, and cut-off values. Below, we provide several recommendations to overcome this issue in future studies. Although we did not find any evidence of a statistically significant effect of these confounders on biomarkers’ accuracy, the likelihood ratios indicated moderate increase in probability and several interesting trends were observed. Regarding age, the p-tau/Aβ₄₂ ratio showed optimal diagnostic performance in studies including younger samples (mean age <70.3 years). Sensitivity was very high (91%) in studies including older samples (mean age ≥70.3 years), however at the cost of low specificity (71%). Sensitivity values are usually higher in older populations due to the fact that age-related brain damage is added to the disease-related brain damage [73, 74]. We previously showed higher sensitivity values of the CSF biomarkers in MCI patients older than 70 years compared to MCI patients younger than 70 years [31]. Mattsson and co-workers also found increased sensitivity in MCI patients older than 65 years compared to MCI patients younger than 65 years [75]. Greater AD pathology in older FTLD patients may also be an explanation of worse specificity in comparison with sensitivity, given the high percentage of misdiagnosis in 28% of FTLD patients [16, 17].

Regarding global cognitive impairment, normal CSF levels of p-tau may help to rule-out AD in patients with lower MMSE score (mean <19.3). The opposite was observed for the t-tau/Aβ₄₂ ratio, where normal values of the ratio may help to rule-out AD in patients with higher MMSE score (mean >19.2). To our knowledge, previous meta-analyses have not evaluated CSF biomarkers’ diagnostic performance in relation to disease severity. However, greater brain structural changes and alterations in brain connectivity, as markers of disease severity, are associated with more pathological CSF biomarker levels [76 –78]. Therefore, it seems reasonable to expect higher diagnostic performance in more severely impaired patients.

Specific subgroup analyses could not be performed for other confounders due to insufficient information available. Nonetheless, clinical-pathophysiological heterogeneity of FTLD as well as postmortem diagnostic confirmation is of relevance and results from individual studies deserve some discussion. An important candidate to influence CSF biomarkers’ diagnostic performance is the clinical and pathophysiological heterogeneity of the FTLD spectrum. The different clinical FTLD syndromes, i.e., FTD (including bvFTD, progressive non-fluent aphasia, and semantic dementia), progressive supranuclear palsy syndrome, and corticobasal syndrome, could imply different profiles of CSF biomarkers. However, data reported in the articles did not enable subgroup analyses. Unsatisfactory results were obtained in the few studies including or reporting data separately for bvFTD [45, 50], except in de Souza et al. [51]. This last study showed that Aβ₄₂ was less sensitive but more specific when AD was compared with bvFTD than when compared with semantic dementia. The opposite occurred for tau. Moreover, progressive non-fluent aphasia could not clearly be differentiated from AD. Grossman et al. [53] showed that semantic dementia was the variant with more AD-like CSF biomarkers profile, and significantly different from the other FTLD variants. Regarding pathophysiological heterogeneity in FTLD, three subtypes have been described depending on the main protein deposits [14, 79]: tau, TDP-43 (TAR DNA-binding protein 43), and FUS (tumor associated protein fused in sarcoma). There is a strong correspondence between the clinical diagnosis of semantic dementia and the TDP-43 subtype [7 , 81], and both progressive supranuclear palsy and corticobasal syndromes with the tau subtype [82, 83]. However, the clinical diagnosis of bvFTD has not been clearly associated with any of these [7]. Hence, results from both primary studies and the current meta-analysis should thus be interpreted in this context of FTLD heterogeneity. Some recommendations are given below.

Another source of heterogeneity is the frequent use of clinical diagnosis as reference standard instead of postmortem confirmed diagnosis. Previous research has evidenced considerable variability in false positive rates of FTLD clinical diagnosis when pathologically confirmed [84 –86]. Irwin et al. [56] found that clinical diagnosis yielded 67% sensitivity and 87% specificity when predicting postmortem confirmed diagnosis. Toledo et al. [69] reported that the use of clinical diagnosis instead of pathologically confirmed diagnosis underestimates sensitivity and specificity values in about 10–20% for CSF tau and Aβ biomarkers. However, studies with pathological diagnosis are still scarce and normally include small samples. In this review, only eight studies included samples with postmortem diagnostic confirmation, some of them overlapped, and they did not evaluate the same biomarkers. However, the four studies that evaluated the ratio t-tau/Aβ₄₂ obtained an excellent pooled result (88% sensitivity and 95% specificity). Nonetheless, CSF biomarkers’ diagnostic performance in pathologically confirmed AD still remains uncertain. It must be taken into account that most of the interpretations from primary studies on the potential mechanisms involved in FTLD are derived from clinically diagnosed patients.

Variability in cut-off values to discriminate between diagnostic groups is common in the field. Cut-offs are usually obtained from the same sample in which sensitivity and specificity values are calculated [60, 87]. Other studies have used external cut-off values [55 , 67]. We did not find statistically significant differences between the two methods. Besides potential differences in the samples evaluated, technical factors regarding the analytical platforms used to measure the CSF biomarkers could also introduce bias in the evaluation of their diagnostic performance. Results from a multicenter study comparing the analytical accuracy of CSF t-tau measurements show an inter-center coefficient of variation for t-tau levels of 16% [88]. Currently, efforts are being made to harmonize biomarker methods and research protocols with the hope of making results comparable among centers [89, 90].

Some limitations should also be discussed. The assessment of methodological quality reflected possible patient selection bias in eight of the included studies. This was related to inclusion of not completely consecutive or random samples and not perfect avoidance of inappropriate exclusions. In particular, patients were normally selected from specialized centers on the basis of availability of CSF data, a procedure not always performed in all incoming patients. Further, underreporting of data was observed. Most of the studies did not report information on confounders other than age, disease duration, global cognitive impairment, clinical diagnostic criteria for FTLD, and cut-offs values for discriminating between diagnostic groups. Further, multivariate analysis on the effect of these confounding factors was not possible. This was supplied by performing subgroup analyses. Although these analyses provided relevant information they need to be considered with caution due to aspects such as the use of univariate approach or use of arbitrary thresholds (medians) to define the subgroups. Beta CIs were wide in some analyses, and the assumption of symmetry (equal accuracy for different thresholds) must thus be considered with caution. Another limitation is that MMSE is commonly used as a measure of global cognition and disease severity in AD. However, other instruments such as the Frontotemporal Dementia Rating Scale [91] are more appropriate for the same purpose in FTLD, which could vary the results of the subgroup analyses on MMSE. Finally, FTLD is a heterogeneous disorder both clinically and pathophysiologicaly. Primary studies do not normally specify the FTLD subtype included or provide CSF biomarkers values for combined groups, therefore it was not possible to conduct separate meta-analyses for the different subtypes. This is an important limitation and future primary studies should provide more information in thisregard.

CONCLUSIONS

At present, differential diagnosis between AD and FTLD is one of the main challenges and great controversy exists on the clinical utility of core CSF biomarkers. A previous meta-analysis showed unsatisfactory sensitivity and specificity values for CSF t-tau, and better but still suboptimal results for p-tau. Moreover, Aβ₄₂ was not investigated [24]. By using a more appropriated statistical technique and including later studies, we have replicated those results on t-tau and p-tau, and offered meta-analytic evidence on the optimal performance of the p-tau/Aβ₄₂ ratio in the differential diagnosis between AD and FTLD. In particular, the p-tau/Aβ₄₂ ratio may be especially indicated for young AD and FTLD patients, where differential diagnosis is indeed more frequent in the clinical practice. Moreover, depending on the level of global cognitive impairment, the use of p-tau may be preferable (high MMSE scores); otherwise the t-tau/Aβ₄₂ ratio could be more appropriate (low MMSE scores).

Based on the findings of the current study and the systematic review performed, the following bullet points are meant to conclude on the current state of the topic and propose recommendations for future research:

The reviewed core CSF biomarkers are currently used in some centers for the differential diagnosis of AD [92]. If methodological limitations are overcome in the future, the use of CSF biomarkers may be generalized to other hospitals and be definitively implemented as part of the clinical routine.

Postmortem studies are needed to clarify the potential bias implied in using clinical diagnosis as reference standard. Most of current interpretations on the potential mechanisms involved in FTLD derive from clinically diagnosedpatients.

Given the significant influence of several confounder factors on the CSF biomarkers, it is of utmost importance that future studies report information on age, years of education, disease duration, disease severity, and global cognitive impairment of the studied groups. The clinical diagnostic criteria used as well as the type of technique used to measure CSF biomarker levels, and cut-offs values for discriminating between diagnostic groups, should also be clearly stated.

Ideally, specific research should be carried out in order to understand the mechanisms behind the influence of these confounding factors alone and in interaction on the CSF biomarkers.

A more complete reporting of results in future studies is also required to better understand the role of the CSF biomarker levels in dementia. In particular, this report should at least clearly specify the CSF biomarkers investigated (including epitopes), ideally with separate reporting of values (in case combinations are also reported), and careful phenotyping and stratification of patients with different disease subtypes of AD and particularly patients with FTLD. Overlapped samples with previous studies should also be clearly specified.

To advance in these aspects is expected to make it possible to definitively identify those conditions where the CSF biomarkers are useful both for clinical practice and research.

Footnotes

¹AR: Amado Rivero-Santana; LP: Lilisbeth Perestelo-Perez.

²AR: Amado Rivero-Santana; DF: Daniel Ferreira.

ACKNOWLEDGMENTS

The authors would like to thank the Spanish Ministry of Health, Social Services, and Equality; the Swedish Brain Power; the Strategic Research Programme in Neuroscience at Karolinska Institutet (StratNeuro); the Swedish Medical Society; and the regional agreement on medical training, clinical research (ALF) between Stockholm County Council and Karolinska Institutet.

Authors’ disclosures available online ().

Appendix

References

Brunnström

, Gustafson

, Passant

, Englund

(2009) Prevalence of dementia subtypes: A 30-year retrospective survey of neuropathological reports. Arch Gerontol Geriatr 49, 146–149.

Ferri

, Prince

, Brayne

, Brodaty

, Fratiglioni

, Hall

, Hasegawa

, Hendrie

, Huang

, Jorm

, Menezes

, Rimmer

, Scazufca

(2010) Global prevalence of dementia: A Delphi consensus study. Lancet 366, 2112–2117.

Blennow

, de

Leon MJ

, Zetterberg

(2006) Alzheimer’s disease. Lancet 368, 387–403.

Braak

, Zetterberg

, Del Tredici

, Blennow

(2013) Intraneuronal tau aggregation precedes diffuse plaque deposition, but amyloid-β changes occur before increases of tau in cerebrospinal fluid. Acta Neuropathol 126, 631–641.

Onyike

, Diehl-Schmid

(2013) The epidemiology of frontotemporal dementia. Int Rev Psychiatry 25, 130–137.

Seelaar

, Rohrer

, Pijnenburg

, Fox

, van Swieten

(2011) Clinical, genetic and pathological heterogeneity of frontotemporal dementia: A review. J Neurol Neurosurg Psychiatry 82, 476–486.

Josephs

, Hodges

, Snowden

, Mackenzie

, Neumann

, Mann

, Dickson

(2011) Neuropathological background of phenotypical variability in frontotemporal dementia. Acta Neuropathol 122, 137–153.

Neary

, Snowden

, Gustafson

, Passant

, Stuss

, Black

, Freedman

, Kertesz

, Robert

, Albert

, Boone

, Miller

, Cummings

, Benson

(1998) Frontotemporal lobar degeneration: A consensus on clinical diagnostic criteria. Neurology 51, 1546–1554.

Gustafson

(1987) Frontal lobe degeneration of non-Alzheimer type. II. Clinical picture and differential diagnosis. Arch Gerontol Geriatr 6, 209–223.

10.

No authors listed (1994) Clinical and neuropathological criteria for frontotemporal dementia. The Lund and Manchester Groups. J Neurol Neurosurg Psychiatry 57, 416–418.

11.

McKhann

, Albert

, Grossman

, Miller

, Dickson

, Trojanowski

(2001) Clinical and pathological diagnosis of frontotemporal dementia: Report of the Work Group on Frontotemporal Dementia and Pick’s Disease. Arch Neurol 58, 1803–1809.

12.

Neary

, Snowden

(1996) Fronto-temporal dementia: Nosology, neuropsychology, and neuropathology. Brain Cogn 31, 176–187.

13.

Rascovsky

, Hodges

, Knopman

, Mendez

, Kramer

, Neuhaus

, van Swieten

, Seelaar

, Dopper

, Onyike

, Hillis

, Josephs

, Boeve

, Kertesz

, Seeley

, Rankin

, Johnson

, Gorno-Tempini

, Rosen

, Prioleau-Latham

, Lee

, Kipps

, Lillo

, Piguet

, Rohrer

, Rossor

, Warren

, Fox

, Galasko

, Salmon

, Black

, Mesulam

, Weintraub

, Dickerson

, Diehl-Schmid

, Pasquier

, Deramecourt

, Lebert

, Pijnenburg

, Chow

, Manes

, Grafman

, Cappa

, Freedman

, Grossman

, Miller

(2011) Sensitivity of revised diagnostic criteria for the behavioural variant of frontotemporal dementia. Brain 134, 2456–2477.

14.

Mackenzie

, Neumann

, Bigio

, Cairns

, Alafuzoff

, Kril

, Kovacs

, Ghetti

, Halliday

, Holm

, Ince

, Kamphorst

, Revesz

, Rozemuller

, Kumar-Singh

, Akiyama

, Baborie

, Spina

, Dickson

, Trojanowski

, Mann

(2010) Nomenclature and nosology for neuropathologic subtypes of frontotemporal lobar degeneration: An update. Acta Neuropathol 119, 1–4.

15.

Gorno-Tempini

, Hillis

, Weintraub

, Kertesz

, Mendez

, Cappa

, Ogar

, Rohrer

, Black

, Boeve

, Manes

, Dronkers

, Vandenberghe

, Rascovsky

, Patterson

, Miller

, Knopman

, Hodges

, Mesulam

, Grossman

(2011) Classification of primary progressive aphasia and its variants. Neurology 76, 1006–1014.

16.

Graham

, Davies

, Xuereb

, Halliday

, Kril

, Creasey

, Graham

, Hodges

(2005) Pathologically proven frontotemporal dementia presenting with severe amnesia. Brain 128, 597–605.

17.

Kertesz

, McMonagle

, Blair

, Davidson

, Munoz

(2005) The evolution and pathology of frontotemporal dementia. Brain 128, 1996–2005.

18.

Liscic

, Storandt

, Cairns

, Morris

(2007) Clinical and psychometric distinction of frontotemporal and Alzheimer dementias. Arch Neurol 64, 535–540.

19.

Stopford

, Snowden

, Thompson

, Neary

(2008) Variability in cognitive presentation of Alzheimer’s disease. Cortex 44, 185–195.

20.

McKhann

, Knopman

, Chertkow

, Hyman

, Jack

, Kawas

, Klunk

, Koroshetz

, Manly

, Mayeux

, Mohs

, Morris

, Rossor

, Scheltens

, Carrillo

, Thies

, Weintraub

, Phelps

(2011) 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 7, 263–269.

21.

Mitchell

(2009) CSF phosphorylated tau in the diagnosis and prognosis of mild cognitive impairment and Alzheimer’s disease: A meta-analysis of 51 studies. J Neurol Neurosurg Psychiatry 80, 966–975.

22.

Bloudek

, Spackman

, Blankenburg

, Sullivan

(2011) Review and meta-analysis of biomarkers and diagnostic imaging in Alzheimer’s disease. J Alzheimers Dis 26, 627–645.

23.

Hampel

, Bürger

, Teipel

, Bokde

ALW

, Zetterberg

, Blennow

(2008) Core candidate neurochemical and imaging biomarkers of Alzheimer’s disease. Alzheimers Dement 4, 38–48.

24.

Van Harten

, Kester

, Visser

P-J

, Blankenstein

, Pijnenburg

, van der Flier

, Scheltens

(2011) Tau and p-tau as CSF biomarkers in dementia: A meta-analysis. Clin Chem Lab Med 49, 353–366.

25.

Tang

, Huang

, Wang

Z-Y

, Yao

Y-Y

(2014) 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 345, 26–36.

26.

Tang

, Huang

, Yao

Y-Y

, Wang

, Wu

Y-L

, Wang

Z-Y

(2014) 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 121, 1541–1553.

27.

Dahabreh

, Trikalinos

, Lau

, Schmid

(2012) An empirical assessment of bivariate methods for meta-analysis of test accuracy [Internet]. Agency for Healthcare Research and Quality (US), Rockville, MD, 2012 Nov. Report No.: 12(13)-EHC136-EF.

28.

Harbord

, Whiting

, Sterne

, Egger

, Deeks

, Shang

, Bachmann

(2008) An empirical comparison of methods for meta-analysis of diagnostic accuracy showed hierarchical models are necessary. J Clin Epidemiol 61, 1095–1103.

29.

Rutter

, Gatsonis

(2001) A hierarchical regression approach to meta-analysis of diagnostic test accuracy evaluations. Stat Med 20, 2865–2884.

30.

Ferreira

, Perestelo-Pérez

, Westman

, Wahlund

L-O

, Sarría

, Serrano-Aguilar

(2014) Meta-review of CSF core biomarkers in Alzheimer’s disease: The state-of-the-art after the new revised diagnostic criteria. Front Aging Neurosci 6, 1–24.

31.

Ferreira

, Rivero-Santana

, Perestelo-Pérez

, Westman

, Wahlund

L-O

, Sarría

, Serrano-Aguilar

(2014) Improving CSF biomarkers’ performance for predicting progression from mild cognitive impairment to Alzheimer’s disease by considering different confounding factors: A meta-analysis. Front Aging Neurosci 6, 287.

32.

Valcárcel-Nazco

, Perestelo-Pérez

, Molinuevo

, Mar

, Castilla

, Serrano-Aguilar

(2014) Cost-effectiveness of the use of biomarkers in cerebrospinal fluid for Alzheimer’s disease. J Alzheimers Dis 42, 777–788.

33.

McKhann

, Drachman

, Folstein

, Katzman

, Price

, Stadlan

(1984) Clinical diagnosis of Alzheimer’s disease: Report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer’s Disease. Neurology 34, 939–944.

34.

Whiting

, Rutjes

AWS

, Westwood

, Mallett

, Deeks

, Reitsma

, Leeflang

MMG

, Sterne

JAC

, Bossuyt

PMM

(2011) QUADAS-2: A revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med 155, 529–536.

35.

Liberati

, Altman

, Tetzlaff

, Mulrow

, Ioannidis

JPA

, Clarke

, Devereaux

, Kleijnen

, Moher

(2009) The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: Explanation and elaboration. Ann Intern Med 151, 65–94.

36.

Moher

, Liberati

, Tetzlaff

, Altman

(2010) Preferred reporting items for systematic reviews and meta-analyses: The PRISMA statement. Int J Surg 8, 336–341.

37.

Macaskill

(2004) Empirical Bayes estimates generated in a hierarchical summary ROC analysis agreed closely with those of a full Bayesian analysis. J Clin Epidemiol 57, 925–932.

38.

Harbord

, Whiting

(2009) Metandi: Meta-analysis of diagnostic accuracy using hierarchical logistic regression. Stata J 9, 211–229.

39.

(1998) Consensus report of the Working Group on: Molecular and Biochemical Markers of Alzheimer’s Disease. The Ronald and Nancy Reagan Research Institute of the Alzheimer’s Association and the National Institute on Aging Working Group. Neurobiol Aging 19, 109–116. Erratum in: Neurobiol Aging 19, 285, 1998.

40.

Qizilbash

(2002) Evidenced-based Dementia Practice, Blackwell Publishing, Oxford.

41.

Baldeiras

, Santana

, Leitão

, Ribeiro

, Pascoal

, Duro

, Lemos

, Santiago

, Almeida

, Oliveira

(2015) Cerebrospinal fluid Aβ40 is similarly reduced in patients with frontotemporal lobar degeneration and Alzheimer’s disease. J Neurol Sci 358, 308–316.

42.

Bian

, Van Swieten

, Leight

, Massimo

, Wood

, Forman

, Moore

, de Koning

, Clark

, Rosso

, Trojanowski

, Lee

VM-Y

, Grossman

(2008) CSF biomarkers in frontotemporal lobar degeneration with known pathology. Neurology 70, 1827–1835.

43.

Bibl

, Gallus

, Welge

, Esselmann

, Wolf

, Rüther

, Wiltfang

(2012) Cerebrospinal fluid amyloid-β 2-42 is decreased in Alzheimer’s, but not in frontotemporal dementia. J Neural Transm 119, 805–813.

44.

Blasko

, Lederer

, Oberbauer

, Walch

, Kemmler

, Hinterhuber

, Marksteiner

, Humpel

(2006) Measurement of thirteen biological markers in CSF of patients with Alzheimer’s disease and other dementias. Dement Geriatr Cogn Disord 21, 9–15.

45.

Blennow

, Wallin

, Agren

, Spenger

, Siegfried

, Vanmechelen

(1995) Tau protein in cerebrospinal fluid: A biochemical marker for axonal degeneration in Alzheimer disease? Mol Chem Neuropathol 26, 231–245.

46.

Brunnström

, Rawshani

, Zetterberg

, Blennow

, Minthon

, Passant

, Englund

(2010) Cerebrospinal fluid biomarker results in relation to neuropathological dementia diagnoses. Alzheimers Dement 6, 104–109.

47.

Buerger

, Zinkowski

, Teipel

, Tapiola

, Arai

, Blennow

, Andreasen

, Hofmann-Kiefer

, DeBernardis

, Kerkman

, McCulloch

, Kohnken

, Padberg

, Pirttilä

, Schapiro

, Rapoport

, Möller

H-J

, Davies

, Hampel

(2002) Differential diagnosis of Alzheimer disease with cerebrospinal fluid levels of tau protein phosphorylated at threonine 231. Arch Neurol 59, 1267–1272.

48.

Clark

, Xie

, Chittams

, Ewbank

, Peskind

, Galasko

, Morris

, McKeel

, Farlow

, Weitlauf

, Quinn

, Kaye

, Knopman

, Arai

, Doody

, DeCarli

, Leight

, Lee

VM-Y

, Trojanowski

(2003) Cerebrospinal fluid tau and beta-amyloid: How well do these biomarkers reflect autopsy-confirmed dementia diagnoses? Arch Neurol 60, 1696–1702.

49.

De Jong

, Jansen

RWMM

, Pijnenburg

, van Geel

, Borm

, Kremer

HPH

, Verbeek

(2007) CSF neurofilament proteins in the differential diagnosis of dementia. J Neurol Neurosurg Psychiatry 78, 936–938.

50.

De Rino

, Martinelli-Boneschi

, Caso

, Zuffi

, Zabeo

, Passerini

, Comi

, Magnani

, Franceschi

(2012) CSF metabolites in the differential diagnosis of Alzheimer’s disease from frontal variant of frontotemporal dementia. Neurol Sci 33, 973–977.

51.

De Souza

, Lamari

, Belliard

, Jardel

, Houillier

, De Paz

, Dubois

, Sarazin

(2011) Cerebrospinal fluid biomarkers in the differential diagnosis of Alzheimer’s disease from other cortical dementias. J Neurol Neurosurg Psychiatry 82, 240–246.

52.

Gabelle

, Dumurgier

, Vercruysse

, Paquet

, Bombois

, Laplanche

J-L

, Peoc’h

, Schraen

, Buée

, Pasquier

, Hugon

, Touchon

, Lehmann

(2013) Impact of the 2008-2012 French Alzheimer Plan on the use of cerebrospinal fluid biomarkers in research memory center: The PLM Study. J Alzheimers Dis 34, 297–305.

53.

Grossman

, Farmer

, Leight

, Work

, Moore

, Van Deerlin

, Pratico

, Clark

, Coslett

, Chatterjee

, Gee

, Trojanowski

, Lee

VM-Y

(2005) Cerebrospinal fluid profile in frontotemporal dementia and Alzheimer’s disease. Ann Neurol 57, 721–729.

54.

Hampel

, Buerger

, Zinkowski

, Teipel

, Goernitz

, Andreasen

, Sjoegren

, DeBernardis

, Kerkman

, Ishiguro

, Ohno

, Vanmechelen

, Vanderstichele

, McCulloch

, Moller

, Davies

PBK

(2004) Measurement of phosphorylated tau epitopes in the differential diagnosis of Alzheimer disease: A comparative cerebrospinal fluid study. Arch Gen Psychiatry 61, 95–102.

55.

Herbert

, Aerts

, Kuiperij

, Claassen

JAHR

, Spies

, Esselink

RAJ

, Bloem

, Verbeek

(2014) Addition of MHPG to Alzheimer’s disease biomarkers improves differentiation of dementia with Lewy bodies from Alzheimer’s disease but not other dementias. Alzheimers Dement 10, 448–455.e2.

56.

Irwin

, McMillan

, Toledo

, Arnold

, Shaw

, Wang

L-S

, Van Deerlin

, Lee

VM-Y

, Trojanowski

, Grossman

(2012) Comparison of cerebrospinal fluid levels of tau and Aβ 1-42 in Alzheimer disease and frontotemporal degeneration using 2 analytical platforms. Arch Neurol 69, 1018–1025.

57.

Kapaki

, Paraskevas

, Papageorgiou

, Bonakis

, Kalfakis

, Zalonis

, Vassilopoulos

(2008) Diagnostic value of CSF biomarker profile in frontotemporal lobar degeneration. Alzheimer Dis Assoc Disord 22, 47–53.

58.

Koopman

, Le Bastard

, Martin

J-J

, Nagels

, De Deyn

, Engelborghs

(2009) Improved discrimination of autopsy-confirmed Alzheimer’s disease (AD) from non-AD dementias using CSF P-tau(181P). Neurochem Int 55, 214–218.

59.

Paraskevas

, Kapaki

, Liappas

, Theotoka

, Mamali

, Zournas

, Lykouras

(2005) The diagnostic value of cerebrospinal fluid tau protein in dementing and nondementing neuropsychiatric disorders. J Geriatr Psychiatry Neurol 18, 163–173.

60.

Petzold

, Chapman

, Schraen

, Verwey

, Pasquier

, Bombois

, Brettschneider

, Fox

, von Arnim

CAF

, Teunissen

, Pijnenburg

, Riepe

, Otto

, Tumani

, Scheltens

, Buee

, Rossor

(2010) An unbiased, staged, multicentre, validation strategy for Alzheimer’s disease CSF tau levels. Exp Neurol 223, 432–438.

61.

Pijnenburg

YAL

, Schoonenboom

NSM

, Rosso

, Mulder

, Van Kamp

, Van Swieten

, Scheltens

(2004) CSF tau and Abeta42 are not useful in the diagnosis of frontotemporal lobar degeneration. Neurology 62, 1649.

62.

Riemenschneider

, Wagenpfeil

, Diehl

, Lautenschlager

, Theml

, Heldmann

, Drzezga

, Jahn

, Förstl

, Kurz

(2002) Tau and Abeta42 protein in CSF of patients with frontotemporal degeneration. Neurology 58, 1622–1628.

63.

Schoonenboom

NSM

, Pijnenburg

YAL

, Mulder

, Rosso

, Elk

, Van Kamp

, van Swieten

, van Scheltens

(2004) Amyloid beta (1-42) and phosphorylated tau in CSF as markers for early-onset Alzheimer disease. Neurology 62, 1580–1584.

64.

Seeburger

, Holder

, Combrinck

, Joachim

, Laterza

, Tanen

, Dallob

, Chappell

, Snyder

, Flynn

, Simon

, Modur

, Potter

, Wilcock

, Savage

, Smith

(2015) Cerebrospinal fluid biomarkers distinguish postmortem-confirmed Alzheimer’s disease from other dementias and healthy controls in the OPTIMA cohort. J Alzheimers Dis 44, 525–539.

65.

Sjögren

, Rosengren

, Minthon

, Davidsson

, Blennow

, Wallin

(2000) Cytoskeleton proteins in CSF distinguish frontotemporal dementia from AD. Neurology 54, 1960–1964.

66.

Sjögren

, Davidsson

, Tullberg

, Minthon

, Wallin

, Wikkelso

, Granérus

, Vanderstichele

, Vanmechelen

, Blennow

(2001) Both total and phosphorylated tau are increased in Alzheimer’s disease. J Neurol Neurosurg Psychiatry 70, 624–630.

67.

Skillbäck

, Farahmand

, Rosén

, Mattsson

, Nägga

, Kilander

, Religa

, Wimo

, Winblad

, Schott

, Blennow

, Eriksdotter

, Zetterberg

(2015) Cerebrospinal fluid tau and amyloid-β1-42 in patients with dementia. Brain 138, 2716–2731.

68.

Struyfs

, Van Broeck

, Timmers

, Fransen

, Sleegers

, Van Broeckhoven

, De Deyn

, Streffer

, Mercken

, Engelborghs

(2015) Diagnostic accuracy of cerebrospinal fluid amyloid-β isoforms for early and differential dementia diagnosis. J Alzheimers Dis 45, 813–822.

69.

Toledo

, Brettschneider

, Grossman

, Arnold

, Hu

, Xie

, Lee

VM-Y

, Shaw

, Trojanowski

(2012) CSF biomarkers cutoffs: The importance of coincident neuropathological diseases. Acta Neuropathol 124, 23–35.

70.

Verwey

, Kester

, van der Flier

, Veerhuis

, Berkhof

, Twaalfhoven

, Blankenstein

, Scheltens

, Pijnenburg

(2010) Additional value of CSF amyloid-beta 40 levels in the differentiation between FTLD and control subjects. J Alzheimers Dis 20, 445–452.

71.

Holtzman

(2011) CSF biomarkers for Alzheimer’s disease: Current utility and potential future use. Neurobiol Aging 32(Suppl 1), S4–S9.

72.

Anoop

, Singh

, Jacob

, Maji

(2010) CSF biomarkers for Alzheimer’s disease diagnosis. Int J Alzheimers Dis 2010, 606802.

73.

Lorenzi

, Pennec

, Frisoni

, Ayache

(2015) Disentangling normal aging from Alzheimer’s disease in structural magnetic resonance images. Neurobiol Aging 36(Suppl 1), S42–S52.

74.

Bouwman

, Schoonenboom

NSM

, Verwey

, van Elk

, Kok

, Blankenstein

, Scheltens

, van der Flier

(2009) CSF biomarker levels in early and late onset Alzheimer’s disease. Neurobiol Aging 30, 1895–1901.

75.

Mattsson

, Rosén

, Hansson

, Andreasen

, Parnetti

, Jonsson

, Herukka

S-K

, van der Flier

, Blankenstein

, Ewers

, Rich

, Kaiser

, Verbeek

, Olde Rikkert

, Tsolaki

, Mulugeta

, Aarsland

, Visser

, Schröder

, Marcusson

, de Leon

, Hampel

, Scheltens

, Wallin

, Eriksdotter-Jönhagen

, Minthon

, Winblad

, Blennow

, Zetterberg

(2012) Age and diagnostic performance of Alzheimer disease CSF biomarkers. Neurology 78, 468–476.

76.

, Li

T-Q

, Andreasen

, Wiberg

, Westman

, Wahlund

L-O

(2014) The association between biomarkers in cerebrospinal fluid and structural changes in the brain in patients with Alzheimer’s disease. J Intern Med 275, 418–427.

77.

, Li

T-Q

, Andreasen

, Wiberg

, Westman

, Wahlund

L-O

(2013) Ratio of Aβ42/P-tau181p in CSF is associated with aberrant default mode network in AD. Sci Rep 3, 1339.

78.

Orellana

, Ferreira

, Mecocci

, Vellas

, Tsolaki

, Kłoszewska

, Soininen

, Lovestone

, Simmons

, Wahlund

L-O

, Westman

(2016) Measuring global brain atrophy with the brain volume / cerebrospinal fluid index: Normative values, cut-offs, and clinical associations. Neurodegener Dis 16, 77–86.

79.

Hales

, Hu

(2013) From frontotemporal lobar degeneration pathology to frontotemporal lobar degeneration biomarkers. Int Rev Psychiatry 25, 210–220.

80.

Hodges

, Mitchell

, Dawson

, Spillantini

, Xuereb

, McMonagle

, Nestor

, Patterson

(2010) Semantic dementia: Demography, familial factors and survival in a consecutive series of 100 cases. Brain 133, 300–306.

81.

Josephs

, Stroh

, Dugger

, Dickson

(2009) Evaluation of subcortical pathology and clinical correlations in FTLDU subtypes. Acta Neuropathol 118, 349–358.

82.

Josephs

, Petersen

, Knopman

, Boeve

, Whitwell

, Duffy

, Parisi

, Dickson

(2006) Clinicopathologic analysis of frontotemporal and corticobasal degenerations and PSP. Neurology 66, 41–48.

83.

Snowden

, Neary

, Mann

(2007) Frontotemporal lobar degeneration: Clinical and pathological relationships. Acta Neuropathol 114, 31–38.

84.

Snowden

, Thompson

, Stopford

, Richardson

AMT

, Gerhard

, Neary

, Mann

DMA

(2011) The clinical diagnosis of early-onset dementias: Diagnostic accuracy and clinicopathological relationships. Brain 134, 2478–2492.

85.

Knopman

, Boeve

, Parisi

, Dickson

, Smith

, Ivnik

, Josephs

, Petersen

(2005) Antemortem diagnosis of frontotemporal lobar degeneration. Ann Neurol 57, 480–488.

86.

Forman

, Farmer

, Johnson

, Clark

, Arnold

, Coslett

, Chatterjee

, Hurtig

, Karlawish

, Rosen

, Van Deerlin

, Lee

VM-Y

, Miller

, Trojanowski

, Grossman

(2006) Frontotemporal dementia: Clinicopathological correlations. Ann Neurol 59, 952–962.

87.

Blennow

, Hampel

(2003) CSF markers for incipient Alzheimer’s disease. Lancet Neurol 2, 605–613.

88.

Verwey

, van der Flier

, Blennow

, Clark

, Sokolow

, De Deyn

, Galasko

, Hampel

, Hartmann

, Kapaki

, Lannfelt

, Mehta

, Parnetti

, Petzold

, Pirttila

, Saleh

, Skinningsrud

, Swieten

JCV

, Verbeek

, Wiltfang

, Younkin

, Scheltens

, Blankenstein

(2009) A worldwide multicentre comparison of assays for cerebrospinal fluid biomarkers in Alzheimer’s disease. Ann Clin Biochem 46, 235–240.

89.

Shaw

, Vanderstichele

, Knapik-Czajka

, Figurski

, Coart

, Blennow

, Soares

, Simon

, Lewczuk

, Dean

, Siemers

, Potter

, Lee

VM-Y

, Trojanowski

(2011) Qualification of the analytical and clinical performance of CSF biomarker analyses in ADNI. Acta Neuropathol 121, 597–609.

90.

Mattsson

, Andreasson

, Persson

, Arai

, Batish

, Bernardini

, Bocchio-Chiavetto

, Blankenstein

, Carrillo

, Chalbot

, Coart

, Chiasserini

, Cutler

, Dahlfors

, Duller

, Fagan

, Forlenza

, Frisoni

, Galasko

, Galimberti

, Hampel

, Handberg

, Heneka

, Herskovits

, Herukka

S-K

, Holtzman

, Humpel

, Hyman

, Iqbal

, Jucker

, Kaeser

, Kaiser

, Kapaki

, Kidd

, Klivenyi

, Knudsen

, Kummer

, Lui

, Lladó

, Lewczuk

, Li

Q-X

, Martins

, Masters

, McAuliffe

, Mercken

, Moghekar

, Molinuevo

, Montine

, Nowatzke

, O’Brien

, Otto

, Paraskevas

, Parnetti

, Petersen

, Prvulovic

, de Reus

HPM

, Rissman

, Scarpini

, Stefani

, Soininen

, Schröder

, Shaw

, Skinningsrud

, Skrogstad

, Spreer

, Talib

, Teunissen

, Trojanowski

, Tumani

, Umek

, Van Broeck

, Vanderstichele

, Vecsei

, Verbeek

, Windisch

, Zhang

, Zetterberg

, Blennow

(2011) The Alzheimer’s Association external quality control program for cerebrospinal fluid biomarkers. Alzheimers Dement 7, 386–395.e6.

91.

Mioshi

, Hsieh

, Savage

, Hornberger

, Hodges

(2010) Clinical staging and disease progression in frontotemporal dementia. Neurology 74, 1591–1597.

92.

Ferreira

, Jelic

, Cavallin

, Oeksengaard

, Snaedal

, Høgh

, Bo Andersen

, Naik

, Engedal

, Westman

, Wahlund

(2016) Electroencephalography is a good complement to currently established dementia biomarkers. Dement Geriatr Cogn Disord 42, 80–92.

Cerebrospinal Fluid Biomarkers for the Differential Diagnosis between Alzheimer’s Disease and Frontotemporal Lobar Degeneration: Systematic Review,HSROC Analysis,and Confounding Factors

Abstract

Keywords

INTRODUCTION

MATERIAL AND METHODS

Data source and study selection

Data collection, risk of bias, and quality assessment

Statistical analysis

RESULTS

Total tau (t-tau)

Phosphorylated tau (p-tau)

Amyloid-beta 1-42 (Aβ42)

T-tau/Aβ42 and p-tau/Aβ42 ratios

Other combinations of CSF biomarkers

DISCUSSION

CONCLUSIONS

Footnotes

ACKNOWLEDGMENTS

Appendix

References

Amyloid-beta 1-42 (Aβ₄₂)

T-tau/Aβ₄₂ and p-tau/Aβ₄₂ ratios