1、Cerebrospinal fluid biomarkers in the differentialdiagnosis of Alzheimers disease from othercortical dementiasLeonardo Cruz de Souza,1,2Foudil Lamari,3Serge Belliard,4Claude Jardel,3Caroline Houillier,2Raphael De Paz,2Bruno Dubois,1,2Marie Sarazin1,2ABSTRACTBackground Considering that most semantic
2、dementia(SD) and frontotemporal dementia (FTD) patients showno post-mortem Alzheimers disease (AD) pathology,cerebrospinal fluid (CSF) biomarkers may be of value fordistinguishing these patients from those with AD.Additionally, biomarkers may be useful for identifyingpatients with atypical phenotypi
3、c presentations of AD,such as posterior cortical atrophy (PCA) and primaryprogressive non-fluent or logopenic aphasia (PNFLA).Methods The authors investigated CSF biomarkers(beta-amyloid 1e42 (Ab42), total tau (T-tau) andphosphorylated tau (P-tau) in 164 patients with AD(n60), PCA (n15), behavioural
4、 variant FTD (n27),SD (n19), PNFLA (n26) and functional cognitivedisorders (FCD, n17). The authors then examined thediagnostic value of these CSF biomarkers indistinguishing these patients from those with AD.Results The P-Tau/Ab42ratio was found to be the bestbiomarker for distinguishing AD from FTD
5、 and SD, witha sensitivity of 91.7% and 98.3%, respectively, anda specificity of 92.6% and 84.2%, respectively. Asexpected, biomarkers were less effective indifferentiating AD from PNFLA and PCA, as significantproportions of PCA and PNFLA patients (60% and 61.5%,respectively) had concurrent alterati
6、ons of both T-tau/Ab42and P-Tau/Ab42ratios. None of the FCD patientshad a typical AD CSF profile or abnormal T-tau/Ab42orP-Tau/Ab42ratios.Conclusion The P-Tau/Ab42ratio is a useful tool todistinguish AD from both FTD and SD, which are knownto involve pathological processes distinct from AD.Biomarker
7、s could be useful for identifying patients withan atypical AD phenotype that includes PNFLA and PCA.INTRODUCTIONWith the prospect of disease-modifying drugs thatwill target the physiopathological process ofAlzheimers disease (AD), it is crucial to improvethe efcacy of the differential diagnosis betw
8、eenAD and other cortical dementias. The determina-tion of cerebrospinal uid (CSF) biomarker prolesof cortical dementias may ameliorate the distinc-tion between AD and newly characterised non-ADdementias, such as frontotemporal dementia (FTD)and semantic dementia (SD). In addition, it mayidentify pot
9、ential patients with atypicalphenotypic presentation of AD, such as posteriorcortical atrophy (PCA) and primary progressiveaphasia (PPA), who may be candidates for emergingtherapies.The CSF biomarkers total tau (T-tau), phos-phorylated tau (P-tau) and beta-amyloid peptide1e42 (Ab42) can distinguish
10、controls from ADsubjects,1 2even in early stages of the disease.3e8The accuracy of CSF biomarkers in the differentialdiagnosis between AD and other cortical demen-tias, however, is not well understood. Previousstudies have primarily focused on the interest inbiomarkers to distinguish AD from frontal
11、 variantsof frontotemporal lobar degenerations (FTLD),which are dened by behavioural changes. Thesestudies have found that CSF biomarkers are usefulin distinguishing AD from FTD,9e17and onestudy11demonstrated the high sensitivity andspecicity of biomarkers in distinguishing AD fromproven FTD (conrma
12、tion via autopsy andgenetics).Besides these data, some important questionsremain unresolved. First, the ability of CSFbiomarkers to distinguish AD from non-behav-ioural variants of FTLDs, such as SD, remains to bestudied. The diagnosis of SD can be difcult and isa common cause of the misdiagnosis of
13、 AD.18While neuropathological studies have reporteda lack of AD lesions in most SD patients,18e21onestudy (which included a small sample of patients)reported that CSF biomarkers in SD subjectsshowed a similar CSF prole to that of ADpatients,14showing an unexpected low power ofbiomarkers to distingui
14、sh between these twodistinct pathologies.The second unsolved question is the ability ofCSF biomarkers to identify AD pathology in atyp-ical focal cortical presentations of AD, such as PCAand logopenic PPA.18 19 22Different from AD, PCAand logopenic PPA are characterised by specicclinical presentatio
15、ns (eg, visualespatial difcultiesin PCA), with relative respect to episodic memoryand focal atrophy. The most frequent underlyingpathology found in autopsy studies, however, isAD.18Thus, these conditions have been coinedfocal cortical presentations of AD.18Biomarkersmay be able to identify the biolo
16、gical mechanismsof PCA and PPA and may also be useful for selectingpatients who would benet from new disease-modifying therapies.The main objective of this study was to analysethe usefulness of CSF biomarkers for distinguishingAD from FTD, SD, PCA and primary progressivenon-uent or logopenic aphasia
17、 (PNFLA). Weanalysed individual data to identify how CSFbiomarkers can be included in the differentialdiagnosis of AD in clinical practice.PFNLASDFTDFCD, for both biomarkers).This was especially true with the FCD, FTD and SD groups(p PNFLAPCAFTDSDFCD).Ab42levels and the two ratios differed signicant
18、ly betweenthe FCD group and the FTD group. However, there were nodifferences in the CSF biomarker levels between the FTD and SDgroups. All biomarker measures were signicantly different inthe PNFLA and PCA groups when compared with the FTD orSD groups.No correlation was found between Ab42and T-tau or
19、 P-Taulevels for the entire population or within each group. A linearregression analysis between CSF T-tau and P-Tau levels showeda signicant correlation for all groups studied (data not shown).We also analysed the statistical correlation (Spearman testwith Bonferroni correction) between different C
20、SF biomarkersand clinical data (age, disease duration or MMSE). The differentCSF biomarkers and ratios (Ab42, T-tau, P-Tau, T-tau/Ab42andP-Tau/Ab42) showed no signicant correlations with age, gender,disease duration or MMSE for the entire population or withineach group (the corrected p was not signi
21、cant for any of thecorrelations).Determination of the sensitivity, specificity and optimalbiomarker cut-off for differential AD diagnosisA pairwise comparison of the area under the ROC curvesrevealed that T-tau/Ab42and P-tau/Ab42were better than eachseparated CSF biomarkers in distinguishing AD pati
22、ents fromFCD, FTD, SD, PNFLA and PCA (see gure 1). Based on theAUC, sensitivity and specicity, the P-tau/Ab42ratio distin-guished AD from FCD and other cortical dementias better thanAb42, T-tau, P-tau and T-tau/Ab42.Table 2 shows the sensitivity, specicity, AUC and optimalcut-offs determined by ROC
23、analysis for distinguishing the ADgroup from each clinical group.The T-tau/Ab42and the P-tau/Ab42ratios differentiated ADpatients from SD or FTD patients better than the individualCSF markers alone. The T-tau/Ab42ratio distinguished ADfrom SD or FTD patients with a 95% sensitivity (for bothgroups),
24、and an 84.2% and 85.2% specicity, respectively.Similarly, the sensitivity for the P-tau/Ab42ratio for distin-guishing AD from FTD or SD was 91.7% and 98.3%, respec-tively, whereas the specicity was 92.6% and 84.2%,respectively (see table 2). In contrast, CSF biomarkers and theirratios were less effe
25、ctive at distinguishing AD from PNFLA andPCA patients.Because we wanted to determine the optimal cut-off fora differential diagnosis of AD, we grouped SD and FTD patientsinto an FTLD group. ROC curve analysis showed that sensi-tivity and specicity of P-tau/Ab42ratio remained high (91% andTable 1 Dem
26、ographic and cerebrospinal fluid data of studied groupsAlzheimersdiseaseBehavioural variantof frontotemporaldementiaSemanticdementiaFunctional cognitivedisordersPosteriorcortical atrophyProgressive non-fluentor logopenic aphasiaN60 27 19 17 15 26Age 63.6 (57 to 72) 66 (59 to 73) 60 (57 to 68) 56 (52
27、 to 58)y 62 (57 to 67) 65.5 (62 to 71)Sex ratio(male/female)30/30 15/12 9/10 12/9 3/12 10/16Disease duration(years)3 (2 to 4) 3 (2 to 5) 4 (2 to 7) Not applicated 3 (2 to 5) 4 (3 to 5)Mini-MentalState Examination20 (17 to 24) 21 (18 to 23) 22 (21 to 27)z 27 (26 to 28)* x 19 (16 to 22) 21 (20 to 24)F
28、rontal AssessmentBattery13 (11 to 14) 13 (6 to 14) 15 (12 to 17) 15 (13 to 17)* 9 (6 to 14) 14 (10 to 17)Ab42 (ng/ml) 237.6 (155.5 to 315.5) 394 (306.2 to 473)* 532 (308.2 to 682.2)* 500 (417 to 614.5)* x 241 (227.2 to 382.7)NSx 263.5 (206 to 351)NSxT-tau (ng/ml) 571 (419 to 796.5) 239 (137 to 407.4
29、)* 317 (209.5 to 433.7)* 185 (141.2 to 262.7)* 436 (368.3 to 496.5)NSx 414 (288 to 556)yxP-tau (ng/ml) 84.2 (69 to 107.5) 43.5 (32.5 to 57)* 46.2 (38.4 to 67.0)* 34 (30.2 to 56.9)* 58 (47.3 to 77.7)zx 59.95 (49 to 86)zxT-tau/Ab42 2.450 (1.688 to 3.433) 0.582 (0.396 to 1.055)* 0.492 (0.35 to 0.928)*
30、0.335 (0.31 to 0.51)* x 1.759 (1.13 to 2.46)zx 1.528 (0.85 to 2.31)zxP-tau/Ab42 0.43 (0.26 to 0.48) 0.13 (0.08 to 0.15)* 0.18 (0.06 to 0.13)* 0.074 (0.066 to 0.093)* x 0.26 (0.14 to 0.31)yx 0.30 (0.16 to 0.39)yxData are presented as median (25th to 75th percentile) for each variable. Comparison betw
31、een Alzheimers disease patients and other groups was performed using a non-parametricManneWhitney U test.*p1.23 and P-tau/Ab420.211) distinguished AD patients from(A) PCA, (B) FTD, (C) SD and (D) PNFLA patients with 95% sensitivityand 94% specificity, with a positive predictive value of 93% anda neg
32、ative predictive value of 95.4%.J Neurol Neurosurg Psychiatry 2011;82:240e246. doi:10.1136/jnnp.2010.207183 243Research on November 13, 2015 - Published by http:/ from PNFLA and AD (see gure 2), are in agreement with theseneuropathological and PETstudies.Similarly, CSF P-Tau/Ab42 ratio does not dis
33、tinguish PCAfrom AD with the same power that we observed in FTD, SD andFCD. Clinicopathological series found that AD was the mostfrequent cause of PCA, accounting for 80e100% of all cases.18 35 40An in vivo study of amyloid deposition by PET conrmed ina single patient that PCA is associated with amy
34、loidosis.41Interestingly, individual data showed abnormal ratios for bothT-tau/Ab42 and P-Tau/Ab42 in 61.5% of PNFLA patients and in60% of PCA patients. A recent neuropathological studydemonstrated that the combination of low Ab42 and high T-taulevel in antemortem CSF predicted the presence of AD-as
35、soci-ated pathological changes with high accuracy and stronglysupported the possibility of AD.42The P-tau/Ab42 ratioexhibited the best sensitivity and the highest specicity forpatients with a conrmed AD diagnosis. Consistent with thesedata, patients with atypical cortical presentation of AD, such as
36、PNFLA and PCA, could be identied by CSF biomarkers,particularly T-tau/Ab42 and P-Tau/Ab42 ratios.Table 2 Results of receiver operating characteristics analyses for discrimination of Alzheimers disease(AD) from other clinical diagnosesCut-off(pg/ml)Sensitivity(%)Specificity(%)Area under the curve(95%
37、 CI) p Value*AD versus functional cognitive disordersAb42395 93.3 82.4 0.940 (0.862 to 0.981) 0.0001T-Tau 289 90 94.1 0.931 (0.849 to 0.976) 0.0001P-Tau 64 83.3 94.1 0.900 (0.811 to 0.957) 0.0001Tau/Ab420.77 98.3 100 0.988 (0.930 to 1.00) 0.0001P-Tau/Ab420.140 98.3 100 0.984 (0.925 to 0.99) 0.0001AD
38、 versus frontotemporal dementiaAb42292.1 68.3 85.2 0.817 (0.720 to 0.892) 0.0001T-Tau 458 70 88.9 0.832 (0.736 to 0.903) 0.0001P-Tau 62.5 83.3 85.2 0.851 (0.759 to 0.918) 0.0001Tau/Ab421.23 95 85.2 0.926 (0.849 to 0.971) 0.0001P-Tau/Ab420.211 91.7 92.6 0.942 (0.871 to 0.981) 0.0001AD versus semantic
39、 dementiaAb42425 98.3 68.4 0.810 (0.706 to 0.889) 0.0001T-Tau 493 65 89.5 0.813 (0.709 to 0.892) 0.0001P-Tau 74 63.3 94.7 0.825 (0.723 to 0.901) 0.0001Tau/Ab421.05 95 84.2 0.876 (0.783 to 0.940) 0.0001P-Tau/Ab420.143 98.3 84.2 0.864 (0.768 to 0.931) 0.0001AD versus progressive non-fluent or logopeni
40、c aphasiaAb42234 50 69.2 0.581 (0.470 to 0.687) 0.2346T-Tau 616 46.7 88.5 0.686 (0.576 to 0.781) 0.0017P-Tau 62 85 53.8 0.671 (0.562 to 0.769) 0.0044Tau/Ab421.678 75 65.4 0.713 (0.606 to 0.806) 0.0002P-Tau/Ab420.296 68.3 69.2 0.688 (0.579 to 0.783) 0.0014Alzheimers disease versus posterior cortical
41、atrophyAb42221 41.7 86.7 0.614 (0.494 to 0.724) 0.1796T-Tau 497 63.3 80 0.663 (0.545 to 0.768) 0.0251P-Tau 59.5 86.7 60 0.705 (0.588 to 0.805) 0.0027Tau/Ab421.33 88.3 46.7 0.706 (0.589 to 0.805) 0.0026P-Tau/Ab420.311 61.7 80 0.733 (0.618 to 0.828) 0.0003AD versus semantic dementia+frontotemporal dem
42、entiaAb42379 91.7 60.9 0.814 (0.727 to 0.883) 0.0001T-Tau 465 68.3 87 0.824 (0.738 to 0.891) 0.0001P-Tau 64 83.3 80.4 0.840 (0.757 to 0.904) 0.0001Tau/Ab421.23 95 84.8 0.905 (0.833 to 0.954) 0.0001P-Tau/Ab420.211 91.7 89.1 0.911 (0.839 to 0.957) 0.0001AD versus semantic dementia+frontotemporal demen
43、tia+functional cognitive disordersAb42379 91.7 66.7 0.848 (0.772 to 0.906) 0.0001T-Tau 341 88.3 73 0.853 (0.778 to 0.910) 0.0001P-Tau 64 83.3 84.1 0.857 (0.782 to 0.913) 0.0001Tau/Ab421.23 95 88.9 0.928 (0.866 to 0.966) 0.0001P-Tau/Ab420.211 91.7 92.1 0.930 (0.870 to 0.968) 0.0001*Significance of di
44、scrimination (p465 pg/ml 41/60 (68.3) 2/17 (11.76) 3/27 (11.1) 4/19 (21.05) 6/15 (40) 10/26 (38.46)P-Tau64.3 pg/ml 50/60 (83.3) 1/17 (5.88) 4/27 (14.81) 5/19 (26) 6/15 (40) 12/26 (46.1)T-tau/Ab421.23 57/60 (95) 0/17 4/27 (14.81) 3/19 (15.8) 11/15 (73) 17/26 (65.3)P-tau/Ab420.211 54/60 (90) 0/17 2/27
45、 (7.4) 3/19 (15.8) 9/15 (60) 16/26 (61.5)T-tau/Ab421.23 andP-tau/Ab420.21154/60 (90) 0/17 1/27 (3.7) 3/19 (15.8) 9/15 (60) 16/26 (61.5)The cut-offs used were those found from receiver operating characteristics curves for distinguishing Alzheimers disease from frontotemporal dementia pooled with sema
46、ntic dementia.J Neurol Neurosurg Psychiatry 2011;82:240e246. doi:10.1136/jnnp.2010.207183 245Research on November 13, 2015 - Published by http:/ from REFERENCES1. Blennow K, Hampel H. CSF markers for incipient Alzheimers disease. Lancet Neurol2003;2:605e13.2. Hulstaert F, Blennow K, Ivanoiu A, et a
47、l. Improved discrimination of AD patientsusing beta-amyloid(1e42) and tau levels in CSF. Neurology 1999;52:1555e62.3. Ewers M, Buerger K, Teipel SJ, et al. Multicenter assessment of CSF-phosphorylated tau for the prediction of conversion of MCI. Neurology2007;69:2205e12.4. Li G, Sokal I, Quinn JF, e
48、t al. CSF tau/Abeta42 ratio for increased risk of mildcognitive impairment: a follow-up study. Neurology 2007;69:631e9.5. Mattsson N, Zetterberg H, Hansson O, et al. CSF biomarkers and incipientAlzheimer disease in patients with mild cognitive impairment. JAMA2009;302:385e93.6. Brys M, Pirraglia E,
49、Rich K, et al. Prediction and longitudinal study of CSF biomarkersin mild cognitive impairment. Neurobiol Aging 2009;30:682e90.7. Visser PJ, Verhey F, Knol DL, et al. Prevalence and prognostic value of CSF markersof Alzheimers disease pathology in patients with subjective cognitive impairment ormild cognitive impairment in the DESCRIPA study: a prospective cohort study. LancetNeurol 2009;8:619e27.8. Hansson O, Zetterberg H, Buchhave P, et al. Association between