CD45 isoform alteration in CD4+ T cells as a potential diagnostic marker of Alzheimer's disease

Journal of Neuroimmunology 132 (2002) 164 – 172 www.elsevier.com/locate/jneuroim

CD45 isoform alteration in CD4+ T cells as a potential diagnostic marker of Alzheimer’s disease Jun Tan a,*,1, Terrence Town a,1, Laila Abdullah a, Yajaun Wu a, Andon Placzek a, Brent Small a, Jodi Kroeger b, Fiona Crawford a, Dan Richards a, Michael Mullan a b

a Department of Psychiatry, Roskamp Institute, University of South Florida, 3515 E. Fletcher Ave., Tampa, FL 33613, USA H. Lee Moffitt Cancer Center and Research Institute, University of South Florida, 12902 Magnolia Dr., Tampa, FL 33612-9497, USA

Received 6 March 2002; received in revised form 15 August 2002; accepted 15 August 2002

Abstract Aging represents the greatest risk for development of Alzheimer’s disease (AD), and changes in peripheral immune cell phenotypes have been found to be associated with aging. Using flow cytometry, we measured the relative expression levels of CD45 isoforms, a marker of naı¨ve versus memory CD4+ T cell status, on isolated CD4+ T lymphocytes from patients with a clinical diagnosis of probable Alzheimer’s disease, normal elderly, cognitively abnormal elderly, and patients with clinically diagnosed other forms of dementia. Data show significantly lower levels of CD45RA, and an increase in the CD45RO/CD45RA ratio, on CD4+ T cells in patients diagnosed with probable Alzheimer’s disease (n = 46) and in cognitively abnormal individuals (n = 37) compared to age-matched normal participants (n = 90). Patients diagnosed with other forms of dementia (n = 19) did not significantly differ from normal individuals. Both CD45RA and the CD45RO/CD45RA ratio had higher positive and negative predictive values and were more sensitive biomarkers of probable AD than the apolipoprotein E q4 allele, and had greater predictive ability for probable AD by regression analyses. Additionally, a testing strategy employing apolipoprotein E genotyping and CD45RA or the CD45RO/CD45RA ratio revealed increased sensitivity, positive and negative predictive values, and predictive ability over the apolipoprotein E q4 allele. These data show altered peripheral immunity in AD patients, and raise the possibility that a testing strategy using CD45 isoform alteration on CD4+ T cells and apolipoprotein E genotype may be clinically valuable for diagnosing probable AD. D 2002 Elsevier Science B.V. All rights reserved. Keywords: T lymphocyte; Diagnostic marker; Apolipoprotein E

1. Introduction Alzheimer’s disease (AD) is the most commonly reported dementing illness, producing a progressive loss of memory and other higher cognitive functions, eventually leading to death. For patients presenting with AD, a definitive diagnosis of the disease can only be obtained through pathological observation of both senile plaques and neurofibrillary tangles (McKhann et al., 1984; Khachaturian, 1985). Current methods of early diagnosis rely primarily on direct patient assessment and interviews with family members, often supplemented by neuroimaging studies. However, at specialist centers, evaluation for AD (clinical *

Corresponding author. Tel.: +1-813-974-3722; fax: +1-813-974-3915. E-mail address: [email protected] (J. Tan). 1 Both authors contributed equally to this work.

neuropsychometric and neuroradiological workups) takes several hours of patient and clinician time, which a quick and easy diagnostic test would dramatically reduce. Additionally, an early biological marker of AD would be useful to allow early administration of treatments aimed at slowing disease progression or delaying full disease onset, thereby optimizing therapeutic benefit (Brookmeyer and Zeger, 1996; Brookmeyer et al., 1998). With this in mind, we sought to identify a biomarker of AD that would allow for a rapid, simple, inexpensive, and minimally invasive substitute for the standard diagnostic assessment. Of the tests currently commercially available for diagnosis of AD, some (the ADmark tau/Ah1 – 42 and Nymox AD7C-NTP tests) are CSF-based tests and, therefore, are somewhat invasive (Mayeux, 1998). A test that combines measurement of CSF tau and Ah1 – 42 and utilizes cutoff values that best separate AD cases from others, has sensi-

0165-5728/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 5 - 5 7 2 8 ( 0 2 ) 0 0 3 0 9 - 0

J. Tan et al. / Journal of Neuroimmunology 132 (2002) 164–172

tivity values ranging between 50% and 85% and specificity values between 86% and 96% (Motter et al., 1995; Shoji et al., 1998; Tapiola et al., 2000; Hulstaert et al., 1999). However, the combined tau/Ah1 42 test has previously been criticized for its high false negative rate (which is due to AD cases with low CSF tau levels), and no diagnostic value when both values are either high or low (approximately 1/3 of subjects) (Ghanbari and Ghanbari, 1998). The AD7C-NTP biomarker test (Nymox) is based on raised neural thread protein level in AD brain and is a CSF test (although a large clinical study for the efficacy of measurement in urine is underway) (Ghanbari and Ghanbari, 1998; Ghanbari et al., 1998). As a CSF-based test, the sensitivity and specificity are high (89%) for AD compared to agematched, non-demented controls. Both the ADmark tau/ Ah1 – 42 and the Nymox AD7C-NTP tests were developed in known cases of dementia versus non-demented controls. The use of a predictive test for cognitive impairment would also be useful, as these subjects frequently progress to AD (Petersen, 2000). The apolipoprotein E gene (APOE) has consistently been found to be associated with both sporadic and familial AD (Saunders et al., 1993; Poirier et al., 1993). However, the use of APOE genotyping in the diagnosis of AD is limited due to the non-Mendelian relationship between APOE genotype and AD. In a large autopsy-based study, the use of APOE genotyping alone (presence or absence of an q4 allele) produced a sensitivity of 65% and a specificity of 68% (Mayeux et al., 1998). In the same study, the sequential use of clinical diagnosis and APOE genotyping reduced the false positive rate of the clinical diagnosis, increasing specificity from 55% to 84% and modestly increasing the positive predictive value from 90% to 94% (Mayeux, 1998; Mayeux et al., 1998). Thus, while the use of APOE alone as a predictive test in not recommended (Roses, 1995), a testing strategy that combines APOE and another biomarker may well be beneficial. Aging remains the greatest risk factor for AD, and changes in peripheral immune cell phenotype associated with aging have been documented in previous studies. For example, an increase in the ratio of memory to naı¨ve CD4+ T cells (indicated by an increase in the CD45RO/CD45RA ratio) has been shown to correlate with aging (Flurkey et al., 1992; Utsuyama et al., 1992). Similar shifts in the numbers, and naı¨ve versus memory status of circulating T cell subpopulations have been postulated to account for a dysfunctional immune response observed in the elderly and even more so in AD patients (Ikeda et al., 1991; Hu et al., 1995). Given the evidence for alterations in T cell phenotype and potentially dysfunctional peripheral immunity in AD patients, we investigated the measurement of differential CD45 isoform expression (CD45RA, CD45RO, and CD45RB) in patients with a clinical diagnosis of probable AD, normal elderly, cognitively abnormal elderly, and patients with clinically diagnosed other forms of dementia. We performed Western analysis on whole leukocyte lysates

165

from these study subjects, and we found evidence for altered CD45 isoform expression in AD patients and cognitively abnormal elderly. Using flow cytometry, we then measured the relative expression levels of CD45 isoforms on isolated CD4+ T lymphocytes from these study subjects in order to evaluate the potential usefulness of CD45 isotyping of CD4+ T lymphocytes as a clinical diagnostic test for AD. Finally, we critically evaluated this biomarker by determining its predictive ability by multiple hierarchical logistic regression analyses and by assessing sensitivity, specificity, positive predictive value, negative predictive value.

2. Materials and methods 2.1. Subject recruitment and neurological and psychological evaluation The sample was comprised of 192 individuals (90 cognitively normal, 37 cognitively abnormal, 46 diagnosed with probable AD, and 19 with other dementias) originally from 31 different states who were referred to the University of South Florida Memory Disorder Clinic or who were screened for cognitive disorder in the community by clinic staff. Those screened in the community were elderly volunteers who agreed to complete a Mini Mental State Examination (MMSE) questionnaire (Folstein et al., 1975) and donate a blood sample. Of those screened in the community, 90 subjects scored 27 or greater on the MMSE and were thus considered cognitively normal (Normal, 40 male, mean age = 75.54 F 7.52 S.D., MMSE = 29.0 F 1.12 S.D.). During cognitive screening, 37 participants scored V 26 on the MMSE and were thus considered cognitively abnormal (Cogn. Abn., 19 male, mean age = 77.29 F 10.59 S.D., MMSE = 20.6 F 6.39 S.D.). For those seen in the community screen, the presence of obvious depressive symptomatology was either ruled out during face-to-face clinical interview or by scores of less than 14 on a 17-item Hamilton Depression Scale. Those seen and diagnosed with dementia in the clinic underwent extensive evaluation before a consensus diagnosis (involving all evaluating professionals) was made. Assessment generally included full medical and neurocognitive evaluation by a physician specialist in dementia, neuropsychological, and neuroradiological (magnetic resonance imaging analysis) assessment, with routine blood tests. Full details of this evaluation are available upon request. For this study, 46 cases were diagnosed as having probable AD (AD Prob., 10 male, mean age = 75.47 F 6.81 S.D., MMSE = 16.8 F 5.5 S.D.) according to NINCDS-ADRDA criteria. In addition, patients with other dementia [Other Dem.; including frontotemporal dementia (n = 2 females, diagnosed in accordance with consensus criteria, Neary et al., 1998), dementia with Lewy bodies (n = 2 males and 1 female diagnosed in accordance with the consensus guidelines from the DLB International Workshop, McKeith et al.,

166

J. Tan et al. / Journal of Neuroimmunology 132 (2002) 164–172

1996), and patients clinically diagnosed with stroke (n = 10 male and 4 female, these individuals had an MMSE score of V 22, impairment in more than two cognitive domains, and impaired independent activities of daily living); total n = 19, 12 male, mean age = 72.53 F 10.45 S.D., MMSE = 18.9 F 6.6 S.D.] were included as a non-AD demented sample. Thus, the Normal and Cogn. Abn. groups were drawn from a community-based sample, while the AD Prob. and Other Dem. groups were sampled from the clinic setting. None of the subjects exhibited symptoms of systemic infection at the time of blood collection. Controls and AD Prob. patients were all z 60 years of age, and each group was not significantly different for age (by one-way ANOVA, p = 0.296). The Normal, Cogn. Abn., and Other Dem. groups were not significantly different in gender ( p = 0.153), while AD Prob. patients significantly differed from the other groups in gender ( p < 0.01). However, no significant correlation was found between gender and any of the CD45 isoforms investigated (CD45RA, p = 0.345; RB, p = 0.389; RO, p = 0.450; the CD45RO/RA ratio, p = 0.589), suggesting that this difference did not confound our analyses. 2.2. Western immunoblotting Leukocytes were isolated from whole, sodium heparintreated blood samples (10 ml) using RBC lysis solution. Immediately following isolation, leukocytes were washed in ice-cold phosphate-buffered saline (PBS) three times and lysed in an ice-cold lysis buffer containing 20 mM Tris– HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA, 10% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM h-glycerophosphate, 1 mM Na3VO4, 1 Ag/ml leupeptin, and 1 mM PMSF. After incubating for 30 min on ice, samples were centrifuged at high speed for 15 min, and supernatants were collected. Total protein content was estimated using the Bio-Rad protein assay (Bio-Rad, Hercules, CA). An aliquot corresponding to 50 Ag of total protein of each sample was separated by SDSPAGE and transferred electrophoretically to Immune-Blotk PVDF membranes (Bio-Rad). Nonspecific antibody binding was blocked with 5% nonfat dry milk for 1 h at room temperature in TBS (20 mM Tris, 500 mM NaCl, pH 7.5). Immunoblotting was carried out with a primary antibody (1:100 dilution, purified mouse anti-human anti-CD45, antiCD45RA, or anti-CD45RO monoclonal antibodies, Santa Cruz Biotechnology, Santa Cruz, CA) followed by an antimouse alkaline phosphatase-conjugated IgG secondary antibody (1:1500 dilution, Santa Cruz Biotechnology) as a tracer. The Immun-StarR chemiluminescent substrate (BioRad) was used to develop the blots. 2.3. Flow cytometry All PE- and FITC-conjugated monoclonal mouse antihuman antibodies (mAbs, mouse anti-CD45RA, antiCD45RB, anti-CD45RO, anti-CD4, and IgG control antibodies) were purchased from PharMingen (Los Angeles,

CA). Twenty-four hours after collection, whole, sodium heparin-treated blood samples (200 Al) were double-labeled with FITC-conjugated anti-CD4 and with PE-conjugated anti-CD45RA, anti-CD45RB, or anti-CD45RO antibodies (1:50 dilution for each). Aliquots of each sample were also double-labeled with FITC-conjugated and PE-conjugated mouse IgG (isotype-matched) control antibodies (1:50 dilution for each). Erythrocytes were then lysed in a red blood cell lysis solution (Gentra Systems, Minneapolis, MN), and samples were centrifuged and cell pellets suspended in 300 Al of a 10% formalin solution (Sigma) for 5 min. Cells were washed three times in PBS, then resuspended in 250 Al of flow buffer (PBS containing 0.1% sodium azide and 1.0% fetal bovine serum) for flow cytometric analysis, according to our previously described methods (Tan et al., 1999). CD4+ T cells (10,000) were collected and analyzed by flow cytometry for each study participant to exclude the possibility that observed differences on CD45 isoform expression may be due to different numbers of CD4+ T cells analyzed. Scatterplots and percentages of fluorescent events were generated using CellQuest flow cytometric analysis software (Becton Dickinson, San Jose, CA). Using scatterplots, cells were gated based on morphological characteristics such that T lymphocytes were selected. Data are represented as the percentage of cells double-positive for CD4 and the particular CD45 isoform minus the percentage of cells double-positive for the isotype-matched control antibodies. To ensure a valid flow cytometric analysis and to protect against artifacts, additional steps were taken. To test the specificity of the anti-CD45RA and anti-CD45RO antibodies, we performed flow cytometric analyses on cultures of human Jurkat T cells or CD45-deficient Jurkat T cells. Results showed that f 35% of Jurkat T cells stained positive for CD45RA, and >50% of these cells stained for CD45RO. Further, neither CD45RA nor CD45RO were detected on CD45-deficient Jurkat T cells. Titre curves were generated for each antibody from 1:50 to 1:400, and a 1:200 dilution was generally found to give the highest signal to noise ratio, thereby allowing the greatest discriminative ability between positive and negative cells. Further, antibody aggregates were minimized by storing all antibodies at 4 jC and discarding antibodies after 6 months of storage, and cell death in sample preparations was minimized by keeping cells at 4 jC throughout the staining procedure and by supplementing flow buffer with 1.0% fetal bovine serum as described above. We tested two different methods of flow cytometric analysis on data obtained from our study cohort. The first method was to gate on all lymphocytes by forward and side-scatter criteria, and to then generate a four-quadrant dot-plot. The upper-right quadrant percentage was then divided by the sum of the upper-right and lower-right quadrants, and subtracted from the figure obtained with the isotype-matched control antibodies, yielding percentage of CD4+ lymphocytes double-positive for the CD45 isoform under consideration. In the second method, we gated on CD4+ lymphocytes, and then simply generated histograms

J. Tan et al. / Journal of Neuroimmunology 132 (2002) 164–172

to determine percentage shift for CD45 isoforms by subtracting peak shifts from any shifts observed with the isotype-matched control antibody, again yielding percentage of CD4 and CD45 isoform double-positive cells. When statistically comparing these methods, no significant differences were found ( p>0.05), and we therefore adopted the more conventional four-quadrant method. 2.4. Statistical analysis For all quantitative data, the one-sample Kolmogorov – Smirnov test was used to test if data were normally or nonnormally distributed. If data were normally distributed, one-way analysis of variance (ANOVA) followed by post hoc comparison using either Bonferroni’s or Dunnett’s T3 method (where Levene’s test for homogeneity of the variance was used to determine the method of post hoc analysis) was used to analyze the data. In the case of nonnormally distributed data, the Kruskal – Wallis H test followed by post hoc comparison using the Mann – Whitney U-test was used to perform the analysis. Between-groups comparisons on gender were performed using the likelihood ratio v2 statistic, and comparisons on study subject ages were performed using one-way ANOVA. Bivariate correlations were determined by first rank-ordering data according to Spearman’s method, and then determining Pearson’s r with associated significance. Sensitivity (percentage of the AD Prob. population positive for the test) and specificity (percentage of the non-AD Prob. population negative for the test) for APOE q4 were calculated based on 2  2 contingency tables, and

167

Receiver Operator Characteristic (ROC) curve analysis was performed to determine these values for CD45 isoforms, both alone or when sequentially tested following APOE q4 genotyping as previously described (Mayeux et al., 1998). For ROC analyses, cutoff scores, sensitivity, and specificity for CD45RA levels or the CD45RO/RA ratio on CD4+ T cells were determined based on the area under the curve. Positive predictive value (probability that an individual is actually diagnosed as AD Prob. when the test is positive) and negative predictive value (probability that a patient is truly not diagnosed as AD Prob. when the test is negative) were calculated as previously described (Saunders, 1996). Multiple hierarchical logistic regression models controlling for gender were used to assess predictive ability of APOE q4 and CD45 status (CD45RA or the CD45RO/RA ratio) when modeled alone or together. R2 values were calculated according to Nagelkerke’s method, and likelihood ratio testing was used to make between-models comparisons. For all analyses, a levels were set at 0.05, and SPSS software release 10.0.5 was used.

3. Results 3.1. Reduced CD45RA expression and elevated CD45RO/ RA ratio in leukocytes from probable AD and cognitively abnormal individuals In order to determine whether expression levels of CD45RO and/or CD45RA might be altered in AD patients compared to controls, we initially performed Western immu-

Fig. 1. Differential expression of CD45 isoforms in leukocytes from AD patients and cognitively abnormal participants compared to controls. (A) Representative Western blot indicating a decrease in CD45RA expression in both probable AD cases (AD Prob.) and cognitively abnormal (Cogn. Abn.) subjects compared to normal individuals (Normal) or other dementia (Other Dem.) patients. Total CD45 was used as an internal reference for semi-quantitative measurement of CD45 isoform expression by densitometric analysis shown in B and C. (B) Graph showing the band density ratio of CD45RA to total CD45 (mean F 1 S.E.M., n = 10 for each group). (C) Graph showing the band density ratio of CD45RO/CD45RA (mean F 1 S.E.M., n = 10 for each group). For B and C, ANOVA revealed significant between-groups differences ( p < 0.001), and post hoc comparison showed significant differences between Normal and either AD Prob. or Cogn. Abn. individuals ( p < 0.001). No significant differences were observed between normal subjects and Other Dem. patients ( p>0.05).

168

J. Tan et al. / Journal of Neuroimmunology 132 (2002) 164–172

noblotting on total peripheral leukocytes isolated from a subset of our study subjects (10 subjects randomly selected from each group) (Fig. 1A). There was a statistically significant ( p < 0.001) decrease in CD45RA in AD Prob. cases and Cogn. Abn. individuals as compared to normal subjects (Fig. 1B). However, no significant difference was noted between Other Dem. subjects and Normals. When

considering the CD45RO/CD45RA ratio in peripheral leukocytes, data showed a statistically significant ( p < 0.001) elevation of this ratio in both AD Prob. patients and Cogn. Abn. individuals as compared to Normals, who did not significantly differ from Other Dem. patients (Fig. 1C). These data led us to specifically evaluate expression levels of CD45 isoforms on CD4+ T cells, the cell type where

Fig. 2. CD4+ T cells from probable AD patients and cognitively abnormal participants have altered CD45RA isoform expression profiles compared to controls. (A) FACS dot-plots of representative samples indicating differential expression of CD45 isoforms on CD4+ T cells (top row, CD45RA; bottom row, CD45RO) from normal subjects (a, e), Cogn. Abn. participants (b, f), AD Prob. patients (c, g) and Other Dem. patients (d, h). The x-axis (FL1-H) represents log fluorescence intensity of the FITC-conjugated anti-CD4 mAb, and the y-axis (FL2-H) represents log fluorescence intensity of the PE-conjugated anti-CD45 mAb. The upper right-hand portion of the plot indicates % of double-positive cells for CD4 and CD45RA or CD45RO. (B) Scatterplot showing percentage of CD45RA-positive CD4+ T cells. The means are represented as a solid bar. (C) Graph summarizing expression levels (%) of CD45RA and CD45RO (mean F 1 S.E.M.; n = 90 for Normals; n = 37 for Cogn. Abn.; n = 46 for AD Prob.; n = 19 for Other Dem.). For CD45RA, Kruskal – Wallis H testing revealed significant between-groups differences ( p < 0.001), and post hoc comparison revealed significant differences between Normal participants and either AD Prob. patients or Cogn. Abn. individuals ( p < 0.001), but not between Other Dem. subjects and Normals ( p>0.05). No significant between-groups differences were observed on CD45RO status ( p>0.05).

J. Tan et al. / Journal of Neuroimmunology 132 (2002) 164–172

169

Fig. 3. CD4+ T cells from probable AD patients and cognitively abnormal participants have an altered ratio of CD45RO/CD45RA compared to controls. (A) Scatterplot showing the ratio of CD45RO/CD45RA on CD4+ T cells. The means are represented as a solid bar. (B) Graph summarizing the ratio of CD45RO/ CD45RA expression levels shown in A (mean F 1 S.E.M.; n = 90 for Normals; n = 37 for Cogn. Abn.; n = 46 for AD Prob.; n = 19 for Other Dem.). Kruskal – Wallis H testing revealed significant between-groups differences ( p = 0.001), and post hoc comparison revealed significant differences between Normal participants and either AD Prob. patients or Cogn. Abn. individuals ( p < 0.01). No significant difference was observed between Other Dem. subjects and Normals ( p>0.05).

differential expression of CD45 isoforms has been shown to be a marker of naı¨ve versus previously activated status (Ikeda et al., 1991; Flurkey et al., 1992; Utsuyama et al., 1992; Hu et al., 1995). 3.2. CD45RA is decreased and the CD45RO/RA ratio is increased on CD4+ T cells from probable AD patients and cognitively abnormal individuals We examined CD4+ T cell subsets in Normals, Cogn. Abn. individuals, AD Prob. cases, and Other Dem. patients by flow cytometry (Fig. 2A). Fig. 2B shows the distribution of the percentage of CD45RA-positive CD4+ T cells, and, as shown in Fig. 2C, percentages of CD45RA-expressing CD4+ T cells were significantly ( p < 0.001) decreased in AD Prob. and Cogn. Abn. subjects compared to Normal individuals. Also, as summarized in Fig. 2C, no significant differences were detected between any of the subject groups on CD45RO-expressing CD4+ T cells. When considering the ratio of CD45RO/CD45RA expression on CD4+ T cells, this ratio was significantly ( p < 0.01) increased in both AD Prob. and Cogn. Abn. patients compared to Normals, suggesting an increase in the CD4+ T cell memory phenotype in these individuals (scatterplot, Fig. 3A; summary data, Fig. 3B). When considering CD45RA or the CD45RO/ CD45RA ratio on CD4+ T cells, no significant differences were observed between Other Dem. subjects and Normal participants. Additionally, when examining CD45RB isoform expression on CD4+ T cells by this method, no significant differences were found between subject groups ( p>0.05, data not shown). A previous study found that

CD45R isoforms were decreased on CD4+ T cells in patients diagnosed with probable AD, although those authors were not able to identify the specific isoforms responsible for this effect (Ikeda et al., 1991). Our data now show that CD45RA, but not CD45RO or CD45RB, is specifically decreased on CD4+ T cells in AD Prob. patients and Cogn. Abn. individuals. 3.3. Evaluation of CD45 isoform alteration on CD4+ T cells as a biomarker for probable AD For a biomarker of AD to be useful, it must be sensitive, specific, and have positive predictive value (PPV) and negative predictive value (NPV) (Saunders et al., 1993; Mayeux et al., 1998). Thus, we wished to evaluate CD45RA Table 1 Validity of APOE q4, CD45 isoforms, and the sequential test APOE q4 CD45RA APOE q4 + CD45RA CD45RO/RA APOE q4 + CD45RO/RA

Cutoff

Sensitivity Specificity

PPV

NPV

z q4 68.14 63.90 1.21 1.22

0.51 0.76 0.82 0.68 0.74

0.46 0.55 0.80 0.54 0.68

0.75 0.86 0.79 0.81 0.78

0.71 0.70 0.73 0.61 0.69

Sensitivity, specificity, PPV, and NPV were calculated as described in Materials and methods, and the sequential test refers to APOE genotyping followed by CD45 isoform expression level testing. Values were obtained by comparing AD Prob. patients to all other subjects but Cogn. Abn. individuals; however, similar figures were obtained when grouping AD Prob. and Cogn. Abn. subjects together. Note the increased sensitivity, PPV, and NPV of CD45RA or the CD45RO/RA ratio, alone or when sequentially tested, compared to APOE q4 alone.

170

J. Tan et al. / Journal of Neuroimmunology 132 (2002) 164–172

or the CD45RO/CD45RA ratio using these criteria. In addition, we genotyped our study subjects for APOE q4, a known diagnostic marker of AD (Roses, 1995; Mayeux, 1998; Mayeux et al., 1998) according to previously published methods (Wenham et al., 1991). This enabled us to compare sensitivity, specificity, PPV, and NPV for CD45RA or the CD45RO/CD45RA ratio to APOE q4 and to determine if testing for APOE q4 followed by CD45RA or the CD45RO/CD45RA ratio might increase these measures over APOE q4 alone. The sensitivity, specificity, PPV, and NPV values that we obtained for APOE q4 on prediction of AD were similar to previously published data (Roses, 1995; Saunders, 1996; Mayeux, 1998; Mayeux et al., 1998). As shown in Table 1, both CD45RA and the CD45RO/CD45RA ratio demonstrated sensitivity, PPV and NPV values that were increased compared to APOE q4, with CD45RA measurement showing the largest increases. Additionally, testing for APOE q4 followed by CD45RA or the CD45RO/ CD45RA ratio increased sensitivity, PPV, and NPV over that of APOE q4 alone while maintaining specificity comparable to that of APOE q4, showing that these measures are providing additional information in the prediction of AD. Increases in these values were most pronounced when considering APOE q4 together with CD45RA. To further confirm that CD45RA or the CD45RO/ CD45RA ratio are valuable biomarkers of probable AD, multiple hierarchical logistic regression models were employed. As shown in Table 2, CD45RA and the CD45RO/CD45RA ratio were better predictors of disease

Table 2 Predictive ability of APOE q4 and CD45 isoforms Model/description

Class prediction (%)

R2

df

0: 1: 2: 3:

– 68.4 75.5 75.2

– 0.139 0.277 0.304

– 2 2 3

74.2

– 0.175

1 2

75.2

0.218

3

< 0.001

–

1

*< 0.015 – 1

h0 h0 + APOE q4 h1 h0 + CD45RA h1 h0 + APOE q4 h1 + CD45RA h2

4: h0 + CD45RO/ RA h1 5: h0 + APOE q4 h1 + CD45RO/RA h2

p

– 0.001 < 0.001 < 0.001 *< 0.00013 – 1 < 0.001

Multiple hierarchical logistic regression models were carried out as described in Materials and methods, and model summary data are displayed. Values were obtained by comparing AD Prob. patients to all other subjects but Cogn. Abn. individuals; however, similar figures were obtained when grouping AD Prob. and Cogn. Abn. subjects together. Class predictions and R2 values for CD45RA or the CD45RO/CD45RA ratio when modeled alone compared to APOE q4 alone are increased, suggesting that CD45 isoforms are a better predictor of disease than APOE q4. Likelihood ratio testing between models shows that CD45RA or the CD45RO/RA ratio, when modeled together with APOE q4, add significantly more predictive information than APOE q4 when modeled alone, as indicated by the significant p values shown. *x – y denotes comparison of model x to model y by likelihood ratio testing.

than APOE q4, with CD45RA measurement showing the greatest predictive ability, as indicated by increased class prediction and R2 values compared to APOE q4. Additionally, when CD45RA or the CD45RO/CD45RA ratio were modeled together with APOE q4, class prediction and R2 values were increased compared to APOE q4 alone, and likelihood ratio testing between models indicated that CD45RA or the CD45RO/CD45RA ratio in combination with APOE q4 was significantly adding more predictive information for disease than APOE q4 alone. Again, this effect was most pronounced when considering the APOE q4 and CD45RA combination.

4. Discussion Changes in peripheral immune cell phenotypes, including increases in the ratio of circulating memory versus naı¨ve T cells, and changes in T cell numbers, have been found with age and even more so in AD (Ikeda et al., 1991; Flurkey et al., 1992; Utsuyama et al., 1992; Hu et al., 1995). These data provide evidence for a dysfunctional peripheral immune response in AD patients, and we sought to evaluate whether CD45 isoform alteration on CD4+ T cells might be a valuable simple and minimally invasive biomarker for the disease. Results showed that CD4+ T cell CD45 isoform alteration is a useful biomarker of clinically diagnosed AD, and allows for the identification of individuals who are cognitively abnormal, many of whom later progress to AD (Petersen, 2000). In particular, a single measurement of the CD45RA isoform on CD4+ T cells, as opposed to flow analysis of both CD45RO and CD45RA on CD4+ T cells, appears to be adequate in terms of sensitivity, specificity, NPV, and PPV. This may reflect the fact that expression of CD45RA versus CD45RO is generally regarded as mutually exclusive, as it is the alternate splicing of the identical mRNA molecule that gives rise to the different isoforms after initial T cell activation. Finally, combination of CD45 isoform alteration with APOE q4 genotyping resulted in additional predictive value, thereby increasing the utility of CD45 isoforms as a biomarker for AD and cognitive abnormality. Previous studies have shown that the CD45RO/CD45RA ratio on CD4+ T cells increases with aging, indicating greater numbers of memory T cells in the elderly (Flurkey et al., 1992; Utsuyama et al., 1992). Our data are in accordance with this, as we find a modest but statistically significant positive correlation of the CD45RO/CD45RA ratio on CD4+ T cells with age (r = 0.170, p < 0.05). As we found that CD45 isoforms were markers of probable AD and cognitive abnormality, we further assessed a possible correlation between CD45 isoforms and Mini-Mental State Examination score (Folstein et al., 1975), and found a significant negative correlation when including all subjects between the CD45RO/CD45RA ratio on CD4+ T cells and MMSE score (r = 0.222, p < 0.01), as well as a significant positive correlation between CD45RA on CD4+ T cells and

J. Tan et al. / Journal of Neuroimmunology 132 (2002) 164–172

MMSE score (r = 0.313, p < 0.001). Mild cognitive impairment has been suggested to be an early stage of AD (Petersen, 2000), and our finding that differential CD45 isoform expression is a predictor of cognitive abnormality raises the possibility that CD45 isoform alteration CD4+ T lymphocytes may be a biomarker for cognitive abnormality prior to diagnosis of AD. Additionally, as 95% of patients who are clinically diagnosed with probable AD and who are subsequently brought to autopsy have pathologically confirmed AD (Price et al., 1993), our results raise the possibility that differential CD45 isoform expression may be a valuable marker of postmortem-confirmed AD. It is becoming increasingly substantiated that there is a brain immune/inflammatory component in the pathogenesis of AD, which is characterized by activation of microglia and astrocytes resulting in secretion of a gamut of proinflammatory cytokines and acute-phase reactants (for a review, see Selkoe, 2001; Wyss-Coray and Mucke, 2002). However, there is much less evidence for a peripheral immune cell response in AD patients. While the data presented in this manuscript provide such evidence, the mechanism underlying differential CD45 isoform expression on CD4+ T cells in AD patients remains unclear. For example, it is possible that, during the course of the disease, a soluble factor is released from the dysfunctional AD blood – brain barrier (Kalaria, 1997; Skoog et al., 1998), and that this factor interacts with circulating T cells, producing a variation in CD45 expression. It is also possible that cell surface markers expressed on endothelial cells lining cerebral vessels in AD brain interact with passing T cells, a phenomenon referred to as T cell rolling (Hunt et al., 1996), resulting in differential CD45 isoform expression on these cells. Whereas these explanations rely on immune cross-talk between the CNS and the periphery, it also remains possible that the alterations observed here on CD45 isoform status are due to an independent compartmentalized peripheral immune response. Additionally, CD45 isoform alteration on CD4+ T cells in AD patients may be consequent upon other changes that co-occur with AD, such as infection, weight loss, poor nutritional status, and decreased energy expenditure (Poehlman and Dvorak, 2000). Finally, it should be noted that alterations in this biomarker may not be regarded as specific to AD, as there are a great number of conditions (such as infection and autoimmune disorder) that could alter CD45 isoform expression levels on CD4+ T cells. Further study into non-AD individuals, AD patients, and cognitively abnormal individuals with and without infection or autoimmune disorder is necessary to determine to what extent these conditions could impact the usefulness of this biomarker. It should be noted that, while this study did not aim to address etiological explanations for why CD45 isoform alteration on CD4+ T cells is associated with AD and cognitive abnormality, additional studies that are designed to identify the causes and molecular mechanisms of these variations may provide indirect information about the etiol-

171

ogy of the disease and aid in explaining why these changes appear to be specific to AD. In particular, we would like to examine the correlation between CD45 isoform alteration and neuropathologic or clinico-neuropathologic diagnosis of AD, to explore the relationship between aberrant CD45 isoform alteration in cognitively abnormal individuals and later diagnosis of AD, and to perform a prospective study in normal aging individuals to determine whether CD45 isoform alteration is a predictor of subsequent cognitive impairment and AD. Such studies will allow for determination of whether CD45 isoforms may be etiologically involved in dementia, markers of some other process that causes dementia, or are an epiphenomenon related to some other condition co-morbid with dementia such as disability, nutrition, or infection. Additionally, these future studies will address whether CD45 isoform testing will be a good biomarker for dementia in patients before they become demented, during the preclinical phases of dementia, or in patients with mild cognitive impairment (where diagnosis is most difficult). Finally, recent data have indicated that use of additional markers of naı¨ve versus memory CD4+ T cells (such as CD11a, CD62L, or CD27) may provide for a better discrimination between these subsets (Roederer and Hardy, 2001), and future studies should go on to address whether or not the addition of such markers might improve the utility of CD45 isotyping on CD4+ T cells for diagnosis of cognitive abnormality and AD.

Acknowledgements The authors are grateful to Mr. and Mrs. Robert Roskamp for their generous support, which helped to make this work possible.

References Brookmeyer, R., Zeger, S., 1996. Statistical issues in prevention and therapeutic trials of Alzheimer’s disease. Alzheimer Dis. Assoc. Disord. 10, 27 – 30. Brookmeyer, R., Gray, S., Kawas, C., 1998. Projection of Alzheimer’s disease in the United States and the public health impact of delaying disease onset. Am. J. Public Health 88, 1337 – 1342. Flurkey, K., Stadecker, M., Miller, R.A., 1992. Memory T lymphocyte hyporesponsiveness to non-cognate stimuli: a key factor in age-related immunodeficiency. Eur. J. Immunol. 22, 931 – 935. Folstein, M.F., Folstein, S.E., McHugh, P.R., 1975. ‘‘Mini-mental state’’. A practical method for grading the cognitive state of patients for the clinician. J. Psychiatr. Res. 12, 189 – 198. Ghanbari, K., Ghanbari, H.A., 1998. A sandwich enzyme immunoassay for measuring AD7C-NTP as an Alzheimer’s disease marker: AD7C test. J. Clin. Lab. Anal. 6, 379 – 383. Ghanbari, H., Ghanbari, K., Beheshti, I., 1998. Biochemical assay for AD7C-NTP in urine as an Alzheimer’s disease marker. J. Clin. Lab. Anal. 12, 285 – 288. Hu, G.R., Walls, R.S., Creasey, H., McCusker, E., Broe, G.A., 1995. Peripheral blood lymphocyte subset distribution and function in patients with Alzheimer’s disease and other dementias. Aust. N.Z. J. Med. 25, 212 – 217. Hulstaert, F., Blennow, K., Ivanoiu, A., Schoonderwaldt, H.C., Riemensch-

172

J. Tan et al. / Journal of Neuroimmunology 132 (2002) 164–172

neider, M., De Deyn, P.P., Bancher, C., Cras, P., Wiltfang, J., Mehta, P.D., Iqbal, K., Pottel, H., Vanmechelen, E., Vanderstichele, H., 1999. Improved discrimination of AD patients using beta-amyloid (1 – 42) and tau levels in CSF. Neurology 52, 1555 – 1562. Hunt III, S.W., Harris, E.S., Kellermann, S.A., Shimizu, Y. 1996. T-lymphocyte interactions with endothelium and extracellular matrix. Crit. Rev. Oral Biol. Med. 7, 59 – 86. Ikeda, T., Yamamoto, K., Takahashi, K., Yamada, M., 1991. Immune system-associated antigens on the surface of peripheral blood lymphocytes in patients with Alzheimer’s disease. Acta Psychiatr. Scand. 83, 444 – 448. Kalaria, R.N., 1997. Cerebrovascular degeneration is related to amyloidbeta protein deposition in Alzheimer’s disease. Ann. N.Y. Acad. Sci. 826, 263 – 271. Khachaturian, Z.S., 1985. Diagnosis of Alzheimer’s disease. Arch. Neurol. 42, 1097 – 1105. Mayeux, R., 1998. Evaluation and use of diagnostic tests in Alzheimer’s disease. Neurobiol. Aging 19, 139 – 143. Mayeux, R., Saunders, A.M., Shea, S., Mirra, S., Evans, D., Roses, A.D., Hyman, B.T., Crain, B., Tang, M.X., Phelps, C.H., 1998. Utility of the apolipoprotein E genotype in the diagnosis of Alzheimer’s disease. Alzheimer’s disease centers consortium on apolipoprotein E and Alzheimer’s disease. N. Engl. J. Med. 338, 506 – 511. McKeith, I.G., Galasko, D., Kosaka, K., Perry, E.K., Dickson, D.W., Hansen, L.A., Salmon, D.P., Lowe, J., Mirra, S.S., Byrne, E.J., Lennox, G., Quinn, N.P., Edwardson, J.A., Ince, P.G., Bergeron, C., Burns, A., Miller, B.L., Lovestone, S., Collerton, D., Jansen, E.N., Ballard, C., de Vos, R.A., Wilcock, G.K., Jellinger, K.A., Perry, R.H., 1996. Consensus guidelines for the clinical and pathologic diagnosis of dementia with Lewy bodies (DLB): report of the consortium on DLB international workshop. Neurology 47, 1113 – 1124. McKhann, G., Drachman, D., Folstein, M., Katzman, R., Price, D., Stadlan, E.M., 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. Motter, R., Vigo-Pelfrey, C., Kholodenko, D., 1995. Reduction of h-amyloid peptide42 in the cerebrospinal fluid of patients with Alzheimer’s disease. Ann. Neurol. 38, 643 – 648. Neary, D., Snowden, J.S., Gustafson, L., Passant, U., Stuss, D., Black, S., Freedman, M., Kertesz, A., Robert, P.H., Albert, M., Boone, K., Miller, B.L., Cummings, J., Benson, D.F., 1998. Frontotemporal lobar degeneration: a consensus on clinical diagnostic criteria. Neurology 51, 546 – 554. Petersen, R.C., 2000. Mild cognitive impairment: transition between aging and Alzheimer’s disease. Neurology 15, 93 – 101. Poehlman, E.T., Dvorak, R.V., 2000. Energy expenditure, energy intake, and weight loss in Alzheimer disease. Am. J. Clin. Nutr. 71, 650S – 655S.

Poirier, J., Davignon, J., Bouthillier, D., Kogan, S., Bertrand, P., Gauthier, S., 1993. Apolipoprotein E polymorphism and Alzheimer’s disease. Lancet 342, 697 – 699. Price, B.H., Gurvit, H., Weintraub, S., Geula, C., Leimkuhler, E., Mesulam, M., 1993. Neuropsychological patterns and language deficits in 20 consecutive cases of autopsy-confirmed Alzheimer’s disease. Arch. Neurol. 50, 931 – 937. Roederer, M., Hardy, R.R., 2001. Frequency difference gating: a multivariate method for identifying subsets that differ between samples. Cytometry 45, 56 – 64. Roses, A.D., 1995. Apoplipoprotein E genotyping in the differential diagnosis, not prediction, of Alzheimer’s disease. Ann. Neurol. 38, 6 – 14. Saunders, A.M., 1996. Specificity, sensitivity, and predictive value of apolipoprotein-E genotyping for sporadic Alzheimer’s disease. Lancet 348, 90 – 93. Saunders, A.M., Strittmatter, W.J., Schmechel, D., George-Hyslop, P.H., Pericak-Vance, M.A., Joo, S.H., Rosi, B.L., Gusella, J.F., Crapper-MacLachlan, D.R., Alberts, M.J., et al., 1993. Association of apolipoprotein E allele epsilon 4 with late-onset familial and sporadic Alzheimer’s disease. Neurology 43, 1467 – 1472. Selkoe, D.J., 2001. Alzheimer’s disease: genes, proteins, and therapy. Physiol. Rev. 81, 741 – 766. Shoji, M., Matsubara, E., Kanai, M., Watanabe, M., Nakamura, T., Tomidokoro, Y., Shizuka, M., Wakabayashi, K., Igeta, Y., Ikeda, Y., Mizushima, K., Amari, M., Ishiguro, K., Kawarabayashi, T., Harigaya, Y., Okamoto, K., Hirai, S., 1998. Combination assay of CSF tau, A beta 1 – 40 and A beta 1 – 42(43) as a biochemical marker of Alzheimer’s disease. J. Neurol. Sci. 158, 134 – 140. Skoog, I., Wallin, A., Fredman, P., Hesse, C., Aevarsson, O., Karlsson, I., Gottfries, C.G., Blennow, K., 1998. A population study on blood – brain barrier function in 85-year-olds: relation to Alzheimer’s disease and vascular dementia. Neurology 50, 966 – 971. Tan, J., Town, T., Paris, D., Placzek, A., Parker, T., Crawford, F., Yu, H., Humphrey, J., Mullan, M., 1999. Activation of microglial cells by the CD40 pathway: relevance to multiple sclerosis. J. Neuroimmunol. 97, 77 – 85. Tapiola, T., Pirttila, T., Mehta, P.D., Alafuzofff, I., Lehtovirta, M., Soininen, H., 2000. Relationship between apoE genotype and CSF betaamyloid (1 – 42) and tau in patients with probable and definite Alzheimer’s disease. Neurobiol. Aging 21, 735 – 740. Utsuyama, M., Hirokawa, K., Kurashima, C., Fukayama, M., Inamatsu, T., Suzuki, K., Hashimoto, W., Sato, K., 1992. Differential age-change in the numbers of CD4 + CD45RA and CD4 + CD29+ T cell subsets in human peripheral blood. Mech. Ageing Dev. 63, 57 – 68. Wenham, P.R., Price, W.H., Blandell, G., 1991. Apolipoprotein E genotyping by one-stage PCR. Lancet 337, 1158 – 1159. Wyss-Coray, T., Mucke, L., 2002. Inflammation in neurodegenerative disease—a double-edged sword. Neuron 35, 419 – 432.