Application of the IWG-2 Diagnostic Criteria for Alzheimer’s Disease to the ADNI

ADNI dataset

The ADNI is a large, multicenter, longitudinal neuroimaging study, launched in 2003 by the National Institute on Aging, the National Institute of Biomedical Imaging and Bioengineering, the Food and Drug Administration, pharmaceutical companies, and nonprofit organizations [22]. The initial goal of ADNI was to recruit 800 subjects but the ADNI has been followed by ADNI-GO and ADNI-2 with additional recruitment. To date, the three protocols have recruited over 1,500 adults, ages 55 to 90, including cognitively normal older individuals (CN), patients with mild cognitive impairment (MCI), and patients with early dementia due to AD. The study was approved by the institutional review boards of all participating centers and written informed consent was obtained from all participants or authorized representatives after extensive description of the ADNI based on the 1975 Declaration of Helsinki.

Subjects

The participants used in this study were obtained from the ADNI database (http://adni.loni.usc.edu). Inclusion criteria for AD subjects included National Institute of Neurological and Communication Disorders/Alzheimer’s Disease and Related Disorders Association criteria for probable AD (NINCDS–ADRDA) [23] with a Mini-Mental State Examination score between 20 and 26, a global Clinical Dementia Rating of 0.5 or 1, a sum-of-boxes Clinical Dementia Rating of 1.0 to 9.0. All amnestic MCI subjects fulfilled a Mini-Mental State Examination score of 24 to 30 and a Memory Box score of at least 0.5. On the other hand, subjects who had any serious neurological disease other than possible AD, any history of brain lesions or head trauma, or psychoactive medication use (including antidepressants, neuroleptics, chronic anxiolytics, or sedative hypnotics) were excluded from this study.

Other details about the ADNI cohort can be foundonline (http://adni.loni.usc.edu/wp-content/uploads/2010/09/ADNI_GeneralProceduresManual.pdf).

Memory function test

The memory function was largely measured by Rey Auditory Verbal Learning Test (RAVLT) scale, and the memory tests and corresponding scores can be found on website (http://adni.loni.usc.edu/upload-data/). Each episodic memory score were converted into a Z score within the population. Although these memory tests are continuous measures, the new IWG criteria required the classification of every test as either normal or abnormal [24], and the cut-off values for these measurements were based on a cohort that was independent of the included subjects. For the purposes of this study, subjects below the tenth percentile of the cognitively normal sample (less than –1.25 SD) were considered to have episodic memory impairment and abnormal [18 , 25].

CSF proteins examination

The detailed methods of CSF proteins, amyloid-beta (Aβ_1-42), total tau (T-tau), and phosphorylated-tau (P-tau₁₈₁) measurement were described elsewhere [26]. The CSF data used here was from the University of Pennsylvania (UPENN), which covered baseline CSF samples in ADNI, ADNI-GO, and ADNI-2. Similarly, cutoff points were necessary to categorize these continuous protein concentrations into normal and abnormal levels. Based on Shaw et al., the cutoff point for CSF Aβ_1-42 was 192 pg/ml, which meant the levels of ≤192 pg/ml were considered abnormal, and the CSF T-tau and P-tau₁₈₁ cutoff level were 93 pg/ml and 23 pg/ml respectively [2, 27].

Neuroimaging methods

The PET imaging data with amyloid tracer, florbetapir (AV-45), were obtained from UC Berkeley – AV45 analysis dataset on website (http://adni.loni.usc.edu/upload-data/). The institute used a native-space MRI scan for each subject that is segmented with Freesurfer (version 4.5.0) to define cortical grey matter regions of interest (ROI) (frontal, anterior/posterior cingulate, lateral parietal, lateral temporal) that make up a summary cortical ROI. The cerebellum was defined as reference region. Each florbetapir scan was applied to the corresponding MRI and the mean florbetapir uptake calculated within the cortical and reference region. Finally, florbetapir standard uptake value ratios (SUVRs) were created by averaging across the 4 cortical regions and dividing this cortical summary ROI by the whole cerebellum. The florbetapir cutoff was determined as 1.11 using the whole cerebellum as reference region. Likewise, other neuroimaging data, such as regional volume on MRI, and cerebral metabolic rate for glucose (CMRgl) on FDG-PET were downloaded from the corresponding dataset in ADNI dataset. The detailed description on acquiring and treating imaging data from ADNI was previously described [28]. Referring to the evidence presented in the refined criteria, we used hippocampal atrophy as MRI-related markers and the CMRgl of bilateral posterior cingulate as FDG-PET markers for our analysis.

Statistical analysis

Given that subjects who had any serious neurological disease other than possible AD had been excluded from ADNI, the criteria for typical AD required both significant memory impairment and markers of Alzheimer’s pathology (either decreased Aβ_1-42 with elevated T-tau/P-tau₁₈₁ in CSF or elevated SUVR on amyloid PET). Asymptomatic AD required normal memory function with abnormal pathophysiological markers (CSF or amyloid imaging biomarkers). Subjects that were positive for cognitive impairment but negative for pathophysiological markers were MCI patients based on the IWG-2 criteria. The normal control group was defined as the individuals with normal cognitive function (including memory function) and without biomarkers of pathology.

Baseline differences between the three stages were analyzed using one-way analysis of variance (ANOVA) for continuous variables and Pearson chi2 analysis and logistic regression models for categorical variables. The Clinical Dementia Rating scale sum of boxes (CDRSB) represented the measure of cognitive impairment in our study [25]. In addition, we adopted a multiple linear regression model to identify correlation between biomarkers and AD severity. In this model, the levels of AD biomarkers were dependent variables, CDRSB was the independent variable, and age, gender, education, and APOE ɛ4 status were used as covariates to estimate coefficients and the 95% confidence interval (CI). Furthermore, to investigate whether topographical biomarkers changed over time, a regression model with change of topographic markers as explanatory variable, the types of population (typical AD and normal group) as independent variable, and follow-up time, age, gender, education, and APOE ɛ4 status as covariates was used. All statistical analyses were performed by Stata 12.0. The criterion for significant difference was p <  0.01.

IWG-2 categorization of ADNI subjects

Figure 1 lists the details of classification. A total of 1,735 individuals were recruited from ADNI database. In this sample 1,727 subjects (AD = 339, MCI = 870, CN = 518) were measured with the memory scales, 389 subjects (AD = 25, MCI = 257, CN = 107) underwent CSF biomarkers test, and 852 individuals (AD = 138, MCI = 458, CN = 256) received amyloid biomarkers examination. Given that only participants with episodic memory evaluation and at least a marker of AD pathology (CSF or amyloid imaging markers) were eligible for our investigation, 857 subjects (139 AD, 462 MCI, and 256 CN) entered into the categorization process. According to the IWG-2 criteria 290 subjects were diagnosed with typical AD, consisting of 117 ADNI AD (84.2% of AD group) and173 MCI (37.4% of MCI group) participants, and 105 CN subjects (41.0% ) were diagnosed as asymptomatic AD. The 289 MCI subjects who showed cognitive impairment without evidence of AD pathology were still considered as MCI group. In addition, 151 CN subjects were viewed as normal control group, defined as those with normal episodic memory and without markers of AD pathology. Finally, 22 subjects from ADNI AD dementia group (15.8% ) did not fulfill both asymptomatic AD and typical AD criteria. Among them, four patients with markers of AD pathology but without memory impairment should be considered as atypical AD, whereas the remaining 18 patients lacked evidence of AD pathology and should be considered as other types of dementia.

Demography, cognition, and neuroimaging at baseline

The baseline demographics, cognitive function, and neuroimaging markers levels at baseline are presented in Table 1. There were significant differences in age, gender, and education years between normal individuals, asymptomatic AD, MCI group, and typical AD. As expected, the ɛ4 allele of APOE gene substantially increased the risk of AD. Typical AD patients had the worst cognitive function, and normal group had the best cognitive performances based on CDRSB, ADAS-cog, RALVT, and MMSE scales. The three neuroimaging markers (hippocampus, entorhinal area, and medial temporal cortex) showed considerable atrophy in typical AD group than those in MCI, asymptomatic AD, and normal group; and typical AD patients had the largest ventricles across the four groups. Furthermore, FDG-PET examination detected that typical AD patients had the lowest CMRgl, followed by MCI, asymptomatic AD, and normal grouprespectively.

Diagnostic markers and Alzheimer’s disease

Based on the IWG-2 criteria, CSF proteins (Aβ_1-42, T-tau, and P-tau₁₈₁) and brain amyloid retention (SUVR) on PET imaging are diagnostic markers. Here, we tested the relationships between these diagnostic markers and the severity of dementia (via CDRSB) within the typical AD group in a multi-linear regression model where levels of diagnostic markers was regarded as dependent variable, the CDRSB scores as independent variable, and age, gender, education, and APOE ɛ4 allele as covariates. None of these diagnostic markers (Aβ_1-42: p = 0.013, T-tau: p = 0.144, P-tau₁₈₁: p = 0.35, SUVR: p = 0.05) showed significant association with CDRSB scores at the significant level of 0.01 (see Table 2).

Progression markers and Alzheimer’s disease

Topographic biomarkers were defined as progression markers by the revised IWG algorithm. Among the topographic markers, hippocampus atrophy is the most robust [29 –31]. Our study suggested that both left hippocampus (p = 0.003) and right hippocampus atrophy (p <  0.001) were positively correlated with the severity of dementia based on CDRSB scale within typical AD group in cross-section analysis, after controlling for age, gender, education and APOE ɛ4 allele (Fig. 2A, B). Furthermore, typical AD patients showed a trend of increasing atrophy on bilateral hippocampus (left hippocampus: p = 0.002, right hippocampus: p = 0.002) compared to normal control group in follow-up, after adjusting for follow-up time, age, gender, education, and APOE ɛ4 allele (Fig. 2C, D).

In addition, cortical hypometabolism on FDG-PET was sensitive to early dysfunction in AD and may follow disease progression. Based on previous evidence, the CMRgl in bilateral posterior cingulate were treated as the region of interest [32]. Using multiple regression analysis with age, gender, education, and APOE4 allele as covariates, we detected that FDG uptake on bilateral posterior cingulate was closely associated with the decline of cognition in typical AD group (p <  0.001) (Fig. 3A). Likewise, typical AD patients had decreasing FDG uptake (p = 0.01) compared to normal control group longitudinally, when corrected for follow-up time, age, gender, education, and APOE ɛ4 allele (Fig. 3B).