Skip to main content

Table 3 Frequency of the hippocampal-sparing AD subtype across studies and from our original data

From: The hippocampal sparing subtype of Alzheimer’s disease assessed in neuropathology and in vivo tau positron emission tomography: a systematic review

A

Data from the systematic review: postmortem studies

Study

Murray et al. 2011 [7]

Whitwell et al. [9]

Petersen et al. [25]

Corder et al. [28]

Uretsky et al. [26]

Smirnov et al. [27]

Data modality

Postmortem (NFT count)

Postmortem (NFT count)

Postmortem (NFT count)

Postmortem (NFT count)

Postmortem (NFT count)

Postmortem (NFT count)

Braak’s tau NFT stage

V or VI

V or VI

V or VI

I to VI

V or VI

V or VI

Subtyping algorithm

Murrayb

Murrayb

Murrayb

Data-driven (GoM)

Approximation of Murrayb

Approximation of Murrayb

Sample size

889

177

74

249

292

121

Percentage of individuals with hippocampal-sparing AD

11%

11%

7%

–

8%

19%

Percentage of individuals with NFT count or tau PET uptake completely sparing the hippocampus, according to:

      

Neuropathologic definition

0%

0%

0%

0%

0%

0%

B

Data from the systematic review: in vivo tau PET studies

Study

Schwarz et al. [29]

Schwarz et al. [30]

Palleis et al., [32]

Rullmann et al. [33]

Krishnadas et al. [34]

Toledo et al. [31]

Data modality

PET (flortaucipir)

PET (flortaucipir)

PET (PI-2620)

PET (PI-2620)

PET (MK-6240)

PET (flortaucipir)

Braak’s tau NFT stage

0 to VIa

0 to VIa

Unknown

I to VIa

Unknown

Unknown

Subtyping algorithm

Braak staginga

Braak staginga,c

Tau PET positivity in cortex in conjunction with tau PET negativity in mesial temporal lobe

Braak staging a

Visual inspection based on Murrayb

Data-driven (Robust collaborative clustering)

Sample size

75

98

10

38

151

 

Percentage of individuals with hippocampal-sparing AD

5%

15%c

7%c

1%c

55%

18%

18%

Identified but frequency not reported

Percentage of individuals with NFT count or tau PET uptake completely sparing the hippocampus, according to:

      

Neuropathologic definition

5%

15% c

7% c

1% c

55%

18%

18%

Identified but frequency not reported

C

Re-analysis from published studies

New data using the ADNI cohort

Study

Whitwell et al. [11]

Charil et al. [6]

Young et al. [12]

Mohanty et al. [8] (Byun’s algorithm15—hippocampus)

Mohanty et al. [8] (Charil’s algorithm6—hippocampus)

Mohanty et al., [8] (Risacher’s algorithm20—hippocampus)

Data modality

PET (flortaucipir)

PET (flortaucipir)

PET (flortaucipir)

PET (flortaucipir)

PET (flortaucipir)

PET (flortaucipir)

Braak’s tau NFT stage

Unknown

V or VI a

Unknown

Unknown

Unknown

Unknown

Subtyping algorithm

Data-driven (K-means clustering)

Murrayd

Approximation of Murrayb

Byune

Murrayd

Murrayf

Sample size

62

45

392

84

84

84

Percentage of individuals with hippocampal-sparing AD

34%

13%

9%

21%

10%

11%

Percentage of individuals with NFT count or tau PET uptake completely sparing the hippocampus, according to:

      

Neuropathologic definition

–

–

–

–

–

–

‘Accuracy-based cut point’

3% g

9%

6%

0%

0%

0%

‘ + 1SD cut point’

34% g

13%

9%

21%

6%

7%

‘10% cut point’

34% g

13%

9%

21%

6%

7%

‘Schöll cut point’

5% g

11%

7%

0%

0%

0%

‘Maass cut point’

2% g

2%

1%

0%

0%

0%

  1. aBraak staging [10] was based on tau PET data
  2. bMurray’s subtyping algorithm is explained in Murray et al. [7]. Briefly, the ratio of the hippocampal to cortical NFT counts was split at the 25th and 75th percentiles of the sample distribution. At a first step, individuals with the ratio < 25% were assigned to hippocampal-sparing AD, those with ratio > 75% were assigned to limbic-predominant AD, and all the rest were assigned to typical AD. At a second step, individuals were re-classified based on median NFT counts in hippocampus and cortex. Brain areas considered in Murray’s subtyping algorithm are hippocampus (CA1 and subiculum), superior temporal cortex, middle frontal cortex, and inferior parietal cortex
  3. cSchwarz et al. [30] used three classification schemes for tau staging. The first scheme is the same than in Schwarz et al. [29] and was designed to mimic Braak staging [10] as closely as possible. The second scheme was a simplified version of the first scheme using fewer and larger ROIs located in medial, lateral, and superior temporal lobes and in the primary visual cortex. The third scheme was even simpler than the first two schemes and used lobar ROIs: temporal, frontal, parietal, and occipital. Percentages are reported in order of appearance (first, second, and third algorithm)
  4. dThe subtyping algorithm in Charil et al. [6] is the same as in Murray et al. [7] with a minor difference in relation to the brain areas considered in the algorithm, substituting CA1 and subiculum areas of the hippocampus for anterior-most position (head) of hippocampus or entorhinal cortex (both methods were tested, for results from the method using entorhinal cortex please see Additional file 1: Table S5). Superior temporal cortex, middle frontal cortex, and inferior parietal cortex are the same as in Murray et al. [7]
  5. eBased on the original algorithm described in Byun et al. [15] regional tau PET uptake measures were adjusted for age using multiple linear regression based on a normative group of amyloid-negative healthy controls from ADNI. Using the normative group, Z-scores of hippocampal/entorhinal cortex, frontal, temporal, and parietal regions were calculated and classified as abnormal when Z-score > 1.0. Subtypes were then determined exactly as in Byun et al. [15] Brain areas considered in Byun’s algorithm are the same as in Murray et al. [7] with a minor difference with substituting CA1 and subiculum areas of the hippocampus for hippocampus or entorhinal cortex (both methods were tested, please see Additional file 1: Table S5). Superior temporal cortex, middle frontal cortex, and inferior parietal cortex are the same as in Murray et al. [7]
  6. fThe subtyping algorithm in Risacher et al.20 is the same as in Murray et al.,7 with a minor difference in relation to the brain areas considered in the algorithm, substituting CA1 and subiculum areas of the hippocampus for hippocampus or entorhinal cortex (both methods were tested, please see Additional file 1: Table S5), and extending the cortical regions to include middle frontal cortex, inferior frontal cortex, superior temporal cortex, inferior parietal cortex, superior parietal cortex, and supramarginal cortex
  7. gBased on the entorhinal cortex instead of the hippocampus. We estimated the approximate proportion of cases who had NFT in the association cortex while completely sparing the hippocampus, based on pooling of all the tau PET data independently of cut point and subtyping method. To do this, each cell from tau PET studies in the “Percentage of individuals with NFT count or tau PET uptake completely sparing the hippocampus” section of the table was treated as an independent study. The total of cases with cortical tau PET uptake completely sparing the hippocampus was computed (n = 372) and divided by the total number of cases included in the studies (N = 5583). The resulting proportion is 8%. GoM = Grade of Membership analysis; AD = Alzheimer’s disease; ADNI = Alzheimer's Disease Neuroimaging Initiative; NFT = neurofibrillary tangles; SD = standard deviation; pc = percentile; PET = positron emission tomography
Back to article page

Acta Neuropathologica Communications

ISSN: 2051-5960

Contact us