Diffusion MRI Indices and Their Relation to Cognitive Impairment in Brain Aging: The Updated Multi-protocol Approach in ADNI3

In the paper

“The TDF model addresses well-established limitations of the standard single-tensor diffusion model—which cannot resolve complex profiles of WM architecture such as crossing or mixing fibers, present in up to 90% of WM voxels (Tournier et al., 2004; Descoteaux et al., 2007, 2009; Jeurissen et al., 2013).”

What the sources say

• ?Not checked — no full text of the sourcecites Tournier et al., 2004 source
• Partly supported

  The Introduction (page not available) describes DTI’s well-known limitations in regions with multiple fiber crossings, and Human Brain Data shows crossings recovered by Q-ball ODFs where DTI profiles are limited. This paper does not establish that crossing or mixing fibers occur in up to 90% of white-matter voxels, and it studies Q-ball imaging rather than the TDF model. (LLM, conf=0.88)

  cites Descoteaux et al., 2007 sourceExperimental — accuracy not yet measured
• Supported

  The Introduction says the single-tensor model is inadequate in voxels with multiple fiber orientations. The Conclusion reports multiple orientations in approximately 90% of white-matter voxels using CSD; this supports the limitation and prevalence attributed to Jeurissen et al., rather than making a claim about TDF itself. (LLM, conf=0.92)

  cites Jeurissen et al., 2013 source

In the paper

“ComBat—originally developed to model and remove batch effects from genomic microarray data (Johnson et al., 2007)—was one of the most effective methods for harmonizing DTI measures in a recent comparison of such techniques (Fortin et al., 2017).”

What the sources say

• ?Not checked — no full text of the sourcecites Johnson et al., 2007 source
• Supported

  The Discussion (section 4) describes a comparison of harmonization methods and states that ComBat performed best at harmonizing FA and MD maps, both DTI measures. This directly supports the claim that it was one of the most effective methods. (LLM, conf=0.95)

  cites Fortin et al., 2017 source

In the paper

“Although many large-scale multi-site DTI studies have obtained consistent results even when acquisition protocols across sites are not harmonized in advance (Jahanshad et al., 2013; Kochunov et al., 2014; Acheson et al., 2017; Kelly et al., 2018), differences in dMRI acquisition parameters, including vendor, voxel size, and angular resolution, are known to affect derived dMRI measures (Alexander”

What the sources say

• Supported

  The Discussion describes datasets with different imaging protocols that were harmonized after collection and reports moderate correlations in heritability estimates across two datasets differing in image acquisition. The Results quantify that correlation as ρ = 0.50. This paper supports the cited sentence as one example, rather than establishing by itself that many studies obtained consistent results. (LLM, conf=0.85)

  cites Jahanshad et al., 2013 source
• Supported

  The Discussion describes a post hoc harmonization protocol for data collected at five sites with diverse imaging protocols, rather than acquisition protocols harmonized in advance. The Results report significant whole-brain FA heritability in all five cohorts, with no significant difference between their estimates. This paper supports the cited sentence as one example of a large multi-site DTI study, not as evidence about how many such studies exist. (LLM, conf=0.88)

  cites Kochunov et al., 2014 source
• Partly supported

  The Discussion (page not specified in our copy) says ENIGMA-DTI was developed to homogenize measurements across diverse imaging protocols and reports a consistent heritability pattern in several large population studies. But this paper’s own reproducibility study used one scanner and the same acquisition protocol (Methods; Conclusions). It does not establish that many multi-site studies obtained consistent results without harmonizing acquisition protocols in advance. (LLM, conf=0.82)

  cites Acheson et al., 2017 sourceExperimental — accuracy not yet measured

In the paper

“Effect size was not correlated with ROI size (Supplementary Figure S13), consistent with prior studies of other disorders (Kelly et al., 2018).”

What the sources say

• Supported

  The Study limitations passage reports a nonsignificant correlation between effect size and ROI size in Kelly et al.’s schizophrenia study, supporting the cited comparison with another disorder. It also notes a medium-sized trend and generally larger effects in larger ROIs, so the finding is best understood as no statistically significant correlation, not proof of no association. (LLM, conf=0.86)

  cites Kelly et al., 2018 source

In the paper

“ComBat uses an empirical Bayes framework to reduce unwanted variation in multi-site data due to differences in acquisition protocol, while preserving the desired biological variation in the data (Fortin et al., 2017).”

What the sources say

• Supported

  Methods §2.3 describes ComBat’s empirical Bayes framework for estimating site effects. The Introduction says it removes unwanted site-induced variation while preserving between-subject biological variability, and Results §3.6 reports that it removed site effects without removing the age-related signal. The paper studies data acquired on different scanners, consistent with the claim’s reference to acquisition differences. (LLM, conf=0.93)

  cites Fortin et al., 2017 source

In the paper

“ADNI’s second phase (ADNI2) introduced to the initiative the use of diffusion-weighted MRI (dMRI) as an additional approach for tracking AD progression (Jack et al., 2015).”

What the sources say

• Supported

  The Introduction says ADNI GO/2 added diffusion MRI as an experimental sequence. The DTI section says an ADNI2 goal was to assess diffusion-weighted MRI’s added value for monitoring brain changes in aging, MCI, and AD, supporting the claim’s description of it as an additional approach for tracking AD progression. (LLM, conf=0.93)

  cites Jack et al., 2015 source

In the paper

“Lower FADTI and higher MDDTI are most frequently reported in studies of AD (Kavcic et al., 2008; Clerx et al., 2012; Nir et al., 2013; Maggipinto et al., 2017; Mayo et al., 2017)”

What the sources say

• Partly supported

  The Results (Linking Performance and DTI; page unavailable) identify decreased FA in posterior regions as distinguishing the AD and older-adult groups. But the DTI Analyses report no statistically significant group difference in either FA or mean diffusivity after correction. The paper does not establish that higher mean diffusivity is a frequent finding across AD studies. (LLM, conf=0.82)

  cites Kavcic et al., 2008 sourceExperimental — accuracy not yet measured
• ?Not checked — no full text of the sourcecites Clerx et al., 2012 source
• Supported

  The Introduction (page not available) says numerous DTI studies of AD and MCI associate greater impairment with lower fractional anisotropy (FA). The Results (page not available) report higher mean diffusivity (MD) in AD patients than in controls. Together, these support the cited sentence’s description of commonly reported AD findings. (LLM, conf=0.88)

  cites Nir et al., 2013 source

In the paper

“dMRI has since been used in numerous studies to understand the effects of AD on white matter (WM) microstructure and brain connectivity (Daianu et al., 2013a,b; Nir et al., 2013; Prasad et al., 2013).”

What the sources say

• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites Daianu et al., 2013a,b
• Supported

  The Discussion describes growing diffusion-imaging evidence of AD-related white matter changes, cites numerous studies, and explicitly lists tract or connectivity analyses among the DTI methods used to assess AD. This supports the broad background claim; the supplied passages do not provide page numbers. (LLM, conf=0.91)

  cites Nir et al., 2013 source
• Supported

  The Introduction says diffusion imaging was added to several large-scale AD studies to examine brain integrity and connectivity. This paper is also an example: it uses diffusion MRI to study white-matter networks in people with AD and comparison groups. (LLM, conf=0.88)

  cites Prasad et al., 2013 source

In the paper

“The participants recruited for ADNI3 tend to be younger and healthier, on average, than those in ADNI2, as they were recruited with the intention of studying the transition from CN to AD (Jack et al., 2015).”

What the sources say

• ?Could not verify

  The ADNI 3 MRI section (page not available in our copy) describes planned scans and imaging aims, but does not compare ADNI 3 recruits with ADNI 2 recruits. Searches for younger or healthier participants and recruitment to study the transition from cognitively normal to AD found no passage supporting those claims. (LLM, conf=0.77)

  cites Jack et al., 2015 sourceExperimental — accuracy not yet measured

In the paper

“but AxDDTI may be more sensitive to unspecific microscopic cellular loss earlier in the disease (O’Dwyer et al., 2011), perhaps making it more sensitive in the healthier participants of the ADNI3 dataset.”

What the sources say

• Partly supported

  The Results (Early detection of diffusional changes in MCI) report axial diffusivity (DA) differences in MCI, supporting sensitivity to early white-matter changes. The Introduction and Discussion relate DA changes to white-matter damage and possible axonal degeneration, but do not establish that DA is specifically sensitive to nonspecific microscopic cellular loss. The ADNI3 inference is the citing authors’ own extrapolation. (LLM, conf=0.84)

  cites O’Dwyer et al., 2011 sourceExperimental — accuracy not yet measured

In the paper

“As a result, improved harmonization of multi-site diffusion data is of great interest (Grech-Sollars et al., 2015; Pohl et al., 2016; Palacios et al., 2017).”

What the sources say

• Partly supported

  The Introduction identifies comparability of diffusion data from different scanners as an important question and discusses harmonization across sites. But the Discussion (> Volunteers; page unavailable) concludes that large multi-centre studies may be feasible without extensive protocol harmonization, rather than advocating improved harmonization. (LLM, conf=0.78)

  cites Grech-Sollars et al., 2015 sourceExperimental — accuracy not yet measured
• Supported

  The Discussion’s “Limitations, considerations, and conclusions” section calls harmonization essential for making data collected across multiple sites and systems comparable. The “Data harmonization across MR system manufacturers” section also reports that scanner differences prevented the authors from combining DTI data without a correction factor. (LLM, conf=0.94)

  cites Pohl et al., 2016 source
• Supported

  The Introduction says imaging biomarkers must be standardized for precise, reproducible measurements across different instruments and procedures. The Discussion adds that compensating for inter-site differences may further reduce variability in multisite diffusion MRI studies, supporting the cited interest in improved harmonization. Page numbers are not provided in our copy. (LLM, conf=0.92)

  cites Palacios et al., 2017 source

In the paper

“The DTI model cannot differentiate these components or even crossing fibers (Tuch et al., 2002; Jbabdi et al., 2010), which are estimated to occur in up to 90% of WM voxels at the typical dMRI resolution (Descoteaux et al., 2009; Jeurissen et al., 2013).”

What the sources say

• ?Not checked — no full text of the sourcecites Tuch et al., 2002 source
• ?Not checked — no full text of the sourcecites Jbabdi et al., 2010 source
• Partly supported

  The Introduction says that many voxels contain multiple crossing or otherwise complex fiber populations at DTI acquisition resolution. Section 4.1 also reports voxels with two- and three-fiber crossings. Neither passage gives the cited estimate of up to 90% of white-matter voxels, so the numerical part attributed to this source is unsupported. (LLM, conf=0.86)

  cites Descoteaux et al., 2009 sourceExperimental — accuracy not yet measured

In the paper

“Among the three cognitive assessments, CDR-sob showed the strongest correlations with dMRI indices, in line with prior ADNI brain imaging studies (Hua et al., 2009; Nir et al., 2013).”

What the sources say

• Partly supported

  The Results section (“Minimal sample size estimates and power analyses”; page unavailable) identifies CDR-SB as the best of the three clinical assessments, but ranks them by the sample sizes needed to detect change. It does not show that CDR-SB had the strongest correlations with dMRI indices; the imaging measure studied was tensor-based morphometry of structural MRI. (LLM, conf=0.83)

  cites Hua et al., 2009 sourceExperimental — accuracy not yet measured
• Partly supported

  The Discussion reports that CDR-sob was the most widely correlated of the three assessments with diffusivity in the MCI group, in both voxel-based and ROI analyses. It does not establish that CDR-sob had the strongest correlations with dMRI indices across the full cohort or across all dMRI measures, as the unqualified claim implies. Page numbers are unavailable in our copy. (LLM, conf=0.86)

  cites Nir et al., 2013 sourceExperimental — accuracy not yet measured

In the paper

“However, the CST ROI from the JHU WMPM atlas is limited to a small region in the inferior portion of the brain and has been shown to be the least reliable and reproducible ROI (Jahanshad et al., 2013; Acheson et al., 2017)”

What the sources say

• Partly supported

  The Methods (Region of interest extraction; page unavailable) identify the JHU white matter parcellation atlas, and the Results (Meta and mega analyses; page unavailable) describe the CST as small and, with the fornix, among the least heritable regions and difficult to register. The Discussion cautions that CST may not be a reliable endophenotype in multisite studies. These passages do not establish that its atlas ROI is confined to an inferior brain region or that it is the least reliable and reproducible ROI. (LLM, conf=0.86)

  cites Jahanshad et al., 2013 sourceExperimental — accuracy not yet measured
• Partly supported

  The Methods (page not provided) say the tract ROIs were derived from the JHU white matter parcellation atlas, and the Results report relatively poor CST reproducibility. But the paper does not describe the CST ROI as a small region confined to the inferior brain, and it does not rank CST as the least reproducible: in the adolescent results, the fornix had higher variance and a lower ICC. (LLM, conf=0.90)

  cites Acheson et al., 2017 sourceExperimental — accuracy not yet measured

In the paper

“In healthy tissue with crossing fibers, the DTI model may show low FA. FADTI may paradoxically appear to increase in regions where crossing fibers deteriorate in neurodegenerative diseases such as AD (Douaud et al., 2011).”

What the sources say

• Supported

  The Image processing section says crossing fibres can cause a substantial drop in FA. The Results and Discussion report higher FA in AD patients than controls in a crossing-fibre region, associated with relative sparing of one pathway compared with the affected pathway. Page numbers were not supplied in our copy. (LLM, conf=0.93)

  cites Douaud et al., 2011 source

In the paper

“This discrepancy between our findings and that of Fortin et al. (2017), may be due to several differences between studies: (1) ADNI3 includes more sites and protocols; (2) in contrast to the number of voxels, the number of ROIs is far less than the number of participants; and (3) the age effects in the elderly populations tested here are stronger than the effects tested in adolescents in Fortin et”

What the sources say

• Partly supported

  Split into 2 atoms — 1 confirmed, 1 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.

  cites Fortin et al. (2017) sourceExperimental — accuracy not yet measured

In the paper

“Each ADNI protocol showed directionally consistent associations in more than 89% of tests (24 ROIs * 5 dMRI indices), except for P36 which was consistent in 81%, but had the smallest sample size (N = 12; Figure 2B; Supplementary Tables S5–S11).”

In the paper

“but the directions of associations with age were largely consistent for ADNI2 and ADNI3 phases separately, and each ADNI3 protocol (Figures 1, 2).”

In the paper

“FATDF and all three diffusivity indices were consistent in ≥96% of tests (24 ROIs * 8 protocols/phases), while FADTI was only consistent in 88% of tests.”

In the paper

“None of the associations in the unexpected direction were significant after multiple comparisons correction, and only two had a p ≤ 0.05.”

In the paper

“FATDF detected robust widespread associations with clinical measures, while FADTI was the weakest of the five indices for detecting associations.”

In the paper

“Pairwise tests comparing the mean age of CN participants scanned with each protocol also showed significant differences between those scanned with S31 and two other protocols: GE54 and S31 (p = 0.026); P33 and S31 (p = 0.0037).”

In the paper

“Of the three cognitive measures, CDR-sob associations showed the largest effect sizes across dMRI indices (in the UNC followed by the CGH for all indices except FADTI)”

In the paper

“Compared to FADTI, FATDF showed larger effect sizes; across cognitive tests, the strongest FATDF associations were detected in the UNC with CDR-sob (d = −1.244; p = 1.39 × 10−20), followed by the CGH (d = −1.213; p = 8.86 × 10−20).”

In the paper

“After using ComBat to harmonize dMRI indices across protocols, ANCOVAs revealed that significant protocol differences in dMRI indices were all but eliminated across ROIs (Supplementary Figure S2A); significant protocol differences were detected only in the CST, for each of the dMRI indices.”

In the paper

“The UNC showed either the first or second largest effect size (alternating with CGH) across diffusivity indices and FATDF tests, but was significant only for cognitive measure associations with FADTI (i.e., three of four clinical tests).”

In the paper

“Greater cognitive impairment was associated with lower anisotropy and higher diffusivity.”

• Also stated in: § Cognitive Measure Associations With Pooled ADNI3 dMRI Indices · plain

  “higher diffusivity was associated with greater cognitive impairment. Lower (D) FADTI (FDR critical threshold p = 0.025) and (E) FATDF (FDR critical threshold p = 7.73 × 10−3) were also associated with greater impairment”

In the paper

“On the other hand, FATDF was significant in 20 out of 24 ROIs, similar to diffusivity indices.”

In the paper

“ANCOVAs revealed significant protocol differences for 22 ROIs for FADTI and FATDF, with the highest overall effect size detected in the anterior limb of the internal capsule (ALIC) and the external capsule (EC) for FADTI (ALIC: d = 0.648; EC: d = 0.652).”

In the paper

“AxDDTI had the smallest effect size, overall, in the splenium of the corpus callosum (SCC; d = 0.106), and only 13 ROIs showed significant AxDDTI differences between protocols.”

In the paper

“In 14 of 15 tests, the CGH consistently showed one of the top two largest effect sizes (CGH FADTI association with CDR-sob was the third largest), along with the uncinate fasciculus (UNC), which was top two in 12 of 15 tests (while significant, cognitive associations with UNC FADTI never showed one of the largest effect sizes).”

In the paper

“In pairwise analyses, AxDDTI was the most stable index across protocols, as significant protocol differences were detected in only 20.6% of pairwise tests (24 ROIs * 21 pairwise tests), compared to FADTI, the most variable index, which showed significant protocol differences in 81.9% of tests (Figure 4B).”

In the paper

“FATDF was significant in 89 of the 96 tests (92.7%; 24 ROIs * 4 clinical tests), while FADTI was only significant in 58 (60.4%).”

In the paper

“FADTI showed the greatest significant differences (largest d-values; dark red) between protocols and AxDDTI the fewest (dark green).”

• Also stated in: § ADNI2 and ADNI3 Protocols in Cognitively Normal Participants · plain

  “AxDDTI was the most stable dMRI index across protocols, while FADTI was the least stable.”

In the paper

“the largest effect sizes across all tests were detected with AxDDTI (UNC: d = 1.344) and MDDTI (UNC: d = 1.342, p = 3.47 × 10−23).”

In the paper

“In a preliminary analysis, we showed that one harmonization method performed on these regional measures, ComBat, reduced cross-site differences in dMRI indices, while preserving biological relationships with age in CN controls.”

In the paper

“While associations with clinical impairment were detected throughout the WM, the region that most frequently showed the lowest effect sizes and was significant in only 3 of the 20 clinical tests, was the corticospinal tract (CST).”

In the paper

“Here, the number of significant ROIs and the magnitude of effect sizes were comparable for ComBat and nested random-effects model approaches.”

In the paper

“As before, we used FDR to account for multiple comparisons.”

In the paper

“Due to the limited available sample size of AD participants (N = 22), and their uneven distribution across the acquisition protocols tested here, we compared only groups of people with CN and MCI diagnoses.”

In the paper

“Age, sex, and age*sex interactions were controlled for as fixed effects, and the protocol and acquisition site were modeled as nested random variables.”

In the paper

“For each dMRI index, we used FDR to correct for multiple comparisons across the 24 ROIs assessed.”

In the paper

“All DWI and T1-weighted images were visually checked for quality assurance.”

In the paper

“In the CN participants from ADNI2 and ADNI3, we ran ComBat on each of the dMRI indices, including age, sex, age*sex, and information from all 24 ROIs to inform the statistical properties of the protocol effects.”

In the paper

“In CN participants, multivariate random-effects linear regressions were used to assess whether dMRI indices from each ADNI protocol individually were associated with age, controlling for sex and age*sex interactions as fixed variables, and acquisition site as a random variable.”

In the paper

“FDR was again used to correct for 24 ROI tests (q = 0.05; Benjamini and Hochberg, 1995).”

What the sources say

• ?Not checked — no full text of the sourcecites Benjamini and Hochberg, 1995 source

In the paper

“We used the false discovery rate (FDR) procedure to correct for multiple comparisons (q = 0.05; Benjamini and Hochberg, 1995) across the 24 ROIs assessed for each dMRI index.”

What the sources say

• ?Not checked — no full text of the sourcecites Benjamini and Hochberg, 1995 source

In the paper

“In CN participants, we tested for significant differences in dMRI indices between the seven ADNI protocols using analyses of covariance (ANCOVAs), adjusting for age, sex, and age*sex interactions as fixed variables, and acquisition site as a random variable.”

In the paper

“Of the 381 participants scanned to date, 55 were excluded after quality assurance: this included ensuring complete clinical and demographic information, and image-level quality control (removing scans with severe motion, missing volumes, or corrupt files).”

In the paper

“To ensure sufficient statistical power to assess differences in data collected with different protocols, we evaluated only those protocols with complete available data for at least 10 participants at the time of download; we did not assess protocol GE36, for which scans from 9 of 12 participants passed quality assurance.”

In the paper

“we did not assess protocol GE36, for which scans from 9 of 12 participants passed quality assurance.”

In the paper

“Even so, separating protocol differences from differences in scanner manufacturer is difficult.”

In the paper

“perhaps due to the larger voxel size in ADNI2 (2.7 mm3 vs. 2.0 mm3 isotropic voxels used in ADNI3).”