Share

February 07, 2023; 100 (6) Research ArticleOpen Access

Association of Regional Atrophy With Naming Decline in Primary Progressive Aphasia

Bonnie L. Breining, Andreia Vasconcellos Faria, Donna Clark Tippett, Melissa Dawn Stockbridge, Erin L. Meier, Brian Caffo, Olivia Hermann, Rhonda Friedman, Aaron Meyer, Kyrana Tsapkini, Argye Elizabeth Hillis
First published November 1, 2022, DOI: https://doi.org/10.1212/WNL.0000000000201491
Full PDF
Short Form
Citation
Permissions

Make Comment

See Comments

Downloads
386

Share

Abstract

Background and Objectives Primary progressive aphasia (PPA) is a neurodegenerative condition that predominantly impairs language. Most investigations of how focal atrophy affects language consider 1 time point compared with healthy controls. However, true atrophy quantification requires comparing individual brains over time. In this observational cohort study, we identified areas where focal atrophy was associated with contemporaneous decline in naming in the same individuals.

Methods Cross-sectional analyses–related Boston Naming Test (BNT) performance and volume in 22 regions of interests (ROIs) at each time point using Least Absolute Shrinkage and Selection Operator (LASSO) regression. Longitudinal analysis evaluated changes in BNT performance and change in volume in the same ROIs.

Results Participants (N = 62; 50% female; mean age = 66.8 ± 7.4 years) with PPA completed the BNT and MRI twice (mean = 343.9 ± 209.0 days apart). In cross-sectional left inferior frontal gyrus pars opercularis, superior temporal pole, middle temporal gyrus, and inferior temporal gyrus were identified as critical for naming at all time points. Longitudinal analysis revealed that increasing atrophy in the left supramarginal gyrus and middle temporal pole predicted greater naming decline, as did female sex and longer intervals between time points.

Discussion Although cross-sectional analyses identified classic language areas that were consistently related to poor performance at multiple time points, it was not increasing atrophy in these areas that lead to further decline: longitudinal analysis of each person's atrophy over time instead identified nearby but distinct regions where increased atrophy was related to decreasing performance. The results demonstrate that directly examining atrophy (in each individual) over time furthers understanding of decline in PPA and reveal the importance of left supramarginal gyrus and middle temporal pole in maintaining naming when areas normally critical for language degenerate. The novel results provide insight into how the underlying disease progresses to result in the clinical decline in naming, the deficit most common among all 3 PPA variants.

Glossary

BNT=
Boston Naming Test;
LASSO=
Least Absolute Shrinkage and Selection Operator;
NACC=
National Alzheimer Coordinating Center;
PALPA=
Psycholinguistic Assessments of Language Processing in Aphasia;
PPA=
primary progressive aphasia;
ROI=
regions of interest

Primary progressive aphasia (PPA) is a neurodegenerative condition with heterogeneous neuropathologic causes.1 It predominantly impairs language, while other cognitive functions remain relatively intact. Three main variants have been distinguished.2 Logopenic variant PPA impairs short-term phonological memory and is linked to atrophy in left posterior perisylvian or parietal regions. Nonfluent/agrammatic variant PPA leads to agrammatism in language production and/or apraxia of speech and involves left posterior frontoinsular atrophy. Semantic variant PPA impairs naming and single-word comprehension and is associated with anterior temporal atrophy, predominantly in the left hemisphere.

Most of what we know about brain-behavior relationships is PPA comes from cross-sectional investigations. In such studies, researchers typically attribute poor performance to dysfunction of regions that seem grossly atrophied or that are reduced in volume or cortical thickness relative to the brains of healthy individuals.3,-,13 Some longitudinal studies have investigated the overall progression of the atrophy in the different variants, finding more diffuse atrophy and widespread language impairment over time.14,-,21 However, there are few reports connecting behavioral decline on specific tasks to increasing atrophy in specific areas (although see Ref. 22 from our group). Our primary research question was as follows: Is further behavioral decline linked to (1) continuing atrophy of critical language regions that are first affected or (2) atrophy of additional regions that may have been playing a compensatory role in processing? To investigate this question, comparisons of each individual's brain images over time are required. Here, we used longitudinally collected data to examine both atrophy and language decline over time to more directly connect brain and behavior changes. We chose to look at naming performance as all 3 variants of PPA typically experience naming difficulty and are likely to show additional decline over time in this function. We recognize that progressive naming difficulty reflects distinct underlying deficits across variants and even across individuals. For example, individuals with nonfluent agrammatic PPA might make naming errors because of apraxia of speech, and those with semantic variant might make naming errors because of impaired semantic representations of the objects being named. These errors would likely be reflected in different naming error types. However, we use naming accuracy to illustrate the usefulness of the approach for evaluating changes in brain volume associated with changes in language, rather than to identify neural regions critical for the various cognitive and motor processes underlying naming. We also conducted more traditional cross-sectional analyses separately considering the data collected at the time of the first and second scan, so the methods could be compared. We hypothesized that cross-sectional analyses at multiple time points would yield results that are complementary to directly examining longitudinal changes in volume and performance over the same period.

Methods

Participants

A convenience sample of 62 individuals with PPA (31 female; mean age = 66.8 ± 7.37 years) recruited from 2 studies that included longitudinal imaging: a PPA treatment study (PI: KT) at Johns Hopkins University School of Medicine and a treatment study (PIs: RF and AM) at Georgetown University and Johns Hopkins participated from 2010 to 2021. PPA diagnoses were based on presentation with progressive language impairment without accompanying cognitive, behavior, or personality changes.2,6 Detailed language and cognitive assessments, history, comprehensive neurologic examination, and available neuroimaging were used to classify participants as having logopenic variant (N = 24), nonfluent/agrammatic variant (N = 26), or semantic variant (N = 11) PPA according to the consensus guidelines.2 The batteries covered the same domains of language (naming, word and sentence comprehension, repetition, syntactic processing, reading, spelling, and nonverbal semantics), but with different tests. For 1 of the studies, the following language tests were administered: Boston Naming Test (BNT)23; Hopkins Assessment of Naming Actions24; Letter Fluency (F,A,S); Semantic Fluency (Fruits and Animals); Subject-relative, Object-relative Active, Passive sentence comprehension25; Spelling to Dictation (JHU Dysgraphia Battery Probability List)26; National Alzheimer Coordinating Center (NACC) Frontotemporal Lobar Degeneration Module27 Sentence Repetition subtest; Temple Assessment of Language and (Verbal) Short-Term Memory in Aphasia28 Sentence Repetition and Nonword Repetition, Psycholinguistic Assessments of Language Processing in Aphasia (PALPA)29 subtest 47: Spoken Word-Picture Matching; and PALPA subtest 48: Written Word-Picture Matching. Conceptual semantics was assessed with the Pyramids and Palm Trees Test short version30 and Kissing and Dancing Test short version (unpublished version of 3). More general cognitive tests included Mini-mental State Examination,32 Digit Span Forward, Digit Span Backward, Spatial Span Forward, Spatial Span Backward, NACC Story Immediate and Delayed Recall (Verbatim & Paraphrase),33 Digit Symbol Substitution, Raven Progressive Matrices,34 and Wechsler Memory Scale Paired Associates (Immediate and Delayed Recall).35 For the other study, the following language tests were administered: the BNT, Northwestern Anagram Test,36 Auditory Word-Picture Matching,37 Written Word-Picture Matching, Reading and Spelling batteries (unpublished), Pseudoword Repetition (unpublished), Picture Description, Basic Word Discrimination, Embedded Sentences, Word and Sentence Repetition, and Action Naming subtests of the Boston Diagnostic Aphasia Examination.38 Lexical semantic tests included the word version of the Pyramids and Palm Trees Test,39 Word version of the Kissing and Dancing Test,31 Word Sorting from the Cambridge Semantic Battery,40 and the Synonym Judgment Test.41 Conceptual semantic processing was assessed with Picture version of the Pyramids and Palm Trees Test, Picture version of the Kissing and Dancing Test, Picture Sorting from the Cambridge Semantic Battery, and the Color Knowledge Test.37 More general cognitive tests included the Montreal Cognitive Assessment42 and the Benson Figure Copy and Recall tasks.43 Demographic information about participants in each variant is presented in Table 1. The variant groups did not significantly differ in age, education, or time since onset. There was a larger proportion of women in the lvPPA group (67% female) as compared with the nfvPPA group (35% female; z = 2.3, p = 0.024). The only language assessment presented in the table is the BNT because it was the only assessment that was given consistently in both studies from which participants were drawn. Most participants with nonfluent/agrammatic PPA had apraxia of speech, as indicated by at least 1 clinician (speech-language pathologist or neurologist) who evaluated them. However, we do not include these data because there was not a consistent and reliable evaluation for apraxia of speech across studies. One participant was unclassified because the individual did not have any of the core criteria for any variant but had progressive impairment in naming and spelling. Each participant took part in concurrently collected behavioral assessment and MRI at 2 time points (mean = 343.9 ± 209.0 days apart) as part of a treatment study; this interval was not significantly different between variant groups. Data from 2 separate treatment studies were used; comparison of the studies and their treatment effects are not considered here because they are ongoing.

View this table:
Table 1

Demographic Information

Standard Protocol Approvals, Registrations, and Patient Consents

The studies were conducted with the approval of the Johns Hopkins University Institutional Review Board; all participants provided written informed consent in line with the Declaration of Helsinki.

Behavioral Assessment

The BNT was used to evaluate object naming performance. Participants named black and white line drawings of objects; 25 participants completed the 60-item full-form version,23 while 39 completed the 30-item short form.44 Scores were normalized to the same scale by doubling short form scores. Uncued first responses were scored. To illustrate our approach with adequate power, we used the global accuracy score, rather than evaluating areas of atrophy associated with change in particular naming error types.

Neuroimaging Collection and Analysis

High-resolution MPRAGE T1-weighted images were acquired using a 3T MRI scanner with the same imaging parameters at both time points (axial orientation, image matrix of 256 × 256 or 224 × 224 mm, and 120–160 slices, voxel size 1 × 1 × 1 or 1 × 1 × 1.1 mm3, TR/TE 8.4/3.9 or 8.1/3.7 ms). Two types of analyses were conducted: cross-sectional analyses conducted separately for the earlier and later time points (i.e., at the time of the first and second scan, respectively) and a longitudinal analysis evaluating change between the 2 time points.

For the cross-sectional analyses, each anatomic scan was segmented into 276 regions through atlas-based analysis in the MRI Cloud platform.45 In this analysis, each participant's brain is registered to multiple geriatric atlases, labeled based on the atlas parcellation, and then warped back to participant space, using the highly accurate large deformation diffeomorphic metric mapping algorithm and a multiatlas fusion label algorithm. The volume of brain tissue in each region was calculated in MNI space. To control for regional atrophy, we normalized volumes by cerebral volume (total brain volume without myelencephalon and CSF). To control for interindividual brain size differences, we also calculated the ratio of cerebral volume to intracranial volume (total brain volume without myelencephalon) for each scan. We selected 22 of the 276 regions from the automatic parcellation as regions of interests (ROIs): 11 left hemisphere regions that are commonly associated with language and their right hemisphere homologues. These included pars opercularis, pars orbitalis, and pars triangularis of inferior frontal gyrus; supramarginal gyrus; angular gyrus; superior, middle, and inferior temporal gyri; superior and middle temporal poles; and fusiform gyrus. These regions were chosen because they are commonly associated with language and often damaged in PPA.2,-,6,8,-,12

For the longitudinal analysis, we quantified change in each ROI. For each individual, we used DiffeoMap46 to align the later scan to the earlier scan, using automated image registration followed by large deformation diffeomorphic metric mapping. The Jacobian determinant, which is the local expansion factor of the large deformation diffeomorphic metric mapping deformation fields, was used to quantify local volume changes at the voxel level. A Jacobian of less than 1 indicates shrinkage relative to the earlier scan, while a Jacobian of greater than 1 indicates expansion. We calculated the mean Jacobian for each of the 22 ROIs used in the cross-sectional analysis, using the parcellation of the earlier scan to define the regions. Figure 1 shows an example Jacobian determinant map. We calculated the mean Jacobian for each of the 22 ROIs used in the cross-sectional analysis, using the parcellation of the earlier scan to define the regions.

Figure 1
Figure 1 Example of a Jacobian Determinant Map

The top of the figure shows the brain of 1 participant with primary progressive aphasia at the earlier time point, labeled day 0. The middle of the figure shows the same individual's brain at the later time point 260 days later. With the naked eye, it is difficult to see exactly where changes are occurring. The bottom of the figure projects a map of the Jacobian determinant for each voxel onto the original brain. A Jacobian determinant of less than 1 indicates shrinkage and is shown in blue/purple, while a Jacobian determinant of greater than 1 indicates expansion and is shown in orange/red. This individual showed expansion of the ventricles and shrinkage in diffuse cortical areas. There is also expansion of sulci in left frontal areas.

Statistical Analysis

We used Least Absolute Shrinkage and Selection Operator (LASSO) regression47 to evaluate relationships between naming performance and volume in the 22 ROIs. This method is useful in this situation with a large number of predictors relative to the sample size because LASSO regression performs automated feature selection and shrinkage.48 The glmnet package in R49 was used to perform LASSO regression with standardized features, using leave-one-out cross-validation to select the lasso penalty parameter (typically labeled λ) value that resulted in the minimum mean cross-validated error. The selective Inference package50 was used to conduct inference testing and to calculate p values for the selected features.

All analyses used 1-tailed LASSO regression, identifying positive predictors where worse naming performance or greater decline was associated with smaller regional volume or greater increase in atrophy. The cross-sectional models included BNT performance as the dependent variable and volume of each ROI normalized by cerebral volume at the corresponding time point as predictors. The longitudinal model included change in BNT performance (BNT performance at the earlier time point—BNT performance at the later time point) as the dependent variable and the mean Jacobian for each ROI as predictors. All models included cerebral/intracranial volume ratio (at the corresponding time point for the single time point models and the first time point for the longitudinal model) as a measure of overall atrophy, the demographic variable of sex, and estimated years since onset of symptoms. The longitudinal model additionly used initial BNT score at the first time point and the number of days between time points as a predictor. Years since onset and days between time points were reverse coded (as −1*the interval) because we expected longer intervals to predict worse performance and wanted to capture these predictors in the 1-tailed model.

Data Availability

Anonymized data not published within this article will be made available by request from any qualified investigator.

Results

The results are presented in Figure 2. Cross-sectional results relating naming performance and volume in 22 ROIs at each time point using LASSO regression are summarized in Table 2. Longitudinal results evaluating changes in naming performance and changes in volume in the same 22 ROIs are summarized in Table 3. Note that LASSO regression selects a group of features that together predict the outcome; although statistical significance can be estimated for specific features within that group, all selected features are considered important predictors. Although p values for the selected features are presented in the tables, they are not discussed in the text. We also created proof-of-concept, exploratory multivariable models with the selected features, both for all participants and for each variant, which are presented in Tables 4 and 5 for cross-sectional and longitudinal analyses, respectively.

Figure 2
Figure 2 ROIs Associated With Naming Performance

Pink regions show areas where smaller volume was associated with worse naming performance on the BNT in the cross-sectional analyses; the same ROIs were identified at both time points. Green regions show areas where greater reduction in volume (i.e., more atrophy) was associated with greater decline in naming performance in the longitudinal analysis. Abbreviations: BNT = Boston Naming Test; ROI = regions of interest.

View this table:
Table 2

LASSO Results—Cross-sectional Analyses Predicting BNT Score at Each Time Point

View this table:
Table 3

LASSO Results—Longitudinal Analysis Predicting Change in BNT Score

View this table:
Table 4

Exploratory Multivariable Models for BNT Score at Each Time Point for the Whole Group and Each Variant, Using Predictors Selected From the Cross-sectional Analyses

View this table:
Table 5

Exploratory Multivariable Models for Change in BNT Score for the Whole Group and Each Variant, Using Predictors Selected From the Longitudinal Analysis

In the cross-sectional analysis at the earlier time point, smaller volumes in the left inferior frontal gyrus pars opercularis, left superior temporal pole, left middle temporal gyrus, and left inferior temporal gyrus were associated with worse naming performance on the BNT. As proof of concept, a multivariable model created with all the selected features was significant (F (5, 56) = 24.4, p < 0.001) and explained 69% of the BNT score at the first time point and 63% after correction for optimism.27 As an additional exploratory analysis to uncover preliminary information about the relationship between these results and the 3 PPA variants, we also created separate multivariable models with the selected features for each variant. We emphasize the exploratory nature of these models and caution against overinterpretation because there is limited power with the relatively small numbers of participants in each variant group. The model for lvPPA was significant (F (5, 18) = 4.1, p = 0.012) and explained 53% of the BNT score at the earlier time point, 27% after correction for optimism. In this model, left inferior temporal gyrus volume was a predictor of BNT score that trended toward significance (p = 0.064). The model for nfvPPA was also significant (F (5, 20) = 3.4, p = 0.021) and explained 46% of the BNT score at the earlier time point and 12% after correction for optimism. Left superior temporal pole (p = 0.012) and left middle temporal gyrus (p = 0.041) volumes were significant predictors of BNT score. The model for svPPA was not significant (F (5,5) = 0.6, p = 0.722) nor were any of the individual predictors. It explained 36% of the BNT score at the earlier time point and <0% after correction for optimism.

At the later time point, as at the earlier, smaller volumes in the left inferior frontal gyrus pars opercularis, left superior temporal pole, left middle temporal gyrus, and left inferior temporal gyrus were associated with lower BNT scores. At this time point, female sex and larger cerebral to intracranial volume ratio (i.e., more overall atrophy) also were associated with worse performance. The proof-of-concept multivariable model created with all selected features was significant (F (7, 54) = 13.8, p < 0 .001) and explained 64% of the BNT score at the second time point and 54% after correction for optimism. We created exploratory models for each variant to examine the relationships at the later time point as well. The model for lvPPA was again significant (F (7, 16) = 3.3, p = 0.023) and explained 59% of the BNT score at the later time point and 2% after correction for optimism. However, none of the individual predictors were statistically significant or trended toward significance. The model for nfvPPA was also significant (F (7, 18) = 3.9, p = 0.009) and explained 60% of the BNT score at the later time point and 18% after correction for optimism. There were 3 significant predictors of BNT score: left inferior frontal gyrus pars opercularis (p = 0.027), left superior temporal pole (p = 0.049), and left middle temporal gyrus (p = 0.006) volumes. The model for svPPA was significant (F (7,3) = 39.9, p = 0.006) and explained 99% of the BNT at the later time point and <0% after correction for optimism, suggesting overfitting. Cerebral to intracranial volume ratio (p = 0.002), left inferior frontal gyrus pars opercularis volume (p = 0.009), and left inferior temporal gyrus volume (p = 0.018) were significant predictors of BNT score; sex (p = 0.088) and left superior temporal pole volume (p = 0.075) also trended toward significance.

In the longitudinal analysis relating change in volume to change in performance, increased atrophy (greater negative change) in the left supramarginal gyrus and left middle temporal pole was associated with increased behavioral decline (greater negative change on the BNT). Greater decline in naming was also associated with female sex and larger intervals between time points. The proof-of-concept multivariable model created with all selected features was significant (F (4, 57) = 9.2, p < 0 .001) and explained 39% of the BNT change and 29% after correction for optimism. As in the cross-sectional analysis, we created exploratory multivariable models with the selected predictors for each variant. The model for lvPPA was significant (F (4, 19) = 3.2, p = 0.037) and explained 40% of the BNT change and 4% after correction for optimism. One predictor trended toward significance in predicting the BNT change: the interval between time points (p = 0.074). The model for nfvPPA was significant as well (F (4, 21) = 6.8, p = 0.001) and explained 57% of the variance and 39% after correction for optimism. Sex (p = 0.040) and the interval between time points (p = 0.028) were significant predictors of BNT change. The model for svPPA was not significant (F (4, 6) = 0.6, p = 0.638) nor were any of the individual predictors. It explained 28% of the BNT score at the earlier time point and <0% after correction for optimism.

Discussion

We identified brain regions that are integral to naming performance in participants with PPA using 2 different but complementary methods. First, we performed more traditional cross-sectional analyses, relating volume in ROIs at a given time to naming scores from the same time. Comparing the results of analyses of the same participants at 2 different time points provides 1 opportunity to evaluate change over time. We were also able to look more directly at change over time with a longitudinal analysis that related change in volume in each area to change in performance. Comparing the cross-sectional and longitudinal results demonstrates the utility of each.

In the cross-sectional analyses at both the earlier and the later time points, the same ROIs were identified. Individuals with lower BNT scores had smaller volumes in the left inferior frontal gyrus pars opercularis, left superior temporal pole, left middle temporal gyrus, and left inferior temporal gyrus. These regions are classic language areas that are commonly associated with naming.2,-,4,7,-,13,21

The results of the earlier and later time point cross-sectional analyses were not identical. At the later time point, greater overall atrophy, as indexed by the ratio of cerebral volume to intracranial volume, predicted worse performance that this factor contributed significantly at the later but not the earlier time point reflects the finding that atrophy becomes more diffuse as PPA progresses.14,-,21 Those with more disease progression are in turn likely to demonstrate more impairment on cognitive tasks.

Although the cross-sectional analyses identified areas that were consistently related to naming performance, the longitudinal analysis allowed us to more directly examine where increasing atrophy predicted increasing behavioral decline. Additional atrophy in the critical areas identified in the cross-sectional analysis (left inferior frontal gyrus pars opercularis, left superior temporal pole, left middle temporal gyrus, and left inferior temporal gyrus) was not linked to additional behavioral decline. Instead, we found that naming performance suffered as the left supramarginal gyrus and left middle temporal pole shrank. These areas are part of the language network, and they are anatomically near and connected to the critical naming areas identified at each time point.2,4,10,12,13,21,22 Because atrophy spreads to these regions, their original language processing functions and any compensatory naming functions they took on are impaired. The longitudinal analysis also showed that time predicted decline: Greater behavioral decline was observed when there were longer intervals between scans. This is unsurprising given that these participants have a progressive neurodegenerative condition. It is of important that the results reveal that left supramarginal gyrus and middle temporal pole are able to maintain naming accuracy when the neural network that normally supports naming degenerates. Findings show that even in neurodegenerative disease, there can be some reorganization of structure-function relationships to compensate for focal atrophy.

One factor predicted naming decline in both analyses: female sex. In the cross-sectional analyses, female sex predicted worse performance at the later time point but not the earlier time point, suggesting that women showed greater decline on the BNT. To confirm this result, we directly compared the BNT performance of men and women at each time point with Welch 2 sample t tests. Men performed better than women at the later time point (t (58.3) = −2.5, p = 0.015), but not at the earlier time point (t (58.7) = −1.4, p = 0.16). Furthermore, the finding is in line with longitudinal results showing that women experienced greater decline in BNT performance. There was no significant difference in the interval between time points for men and women (t (59.7) = -0.28, p = 0.78), meaning that the increased decline for women is not a result of greater time between scans. The effect was numerically present in all 3 variants, although it did not reach significance in any variant (lvPPA: t (17.1) = −1.9, p = 0.072; nfvPPA: t (10.9) = −1.0, p = 0.338; svPPA: t(9.0) = −0.4, p = 0.725), suggesting that was not driven to any 1 variant. The finding that women with PPA had lower BNT scores than men is consistent with previous findings from our group49; however, in contrast to the findings here, our previous work suggested a slower decline in naming for women. The current results are more compatible with aging research suggesting that women older than 50 years experience faster decline in functional status than men.50 Differences in decline may be influenced by factors beyond the scope of this study, such as health comorbidities, differences in social support, or the general sexual dimorphism in brain development and structure observed in healthy individuals.51

Taken together, our results illustrate that directly examining atrophy over time provides a more complete picture of decline in PPA. Our cross-sectional analyses demonstrated that damage in some regions is consistently related to poor naming. However, it is not increasing atrophy in these areas that leads to further decline; our longitudinal analysis instead identified nearby but distinct regions where increased atrophy was related to decreasing performance. The areas of progressive atrophy may be additional language regions in the vicinity of the “core” language areas, and their progressive involvement likely reflects pathologic extension. It is of important that the pathologic extension resulted in further decline in naming, indicating that naming had been supported by these areas for some time. This finding not only advances our understanding of the clinical course of the disorder but also furthers our understanding of the neural instantiation of language.

An important limitation of this study was relatively small sample size because it is unusual for patients with PPA to have multiple scans with the same parameters and contemporaneous language assessment. Because of the small numbers and limited power, we were only able to create exploratory multivariable models for each variant. We were able to identify regions where volumes were individual predictors of naming score for the cross-sectional analysis that differed for the variants in the cross-sectional analysis (e.g., at the earlier time point, left inferior frontal gyrus pars opercularis volume was a trending predictor in the lvPPA group, while left superior temporal pole and left middle temporal gyrus volumes were significant predictors in the nfvPPA group). However, we caution against overinterpretation of these specific results. They provide proof of concept that it is possible to identify different areas driving effects for different variants; however, further analysis is necessary with larger numbers of participants in each variant group to draw specific conclusions. These results were presented as a starting point for further investigation. In the future, it will be interesting to conduct variant-specific analyses considering all ROIs instead of only those selected by analysis of the whole group.

This study has other limitations as well. Because of the particulars of the statistical methods used (e.g., other regions critical for naming that were highly correlated with the selected regions may have been eliminated in the LASSO regression), we cannot make claims about specific regions we did not identify. Notably, some findings could be due to atrophy in regions that are associated with the identified regions. We recognize that the 3 phenotypes of PPA are generally associated with 3 different underlying pathologies that affect different neural networks. However, evidence from several studies shows that peak atrophy in the 3 different phenotypes occurs in 3 rather large ROI,2,4 which we evaluated. It seems likely, based on the literature, that the distinct neural networks heavily and differentially depend on the associated ROI. We were not able to directly evaluate the distinct networks with the imaging we had available, so we indirectly evaluated them by evaluating atrophy in the associated ROI. Another limitation is that the treatment studies that provided data for this study did not collect biomarker data, such as APOE alleles, CSF biomarkers, or amyloid or tau PET, and few participants have had autopsies to date. Nevertheless, our results are reasonable and demonstrate the promise of the longitudinal methodology we used.

In the future, this type of longitudinal analysis should be applied to understand decline in additional linguistic and cognitive functions in PPA and areas where atrophy is correlated with particular biomarkers. That is, it should be generalizable to other cognitive functions and other neurodegenerative diseases. Future studies may allow us to better characterize decline in the different variants of PPA and better predict individual prognoses based on specific atrophy patterns. Similar analyses could benefit clinicians and patients studying and management of other neurodegenerative conditions affecting a variety of cognitive functions. With sufficient numbers, comparable analyses could use a voxel-based approach, identifying atrophy in particular voxels associated with particular functions, diseases, or biomarkers, which could provide complementary information to our parcel-based approach.

Study Funding

This work was supported by NIH Grants R01 DC014475 and R01 DC011317. The MRI equipment in this study was funded by NIH Grant 1S10OD021648.

Disclosure

A.V. Faria reports personal fees from anatomy works, outside the submitted work. The other authors have no relevant disclosures. Go to Neurology.org/N for full disclosures.

Acknowledgment

The authors gratefully acknowledge the participants and sources of research funding. We are grateful to Alexandros Afthinos, PhD, for his assistance accessing the MRI scans for analysis.

Appendix Authors

Table

Footnotes

  • Go to Neurology.org/N for full disclosures. Funding information and disclosures deemed relevant by the authors, if any, are provided at the end of the article.

  • The Article Processing Charge was funded by the authors.

  • Submitted and externally peer reviewed. The handling editor was Linda Hershey, MD, PhD, FAAN.

  • Received June 14, 2022.
  • Accepted in final form September 14, 2022.

This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

References

  1. 1.↵
    1. Mesulam MM
    . Primary progressive aphasia. Ann Neurol. 2001;49(4):425-432. ncbi.nlm.nih.gov/pubmed/11310619
    OpenUrlCrossRefPubMed
  2. 2.↵
    1. Gorno-Tempini ML,
    2. Hillis AE,
    3. Weintraub S, et al.
    Classification of primary progressive aphasia and its variants. Neurology. 2011;76(11):1006-1014. doi: 10.1212/WNL.0b013e31821103e6
    OpenUrlAbstract/FREE Full Text
  3. 3.↵
    1. Cotelli M,
    2. Manenti R,
    3. Paternicò D, et al.
    Grey matter density predicts the improvement of naming abilities after tDCS intervention in agrammatic variant of primary progressive aphasia. Brain Topogr. 2016;29(5):738-751. doi: 10.1007/s10548-016-0494-2
    OpenUrlCrossRef
  4. 4.↵
    1. Gorno-Tempini ML,
    2. Dronkers NF,
    3. Rankin KP, et al.
    Cognition and anatomy in three variants of primary progressive aphasia. Ann Neurol. 2004;55(3):335-346. doi: 10.1002/ana.10825
    OpenUrlCrossRefPubMed
  5. 5.↵
    1. Bruffaerts R,
    2. Schaeverbeke J,
    3. De Weer A-S, et al.
    Multivariate analysis reveals anatomical correlates of naming errors in primary progressive aphasia. Neurobiol Aging. 2020;88:71-82. doi: 10.1016/j.neurobiolaging.2019.12.016
    OpenUrlCrossRef
  6. 6.↵
    1. Collins JA,
    2. Montal V,
    3. Hochberg D, et al.
    Focal temporal pole atrophy and network degeneration in semantic variant primary progressive aphasia. Brain. 2017;140(2):457-471. doi: 10.1093/brain/aww313
    OpenUrlCrossRefPubMed
  7. 7.↵
    1. Leyton CE,
    2. Hodges JR,
    3. Piguet O,
    4. Ballard KJ
    . Common and divergent neural correlates of anomia in amnestic and logopenic presentations of Alzheimer's disease. Cortex. 2017;86:45-54. doi: 10.1016/j.cortex.2016.10.019
    OpenUrlCrossRef
  8. 8.↵
    1. Mesulam MM
    . Slowly progressive aphasia without generalized dementia. Ann Neurol. 1982;11(6):592-598. doi: 10.1002/ana.410110607
    OpenUrlCrossRefPubMed
  9. 9.↵
    1. Meyer AM,
    2. Faria AV,
    3. Tippett DC,
    4. Hillis AE,
    5. Friedman RB
    . The relationship between baseline volume in temporal areas and post-treatment naming accuracy in primary progressive aphasia. Aphasiology. 2017;31(9):1059-1077. doi: 10.1080/02687038.2017.1296557
    OpenUrlCrossRef
  10. 10.↵
    1. Migliaccio R,
    2. Boutet C,
    3. Valabregue R, et al.
    The brain network of naming: a lesson from primary progressive aphasia. PLoS One. 2016;11(2):e0148707. doi: 10.1371/journal.pone.0148707
    OpenUrlCrossRef
  11. 11.↵
    1. Race DS,
    2. Tsapkini K,
    3. Crinion J, et al.
    An area essential for linking word meanings to word forms: evidence from primary progressive aphasia. Brain Lang. 2013;127(2):167-176. doi: 10.1016/j.bandl.2013.09.004
    OpenUrlCrossRefPubMed
  12. 12.↵
    1. Riello M,
    2. Faria AV,
    3. Ficek B, et al.
    The role of language severity and education in explaining performance on object and action naming in primary progressive aphasia. Front Aging Neurosci. 2018;10:346. doi: 10.3389/fnagi.2018.00346
    OpenUrlCrossRef
  13. 13.↵
    1. Snowden JS,
    2. Harris JM,
    3. Thompson JC, et al.
    Semantic dementia and the left and right temporal lobes. Cortex. 2018;107:188-203. doi: 10.1016/j.cortex.2017.08.024
    OpenUrlCrossRef
  14. 14.↵
    1. Brambati SM,
    2. Amici S,
    3. Racine CA, et al.
    Longitudinal gray matter contraction in three variants of primary progressive aphasia: a tenser-based morphometry study. Neuroimage Clin. 2015;8:345-355. doi: 10.1016/j.nicl.2015.01.011
    OpenUrlCrossRefPubMed
  15. 15.↵
    1. Kumfor F,
    2. Landin-Romero R,
    3. Devenney E, et al.
    On the right side? A longitudinal study of left- versus right-lateralized semantic dementia. Brain. 2016;139(3):986-998. doi: 10.1093/brain/awv387
    OpenUrlCrossRefPubMed
  16. 16.↵
    1. Mandelli ML,
    2. Vilaplana E,
    3. Brown JA, et al.
    Healthy brain connectivity predicts atrophy progression in non-fluent variant of primary progressive aphasia. Brain. 2016;139(10):2778-2791. doi: 10.1093/brain/aww195
    OpenUrlCrossRefPubMed
  17. 17.↵
    1. Rohrer JD,
    2. Caso F,
    3. Mahoney C, et al.
    Patterns of longitudinal brain atrophy in the logopenic variant of primary progressive aphasia. Brain Lang. 2013;127(2):121-126. doi: 10.1016/j.bandl.2012.12.008
    OpenUrlCrossRef
  18. 18.↵
    1. Rohrer JD,
    2. Clarkson MJ,
    3. Kittus R, et al.
    Rates of hemispheric and lobar atrophy in the language variants of frontotemporal lobar degeneration. J Alzheimers Dis. 2012;30(2):407-411. doi: 10.3233/JAD-2012-111556
    OpenUrlCrossRef
  19. 19.↵
    1. Tetzloff KA,
    2. Duffy JR,
    3. Clark HM, et al.
    Longitudinal structural and molecular neuroimaging in agrammatic primary progressive aphasia. Brain. 2018;141(1):302-317. doi: 10.1093/brain/awx293
    OpenUrlCrossRef
  20. 20.↵
    1. Wisse LEM,
    2. Ungrady MB,
    3. Ittyerah R, et al.
    Cross-sectional and longitudinal medial temporal lobe subregional atrophy patterns in semantic variant primary progressive aphasia. Neurobiol Aging. 2021;98:231-241. doi: 10.1016/j.neurobiolaging.2020.11.012
    OpenUrlCrossRef
  21. 21.↵
    1. Rogalski E,
    2. Cobia D,
    3. Harrison TM,
    4. Wieneke C,
    5. Weintraub S,
    6. Mesulam M-M
    . Progression of language decline and cortical atrophy in subtypes of primary progressive aphasia. Neurology. 2011;76(21):1804-1810. doi: 10.1212/WNL.0b013e31821ccd3c
    OpenUrlAbstract/FREE Full Text
  22. 22.↵
    1. Faria AV,
    2. Sebastian R,
    3. Newhart M,
    4. Mori S,
    5. Hillis AE
    . Longitudinal imaging and deterioration in word comprehension in primary progressive aphasia: potential clinical significance. Aphasiology. 2014;28(8-9):948-963. doi: 10.1080/02687038.2014.911241
    OpenUrlCrossRef
  23. 23.↵
    1. Kaplan E,
    2. Goodglass H,
    3. Weintraub SL&F
    . Boston Naming Test. Lea & Febige; 1983.
  24. 24.↵
    1. Breining BL,
    2. Faria AV,
    3. Caffo B, et al.
    Neural regions underlying object and action naming: complementary evidence from acute stroke and primary progressive aphasia. Aphasiology. 2022;36(6):732-760. doi: 10.1080/02687038.2021.1907291
    OpenUrlCrossRef
  25. 25.↵
    1. Love T,
    2. Oster E
    . On the categorization of aphasic typologies: the SOAP (a test of syntactic complexity). J Psycholinguist Res. 2002;31(5):503-529. doi: 10.1023/a:1021208903394
    OpenUrlCrossRefPubMed
  26. 26.↵
    1. Chapey R
    1. Beeson P,
    2. Hillis AE
    . Comprehension and production of written words. In: Chapey R, ed. Language Intervention Strategies in Aphasia and Related Neurogenic Communication Disorders. 4th ed; 2001:572-604.
  27. 27.↵
    1. Gefen T,
    2. Teylan MA,
    3. Besser L,
    4. Pollner E,
    5. Moshkovich A,
    6. Weintraub S
    . Measurement and characterization of distinctive clinical phenotypes using the Frontotemporal Lobar Degeneration Module (FTLD-MOD). Alzheimers Dement. 2020;16(6):918-925. doi: 10.1002/alz.12098
    OpenUrlCrossRef
  28. 28.↵
    1. Obermeyer J,
    2. Schlesinger J,
    3. Martin N
    . Evaluating the contribution of executive functions to language tasks in cognitively demanding contexts. Am J Speech-language Pathol. 2020;29(1S):463-473. doi: 10.1044/2019_AJSLP-CAC48-18-0216
    OpenUrlCrossRef
  29. 29.↵
    1. Kay J,
    2. Lesser R,
    3. Coltheart M
    . Psycholinguistic assessments of language processing in aphasia (PALPA): an introduction. Aphasiology. 1996;10:159-180.
    OpenUrlCrossRef
  30. 30.↵
    1. Breining BL,
    2. Lala T,
    3. Martínez Cuitiño M, et al.
    A brief assessment of object semantics in primary progressive aphasia. Aphasiology. 2015;29(4):488-505. doi: 10.1080/02687038.2014.973360
    OpenUrlCrossRef
  31. 31.↵
    1. Mansur LL,
    2. Carthery-Goulart MT,
    3. Bahia VS,
    4. Bak TH,
    5. Nitrini R
    . Semantic memory: nouns and action verbs in cognitively unimpaired individuals and frontotemporal lobar degeneration. Dement Neuropsychol. 2013;7(1):48-54. doi: 10.1590/S1980-57642013DN70100008
    OpenUrlCrossRef
  32. 32.↵
    1. Folstein MF,
    2. Folstein SE,
    3. McHugh PR
    . “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;12(3):189-198. doi: 10.1016/0022-3956(75)90026-6
    OpenUrlCrossRefPubMed
  33. 33.↵
    1. Besser L,
    2. Kukull W,
    3. Knopman DS, et al.
    Version 3 of the National Alzheimer's Coordinating Center's uniform data set. Alzheimer Dis Assoc Disord. 2018;32(4):351-358. doi: 10.1097/WAD.0000000000000279
    OpenUrlCrossRef
  34. 34.↵
    1. Raven J,
    2. Raven JC,
    3. Court JH
    . Manual for Raven's Progressive Matrices and Vocabulary Scales. Harcourt Assessment; 2004.
  35. 35.↵
    Wechsler Memory Scale, Version 3. Psychological Corporation. Harcourt Brace & Company; 1997.
  36. 36.↵
    1. Weintraub S,
    2. Mesulam MM,
    3. Wieneke C,
    4. Rademaker A,
    5. Rogalski EJ,
    6. Thompson CK
    . The northwestern anagram test: measuring sentence production in primary progressive aphasia. Am J Alzheimers Dis Other Demen. 2009;24(5):408-416. doi: 10.1177/1533317509343104
    OpenUrlCrossRefPubMed
  37. 37.↵
    1. Rogers SL,
    2. Friedman RB
    . The underlying mechanisms of semantic memory loss in Alzheimer's disease and semantic dementia. Neuropsychologia. 2008;46(1):12-21. doi: 10.1016/j.neuropsychologia.2007.08.010
    OpenUrlCrossRefPubMed
  38. 38.↵
    1. Goodglass H,
    2. Kaplan E,
    3. Weintraub S
    . The Boston Diagnostic Aphasia Examination. Lippincott Williams & Wilkins; 2001.
  39. 39.↵
    1. Howard D,
    2. Patterson K
    . The Pyramids and Palm Trees Test: A Test of Semantic Access from Words and Pictures. Thames Valley Test Company; 1992.
  40. 40.↵
    1. Adlam A-LR,
    2. Patterson K,
    3. Bozeat S,
    4. Hodges JR
    . The Cambridge Semantic Memory Test Battery: detection of semantic deficits in semantic dementia and Alzheimer's disease. Neurocase. 2010;16(3):193-207. doi: 10.1080/13554790903405693
    OpenUrlCrossRefPubMed
  41. 41.↵
    1. Jefferies E,
    2. Patterson K,
    3. Jones RW,
    4. Lambon Ralph MA
    . Comprehension of concrete and abstract words in semantic dementia. Neuropsychology. 2009;23(4):492-499. doi: 10.1037/a0015452
    OpenUrlCrossRefPubMed
  42. 42.↵
    1. Nasreddine ZS,
    2. Phillips NA,
    3. Bédirian V, et al.
    The Montreal Cognitive Assessment, MoCA: a brief screening tool for mild cognitive impairment. J Am Geriatr Soc. 2005;53(4):695-699. doi: 10.1111/j.1532-5415.2005.53221.x
    OpenUrlCrossRefPubMed
  43. 43.↵
    1. Possin KL,
    2. Laluz VR,
    3. Alcantar OZ,
    4. Miller BL,
    5. Kramer JH
    . Distinct neuroanatomical substrates and cognitive mechanisms of figure copy performance in Alzheimer's disease and behavioral variant frontotemporal dementia. Neuropsychologia. 2011;49(1):43-48. doi: 10.1016/j.neuropsychologia.2010.10.026
    OpenUrlCrossRefPubMed
  44. 44.↵
    1. Mack WJ,
    2. Freed DM,
    3. Williams BW,
    4. Henderson VW
    . Boston Naming Test: shortened versions for use in Alzheimer's disease. J Gerontol. 1992;47(3):P154-P158. doi: 10.1093/geronj/47.3.p154
    OpenUrlCrossRefPubMed
  45. 45.↵
    MRICloud. Accessed November 4, 2022. mricloud.org
  46. 46.↵
    MRIStudio. Accessed November 4, 2022. MRIStudio.org
  47. 47.↵
    1. Tibshirani R
    . Regression shrinkage and selection via the lasso. J R Stat Soc Ser B. 1996;58(1):267-288. doi: 10.1111/j.2517-6161.1996.tb02080.x
    OpenUrlCrossRefPubMed
  48. 48.↵
    1. Meinshausen N,
    2. Yu B
    . LASSO-type recovery of sparse representations for high-dimensional data. Ann Stat. 2009;37(1):246-270. doi: 10.1214/07-AOS582
    OpenUrlCrossRef
  49. 49.↵
    glmnet: Lasso and Elastic-Net Regularized Generalized Linear Models. November 4, 2022. cran.r-project.org/web/packages/glmnet/index.html
  50. 50.↵
    SelectiveInference: tools for post-selection inference. Accessed November 4, 2022. cran.r-project.org/web/packages/selectiveInference/selectiveInference.pdf
  51. 51.↵
    1. Sebastian R,
    2. Thompson CB,
    3. Wang NY, et al.
    Patterns of decline in naming and semantic knowledge in primary progressive aphasia. Aphasiology. 2018;32(9):1010-1030. doi: 10.1080/02687038.2018.1490388
    OpenUrlCrossRef
  52. 52.
    1. Liang J,
    2. Bennett JM,
    3. Shaw BA, et al.
    Gender differences in functional status in middle and older age: are there any age variations? Journals Gerontol Ser B Psychol Sci Soc Sci. 2008;63(5):S282-S292. doi: 10.1093/geronb/63.5.S282
    OpenUrlCrossRefPubMed
  53. 53.
    1. Sun Y,
    2. Lee R,
    3. Chen Y, et al.
    Progressive gender differences of structural brain networks in healthy adults: a longitudinal, diffusion tensor imaging study. PLoS One. 2015;10(3):e0118857. doi: 10.1371/journal.pone.0118857
    OpenUrlCrossRefPubMed

Letters: Rapid online correspondence

No comments have been published for this article.
Comment

REQUIREMENTS

If you are uploading a letter concerning an article:
You must have updated your disclosures within six months: http://submit.neurology.org

Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.

If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.

Submission specifications:

  • Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
  • Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
  • Submit only on articles published within 6 months of issue date.
  • Do not be redundant. Read any comments already posted on the article prior to submission.
  • Submitted comments are subject to editing and editor review prior to posting.

More guidelines and information on Disputes & Debates

Compose Comment

More information about text formats

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
Author Information
NOTE: The first author must also be the corresponding author of the comment.
First or given name, e.g. 'Peter'.
Your last, or family, name, e.g. 'MacMoody'.
Your email address, e.g. higgs-boson@gmail.com
Your role and/or occupation, e.g. 'Orthopedic Surgeon'.
Your organization or institution (if applicable), e.g. 'Royal Free Hospital'.
Publishing Agreement
NOTE: All authors, besides the first/corresponding author, must complete a separate Publishing Agreement Form and provide via email to the editorial office before comments can be posted.
CAPTCHA
This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.

Vertical Tabs

You May Also be Interested in

Back to top
Advertisement

Related Articles

  • No related articles found.

Topics Discussed

Alert Me