Performance in Specific Language Tasks Correlates With Regional Volume Changes in Progressive Aphasia

Serena Amici, Jennifer M. Ogar, Simona M. Brambati, Bruce L. Miller, John Neuhaus, Nina Dronkers, Maria Luisa Gorno‐TempiniCognitive 2007 · DOI

TL;DR

This study uses advanced brain imaging (voxel-based morphometry) to show that performance on specific language tasks directly correlates with the location and extent of gray matter loss in patients with progressive aphasia. These findings suggest that simple language tests can serve as valuable, non-invasive tools for diagnosing different subtypes of this neurodegenerative condition.

What Problem Does This Paper Address?

Progressive aphasia (PA) is a group of neurodegenerative diseases that primarily affect language abilities. Diagnosing the specific subtype of PA can be challenging because the symptoms can overlap, and the underlying brain damage is often diffuse rather than localized like in stroke patients. Traditionally, patterns of language impairment have been used to infer brain damage, but this approach is less straightforward in neurodegenerative diseases. The researchers aimed to bridge this gap by investigating whether performance on well-established language tasks could reliably predict specific patterns of brain atrophy in PA patients, thereby aiding in differential diagnosis.

Methods

The core of this study lies in the application of voxel-based morphometry (VBM), a neuroimaging technique that allows for the statistical comparison of brain structures between groups of individuals. Here's how they used it:

  • MRI Scans: High-resolution T1-weighted MRI scans were acquired from 51 patients with progressive language impairment.
  • VBM Preprocessing: The MRI images were processed to normalize them to a standard brain template (MNI space), segment them into gray and white matter, and then smooth them to reduce noise and improve statistical analysis.
  • Statistical Analysis: VBM was used to correlate scores on four specific language tasks with regional gray matter volumes. This means they looked for areas where lower scores on a task were associated with less gray matter.
  • Region of Interest (ROI) Analysis: To increase the sensitivity of their findings, the analysis was focused on a predefined region of interest encompassing the left perisylvian region and temporal lobes, areas known to be crucial for language processing.
  • Prediction Error Analysis: To assess how well the language tasks predicted brain volume, they calculated R-squared values and used leave-one-out cross-validation to estimate prediction errors.

The language tasks used were:

  • Boston Naming Test (BNT): Assesses confrontation naming (identifying objects from pictures).
  • Western Aphasia Battery (WAB) Repetition Subtest: Measures the ability to repeat words and sentences.
  • WAB Sequential Commands Subtest: Evaluates sentence comprehension.
  • WAB Spontaneous Speech Language Fluency Subtest: Assesses language production fluency.

Datasets

The study utilized MRI scans from 51 patients diagnosed with various forms of neurodegenerative disease presenting with prominent speech and language symptoms. Of these, 41 met the clinical criteria for Primary Progressive Aphasia (PPA), including subtypes like Progressive Nonfluent Aphasia (PNFA), Semantic Dementia (SD), and Logopenic Progressive Aphasia (LPA). The remaining 10 patients had other neurodegenerative diagnoses (e.g., Frontotemporal Dementia, Alzheimer's Disease). The study did not introduce a new dataset but rather analyzed existing patient data.

Key Insights and Ideas

  • Language Tasks as Brain Maps: The most significant insight is that specific language tasks act like functional maps of the brain. Performance on these tasks isn't just about language ability; it's a direct reflection of the integrity of specific brain regions.
  • Temporal Lobes for Naming: The study strongly links difficulties in naming objects (BNT) to atrophy in the temporal lobes, particularly the inferior temporal gyri. This is crucial for understanding Semantic Dementia, where naming deficits are prominent.
  • Superior Temporal Gyrus for Repetition: Sentence repetition, a key feature of Logopenic Progressive Aphasia, was found to be specifically associated with damage to the posterior part of the left superior temporal gyrus. This highlights the role of this region in holding and processing auditory information for repetition.
  • Frontal Lobes for Comprehension and Fluency: Sentence comprehension was linked to atrophy in the dorsal parts of the left middle and inferior frontal gyri, while language fluency was associated with more ventral areas of the left middle frontal gyrus. This reinforces the role of the frontal lobes in higher-level language production and understanding.
  • Clinical Utility: The findings suggest that these relatively simple language tests, when interpreted in conjunction with neuroimaging, can provide valuable clues for differentiating between PPA subtypes, potentially improving early diagnosis and management.

Key Results

  • Naming: Impaired confrontation naming correlated with gray matter atrophy in the inferior temporal gyri, fusiform gyri, and temporal poles. The imaging models accurately predicted brain volume loss in this area (R² = 0.53).
  • Sentence Repetition: While single-word repetition showed no significant correlation, sentence repetition scores correlated with atrophy in the posterior portion of the left superior temporal gyrus. The predictive power for repetition was weaker (R² = 0.26).
  • Sentence Comprehension: Lower performance on sentence comprehension was significantly associated with gray matter loss in the dorsal portions of the left middle and inferior frontal gyri. The imaging models showed good predictive accuracy (R² = 0.58).
  • Language Fluency: Reduced language fluency correlated with decreased gray matter volume in the ventral portion of the left middle frontal gyrus, as well as superior and inferior frontal gyri. The predictive accuracy was moderate (R² = 0.36).

Limitations and Future Work

  • Sample Size and Heterogeneity: While the study included 51 patients, the authors acknowledge that the VBM technique has limitations in spatial resolution and can be susceptible to registration errors. Furthermore, the patient group was heterogeneous, including patients with PPA and other dementias, which, while increasing statistical power, means the findings might not be perfectly generalizable to a pure PPA cohort.
  • Clinical Assessment: The study notes that language tasks were primarily evaluated by a single clinician, though controversial cases were reviewed by a consensus group.
  • Future Directions: The authors suggest that these language tests, when combined with neuroimaging, can be instrumental in the differential diagnosis of PPA variants. Future work could involve validating these findings in larger, more homogenous patient groups and exploring how these correlations evolve over the course of the disease. The potential for using these tasks as biomarkers for disease progression or treatment response is also an avenue for future research.

Summary updated

Abstract

Performance on specific language tasks corresponds to regional anatomic damage in aphasia owing to neurodegenerative disorders. These language tests might be useful in the differential diagnosis of primary progressive aphasia variants that have been previously associated with damage to corresponding anatomic regions.
BibTeX
@inproceedings{amici2007performance,
  author = {Serena Amici and Jennifer M. Ogar and Simona M. Brambati and Bruce L. Miller and John Neuhaus and Nina Dronkers and Maria Luisa Gorno‐Tempini},
  title = {{Performance in Specific Language Tasks Correlates With Regional Volume Changes in Progressive Aphasia}},
  booktitle = {Cognitive},
  year = {2007},
  doi = {10.1097/wnn.0b013e31815e6265},
}
Figure