Musicianship-Related Structural and Functional Cortical Features Are Preserved in Elderly Musicians

Abstract

Background: Professional musicians are a model population for exploring basic auditory function, sensorimotor and multisensory integration, and training-induced neuroplasticity. The brain of musicians exhibits distinct structural and functional cortical features; however, little is known about how these features evolve during aging. This multiparametric study aimed to examine the functional and structural neural correlates of lifelong musical practice in elderly professional musicians.

Methods: Sixteen young musicians, 16 elderly musicians (age >70), and 15 elderly non-musicians participated in the study. We assessed gray matter metrics at the whole-brain and region of interest (ROI) levels using high-resolution magnetic resonance imaging (MRI) with the Freesurfer automatic segmentation and reconstruction pipeline. We used BrainVoyager semiautomated segmentation to explore individual auditory cortex morphotypes. Furthermore, we evaluated functional blood oxygenation level-dependent (BOLD) activations in auditory and non-auditory regions by functional MRI (fMRI) with an attentive tone-listening task. Finally, we performed discriminant function analyses based on structural and functional ROIs.

Results: A general reduction of gray matter metrics distinguished the elderly from the young subjects at the whole-brain level, corresponding to widespread natural brain atrophy. Age- and musicianship-dependent structural correlations revealed group-specific differences in several clusters including superior, middle, and inferior frontal as well as perirolandic areas. In addition, the elderly musicians exhibited increased gyrification of auditory cortex like the young musicians. During fMRI, the elderly non-musicians activated predominantly auditory regions, whereas the elderly musicians co-activated a much broader network of auditory association areas, primary and secondary motor areas, and prefrontal and parietal regions like, albeit weaker, the young musicians. Also, group-specific age- and musicianship-dependent functional correlations were observed in the frontal and parietal regions. Moreover, discriminant function analysis could separate groups with high accuracy based on a set of specific structural and functional, mainly temporal and occipital, ROIs.

Conclusion: In conclusion, despite naturally occurring senescence, the elderly musicians maintained musicianship-specific structural and functional cortical features. The identified structural and functional brain regions, discriminating elderly musicians from non-musicians, might be of relevance for the aging musicians' brain. To what extent lifelong musical activity may have a neuroprotective impact needs to be addressed further in larger longitudinal studies.

Keywords: aging; auditory cortex; elderly; fMRI; functional; musicians; musicianship; structural.

FIGURE 1

Musicianship-dependent scores. Average musicianship scores of young musicians (YMs), old musicians (OMs), and old non-musicians (ONMs): (A) tonal, rhythm, and total scores of the AMMA test; (B) total duration of musical activity (y) across the lifespan; (C) mean intensity of musical activity (h/w) presented for the lifespan; (D) cumulative musical practice (y*h/w) across the lifespan; (E) mean intensity of musical activity (h/w) presented for each decade separately, and error-bars represent SEM. (A–D) Whiskers in the boxplots represent data range. For more details, see corresponding Table 1. YMs, young musicians; OMs, old musicians; ONMs, old non-musicians; AMMA, Advanced Measures of Music Audiation test; y, years; h/w, hours/week; y*h/w, years*hours/week. **Significance level p < 0.001; *significance level p < 0.01; n.s., not significant.

FIGURE 2

Average structural metrics. Group averaged maps for (A) thickness, (B) volume, and (C) surface area projected on group-averaged inflated pial surfaces. Color scales present smaller values in red and higher values in blue (set up using FreeView). See Supplementary Table 3 for more details. YMs, young musicians; OMs, old musicians; ONMs, old non-musicians; THK, thickness; VOL, volume; SA, surface area; LH/RH, left hemisphere/right hemisphere.

FIGURE 3

Age- and musicianship-dependent structural differences. Age-dependent effects: clusters that showed significant changes with age in (A) thickness, (B) volume, and (C) surface area between the specific groups (output from FreeSurfer age-slope difference analysis). Musicianship-dependent effects: (D) cluster that showed significant association between the mean intensity of musical activity (h/w) and volume in the YMs (results from covariance analysis); (E) cluster that showed significant surface area differences between the YMs and the OMs with changes in AMMA tonal score (output from FreeSurfer AMMA-slope difference analysis). All depicted clusters evolved from the general linear model (GLM) analysis. The multiple comparison correction method is Monte Carlo 10,000 simulations with a setup clustering threshold of log10(p-value), which was set at p = 0.05; hence the color bar indicates positive (red/yellow) vs. negative (blue/turquoise) significant values. Cluster details can be found in Table 2. Note that the depicted brain clusters only refer to specific group differences found in the GLM (age-slope analysis: A–C; covariance analysis: D; AMMA-slope analysis: E), while the scatterplots include values for all the three groups. Scatterplots that included all the three groups were only inserted for descriptive purposes and better readability of the data and were not part of the GLM analysis. The gray area around the lines represents the 95% confidence level interval for predictions from a linear model (“lm”). YMs, young musicians; OMs, old musicians; ONMs, old non-musicians; THK, thickness in mm; VOL, volume in mm³; SA, surface area in mm²; LH/RH, left hemisphere/right hemisphere; AMMA, Advanced Measures of Music Audiation test; y, years, h/w, hours/week.

FIGURE 4

Inter-individual shape differences of the auditory cortex (AC). (A) Subsample of Supplementary Figure 1 demonstrating representative AC surfaces of six exemplary participants, and depicting the characteristic individual shapes of the left (blue) and right (red) AC within each group. Further AC reconstructions of all the subjects are presented in Supplementary Figure 1. (B) Observed frequencies of the two defined AC types, single and duplicated Heschl’s gyrus (HG), are counted separately per hemisphere. A more detailed definition of HG morphotypes can be found in Supplementary Figure 1B. LH/RH, left hemisphere/right hemisphere; YMs, young musicians; OMs, old musicians; ONMs, old non-musicians; HG, Heschl’s gyrus; n.s., not significant (contrast).

FIGURE 5

Functional group activations and correlation analyses. (A) Functional results for the attentive tone listening functional MRI (fMRI) task: activation maps from contrast tone listening vs. baseline depicted for each group (YMs, OMs, and ONMs). The results are presented at the threshold FDR (p < 0.01) and projected on the inflated pial surface of a standard average brain. Red-yellow represents voxels showing increased activity during task (task > baseline). Additional group contrast results can be found in the Supplementary Tables 5–7. (B–D) Results from the random effects (RFX) group analysis. Only views with significant clusters are presented. Depicted are functional clusters correlating with: (B) age of musicians (YMs and OMs); (C) cumulative musical practice (y*h/w) in the elderly (OMs and ONMs); (D) AMMA tonal score in the elderly (OMs and ONMs). The results have been corrected at cluster level (p < 0.02). Red-yellow represents positive and blue-green negative correlations, respectively. For more details, see Table 3. YMs, young musicians; OMs, old musicians; ONMs, old non-musicians; LH/RH, left hemisphere/right hemisphere; RFX, random effects group analysis; AMMA, Advanced Measures of Music Audiation test; M1, primary motor cortex; SMA, supplementary motor area; PMA, premotor area; S2, secondary somatosensory cortex; IPL, inferior parietal lobule; SMG, supramarginal gyrus; HG, Heschl’s gyrus; STG, superior temporal gyrus; MTG, middle temporal gyrus; ITG, inferior temporal gyrus; VPMC, ventral premotor cortex; LPFC, lateral prefrontal cortex; MFG, middle frontal gyrus; PreCun, Precuneus; Cun, Cuneus; ACC, anterior cingulate cortex; PCC, posterior cingulate cortex; CMA, cingulate motor area.

FIGURE 6

Discriminant function analysis based on structural ROIs. ROIs that separated the groups based on structural metrics. (A) Visualization of the ROIs that resulted to be significant in the discriminant function analysis. The left and right hemispheres as well as lateral and medial views are presented based on the FreeSurfer Desikan atlas. The color code (yellow = THK, violet = VOL, and turquoise = SA) indicates the respective GM metric in the specific ROI. (B) Discriminant function analysis plot visualizing the separation between the young and old subjects on function 1 and musicians and non-musicians on function 2 based on Desikan atlas structural ROIs. For more details, see Table 4. Based on the structural ROIs in panel A, the first discriminant function discriminated best between the OMs/ONMs and the YMs, i.e., young vs. old, while the second function differentiated the OMs from the ONMs, i.e., musicians vs. non-musicians. YMs, young musicians; OMs, old musicians; ONMs, old non-musicians; LH/RH, left hemisphere/right hemisphere; THK, thickness; VOL, volume; SA, surface area; ROI, region of interest; group mean, mean variate score for each group called centroid in the discriminant analysis; STG, superior temporal gyrus; SMG, supramarginal gyrus; aHG, anterior Heschl’s gyrus.