In vivo visualization of age-related differences in the locus coeruleus

Abstract

The locus coeruleus (LC), the major origin of noradrenergic modulation of the central nervous system, may play an important role in neuropsychiatric disorders including Parkinson's disease and Alzheimer's disease. The pattern of age-related change of the LC across the life span is unclear. We obtained normalized, mean LC signal intensity values, that is, contrast ratios (CRs), from magnetization transfer-weighted images to investigate the relationship between LC CR and age in cognitively normal healthy adults (N = 605, age range 18-88 years). Study participants were part of the Cambridge Centre for Ageing and Neuroscience-an open-access, population-based data set. We found a quadratic relationship between LC CR and age, the peak occurring around 60 years, with no differences between males and females. Subregional analyses revealed that age-related decline in LC CR was confined to the rostral portion of the LC. Older adults showed greater variance in overall LC CR than younger adults, and the functional and clinical implications of these observed age-related differences require further investigation. Visualization of the LC in this study may inform how future scanning parameters can be optimized, and provides insight into how LC integrity changes across the life span.

Keywords: Aging; Locus coeruleus; Magnetic resonance imaging; Neuromelanin; Noradrenergic system.

Fig. 1

Postmortem studies have reported an inverted U-shaped (A, B) or linear (C) pattern of neuromelanin accrual in LC neurons, and a linear decline (B) or no change (D) in LC neuron number with increasing age. Graph A (reproduced with permission from Mann and Yates, 1974) shows mean melanin content of LC neurons against age from 45 human cases (age range 0–91 years). Participants did not have neurological illness or abnormal neuropathological findings. Graph B (reproduced with permission from Manaye et al., 1995) shows the total number of LC neurons (neuromelanin [NM] and tyrosine hydroxylase [TH] cell counts) on one side of the brain against age from 17 human cases (age range 1–104 years). Fourteen brains were from individuals with no known neurological or psychiatric disease, two had psychiatric illness and one had Down's syndrome. Graph C (reproduced with permission from [Zucca et al., 2006]) shows the mean neuromelanin content in LC neurons against age from healthy adults (age range 14–97 years) with no known neuropsychiatric or neurodegenerative disease. Graph D (reproduced with permission from Ohm et al., 1997) shows the stereological LC cell count from one side of the brain against age from 20 healthy adults (age range 49–98 years). Abbreviation: LC, locus coeruleus.

Fig. 2

Cam-CAN MT-weighted group template images showing LC (axial B,D, coronal C,E) with LC segmentations (D,E in red). A multivariate template approach was used to create an MT-weighted (F) and T1-weighted (G) group image. The substantia nigra could be visualized on the MT-weighted group template (A). In (B) and (F), two areas of signal hyperintensity are visible which correspond to the neuroanatomical positions of right and left LC in the lateral floor on the 4th ventricle in (G). The left LC can be seen at the crosshair in (F) and its position in (G) is indicated by the corresponding crosshair. Segmentation of these areas was performed on the axial slices on the MT-weighted template (D). The substantia nigra, a region with similar macromolecular properties to the LC, could also be visualized as two areas of signal hyperintensity on the MT-weighted group template (A). Abbreviations: LC, locus coeruleus; Cam-CAN, Cambridge Centre for Ageing and Neuroscience; MT, magnetization transfer. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 3

Cam-CAN MT-weighted images from three individual participants with LC and reference region (pons) segmentations (second row, red). The number above each image is the mean LC CR value for an individual participant, demonstrating that the LC can be visualized on individual scans as two areas of signal hyperintensity (first row) across the range of LC CR values, shown in axial (left) and coronal (right) views. In the second row, the large red square represents the reference region (pons). These images for each participant were visually assessed to ensure correct identification of the LC and to detect any significant movement artifacts. Abbreviations: CR, contrast ratio; LC, locus coeruleus; Cam-CAN, Cambridge Centre for Ageing and Neuroscience; MT, magnetization transfer. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 4

Spatial overlap between the Cam-CAN LC mask and three other published LC masks in MNI space. The Cam-CAN LC mask (white) showed spatial overlap with LC masks from three previous studies (Betts et al., 2017, Dahl et al., 2018, Keren et al., 2009), overlapping 94% relative to the Dahl mask (top row, green), 48% relative to the Keren 1SD mask (middle row, red), and 19% relative to the Betts mask (bottom row, blue). As was the case for the Dahl mask (Dahl et al., 2018), it can be observed that the Cam-CAN LC mask does not extend as far caudally as did the Keren and Betts masks, explaining the relatively lower overlap, and the Betts mask (which had the lowest overlap) did not extend as far rostrally as did the other masks. Abbreviations: LC, locus coeruleus; Cam-CAN, Cambridge Centre for Ageing and Neuroscience. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 5

Graphs showing the relationship between mean LC CR (A, B, E), raw LC mean signal intensity (C), or mean pontine reference region signal intensity (D) and age in years. Unfilled circles (A, B, C, D) represent individual data points (n = 605); solid red circles represent males (n = 295) and solid blue circles represent females (n = 310) (E). In A, a quadratic model (adjusted R² = 0.126, p < 0.001) provided a better fit than a linear model (adjusted R² = 0.086, p < 0.001) according to the AIC. The quadratic relationship remained significant (adjusted R² = 0.181, p < 0.001) after controlling for reference region signal intensity and TR scan parameter (30 or 50 ms). In (B), a post hoc “two-lines” test (proposed to be a valid alternative to testing a U-shaped relationship) provided a breakpoint of 57 years but the second slope was not individually significant (p = 0.92); not fully supporting the presence of a U-shaped relationship between LC CR and age. This may have been related to a smaller age range and higher variance (F = 1.46, p < 0.001) in the “older” (>57 years) versus “younger” group (<57 years). C shows a nonsignificant, weak, negative linear relationship between the raw LC signal intensity mean scores and age (r = −0.039, adj R² = 0.0002, p = 0.341) and (D) shows the observed significant negative linear relationship between reference region (pontine) signal intensity and age (r = −0.1, adj R² = 0.005, p = 0.049). In (E), no significant difference in mean LC CR was found between males (n = 295, solid red circles) and females (n = 310, solid blue circles) using a Bayesian approach. Abbreviations: LC, locus coeruleus; CR, contrast ratio; TR, repetition time; AIC, Akaike information criterion. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 6

Graphs showing regional age-related differences in LC CR, plotting maximum (A) or mean (B, C, D) LC CR for rostral/caudal regions of LC and age in years. Solid red and blue circles (A and B) represent individual data points for rostral and caudal LC CR values, respectively. Unfilled circles (C and D) represent individual data points (n = 605). Age-related differences were observed for maximum (A) and mean (B) LC CR within rostral and caudal regions of LC, and the rostral region showed significantly higher LC CR values than the caudal region. Quadratic models were a better fit for the data compared with a linear model according to the AIC. A subsequent two-lines test revealed an inverted U-shaped relationship (with breakpoints shown in green) between mean LC CR and age in the rostral region but not in the caudal region, as both the upward (p < 0.001) and downward (p = 0.0426) slopes for the former reached statistical significance. Abbreviations: LC, locus coeruleus; CR, contrast ratio; AIC, Akaike information criterion. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)