OUP user menu

Preservation of midbrain catecholaminergic neurons in very old human subjects

N. Kubis , B. A. Faucheux , G. Ransmayr , P. Damier , C. Duyckaerts , D. Henin , B. Forette , Y. Le Charpentier , J.-J. Hauw , Y. Agid , E. C. Hirsch
DOI: http://dx.doi.org/10.1093/brain/123.2.366 366-373 First published online: 1 February 2000


Parkinson's disease is characterized by a progressive degeneration of dopaminergic neurons in the midbrain, yet the cause of this neuronal loss is still unknown. It has been hypothesized that Parkinson's disease could be the consequence of accelerated ageing. In order to reveal a possible common process during ageing and Parkinson's disease neurodegeneration, catecholaminergic neurons of five anatomical regions of the brainstem (substantia nigra, central grey substance, ventral tegmental area, peri- and retrorubral area, and locus coeruleus) have been quantified using immunohistochemical staining for tyrosine hydroxylase (TH) on regularly spaced sections, between the rostral and caudal poles of the mesencephalon and in the rostral pole of the pons, in post-mortem samples of 21 control subjects who died at ages 44–110 years. No statistically significant loss of TH positive neurons was observed in the older subjects, either in the substantia nigra or in the other midbrain regions that are known to degenerate to a lesser degree in Parkinson's disease. Furthermore, in the later regions no neuronal loss was observed from age 44 to 80 years, indicating that this result is not dependent on the inclusion of `supernormal' very old people. These results suggest that from age 44 to 110 years, ageing in control adults is not, or is scarcely, accompanied by catecholaminergic cell loss in the midbrain and hence Parkinson's disease is probably not caused by an acceleration of a degenerative process during ageing.

  • substantia nigra
  • tyrosine hydroxylase
  • ageing
  • human
  • A8 = peri- and retro-rubral catecholaminergic cell group A8
  • AChE = acetylcholinesterase
  • CGS = central grey substance
  • LC = locus coeruleus
  • SNpc = substantia nigra pars compacta
  • TH = tyrosine hydroxylase
  • VTA = ventral tegmental area


The loss of nigral dopaminergic neurons in Parkinson's disease leads to a striatal dopamine depletion responsible for the major symptoms of the disease (Hornykiewicz, 1966). Although the dopaminergic neurons of the substantia nigra pars compacta (SNpc) are mostly affected, other cell nuclei, such as the ventral tegmental area (VTA), the peri- and retrorubral catecholaminergic cell group A8 (A8) and the locus coeruleus (LC) also degenerate, but to a lesser extent. In contrast, catecholaminergic neurons are relatively preserved in the central grey substance (CGS). The late onset of the disease and its long duration for most patients suggest that ageing associated events may contribute to pathological cell death mechanisms. Indeed, during normal human ageing there is a progressive motor impairment which has been associated with nigrostriatal dysfunction (Newman et al., 1985), while biochemical studies have shown that dopaminergic systems are altered during normal ageing (Calne and Peppard, 1987; Scherman et al., 1989; De Keyser et al., 1990; Rinne et al., 1990). Lewy bodies, which are hyaline neuronal inclusions that are detected at pathological examination in the SNpc and the LC of parkinsonian patients, are sometimes also found at autopsy in the brain of aged individuals who died without neurological or psychiatric illness (Smith et al., 1991). Some post-mortem studies have also demonstrated a loss of nigral neurons during ageing (Hirai, 1968; Mann and Yates, 1983; Fearnley and Lees, 1991; Gibb and Lees, 1991), while other studies have suggested that the pathogenesis of Parkinson's disease is not linked to ageing. Epidemiological studies have shown that the frequency of Parkinson's disease does not increase with age (Koller et al., 1987) and that levodopa therapy does not alleviate the motor disturbances associated with ageing, indicating that these disorders should not be the consequence of lesions in dopaminergic nerve terminals (Newman et al., 1985). In vivo studies using PET scans have shown that the spatial distribution of [18F]fluorodopa uptake in the striatum differs between normal aged individuals and Parkinson's disease patients: [18F]fluorodopa uptake is severely impaired in the posterior part of the putamen of Parkinson's disease patients (Brooks et al., 1990), while in healthy aged subjects it has been reported not to be impaired in the caudate nucleus and the putamen (Sawle et al., 1990) or in the whole striatum (Martin et al., 1989). In addition, Lewy bodies, which are considered as hallmarks of Parkinson's disease, have not been observed in the brains of aged non-human primates (Price et al., 1991). No loss of nigral neurons with advancing age has been reported in some post-mortem studies (McGeer et al., 1977; Morris et al., 1989; Thiessen et al., 1990; Muthane et al., 1998), but other studies (Hassler, 1938; Fearnley and Lees, 1991) have shown that there is a differential regional cell loss within the SNpc between controls and patients: in elderly control subjects the predominant cell loss occurs in the dorsal part of the SNpc (SNpc-d), but in patients with Parkinson's disease the cell loss takes place mainly in the ventral part of the SNpc (SNpc-v). The discrepancies between results reported in studies on nigral neuronal loss during normal ageing are probably due, in part, to differences in the methods used and in the case samples examined. In this study, we quantified catecholaminergic neurons in the midbrains of a group of control subjects with ages at death ranging from 44 to 110 years. The total number of dopaminergic neurons located in the whole SNpc was measured as well as those in the SNpc-d and SNpc-v, and in the other catecholaminergic regions that are less affected in Parkinson's disease (not previously investigated): VTA, A8, CGS and LC. For each of these regions an estimate of the total number of neurons stained for tyrosine hydroxylase (TH), using immunohistochemical procedures, was calculated by integration after computer-assisted counting of all immunoreactive neurons, on regularly spaced sections between the rostral and the caudal poles of the midbrain.



Brainstems of 36 subjects who had died without drug abuse, neurological or psychiatric disease history, as evidenced by retrospective clinical chart analysis, were collected. Fifteen were excluded after pathological examination because of tissue alteration, incomplete dissection or macroscopical evidence of neurological or vascular disease that had not been diagnosed during life-time. The 21 remaining subjects had no Lewy bodies and no significant counts of senile plaques or neurofibrillary tangles. Their age at death ranged from 44 to 110 years. There were 11 females and 10 males. Post-mortem delay (time between death and tissue fixation) ranged from 8.5 to 56 h (mean time 17 h) (Table 1).

View this table:
Table 1

Characteristics of the patients

CaseSexAge at death (years)Immediate cause of death
1M 44Acute myocardial infarct
2F 47Cancer
3M 53Acute myocardial infarct
4M 63Sepsis
5M 68Sepsis
6M 70Cardiac insufficiency
7M 70Peritonitis
8F 71Acute myocardial infarct
9M 73Cardiac insufficiency
10F 82Acute myocardial infarct
11F 82Cardiac insufficiency
12M 83Cardiac insufficiency
13M 85Peritonitis
14M 85Cardiac insufficiency
15F 86Cardiac insufficiency
16F 87Pneumonitis
17F 91Cancer
18F 99Undetermined
19F101Pulmonary embolism, cardiac insufficiency
20F101Pulmonary embolism, sepsis

Tissue preparation

Within 2 h after autopsy the brain was hemisected and one hemisphere was fixed in formalin for pathological examination. The brainstem was removed from the other hemibrain at the level of the mammillary bodies and cut into two blocks at the level of the LC. The tissue blocks were fixed by immersion for 72 h at 4°C in 4% (w/v) paraformaldehyde and 15% (v/v) saturated picric acid in 0.1 M potassium phosphate buffer at pH 7.4, then washed in 0.1 M potassium phosphate buffer at pH 7.4, containing consecutively 0, 5, 10, 15 and 20% (w/v) of sucrose (24 h each). They were then frozen in dry-ice reduced to powder and stored at –80°C until processing. The blocks were cut into 40 μm serial sections using a sliding microtome, the plane of the sections being perpendicular to the rostrocaudal axis of the mesencephalon. The sections were immersed free-floating into 0.25 M Tris–HCl, 0.9% NaCl at pH 7.4 (Tris-buffered saline) containing 0.1% (w/v) sodium azide, and stored at 4°C.


TH immunohistochemistry was performed on tissue sections regularly spaced (every 36th section) along the whole mesencephalon, from the rostral pole of the SNpc to the rostral third of the LC. Mouse monoclonal antibody directed against human TH (INCSTAR, Sorin Biomedica, Brussels, Belgium) was used. Sections were pretreated with: (i) 20% (v/v) methanol and 0.9% (v/v) hydrogen peroxide for 5 min, to inhibit endogenous peroxidase; (ii) 0.2% (v/v) Triton X-100 for 5 min, to facilitate antibody penetration; and (iii) normal goat serum diluted 1 : 30 in 0.25 M Tris-buffered saline for 30 min, to inhibit non specific binding. Incubation with primary antibody was performed at 4°C for 48 h with gentle agitation, at a 1 : 250 dilution in 0.25 M Tris-buffered saline solution containing 1% (v/v) normal goat serum, 1% (v/v) normal human serum and 0.01% (w/v) thimerosal. Immunoreactivity in tissue sections was revealed by the avidin–biotin method (Vectastain anti-mouse kit, Vector Laboratories, Burlingame, Calif., USA). The horseradish peroxidase enzyme attached to the secondary antibody was revealed by incubation of the sections in 3,3′-diaminobenzidine solution (0.05 M). To test non-specific staining due to the secondary antibody, some sections were incubated without the primary antibody; no staining was observed under such conditions.

Ubiquitin immunohistochemistry was performed to detect Lewy bodies. Two sections (one taken at the caudal level of the SNpc and one taken at the level of the LC were immunostained for each subject with a rabbit polyclonal antibody directed against ubiquitin (Dako, Trappes, France). Immunolabelling was detected by the double-bridge peroxidase-antiperoxidase (PAP) method (Vacca, 1982) using 3,3′-diaminobenzidine as a co-substrate.

Acetylcholinesterase histochemistry

The limits of the anatomical regions were identified by acetylcholinesterase (AChE) histochemistry performed on sections that were adjacent to those treated for TH immunohistochemistry. A slightly modified version of the Geneser–Jensen and Blackstad method was used (Graybiel and Ragsdale, 1978). The limits of the red nucleus, medial lemniscus and regions of high or low AChE staining were drawn, stored in a computer and projected on to cartography of TH immunostained sections.

Anatomical regions

The study was performed in five anatomical regions which were delineated as described by Hirsch and colleagues (Hirsch et al., 1988). The arbitrary limits were as follows: the SNpc was distinguished from the VTA by a line parallel to the midline of the brain passing by the midwidth of the red nucleus and more caudally the decussation of the brachium conjunctivum; the VTA was located medially to this line and ended at the level of the CGS dorsally; dopaminergic neurons located in the medial lemniscus and above it, in the midbrain tegmentum, were considered as part of A8; CGS was located around the sylvian aqueduct; the LC was analysed through its rostral third, which extended 4.32 mm from the rostral pole of the nucleus (identified by TH immunostaining). The SNpc-d and SNpc-v were also analysed separately and two different types of criteria were used in order to compare them. For one set of analyses, the SNpc was arbitrarily divided into a dorsal and a ventral portion of equal width, along the long axis of the structure (Fig. 1A). For the second set of analyses, the SNpc was divided into a dorsal part, a ventrolateral part and a ventromedian part, according to the nigral subdivisions and criteria defined by Fearnley and Lees (Fearnley and Lees, 1991) (Fig. 1B).

Fig. 1

Charts of TH-positive neurons in the mesencephalon of representative control subjects aged 44 years (A), 70 years (B) and 101 years (C). The LC, which is more caudally located, is not represented.

Measurement of the mean area of the neuronal cell bodies

The mean size of the soma of neurons stained for TH was estimated in each region by the measurement of the cross-sectional surface of 40 immunostained neurons (when possible) in each anatomical region for each of the 21 individuals.

Image analysis

For each brain, xy plots of every TH-positive neuron were obtained for regularly spaced (1.44 mm) transverse sections, using a computer-assisted microscope system (Historag, Biocom, Les Ulis, France). With this system, each stained neuron was plotted only once. The total number of the TH-positive neurons located in the defined anatomical region was counted from the most rostral level (sections including the subthalamic nucleus) to a caudal level (corresponding to the midlength of the LC). These values were then plotted against the cumulative distance between the sections and the surface under this curve was taken as an estimate of the total number of neurons in each region from its rostral to its caudal extent.

Abercrombie correction

The total number of neurons was counted using the Abercrombie formula with a section thickness of 40 μm and a cross-section diameter of the cells measured from the mean area of the cells determined, as described above (Abercrombie, 1946).

Statistical analysis

A possible statistical correlation was checked between measurements of the mean area of the neuronal somas and the age of the individuals. Statistical relationships between the numbers of catecholaminergic neurons and the age of the individuals was analysed in the whole SNpc as well in the SNpc-d and SNpc-v, and in the VTA, A8, CGS and LC using both linear regression and polynomial regression analyses for all subjects from 44 to 110 years. Statistical analyses were performed using Sigmastat Statistical Software (version 2.0, SPSS Inc., Chicago, Ill., USA).


Cell body size

The mean size of the TH-positive cell body measured for neurons, located in any of the five studied regions, was not significantly correlated with age. In this population of control subjects, no shrinkage of neuronal cells was thus observed with ageing.

Analyses of cell counts as a function of age

The distribution of catecholaminergic neurons is shown in the mesencephalon of representative control subjects aged 44, 70 and 101 years (Fig. 1). No apparent loss of neurons is seen in the mesencephalon of the older subjects. Using linear regression analyses, no statistically significant variations in the numbers of catecholaminergic neurons were found with age, whichever anatomical region was studied (Figs 2 and 3; Table 2). The results observed for SNpc-d and SNpc-v were similar whether the anatomical limits were those defined by Fearnley and Lees (Fearnley and Lees, 1991) or those obtained with arbitrary delimitations. In the LC there was a trend for the number of noradrenergic neurons to decrease with advancing age, although statistical level of significance was not reached.

View this table:
Table 2

Linear regression of estimated numbers of TH immunoreactive neurons located in the whole SNpc, SNpc-d, SNpc-v, VTA, A8, CGS and LC, as a function of age of 21 individuals with ages ranging from 44 to 110 years

Anatomical regionCorrelation coefficient (r)Linear regression (y = ax + b)P
Nv = estimated number of TH immunoreactive neurons, calculated according to Abercrombie, with a correction for neuronal size and section thickness; NS = non-significant.
SNpc0.33Nv SNpc = −527 × age + 187 480NS (0.14)
SNpc-d0.12Nv SNpc-d = 114 × age + 39 082NS (0.61)
SNpc-v0.34Nv SNpc-v = −476 × age + 133 298NS (0.13)
VTA0.32Nv VTA = −265 × age + 54 473NS (0.16)
A80.06Nv A8 = −22 × age + 23 731NS (0.79)
CGS0.20Nv CGS = 24 × age + 1793NS (0.38)
LC0.16Nv LC = −14 × age + 3380NS (0.52)
Fig. 2

Number of TH immunoreactive neurons in the SNpc (A), VTA (B), A8 (C), CGS (D) and LC (E), as a function of age.

Fig. 3

Number of TH immunoreactive neurons in the SNpc-d (A) and SNpc-v (B), as a function of age.

When the variation of cell counts was examined using polynomial analyses, a statistically significant U-shaped relationship between age and cell counts was observed for data of the SNpc taken as a whole (P < 0.03), the SNpc-d (P < 0.02) and the SNpc-v (P < 0.04). No statistically significant relationship was found for the other regional quantifications (Table 3).

View this table:
Table 3

Polynomial regression, order 2, of estimated numbers of TH immunoreactive neurons located in the whole SNpc, SNpc-d, SNpc-v, VTA, A8, CGS and LC, as a function of age of 21 individuals with ages ranging from 44 to 110 years

Anatomical regionCorrelation coefficient (r)Polynomial regression, order 2 (y = ax2 + bx + c)P
Nv = estimated number of TH immunoreactive neurons, calculated according to Abercrombie, with a correction for neuronal size and section thickness; NS = non-significant.
SNpc0.57Nv SNpc = 37 age2 − 6096 × age + 388 259<0.03
SNd0.59Nv SNpc-d = 27 age2 + 3976 × age + 186 553<0.02
SNv0.56Nv SNpc-v = 30 age2 − 5044 × age + 297 984<0.04
VTA0.32Nv VTA = 0.9 age2 − 401 × age + 59 390NS (0.38)
A80.45Nv A8 = 8 age2 − 1247 × age + 67 891NS (0.13)
CGS0.22Nv CGS = −0.5 age2 + 93 × age − 708NS (0.65)
LC0.23Nv LC = −0.7 age2 + 95 × age − 494NS (0.67)


The major finding of this study is the absence of any statistically significant loss of mesostriatal dopaminergic neurons in the SNpc from age 44 to 110 years, in both dorsal and ventral parts, and in the other regions also known to degenerate in Parkinson's disease such as in VTA and A8. These data are in agreement with a recent report showing that no neuronal loss occurred in the SNpc with ageing in a group of Indian subjects (Muthane et al., 1998). Discrepancies between some results reported in previous studies may be partly explained in terms of methodological differences. For example, some studies have been performed at a single mesencephalic level of the SNpc. In addition, the planes of sections and the rostrocaudal levels were not identical for each subject and in all studies (Morris et al., 1989; Thiessen et al., 1990; Gibb and Lees, 1991; Muthane et al., 1998). Our data were obtained from quantifications of regularly spaced levels covering the whole span of the mesencephalic regions, and all neurons that were immunoreactive for TH were counted. Moreover, the limits of the anatomical boundaries of the SNpc have not been clearly defined in most published articles, except for the study performed by Fearnley and Lees (Fearnley and Lees, 1991). Another reason for discrepancy between studies is the estimation of the neuronal cell loss, which is obtained from cellular density quantifications, which does not account for the progressive cerebral atrophy that occurs during normal ageing and overestimates the number of counted cells. We tried to avoid this by quantifying neurons from regularly spaced sections between the rostral and caudal poles of the cell groups and by using Abercrombie correction. We also ascertained that there was no cell shrinkage with age in our population, which could have led us to underestimate the number of neurons. Moreover, in many studies neuromelanin content was used to identify dopaminergic neurons, but all dopaminergic neurons of the brainstem do not contain visible neuromelanin and quantification from TH immunoreactivity is more accurate (Hirsch et al., 1988). We did not observe any fainting or exacerbation of TH immunoreactivity in our subjects with advancing age. Finally, the inclusion of presymptomatic Parkinson's disease patients in our study of normal ageing was avoided by checking for the absence of Lewy bodies in the SNpc and LC of our subjects. Our data show (i) a continuous `decrease' in the number of dopaminergic neurons located in the SNpc for individuals younger than 80 years, and (ii) higher counts in older individuals, especially in the centenarians.

The `absence' of neuronal loss shown in elderly subjects of over 80 years may result from inter-individual differences. We could hypothesize that a specific subgroup of people born with a different initial number of neurons was selected and that their rate of neuronal loss was similar to that of younger individuals. However, such an explanation seems unlikely since it would imply a great variability in the number of neurons measured in the younger subjects, a result not observed by us or by Fearnley and Lees (Fearnley and Lees, 1991). Higher cell counts in very old subjects may reflect a higher resistance to death of their catecholaminergic neurons, which would make them `super-normal individuals'. They would be better protected against the neurodegenerative processes associated with ageing and thus be resistant to senescence. This explanation is supported by observations made in infra-human primates. Indeed, in Microcebus murinus it has been shown that the number of neurons located in the basal forebrain cholinergic column was higher in very old than in middle-aged animals (Mestre and Bons, 1993). Similar results have been found in studies estimating the mortality rate of medflies according to their age (Carey et al., 1992; Curtsinger et al., 1992); the death rate among flies increased until late middle-age and then levelled off.

A sex-linked contribution might be involved since men and women were not equally represented in our sample with advancing age. However, a MRI study of morphometric changes with normal ageing in the human mesencephalon (Doraiswamy et al., 1992) showed no relationship between gender and age, although differences between males and females have already been reported in cortical regions (Cowell et al., 1994).

Our study does not confirm the results of studies that show a continuous neuronal loss in the SNpc-v and a more severe involvement of the SNpc-d during normal ageing (Hassler, 1938; German et al., 1988; Fearnley and Lees, 1991; Gibb et al., 1991). However, it is in agreement with and confirms the observations reported by Fearnley and Lees between the fifth and the end of the ninth decade (Fearnley and Lees, 1991). When cell counts reported by these authors and cell numbers estimated by us are analysed for the subjects aged from 44 to 91 years, linear regression analyses show a negative association between the numbers of neurons in the SNpc and age, with a statistically significant coefficient of correlation (Fearnley and Lees: n = 32, r = −0.50, P < 0.01; our study: n = 17, r = −0.52, P < 0.05) and regression (Fearnley and Lees: F(1,30) = 9.98, P < 0.005; our study: F[1,15] = 5.62, P = 0.03). Our results did not depend upon the subdivision of the SNpc; there was no significant loss from 44 to 110 years (i) with an arbitrary but reproducible subdivision of the SNpc into two bands of equal width, or (ii) with the subdivisions described by Fearnley and Lees (Fearnley and Lees, 1991). Accordingly, we do not confirm a neuronal loss in the SNpc-v of very old human subjects where it should have been preponderant if Parkinson's disease was linked to accelerated ageing.

Within the VTA, A8 and CGS, no statistically significant variations in the total numbers of neurons were observed from age 44 to 110 years. This is an important difference from observations in Parkinson's disease where, besides the severe loss of nigral neurons, half the dopaminergic neurons also degenerate in the VTA and A8 areas, and a small proportion (~5%) of dopaminergic neurons are also lost in the CGS (Hirsch et al., 1988).

Finally, we did not observe a significant variation with age in the number of neurons located in the rostral part of the LC. It has been suggested that these noradrenergic neurons are vulnerable to normal ageing (Vijayashankar and Brody, 1979), with a more severe neuronal loss in the rostral than in the caudal portion of the LC during ageing (Chan Palay and Asan, 1989a); according to Chan Palay and Asan, the rostral and the caudal parts of the LC are equally affected in Parkinson's disease (Chan Palay and Asan, 1989b), another difference in the distribution of neuronal loss between normal ageing and Parkinson's disease. We analysed three levels in the rostral LC and did not observe any statistical variation in the number of TH-positive neurons with ageing, which emphasizes the differences between the patterns of cell loss during normal ageing and in Parkinson's disease.

In conclusion, this study provides data on neuronal cell counts in the SNpc and in the other dopaminergic nuclei of the midbrain in a rather large group of control subjects which includes very old individuals. With a reliable count of the total number of dopaminergic neurons, a decreased number of neurons was observed in the SNpc as a function of age before the age of 80 years. This effect was not seen in the older subjects who may represent `super-normal' individuals. Furthermore, our study clearly shows that, even before the age of 80 years, there is no loss of dopaminergic neurons as a function of age in the VTA, A8 and LC, where neurodegeneration is known to occur during Parkinson's disease. The factors contributing to neuronal death that have been implicated in ageing and in Parkinson's disease probably belong to a common final pathway but are not the primary cause that initiates neuronal death.


We wish to thank M. Polivka, MD and J. Mikol, MD (Hôpital Lariboisière, Paris) for providing some autopsy material. This study was supported in part by the Institut National de la Santé et de la Recherche Médicale, the Fondation pour la Recherche Médicale (N.K.), the Association Claude Bernard pour le Développement des Recherches Biologiques et Médicales dans les Hôpitaux de L'Assistance Publique à Paris (B.A.F. and J.-J.H.), the Scientific co-operation programme between France and Austria of the French `Ministère des Affaires Etrangères' (Project No. 95.45.18) and the Parkinson Foundation Inc., Miami, Fla., USA (Y.A.).


View Abstract