Introduction
Acromegaly is characterized by excessive levels of growth hormone (GH) and insulin-like-growth factor 1 (IGF-1) that are caused by pituitary adenomas in most cases. The clinical condition is associated with a vast range of complications and comorbidities such as organomegaly, skeletal deformations with secondary arthropathy, diabetes mellitus, arterial hypertension, secondary pituitary deficiency, and neurological complications such as nerve compression syndromes [
1]. With regard to psychiatric syndromes, both personality changes and a psychopathological risk profile with an increased prevalence of affective disorders have been suspected [
2‐
7].
The increasing attention for neuropsychiatric aspects of acromegaly is concurring with preclinical research that has gained insight into the role of GH and IGF-1 in brain function. Both factors are involved in the development of the brain [
8‐
10]. More specifically, GH and IGF-1 play a part in the regulation of brain growth including neuronal cell proliferation, differentiation and myelination during development [
11‐
13]. Further, both postnatally and in the adult rodent, IGF-1 promotes neurogenesis and synaptogenesis in the hippocampal dentate [
13,
14]. In turn, the experimental disruption of IGF-1 gene function results in reduced brain size, CNS hypomyelination and loss of hippocampal granule and striatal parvalbumin-containing neurons [
15].
While many of the effects of IGF-1 and GH in the brain are mediated by locally produced hormones, also circulating hormones can reach the brain after crossing or bypassing the blood brain barrier [
16]. In acromegaly, from 6 to 10 years of hormone excess passes until clinical signs or symptoms of acromegaly lead to the diagnostic work-up and treatment [
17]. Such long-lasting exposure to IGF-1 and GH interferes with regulation of trophic processes in many organ systems [
1], and may be supposed to affect brain function and structure. Here, magnetic resonance imaging (MRI) provides a number of techniques that allow a non-invasive characterization of neuropathological changes, yet no quantitative analyses of brain changes in acromegaly have been presented so far.
We investigated if the GH and IGF-1 excess in acromegalic patients alters the macroscopic structure of the brain as determined by MRI. More specifically, we gained volumes of the main compartments of the brain—gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF)—from high resolution MRI. The GM compartment includes the whole cortex and the deep nuclei, whereas the WM compartment represents the entirety of the fiber tracts. In addition, we explored if acromegalic patients show MRI signs of neurovascular pathology that may add to their neuropsychiatric comorbidity.
Statistical analysis
Comparison of comorbidities and WML
The frequency of comorbid diagnoses and anthropometric measures were compared using Fisher’s exact tests or t-tests, respectively. The percentage of subjects in each WML class was compared between the patient and control group using two-sided Pearson chi square tests. The analysis was repeated separately for patients with shorter and longer exposure to hormone excess.
Analysis of volumes of gray matter, white matter, cerebrospinal fluid, and total intracranial volume
GMtotal, WMtotal, CSFtotal, and TIV were compared between patients and controls using analysis of covariance (ANCOVA) with age and gender as covariates. As a large proportion of the variance of the brain’s volume is determined the TIV, the analyses were repeated with GMnorm, WMnorm, and CSFnorm.
Relation of volumetric measures to estimated disease duration and current hormonal status
To detect a relationship between the time of hormonal excess and volumetric changes, patients were stratified into subgroups according to a median (10.9 years) split. ANCOVA was performed using a three-level group factor (controls, patients with below median [subgroup I] and above median [subgroup II] time of hormonal excess) and age and gender as covariates. Dependent variables were absolute brain volume (GMtotal + WMtotal), GMtotal, WMtotal, CSFtotal, the respective normalized volumes, and TIV. Subgroups did not differ with regard to gender distribution (women/men 13/9 and 10/12, P = 0.547, Fisher’s exact test) and mean age (56.5 ± 14.0 years and 52.2 ± 13.8 years, P = 0.305, t-test). Expectedly, the mean age at diagnosis was lower in subgroup II (51.4 ± 14.4 and 42.5 ± 13.1 years, P = 0.037, t-test). Covariation of absolute and normalized volumes of the whole brain, GM, WM and CSF, and TIV with the current hormonal status was investigated by (a) multiple regression of the volumetric parameters on age, gender and xULNIGF-1, and (b) by comparing patients with (N = 27) and without (N = 17) biochemical control at the time of MRI, covarying for age and gender. The statistical significance threshold was set to 0.05.
Relation of brain measures with vascular comorbidity status
After obtaining the main finding of an association between acromegaly and brain volume, cardiovascular comorbidities found to be different in frequency between patients and controls (arterial hypertension and diabetes, see result section) were summarized to one factor, and two-factoral ANCOVA was performed to investigate if this comorbidity status or its interaction with acromegaly underlies these effects. Age and gender were used as covariates.
Discussion
Due to the widespread functions of IGF-1 and GH in many organ systems, acromegaly is a complex systemic disease. Similarly to other organ systems, the central nervous system may be affected by long-term exposure to increased levels of GH and IGF-1, however, no in vivo studies on brain structure (or metabolism) in acromegaly have been reported to our knowledge. In brief, the data from this study demonstrates that: (1) the excess of GH and IGF-1 is associated with a volumetric expansion of the GM (+3.7%) and WM compartment (+5.1%) of the brain, accompanied by compensatory CSF volume decreases and total intracranial volume increases, (2) these changes seem to occur during the first 10 years of hormone excess, and (3) patients exhibit more focal hyperintense WM lesions that served as marker of neurovascular pathology.
The volumetric expansion of GM and WM may be the result of the long-lasting exposure of neuronal or glial cells towards systemically elevated GH and IGF-1. Both factors are known to mediate the maturation of these cell types during brain development and continue to have pleiotropic effects in adulthood [
10,
21]. IGF-1 overexpression—though not directly comparable to systemically elevated IGF-1/GH levels—has been reported to lead to markedly increased brain size in mice [
22], whereas IGF-1 knockout mice showed smaller brains [
15]. These reports suggest that IGF-1 mediated trophic effects on the brain may indeed reach a macroscopic scale. It is not known, however, to what degree the altered brain phenotypes translated into behavioral changes or more generally, functional deficits, since data on these characteristics were not systemically evaluated [
15,
21,
22]. Central application of IGF-1 in adult rodents resulted in decreased depression-like behavior, however, no volumetric information was provided in this specific study [
23].
With regard to WM, both increases of the axonal volume (physiologically about 46% of WM) and of myelin sheaths (i.e., oligodendrocytes, physiologically about 24% of WM) are conceivable [
24]. In view of reports on swelling of myelin sheaths in consequence to increased IGF-1 levels, and an increase of the myelin content in IGF-1-overexpressing mice, expansion of the myelin compartment may explain our observation [
22]. Advanced MRI including diffusion tensor imaging and
1H-MR-spectroscopy may be used to assign the volume changes more specifically to a histopathological substrate. With regard to GM, the observed changes may reflect expansion of axons and dendrites of fully differentiated neurons. Experimental histopathological work and theoretical considerations, however, suggest that the glia cell compartment may be an equally important source of macroscopical volume changes [
25]. Neurogenesis triggered by IGF-1 and GH, i.e., an increased number of neurons, seems more unlikely as neurogenesis is restricted to few brain regions such as the hippocampus.
In general, decreases of local or global GM or WM volume due to primary neurodegeneration or secondary to inflammatory conditions, neurovascular pathology or toxic agents are found more often than volume increases [
26]. We investigated the influence of arterial hypertension and diabetes, the two cardiovascular factors significantly more frequent in the acromegaly group in this sample, finding trend decreases of GM
total volumes across in association with the presence of these comorbidities across both groups, and no effects (
P > 0.10) for WM
total and BV
total. This strengthens that volume effects are unlikely to be driven by cardiovascular comorbidity but are rather specific to acromegaly. Disease specific increases of GM volume are rather exceptional and mostly represented in the spectrum of neurodevelopmental disorders such as hemimegalencephaly or cerebral gigantism. Patients with these disorders often present with general developmental delay or seizures. Also cortical dysplasia may be volumetrically expansive, however is highly focal in most cases [
27]. Increases of basal ganglia volumes have been observed as consequence of long-standing neuroleptic drug treatment [
28].
The split into patient subgroups with shorter and longer hormone excess allowed to generate a number of new hypotheses: first, no linear relationship between the estimated time of hormone excess and brain size could be observed. Rather, for both GM and WM, the volume increases seem to saturate after about 10 years. Even a slight decrease was seen after 10 years that, however, failed to reach significance. Second, as far as extractable from cross-sectional data, CSF spaces were decreased first, while TIV continued to increase, giving space to CSF that returned to normal absolute values. This pattern may indicate that the brain volume increase preceeds (or overrides) changes of the intracranial vault in the early phase. As some IGF-1/GH effects in acromegaly, for instance on bones, skull and joints, are reversible only at initial stages, but not if the disease is left untreated for several years [
1], it remains pivotal to determine if there is also a critical limit up to which cerebral changes are reversible. We also revealed no correlation between the current hormone status and brain volumes, confirming that brain volume changes are long-term effects and unlikely to follow short term fluctuations of IGF-1 (or GH).
Although widespread anatomical systems may be affected by the ubiquitary presence of IGF-1 and GH in the brain, some regions may be more susceptible than others [
29]. Clinically, however, a large variety of neuropsychiatric symptoms rather argues for disturbances of multiple circuits, including the regulation of affect and anxiety related behavior, pain perception and higher cognitive functions. It should be mentioned that we do not know yet to which degree the observed structural changes actually affect brain function. Future studies in acromegalic patients are underway to determine if psychopathological or cognitive abnormalities are indeed associated with these morphological changes.
Differences of WML counts were subtle with patients being overrepresented in only one of the higher WML lesion categories. The effect of acromegaly on this simple neurovascular parenchymal marker was into the expected direction, as acromegaly provokes cardiovascular and metabolic complications including arterial hypertension and diabetes. Both diagnoses were indeed more frequent in patients than controls and are known to be associated with increased number of WML [
30,
31]. Therefore, the WML finding of this study probably reflects the increased cardiovascular and metabolic comorbidity. In addition, the WML effect was stronger in the subgroup with longer hormone excess, which further concurs with the hypothesis that hypertension and cardiovascular diseases develop as a consequence of chronic alterations of the endothelial and cardiac function in acromegaly [
1]. We cannot exclude, however, that IGF-1/GH increase the susceptibility to ischemia by different mechanismus, e.g., by reducing the remyelination potential. In any case, structural damage to the fiber tracts by microangiopathy can add to cognitive decline [
32] and makes careful monitoring and treatment of cardiovascular and metabolic comorbidities mandatory.
The study is mainly limited by its cross-sectional design and heterogeneous patient sample with different disease duration. In addition, the method of stratification into shorter and longer duration of hormone excess is debatable. We performed a supportive analysis using linear regression to explore if the estimated time of hormone excess allows correlates with total brain size, yet found no improvement in the model fit compared with the group stratification (data not shown). Further, as the variable is a compound of three other variables (time of diagnosis, estimated 8 years of previous hormone excess and time since biochemical control), its reliability is generally limited. Methodologically, TIV is a standard intraindividual reference for brain volume, however, in the case of acromegaly it may not be stable. Despite this, we also analyzed volumetric measures normalized to TIV, as TIV explains a large proportion of the variance of the brain volume. Subtle changes of the group statistics confirmed that TIV covaried with the disease process which underlines the importance to match for other factors co-correlating with TIV, particularly gender and body height. Furthermore, the calculation of total CSF volume including the external subarachnoid spaces is challenging for automated methods that do not explicitly model extradural tissue. This may in part explain the larger group overlap for analysis of CSF and TIV.
In conclusion, we provide first in vivo evidence of a disturbed macroscopic brain architecture in acromegaly with increases of total GD and WM volumes. These changes seem to occur during the first 10 years of hormone excess at the expense of the CSF compartment, with parallel increases of the total intracranial volume. In addition, a tendency for increased neurovascular pathology is demonstrated, most likely due more frequent metabolic and cardiovascular comorbidities. Future work will need to clarify which brain functions and neuropsychiatric syndromes relate to these changes, and if treatment can reverse them.