Survivors of traumatic brain injury (TBI) often suffer from several medical complications, which may have implications for recovery. Most of these complications, such as neurological and neuropsychiatric disorders, are familiar to physicians and receive the necessary attention during the early period after trauma and later on during the course of rehabilitation. Among the less well recognised and potentially underdiagnosed complications is post-traumatic hypopituitarism (PTHP), despite post-mortem evidence dating back several decades showing pituitary gland infarction to be present in up to one-third of patients who died shortly after TBI.1^–3 This may be explained by the fact that many of the symptoms of hypopituitarism, such as fatigue, neuropsychiatric and cognitive deficits, are often attributed to the post-concussional syndrome.4 Over the past few years, attention has focused on evaluating the hypothalamic-pituitary function in survivors of TBI, and several systematic studies have shown that PTHP is far more common than previously thought and can be associated with adverse outcomes. These findings strongly suggest that hypopituitarism remains largely undiagnosed and untreated in most cases after TBI, which can have significant implications for patient recovery and rehabilitation.

In this review, we will examine the prevalence and the natural history of pituitary dysfunction after TBI and the potential implications of this complication for the care of affected patients. For this purpose, we searched PubMed for articles using the following terms: “traumatic brain injury with hypopituitarism”, “growth hormone deficiency”, “cortisol deficiency”, “hypogonadism”, “hypothyroidism”, “diabetes insipidus”, “syndrome of inappropriate anti-diuretic hormone secretion” and “neuroendocrine dysfunction”. There were no restrictions regarding date or language of publication. All original systematic articles with prospective and/or retrospective design of acute and/or chronic TBI were included, as were review articles and case series. However, single case reports were excluded.

Epidemiology of TBI

TBI is a non-degenerative, non-congenital insult to the brain from an external force, resulting in transient or permanent neurological dysfunction.5 6 It is a disorder of major public health significance, as it is the leading cause of death and disability in young adults.7^–9 Approximately 180–250 persons per 100 000 per year die or are hospitalised in industrialised countries as a result of TBI.10 In the USA in 2003, there were an estimated 1.5 million TBI cases, with 1.2 million emergency department attendances, 290 000 hospitalisations and 51 000 deaths.11 It is estimated that 2% of the US population today are living with the consequences of TBI, requiring assistance with activities of daily living,12 and that this level of disability results in a lifetime cost of US$ 600 000 to US$ 1.9 million per person.13 People most at risk of TBI are those aged between 15–24 years, children less than 5 years of age; another peak in incidence occurs in those over 75 years old.8 Males are twice as likely to experience TBI than females; however, the number of females is rising steadily.13 It is worth noting that the prevalence of TBI is probably underestimated, as many episodes of mild injury are not reported and therefore are not included in the statistics.

The majority of TBI cases are caused by motor vehicle accidents (50%), whereas falls and violence-related injury account for approximately 30% and 20%, respectively. The predominant cause of TBI changes at both ends of the life spectrum. Falls are more common in the elderly, whereas child abuse (such as shaken baby syndrome) accounts for a larger proportion of TBI in young children.8

PITUITARY DYSFUNCTION IN TBI

Historical perspective

Post-traumatic hypopituitarism (PTHP) has been recognised since 1918 when Cyran published a report of pituitary damage following a fracture in the base of the skull.14 However, PTHP was considered a rare consequence of TBI, accounting for 0.7% of hypopituitarism in a series reported by Escumilla & Lisser15 Although several published autopsy series since the 1950s have shown pituitary and/or hypothalamic damage to occur frequently following fatal head injuries,1^–3 16 17 no systematic study of pituitary function was carried out until recently. Two large case series of pituitary function were published between 1986 and 2000.18 19 The larger of the two case series was reported in 2000 by Benvenga et al, which consisted of 367 known cases of PTHP.19 In that series, the prevalence of gonadotropin, adrenocorticotropic hormone (ACTH), thyroid-stimulating hormone (TSH) and growth hormome (GH) deficiencies and hyperprolactinaemia were 100%, 52.8%, 44.3%, 23.7% and 47.7%, respectively. As this was a case series, selection bias may have contributed to the high reported frequency of gonadotrophin deficiency compared with GH deficiency in those patients. A history of coma of varying durations was noted in 93% of patients. Interestingly, the diagnosis of hypopituitarism was made more than 5 years following head injury in 15% of patients.

NEUROENDOCRINE DYSFUNCTION IN THE ACUTE PHASE OF TBI

A number of studies have assessed the acute neuroendocrine changes following TBI. Earlier studies were not specifically designed to examine the frequency of acute hypopituitarism after TBI, but rather attempted to correlate neuroendocrine changes with the severity of head trauma and/or to define the metabolic derangements and other variables that may predict outcome.20^–23 More recently, attention has been given to diagnosing the frequency of hormone deficiencies in cohorts of TBI patients.24^–26 The design, methodologies and findings of these studies are summarised in table 1.

As demonstrated in table 1, there is little agreement among the studies about the nature of neuroendocrine changes post-TBI and these variations may reflect differences in patient selection, severity of injury, study design, and methodology and timing of the assessment. There is also inconsistency regarding the biochemical parameters that can predict outcome. Cernak et al noted that serum testosterone correlated with severity of injury;23 however, this was not supported by the findings of the study by Lee et al.27 Hackl reported that impaired GH response to arginine stimulation and a flat TSH response to thyrotropin-releasing hormone (TRH) were associated with poor outcome,21 whereas the study by Della Corte et al showed a paradoxical rise in growth hormone response to TRH, which predicted poor outcome.22 The authors and their colleagues noted that serum prolactin correlated negatively with Glasgow Coma Scale (GCS) scores.24 Tanriverdi et al found a positive correlation between severity of injury and cortisol levels in mild and moderate TBI, but this was not demonstrable in patients with severe TBI.26 In the study by Cohan et al, subnormal cortisol levels were associated with higher injury severity scores and early ischaemic insults.25

The occurrence of adrenal insufficiency (ACTH-cortisol deficiency) in the acute phase of TBI deserves special attention, as this complication is particularly serious and can be life-threatening. We found abnormal cortisol response to glucagon stimulation in 16% of patients with TBI in the acute phase.24 However, Cohan et al reported a higher frequency of 53%, which was based on a basal serum cortisol concentration below the 25^th percentile of the expected response in stressed, non-TBI trauma patients.25

Some of the acute-phase changes—in particular, hypogonadism and hyperprolactinaemia—are not specific to TBI and may reflect adaptive responses to injury and critical illness with uncertain clinical significance. However, adrenal insufficiency appears to be associated with serious consequences in some patients. This was highlighted by a case series published by the authors and their colleagues describing acute adrenal crises presenting in patients with severe head trauma, who subsequently showed dramatic responses to glucocorticoid replacement therapy,28 whereas Cohan et al also showed that hypotension (requiring vasopressor support) was more common in patients with adrenal insufficiency.25

NEUROENDOCRINE DYSFUNCTION IN LONG-TERM SURVIVORS OF TBI

Anterior pituitary dysfunction

Anterior pituitary dysfunction following TBI has been widely studied in recent years. The findings of 12 of these systematic studies are summarised in table 2. In the first of these, Kelly et al studied 22 patients with moderate to severe TBI assessed at a median of 26 months post-injury.29 The clinical significance of the hormone deficiencies that they found were not reported, and interpretation was complicated by the fact that all patients who were found to have pituitary dysfunction had experienced hypotensive or hypoxic insults during the episode of TBI. Lieberman et al followed this with a study of 70 patients with TBI recruited from a residential rehabilitation centre.30 The patients underwent initial hormonal screening, and only those with abnormalities went on to have dynamic pituitary stimulation tests. GH reserves were assessed in 48 patients using the glucagon stimulation test.31 Seven patients (14.6%) had a subnormal GH response, 5 of whom went on to have the L-dopa test, which confirmed GH deficiency in all 5 patients. A total of 45% of the cohort had morning basal cortisol levels <7 μg/dl (193 nmol/l) but only 5 patients failed the short Synacthen (ACTH) stimulation test. Overall, 51.4% of patients had a single pituitary hormone abnormality and 17.1% of this cohort had dual abnormalities.

The authors and their colleagues examined 102 consecutive unselected survivors of moderate or severe TBI at a median of 17 months after injury.32 To ensure rigorous assessment of the somatotrophic and gonadotrophic axes, each patient underwent two separate stimulation tests; patients were defined as GH or ACTH deficient if they failed both provocative tests. Twenty nine percent of patients had at least one pituitary hormone abnormality; 23% had isolated abnormalities. A total of 11% of patients had GH deficiency, including 8% with severe deficiency—peak GH <3 μg/l, which is an indication for replacement. In addition, 13% of patients had ACTH deficiency, 12% had gonadotropin deficiency, 1% TSH deficiency and 13% of patients had hyperprolactinaemia. Only one patient was found to have panhypopituitarism. Several other studies have confirmed the findings of the above initial reports with varying frequencies of hormone deficiencies (table 2).26 33^–40 Leal-Cerro et al only performed systematic neuroendocrine testing on patients who had neuropsychiatric abnormalities based on their own specifically designed questionnaire36 and on the Quality of Life-Assessment of Growth Hormone Deficiency in Adults (QoL-AGHDA) questionnaire, comprising 25 yes/no questions relating to issues of memory, concentration, social isolation, energy and mood.41 Of the patients who underwent initial basal testing, 44 patients had low insulin growth factor 1 (IGF-1) and went on to have assessment with one or more of three dynamic tests, 10 of whom had GH deficiency. However, it is possible that the true prevalence of GH deficiency in this cohort was underestimated, as approximately 50% of adults with GH deficiency have serum IGF-1 in the normal reference range.42

Most of the reported studies failed to show an association between the severity of the head injury, as measured by the GCS, and the risk of anterior hypopituitarism; however, Bondanelli et al found slightly lower GCS scores in their patients with PTHP compared with unaffected survivors.34 In addition, a recent study by Klose et al noted that patients with anterior pituitary hormone deficiencies had more often suffered severe TBI, increased intracranial pressure (ICP) and longer intubation than the patients who were without pituitary insufficiency.40 One study found that hypogonadism at 3 months is closely related to the severity of TBI,38 although this was postulated to be an effect of physiological downregulation of the gonadal axis, rather than being related specifically to TBI. The authors found no association between anterior pituitary dysfunction and acute-phase CT scan appearance or the Glasgow Outcome Scale, nor were we able to show a difference in frequency of PTHP between patients with different mechanisms of injury (eg, motor vehicle accidents versus falls).27

Although the different studies reported varying results on the overall prevalence of long-term hypopituitarism and on the relative frequency of individual hormone deficiencies, there is a broad agreement that PTHP is a common finding after head injury with an estimate of about 25% among long-term survivors.

NEUROHYPOPHYSEAL DYSFUNCTION

Diabetes insipidus is well recognised following TBI, with a reported frequency of 3–26% in the acute phase.43^–46 Acute-phase diabetes insipidus has an association with more severe head injury and the presence of cerebral oedema on brain CT.45 In a series of 102 patients with a history of severe or moderate TBI who were evaluated at a median of 17 months after injury, diabetes insipidus was diagnosed in 7% of patients using the gold-standard water deprivation test. Interestingly, neither transient nor permanent diabetes insipidus was shown to be related to the presence of anterior hypopituitarism.47

Hyponatraemia, mainly caused by the syndrome of inappropriate antidiuretic hormone secretion (SIADH), has been reported to occur in 2.3–36% individuals post-TBI.48^–52 The conflicting figures in the literature reflect differences in patient selection, diagnostic criteria for SIADH and length of monitoring of plasma sodium after TBI. In most cases, SIADH manifests within 2 days of the injury, but can be delayed for up to 18 days and is nearly always transient. SIADH is unrelated to the severity of head trauma.45 46 Cerebral salt wasting syndrome has also been suggested as a possible cause of hyponatraemia following TBI,53 54 with an incidence of about 1%;45 however, some authors have disputed the existence of this diagnostic entity entirely.55

TBI IN CHILDHOOD AND ADOLESCENCE

Almost 180 children per 100 000 population experience closed head injury per year. The estimated incidence of TBI doubles between the ages of 5 and 14 years and peaks in early adulthood, with incidence being similar to that in the adult population.56 TBI in children is secondary to unintentional injury in 75% of cases;8 however, child abuse is an important cause, with an annual incidence of 24.6 per 100 000 children younger than 1 year.57 In the first systematic studies in a paediatric population, Einaudi et al evaluated 22 TBI children retrospectively at 0.7–7.25 years post-TBI (mean age 12.8 years), and 30 children prospectively (mean age 9.06 years).58 A proportion of the prospective cohort was re-evaluated at 6 (86%) and 12 months (66%) after injury. Basal tests were performed in all patients and dynamic testing in selected patients. Hormonal abnormalities were present in 16% and 8% in the retrospective and prospective groups, respectively, with GH deficiency being the most common deficit. The prevalence of PTHP in this study was lower than that in the adult populations, possibly due to the small numbers studied and the reliance on basal hormone measurements. By contrast, a second retrospective study by Niederland et al, consisting of 26 children (mean age 11.47 years), studied at a mean of 30 months following TBI, reported pituitary dysfunction to be present in 61% of the subjects on the basis of both basal and stimulated hormone measurements, with GH deficiency being the most common biochemical abnormality detected in 42% of the children.59 A non-significant decrease in height was noted in the GH-deficient patients with TBI, when compared with the GH-sufficient patients with TBI. GH deficiency was diagnosed using the L-dopa test, with a peak GH <7 ng/ml, a cut-off that may be too high for L-dopa, which is a weak GH secretagogue. This may have led to an overestimation of the frequency of GH deficiency and hence hypopituitarism in this group.59

Notwithstanding the potential methodological problems with these two studies, they help to raise awareness of post-traumatic hypopituitarism in children and the need for properly designed prospective longitudinal studies to assess not only the biochemical responses, but also auxological changes, which will improve the reliability of the data reported.60^–62

THE NATURAL HISTORY OF PTHP

Four prospective longitudinal studies have evaluated pituitary function in a cohort of TBI patients with predominately severe or moderate injury.26 37 38 63 The authors and colleagues have prospectively assessed anterior pituitary function in 50 patients following moderate or severe TBI.63 The results demonstrated that anterior pituitary hormone deficiencies that occur soon after TBI recover in some patients and recovery was usually evident by 6 months. Gonadotropin deficiency and hyperprolactinaemia were most likely to recover completely in the majority of patients. A return to normal GH production was seen in 66% and cortisol production in 50% of patients. Persistent deficiency of these two axes was associated with more severe acute-phase GH and cortisol hyposecretion. However, some patients with normal responses in the acute phase developed late pituitary dysfunction, particularly ACTH deficiency, which was apparent at the 6-month assessment. No new abnormalities were evident later than 6 months. Aimaretti et al examined 70 patients with TBI at 3 and 12 months after injury.37 Panhypopituitarism that was diagnosed at 3 months remained unchanged at 12 months, but many of the isolated or multiple hormone deficiencies recovered or improved between 3 and 12 months. A total of 13% of the isolated deficiencies at 3 months became multiple by 12 months and 5.5% of patients without abnormalities at 3 months had isolated deficiencies by 12 months, demonstrating the dynamic nature of pituitary hormone changes over time. Two subsequent studies reported similar findings.26 38 Therefore, it seems that early post-traumatic pituitary dysfunction recovers in a significant proportion of patients and, conversely, hypopituitarism can evolve over time and become detectable in the post-acute phase in others. These findings have implications for the timing and the frequency of pituitary assessment after TBI.

PATHOPHYSIOLOGY OF TBI-INDUCED HYPOPITUITARISM

The pituitary gland is located within the bony sella turcica, where it is lined superiorly by a dense layer of connective tissues—the diaghragma sellae. A total of 70–90% of the blood supply to the anterior lobe of the pituitary gland is derived from the long hypophyseal vessels—these give rise to the hypophyseal portal circulation that carries hypothalamic neuropeptides from the hypothalamic neurones to the adenohypophysis. Blood supply to a small part of the adenohypophysis adjacent to the posterior lobe and the entire neurohypophysis is derived directly from the inferior hypophyseal artery, which arises from the internal carotid artery just below the diaghragma sellae.64

A number of potential pathological mechanisms have been proposed that may lead ultimately to anterior pituitary infarction. These include compression of the pituitary gland, hypothalamic nuclei or interruption of the long hypophyseal vessels by oedema, haemorrhage, skull fracture, raised intracranial pressure or any cause of hypoxic insult.18 29 65 Hypopituitarism can also result from direct mechanical injury to the hypothalamus, the pituitary stalk and/or the pituitary gland itself. The inferior hypophyseal blood vessels are not transected with stalk rupture and therefore infarction of the posterior lobe is avoided. However, it may become denervated as a result of injury to the paraventricular and supraoptic hypothalamic neuclei, the pituitary stalk or the axon terminals in the posterior pituitary leading to diabetes insipidus.2 18 Inflammatory oedema around the hypothalamus or the posterior pituitary may also cause diabetes insipidus, which recovers as the oedema resolves.

Different autopsy series have reported injury to the hypothalamus, pituitary gland or pituitary stalk in 26–86% of patients who died shortly after TBI.2 3 16 17 The site of the injury varies according to the study but necrosis of the anterior lobe was more frequent in patients who died within 1 week of TBI.19 Daniel et al described autopsy findings in five cases of extensive infarction of the anterior lobe of the pituitary gland due to rupture of the pituitary stalk after head trauma.2 The sixth case had traumatic stalk interruption at the point of attachment of the stalk to the hypothalamus; the stalk itself remained intact, as did the vessels that supply the stalk with no evidence of anterior lobe infarction. This case demonstrated that lesions high on the stalk that do not interfere with its blood supply do not cause infarction of the anterior lobe.2

These autopsy studies highlighted that the anterior pituitary and the stalk are common sites of injury. Although these were studies of fatal head trauma cases and the correlation between the autopsy and the biochemical finding is unknown, the results add support to the subsequent biochemical data that demonstrated a high prevalence of hypopituitarism after TBI.

CLINICAL CONSEQUENCES OF POST-TRAUMATIC HYPOPITUITARISM

The studies discussed in this review clearly demonstrate that pituitary dysfunction following TBI occurs with a much higher frequency than previously thought and raise important questions about the potential contribution of this complication to the morbidity and possibly the mortality associated with head injuries. Some of the abnormalities identified in the various studies are partial deficits and may be of uncertain clinical significance in an otherwise healthy individual. However, they may have added importance in patients with TBI who have a high burden of physical and neuropsychiatric disabilities, resulting in increased morbidity and impaired recovery. Glucocorticoid deficiency can be life-threatening, particularly in acutely ill patients. Two recent reports have highlighted the serious morbidity associated with post-traumatic adrenal insufficiency in acutely ill patients with TBI,25 28 with one series showing a dramatic response to glucocorticoid replacement.28

GH deficiency impairs linear growth and the attainment of normal body composition in children. In adults, it causes reduced lean body mass,66 67 decreased exercise capacity,68 impaired cardiac function69 70 and reduced bone mineral density,71^–73 which may be of particular significance in immobilised patients. In addition, the adult GH deficiency syndrome may result in neuropsychiatric manifestations and diminished quality of life.74 Kelly et al reported increased neuropsychiatric morbidity in patients with post-TBI GH deficiency and/or insufficiency.29

Testosterone deficiency in males is associated with reduced lean body mass, muscle weakness and impaired exercise tolerance.75 In both males and females, sex-steroid deficiency results in reduced bone mineral density and osteoporosis.76 77 This situation can be exacerbated by long periods of immobility that occur after serious TBI and by the co-existence of other anterior pituitary hormone deficiencies. Hypothyroidism leads to lethargy, fatigue and neuropsychiatric manifestations. Untreated diabetes insipidus causes dehydration if water intake is not adequate to compensate due to impaired cognition, physical disability or co-existent hypodipsia. It is, therefore, reasonable to infer that unrecognised and untreated hypopituitarism can have serious adverse consequences for patients with TBI and may impair recovery, rehabilitation and adds significantly to the high morbidity seen in this condition. In a recent large study of 104 patients with TBI, post-traumatic hypopituitarism was independently associated with poor quality of life (particularly in scores of energy, sleep and physical mobility), abnormal body composition and adverse metabolic profile,78 therefore confirming the clinical and public health importance of this condition.

SCREENING FOR POST-TRAUMATIC HYPOPITUITARISM

Currently, the available evidence supports screening patients with severe or moderate TBI, whereas screening those with mild injury may be done if clinically indicated based on symptoms and signs.79 Certain categories of patients with TBI may be at a greater risk of PTHP, including those with diffuse axonal injury, basal skull fracture and the older age group,80 and they should be regarded as a high priority for pituitary assessment.

In the acute phase, adrenal insufficiency should not be missed because it is potentially life-threatening.25 28 As dynamic tests are not practical in acutely ill patients, we recommend daily measurement of serum cortisol concentration in the first 7 days after TBI and we regard an AM cortisol level of less than 200 nmol/l (in the absence of dexamethasone or prednisolone administration) to be highly suggestive of glucocorticoid deficiency, even allowing for changes in plasma protein concentration in this setting—this can influence the assay results. Morning cortisol values between 200 and 500 nmol/l have to be interpreted in the clinical context and glucocorticoid replacement should be started if there are any features suggestive of hypoadrenalism, such as hypotension, hyponatraemia or hypoglycaemia.28

Assessment of the GH, gonadal and thyroid axes is not necessary in the acute phase as there is currently no evidence to show that replacing these hormones in the acute phase is of benefit. Patients with severe and moderate TBI should undergo assessment of the adrenal (using the short Synacthen test), thyroid (basal free T4 and TSH) and gonadal (sex-steroid concentrations, LH and FSH, menstrual history) functions at between 3 and 6 months. If abnormalities are detected, hormone replacement should be instituted and repeat assessment at 1 year should be considered if the clinical or biochemical parameters suggest delayed recovery. Assessment of GH reserves requires dynamic tests, which can be time-consuming and need to be performed in specialist pituitary units, but should be considered at 1 year (early GH deficiency can be transient in many cases)—at least in selected patients who continue to have significant cognitive and/or neuropsychiatric dysfunction or who have evidence of additional pituitary hormone deficiencies. Figure 1 contains a suggested algorithm for screening patients following TBI.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Suggested algorithm for assessment of patients following traumatic brain injury. GH, growth hormone; TBI, traumatic brain injury.

CONCLUSIONS

TBI is a major public health problem that exerts a high cost for both the individual and society at large. The studies discussed in this review provide convincing evidence that post-traumatic hypopituitarism is common following TBI and can contribute to the morbidity associated with head trauma, but remains undiagnosed and treated in the majority of patients. Considering the high incidence of TBI, post-traumatic hypopituitarism becomes a disorder of major public health importance. As hormone replacement therapy has the potential to reduce morbidity and improve outcome, screening programmes for post-traumatic hypopituitarism need to become part of standard clinical care for patients with head injury. Such programmes need multidisciplinary collaboration between endocrinologists, neurosurgeons, psychologists, rehabilitation physicians and other interested disciplines in order to ensure optimal delivery of care to maximise the potential for recovery and return to normality for the victims of head injury.