Predictors of Hypopituitarism in Patients with Traumatic Brain Injury

Introduction

Traumatic Brain Injury (TBI) is a global health problem and the leading cause of death and disability among young adults in industrialized countries, affecting 823.7 per 100,000 persons per year in 2010. It accounts for an estimated 2.2 million emergency department (ED) visits, 280,000 hospitalizations, and more than 50,000 deaths in the United States annually. ¹ The most common causes of these ED visits were due to falls, motor vehicle accidents (MVAs), or being struck by/against an object. ¹ During 1997–2007, an annual average of 53,014 TBI-related deaths (18.4 per 100,000 population; range: 50,680–54,906) occurred among U.S. residents. ² Among military service members, the incidence rate of TBI in 2010 was estimated to be 1,763.6 per 100,000. From 2010–2011, 4.2% of all service members were diagnosed with TBI, with more than 75% of those classified as mild. ³

In the last decade, several recent cross-sectional studies have found that patients who suffered TBI are at considerable risk of hypopituitarism. In a meta-analysis of 13 studies, including 911 patients with TBI or subarachnoid hemorrhage (SAH), Schneider and colleagues reported that the prevalence of pituitary dysfunction among patients with TBI was 27.5%. ⁴ However, studies have shown a wide range in the prevalence of hypopituitarism between 5 and 70% of patients with TBI. ^{5 –9} Little is known about the causes of these large variations in prevalence, which may be explained as a consequence of different types of injury, diverse diagnostic criteria used to establish hypopituitarism, different time-points for evaluation after TBI, patient selection, and different study design (including cross-sectional versus prospective). Some studies have suggested that the prevalence of hypopituitarism is not clearly related to the severity of trauma (measured with the Glasgow Coma Scale [GCS]), patient age, or gender. ^4,10 Some patients with mild TBI appear to be at substantial risk for hypopituitarism. ¹¹ A recent study revealed that dynamic endocrine changes may evolve for 5 years after TBI. ¹² Identification of risk factors for development of hypopituitarism in patients who have experienced TBI is needed to guide clinicians in deciding whom to test for hypopituitarism.

To characterize possible predictors of hypopituitarism or individual pituitary hormone deficiencies, we reviewed the records of all patients with TBI who underwent neuroendocrine evaluation at our tertiary care pituitary center from 2007 until 2013.

Methods

This is a retrospective study involving a review of case records, laboratory tests, and magnetic resonance imaging (MRI) or computed tomography (CT) data of TBI patients who underwent neuroendocrine evaluation. Between 2007 and 2013, a total of 166 adult patients were evaluated for hypopituitarism, mostly during the rehabilitation phase after TBI. These patients were largely referred for neuroendocrine testing from TBI-expert physiatrists as a part of the patients' multi-disciplinary evaluation and care. Most patients referred for neuroendocrine evaluation had persistent symptoms for longer than 60 days before testing.

The following data were extracted from case records: patient demographics, work status, cause and severity of brain trauma, TBI-related hospital admission chart notes, GCS score, brain imaging, and post-injury symptomatology, including headaches, seizures, depression (when diagnosed by a psychiatrist and noted in the psychiatry note), and post-traumatic stress disorder (PTSD). Data were also collected regarding medication use, pre-existing endocrinopathy, Diagnostic and Statistical Manual of Mental Disorders, 4th edition (DSM)-IV Axis V Global Assessment of Functioning (GAF) score, and the results of neuroendocrine laboratory testing. We reviewed neuroimaging (86 MRI and 23 CT scans) records to identify possible structural predictors of hypopituitarism following TBI. Patients who had evidence of hypopituitarism on endocrine testing underwent imaging using a dedicated pituitary protocol, whereas the remainder underwent brain imaging based on clinical indications. The MRI or CT reports of all patients with TBI were extracted, reviewed, and coded. For CT report coding, a “Form I CT data” sheet, developed and utilized by the Traumatic Brain Injury Model Systems (TBIMS), was used. ¹³ TBIMS is a project funded by the National Institute on Disability and Rehabilitation and is an ongoing, longitudinal, prospective, multi-center study. The MRI grading system was adapted from the TBIMS CT coding sheet, because there is no equivalent coding sheet for MRI data after TBI.

Neuroendocrine testing included the following tests:To assess the pituitary-adrenal axis, a cosyntropin-stimulation test was performed (at least 6 weeks after TBI). During this test, 250 mcg of cosyntropin was administered as an intravenous bolus, and cortisol was measured at 0, 30, and 60 min. A peak cortisol level <18 mcg/dL was considered abnormal. The pituitary-thyroid axis was assessed with measurement of free thyroxine (FT4) and thyroid-stimulating hormone (TSH). Secondary hypothyroidism was diagnosed when free T4 was below the lower limit of normal without elevated serum thyrotropin levels. The pituitary-gonadal axis was assessed with a morning testosterone (men) or random estradiol (women), luteinizing hormone (LH), follicle-stimulating hormone (FSH), and prolactin (PRL). Secondary hypogonadism was defined as morning serum testosterone level below the lower limit of normal for the assay (in men) and oligomenorrhea or amenorrhea (in women), without elevated serum gonadotropins (for patients of both genders).

The growth hormone/insulin-like factor-1 (GH/IGF-1) axis was evaluated with a basal IGF-1 level and a glucagon stimulation test (GST) in 121 patients. During this test, glucagon (1 mg) was administered intramuscularly, and glucose and GH levels were measured at 0, 30, 45, 60, 90, and 120 min. The GST was chosen because it has been shown to compare favorably with the insulin tolerance test (ITT), which is considered the criterion standard for provocative testing for GH deficiency. ¹⁴ We analyzed data using two different diagnostic cutpoints for GH deficiency (either a peak GH <3 ng/mL for all patients, or body mass index [BMI]-dependent cutpoints, including <1 ng/mL for patients with BMI >25 kg/m² and <3 ng/mL for patients with BMI <25 kg/m²). ¹⁵ Evaluation of the gonadal and GH axes took place at a minimum of 5–6 months after TBI. Diabetes insipidus (DI) was documented based on review of the patients' records and was defined as the presence of inappropriate hyposthenuria responding to desmopressin administration.

The GAF scores were retrospectively collected from psychiatrists' notes. The GAF score is a reflection of a patient's ability to function in daily life. It is a numeric scale (ranging from 0 to 100) rating the social, occupational, and psychological functioning of adults, with higher scores being indicative of better function. ¹⁶ The severity of TBI was categorized based upon the classification scheme from the Department of Defense and Department of Veterans Affairs TBI classification system (shown in Table 1).

Results

We identified 166 patients [117 (70%) males], who underwent neuroendocrine evaluation at our center. Clinical characteristics of the study population, stratified by the presence or absence of hypopituitarism, are shown in Table 2. Subjects' age ranged between 18 and 76 years (median age: 41.6 years). The interval time between brain injury and evaluation ranged between <1 month and 430.4 months (median: 40.4 months). The majority of these patients were examined in the chronic phase after TBI, including 75% of patients who were examined at ≥12 months and 85% of patients who were examined at ≥5 months after TBI. The most common causes of TBI were MVA (36%), fall (18%), military blast exposure (16%), and sports-related injury (15%). Sports injuries occurred while patients were participating in the following sports: football (12 patients), rugby (3 patients), skiing (4 patients), ice hockey (4 patients), soccer (1 patient), and swimming (1 patient). Of the 154 patients who had BMI data available, 31% were obese (BMI >30 kg/m²). In addition, 20% were military personnel, and 48% were hospitalized after the TBI.

Presenting symptoms (%, number of patients)were fatigue (91%, 138/152), headache (52%, 87/166), nausea (33%, 51/153), dizziness (53%, 80/152), seizure (7%, 11/165), decreased memory/attention (96%, 133/138), back/extremity pain (62%, 93/149), low libido (38%, 36/95), erectile dysfunction (26%, 26/101), oligomenorrhea not attributable to menopause or oral contraceptive use (28%, 7/25), depressive symptoms (43%, 72/166), and anxiety (28%, 46/166). Thirty-four patients met diagnostic criteria for major depression and 33 patients for PTSD.

One hundred fourteen (69%) patients had suffered mild injury and 51 patients (31%) had experienced moderate or severe TBI. Most of the injuries caused by blast and sports injury were mild, whereas injuries caused by MVA were more likely to be moderate or severe. Data on the prevalence of individual pituitary hormone deficiencies, stratified by severity of TBI, are shown in Table 3. Patients who had moderate or severe injury were more likely to have DI (p < 0.001), headaches (p = 0.004), and PTSD (p = 0.03). On neuroimaging, the presence of intracranial hemorrhage, petechial hemorrhages, and SAH were all associated with greater severity of TBI (p values <0.001, 0.002, and 0.001, respectively). There was no difference in prevalence of hypopituitarism between hospitalized versus non-hospitalized patients with mild TBI (data not shown).

Fifty-one (31%) patients were diagnosed with one or more pituitary hormone deficiencies, and had higher BMI (p = 0.0002), but were comparable in age and gender distribution to those without deficiencies. There was no association between the cause of injury and the presence of hypopituitarism. Thirty-three (29%) of patients who suffered mild TBI had evidence of pituitary dysfunction, and 18 (35%) of patients who suffered moderate or severe TBI had at least one pituitary hormone deficiency (difference between groups, NS). There was a trend for patients who had any pituitary hormone deficiency to report headaches less frequently (43% vs. 60%, p = 0.07) and a lower GAF score (16 vs. 27, p = 0.06). No symptoms were more commonly reported among patients with one or more pituitary hormone deficiencies (data not shown). History of depression or PTSD was not associated with any pituitary hormone deficiency; patients with any pituitary hormone deficiency were not more likely to require a neurosurgical procedure than those without evidence of hypopituitarism (data not shown). GAF score was not significantly associated with age, number of TBIs, time since most recent TBI, or diagnosis of PTSD (data not shown).

A total of 57 patients who had no evidence of neuroendocrine deficiency had an MRI or CT examination versus 34 patients with any pituitary hormone deficiency. There was no imaging abnormality that predicted the presence of hypopituitarism (data not shown).

Clinical and endocrine data, stratified by the presence or absence of any evidence of hypopituitarism or individual pituitary hormone deficiencies, are shown in Table 4a and 4b. Of the 144 (87%) patients who underwent a cosyntropin-stimulation test, adrenocorticotropic hormone (ACTH) deficiency was present in 14 (9.7%) patients. Patients with ACTH deficiency were more likely than those without hypoadrenalism to have suffered MVA (64% vs. 34%, p = 0.04), experienced post-traumatic seizures (21% vs. 5%, p = 0.04), demonstrated any intracranial hemorrhage (67% vs. 31%, p = 0.05), petechial brain hemorrhages (33% vs. 10%, p = 0.017), or focal cortical (frontal) parenchymal contusions (56% vs. 18%, p = 0.02) on MRI. There was no significant difference between patients who had and did not have adrenal insufficiency with regards to severity of trauma. Of 14 patients with ACTH deficiency, 7 had TSH deficiency, 6 had LH/FSH deficiency, 4 had DI, and 3 had GH deficiency.

Among 156 patients who had serum free T4 levels tested, 12 patients (7.8%) had central hypothyroidism. These patients had a higher mean BMI than those without thyrotropin deficiency (p = 0.02) and were more likely than those without hypothyroidism to have suffered an MVA (75% vs. 34%, p = 0.0094). Hypothyroid patients showed a trend toward more frequent focal cortical right frontal hemorrhage than those without hypothyroidism (50% vs. 19%, p = 0.06). Otherwise, there was no imaging abnormality that was predictive of TSH deficiency. Of all patients with TSH deficiency, 9 (75%) also had LH/FSH deficiency.

A total of 19 patients were diagnosed with central hypogonadism (after excluding 23 patients taking opioids). There were 19 of 139 patients (four women and 15 men) who had gonadotropin deficiency, which was associated with higher BMI (31.9 vs. 27.7 kg/m², p = 0.03). Patients with FSH/LH deficiency were less likely to be working after TBI (25% vs. 62%, p = 0.002). Male patients with gonadotropin deficiency were more likely to have low libido (54% vs. 32%, p = 0.050) or erectile dysfunction (61% vs. 20%, p = 0.004) in comparison with those without hypogonadism. Lower GAF scores were associated with FSH/LH deficiency, (47 vs. 55, p = 0.03). We found no association between the presence of gonadotropin deficiency and neuroimaging findings (data not shown).

Of the 166 subjects studied, 121 patients underwent GST. Thirty-four patients (21%) had a peak GH <3ng/mL. Such patients had a higher mean BMI (p = 0.0005) and were less likely to be employed after TBI (41% vs. 63%, p = 0.04) than those without GH deficiency. Patients with GH deficiency also had lower mean total testosterone (311 ng/dL vs. 464 ng/dL, p = 0.0005) and calculated free testosterone (7.0pg/mlL vs. 9.5 pg/mL, p = 0.01), using the law of mass action. There were no significant differences in serum IGF-1 values between these two groups (data not shown). When other pituitary hormone deficits in 34 GH-deficient patients were individually analyzed, 11 of the 34 GH-deficient patients also had FSH/LH deficiency.

Recent data have suggested that the GH response on GST is BMI-dependent in healthy adults. ¹⁵ We re-analyzed our data, defining cut-points diagnostic of GH deficiency according to BMI. More specifically, patients with BMI >25 kg/m² were considered to be GH deficient when peak GH during the GST was <1 ng/mL. Using this new diagnostic cut-point, the number of GHD patients decreased to 18 (15%). Those patients were more likely to be hospitalized for TBI (76% vs. 46%, p = 0.03). In addition, patients with GHD deficiency had lower total testosterone (298.5 ng/dL vs. 448.7 ng/dL, p = 0.004) and free testosterone (6.6 pg/mL vs. 9.4 pg/mL, p = 0.03). We also found a trend toward a more frequent presence of intracranial hemorrhage on neuroimaging in this group, (60% vs. 29%, p = 0.07). There was no association between the cause of the injury and the presence of GH deficiency (data not shown).

Ten patients (6%) had central DI and were more likely to have suffered MVA (100% vs. 32%, p = 0.001), have been hospitalized for TBI (100% vs. 46%, p = 0.0006), have suffered post-traumatic seizures (30% vs. 5%, p = 0.02), and have demonstrated to have any degree intracranial hemorrhage (100% vs. 30%, p = 0.003), SAH (60% vs. 8%, p = 0.009), or focal cortical (frontal) contusions (100% vs. 17%, p = 0.0003) on MRI than patients without DI. Moderate and severe TBI was more frequent among patients with DI (p < 0.001).

Discussion

In the past several years, a number of studies have found that hypopituitarism following TBI is much more common than previously thought. However, there are few published data on risk factors associated with the presence of pituitary hormone deficiencies in this population. A major aim of our study was to identify predictors of the development of pituitary hormone deficiencies after TBI to provide some guidance to clinicians with regard to patients at highest risk and therefore most in need of testing for pituitary hormone deficiencies. Our data showed that post-traumatic seizures, intracranial hemorrhages, petechial brain hemorrhages, and focal cortical contusions are all associated with an increased risk for central adrenal insufficiency. TBI caused by MVA was a risk factor for both TSH and ACTH deficiency. Hospitalization after TBI and the presence of intracranial hemorrhage on neuroimaging were associated with GH deficiency (imaging abnormalities showed a trend toward statistical significance). All patients with DI had suffered MVAs and had evidence of intracranial hemorrhage, SAH, or focal cortical contusion on MRI. Therefore, TBI patients who have experienced MVA and those with post-traumatic seizures, intracranial hemorrhage, petechial brain hemorrhages, and/or focal cortical contusions are at high risk for endocrine dysfunction, including central adrenal insufficiency and DI, and should be referred for pituitary hormone testing. It should be noted that a substantial number of patients without these risk factors also developed pituitary hormone deficiencies. Therefore, patients with TBI should be monitored for any signs or symptoms that indicate hypopituitarism.

In the present study, we found the overall prevalence of hypopituitarism after TBI to be broadly comparable with previously published data. The prevalence of central hypoadrenalism has been reported to vary between 4 and 53% in the acute phase. ^{5,17 –20} Kopczak and colleagues tested 56 patients (41 with TBI and 15 with SAH) and found that 37.5% of them had hypoadrenalism, based on the results of ITT. ²¹ More recently, Kopczak and associates tested 509 patients after TBI and SAH (time between injury and testing ranging between <1 month and 39 years) and reported adrenal insufficiency in 1.4% of patients with TBI, using a basal morning cortisol level <5 mcg/dL as the diagnostic cutpoint. ²² Based on the results of cosyntropin-stimulation testing, the prevalence of hypoadrenalism increased to 11.5%. ²²

Most prospective studies have shown that GH and gonadotropin deficiency are the most common pituitary hormone deficiencies in patients with TBI. ^{5 –9,19} However, few studies have used rigorous stimulation testing to assess the somatotroph axis post-TBI. To diagnose GH deficiency, we used the GST, which is a validated test for this purpose. ^{23 –27} Growth hormone secretion is reduced in otherwise healthy overweight and obese individuals without TBI, and approximately one third of our cohort was overweight or obese. Therefore, although the widely used GH deficiency diagnostic cutpoint is 3 ng/mL, we also analyzed our data using the recently proposed diagnostic cutpoint of 1 ng/mL for overweight and obese individuals. ^15,28,29 Without the BMI adjustment and using the widely used cutpoint for GHD, 21% of the patients were diagnosed as GH deficient. However, using BMI-adjusted criteria, the prevalence decreased to 15%. These data underscore the importance of rigorous diagnostic criteria in order to accurately identify pituitary hormone deficiencies. The prevalence of hypogonadism in our study was 8.6% in contrast to higher rates reported in the literature, including a study by Tanriverdi and co-workers that reported a 41.6% prevalence in a prospective study of 52 patients evaluated in the acute phase (within 24 h). ³⁰ Recently, Kopczak and colleagues reported hypogonadism in 40.7% of all tested men (322 patients) with an interval between TBI and evaluation ranging between <1 month and 39 years. ²² One factor that may have contributed to the very high prevalence of hypogonadism in these studies may be the inclusion of patients with acute injury, as the effect of severe, acute illness causing transient hypogonadotropic hypogonadism is well established. DI has been well recognized in TBI patients for several decades. Published data suggest that the prevalence of DI ranges between 3% and 26%. ^{31 –33} In this study we found a prevalence of 6%.

Regarding blast-induced TBI, we found 7 patients with evidence of pituitary deficiency (representing 14% of the study population and 39% of patients with blast-induced TBI). The prevalence of hypopituitarism in patients with a history of blast-induced TBI has been reported previously as 42% (11/26 of patients in the study by Wilkinson and associates) and 32% (6/19 in the study by Baxter and co-workers), possibly reflecting differences in study populations, timing of the assessment, or diagnostic criteria for hypopituitarism. ^34,35

Several previous studies reported no association between the severity of the trauma and pituitary dysfunction. ^5,6,19 In contrast, other authors have suggested that the severity of the trauma is an indicator of the risk for hypopituitarism. ^7,18 In their meta-analysis of 19 studies reporting pituitary function after TBI or SAH (including a total of 1137 patients), Schneider and colleagues found that the risk of hypopituitarism was higher in patients with severe TBI in comparison with those with mild trauma (35.3% vs. 16.8%). ⁴ They proposed that diffuse axonal injury and basal fractures might be predictive of pituitary deficiency. However, predictors of individual pituitary hormone deficiencies were not examined. In our study, the severity of TBI was partly based on the initial post-resuscitation GCS score, with a score of 3–8 indicating severe, 9–12 moderate, and 13–15 mild TBI. ³⁶

We described potential predictors of hypopituitarism, so that patients at risk could be promptly evaluated. However, our data suggest that no single variable accurately predicts the presence of hypopituitarism. Thus we propose that patients with mild TBI and persistent symptoms, as well as all patients with moderate to severe TBI, be considered for possible hypopituitarism. Hypoadrenalism and DI should be considered both in the acute and chronic phase after TBI. In contrast, other pituitary axes should be evaluated after recovery from acute illness to minimize the confounding effects of acute injury. In the present retrospective study, patients were generally evaluated at varying intervals during recovery after TBI, depending on the timing of clinic referral. However, the large majority of our patients were seen ≥12 months after TBI. As a consequence, no insights regarding the optimal timing of endocrine evaluation can be gleaned from our data.

Strengths of our study include the substantial number of patients that underwent detailed endocrine evaluation using stimulation testing. In addition, our study is the first one to use BMI-dependent diagnostic cutpoints for GH levels measured during GST. However, our data need to be interpreted cautiously due to potential limitations arising from the nature of this cross-sectional and retrospective study. Potential referral bias should also be considered, as we only tested patients who had symptoms and were referred to our neuroendocrine center. With the exception of post-traumatic seizures, low libido or erectile dysfunction, our study did not find evidence of a significant association between symptoms or comorbidities examined and the presence of hypopituitarism.

In summary, we found that predictor variables associated with the presence of individual hormone deficiencies (central adrenal insufficiency and DI) include the type of injury, specifically MVA, and the presence of specific abnormalities on neuroimaging. These patients warrant detailed pituitary hormone testing. However, some patients who did not meet these criteria also developed hypopituitarism, which can have serious medical consequences if unrecognized. History of mild injury does not exclude the development of pituitary hormone deficiencies. Therefore, patients with TBI and signs or symptoms suggestive of hypopituitarism should be evaluated by an endocrinologist regardless of severity of TBI, mode of injury, or lack of imaging abnormalities.