Growth hormone deficiency testing and treatment following mild traumatic brain injury

Pituitary dysfunction, specifically growth hormone (GH) deficiency, can occur following traumatic brain injury. Our objective was to characterize the prevalence of GH deficiency (GHD) testing and response to recombinant human GH (rhGH) treatment in adults with persistent symptoms following mild traumatic brain injury (mTBI) referred for assessment of pituitary dysfunction. A retrospective chart review was conducted of patients seen at an outpatient brain injury clinic with a diagnosis of mTBI and persistent post-concussive symptoms who were referred to endocrinology. Clinical assessments of symptoms were collected. Investigations and results of GHD were collected, including initiation of rhGH treatment and treatment response. Of the 253 patients seen in both brain injury and endocrinology clinics, 97 with mTBI were referred for investigation of pituitary dysfunction and 73 (75%) had dynamic testing for assessment of GHD. Of the 26 individuals diagnosed with GHD, 23 (88%) started rhGH. GH therapy was inconsistently offered based on interpretation of GH dynamic testing results. Of those who started rhGH, 18 (78%) had a useful treatment response. This study suggests that clinical management of these patients is varied, highlighting a need for clear guidelines for the diagnosis and management of GHD following mTBI.

. Demographics and past medical history of patients referred to OECs and who completed dynamic testing. mTBI mild traumatic brain injury, OECs outpatient endocrinology clinics. www.nature.com/scientificreports/ Growth hormone (GH) therapy. Peak GH thresholds used for prescription of recombinant human GH (rhGH) therapy differed between clinicians. Twenty-six patients with a mean peak GH of 2.3 ± 1.8 µg/L (range 0.1-5.5 µg/L) were offered rhGH therapy. Of those offered therapy, 8 (31%) had a peak GH > 3 µg/L. Three patients (12%) did not proceed with rhGH due to cost or lack of insurance. Therapy was inconsistently offered for peak GH results of 3.6-5.5 µg/L. The dynamic testing results of 4 patients with peak GH result of 3.6-5.5 µg/L were interpreted as normal and therefore they were not offered rhGH (Fig. 5).
Response. Timepoint of response determination varied depending on when patients started rhGH.
Response (responder versus non-responder) was determined at their most recent clinic assessment. For 9 (39%) patients, response was determined 3 months following treatment start. Response determination for 14 (61%) patients was at 1 year or greater. At time of response determination, 18 (78%) patients had responded to treat- www.nature.com/scientificreports/ ment with self-reported benefit, deciding to continue rhGH. Of the 18 patients who responded to therapy, 13 (72%) had severe GHD and 5 (28%) had GHI. After starting rhGH, 5 (22%) patients discontinued treatment.
Of those who discontinued treatment, 2 stopped after 1 year or more due to lack of continued benefit, 2 did not have self-reported benefit within the first 3 months, discontinuing rhGH and one patient reported side effects of enlarged hands and feet and soft tissue swelling.

Discussion
It is only recently that the risk of hypopituitarism following mTBI has been recognized 14 . Many of the psychosocial symptoms of GHD overlap with persistent post-concussive symptoms suggesting that in some cases, untreated endocrinopathy may contribute to post-injury impairment 15 . Previous studies examining GHD following mTBI have focussed on combat-related 16,17 or repetitive sport-related concussion cohorts 10 . This study presents one of the largest reported cohorts of patients assessed for GHD following mTBI. Our findings highlight the importance of screening for pituitary dysfunction in adults with persistent symptoms following mTBI, further suggesting that the majority of patients with GHD who undergo treatment improve. However, we observed great variability in the clinical management of these patients; only a small number of individuals presenting to Figure 2. GHD prevalence using alternative diagnostic cut-offs based on peak GH during dynamic testing. Numbers of patients meeting differential diagnostic cut-offs with percentages calculated out of total number of individuals tested (n = 73). GH growth hormone, GHD growth hormone deficiency. www.nature.com/scientificreports/ outpatient brain injury clinic services were referred for endocrinological assessement, and even fewer progressed to GH testing and subsequent therapeutic GH replacement, possibly due to variable interpretations of GH testing. In light of the high prevalence of GHD/GHI in patients tested in this sample, and the suggested benefit of rhGH treatment, there is a great need for collaboration between specialized brain injury clinicians and endocrinologists. Further research and development of care pathways, specifically between brain injury and endocrinology services, may allow for better clinical management of those with persistent post-concussive symptoms. Previously, post-traumatic hypopituitarism was not thought to result from milder injury. However, a systematic review of post-traumatic pituitary dysfunction reported a pooled prevalence of hypopituitarism of 16.8% following mTBI 14 . Despite this finding, consensus guidelines and expert opinions have recommended that only patients with moderate or severe TBI, an mTBI requiring hospitalization for at least 24 h, or an mTBI with an abnormality on initial CT be screened for post-traumatic pituitary dysfunction 9,18,19 . The lack of investigation of GHD following uncomplicated mTBI has likely contributed to the absence of screening recommendations for this population. Therefore, determination of which patients should be tested for GHD remains an important clinical challenge. This study provides evidence of GHD following uncomplicated mTBI, which raises questions of if and when these patients should be screened for post-traumatic hypopituitarism. Collaborative development of guidelines by brain injury and endocrinology clinicians to guide referral of patients following uncomplicated mTBI for investigation of post-traumatic GHD is an important next step.
Post-traumatic GHD may be transient, suggesting that screening acutely post-injury is not of clinical benefit 20 . However, testing in patients with persistent symptoms greater than one-year post-injury is advised as changes in pituitary function have been shown to persist at 3 and 5 years post-injury 8,21 . In this study, patients underwent dynamic testing a mean of 25 ± 11.3 months post-injury. Testing at this timepoint may be explained by several factors. Many patients did not present to the outpatient brain injury clinic less than one year post-injury and therefore may not have been referred to endocrinology sooner. Treating physicians may also elect to first try other less invasive treatment options prior to referring for assessment of pituitary dysfunction. While assessment of pituitary dysfunction at one year post-injury is appropriate as soon as referral for dynamic testing is made by the treating endocrinologist, depending on the healthcare system, wait times may determine the assessment  www.nature.com/scientificreports/ time course. In our health care system, it may be several months from the time of referral to assessment by an endocrinologist and subsequent testing. Of those tested in our sample, 63% met criteria for severe GHD or GHI. This supports previous data that GHD can persist at one year or more following mTBI 8,21 . Several factors may contribute to the low rate (3%) of referral of patients seen at CBIP for endocrinological assessment. Patients seen at CBIP may have moderate to severe brain injury or other acquired brain injury and therefore would not be included in this study. Additionally, many patients may present with vague clinical signs, or symptoms that may not be recognized as post-traumatic GHD. Treating physicians may also elect to pursue less invasive therapies or testing first.
Our findings suggest there is a need for better standardization of timepoints for post-traumatic pituitary dysfunction screening in this population. With an estimated prevalence of GHD of 5-40% 5-8 at 1 year post-mTBI, it is possible that the prevalence of GHD in the total CBIP sample would be higher than reported in this study. The high treatment response rate (78%) to rhGH suggests that diagnosis and treatment of post-traumatic GHD may be especially beneficial to those presenting to outpatient brain injury clinics with persistent symptoms. However, there are currently no specific guidelines or screening tools that can aid clinicians in determining who should be referred for screening of post-traumatic GHD. Given that dynamic testing is labor intensive and expensive, it is not feasible to test all patients with mTBI and persistent symptoms one year post-injury. There is a need for a more cost-effective screening measure for patients with persistent symptoms following uncomplicated mTBI. There is also a need for a comprehensive prospective study evaluating reason for referral for assessment of pituitary dysfunction following mTBI by brain injury clinicians.
Dynamic testing is required for assessment of GHD as serum IGF-1 lacks diagnostic sensitivity and is not recommended by clinical practice guidelines 22 . In a mixed-severity traumatic brain injury cohort, of the 25 individuals diagnosed with GHD based on dynamic testing, IGF-1 level was within the age and gender specific range for all patients, confirming lack of diagnostic utility of IGF-1 in the diagnosis of GHD in this patient population 23 . Despite this, 3 patients referred to endocrinology had serum IGF-1 as the only assessment of GHD. As the remainder of patients received dynamic testing for assessment of GHD, IGF-1 data are not available for the majority of this cohort and thus comparison between dynamic testing results and IGF-1 was not possible. Unfortuantely, in some cases, recent studies have still used IGF-1 as a measure of GHD 16 .
Interpretation of post-traumatic GHD literature is also complicated by the use of various dynamic tests. The ITT is often cited as the gold-standard, while GST is frequently used when ITT is contra-indicated. GHRH + Arginine and GHRH + AGRP-6 tests have also been used, although these are no longer available in Canada. The majority of this sample underwent GST (94%) for assessment of GHD. Additionally, diagnostic cut off values vary across groups and may or may not be body mass index (BMI)-adjusted. The most recent American Association of Clinical Endocrinologists (AACE) guidelines suggest a normal response to ITT is indicated by a peak GH > 5 µg/L, while a normal response to GST should be > 3 µg/L for BMI < 25 kg/m 2 and > 1 µg/L for BMI > 25 kg/ m 222 . Few studies have used BMI-adjusted interpretation of GST data and a variety of other cut-offs are used in the literature. A recent randomized control trial of rhGH for treatment of GHD following mTBI used a broader diagnostic criteria, including individuals with a peak GH < 8 µg/L during GST 11 . Using this criteria, 14 additional patients in this sample would have met criteria for rhGH treatment.
We observed a wide variation in the interpretation of test results and decisions to offer rhGH replacement therapy among endocrinologists. This may reflect the variation in biochemical thresholds for diagnosis recommended by different guidelines. As well, most studies of GHD in the endocrinology literature are conducted in patients with pituitary surgery. Therefore, many endocrinologists may be unware or uncertain as to whether mTBI patients should be diagnosed with GHD following the same guidelines. Lastly, there are very few studies of rhGH replacement in those with mTBI such that endocrinologists may be reluctant to offer rhGH therapy due to uncertainty of benefit. Our findings highlight the benefit of treating individuals with GHD following mTBI. A large randomized controlled trial evaluating the efficacy and benefits of GH replacement is needed. There is also a great need for determination of response according to peak GH during dynamic testing to inform implementation of a standardized cut-off point for treatment of these patients.
Few clinical determinants of GHD following mTBI have been identified. Previous literature has reported associations between post-traumatic GHD and older age, increased BMI and waist circumference 6,17,[24][25][26] . Injury severity based on GCS has not been found to be associated with GHD 27,28 . In this study, we found that none of the clinical measures commonly collected in the outpatient brain injury clinic setting are significantly associated with severe GHD, although this may be due in part to the relatively small sample size. A larger, adequately powered study is needed to evaluate clinical predictors of GHD in individuals with uncomplicated mTBI.
We did not find a significant difference in cognitive impairment between individuals with and without severe GHD. While not significant, the estimated marginal mean MoCA score in those with GHD met criteria for mild cognitive impairment, while the MoCA score for those without GHD did not. Similarly, decreased inhibition of cognitive interference has been reported in veterans with GHD following combat-related mTBI 17 . The measures used in these studies cannot be directly compared and the study samples were notably different (civilians versus veterans). Associations between cognition and GHD have primarily been reported in cohorts following moderate or severe TBI. Several studies have reported lower scores on measures of attention and memory in those with GHD 26,28 . Additionally, lower scores on the Level of Cognitive Functioning Scale at hospital discharge were positively correlated with peak GH levels (during dynamic testing) in a sample 6-12 months following moderate or severe TBI 29 . In non-TBI individuals with GHD, cognitive performance was shown to improve with GH replacement in a meta-analysis of 340 patients 30 . Future work should assess cognition in individuals with and without GHD following mTBI using a full neuropsychological battery to examine if a specific sub-test(s) may be a useful preliminary screen of hypopituitarism. Establishing one or multiple clinical measures that are most predictive of post-traumatic GHD would help to inform screening guidelines and be of great clinical utility. www.nature.com/scientificreports/ Although not found in this sample, depression has been associated with GHD in other cohorts. Higher depression scores using the Beck Depression Inventory were reported in those with GHD following combatrelated mTBI 17 . Higher incidence of depression has also been reported in mixed-severity cohorts using a variety of measures 15,31 . Again, the discrepancy in findings is likely due to differences in injury severity, mechanism of injury and outcome measures between cohorts. Our findings suggest there is a high prevalence of depression in those with persistent post-concussive symptoms presenting to outpatient brain injury services, and therefore depression alone this is not specifically related to post-traumatic GHD.
A correlation with medium effect size was previously reported between severe GHD and Qol-AGHDA scores following combat-related mTBI 17 . Similarly, we reported a correlation between peak GH and Qol-AGHDA scores. While this measure should not be used diagnostically, it may be a superior measure for initial screening of GHD in the mTBI population, compared to specific symptom questionnaires. In a cohort following moderate-severe TBI, scores on the Qol-AGHDA were not significantly different between those with and without GHD. However, the analysis included individuals with both GHD and GHI 15 . On the Short Form (36) Health Survey, a quality of life measure, decreased scores in domains of energy and fatigue and emotional well-being have been reported in two separate mixed-severity TBI cohorts with GHD 15,31 . Our findings support previous work, demonstrating reduced quality of life in those with GHD following TBI and highlights the importance of screening and diagnosis of GHD in these patients.
Of those offered therapy, 88% started rhGH. This illustrates the willingness of patients to follow injection protocols and highlights the effect of persistent symptoms on quality of life and function in this cohort. The benefits of rhGH for GHD have been studied following moderate and severe TBI 12,13 , but few studies have evaluated the response to treatment following mTBI 11 .
The literature is unclear on whether or not GH therapy improves cognition. In a placebo-controlled crossover study of individuals with GHD following mTBI, there was not a significant improvement on neuropsychological testing measures following GH replacement 11 . However, in a sample of individuals with moderate or severe injury, several measures on a neuropsychological battery did improve following a year of GH replacement 12 . While improvements in cognition were not reported with GH replacement in those with mTBI, treatment lead to improvements in fatigue, sleep, mood and an increase in lean body mass 11 . In our sample, 78% of patients who started GH therapy had self-reported benefit and decided to continue with treatment. Unfortunately, lack of testing resources, specialized clinicians and treatment cost may all limit accessibility to treatment. In this cohort, three patients did not proceed with treatment due to associated cost or lack of insurance coverage.
There is evidence to suggest that greater improvement with rhGH treatment may occur over time, specifically in the first year 12 . Therefore, we suspect that if individuals reported improvement of symptoms at 3 months, the response would be even more pronounced at 1 year. Interestingly, two patients decided to stop GH replacement following 1 year or more of therapy. Following prescription of rhGH, current practice is to follow-up with patients at 3 months and yearly thereafter. Response to intervention is assessed at follow-up visits by self-reported benefit and willingness to continue therapy, as described in this paper. It may be of interest to administer common clinical outcomes for objective assessment of symptom improvement over time. The timecourse of response to rhGH in those with post-traumatic GHD is not well understood and further studies are needed to better understand response over time.
This study had several limitations. This cohort is not representative of all adults with mTBI, but rather individuals with months to years of symptoms following injury. Although this was a specific patient population, the results may be generalizable to patients with persistent post-concussive symptoms greater than 1 year. The variable time since injury at clinic assessment and dynamic testing in addition to variable timepoint of response determination are limitations. While we did not have access to BMI data, it is an important factor in the interpretation of dynamic testing results. Future studies with consistent timepoints for testing and treatment response in individuals with mTBI are required.
Additional studies are needed to establish the prevalence of GHD following mTBI. This study only included patients specifically referred for assessment of post-traumatic hypopituitarism and therefore is not indicative of overall prevalence. To date, prevalence data is based on small cohorts following complicated or combat-related mTBI. There is also a great need to identify risk factors of post-traumatic GHD following mTBI. This would serve to inform referral for testing and enable efficient allocation of testing resources. Additional studies are needed to systematically evaluate change in clinical outcome measures (i.e., cognition, quality of life, post-concussive symptoms) following rhGH treatment across timepoints (i.e., 6, 12 months).

Conclusion
Hypopituitarism, specifically GHD, is likely more common in individuals following mTBI than previously appreciated. Treatment for individuals with persistent post-concussive symptoms is largely symptom-based and symptom etiology is often difficult to establish. Our findings provide further evidence that pituitary dysfunction following mTBI should not be ignored and may contribute to chronic symptoms. However, lack of screening tools and specific guidelines to aid clinicians in determining which patients should be screened following uncomplicated mTBI likely contributes to the great variability in screening of these patients. Post-traumatic GHD is a serious, but treatable complication of mTBI. This study demonstrates the need for consensus on referral for testing, interpretation of dynamic testing results and offering of rhGH treatment. Currently the clinical management of these patients is varied and there is no evidence-based gold standard diagnostic cut-off for offering rhGH based on dynamic testing results for post-traumatic GHD. Many patients with post-traumatic GHI, generally defined as a peak GH of 3-10 µg/L, may also benefit from rhGH therapy, and the use of narrow diagnostic criteria should be reconsidered for this patient population. Data collection. The following clinical questionnaire data collected at patients' initial brain injury clinic assessment was extracted.
1. The Rivermead Post Concussion Symptoms Questionnaire (RPQ) is a commonly used questionnaire where 16 post-concussion symptoms are scored between 0 (not experienced at all) and 4 (a severe problem) 33 . A score out of 64 is generated. 2. The Patient Health Questionnaire-9 (PHQ-9) is a brief measure of depressive symptoms 34 . This 9-item measure (scored out of 27) is a valid assessment of depression following TBI 35 . 3. The MoCA is a brief measure of cognitive impairment scored out of 30 36 .
A subset of patients completed the Qol-AGHDA at their initial endocrinology clinic assessment. This is a 25-item assessment of quality of life designed as a preliminary screening tool for GHD where higher scores indicate worse quality of life 20 .
Investigations of GHD ordered by treating endocrinologists were collected, including whether or not an ITT or GST was performed. The decision to proceed with GHD testing, test result interpretation and decision to offer GH replacement were at the discretion of the attending endocrinologist. Presence of other pituitary dysfunction, such as hypogonadism, was also collected. Completion of ITT versus GST was based on availability of nursing staff or contraindication to ITT such as severe cardiovascular disease. All dynamic GH testing (GST or ITT) was completed at testing centers under the supervision of a specialized nurse and followed standard published protocols 37 . Results of dynamic testing were recorded. In accordance with clinical guidelines, peak stimulated GH levels < 3 µg/L were considered severe GHD 38 . GH insufficiency (GHI) was indicated by a peak stimulated GH of 3-10 µg/L, while a response > 10 µg/L was considered normal.
Patients seen by an endocrinologist who did not undergo dynamic testing were classified from electronic medical records as follows: (1) patient reported symptom improvement; (2) patient declined testing; (3) endocrinologist did not order testing/determined it was not indicated; (4) unknown.
Number of patients who started rhGH therapy was collected. Treatment was offered based on the clinical expertise of the treating endocrinologist. For patients offered rhGH who did not proceed with treatment, reasons for declining were recorded. Patients who started rhGH were dichotomized as either treatment responders or non-responders in accordance with our previously published approach whereby patients self-determine the overall value of GH replacement 39 . An individual was classified as a responder if they had self-reported symptom improvement and decided to continue therapy at their most recent clinic visit. Response determination timepoints were at 3 months following treatment start, or greater than one year. Individuals who had started but elected to discontinue rhGH therapy were classified as non-responders. Statistical analysis. Participant demographics, past medical history, injury characteristics and endocrinology assessment data were reported. Cut-offs based on results of dynamic testing were used for classification of GHD. Between group differences were analysed using either chi square or independent-samples t-tests where appropriate. A sensitivity analysis for GHD prevalence was performed using different peak serum GH cut-offs (during dynamic testing) commonly used in the literature for diagnosis of GHD, while also accounting for assay uncertainty (± 0.3 µg/L) 4,11 . Therefore, cut-offs of < 1.0 µg/L, < 1.3 µg/L, < 3.0 µg/L, < 3.3 µg/L, < 5.0 µg/L, < 5.3 µg /L, < 8.0 µg/L and < 8.3 µg/L were included in the analysis. Differences in clinical outcomes between those with and without severe GHD were compared using one-way analysis of covariance (ANCOVA), controlling for age and sex as these may influence GH levels. Association between QoL-AGHDA scores and peak GH on dynamic testing was analyzed using Pearson's correlation.

Data availability
The data from this study are available upon request. www.nature.com/scientificreports/