Mapping the Steroid Response to Major Trauma From Injury to Recovery: A Prospective Cohort Study

Abstract Context Survival rates after severe injury are improving, but complication rates and outcomes are variable. Objective This cohort study addressed the lack of longitudinal data on the steroid response to major trauma and during recovery. Design We undertook a prospective, observational cohort study from time of injury to 6 months postinjury at a major UK trauma centre and a military rehabilitation unit, studying patients within 24 hours of major trauma (estimated New Injury Severity Score (NISS) > 15). Main outcome measures We measured adrenal and gonadal steroids in serum and 24-hour urine by mass spectrometry, assessed muscle loss by ultrasound and nitrogen excretion, and recorded clinical outcomes (ventilator days, length of hospital stay, opioid use, incidence of organ dysfunction, and sepsis); results were analyzed by generalized mixed-effect linear models. Findings We screened 996 multiple injured adults, approached 106, and recruited 95 eligible patients; 87 survived. We analyzed all male survivors <50 years not treated with steroids (N = 60; median age 27 [interquartile range 24–31] years; median NISS 34 [29–44]). Urinary nitrogen excretion and muscle loss peaked after 1 and 6 weeks, respectively. Serum testosterone, dehydroepiandrosterone, and dehydroepiandrosterone sulfate decreased immediately after trauma and took 2, 4, and more than 6 months, respectively, to recover; opioid treatment delayed dehydroepiandrosterone recovery in a dose-dependent fashion. Androgens and precursors correlated with SOFA score and probability of sepsis. Conclusion The catabolic response to severe injury was accompanied by acute and sustained androgen suppression. Whether androgen supplementation improves health outcomes after major trauma requires further investigation.


INTRODUCTION 59
Over 5 million people worldwide die each year from serious injury (1), with almost 25% caused by road 8 RTC (n=11; 18%) were the most common causes of injury. Twenty-five (42%) patients had at least one septic 157 episode and most occurred in the second week (Fig 2C). 158

Androgen Biosynthesis and Activation After Major Trauma 170
Serum concentrations of the adrenal androgen precursor dehydroepiandrosterone (DHEA) were very low 171 after injury (p<0·0001, compared with healthy controls) but recovered to the normal range by three months post-172 injury (Fig. 3D, Suppl. Fig. 1) (34). In contrast, its sulfate ester, DHEAS, demonstrated sustained suppression; 173 median serum DHEAS concentrations did not recover to values within the healthy reference range, even at the end 174 of the 6-month study period (Fig. 3E). Consequently, the serum DHEA-to-DHEAS ratio (Fig. 3F) increased by 175 week 2 compared with controls and failed to return to normal during the 6-month study period. The serum cortisol-176 to-DHEAS ratio (Fig. 3G) increased post-injury, peaking at 2 weeks, followed by a gradual decrease, but without 177 returning to normal by the end of the 6-month study period. 178 Serum concentrations of the androgen precursor androstenedione (Fig. 3H) were below the reference 179 range immediately after injury, recovering to the mid reference range at 2 weeks post-injury. Thus, serum 180 androstenedione concentrations recovered much faster than DHEA, suggestive of rapid downstream activation of 181 DHEA to androstenedione. 182 Serum testosterone (Fig. 3I, Suppl. Fig. 4A+B) (34) was very low following injury, starting to increase 183 after two weeks, and recovering to the healthy sex-and age-matched reference range approximately eight weeks 184 after injury. This was mirrored by acute suppression of serum LH immediately after injury, followed by recovery 185 to the normal range approximately 2 weeks after injury (Suppl. decrease in DHEA sulfation ( Fig. 3D-F). The overall decrease in androgen production was paralleled by a 194 profound decrease in systemic 5a-reductase activity (Suppl. Fig. 4E+F) (34), and hence in androgen activation, 195 as 5a-reductase is responsible for converting testosterone to the most potent androgen 5a-dihydrotestosterone. 196

Protein Catabolism After Major Trauma 197
The 24-hour total urinary nitrogen (TUN) excretion increased immediately after trauma, peaking at 198 25.0±16.1 g/day at the end of the first week, returning to below 15·0 g/day by week-4. The mean maximum rate 199 of nitrogen excretion was 33.0±21.3 g/day (Fig. 4A). The normalization of TUN excretion coincided with the 200 gradual recovery of adrenal and gonadal androgen production (Fig. 4B+C). 201 The biceps brachii muscle was the most reliable site for ultrasound measurement of muscle thickness; 202 dressings, amputations and other wounds hampered the measurements of the other muscle areas. Changes in 203 biceps brachii muscle thickness followed a U-shaped curve after injury, reaching a nadir at 6 weeks (day-1 after 204 trauma compared with week-6, p=0·001). The mean muscle loss was 22.7±12.5% (Fig. 4D). Similar to TUN, 205 muscle thickness recovered alongside gradually increasing adrenal and gonadal androgen production (Fig. 4E+F). 206

Clinical Course of Post-Traumatic Recovery and Serum Androgen Dynamics 207
The relationship between adrenal and gonadal androgens and the Sequential Organ Failure Score (SOFA) 208 and probability of sepsis are illustrated in Fig. 5. During the first four weeks, serum DHEA, DHEAS, and 209 testosterone all correlated with the clinical SOFA score (autocorrelation factor (ACF) = 0·85, 0·90 and -0·79, 210 respectively). The serum concentrations of all three steroids also showed strong associations with the probability 211 of sepsis (R=-0·85, 0·85 and -0·97 for serum DHEA, DHEAS and testosterone, respectively). SOFA score and 212 probability of sepsis also correlated strongly with the DHEA:DHEAS ratio (autocorrelation factor (ACF) = -0·94 213 and -0·96 respectively) and with the serum cortisol-to-DHEAS ratio, negatively for the SOFA score but positively 214 for probability of sepsis (autocorrelation factor (ACF) = -0·81 and 0·89, respectively) (Suppl.

Opioid administration and endocrine recovery 216
To examine whether opioid administration affected endocrine recovery, we modelled the impact of the 217 total cumulative in-patient opioid dose on circulating steroid concentrations during recovery from major trauma. 218 For this purpose, we categorised patients according to cumulative opioid dose. Modelling took into account the 219 differences in ISS, length-of-stay (LOS), ICU LOS, and SOFA score. 220 The adjusted modelling revealed a dose-dependent impact of opioid treatment, with a higher initial peak 221 of serum cortisol and the cortisol/cortisone ratio in those on higher doses (≥3000mg) while those on lower doses 222 had initially lower serum cortisol concentrations but showed better recovery of cortisol and cortisol/cortisone 2 223 months into the recovery period, with broad interindividual variability in those with high cumulative opioid doses 224 ( Fig. 6). 225 Opioid administration showed a pronounced, dose-dependent effect on adrenal and gonadal androgen 226 production, with significantly delayed recovery of serum DHEA and DHEAS in patients on higher opioid doses 227 (p=0.029, p=<0.001 respectively; Fig. 7). By contrast, serum testosterone concentrations, which were initially 228 equally suppressed in all cumulative dose groups, showed a much faster recovery in individuals who received 229 higher (≥3000mg) total cumulative opioid doses. However, these confidence intervals were large for these model 230 estimates (Fig. 7). 231

DISCUSSION 232
In this study, we have characterized the response of adrenal and gonadal steroids and catabolic metabolism 233 to severe injury, describing the related dynamic changes for six months post-injury. Modelling the data has 234 allowed us to provide a detailed description of the transition from catabolism to anabolism during recovery from 235 severe injury, including investigating the impact of cumulative in-patient opioid dose. Our data are the first to 236 provide detailed adrenal and gonadal steroids beyond the first days after trauma in a large cohort of young patients, 237 with all patients recruited prospectively and steroid analysis carried out by tandem mass spectrometry. 238 As summarized in a recent meta-analysis (35), previous data on serum cortisol after injury are limited to 239 small cohorts derived from elective surgery, rarely followed up for more than two days. In our study, serum 240 cortisol quickly returned to normal following slight initial increases after acute trauma. In contrast, serum 241 cortisone remained low for three months post-injury. Our study revealed an initial phase of minor glucocorticoid 242 activation with a transient increase in the serum cortisol-to-cortisone ratio, with changes in urinary glucocorticoid 243 metabolites indicative of increased 11β-HSD1 activity. The cortisol-activating enzyme 11β-HSD1 is the major 244 enzyme converting inactive cortisone to cortisol and has been shown to be upregulated systemically and locally in 245 response to inflammation, thereby dampening the inflammatory response (36,37). Skeletal muscle expresses 11β-246 HSD1 (38), Previous studies reported increased 11β-HSD1 activity in an animal model of trauma haemorrhage 247 (39). and improved wound healing in mice treated with 11β-HSD1 inhibitors (40). However, human data after 248 trauma are lacking. There is substantial evidence indicating a reduced cortisol clearance in critical illness, due to 249 decreased cortisol inactivation in liver and kidney (41), this mechanism could also be responsible for the slight 250 changes in cortisol and cortisone we observed. This was corroborated by the observed reduction in the urinary 251 cortisol-to-cortisone ratio, which is reflective of 11β-HSD2 activity. 252 Interestingly, patients on higher opioid doses, showed a higher early peak in cortisol production after 253 trauma, followed by persistently lower circulating cortisol during the recovery period, as compared with patients 254 on lower opioid doses. Previous reports have described suppressive effects of opioids on the HPA axis, though 255 studies in smaller mammals have indicated an acute stimulatory effect of opioid administration on serum cortisol 256 concentrations (42,43).

12
We observed a pronounced and sustained loss of adrenal and gonadal androgen synthesis within the first 258 24 hours following acute major trauma. The recovery of circulating DHEA and testosterone concentrations took 259 two and four months post-injury, respectively, and DHEAS remained pathologically suppressed at the end of the 260 six-month follow-up period. In a mouse model of acute inflammation, sustained suppression of the expression of 261 the DHEA sulfotransferase SULT2A1 and its sulfate donor enzyme, PAPSS2, have been described (44). We 262 reviewed 23 previous studies that measured serum DHEA and DHEAS in critically ill patients (Suppl . Table 1) 263 (45), but most studies followed patients for only a few days and relatively few patients suffered from acute trauma. 264 One previous study measured both serum DHEA and DHEAS in 181 patients with septic shock, and 31 patients 265 with acute hip fracture (46). Serum DHEAS was decreased in both groups, while DHEA was increased in sepsis 266 but decreased after trauma. This suggested an inflammation-mediated downregulation of DHEA sulfation after 267 trauma, resulting in a dissociation of serum DHEA and DHEAS. In our study, this was also observed, as indicated 268 by a sustained increase in the serum DHEA/DHEAS ratio and persistently low serum DHEAS concentrations. studies have demonstrated that cortisol decreases neutrophil superoxide production, which is counteracted by 272 coincubation with DHEAS (47). Furthermore, we have previously shown that DHEAS, but not DHEA, directly 273 enhances neutrophil superoxide generation; a key mechanism of human bactericidal function via activation of 274 protein kinase C-β, independent of androgen receptor signalling (51). In the present study, carried out in severely 275 injured men younger than 50 years of age, we observed suppression of both serum DHEA and DHEAS post-injury, 276 indicating that the loss of adrenal androgen synthesis is a trauma-related event. Importantly, we showed for the 277 first time that this decrease in circulating adrenal androgen precursors is sustained for several months, and that 278 DHEAS remains low even six months post-injury. 279 Alongside the decrease in adrenal androgen synthesis, we observed a near complete loss of gonadal 280 testosterone production and pituitary LH secretion immediately after trauma. Both the gradual recovery of adrenal 281 and gonadal androgen production paralleled the decrease in catabolism, as assessed by urinary nitrogen excretion 282 and biceps muscle thickness. The suppression of the hypothalamus-pituitary-gonadal (HPG) axis after severe 283 13 injury shown in our study is supported by the literature (52-55). Our prospective, longitudinal data demonstrate 284 that suppression of the HPG axis is of shorter duration than that of the HPA axis. In traumatic brain injury studies, 285 a significant proportion of patients go on to develop anterior pituitary dysfunction including secondary 286 hypogonadism (56). However, in our study traumatic brain injury was an exclusion criterion. While limited data 287 from patients with burns and critical illness have suggested a central, hypothalamic-pituitary cause of trauma-288 related hypogonadism (57,58), the evidence prior to our study has been limited. Our data indicated a central cause 289 of suppression to the gonadotrophic axis, with a decrease in both pituitary LH and gonadal testosterone. 290 Interestingly, we observed a differential impact of the cumulative opioid dose on adrenal and gonadal androgens, 291 respectively, with a significantly delayed recovery of DHEA and DHEAS, but a trend towards faster recovery of 292 gonadal testosterone synthesis in patients with higher cumulative opioid doses. Previous data on opioid effects on 293 adrenal androgen production are very scarce, but our findings with respect to gonadal testosterone biosynthesis 294 contrasted previous studies describing suppressive opioid effects on the HPG axis (42,43).

295
Our study revealed a loss of both adrenal and gonadal androgen production in young and middle-aged 296 men after major trauma. This effect was further enhanced by long-lasting suppression of androgen-activating 297 systemic 5a-reductase activity, as demonstrated by urinary steroid metabolite analysis. Androgens are important 298 in wound healing, erythrocytosis, bone density and muscle mass (59). The catabolic state that occurs following 299 trauma thus presents a significant challenge. The use of androgens to ameliorate catabolism has some precedent, 300 as evidenced by the use of the synthetic androgen, oxandrolone, that has some proven benefit in treating burn 301 injury (10).  The strengths of our study include its prospective nature, narrow age range of the patients, single 307 gender, single site for recruitment and analysis and detailed follow-up over six months as well as the 308 measurement of circulating (and in a smaller cohort also excreted) steroid hormones by state-of-the-art 309 14 mass spectrometry assays. Analysing a young to middle-aged patient cohort has also reduced the 310 confounding effects of age-related co-morbidities. Another strength is the unique opportunity our study 311 offered for analysis of the opioid effects on endocrine recovery, facilitated by detailed prospective, 312 longitudinal phenotyping with dedicated software.

313
Our study was limited by the diverse nature of major trauma patients in relation to injury pattern and the 314 involvement of military casualties. The timing and number of observations during our study was pragmatic and 315 some statistical comparisons were made using modelled data. While we measured total cortisol and cortisone by 316 tandem mass spectrometry, we did not measure free cortisol or cortisol-binding globulin. We were only able to 317 measure urinary steroid excretion in a sub-cohort of patients, as accurate and repeated collection of 24-h urine 318 proved very challenging under ICU conditions. The estimation of nitrogen excretion was pragmatic due to the 319 diverse nature of the patients and we were not able to record nitrogen intake. Ultrasound estimation of muscle 320 thickness was performed at four different body sites, but many individuals had limbs missing or extensive wounds 321 that prevented measurements. While imperfect, the longitudinal nature of these measurements allowed us to model 322 these changes over time. 323 In conclusion, in this most detailed and first prospective study of the steroid response to major trauma, we 324 followed the patients from severe injury to six months of recovery, revealing pronounced and sustained decreases 325 in adrenal and gonadal androgen biosynthesis. Recovery of androgen production in the severely injured patients 326 was mirrored by a switch from catabolism to anabolism as reflected by recovery of muscle mass and a decrease in 327 nitrogen loss. Adrenal and gonadal androgens correlated with risk of sepsis. It is tempting to suggest that an 328 anabolic intervention with androgens or androgen precursors could have a beneficial effect on health outcomes 329 during recovery from major trauma. However, this will need to be investigated by future intervention studies.

15.
Le Gall J-R, Lemeshow S, Saulnier F.     Data are represented after modelling of the raw data (Suppl. Fig. 1) using a non-linear mixed effects model that 532 accounts for unbalanced repeated measures using a 4-knot cubic spline. Modelled data are shown as means and 533 95% confidence intervals. 534   days 1-7-14-28, of the burned at 1-7-14-28-56, as well as after 6 and 12 months.
All androgens (T, DHT, FT) decreased significantly in the males, DHEAS decreased in male and females.