SEARCH

SEARCH BY CITATION

Keywords:

  • cortisol;
  • hypothalamic–pituitary axis;
  • stroke

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Conflict of interest statement
  9. References
  10. Appendices

Abstract.  Neidert S, Katan M, Schuetz P, Fluri F, Ernst A, Bingisser R, Kappos L, Engelter ST, Steck A, Müller B, Christ-Crain M (University Hospital Basel, Basel, Switzerland; SphingoTec GmbH, Borgsorf, Germany; Kantonsspital Aarau, Aarau, Switzerland) Anterior pituitary axis hormones and outcome in acute ischaemic stroke. J Intern Med 2011; 269: 420–432.

Background.  Early and accurate prediction of outcome in acute stroke is important and influences risk-optimized therapeutic strategies. Endocrine alterations of the hypothalamic–pituitary axis are amongst the first measurable alterations after cerebral ischaemia. We therefore evaluated the prognostic value of cortisol, triiodothyronine (T3), free thyroxine (fT4), thyroid-stimulating hormone (TSH) and growth hormone (GH) in patients with an acute ischaemic stroke.

Methods.  In an observational study including 281 patients with ischaemic stroke, anterior pituitary axis hormones (i.e. cortisol, T3, fT4, TSH and GH) were simultaneously assessed to determine their value to predict functional outcome and mortality within 90 days and 1 year.

Results.  In receiver operating characteristic curve analysis, the prognostic accuracy of cortisol was higher compared to all measured hormones and was in the range of the National Institutes of Health Stroke Scale (NIHSS). Cortisol was an independent prognostic marker of functional outcome and death [odds ratio (OR) 1.0 (1.0–1.01) and 1.62 (1.37–1.92), respectively, P < 0.0002 for both, adjusted for age and the NIHSS] in patients with ischaemic stroke, but added no significant additional predictive value to the clinical NIHSS score.

Conclusion.  Cortisol is an independent prognostic marker for death and functional outcome within 90 days and 1 year in patients with ischaemic stroke. By contrast, other anterior pituitary axis hormones such as peripheral thyroid hormones and GH are only of minor value to predict outcome in stroke.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Conflict of interest statement
  9. References
  10. Appendices

Stroke is the third commonest cause of mortality worldwide and a major cause of long-term disability [1]. Stroke strongly influences an individual’s emotional and socio-economic quality of life [1]. It is estimated that 795 000 people suffered an acute stroke in the United States of America in 2009, of whom 15–30% remain permanently disabled [2]. An early risk assessment with an estimate of the severity of disease and prognosis is pivotal for optimized care and allocation of healthcare resources. The National Institutes of Health Stroke Scale (NIHSS) is a standardized and widely used assessment measure to predict 3-month outcome in acute cerebrovascular events but its use implies special training and there is inter-observer variability [3]. In this context, rapidly measurable markers to predict illness development, outcome and mortality might improve the prognostic accuracy of clinical scores and traditional risk factors.

The classical ‘stress response’ of the body occurs after stimulation of the hypothalamic–pituitary axis (HPA axis) by a stressor. It is characterized by an increase in cortisol levels, depression of thyroid function and a functional deficit of anabolic hormones such as growth hormone (GH) and insulin [4]. Anabolic resistance might contribute to the prolonged whole-body protein breakdown with increased susceptibility for infections and delayed recovery. In cerebral ischaemia, endocrine changes of the HPA axis are one of the first measurable alterations [5, 6]. In addition, low triiodothyronine (T3) levels, as in the ‘low T3 syndrome’, have been described as a prognostic risk factor for death and functional outcome in stroke patients [7]. We therefore evaluated the prognostic value of cortisol, T3, free thyroxine (fT4), thyroid-stimulating hormone (TSH) and human GH on hospital admission in a well-described cohort of 281 patients with ischaemic stroke.

Material and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Conflict of interest statement
  9. References
  10. Appendices

Study design and setting

The design of this prospective cohort study at the University Basel, Switzerland (Clinical Trials.gov number, NCT00390962), has been described in detail elsewhere [8, 9]. Briefly, from November 2006 until November 2007, consecutive patients presenting with acute ischaemic stroke were included. Informed consent was obtained from the patient if possible, otherwise from a relative or from the patient’s physician in the absence of a relative. This study adheres to the consolidated standards for the reporting of observational trials [10] and was approved by the Ethics Committee of Basel, Switzerland.

Patients

Patients were eligible for the study if they were admitted with acute ischaemic stroke according to the World Health Organization criteria [11], with symptom onset within 72 h.

Of a total of 362 eligible patients, blood was collected on day 1 after admission for standardized measurement of cortisol, thyroid hormones and GH in 281 patients; these measurements could not be performed in 81 patients (two patients died, 68 were discharged to another institution before the standardized blood sampling on the day after admission could be performed and blood sampling was omitted in 11 patients by mistake). However, these 281 patients were similar in terms of baseline characteristics [age (P = 0.50), gender (P = 0.82), NIHSS (P = 0.65) and weight (P = 0.89)] compared to the overall cohort. Of the original 281 stroke patients, 268 completed the 1-year follow-up and were available for long-term analysis.

Clinical variables

Within the first 24 h after admission, the following data were recorded: vital signs; relevant co-morbidities assessed by the Charlson comorbidity index (CCI) adjusted for stroke (the CCI is a comorbidity scoring system that includes weighting factors on the basis of disease severity according to the ICD-9-CM system) [12]; medication; traditional risk factors (i.e. age, gender, smoking habits, hypercholesterolaemia, history of hypertension, diabetes mellitus or transient ischaemic attack (TIA)/ischaemic stroke, or positive family history of myocardial infarction, stroke or TIA); and severity of stroke as assessed by the NIHSS [13] by a neurologist certified in the use of this scale. The clinical stroke syndrome was determined by applying the criteria of the Oxfordshire Community Stroke Project: total anterior circulation syndrome (TACS); partial anterior circulation syndrome (PACS); lacunar syndrome (LACS); and posterior circulation syndrome (POCS) [14]. Patients underwent routine laboratory testing and standardized diagnostic work-up to evaluate stroke aetiology. Stroke aetiology was determined according to the criteria of the TOAST classification [15], which distinguishes between large-artery arteriosclerosis, cardioembolism, small-artery occlusion and other or undetermined aetiologies.

Neuroimaging

To exclude intracranial haemorrhage, cranial computed tomography was performed in all patients on admission. Thereafter magnetic resonance imaging (MRI) was performed on a clinical 1.5 T MR Avanto system (SIEMENS, Erlangen, Germany) using a stroke protocol, including T1-, T2- and diffusion-weighted imaging (DWI) sequences, and magnetic resonance angiography. MRI with DWI data was available for 169 patients (60.1%). In these patients, DWI lesion volumes were determined by consensus of two experienced neuroradiologists unaware of the clinical and laboratory results. The lesion size was calculated by the commonly used semi-quantitative method [16]. Lesions were classified into three sizes to represent typical stroke patterns: (i) small lesions with a volume of <10 mL; (ii) medium lesions of 10–100 mL; and (iii) large lesions with a volume of more than 100 mL [17].

Assays

Blood samples were obtained from an indwelling venous catheter the first morning after admission. Routine blood count and C-reactive protein (CRP) levels were measured in all patients. Plasma was collected at the time of blood sampling in plastic tubes containing ethylenediaminetetraacetic acid. The tubes were placed on ice and centrifuged at 3000 g, and plasma was stored at −70 °C until required for assay.

Cortisol was measured with a competitive chemiluminescence immunoassay (IMMULITE 2000; Siemens Medical Solution Diagnostics, Los Angeles, CA, USA) with a calibration range from 28 to 1380 nmol L−1. T3 (nmol L−1), fT4 (pmol L−1) and TSH (mIU L−1) were measured by an electrochemical luminescence immunoassay (Roche Diagnostics, Mannheim, Germany). GH was measured using a high-sensitivity chemiluminescence immunoassay with a functional assay sensitivity of 0.027 ng mL−1 as described recently [18]. For all measurements, levels that were not detectable were considered to have a value equal to the lower limit of detection of the assay.

Outcomes

The primary end-point was functional outcome on day 90. It was assessed by two trained medical students, blinded to hormone levels, with a structured follow-up telephone interview with the patient or, if not possible, with the closest relative or family doctor if no close relatives were available. Functional outcome was assessed by the modified Rankin Scale (mRS) [19]. A favourable functional outcome was defined as an mRS of 0–2 points, whereas an unfavourable outcome was defined as an mRS of >2 points. Secondary end-points were all-cause mortality within 90 days, as well as long-term functional outcome and mortality after 1 year.

Statistical analysis

First, to assess the association between hormones and stroke severity, hormone levels were correlated with NIHSS and with lesion size using Spearman’s rank correlation. Hormone levels were further assessed for different clinical stroke syndromes, and results are presented as median ± interquartile range (IQR). In addition, hormone levels were compared with regard to primary and secondary end-points. Two-group comparisons were performed with the Mann–Whitney U-test and multigroup comparisons with Kruskal–Wallis one-way analysis of variance.

Second, we investigated the association between different hormone levels and both outcomes in univariate logistic regression models, and results are reported as OR. We then adjusted all hormones with significant univariate associations for NIHSS and age, the main outcome predictors within this cohort, as described previously [8]. Further, we performed receiver operating characteristics (ROC) curve analysis to assess discrimination and results are reported as area under the curve (AUC). To study the ability of cortisol to predict mortality, we calculated Kaplan–Meier survival curves and stratified patients by cortisol quartiles. Finally, we calculated reclassification tables [20, 21], and results are reported as net reclassification improvement for outcome and mortality risk categories, as proposed previously [8, 22]. For net reclassification improvement, only those changes in estimated prediction probabilities that imply a change from one risk category to another are considered.

All statistical tests were two-tailed, and P < 0.05 was considered to indicate statistical significance. All statistical analysis was performed with medcalc for windows (version 7.2.1.0.; MedCalc, Mariakerke, Belgium), graph pad prism (version 4; GraphPAd, La Jolla, CA, USA) or stata 9.2 (Stata Corp, College Station, TX, USA).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Conflict of interest statement
  9. References
  10. Appendices

Baseline characteristics of the study population

A total of 281 patients were included in this analysis; 113 (40%) were male and median age was 68 years (IQR 63–82). Median systolic blood pressure was 160 mmHg (IQR 141–180), median diastolic blood pressure was 90 mmHg (IQR 80–100) and median heart rate was 78 beats min−1 (IQR 68–88). A total of 58 (21%) patients were diagnosed with atrial fibrillation, 74 (26%) patients had hypercholesterolaemia, 84 (30%) had a positive family history of cardiovascular events and 98 (35%) were smokers. Median glucose levels were 6.1 mmol L−1 (IQR 5.4–7.4). The main baseline characteristics are summarized in Table 1.

Table 1.   Baseline characteristics of stroke patients
  1. fT4, free thyroxine; IQR, interquartile range; LACS, lacunar syndrome; NIHSS, National Institutes of Health Stroke Scale; PACS, partial anterior circulation syndrome; POCS, posterior circulation syndrome; T3, triiodothyronine; TACS, total anterior circulation syndrome; TSH, thyroid-stimulating hormone.

Demographic characteristics
 Age (years) median (IQR)68 (63–82)
 Male sex (%)59
Clinical findings median (IQR)
 Heart rate (beats min−1)78 (68–88)
 Systolic blood pressure (mmHg)160 (141–180)
 Diastolic blood pressure (mmHg)90 (80–100)
 Temperature (°C)37.0 (36.5–37.5)
 Weight (kg)72 (64–82)
 Height (cm)169 (162–175)
 Body mass index (BMI) (kg m−2)25.2 (24–27.2)
Laboratory findings (median–IQR)
 Cortisol (nmol L−1) (n = 281)480 (344.5–629.8)
 T3 (nmol L−1) (n = 269)1.4 (1.2–1.6)
 fT4 (pmol L−1) (n = 274)15.4 (13.8–17.2)
 TSH (mIU L−1) (n = 275)1.4 (0.9–2.2)
 Growth hormone (ng mL−1) (n = 276)0.4 (0.2–1.1)
 Total cholesterol (mmol L−1) (n = 268)4.4 (3.8–5.1)
 High-density lipoproteins (HDL) (mmol L−1) (n = 281)1.3 (1.1–1.6)
 Low-density lipoproteins (LDL) (mmol L−1) (n = 281)2.4 (1.8–3.0)
 Triglycerides (mmol L−1) (n = 281)1.1 (0.9–1.6)
 C-reactive protein (mg L−1) (n = 281)3.4 (3.0–9.4)
 Glucose (mmol L−1) (n = 281)6.1 (5.4–7.4)
Prognostic scores (median–IQR)
 Modified ranking scale (points)2 (1–4)
 NIHSS (points)5 (2–10)
Stroke syndrome no. (%)
 TACS27 (9.4)
 PACS126 (43.8)
 LACS58 (20.1)
 POCS71 (24.7)
Stroke aetiology no. (%)
 Small-vessel occlusive48 (16.7)
 Large-vessel occlusive54 (18.8)
 Cardioembolic105 (36.5)
 Other13 (4.5)
 Unknown61 (21.2)
Vascular risk factors no. (%)
 Hypertension214 (74)
 Atrial fibrillation58 (20)
 Smoking history98 (34)
 Hypercholesterolaemia74 (26)
 Diabetes mellitus56 (19)
 Coronary heart disease67 (23)
 Prior stroke66 (23)
 Family history of cardiovascular event84 (29)

Pituitary axis hormones and stroke characteristics

NIHSS.  There was a positive correlation between the NIHSS and levels of cortisol (r = 0.32, P < 0.0001) and GH (r = 0.15, P = 0.004), and a negative correlation with levels of T3 (r = −0.27, P < 0.0001) and TSH (= −0.17, P = 0.006). fT4 levels were not correlated with the NIHSS (r = −0.0008, P = 0.78) (Fig. 1).

image

Figure 1.  Correlation between the National Institutes of Health Stroke Scale (NIHSS) and pituitary axis hormone levels. Rank correlation between the NIHSS and levels of cortisol, growth hormone, triiodothyronine, free thyroxine and thyroid-stimulating hormone.

Download figure to PowerPoint

Lesion size.  In patients for whom MRI data were available (n = 169), cortisol levels increased with lesion size. Median cortisol levels in patients with small, medium and large lesions were 461 nmol L−1 (IQR 345–585), 490 nmol L−1 (IQR 367–631) and 749 nmol L−1 (IQR 648–631), respectively (P < 0.0001). There was no difference in T3 or fT4 levels with lesion size. However, TSH levels were lower in patients with large lesions compared to those with small lesion [0.62 mIU L−1 (IQR 0.3–1.1) vs. 1.33 mIU L−1 (IQR 1.1–1.6), P < 0.01], and GH levels increased with lesion size [0.21 ng mL−1 (IQR 0.10–0.94) vs. 0.57 ng mL−1 (IQR 0.16–1.21) vs. 1.65 ng mL−1 (IQR 0.84–2.40), overall P = 0.003].

Clinical stroke syndrome.  Cortisol values were significantly higher in patients with TACS 654 nmol L−1 (IQR 495–839) compared with patients with PACS 472 nmol L−1 (IQR 328–624, P < 0.001), LACS 450 nmol L−1 (IQR 334–587, P < 0.001) or POCS 469 nmol L−1 (IQR 351–613, P < 0.01). Thyroid hormone levels did not vary in patients with different clinical stroke syndromes. GH levels were significantly higher in patients with TACS [0.59 ng mL−1 (IQR 0.36–1.06)] compared to those with LACS [0.31 ng mL−1 (IQR 0.11–0.84), P < 0.05] and tended to be higher compared to those with PACS or POCS [0.35 ng mL−1 (IQR 0.15–1.03) and 0.43 ng mL−1 (IQR 0.14–1.22), respectively, P = 0.05].

Pituitary axis hormones and outcome

Pituitary axis hormones and 90-day outcome.  A total of 172 (61%) patients had a good functional outcome defined as an mRS ≤ 2, whereas 109 patients (39%) had a bad functional outcome defined by an mRS ≥ 3.Thirty patients died within 90 days, and thus the mortality rate was 10.7%. The comparisons of median hormone values in patients with a good/bad functional outcome and in survivors/nonsurvivors are summarized in Table 2.

Table 2.   Pituitary axis hormone levels and outcome after 90 days and 1 year
 Functional outcomeMortality
mRS ≤ 2mRS > 2PSurvivorsNonsurvivorsP
  1. fT4, free thyroxine; GH, growth hormone; IQR, interquartile range; mRS, modified Rankin Scale; T3, triiodothyronine; TSH, thyroid-stimulating hormone.

Pituitary axis hormones and 90-day outcome
 Cortisol (mmol L−1) median (IQR)444 (318.5–558.5)582 (439.5–727)<0.0001466 (337–598)712 (577–1106)<0.0001
 T3 (nmol L−1) median (IQR)1.5 (1.3–1.6)1.3 (1.1–1.6)0.0051.5 (1.3–1.6)1.2 (1.0–1.4)0.002
 fT4 (pmol L−1)  median (IQR)15.6 (14.2–17.8)15.3 (13.7–16.6)0.0315.4 (13.8–16.9)16.3 (14.3–19.6)0.03
 T3/fT4 ratio0.1 (0.08–0.1)0.08 (0.07–0.1)0.00010.09 (0.08–0.11)0.07 (0.06–0.08)0.0001
 TSH (mIU L−1) median (IQR)1.6 (1.1–2.4)1.1 (0.7–2.0)0.00031.4 (0.9–2.3)1.0 (0.6–2.2)0.02
 GH (ng mL−1) median (IQR)0.3 (0.1–1)0.5 (0.2–1.2)0.010.3 (0.2–1.0)0.6 (0.4–1.4)0.02
Pituitary axis hormones and 1-year outcome
 Cortisol (mmol L−1) median (IQR)466 (333–585)512 (377–692)0.004466 (337–590)633 (455–853)<0.0001
 T3 (nmol L−1) median (IQR)1.5 (1.3–1.7)1.3 (1.2–1.7)0.00081.5 (1.3–1.6)1.3 (1.1–1.6)0.06
 fT4 (pmol L−1) median (IQR)15.4 (13.8–16.8)15.6 (14.1–17.8)0.1115.4 (13.8–16.7)17.1(14.3–19.1)0.004
 T3/fT4 ratio0.1 (0.08–0.11)0.08 (0.07–0.1)0.00020.09 (0.08–0.11)0.08 (0.06–0.09)0.001
 TSH (mIU L−1) median (IQR)1.4 (1.0–2.2)1.2 (0.7–2.1)0.041.4 (0.9–2.2)1.1 (0.6–2.2)0.1
 GH (ng mL−1) median (IQR)0.3 (0.1–1.1)0.5 (0.2–1.1)0.060.3 (0.1–1.0)0.6 (0.4–1.5)0.002

Cortisol.  Median cortisol levels in patients with a favourable outcome and in survivors were lower compared to levels in patients with an unfavourable outcome and nonsurvivors, respectively.

Time to death was analysed by Kaplan–Meier survival curves based on cortisol quartiles. Patients in the highest two quartiles (cortisol ≥ 633 nmol L−1) had an increased risk of death compared with patients in the lowest two quartiles (P < 0.001) (Fig. 2).

image

Figure 2.  Kaplan–Meier survival based on cortisol quartiles. Time to death was analysed by Kaplan–Meier curves based on cortisol quartiles. Patients in the lower two quartiles (cortisol <345 nmol L−1 and cortisol between 345 and 480 nmol L−1) had a minor risk of death compared to patients with cortisol levels in the upper two quartile (cortisol between 481 and 632 and ≥633 nmol L−1, P < 0.001).

Download figure to PowerPoint

T3, fT4 and TSH.  Triiodothyronine levels in stroke patients with a favourable outcome were higher compared to levels in patients with an unfavourable outcome; the reverse was found for fT4 levels. T3/fT4 ratios were significantly higher in patients with a favourable outcome compared to ratios in patients with an unfavourable outcome. TSH levels were higher in patients with a favourable outcome compared to levels in patients with an unfavourable outcome.

Patients who died had lower T3/TSH levels and higher fT4 levels compared to levels in patients who survived. T3/fT4 ratios were significantly higher in patients who survived compared to ratios in those who died.

Time to death was analysed by Kaplan–Meier survival curves based on fT4, T3 and TSH quartiles. Patients in the highest quartile of fT4 (P = 0.02) and the lowest quartile of T3 (P = 0.002) and TSH (P = 0.01) had a significantly higher risk of death compared to patients in the other three quartiles.

GH.  Growth hormone levels in patients with a favourable outcome and survivors were lower compared to levels in patients with an unfavourable outcome and nonsurvivors, respectively.

Time to death was analysed by Kaplan–Meier survival curves based on GH quartiles. Patients in the highest two quartiles had a significantly increased risk of death compared to patients in the lower two quartiles (P = 0.001).

Pituitary axis hormones and outcome at 1 year.  Of the original 281 stroke patients, 268 completed the 1-year follow-up and were available for long-term analysis. The comparisons of median hormone values in patients with a good/bad functional outcome and in survivors/nonsurvivors are summarized in Table 2.

Comparison of the pituitary axis hormones with other markers and clinical scores.  We performed ROC curve analysis to compare the overall prognostic accuracy of the NIHSS, cortisol, T3, fT4, TSH and GH. The prognostic accuracy of cortisol was high compared to that of glucose and white blood cell count, but was similar to that of CRP and the CCI. The NIHSS was the most accurate predictor of 1-year outcome (Table 3).

Table 3.   Receiver operating characteristics curve analysis
ParameterFunctional outcome at 90 daysFunctional outcome at 1 year
AUC95% confidence intervalPAUC95% confidence intervalP
  1. AUC, area under the curve; CCI, Charlson comorbidity index; CRP, C-reactive protein; fT4, free thyroxine; GH, growth hormone; NIHSS, National Institutes of Health Stroke Scale; T3, triiodothyronine; TSH, thyroid-stimulating hormone; WBC, white blood count.

Cortisol0.68(0.63–0.74) 0.60(0.53–0.66) 
NIHSS0.75(0.69–0.8)0.080.72(0.66–0.78)0.004
T30.56(0.50–0.63)0.060.61(0.55–0.67)0.71
fT40.58(0.52–0.64)0.030.55(0.49–0.62)0.38
TSH0.60(0.55–0.67)0.290.55(0.49–0.61)0.35
GH0.61(0.55–0.67)0.120.57(0.50–0.63)0.56
Glucose0.55(0.49–0.61)0.010.50(0.44–0.58)0.05
WBC0.56(0.50–0.62)0.020.53(0.49–0.60)0.18
CCI0.62(0.56–0.68)0.260.64(0.57–0.71)0.45
CRP0.60(0.54–0.67)0.160.59(0.52–0.66)0.84
 Mortality at 90 daysMortality at 1 year
Cortisol0.81(0.76–0.86) 0.69(0.63–0.75) 
NIHSS0.85(0.80–0.89)0.440.78(0.72–0.83)0.11
T30.66(0.60–0.70)0.020.59(0.52–0.65)0.10
fT40.60(0.54–0.76)0.0030.63(0.57–0.68)0.29
TSH0.60(0.54–0.67)0.0010.56(0.50–0.62)0.03
GH0.64(0.58–0.70)0.040.65(0.58–0.71)0.50
Glucose0.59(0.53–0.66)0.0020.55(0.48–0.62)0.0.2
WBC0.66(0.53–0.67)0.0040.53(0.47–0.61)0.01
CCI0.59(0.53–0.65)0.0070.59(0.52–0.66)0.08
CRP0.69(0.63–0.75)0.130.64(0.57–0.71)0.27

The AUC to predict mortality was highest for cortisol [0.81 (IQR 0.76–0.86)] and was similar to that of the NIHSS [0.85 (IQR 0.8–0.89), P = 0.44]. Cortisol had a higher prognostic accuracy than the other hormones and laboratory and clinical parameters and tended to have a higher AUC value compared to CRP (Table 3). Again, the NIHSS was the most accurate predictor of mortality at 1 year (Table 3).

When combining the NIHSS with initial cortisol in a logistic regression model for 90-day outcome, we found only a small increase in AUC from 0.75 to 0.77 for functional outcome prediction, which did not reach statistical significance (P = 0.52). Similarly, for mortality prediction, the AUC increased from 0.85 to 0.87 (P = 0.30).

We further calculated in-sample reclassification tables (Appendix 1 and 2). In patients with poor outcome, 14 were classified in higher-risk categories and 10 in lower categories when using the model with the NIHSS and cortisol. Similarly, in patients with good outcome, only 14 were classified in higher-risk categories and 32 in lower categories. Thus, the estimated net reclassification improvement for functional outcome was 0.14 (P < 0.01). Amongst nonsurvivors, seven patients were classified in higher-risk categories and three in lower categories; amongst survivors, 99 patients were classified in lower-risk categories and 50 in higher categories when using the model with the NIHSS score and cortisol (net reclassification improvement 0.33, P < 0.001).

Association between hormones and both functional outcome and mortality in logistic regression analysis

Univariate logistic regression models showed that cortisol, T3 and TSH were associated with functional outcome. Cortisol, fT4 and T3 were also significantly associated with death (Table 2). In a logistic model adjusted for the NIHSS and age, cortisol, but not the other pituitary axis hormones, was independently associated with both functional outcome [OR 1.0 (1.00–1.01), P < 0.0002] and death [OR 1.62 (1.37–1.92), P < 0.0002] (Table 4).

Table 4.   Univariate and multivariate association between hormone levels and outcome
ParameterUnivariate analysisMultivariate analysis
Odds ratioP > z95% confidence intervalOdds ratioP > z95% confidence interval
  1. Multivariate analysis was calculated for all significant predictors in univariate analysis, adjusted for age and the National Institutes of Health Stroke Scale.

  2. fT4, free thyroxine; GH, growth hormone; T3, triiodothyronine; TSH, thyroid-stimulating hormone.

Predictor: functional outcome
 Cortisol (per 100 nmol increase)1.00<0.0002(1.00–1.00)1.23<0.01(1.07–1.43)
 fT41.070.08(0.99–1.14)   
 T30.320.01(0.15–0.72)0.840.09(0.68–1.02)
 TSH0.780.01(0.63–0.95)0.790.60(0.33–1.90)
 GH0.990.81(0.90–1.08)   
Predictor: death
 Cortisol (per 100 nmol increase)1.62<0.0002(1.37–1.92)1.43<0.001(1.17–1.75)
 fT41.100.02(1.01–1.20)1.080.13(0.98–1.19)
 T30.190.01(0.05–0.71)0.690.66(0.13–3.60)
 TSH0.920.50(0.72–1.17)   
 GH1.040.41(0.95–1.14)   

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Conflict of interest statement
  9. References
  10. Appendices

In this study, we simultaneously assessed anterior pituitary axis hormones with regard to their accuracy to predict functional outcome and mortality in patients with acute ischaemic stroke within 90 days and 1 year. Our main finding is that cortisol is an independent prognostic marker of functional outcome and death in patients with ischaemic stroke, but adds no significant additional predictive information to the clinical score of the NIHSS. We demonstrated that cortisol levels increased with lesion size, neurological deficit (assessed by the NIHSS) and the clinical stroke syndrome (i.e. TACS versus PACS, LACS and POCS), reflecting the severity of the stroke. Conversely, T3, fT4, TSH and GH add only limited or no prognostic information to currently used measures and scores.

In previous studies, we found that copeptin is a significant predictor of short- and long-term outcome and mortality [8, 9]. The prognostic performance of cortisol within the same population was similar to that of copeptin, but showed no significant additional predictive value to the NIHSS in contrast to copeptin [8].

Acute ischaemic stroke acts as a stressor and thus stimulates the HPA axis resulting in increased glucocorticoid levels [5, 6, 23]. The higher cortisol levels observed in patients with worse functional outcome or subsequent death reflect a higher degree of stress. These results are in accordance with the results from other studies showing that serum cortisol levels rise proportionally with the degree of stress and correlate with stroke severity [5, 24–26]. A severe stroke per se implies a poor outcome. However, there are several other mechanisms that might explain the unfavourable outcome in patients with higher cortisol levels. Hypercortisolism has been suggested to potentiate ischaemic neuronal injury, especially in hippocampal neurons [27], and the corticosterone synthesis inhibitor metyrapone was able to prevent ischaemia-induced loss of synaptic function in the hippocampus of rats [28]. The hippocampus has an important role in the feedback regulation of the HPA axis. A disturbed hippocampus function might result in a false HPA axis feedback that potentiates hypercortisolism and causes a vicious circle, explaining the worse prognosis in stroke patients with high cortisol levels [29]. In addition, patients with stroke and high cortisol levels have been shown to be more prone to adverse cardiac events (e.g. arrhythmias or myofibrillar degeneration), which might lead to higher mortality rates [25, 30]. Another major cause of a bad prognosis after stroke is the development of infectious disease which is related to an immune dysregulation resulting from neuroendocrine disturbance after stroke.

In our study, T3 levels were lower in patients with a poorer prognosis concerning functional outcome and death. This is in line with a previously published study which showed that the low T3 syndrome was an independent predictor of survival in patients with acute stroke, and predicted disability at 1 year [7]. It should be emphasized that low T3 levels even within the normal range are already associated with poorer prognosis in acute stroke patients [7]; the greater the severity of disease, the lower the serum T3 levels. A decrease in the peripheral production of T3 because of decreased extra-thyroidal conversion of T4 into T3 by the enzyme type I iodothyronine-5′-deiodinase, as in the low T3 syndrome, is a major contributing factor [31]. This is reflected by the lower T3/fT4 ratios in patients with an unfavourable outcome in our cohort. In addition, it has been shown that high levels of corticosteroids suppress TSH secretion and the pituitary response to thyrotropin-releasing hormone in man [32], leading, for example, to low T3 levels. Furthermore, stress-induced elevation of glucocorticoids in rats causes suppression of TSH and T3 levels [33]. Thus, the higher glucocorticoid levels in patients with worse outcome measured in our cohort might result in a higher decrease in T3 levels.

Growth hormone levels in our study were higher in patients who died than in those who survived. Several mechanisms might explain the poorer prognosis of patients with higher GH levels. First, GH levels increase during stress and thus mirror the stress associated with the severity and extent of illness [4]. Second, the higher levels of GH in nonsurvivors might be an attempt by the body to provide energy and postpone anabolism. Third, the higher GH levels in our study observed in nonsurvivors might be related to lower insulin-like growth factor (IGF) levels and reflect hepatic GH resistance. In this context, it has been shown that patients with lower IGF levels suffering from an ischaemic or haemorrhagic stroke are at higher risk of death compared to patients with higher IGF levels [34, 35]. IGF is a potent neurotrophic factor. Its expression is induced in injured brain regions, and its administration reduces the extent of cortical infarction and neuronal death from ischaemic injury in animal models [36].

Limitations

Some limitations of this observational study merit consideration. First, cortisol improved the classification of patients for functional outcome and for death in net reclassification statistics as evidenced by significant net reclassification improvements [21]. However, the lack of significance in the combined ROC curve analysis suggests that these findings need to be interpreted with caution. For this reason, we did not further determine cut-off values for cortisol for use in clinical practice.

Second, we only performed single measurements of hormone levels in the morning after admission. GH and cortisol levels show a large variation during the day because in terms of GH, secretion is pulsatile [37] with almost undetectable serum GH concentrations between the pulses, and in terms of cortisol, it follows a circadian rhythm [38]. A standardized measurement of GH with a functional test in combination with IGF measurement in all patients might have provided a higher accuracy to predict outcome. For cortisol, however, the confounding factors are minimized as the acute illness abolishes its diurnal variation and as we similarly measured the level in the morning after admission in all patients [39]. Also, standardized measurements do not represent clinical routine in the acute emergency room setting.

Third, we analysed patients within 72 h of symptom onset. When comparing initial results with results from hormone levels of patients the next morning after admission and within either 12–24 or 24–72 h after symptom onset, they did not show significant differences. We also tested for effect modification by delay to blood sampling for cortisol measurement and did not find evidence for interaction (between cortisol levels and time delay) with regard to prediction of death and functional outcome (data not shown).

Conclusion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Conflict of interest statement
  9. References
  10. Appendices

In conclusion, in this study, we simultaneously assessed the prognostic accuracy of pituitary axis hormones (i.e. cortisol, T3, fT4 and GH) in a large cohort of patients with acute ischaemic stroke. We found that cortisol levels and, to a lesser extent, T3, fT4 and GH levels mirror stroke severity. Cortisol levels are independently associated with an unfavourable outcome after acute ischaemic stroke.

Conflict of interest statement

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Conflict of interest statement
  9. References
  10. Appendices

AE was an employee of SphingoTec GmbH, the developer of the GH assay. No funding was obtained from commercial sources for this study.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Conflict of interest statement
  9. References
  10. Appendices

Appendices

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Conflict of interest statement
  9. References
  10. Appendices
Table Appendix 1.   Reclassification table for functional outcome
Model with NIHSSModel with NIHSS and cortisol
<3% risk3% to <10% risk10% to <18% risk18% to <35% risk35% to <40% risk40% to 48% risk>48% riskTotal no.
Patients with good functional outcome
 <3% risk00000000
 3% to <10 % risk00000000
 10% to <18% risk0113100015
 18% to <35% risk032284800117
 35% to <40% risk00006309
 40% to 48% risk000035210
 >48% risk0000031720
 Total no.043585171119171
Patients with bad functional outcome
 <3% risk00000000
 3% to <10 % risk00000300
 10% to <18% risk00300003
 18% to <35% risk0012923035
 35% to <40% risk00021205
 40% to 48% risk000235414
 >48% risk0010105052
 Total no.0053371054109
Table Appendix 2.   Reclassification table for mortality
Model with NIHSSModel with NIHSS and cortisol
<1% risk1% to <2% risk2% to <4% risk4% to <9% risk9% to <18% risk18% to 34% risk>34% riskTotal no.
Survivors
 <1% risk00000000
 1% to <2% risk00000000
 2% to <4% risk24334428010130
 4% to <9% risk271431132069
 9% to <18% risk0115103020
 18% to 34% risk1011211319
 >34% risk001222512
 Total no.2741616727198250
Nonsurvivors
 <1% risk00000000
 1% to <2% risk00000000
 2% to <4% risk00110002
 4% to <9% risk00231107
 9% to <18% risk00003205
 18% to 34% risk00000224
 >34% risk0000011112
 Total no.0034461330