The Endocrinologic Changes in Critically Ill Chronic Obstructive Pulmonary Disease Patients

ABSTRACT

Background: Alterations in the neuroendocrine system occur during critical illness. Chronic obstructive pulmonary disease (COPD) itself causes hormonal changes. The aim of this study was to determine neu roendocrine hormones of COPD patients with acute respiratory failure and to investigate the relationship between hormonal changes, mortality, and morbidity.Methods: We enrolled 21 patients (13 F/8 M) with COPD exacerbation requiring artificial airway support. Blood samples were collected on admission to the ICU, and on the day of hospital discharge. Eighteen healthy people were included as controls. Results: Female patients had lower luteinizing hormone (LH), follicle stimulating hormone (FSH), and free triiodothyronine (fT3), and higher prolactin (PRL) levels than controls on admission to the ICU (FSH: 70.3 vs. 29.3 mlU/mL; LH: 26.6 vs. 6.8 mlU/mL; fT3: 2.9 vs. 2.0 pg/mL; PRL: 12.4 vs. 21.3 ng/mL). Male patients had low testosterone and TSH and high PRL but only changes in TSH and PRL reached statistical significance (testosterone: 3.5 vs. 1.5 ng/mL, TSH: 1.1 vs. 0.5 ulU/mL, PRL: 9.7 vs. 14.2 ng/mL). Female patients had lower fT3 than males (fT3_female: 2.7 vs. fT3_male: 2.0 pg/mL). On follow-up, significantly elevated FSH and fT3 and decreased estradiol concentrations were documented among recovered women (FSH: 28.4 vs. 46.6 mlU/mL, fT3,: 2.0 vs. 2.6 pg/mL, E₂: 27.7 vs. 19.0 pg/mL). Patients had high C-reactive protein levels and acute physiologic and chronic health evaluation II scores. Mortality rate was 9.5% and a negative correlation between E₂ and duration of noninvasive mechanical ventilation and length of hospital stay was found in male patients. Conclusion: Men and women with acute respiratory failure in the presence of COPD develop significant changes in the neuroendocrine axis. Hormonal suppression vanishes with disease improvement.

Keywords: :

• Exacerbation of COPD
• Neuroendocrine changes
• Acute respiratory failure
• Artificial airway support
• APACHE II score

INTRODUCTION

Chronic obstructive pulmonary disease (COPD) is a progressive disease characterized by nearly irreversible airflow limitation. Significant chronic inflammation due to harmful particles and gases, especially cigarette smoke, takes the leading role in the pathogenesis of COPD (1). The disease is associated with significant systemic alterations in organ function, such as muscle loss, cardiac disease, and neuroendocrine system changes (2, 3).

There are several studies examining the endocrine system of COPD patients, generally in stable cases (4–8). The effects on thyroid hormones are variable. Some studies report normal serum thyroid hormones in stable and oxygen-dependent COPD patients (9, 10) while others show low triiodothyronine (T3), free triiodothyronine (fT3), and triiodothyronine/thyroxine ratio (T3/T4) in both stable and exacerbated COPD patients (11). Another study demonstrated a positive correlation between PO₂ and T3/T4 in COPD patients with decreased FEV₁ (forced expiratory volume in 1 second) although they had normal serum thyroid hormones (12).

Gonadal axis is the most commonly examined system in COPD patients, predominantly in male patients. The prevalence of low testosterone among male COPD patients was between 22% and 62% (4, 5, 8, 13), and two types of hypogonadism were encountered: hypogonadotrophic hypogonadism (10, 13, 14) and hypergonadotrophic hypogonadism (8, 14, 15). However, there are also some studies that report the low prevalence of hypogonadism among male COPD patients (6, 15). In these studies, prolactin (PRL) was reported normal (7, 15), whereas estradiol (E₂) was measured normal in one study (4) and high in another study (16).

The aim of our study was first to determine changes in the neuroendocrine axis in COPD patients with acute respiratory failure (ARF), who were admitted to the intensive care unit (ICU). The second goal was to investigate the relationship between hormonal changes and the illness scoring system (APACHE II), mechanical ventilation, length of ICU, and hospital stay and mortality.

MATERIALS AND METHODS

Between March 2005 and December 2006, 22 consecutive COPD patients with ARF were enrolled in the study. Diagnosis of COPD was made by using ATS criteria (1). Patients with acute exacerbation of COPD admitted to the ICU were considered for the study. Diagnostic criteria for ARF were respiratory acidosis (pH < 7.35, PaCO₂ > 45 mmHg) and/or PaO₂/FIO₂ < 200 and/or evidence of respiratory distress (respiratory rate ≥ 30/min, accessory respiratory muscle use, paradoxical breathing, or intercostal muscle retraction).

Exclusion criteria were age less than 18 years; preexisting neurological, psychiatric, metabolic or endocrine diseases (except diabetes mellitus); intracranial lesions, hemorrhage or infarction (except lacunar infarct); malignancy (except basal cell cancer); body mass index (BMI) <18.5; renal failure requiring renal replacement therapy; chronic and acute liver disease; autoimmune diseases; HIV positive patients; electrolyte disturbances (Na, K, and Ca); resuscitated patients; heavy alcohol drinkers (≥80 gr/day); premenopausal women; and concomitant treatment with calcium channel blockers, amiodarone, dopamine agonists or antagonists, benzodiazepines, opioids, immunusuppressive drugs, antipsychotic agents, antidepressants, antiepileptics, thyroid hormones, estrogens, and glucocorticoids.

Severity of illness was rated by Acute Physiologic and Chronic Health Evaluation II scoring (APACHE II) calculated 24 hours after admission to the ICU. Spirometric classification of disease severity (mild, moderate, and severe) was used according to ATS criteria (1). If patients did not have a previous pulmonary function test, they underwent pulmonary function testing after clinical stabilization just before hospital discharge. Six patients could not perform the spirometry test. Administration of drugs suspected of affecting the neuroendocrine axis during hospital stay and types of mechanical ventilation were recorded on patient files. Effects of drugs on hormones were analyzed only when at least 5% of patients received a given drug. All patients were followed until hospital discharge. Twenty healthy subjects were included as controls.

Blood samples were collected within the first 4 hours of the ICU admission and on the day of hospital discharge. Since a short glucocorticoid therapy protocol is generally ordered for COPD exacerbation in the ICU (17), control samples were taken after discontinuing steroid treatment at least 5 days. Steroid treatment was prolonged in the case of unresolved broncospasm. If steroid treatment was stopped after hospital discharge, control blood samples were drawn during the first outpatient visit. After centrifugation, sera were kept frozen at −80°C until assayed. Since all patients were taking O₂ when admitted to the ICU, the PO₂/FIO₂ ratio was preferred to PO₂ as a study parameter.

ASSAYS

All hormonal samples from each patient were processed in the same run. The serum concentration of luteinizing hormone (LH), follicle stimulating hormone (FSH), testosterone, thyroid stimulating hormone (TSH), fT3, and free thyroxine (fT4) were measured by the electrochemiluminescence immunoassay “ECLIA” method (Modular Analytics E170, Roche, America. Normal limits: LH: 7.7–58.5 mIU/mL, FSH: 19.3–100.6 mIU/mL for postmenopausal women; LH: 1.7–8.6 mIU/mL, FSH: 1.42–15.4 mIU/mL for men; testosterone: 2.8–8.0 ng/mL; TSH: 0.27–4.2 μIU/mL; fT3: 2.6–4.4 pg/mL; fT4: 0.93–1.7 ng/dL). PRL was measured by the ECLIA method (Elecsys 2010, Roche, America. Normal limits: PRL: 4.79–23.3 ng/mL for women, 4.04–15.2 for men). E₂ was measured by the chemiluminescent enzyme immunoassay (Immulite 2000, Roche, America. Normal limits: E₂: 0–56 pg/mL for postmenopausal women, 0–30 pg/m for men). The intra-assay coefficient variations for all these methods were between 0.6% and 9.9%.

C-reactive protein (CRP) was measured by the particle-enhanced immunoturbidimetric assay (Hitachi 917, Roche, America) and a CRP level > 5 mg/dL was accepted as elevated.

ETHICAL ASPECTS

Informed consent was obtained from the patients and controls. If the patient was unconscious or too ill to communicate, consent was obtained from a first-degree relative. The study protocol was approved by the ethics review board of Marmara University, School of Medicine (Istanbul, Turkey, 2005; MAR-YC-2005-0092).

STATISTICAL ANALYSIS

Since data did not meet parametric test conditions, nonparametric tests were utilized. Study and control groups were compared using the Mann–Whitney U test for continuous variables. The Wilcoxon test was used to compare baseline hormone values with levels after recovery. Correlations between variables were evaluated using the Spearman test. The Mann–Whitney U test was applied for the impact of drugs prescribed during hospital stay. Results were expressed as median with interquartile range unless indicated otherwise. p<.05 was considered significant.

RESULTS

Because one male patient was excluded from the study due to hyperthyroidism diagnosed after the test results, 21 patients (13 F/8 M) were included in the study. The patients and controls were matched by age and BMI (Table 1).

Table 1.  Demographics of the study groups

Sex	Controls	Patients	p
Women (n)	13	13
Age (yr)	72.0 (63.0–83.5)	72.0 (64.5–78.0)	.644
BMI (kg/m^2)	27.4 (24.7–29.8)	25.39 (22.0–27.4)	.130
Men (n)	7	8
Age (yr)	74.0 (52.0–85.0)	70.0 (66.3–72.0)	.562
BMI (kg/m^2)	26.2 (20.3–27.5)	25.1 (22.7–30.0)	.728
Total	20	21

Note: Values are expressed as median and interquartile ranges.

Table 2.  Arterial blood gases, CRP, and APACHE II score of patients

Parameters	Women (n = 13)	Men (n = 8)	p
pH	7.29 (7.24–7.31)	7.26 (7.15–7.32)	.744
PCO2 (mmHg)	73.9 (71.4–94.9)	79.3 (57.5–98.9)	.828
HCO3 (mEq/L)	36.0 (33.0–44.2)	31.7 (29.4–41.8)	.128
PO2/FIO2	197.6 (124.9–253.9)	253.3 (202.5–304.9)	.148
APACHE II	23.0 (17.0–25.0)	19.0 (16.0–24.8)	.558
CRP (mg/dL)	28.5 (12.5–96.4)	45.8 (26.5–103.3)	.396

Note: Values are expressed as median and interquartile ranges.

The cause of COPD in all patients was tobacco exposure. Four patients (3 F/1 M) had moderate and 11 patients (6 F/5 M) had severe COPD. Fourteen patients (9 F/5 M) were O₂ dependent and four of them were also using noninvasive mechanical ventilation (NIMV) at home. The reasons of COPD exacerbation were pneumonia (n = 13), pneumonia and heart failure (n = 3), heart failure (n = 2), treatment noncompliance (n = 2), and heart failure and urinary tract infection (n = 1). The median APACHE II score was 22.0 (16.0–25.0) and CRP was 36.6 (17.2–96.4) mg/dL. Male and female patients had comparable arterial blood gases, CRP levels, and APACHE II scores (Table 2).

Female patients had significantly lower gonadotropins and fT3 and higher PRL concentrations than control females (Table 3). When female patients were reviewed individually, LH levels were low in nine patients and within the physiologic range in four patients; FSH levels were low in five patients and normal in eight patients. Abnormally low values for both LH ≤5.0 mIU/mL and FSH ≤10.0 mIU/mL were found in seven and four females, respectively. Severely depressed values with LH ≤1.0 mIU/mL and FSH ≤5.0 mIU/mL were found in three and two women, respectively.

Table 3.  Hormone levels of female patients on admission to the ICU

Women	Patients (n = 13)	Controls (n = 13)	p
LH (mlU/mL)	6.8 (2.5–12.5)	26.6 (21.0–33.3)	.001
FSH (mlU/mL)	29.3 (9.2–36.0)	70.3 (62.4–88.1)	.000
E2 (pg/mL)	31.0 (21.3–39.0)	27.0 (20.0–31.0)	.589
TSH (μlU/mL)	0.9 (0.3–2.1)	1.8 (1.3–3.0)	.096
fT3 (pg/mL)	2.0 (1.7–2.4)	2.9 (2.5–3.2)	.000
fT4 (ng/dL)	1.2 (1.0–1.4)	1.2 (1.1–1.3)	.939
PRL (ng/mL)	21.3 (14.9–41.4)	12.4 (9.2–19.9)	.043
CRP (mg/dL)	28.5 (12.5–96.4)	0.8 (0.8–0.8)	.000

Note: Values are expressed as median and interquartile ranges.

*p <.05, regarding significant differences between groups.

Male patients had similar gonadotropin concentrations as their controls but higher PRL and lower TSH levels (Table 4). The median serum testosterone tended to be lower but was statistically insignificant. When men were examined individually, testosterone concentrations of five patients (62.5%) and two controls (28.8%) were below the lower limit of the physiologic range. Of the patients with low testosterone, two had high LH and FSH levels compatible with hypergonadotrophic hypogonadism and three had normal LH and FSH levels compatible with hypogonadotrophic hypogonadism. Of the patients with normal testosterone, two had normal gonadotropins and one had raised LH but normal FSH levels.

Table 4.  Hormone levels of male patients on admission to the ICU

Men	Patients (n = 8)	Controls (n = 7)	p
LH (mlU/mL)	8.3 (7.2–13.3)	5.2 (4.4–8.7)	.298
FSH (mlU/mL)	9.6 (5.3–17.2)	4.7 (4.2–22.3)	.908
Testosterone (ng/mL)	1.5 (1.3–3.7)	3.5 (2.3–5.4)	.064
E2 (pg/mL)	28.1 (19.0–38.3)	36.4 (27.3–47.7)	.141
TSH (μlU/mL)	0.5 (0.4–1.1)	1.1 (1.0–1.7)	.028
fT3 (pg/mL)	2.7 (2.3–2.8)	3.1 (2.6–3.9)	.083
fT4 (ng/dL)	1.3 (1.1–1.3)	1.2 (1.1–1.3)	.728
PRL (ng/mL)	14.2 (11.3–53.3)	9.7 (7.6–11.6)	.028
CRP (mg/dL)	45.8 (26.5–103.3)	2.7 (0.8–4.5)	.001

Note: Values are expressed as median and interquartile ranges.

*p <.05 denotes significant differences between groups.

A correlation analysis showed that LH was negatively related to HCO₃ (r = −.753, p = −.000) and E₂ was inversely correlated with PO₂/FIO₂ (r = −.606, p = .028) among the females. fT4 was inversely related to PO₂/FIO₂ (r = −.958, p = .000) and PRL was positively associated with PCO₂ (r = .905, p = .002) but negatively correlated with pH (r = −.714, p = .047) among the men. There was no correlation between the PO₂/FIO₂ ratio and testosterone (r = −.119, p = .779). Hormonal values were similar between home oxygen users and nonusers.

Methylprednisolone treatment was given to 19 patients [median dose: 540 (400–680) mg, therapy days: 7 (4–15)]. Follow-up blood tests of 10 patients (8 F/2 M) were sampled on the day of hospital discharge (21.5 (11–40) days after the first samples). Control samples of three (2 F/1 M) patients using steroids until hospital discharge were drawn at their first outpatient visits, on the 43rd (F: normal LH and FSH), 50th (F: normal LH and FSH), and 53rd days (M: normal testosterone, increased LH, and normal FSH) of the study. After an analysis of follow-up samples, important increases of FSH and fT3 and a decrease of E₂ were documented in females who recovered (Table 5). LH and FSH rose to normal limits in seven and eight women, respectively. Two women had severely depressed LH and FSH on discharge. The median CRP level of the recovered females tended to decline but statistically insignificant. When CRP levels of all discharged patients were compared with their baselines, a significant decrease in CRP was noted (CRP_admission: 34.5 mg/dL, CRP_follow-up: 20.1 mg/dL, p = .013). Since blood samples of only three men were studied on follow-up, statistical analysis could not be done. (On admission: LH: 8.1 mlU/mL, FSH: 7.7 mlU/mL, testosterone: 1.6 mlU/mL, E₂: 28.7 pg/mL, TSH: 0.5 μlU/mL, fT3: 2.7 pg/mL, fT4: 1.1 ng/dL, PRL: 22.9 ng/mL. On follow-up: LH: 19.7 mlU/mL, FSH: 14.4 mlU/mL, testosterone: 2.9 mlU/mL, E₂: 27.2 pg/mL, TSH: 0.7 μlU/mL, fT3: 2.2 pg/mL, fT4: 1.1 ng/dL, PRL: 15.6 ng/mL.) Of the three men, one had normal testosterone with normal gonadotropins on admission and discharge. The second case had low testosterone but elevated gonadotropins. Testosterone of the third patient increased to a normal level with normal gonadotropins during his outpatient visit.

Table 5.  Hormone levels of women on admission to the ICU and after recovery

Women	Admission (n = 10)	Follow-up (n = 10)	p
LH (mlU/mL)	8.1 (3.1–15.3)	19.7 (5.4–25.8)	.333
FSH (mlU/mL)	28.4 (9.6–41.1)	46.6 (26.1–61.6)	.017
E2 (pg/mL)	27.7 (22.4–34.1)	19.0 (19.0–21.8)	.050
TSH (μlU/mL)	1.2 (0.3–2.3)	1.3 (0.8–2.3)	.799
fT3 (pg/mL)	2.0 (1.7–2.3)	2.6 (2.5–2.9)	.037
fT4 (ng/dL)	1.2 (1.1–1.4)	1.3 (1.1–1.4)	.878
PRL (ng/mL)	25.4 (14.9–41.4)	21.5 (9.2–70.0)	.515
CRP (mg/dL)	28.4 (12.5–151.9)	17.9 (7.1–23.7)	.069

Note: Values are expressed as median and interquartile ranges.

*p <.05 denotes significant differences between subsequent results.

When thyroid hormones and PRL of 21 patients were compared with controls, TSH and fT3 were found lower and PRL higher (Table 6). Although patients had lower TSH, none had a TSH value below the physiologic range. A negative correlation between CRP and TSH was demonstrated (r = −.695, p = .001). A mild but insignificant fT3 rise among recovered 13 patients was documented while other hormones stayed unchanged (fT3_admission: 2.1 (1.2–2.7) pg/mL, fT3_follow-up: 2.5 (2.3–2.7) pg/mL, p = .087).

Table 6.  Levels of thyroid hormones and PRL of all patients on admission to the ICU and their comparison with the control group

Thyroid hormones and PRL	Patients (n = 21)	Controls (n = 20)	p
TSH (μlU/mL)	0.7 (0.3–1.7)	1.4 (1.1–2.7)	.010
fT3 (pg/mL)	2.2 (1.9–2.7)	2.9 (2.6–3.3)	.000
fT4 (ng/mL)	1.2 (1.1–1.4)	1.2 (1.1–1.3)	.886
PRL (ng/mL)	20.9 (13.4–41.4)	11.4 (8.9–12.8)	.005

Note: Values are expressed as median and interquartile ranges.

*p <.05 represents differences between the groups.

After documentation of lower TSH levels in male patients compared to controls, possible reasons for differences of thyroid hormones between genders were considered. By comparison, fT3 detection was lower in women than men (fT3_women: 2.0 (1.7–2.4) pg/mL, fT3_men: 2.7 (2.3–2.8) pg/mL; p = .010). When fT3 of each patient was checked, it was discovered that five out of eight male patients had normal serum fT3 while just one female patient out of thirteen had normal serum fT3 on admission to the ICU.

Neuroleptic, antidepressant, Ca channel blocker, and benzodiazepine medications were being used by seven patients (5F/2M), two patients (1F/1M), two women, and one woman, respectively, during hospital stay. The median PRL level was high in neuroleptic users on discharge (PRL_{neuroleptic-user}: 72.0 (27.0–136.6) ng/mL, PRL_nonuser: 9.3 (5.3–19.8) ng/mL; p = .016). A statistical evaluation could not be applied to other medications due to small numbers.

NIMV was applied to all patients in the ICU. Invasive mechanical ventilation (IMV) was applied to seven patients in whom NIMV had failed. The length of NIMV was 166 (128–224) hours, IMV120 (30–307) hours, ICU stay 9 (7–15) days, and hospital stay 16 (12–28) days. A correlation analysis showed that only male patients had a negative correlation between the E₂ level and NIMV duration and hospital stay, although male and female patients had similar E₂ levels on admission to the ICU (p = .780). We did not find any correlation between other hormones and the duration of NIMV or IMV, APACHE II score, hospital stay, and ICU stay (Table 7). Two patients (1 F/1 M) died due to pneumonia. Nineteen patients were discharged. A prescription of NIMV was given to 11 patients (6 F/5 M) and an O₂ concentrator to 16 patients (9 F/7 M).

Table 7.  Correlation between hormones and morbidity and mortality

	NIMV		IMV		ICU stay		Hospital stay		Apache II
Hormones	r	p	r	p	r	p	r	p	r	p
Women (n = 13)
LH	−.511	.074	.600	.400	−.149	.627	−.253	.405	.031	.921
FSH	.044	.887	−.600	.400	.099	.747	.286	.344	.105	.732
E2	−.224	.462	.800	.200	.045	.885	−.310	.303	−.152	.620
Men (n: 8)
LH	.143	.736	.500	.667	−.084	.844	−.262	.531	.110	.795
FSH	.071	.867	.500	.667	.024	.955	−.190	.651	−.074	.862
E2	−.955	.001**	.500	.667	−.591	.162	−.883	.008**	.560	.191
Testosteron	−.240	.955	−.500	.667	−.611	.108	−.238	.570	−.356	.387
All patients (n: 21)
TSH	−.243	.289	−.536	.215	−.287	.207	−.245	.284	−.158	.494
fT3	−.003	.989	.357	.432	.294	.195	.069	.766	.049	.844
fT4	−.092	.691	.036	.939	−.047	.839	.183	.427	.400	.068
PRL	.157	.499	−.710	.879	.140	.546	.051	.825	.199	.388

*Number of female and male patients having IMV were four and three, respectively.

**p <.05 denotes significant correlations between hormones and morbidities.

DISCUSSION

In this study, we measured neuroendocrine hormones of COPD patients with ARF. Female patients had low FSH, LH, and fT3 and high PRL levels. Male patients had significantly low TSH and high PRL levels on admission to the ICU. Low testosterone levels were detected in 62.8% of male patients. Women had lower fT3 than men. On follow-up, FSH and fT3 of recovered female patients increased significantly. A negative correlation between E₂ and NIMV and hospital stay among males was found. CRP levels and APACHE II scores of patients were high, confirming the presence of severe disease and inflammation.

In this study, although the median testosterone level of male patients was not statistically different from that of healthy controls, 62.8% of male patients had low testosterone, compared to 28.8% of controls. Testosterone, one of the more commonly studied hormones in seriously ill patients, decreases in critical illness, as well as in many systemic diseases, and the more severe the disease, the worse is the suppression of testosterone (18–21). Low testosterone concentrations were as well measured among COPD patients, and the prevalence of hypotestosteronemia in this group of patients was reported between 22% and 62% (4, 5, 8, 13).

Two types of hypogonadism have been exhibited in COPD patients: hypergonadotropic hypogonadism (5, 6, 8, 13, 14) and hypogonadotropic hypogonadism (6, 10, 13, 14). In our study, two male patients with hypotestosteronemia out of five had both high serum LH and FSH levels, reminding testicular dysfunction and called hypergonadotrphic hypogonadism. It depicts an increased compensatory gonadotropins secretion from the hypothalamic–pituitary axis in order to correct the low circulating testosterone level (22). Suppressive effects of cytokines on testicular androgenesis are considered accountable for low testosterone (22, 23). In vitro studies illustrated inhibition of Leydig cell steroidogenesis by cytokines (TNFα, Il-6, and IL-1) (22, 24, 25). The study of TNF injection to voluntary men showed a marked decrease in serum testosterone levels in 2 hours (26). Although we did not measure cytokines, we revealed elevated CRP levels in male patients, which is a manifestation of inflammation and can explain low testosterone concentrations. The remaining three patients with low testosterone levels had normal gonadotropin levels regardless of low testosterone concentration, called secondary hypogonadism or hypogonadotropic hypogonadism. Normally, when serum testosterone level is decreased, a negative feedback of sex hormones on the hypothalamic–pituitary axis is lost and secretion of gonadotropins increases (22). In addition, injection of gonadotropin-releasing hormone (GnRH) to healthy subjects stimulates the anterior pituitary gonadotropes, ending up with high serum LH and FSH concentrations (27). These natural responses might be either absent or decreased in some COPD patients. After the injection of GnRH to COPD patients, an increase only in LH level was detected (10). All those imply hypofunctioning hypothalamic–pituitary axis in those patients. Once more, inflammation may take precedence in the inhibition of GnRH secretion as shown after the exposure of the brain to IL-1 (28). Studies in rats indicated suppressed GnRH and LH secretion after exposure to cytokines (22). These two types of hypogonadism were also described in other critically ill patients (29–31), and high serum concentrations of TNFα, IL-6, and CRP were measured in critically ill hypogonadal patients (23). Thus, hypogonadism of our patients can be related to the presence of either critical illness or COPD itself or a combination of both since we studied patients with acute illnesses in the presence of COPD. High CRP levels of patients prove the presence of inflammation, which is thought to be responsible for these changes.

Studies investigating COPD patients mostly enrolled male patients. This study included women who needed ICU support due to COPD exacerbation-related ARF and demonstrated a marked decrease in LH and FSH levels. LH and FSH levels decrease during severe illness (21, 32–35). A decline in gonadotropins occurs within 24 hours of an acute insult and continues until the nadir incline in 4–6 days (21, 26, 35). As the disease becomes more severe, hormonal suppression gets worse, with very low LH (≤0.5 mlU/mL) and FSH (≤1 mlU/mL) levels (21, 32, 33). Hormonal changes are independent of disease type, patient's age, and medications used (21, 22, 29). We also observed severely depressed values with LH ≤1.0 mIU/mL and FSH ≤5.0 mIU/mL in three and two women, respectively. Patients had high APPACHE II scores and CRP levels, which exhibit the severity of illness and the presence of inflammation considered responsible for hormonal alterations (22).

With disease improvement, suppressed hormones rebound as synthesis and secretion of them increase (36, 37). The normalization of hormones may take a long time. Elevation of testosterone to normal values was reported to take 2–12 months (29, 35). We studied testosterone of three recovered patients and found that the patient who was sampled on the 53rd day of the study had normal testosterone and gonadotropins.

We depicted a significant increase of the median level of FSH in recovered female patients. Since LH is more severely affected than FSH, FSH rise occurs first during recovery (33). A return of LH and FSH to normal ranges can take as long as the rise in testosterone (32, 34). Van Steenbergen reported severely depressed FSH and LH after 26 months of acute illness (32). We also had two patients with severely suppressed gonadotropins (LH ≤1.0 mIU/mL and FSH ≤5.0 mIU/mL) on discharge but we could not follow the progress of LH and FSH in these patients. Elevation of fT3 is another hallmark of the recovery in women.

We studied a specific group of patients whose hormonal changes can also be attributed to other factors. One factor is hypoxemia. Serum testosterone was measured low in men exposed to high attitude hypoxia (38), in patients with hypoxemic idiopathic pulmonary fibrosis (39), and hospitalized hypoxemic COPD patients (40). Since all patients were taking O₂ due to hypoxemia while transported to the ICU, we studied the PO₂/FIO₂ ratio instead of PO₂ and we did not show any relation between PO₂/FIO₂ and gonadotropins or testosterone. Moreover, we did not demonstrate any hormonal differences between home oxygen users and nonusers. All of our patients had ARF and were hypoxic on admission. This might explain the lack of difference between O₂ users and nonusers. Another factor blamed for low testosterone is steroid treatment but none of patients were on steroids before inclusion to the study (3, 4).

In contrast to gonadotropins and testosterone, the estrogen level was either normal or high in seriously ill patients (4, 16, 29, 41). Although gonadal synthesis of estrogen is inhibited in severe illnesses, an increased aromatization of adrenal androgens to estrogen in adipose tissues, which mediated by cytokines, is proved to result in normal or increased serum estrogen levels in severe ilnesses (42, 43). We showed normal E₂ levels in both male and female COPD patients. Male patients had a negative correlation between E₂ levels and the duration of NIMV and hospital stay. This correlation can be explained by potential benefits of estrogen on critical illness. The administration of estradiol to male rates with severe illnesses was associated with a higher survival rate, improved myocardial and hepatocellular function, and decreased lung tissue damage (44–46). Moreover, estrogen was found protective against lung injury in one study and regulatory in pulmonary alveolar formation, loss, and regeneration in another study (47, 48). The reason why female patients in this study did not have a causal connection between E₂ and morbidities might be the inclusion of the women with menopause. Studies have demonstrated that women have accelerated alveolar loss after menopause, women who smoke have faster decline in lung function compared to male smokers especially after 45 years of age, and they constitute 75% of never-smokers older than 55 years of age with clinical and lung function evidence of COPD (49, 50). Therefore, although the median level of E₂ in females is the same as in males, menopause might eliminate the beneficial effects of E₂ on morbidity. E₂ concentration among recovered women decreased significantly. This is probably due to reduced or abolished effects of inflammation on aromatization of androgens to estrogen.

In this study, patients had higher PRL levels than controls. PRL is among the first hormones reported to be increased in response to acute physical and psychological stress (36, 41, 51). Although its concentration was reported normal in stable and hypoxemic COPD patients (7, 10), we included COPD patients with respiratory insufficiency. The median PRL level of recovered women was found similar to the baseline value. This is apparently due to neuroleptic use because antipsychotic medication causes hyperprolactinemia by blocking dopamine receptors on lactotropes (52). Females using antipsychotic medication had higher PRL levels than those not using the medication on discharge.

We report low fT3 and TSH values in 21 COPD patients with ARF. Our results are similar to other studies that included either critically ill or COPD patients. Severe physical stress, such as surgery, starvation, trauma, and infection, leads to alterations in the thyroid axis within hours. Circulating T3 drops quickly mainly due to decreased conversion of T4 to T3 (36, 53–55). The magnitude of T3 drop reflects the severity of illness and has prognostic importance (54, 56–58). Although serum TSH usually remains normal, sometimes its concentration decreases below the normal limits and rarely to the detection limit (7, 54, 58, 59). Decreased serum TSH was reported in severe diseases and nonsurvivors (56, 58, 59). Our patients were seriously ill as they all needed artificial breathing support.

Female patients had lower fT3 than males on admission to the ICU. This difference between males and females cannot be explained by disease severity because they had comparable APACHE II scores, blood gas parameters, and CRP levels. This may be due to the effects of gender on hormonal response during critical illness, as reported for other neuroendocrine axises (41, 60). After recovery, female patients had fT3 elevated. This result is similar to other studies showing that conversion of fT4 to fT3 increases following recovery (11, 54, 56).

Plikat and co-workers showed that the median APACHE II score for patients with normal thyroid function, reduced fT3, and reduced fT3 and fT4 was 12.5, 18.0, and 21.0, respectively (58). Spratt et al. demonstrated lower testosterone and FSH levels in patients with APACHE II ≥ 15 than those with APACHE II < 15 (21). In contrast to these studies, we did not find any association between hormones and APACHE II scores. This is most likely due to the inclusion of patients with severe diseases as their median APACHE II score was 22.0 and all had APACHE II score ≥16.0. Another reason of not revealing any relation between disease severity and hormonal changes might be the inclusion of less number of patients. Since we had a low mortality rate, we could not compare survivors and nonsurvivors according to their hormonal states.

This study has some limitations. First, only 21 patients (13 F/8 M) were studied and more studies are needed to confirm data in a large group of patients. Second, we only had COPD patients with respiratory insufficiency. We did not include patients with stable COPD. That is why we could not exactly say that the hormonal changes were related to either COPD or the critical illness itself. Third, this study was performed in a tertiary care center and referred patients were the ones severely ill as proved by a high APACHE II score. Therefore, the patients might have extreme hormonal changes when admitted to the ICU.

In conclusion, we have shown significant changes in the neuroendocrine system of both female and male patients who had respiratory insufficiency in the presence of COPD. Female patients had low gonadotropins and fT3 and high PRL concentrations, while males had high PRL and low TSH concentrations on admission to the ICU. Although the median testosterone level was not statistically different than controls, 62.5% of male patients had low testosterone levels. The hormones increased to normal values following recovery from illness. An inverse correlation between E₂ and the duration of NIMV and hospital stay was detected in male patients although they had normal serum E₂ levels. Since all patients had severe disease with regard to high APACHE II scores and CRP levels, they presented with severe hormonal changes on admission to the ICU.

Declaration of interest

Dr. T. Akbaş, S. Karakurt, G. Ünlügüzel, and T. Çelikel have no conflicts of interest to disclose. Dr. S. Akalın has served on Advisory Boards of Eli Lilly, Novo Nordisk, and has given lectures for Servier and Merck Sharpe and Dohme.

ACKNOWLEDGMENT

We thank Dr. Nadi Bakırcı, M.D., Associate Professor of Health Care and Statistic at Marmara University, School of Medicine, for his help with the statistics of the study.

T. Akbaş designed the study, followed the patients, gathered the raw data of the study and contributed to writing of the paper. S. Karakurt and T. Çelikel helped for the study design and contributed to writing of the paper. G. Ünlügüzel did the laboratory workup and contributed to writing of the assay part of the manuscript. S. Akalın was the chief of the study, designed and followed the study, reviewed the paper and contributed to writing of the manuscript. All authors read, edited, and ultimately approved the final manuscript.