Introduction

Critically ill patients are at risk for developing muscle weakness or intensive care unit-acquired paresis (ICUAP) [1]. ICUAP is frequent, affecting at least 25% of critically ill patients [1], and is a serious neurological disease as it has been associated with increased mortality [2, 3], prolonged mechanical ventilation and hospital stay [4, 5] and long-term disability [6]. It can be related to axonal polyneuropathy (critical illness polyneuropathy), myopathy (critical illness myopathy) or, more frequently, both (critical illness neuromyopathy, CINM) [1, 5, 7]. Various mechanisms have been identified in the pathophysiology of CINM, including systemic inflammatory responses, immobilization, malnutrition, drug toxicity, metabolic disturbances and hyperglycaemia [5]. A possible contribution of endocrine dysfunction in ICUAP is plausible since prolonged critical illness is associated with impaired secretion of various hormones that have a direct effect on muscle metabolism [8]. This hypothesis is also supported by a recent study in which women had a more than fourfold greater risk of developing ICUAP [1]. Therefore, we assessed plasma levels of gonadic hormones and those of hormones both mostly studied in critically ill patients and affecting muscle metabolism, i.e. cortisol, DHEA, TSH and IgF1. The primary objective of the present study was to assess the relationships between plasma levels of these hormones and ICUAP 7 days after awakening in patients with protracted critical illness being weaned from mechanical ventilation [9].

Methods

Patients

Patients were enrolled in a prospective study evaluating the relationship between ICUAP and several outcomes including duration of mechanical ventilation [4] and mortality [3]. Inclusion and exclusion criteria have been previously published [4]. After 7 days of mechanical ventilation, patients were screened daily for awakening and comprehension using five simple verbal commands, as previously described [1, 4]. Patients were enrolled in the study, and hormonal assays were performed on the first day when awakening and comprehension were satisfactory (day 1). The study protocol was approved by the Ethics Committee of Saint-Germain-en-Laye, France, which oversees all participating hospitals. Informed consent was obtained from all patients. One hundred sixteen patients fulfilled the inclusion criteria. Among the 116 patients, plasma hormones levels were available in 102 patients (Fig. 1).

Fig. 1
figure 1

Flow chart

Full size image

As previously described [4], we recorded demographic characteristics, admission diagnosis, comorbidities, ICU admission diagnosis, Simplified Acute Physiologic score II (SAPS II) [10] at admission and day 1, daily organ dysfunction and/or infection (ODIN) score [11] from admission to inclusion as well as use of vasopressor, duration of mechanical ventilation and length of ICU stay prior to inclusion. Mean blood glucose and cumulative dose of corticosteroids (expressed as hydrocortisone equivalent dosage) prior to inclusion were calculated for each patient.

Endocrine measurements

Plasma follicular stimulating hormone (FSH), luteinizing hormone (LH) and prolactin concentrations were measured using radioimmunometric assays (RIA; Immunotech, Beckman Coulter France); intra- and interassay coefficients were less than 8%. Detection limits were 0.2 mUI/ml for FSH and LH, and 0.5 ng/ml for prolactin. Plasma concentrations of testosterone, dehydroepiandrosterone (DHEA) and estradiol were determined by RIA after either extraction. Intra- and interassay coefficients were less than 10%. Detection limits were 0.07 ng/ml for testosterone, 0.3 ng/ml for DHEA and 10 ng/ml for estradiol. Dehydroepiandrosterone sulphate (DHEAS) and progesterone were measured by RIA (Immunotech Beckman Coulter France); intra- and interassay coefficients were less than 10%, detection limits were 60 ng/ml (DHEAS) and 0.05 ng/ml.

Plasma cortisol concentration was measured directly by RIA (Cis Bio International) with intra- and interassay coefficients of 5% and 7%, respectively, and a detection limit of 0.35 μg/dl. Plasma cortisol levels measured in patients still treated with hydrocortisone were not taken into account in the analysis. Plasma concentrations of TSH were determined by a third-generation sandwich immunoassay (Access 2, Beckman Coulter France). Intra- and interassay coefficient of variation (CV) were less than 3%; detection limit was 0.01 mUI/l. IgF1 was measured directly by RIA (Cis Bio International) with intra- and interassay values less than 6%; detection limit was 5 ng/ml. All samples from a given patient were analysed in duplicate in a single assay. Plasma level was considered abnormally low when it was below the lowest normal value (Table 1).

Table 1 Normal values of plasma hormones levels
Full size table

We defined hypogonadism in men, independently of age, as plasma testosterone level of <3 ng/ml [12]. Hypogonadism was considered secondary (SH) when FSH and LH concentrations were <5 mU/l and primary (PH) when FSH and LH levels were >10 mU/l [12].

Women were considered post-menopausal if they were older than 55 years or if they reported amenorrhoea for 1 year or more. Because of the small number of pre-menopausal women (n = 8), plasma levels of sex-dependent hormones (FSH, LH, estradiol (E2), testosterone, DHEA, DHEAS and progesterone) were analysed only in post-menopausal women. In post-menopausal women, primary hypogonadism was considered to be the rule; hypogonadism was considered secondary when LH and FSH levels were inappropriately low (<10 mUI/l) in the presence of a low estradiol level (<10 pg/ml) in post-menopausal women [12].

Detection of ICUAP

Limb muscle strength was assessed in all patients at inclusion and on day 7 by measuring the strength of three muscle groups in the upper and lower limbs using the Medical Research Council (MRC) score. This score ranges from 0 (“paralysis”) to 5 (normal) and composite limb muscle strength (sum score) from 0 to 60 [13]. Muscle strength was assessed by an ICU physical therapist or senior physician in each centre. Consistent with our previous study [1], we defined ICUAP as MRC <48 at day 7.

Endpoints

The endpoints were the association of ICUAP, diagnosed 7 days after awakening, with plasma levels of non-gonadic hormones measured at day 1 in the whole population and with day-1 plasma levels of gonadic hormones separately for post-menopausal women and men.

Statistical analyses

Variables were recorded on admission, between admission and awakening, at awakening and at discharge (Tables 2, 3). Continuous variables were not dichotomized and are reported as median with interquartile range (IQR); categorical variables were coded as 1 or 0 and reported as numbers (%). The Mann–Whitney U test was used for comparison of continuous variables, and the chi-square or Fisher exact test was used for categorical variables.

Table 2 Clinical characteristics and outcome of patients
Full size table
Table 3 Hormonal characteristics of patients at day 1
Full size table

Clinical characteristics, hormonal status and outcome were first compared between women and men by using univariate analyses (Table 2). Then, patients with and without day-7 ICUAP were compared on a priori selected variables, including those previously found to be associated with ICUAP [5, 7], mean blood glucose and hormonal measurements (Table 4). For variables a priori (sex-dependent hormones) or a posteriori associated with gender (p ≤ 0.10) we systematically searched for first-order interactions. Odds ratios (OR) and 95% confidence intervals (CI) were estimated by using asymptotic (Mantel–Haenszel) or exact logistic models as appropriate for variables associated with ICUAP with p < 0.10. Crude ORs are presented when no association with gender was a priori, or a posteriori, observed (p > 0.10), otherwise ORs were adjusted for gender when no interaction could be considered (p > 0.15) or stratified by gender when an interaction could not be excluded. Considering the small number of events, multivariate analysis including all variables associated with ICUAP could not be performed.

Table 4 Comparison between patients with and without ICU-acquired paresis at day 7 (assessed in 86 patients)
Full size table

p ≤ 0.05 was considered statistically significant. All significance tests were two-tailed. Data were analyzed using Stata statistical software (release 8.0, 2003; StataCorp, College Station, TX) and StatXact and LogXact software (CYTEL Software Corporation, 2001).

Results

Patient characteristics (Table 2)

Awakening occurred after a median duration of mechanical ventilation of 10 days (8–14 days). At day 1, median MRC score was 40 (23–51). From day 1 to 7, 16 (16%) patients died or were discharged. MRC score was therefore assessed in 86 (84%) patients at day 7, including 56 (71%) men and 23 (29%) women, among whom there were 18 post-menopausal women. At day 7, median MRC score was 48 (29–56) and 39 (49%) patients had ICUAP. MRC score at day 7 was significantly lower and ICUAP more frequent in women than in men. Twenty-four patients (24%) died during hospitalization, including 15 in the ICU.

Hormonal status (Table 3)

Plasma cortisol levels were in the low range in 14 (14%) patients. Median cumulative dose of hydrocortisone and median delay between discontinuation of corticosteroids and hormonal measurement were not correlated with plasma cortisol level (data not shown). Plasma levels of IgF1 and TSH were low in 81 (79%) and 14 (14%). SAPS II assessed at awakening was inversely correlated with plasma levels of DHEA (rho = 0.22, p = 0.03), DHEAS (rho = 0.20, p = 0.05) and IgF1 (rho = 0.29, p = 0.004). Plasma levels of the non-gonadic hormones were not significantly associated with gender.

In post-menopausal women (n = 29), plasma levels of DHEA and DHEAS were low in 22 (76%) patients. SH was observed in 22 of these women (76%). In men, plasma levels of DHEA and DHEAS were low in 48 (74%) and 56 (86%) patients. Plasma testosterone levels were low in all except 1 patient, and estradiol/testosterone ratio was high in 20 (31%) men. SH was present in 46 (71%) men, alone (n = 33, 51%) or associated with PH (n = 13, 20%). PH and isolated PH were present in 25 (39%) and 12 (19%) men.

Relationships between day-1 hormonal status and ICUAP at day 7 (Table 4)

In the whole population, mean blood glucose was significantly higher in patients with than without ICUAP; the association persisted when the analysis was adjusted for gender. Plasma prolactin levels tended to be greater and secondary hypogonadism more frequent in patients with ICUAP. Plasma levels of cortisol, IgF1 and TSH did not differ between the two groups.

Plasma estradiol tended to be lower in women with than without ICUAP. In men, only estradiol/testosterone ratio significantly differed between the two groups. PH tended to be less frequent in men without ICUAP.

Female gender, number of days with at least two organ failures, duration of mechanical ventilation, septic shock, and use and dose of corticosteroids were also significantly associated with ICUAP.

There was no statistical difference between patients with (n = 11) and without (n = 75) severe ICUAP, defined by MRC score below or above 24, except for plasma IgF1 level, which was significantly lower in patients with severe ICUAP (54 [36–99] versus 74 [56–156], p = 0.02).

Discussion

This study confirms prior data [1] showing that ICUAP is significantly more frequent in women; however, we were not able to identify a hormonal disturbance to explain this association. We also found that protracted critical illness is associated with low plasma levels of IgF1 and with secondary hypogonadism in more than two-third of patients, irrespective of gender. These neuroendocrine alterations have been previously reported in a comparable range [8, 12, 14–17]. We also report that more than 70% of patients had low plasma levels of DHEA and DHEAS, but only 14% of them had low plasma cortisol levels. The combination of low plasma DHEA and DHEAS levels with normal circulating cortisol levels can be interpreted as a sign of exhausted adrenal reserve [18]. In contrast to previous reports [8, 12, 19–21], alterations of TSH and prolactin secretion were relatively uncommon in our prolonged critically ill patients, possibly because few patients were treated with dopamine infusion [22, 23]. In agreement with previous studies [8, 17, 24], we found that prolonged critical illness results in primary hypogonadism (58%) and increased aromatization of androgens in men. We did not find any statistical relationship between ICUAP and plasma levels of cortisol and TSH in both men and women. While plasma levels of IgF1 were significantly lower in patients with severe ICUAP, we believe that this result should be interpreted with caution because of the small number of cases. It should be noted that administration of growth hormone can be highly deleterious for critically ill patients [25]. Therefore, our results rather suggest that non-gonadic anabolic hormones are not involved in muscle weakness.

Although we confirmed our prior findings that women are at increased risk for ICUAP, this gender association was not found in other observational studies [2, 7, 26]. There is no clear explanation for this discrepancy. Our results indicate a potential pathogenic role of some gonadic anabolic hormones. Decreased testosterone activity is a critical endocrine factor associated with muscle weakness in men. This may result from either decrease in synthesis of testosterone or increase in its aromatization, as suggested by the low plasma testosterone levels and high estradiol/testosterone ratio in men who developed ICUAP. In post-menopausal women, muscle weakness tended to be associated only with decrease in estradiol and FSH, both of which have anabolic properties.

The therapeutic implication of these findings is uncertain. The pathogenesis of CINM is highly complex, involving inflammatory, toxic and metabolic factors [7]. We do not know at this time the respective importance of these pathogenic factors. Therefore, it is conceivable that administration of a given anabolic hormone will not improve muscle strength and will not be safe in critically ill patients. For instance, androgens have been shown to have no significant effect on muscle strength in non-critically ill patients [27, 28]. Moreover, the strength of the association between ICUAP and circulating levels of anabolic hormones was weak in comparison with other known risk factors, such as hyperglycaemia, SAPS II and steroids.

Therefore, a preventive approach, such as controlling blood glucose or reducing use of corticosteroids, might be a more relevant therapeutic approach. Hyperglycaemia has been identified as a significant risk factor for electrophysiological abnormalities potentially suggestive of critical illness polyneuropathy [7], but our study is the first to show that increased blood glucose is associated with ICUAP. Although there are no prospective randomized trials evaluating the impact of intensive glycaemic control on ICUAP as a primary outcome, the fact that intensive insulin therapy reduced the severity of electrophysiological abnormalities suggestive of CINM in two large clinical trials suggests that major clinical benefit can be expected [5]. Interestingly, we found that mean blood glucose was higher in post-menopausal women than in men. This discrepancy may be explained by the occurrence of insulin resistance in menopause [29].

Limitations of the study

ICUAP was diagnosed with clinical muscle testing done 7 days after awakening. This delay between assessment of plasma levels of hormones, performed at time of awakening, and diagnosis of ICUAP, done 7 days after awakening, was deliberately selected for the following reasons: (1) to be consistent with our original study, which showed that women were at higher risk for ICUAP 7 days after awakening [1], (2) to select patients with lasting muscle weakness reflecting a more severe subset of CINM and (3) to determine the prognostic value of endocrine variables, which necessarily require a delay between predictors and the predicted event. It is possible that this delay resulted in the selection of patients with critical illness polyneuropathy rather than critical illness myopathy, as the latter recovers more rapidly than the former [6]. Such a bias may account for the fact that ICUAP was not correlated with plasma levels of anabolic hormones but with blood glucose. However, lack of power might be the main explanation for these absences of correlation. This is the case for women, especially pre-menopausal women, who were excluded from the analysis of sex-dependent hormones. We chose to study patients at awakening because it is a prerequisite for clinical assessment of muscle strength and a major milestone in the course of critical illness. It is at this time when important therapeutic decisions are taken such as ventilator weaning or physiotherapy.

According to our previous studies [1, 3, 4] and that of Ali et al. [2], ICUAP was defined by MRC score below 48, which corresponds to an average strength of less than 4 (anti-gravity strength) across all muscles tested. This criterion is also relevant because so-defined ICUAP is independently associated with increased duration of mechanical ventilation [4] and mortality [2, 3]. It has to be noted, except for IgF1, that our results remain unchanged when using a lower cut-off (i.e. MRC score <24) for defining ICUAP.

The biological effects of hormones depend on their circulating levels but also on synthesis and clearance of specific and non-specific circulating hormone binding proteins, and on the expression and regulation of hormone receptors. Since we did not assess binding protein levels nor hormone receptor activity, we cannot exclude that a given hormone is associated with ICUAP based on total serum levels alone. Finally, interpretation of single circulating levels of hormones has to be cautious because levels fluctuate with time and dynamic assessments were not performed [12]. Repeated measurements before and after awakening might have provided very interesting information. However, sampling before awakening would have implied obtaining blood sample from a large number of patients who would die before awakening and would therefore not be evaluated for ICUAP. We reasoned when designing the project that obtaining blood samples after awakening, notably at day 7, might be less acceptable to patients, especially those who would recover normal strength. It has to be noted that plasma levels of many hormones are decreased and fluctuate less post acute phase of critical illness. Therefore, it is likely that plasma hormonal levels performed 7 days after awakening would have been comparable to at day 1.

These results must be interpreted with caution, as a statistical association does not signify a causal relationship. Endocrine dysfunction and ICUAP might be two independent consequences of critical illness. Because of the relatively small number of events, we did not perform multivariate analysis to determine whether specific endocrine imbalances were independently associated with ICUAP or in-hospital mortality. Despite these limitations, our study is the first to assess the relationships between hormonal status and ICUAP.

In conclusion, our study confirms that women developed ICUAP more frequently, perhaps because of the higher prevalence of hyperglycaemia. In men, there is a weak association between ICUAP and a decrease in gonadic anabolic hormones. IgF1 was significantly lower in 11 patients with severe ICUAP. We did not find an association between other non-gonadic anabolic hormones and muscle weakness. However, before envisioning specific clinical trials, the associations demonstrated in this study should be confirmed in a larger cohort and their pathogenic mechanisms elucidated.