Serum concentration of chromogranin A at admission: An early biomarker of severity in critically ill patients

Abstract

Background. Chromogranin A (CGA), a stress marker released with catecholamines by the adrenal medulla, has never been associated with acute inflammation in critically ill patients.

Aim. To determine evidence for a link between serum concentration of CGA, biomarkers of inflammation, and outcome in patients admitted with or without the systemic inflammatory response syndrome (SIRS).

Methods. At admission, we measured in 53 patients and 14 healthy controls the serum concentrations of CGA, procalcitonin, and C-reactive protein. We also assessed the Simplified Acute Physiological Score (SAPS) in the patients.

Results. Serum CGA concentrations were significantly increased in SIRS patients with a median value of 115 µg/L (68.0–202.8), when compared to healthy controls (P<0.001). In cases where infection was associated with SIRS, patients had the highest increase in CGA with a median value of 138.5 µg/L (65–222.3) (P<0.001). CGA concentrations positively correlated with inflammation markers (procalcitonin, C-reactive protein), but also with SAPS. Receiver operating characteristic (ROC) analysis showed that CGA is equivalent to SAPS as an indicator for 28-day mortality (area under curve (AUC) for both: 0.810).

Conclusions. Patients with CGA concentration superior to 71 µg/L have a significantly shorter survival. A Cox model confirmed that CGA and SAPS were independent predictors of outcome.

• Chromogranin A
• critically ill
• outcome
• procalcitonin
• SAPS
• septic shock
• stress

Introduction

Critically ill patients are usually admitted with clinical and biochemical signs of systemic inflammation as a consequence of their disease. These signs are ascribed to the effects of inflammatory mediators produced during the early phase response to a stress of unspecified etiology. The systemic inflammatory response syndrome (SIRS) is the clinical response traditionally associated with this biological inflammation: its incidence can be more than 50% in critically ill patients, many of whom demonstrate or will develop severe infection 1. C-reactive protein (CRP) and procalcitonin are the traditional biomarkers used to evaluate the level of biological inflammation during SIRS. Despite their sensitivity and specificity for the diagnosis of infection, neither procalcitonin nor CRP give significant information as to outcome in such patients 2, who often subsequently die with multiple organ failure. Therefore, clinicians would appreciate new sensitive biomarkers that may reflect a link between clinical SIRS, acute inflammation, and outcome. The latter is currently evaluated by performance of complex scores such as the Simplified Acute Physiological Score (SAPS II) 3.

The stress marker chromogranin A (CGA) is a 48–52-kDa glycophosphoprotein and is known as the first member of the chromogranin/secretogranin family 4. It has been shown to be released by stimulated chromaffin cells together with catecholamines; it has numerous autocrine, paracrine, and endocrine functions 5. It is capable of influencing the vascular tone 6 and the myocardium 7 and is considered to reliably reflect the level of activation of sympathetic tone 8. In addition, some studies suggest that chromogranins could contribute to defensive or adaptative capabilities 5 and may therefore be considered as members of the acute phase response after a life-threatening challenge.

We therefore hypothesized that CGA is released in the bloodstream in critically ill patients, and that its level of release is proportional to the intensity of the stress. This study was then designed 1) to evaluate whether, on admission, in patients at risk of systemic inflammation, increased serum concentrations of CGA can be observed, and 2) whether CGA measuring could be of any help in the care of these patients at high risk of death.

Key messages

• Patients with serum chromogranin A concentration superior to 71 µg/L have a significantly shorter survival.

• This study suggests the opportunity for improving the care management of critically ill patients by measuring the CGA concentration at admission.

Abbreviations

Table

AUC	area under curve
CRP	C-reactive protein
ICU	intensive care unit
ROC	receiver operating characteristic curve

Materials, patients, and methods

Study population

This study protocol was approved by our institutional review board for human experimentation. Written informed consent was obtained before enrolment from each participant or authorized representative. A total of 100 consecutive patients admitted over 2 months in our department were prospectively considered for inclusion: only 53 of them could be included due to exclusion criteria, 5 patients with chronic renal (previous history of kidney disease and documented clearance of creatinine less than 60 mL/min), hepatic (with a previous history of liver disease and/or an abnormality of liver tests), or cardiac failure (with a history of heart disease and/or a pulmonary artery occlusion pressure over 16 mmHg with cardiac output in the range of normal values), on-going steroids 9, proton pump inhibitors 10, and a medical history of neuroendocrine tumors 5. Patients with surgical intervention within 1 month were also excluded. Fourteen healthy volunteers were finally recruited among staff as healthy controls. The flow chart of the study is shown in Figure 1.

Figure 1.  Flow chart of the study. A total of 100 consecutive intensive care unit (ICU) admissions over a 2-month period were screened and 14 healthy controls from our medical staff were also analyzed. Patients were excluded who had former chronic renal, hepatic, or cardiac failures, if they were undergoing treatment with steroids or proton pump inhibitors and if they had a medical history with neuroendocrine tumors. Finally, patients with recent multiple stress or a surgical intervention were also excluded.

Participants were divided into two groups:

1. Non-septic participants that included (a) healthy controls (HC); (b) patients without SIRS and without infection (NSNI) that were patients with benzodiazepines and/or neuroleptic drugs self-poisoning requiring mechanical ventilation; (c) SIRS patients that were admitted for out-of-hospital cardiac arrest without any other exclusion criteria.

2. Septic patients with infection and SIRS including (a) sepsis (S); (b) severe sepsis (SS); and (c) septic shock (SSH) patients. For the diagnosis of infection, special attention was paid to the presence of cultures positive for bacteria.

Patients were volume-resuscitated and managed according to consensual care 11.

Measurements

We measured the concentrations of serum CGA (Cisbio, Saclay, France), procalcitonin (PCT) (Brahms, Berlin, Germany), C-reactive protein (CRP) (Dade Behring, San Francisco, USA), white blood cell (WBC) counts, and creatinine. SAPS II was calculated according to standards 3. Mortality was assessed on day 28 after ICU admission.

Statistical analysis

Results are expressed as means±SEM for continuous variables except for biological values that are expressed in medians and interquartile ranges. Statistical analysis was carried out using Mann-Whitney test for continuous variables and Fisher's exact test for categorical variables. Multivariate analysis was used when necessary. Correlations were performed by Spearman rank test. Receiver operating characteristics (ROC) curves were constructed for predicting mortality; the areas under the curve were calculated, and ROC curves were compared. Finally a Kaplan-Meier analysis of survival was performed. In all cases, P<0.05 was considered to be significant.

A Cox proportional hazard regression model was used to evaluate the effect of CGA levels on the endpoint and calculate hazard ratios with 95% confidence intervals (CI). To assess the independent prognostic value of CGA, a backward elimination procedure was first used.

According to good practice, the data of the healthy controls were included neither in the correlation analysis nor in the survival evaluation.

Results

Characteristics of participants

The final study population consisted of 67 participants (53 participants and 14 healthy controls) whose clinical and biological characteristics are shown in Table I and Table II, respectively. In septic patients (n=31), the infection focus was the respiratory apparatus, the urinary, or biliary tracts (n=17, 8, and 6, respectively). Gram-negative bacteria were involved in 55% of the cases and Gram-positive bacteria in 45%. Healthy controls were younger than intensive care unit (ICU) patients (P<0.05). Within subgroups of patients, there were no differences as far as age, sex ratio, mean time from starting stress to first organ dysfunction, or mean ICU stay were concerned. SIRS and SSH patients underwent significantly more often mechanical ventilation compared with others (P<0.01). SSH patients had significantly higher concentrations of infection biomarkers than did healthy controls (P<0.01). When adjusting for age the results are still significant, and the statistical significance does not change in the multivariate analysis.

Table I.  Characteristics of participants. Data are mean values±SEM. Pairwise comparisons are shown.

	Controls (n=23)		SIRS without infection	Infection with SIRS (n=31)
Characteristics	HC (n=14)	NSNI (n=9)	SIRS (n=13)	S (n=9)	SS (n=8)	SSH (n=14)
Age (yr)	37.6±2.8	64.4±6.8	66.5±5.2	71.9±4.0	70.1±4.4	69.6±2.8
Male sex (%)	57.1	66.7	69.2	66.7	62.5	71.2
SAPS II	ND	39.2±5.0	53.6±3.9	39.7±4.3	47.0±2.1	73.4±4.0^b
Mechanical ventilation (%)	ND	22.2	76.9^b	55.6	50	100^b
Time from first organ dysfunction to admission (h)	ND	26.7±6.9	52.6±23.6	57.3±11.4	31.5±10.6	35.4±7
Mean ICU stay (days)	ND	11.9±3.2	22.2±8.6	11.2±3.1	11.9±3.5	18±3.3
Mortality (%)	0	11.1	23.1	11.1	25	50^a

^aP<0.05; ^bP<0.01 versus other subgroups.

HC = healthy controls; ICU = intensive care unit; ND = not determined; NSNI = patients without systemic inflammatory response syndrome (SIRS) or infection; S = sepsis patients; SAPS = Simplified Acute Physiological Score; SIRS = patients with SIRS but without infection; SS = severe sepsis patients; SSH = septic shock patients.

Table II.  Biological values in the study population. Data are median (interquartile range). The comparisons are versus the healthy control group (HC).

	Controls (n=23)		SIRS without infection (n=13)	Infection with SIRS (n=31)
	HC	NSNI	SIRS alone	S	SS	SSH
Chromogranin A (µg/L)	39.5	53.0	110.0^a	70.0^a	77.5^a	212.5^b
	(35.0–49.8)	(34.0–92.0)	(81.0–143.0)	(60.0–129.0)	(56–187.5)	(106.8–276.5)
C-reactive protein (mg/L)	<4	7.3	68.8	47.7	96.6	105.7
		(4.0–18.8)	(20.3–96.1)	(29.5–88.8)	(14.9–183.3)	(46.7–153.9)
Procalcitonin (µg/L)	<0.1	0.3	0.6	0.41	2.1	28.6
		(0.1–0.5)	(0.2–6.8)	(0.3–5.7)	(0.2–6.7)	(18.0–31.6)
White blood cells (giga/L)	7.5	13.4	13.7	8.9	13.15	13.6
	(6.463–8.625)	(10.8–14.3)	(12.3–15.7)	(7.7–14.8)	(11.55–17.85)	(11.55–16.7)
Creatinine (µmol/L)	<50	107	124	137	139	233
		(88–160)	(87–245)	(94–144)	(106–170)	(157.3–608.3)

^aP<0.05 versus HC. ^bP<0.001 versus HC. Significant differences are not indicated for other parameters.

HC = healthy controls; NSNI = patients without systemic inflammatory response syndrome (SIRS) or infection; S = sepsis patients; SIRS = patients with SIRS but without infection; SS = severe sepsis patients; SSH = septic shock patients.

Serum CGA concentrations study

On admission, acutely stressed patients with a SIRS (including both septic patients and patients with SIRS but without infection, n=44) had significantly higher serum concentrations of CGA than did controls (115 µg/L (68–202.8) versus 40 µg/L (35–52.5), P<0.001). As indicated in Table II, all the subgroups of patients demonstrated higher CGA levels than healthy controls (P < 0.05), apart from patients without either SIRS or infection. The septic group (S, SS, and SSH) had significantly higher CGA levels than patients with SIRS but without infection (138.5 µg/L (65–222.3) versus 110 µg/L (81–143), P < 0.01).

Figure 2.  Serum concentrations of chromogranin A (CGA) on admission in critically ill patients. Controls (SIRS-, Infection -; n=23) included healthy controls (n = 14) and patients with self-poisoning (n=9) without SIRS; SIRS patients (SIRS+, Infection-; n=13) were those with systemic inflammatory response syndrome from non-septic origin; septic patients (n=31) are in gray. The medians (interquartile ranges) for the three groups were 40.0 µg/L (35.0–52.5) for controls, 110.0 µg/L (81.0–143.0) for SIRS patients, and 138.5 µg/L (65–222.3) for septic patients respectively. * P < 0.01; # P < 0.001.

Relationship of CGA with inflammation markers, SAPS II, and outcome

CGA serum levels correlated positively with biomarkers of inflammation (C-reactive protein r²=0.555, P<0.01; procalcitonin r²=0.655, P<0.01; and white cell counts r²=0.306, P=0.012) and also with creatinine (r²=0.503, P<0.01) and with SAPS II (r²=0.405, P=0.003). Multivariate analysis indicated that creatinine and age were independent factors for the level of CGA, but neither SIRS nor infection were. Septic shock patients had a higher death rate than other patients (P < 0.05). In the non-survivor group, CGA on admission was three times as high as in the survivor group (192.5 µg/L (145.8–285.8) versus 65.0 µg/L (36.0–94.0), P < 0.001). As expected, septic shock patients had significantly worse SAPS II than others (P < 0.01). Finally, comparison of areas under the ROC curves showed that the serum CGA level at admission was as efficient as the SAPS II in predicting mortality in our study population (for both AUC 0.810, 95% CI 0.680–0.939 for CGA and 0.698–0.923 for SAPS II, respectively). For a cut-off value of 139 µg/L, the sensitivity of CGA to predict death was 78.6% and its specificity 79.5%.

As shown in Figure 3, the Kaplan-Meier analysis indicated that the admission value of CGA was significantly associated with the survival time: patients with supramedian levels of CGA (>71 µg/L) had significantly shorter survival time (P=0.0013). By multivariable Cox proportional hazards regression, CGA levels (hazard ratio (HR) = 1.022, 95% CI 1.012–1.051) and SAPS II (HR = 1.025, 95% CI 1.015–1.054) at admission were significantly associated with outcome (P<0.05).

Figure 3.  Influence of serum chromogranin A (CGA) concentration at admission on survival time. Kaplan-Meier analysis in the ICU patients (n=53) with inframedian (dashed line) or supramedian (solid line) values (71 µg/L). These data confirmed that the lower the serum concentration on admission, the longer the survival time (log rank test, P=0.0013). Survival time is plotted against survival rate.

Discussion

In this study, we found that 1) patients suffering from critical illnesses had a significant increase in plasma CGA concentrations: a fact which has previously never been reported; 2) a positive correlation between plasma concentrations of CGA and standard biomarkers of inflammation existed in vivo; and 3) CGA plasma concentrations were a further indication of predicting the risk of death in these non-surgical ICU patients.

Origin of CGA in critically ill patients

Increased surem concentrations of CGA have already been reported in other categories of patients. For instance, this glycophosphoprotein is a sensitive and specific tool in the diagnosis of endocrine tumors 5; however, this would not have appeared in our study because of the exclusion criteria. Other mechanisms of high serum CGA levels include a decreased renal clearance in chronic renal failure 12. In our study we have excluded chronic renal failure patients, but acute renal failure may occur and partly contribute to the serum CGA concentration increase. Indeed, this is in agreement both with the correlation we have recorded with creatinine and with the fact that creatinine was an independent predictor of CGA in multivariate analysis. Enhanced sympathetic tone and a production of cytokines have also been reported as increasing CGA in chronic and acute cardiac failure 13, 14, and this must be taken into account in our patients because they frequently demonstrate protracted changes in hemodynamic profiles. The correlations with C-reactive protein and procalcitonin noticed in our study further support the idea that severe acute inflammation can be responsible for an increase in plasma CGA. Additional mechanisms could be involved, such as a release of the protein by immune cells. This would be in line with data of others who have reported 1) a presence of chromogranin-derived peptides in circulating lympho-monocytes of coronary artery bypass patients 15, and 2 with the presence of CGA detected in secretory products from stimulated neutrophils 16. Our data suggest that white blood cells challenged by stress may partially contribute to increased serum CGA release because CGA correlated significantly with white cell counts. We also show a significant increase of serum CGA in acutely stressed patients compared with controls without SIRS (i.e. healthy controls and self-poisoned patients without SIRS). To obtain further information into the mechanism underlying this phenomenon, we compared septic patients and SIRS patients without infection. Although the presence of SIRS is obviously associated with an increase of CGA, we failed to prove SIRS or the presence of infection as independent factors for CGA increase in a multivariate analysis. In contrast, however, their association (which usually indicates severe infection requiring ICU admission) is a condition leading to significant increase of this glycoprotein, possibly because of greater stress. A substantial neuroendocrine activation and a common cytokine-mediated stress mechanism are two ways by which systemic inflammation with proinflammatory cytokines tumor necrosis factor alpha, interleukin-6 (TNF-α, IL-6) activates the hypothalamic-pituitary-adrenal axis and the locus coeruleus-norepinephrine system 17. Thus, in the adrenal medulla, CGA which co-localizes in chromaffin secretory granules with catecholamines will be released in the bloodstream 4. Furthermore, stress-induced release of glucocorticoids may upregulate the CGA synthesis at the gene level 9, providing another additional explanation for early increased serum CGA concentrations in patients suffering from acute stress.

Relationship between CGA and outcome

In this study, correlations between CGA and SAPS II were obtained. Thus, for each range of SAPS II, plasma CGA significantly increased. In addition, CGA concentrations were related to survival time (Figure 3), in accordance with studies performed in end stage cardiac failure 13 but also in patients with acute myocardial infarction 14. Indeed, patients suffering from chronic renal failure only also demonstrate increased CGA serum concentrations but, in the absence of simultaneous inflammation, they do not necessarily develop further organ failures resulting in death. This suggests that the role played by CGA in pathophysiological conditions may change with the occurrence of overwhelming systemic inflammation. On the other hand, organ failures (heart, renal, and hepatic) may also contribute to the increase in serum CGA concentrations: this could result in a vicious circle with accumulation of this glycoprotein, which would then have a protracted noxious action 18. Further studies are necessary to understand whether CGA contributes directly to the development of multiple organ failure or/and if CGA is merely a marker of multiple organ failure severity. Some information is already available indicating that CGA is capable of triggering apoptosis in vitro 18 through a mechanism 19 that is often associated to sepsis 20. Because apoptosis is frequently associated to organ failure in SIRS patients, this mechanism offers a possible explanation. Currently, the way to predict outcome of critically ill patients is organ failure score calculations. On this approach, SAPS II is considered as the gold standard score. The ROC analysis demonstrated that serum CGA concentration on admission is a reliable predictor of 28-day mortality for critically ill patients. It is as efficient as the SAPS II, but the SAPS II requires the analysis of 17 clinical and biochemical parameters to make an accurate prediction 24 hours after admission 3. CGA measured on admission may enable physicians to predict more easily and earlier an ominous outcome.

Study limitations

This study provides data that have never been collected before in critically ill patients. It must be considered as a first study for a further prospective program using CGA as a marker of severity. It is not an explanation for all the reasons of increased CGA in the serum. Because we have strictly excluded all known conditions which can influence the serum concentrations of CGA, we have finally a small number of patients included in our study: this might explain the lack of difference in CGA values observed between SIRS patients without infection and sepsis patients. Also, our results were collected in patients with a single acute stress condition and may therefore not be extended to those undergoing multiple consecutive stresses or a surgical stress followed by a complication.

Conclusions

We have shown an increase of serum CGA levels in non-surgical critically ill patients and a significant correlation between CGA concentrations, the inflammation biomarkers, and the outcome. Our present study suggests the opportunity for improving the care management of critically ill patients by measuring the CGA concentration at admission.

Acknowledgements

We acknowledge support of both the Délégation à la Recherche Clinique des Hôpitaux Universitaires de Strasbourg (grant n°3150, PHRC 2003) and the Department of Foreign Affairs (Charcot grant for DZ). This work was also funded by INSERM and the University Louis Pasteur (Strasbourg, France) and supported by the Fondation Transplantation (PhD grant to DZ). We thank Cisbio (Marcoule, France) for the gift of the CGA kits and Nathalie Mela for careful technical assistance (CNRS, UMR 7191 Institut de Physique Biologique, Strasbourg, France).

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper. D. Zhang and T. Lavaux contributed equally to the writing.