Neuron-specific enolase in diagnosis and prognosis of delirium: a systematic review

Abstract

Delirium, characterized by a sudden onset of neuropsychiatric symptoms, is a highly prevalent syndrome whose diagnosis is defined solely through clinical evaluation. Due to the often challenging reliability of assessments, especially in non-cooperative patients, there is a growing emphasis on exploring new reliable biomarkers, such as Neuron-Specific Enolase (NSE). NSE, an enzyme primarily found in neuronal and neuroendocrine tissues, has been clinically used to assess the prognosis of patients who have experienced traumatic or hypoxic brain injuries. Thus, the primary purpose of the present review is to examine the literature to determine whether NSE is applicable for diagnosis and/or prognosis of patients with delirium. Literature was searched using Pubmed, Lilacs and Scielo databases, and all published reports identified as potentially relevant were independently assessed by each reviewer. All relevant original studies were included and independent extraction of articles was performed by three authors using predefined data fields. Twenty one studies (2,311 patients) satisfied the entry criteria, among which only eight suggest a possible association between NSE and delirium, particularly in intensive care settings, and only one correlate NSE with delirium prognosis. Also, significant heterogeneity was observed among studies, varying across study design, setting, and methodologies. Furthermore, the majority of the selected studies presented severe methodological limitations, particularly small samples. In conclusion, this systematic review underscores the need for further research with larger, standardized studies to establish the reliability and validity of NSE as a diagnostic and prognostic tool for delirium. The current evidence does not sufficiently support its routine clinical application in assessing patients with delirium.

Background

Delirium is a neuropsychiatric syndrome defined as a confusional state of sudden onset and fluctuating course, in which disturbances in attention and awareness represent a change from the baseline, and that is not better explained by an underlying neurocognitive disorder [1, 2].

The importance of delirium is evident in its incidence, morbidity, and costs for the health system. At least one-third of the hospitalized patients develop delirium, half of which at the admission and the other half during their stay [3]. Moreover, delirium is estimated to occur in around 15 to 25% of elderly patients submitted to major surgery, and 75% of patients at intensive care units under mechanical ventilation [3]. This high incidence leads to healthcare costs that are over 160 billion dollars each year in the USA [3].

Furthermore, even though delirium is, by definition, an acute condition, a significant number of patients maintain a degree of neurocognitive disability even after its resolution [4].

To diagnose delirium, medical professionals can only rely on their clinical suspicion, which proves to be a challenge, given that the neurocognitive impairment may be wrongly attributed to old age and other humor disorders. Therefore, it is estimated that only up to 12 to 35% of delirium cases are recognized [5].

The diagnostic criteria are stated in the Diagnostic and Statistical Manual of Mental Disorders (DSM-5), and they are very much similar to the very definition [2]. Hence, several questionnaires were developed and validated to help physicians assisting patients with disturbances of consciousness. The most widely used are the Confusion Assessment Method (CAM) and its brief version (bCAM); however, these tools might be impaired by the non-cooperative patient [6].

Considering the facts above, it is clear the importance of a biomarker that could identify patients with delirium [7]. A biomarker is defined as a characteristic that can be measured and assessed as an indicator of biological processes (normal or pathological) or as a response to therapeutic interventions [8, 9].

A biomarker’s relevance is its ability to provide information on questions of interest adequately, and the validity represents the effectiveness in doing so [9]. So far, several biomarkers have been described as being able to diagnose delirium; however, none of them have been widely incorporated in clinical practice.

Neuron Specific Enolase (NSE) is a 78 kD gamma-homodimer and represents the dominant enolase-isoenzyme found in neuronal and neuroendocrine tissues. Its primary function is to catalyze the conversion of 2-phosphoglycerate to phosphoenolpyruvate and its levels in other tissues, except erythrocytes, are negligible [10].

Due to this organ-specificity, concentrations of NSE in cerebrospinal fluid and serum are often elevated in diseases resulting in acute neuronal destruction [11]. Elevated serum NSE levels can be found in coma patients after a hypoxic insult [12] or head trauma [13] and are usually related to a poor prognosis.

Therefore, the present review’s primary purpose is to examine the literature to determine whether NSE is applicable for diagnosis and/or prognosis of patients with delirium.

Methods

Study design

This was a systematic review to evaluate NSE as a diagnostic and prognostic tool in delirium patients.

This study followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.

Eligibility criteria

We included original observational studies (cohort, case-control and cross-sectional studies) that evaluated plasma or cerebrospinal fluid NSE levels among adult patients with delirium. There was no restriction in terms of year of publication, intervention or outcome. Non English or adult human studies, case reports, narrative reviews, and opinion articles were excluded.

Search strategy

A strategy for a literature search was developed and executed by a medical doctor with inputs from the study investigators. The search strategy was created using a combination of keywords and standardized index terms related to NSE and delirium.

The search strategy included the search terms (Delirium) AND (Neuron specific enolase OR enolase OR NSE).

The initial search was run in July 2024 in PubMed (66), Scielo (1) and Lilacs (2).

Study selection and data extraction

In the first phase, 3 investigators independently screened all titles and abstracts for eligibility.

In the second phase, all studies considered potentially relevant were retrieved as full text and independently assessed for eligibility. The investigators were not blinded to the authors, journals, or results of the studies.

Pertinent data were independently extracted for all the studies using a standardized, predefined extraction form. The extracted data included author, journal, year, country, study type, included patients characteristics and main findings, which were paraphrased and adapted from the original publication results and conclusions. Unadjusted and adjusted effect estimates reported by the studies were extracted. Only data available in published manuscripts and abstracts were used.

Results

Initially, 69 articles were selected after searching the databases, and 2 duplicates were identified and excluded. Thus, 67 articles remained for the first phase of the analysis.

After reading the title and abstract, 23 articles were selected (not related to NSE and delirium, non english or adult human studies, editorials or reviews were excluded), and then read in full by the reviewers. 2 studies were excluded in the second phase because they did not meet the inclusion criteria (lack of data about NSE or delirium-NSE correlation).

The 21 selected articles constitute this systematic review (Fig. 1; Tables 1 and 2).

Fig. 1

Diagram of studies selection

Table 1 Key characteristics of included studies

Table 2 Included patients characteristics and main findings of the selected studies

Study design

Out of the twenty one articles selected, two were randomized clinical trials, six were case-control studies, and thirteen were cohorts.

Methodologies

Each study’s sample sizes varied from 13 to 194, with a median value of 74 and an interquartile range of 44 to 120 patients.

Three out of twenty one studies determined a cutoff value for defining delirium, utilizing plasmatic NSE levels above the 95th percentile or above 12,5 μg/L [22, 24, 26].

Two out of twenty one studies took samples in varying time spams to evaluate NSE kinetics and release patterns before and after situations related to brain dysfunction [15, 17].

Most studies analyzed serum samples to determine NSE concentrations, as only Caplan et al. [19] and Zhang et al. [34] considered concentrations in cerebrospinal fluid.

The delirium diagnosis was confirmed with a clinical evaluation, carried out most frequently before and after exposure. Nonetheless, the delirium assessment tools varied, as five studies utilized the latest DSM criteria available at the time and thirteen studies utilized the Confusion Assessment Method (CAM) and its variations (CAM-ICU). Only three studies considered neither DSM criteria nor CAM to diagnose delirium.

Results of individual studies

Eight studies found statistically significant variation in plasma or CSF NSE levels in patients with and without delirium [15, 19, 20, 22, 24, 28, 30, 33, 34], while other eleven did not [14, 16,17,18, 23, 25,26,27, 29, 31, 32].

One study concluded that NSE might have prognostic value, for its higher levels being related to a higher risk of delirium and overall mortality [20], while other three did not [15, 21, 34].

Limitations and bias

Ten authors acknowledged that they worked with small sample sizes [15, 19, 20, 23, 24, 26, 27, 32,33,34]. Other limitations stated were also related to the small number of participants, such as lower incidence of cases [20, 22, 27, 33] and the rigidity of exclusion criteria [21, 23, 27, 28, 31].

Three out of twenty one authors declared limitations in the post-analytical phase, for instance, hemolysis [14] and the difference between peripheral and CNS samples [18, 32].

We did not observe any limitations reported in four studies [16, 17, 25, 30], even though some of the limitations discussed by other authors could also be applied to them.

Discussion

Delirium is the outcome of a multifactorial process that culminates in the acute confusional state and not a proper disease [1]. Therefore, it may be better comprehended as an “acute brain failure” in many ways, similar to heart failure and other organic dysfunctions. The difference is the lack of biomarkers that could reliably define the diagnosis and prognosis, which would serve a similar purpose that troponin and brain natriuretic peptides do for heart failure [35, 36].

Neuron Specific Enolase is an intracellular enzyme, located almost exclusively in neurons and neuroendocrine tissues [10]. This fact prompted its use as a marker of neuronal injury. There is no consistent evidence that, in patients with delirium, neuronal death occurs. However, other central nervous system injury markers, like S100B, have been reported to reflect the delirium severity [7].

Therefore, in this review, we evaluated serum NSE levels’ applicability as a diagnostic and prognostic tool to assess patients with delirium. We selected twenty one papers, published from 1995 to 2024, that evaluated the relationship between serum or CSF NSE levels and the occurrence of delirium. Due to the wide variation in study design, participants, interventions, data, and outcomes reported, we decided to describe the studies, their results, their limitations, and possible bias.

Only eight studies concluded that NSE could have any value for identifying or giving the prognosis of patients with delirium [15, 19, 20, 22, 24, 28, 30, 33, 34]. All of these studies, except one [19], were conducted in intensive care units. Although there is no logical explanation for this finding, it can be supposed that these patients received a higher degree of attention to changes in consciousness levels, making it more probable that delirium could be identified.

This systematic review suffers significantly from its study pool’s vast heterogeneity, which leads to an impossibility to create a meta-analysis. The data published in these studies have almost no similarity between each other, as seen in the delirium assessment: the assessment tool varied significantly, and the time of evaluation varied even further.

Besides that, the number of patients included in each study was very small for the designed purpose. The three largest studies included around 177 patients [24, 29, 32] and, notoriously, presented opposite results. As a comparison, one of the first studies to report the usefulness of troponin to diagnose myocardial infarction enrolled 388 patients [37]. Also, selection bias is also very plausible in this systematic review, the size of which we could not predict due to the incapability of creating a funnel plot.

Although a discussion about the limitations was supposed to be found in each paper, some did not present it. All the others discussed the matter, pointing out the small samples and even analytical troubles responsible for negative results. Anderson and colleagues [22] even stated that the varying results across studies might be explained by different outcome definitions or the difference in pathophysiologic mechanisms leading to delirium in different patient populations.

Conclusion

In conclusion, the present data and findings concerning NSE measurements are not sufficient to be clinically applied in the neurobehavioral and/or neurocognitive diagnoses or prognosis of a patient that presents to the emergency room or to the intensive care unit. Thus, further data is necessary to support any claims that NSE is a diagnostic or prognostic tool for patients with delirium.