Cerebrospinal Fluid Concentration of Neurogranin in Hip Fracture Patients with Delirium

INTRODUCTION

Delirium is a severe neuropsychiatric syndrome characterized by acute disturbances in attention, awa-reness, and cognition. It affects up to 50%of hospitalized older adults [1] and arises as a result of a medical condition or substance intoxication or withdrawal [2]. Cognitive impairment due to underlying neurodegenerative disorders (NDD) is a major risk factor [3, 4]. Serious deleterious outcomes are associated with delirium, including incident dementia and acceleration of existing cognitive symptoms and dementia [5–8]. The underlying pathophysiology of delirium is poorly understood. In recent years, biomarkers of importance in NDD have been explored in relation to delirium pathophysiology, suggesting a bidirectional relationship between delirium and NDD [9–11].

Regulation of synaptic signaling is essential for the coordinated relay of information in the brain. Red-uced synaptic density and efficacy of signal transmission have been linked to several NDD. For instance, in the early phases of Alzheimer’s disease (AD), cogni-tive impairment has been associated with hippocam-pal synaptic dysfunction, prior to definite neuronal cell death [12]. Jarquin-Valdivia and Major hypothesize that synaptic disruptions, and particularly changes in Hebbian Spike-timing-dependent plasticity, are key pathological etiologies in both delirium and neurodegenerative disease [13]. An experimental mouse study showed that increasing axonal and synaptic pathology were associated with a higher risk of acute cognitive impairment, as seen in delirium [14], but no further synaptic degeneration was observed following the episode of acute cognitive im-pairment. This suggests that synaptic degeneration may be involved in delirium, but that the pathophysio-logical mechanisms causing delirium do not aggrava-te synaptic degeneration. As proposed by Maldonado, neurobehavioral symptoms of delirium may be expl-ained by a temporary breakdown of functional integration between connected brain systems, resulting in pathological signaling and altered neurotransmitter homeostasis, which may further trigger neurotoxic signaling with ensuing neuronal apoptosis, leading to long-term cognitive symptoms [15].

Neurogranin is a postsynaptic protein which has a central role in long-term potentiation through regulation of calmodulin availability [16]. It is expressed mainly in neurons in the hippocampus, associative cortex, and amygdala, which are main brain areas affected by pathological changes in AD [17]. Studies have shown that neurogranin expression may be regulated through synaptic activity in hippocampal cell cultures [18] and decreases with age in mouse models [19]. At autopsy, neurogranin concentrations in the frontal cortex and hippocampus are lower in AD patients, likely reflecting reduced synaptic density [20]. In cerebrospinal fluid (CSF), neurogranin concentration increases from the early asymptomatic stages of AD [21–23] and predicts cognitive decline [24] and increased brain atrophy in early AD and mild cognitive impairment [23]. The increase may indicate ongoing synaptic loss and/or dysfunction with leakage to the CSF. The topographical distribution of neurogranin in the brain may explain that changes in neurogranin appear to be specific for AD [21] and Creutzfeldt-Jakob disease [25]. These areas include neuroanatomical structures that are likely involved in delirium [14]. Alternatively, there may be increased neurogranin release from AD-affected neurons, possibly in response to amyloid-β (Aβ) pathology, by similar mechanisms as proposed for the AD-specific increase in CSF phosphorylated tau (p-tau) and total tau (t-tau) protein concentration [26].

To our knowledge, the relationship between delirium and CSF neurogranin, as a biomarker of postsynaptic integrity, has never been studied. We hypothesized that increased levels of CSF neurogranin were associated with delirium in hip fracture patients (Fig. 1), either as a marker of the processes causing delirium and/or contributing to the patient’s vulnerability to delirium.

OPEN IN VIEWER

Fig. 1

Graphical abstract.

Our main study population consisted of demented and non-demented hip fracture patients with and without delirium. We chose this patient group be-cause the prevalence of delirium is high in these patients and CSF was obtained at the time of spinal anesthesia. We performed subgroup analysis based on dementia status and core AD biomarkers—since dementia is a main risk factor for delirium, and time of delirium onset—to better untangle the pathophysiological implications of a possible association between delirium and neurogranin. Two comparison groups were included: AD patients and CUA, to help dissociate changes in CSF neurogranin concentration due to delirium, from changes related to ageing and AD in the hip fracture population.

METHODS

Cohorts

Hip fracture cohort

Patients with hip fractures (n = 332) were enrolled in the Oslo Orthogeriatrics Trial, a randomized controlled trial evaluating the effect of orthogeriatric care on cognitive function, at Oslo University Hospital from September 2009 to January 2012, as described previously [27, 28]. Patients with terminal illness or high-energy trauma were excluded. The orthogeriatric intervention did not influence delirium incidence [27] and all participants were assembled in the present study. 130 participants had available CSF, of which two were excluded due to missing delirium status, yielding a final sample of 128 hip fracture patients (Fig. 2).

OPEN IN VIEWER

Fig. 2

Main cohort of hip fracture patients, control cohort of cognitively unimpaired older adults and comparison cohort of patients with Alzheimer’s disease. Cognitive status was assessed through ¹retrospective consensus (diagnosis of dementia based on available information at and after hospital admittance) or ² clinical examination including extensive psychometric test batteries.

The presence of delirium was assessed daily by trained investigators in all participants preoperatively and until the fifth postoperative day (all) or discharge (patients with delirium), using the Confusion Assessment Method (CAM) [29]. The study physician or study nurse scored CAM based on a 10- to 30-min interview with participants and information from relatives, nurses, and hospital records. Delirium status was defined as a binary variable (delirium/no delirium). The group without delirium consisted of patients who did not develop delirium at any time point during the study. In our main analysis (delirium versus no delirium), patients with subsyndromal delirium (SSD), were included in the no delirium group. SSD was defined as fulfilling at least two, but not all, required CAM criteria for the full syndrome. Within the group with delirium, participants were classified as having preoperative or incident delirium, depending on the time of delirium onset. Delirium severity was evaluated using The Memorial Delirium Assessment Scale (MDAS) [30].

One geriatrician and one old age psychiatrist independently evaluated whether participants met the ICD-10 criteria for dementia prior to the fracture, based on all prevailing data at baseline and 12-month follow-up (except delirium status during admission), including the Informant Questionnaire on Cognitive decline in the Elderly (IQCODE) and hospital records. The inter-rater consensus agreement upon the dementia diagnosis was acceptable (kappa 0.87) and disagreements were resolved through discussion.

Control group of cognitively unimpaired older adults (CUA)

The control group included 172 patients admitted for elective gynecological, orthopedic, or urological surgery in spinal anesthesia, aged 65 years or older the year of inclusion, who were recruited to the COGNORM-study from 2012–2013 at Oslo University Hospital and Diakonhjemmet Hospital, Oslo, as previously described [31]. Exclusion criteria were dementia, previous stroke with sequelae, Parkinson’s disease, and other acknowledged or suspected brain disease likely to influence cognition. Participants were assessed with cognitive tests prior to surgery to assure the absence of cognitive impairment, as described elsewhere [31]. Participants with a baseline Mini-Mental Status Examination score [32] of < 28 (n = 16) or suspected undiagnosed dementia (based on test scores and clinical data) with referral to a memory clinic by a geriatrician during six years of follow-up (n = 14) were excluded. Furthermore, 15 participants did not have available CSF samples. All patients were free from delirium at the time of CSF sampling, based on the cognitive tests prior to surgery. In addition, we examined case notes (all sections), using the chart method described by Inouye et al. [33], to confirm that no patients had developed delirium in the time from cognitive testing to the day of surgery (mean 11 days) to confirm that no patients had developed delirium in the time from cognitive testing to the day of surgery (mean 11 days). The final sample consisted of 127 CUA (Fig. 1).

Comparison group of patients with Alzheimer’s disease

The Norwegian Registry of Persons Assessed for Cognitive Symptoms (NorCog) is a consent-based national registry and contains clinical data for patients referred for examination of dementia in outpatient clinics [34]. The patients go through cognitive testing and tests of physical function, and blood tests and an MRI/CT of the brain are performed, as previously described [34, 35]. As a comparison group of patients with AD (Fig. 1), 46 patients enrolled in NorCog at Oslo University Hospital from 2009 to 2012, fulfilling the core clinical NIAA-criteria for probable anamnestic AD dementia [36] were eligible for analyses of neurogranin in the CSF. Cut-offs used for CSF AD-biomarkers were as follows: Aβ₄₂ < 700 pg/mL, p-tau > 80 pg/mL, and t-tau > 300 (age < 50 years), > 450 (50–70 years), and > 500 (> 70 years) pg/mL[35].

RESULTS

Hip fracture patients

Association between neurogranin and delirium/dementia status

Main analyses. No difference in CSF neurogranin concentration was found between patients with and without delirium (median [IQR] 201 [150,248] versus 197 [146,235]; p = 0.68, Table 1, Fig. 3). No correlation was detected between delirium severity and CSF neurogranin concentration (ρ= –0.075, p = 0.47). Adjusting for age did not alter the findings significantly.

OPEN IN VIEWER

Fig. 3

CSF neurogranin concentration in hip fracture patients with and without delirium. The black lines represent the median with the interquartile range. The p-value stems from Mann Whitney U analysis.

DISCUSSION

In contrast to our main hypothesis, we did not find that preoperative or incident delirium were associated with changes in CSF neurogranin concentration. With worsening neurodegenerative changes and cognitive impairment, the risk of delirium appears to gradually increase [14]. One might therefore expect that increasing CSF levels of neurogranin, as a measure of synaptic dysfunction and/or deterioration, might indicate an increased risk of delirium. In fact, a recent study found that blood neurogranin was elevated in critically ill patients prior to and at the time of delirium, compared to controls [43]. Importantly, neurogranin was measured in blood, and plasma neurogranin does not correlate with CSF neurogranin nor clinical outcomes in neither AD [44] nor acute stroke [45], likely due to extracerebral sources of neurogranin and proteolytic activity in blood [44]. Moreover, a study showed that the apical tree of CA1 neurons in aged mice was remodeled in response to acute stress such as during delirium [46]. The authors hypothesized that synaptic dysfunction in delirium may initially be adaptive. However, under pathological conditions with reduced synaptic plasticity, we advocate that the transient remodeling may become more permanent and lead to synaptic dysfunction and loss. Synaptic loss [47] and higher CSF neurogranin [48] have been shown to correlate with cognitive decline in early stages of AD. An association between delirium and neurogranin could thus suggest that synaptic loss caused by the mechanisms resulting in delirium might contribute to accelerated dementia and/or incident dementia after delirium.

However, our negative findings suggest that alt-hough neurogranin is expressed in neuroanatomical structures that are likely involved in delirium symptomatology [14], symptoms present in delirium are complex, and widespread cerebral dysfunction in other key areas may be more prominent. In addition, even though synaptic dysfunction with release of neurogranin may theoretically occur as a result of delirium, the rise in CSF neurogranin may be too insignificant to be registered in the hip fracture population, as discussed below.

Our second aim was to compare CSF neurogranin in the hip fracture population, which included patients with and without dementia, to neurogranin concentrations in CUA and AD patients (Fig. 1). CUA serve as a control group for both of the other cohorts. In contrary to acutely admitted hip fracture patients, they were thoroughly tested prior to CSF sampling and represent a group that with a high degree of certainty have neither dementia nor delirium. Increased concentrations of CSF neurogranin have been demonstrated repeatedly in AD [21–23]. Comparing CSF neurogranin concentrations in AD and hip fracture patients, particularly hip fracture patients with dementia, may help discriminate changes in CSF neurogranin due to delirium from changes related to AD. While the AD patients from the memory clinic underwent thorough clinical and biochemical testing confirming probable AD, the type of dementia was only registered for a minor subset of hip fracture patients. Other undetermined etiologies, such as cerebrovascular disease, might therefore have caused dementia in the hip fracture cohort. Accordingly, in agreement with previous studies, we found that AD patients had significantly higher levels of neurogranin than CUA and hip fracture patients [21–23]. Despite undetermined dementia etiologies, based on existing demographics concerning the prevalence of dementia subtypes in the oldest population [49], one might assume that AD or mixed pathology involving AD-specific changes were the leading etiologies also in the hip fracture patients [21–23]. We were therefore surprised to find that after adjusting for age, hip fracture patients had lower levels of neurogranin than CUA, and that this difference seemed to be driven by lower neurogranin in hip fracture patients with dementia. We suggest that neurogranin expression in the brain likely needs to be of a certain magnitude for concentrations of neurogranin to increase detectably in CSF. The hip fracture patients with dementia were in clinically advanced disease stages based on informant questionnaires (IQCODE) and scored significantly higher on IQCODE than the AD patients enrolled at the Memory Clinic. Advanced AD in the hip fracture patients may result in reduced neurogranin expression in the brain and/or neurogranin release to CSF due to reduced neurogranin production, loss of synapses and/or low disease intensity. In line with this, decreased levels of neurogranin have been detected at autopsy in cortical regions in the AD brain [20], with greater decreases in late stages of AD [49]. Furthermore, while levels of neurogranin have been shown to rise in early phases of AD-pathology [22], several studies have reported a negative correlation between AD duration and CSF levels of neurogranin [21, 48]. In our study, only hip fracture patients with delirium and normal levels of t-tau (N-) tended to have higher levels of neurogranin in delirium, supporting that underlying neurodegenerative changes could masque delirium-associated changes in synaptic function and neurogranin. Lastly, neurogranin expression has been shown to decrease with age in mouse models [19], which suggests that the release of neurogranin to the CSF may not be as prominent in synaptic dysfunction in the oldest old. The hip fracture patients were significantly older than participants in the two control groups (Table 1). Although we found a positive correlation between age and neurogranin in the CUA, we did not find any correlation in the hip fracture patients, possibly due to an age-related plateau effect. Taken together, CSF neurogranin may not be representative of the degree of synaptic dysfunction and/or degeneration in the hip fracture population, comprising the oldest old, including dementia patients likely to suffer in part from advanced AD [14, 46–50]. Limitations of our study include retrospective diagnosis of dementia with missing dementia etiology in most hip fracture patients. Characterization of dementia etiology is important as neurogranin appears to be AD specific. Analysis of CSF core AD biomarkers (Aβ₄₂, t-tau, and p-tau) at two different laboratories impeded direct comparison between cohorts. Although the hip fracture cohort was large in the setting of delirium research, lack of power may affect results, especially in subgroup analyses. The use of two comparison groups was a strength of our study. Also, neurogranin was analyzed at the same time at the same laboratory in all participants. Furthermore, delirium was assessed daily based on validated instruments by trained investigators.

No association was found between delirium and CSF neurogranin concentration in the hip fracture cohort. Studies on neurogranin as a biomarker for synaptic dysfunction in delirium pathophysiology should be repeated, possibly in a younger and cognitively healthier population.

AVAILABILITY OF DATA AND MATERIALS

Legal restrictions, imposed by the owners of the Norwegian Registry of Persons Assessed for Cognitive Symptoms and the ethical committee, prevent us from publicly sharing the de-identified dataset regarding the AD-patients due to sensitive patient in-formation. The clinical data may be requested at e-mail: post@aldringoghelse.no. However, data availability is dependent on approval from the REC South East, contact at e-mail: post@helseforskning.etikkom.no. The data that supports the findings in the hip fracture patients and the cognitively unimpaired control group are available from the corresponding author upon reasonable request. The data are not publicly available due to privacy or ethical restrictions.