Prospective Investigation of Markers of Elevated Delirium Risk (PRIMED Risk) study protocol: a prospective, observational cohort study investigating blood and cerebrospinal fluid biomarkers for delirium and cognitive dysfunction in older

Delirium is a common post-operative complication, Background: particularly in older adults undergoing major or emergency procedures. It is associated with increased length of intensive care and hospital stay, post-operative mortality and subsequent dementia risk. Current methods of predicting delirium incidence, duration and severity have limitations. Investigation of blood and cerebrospinal fluid (CSF) biomarkers linked to delirium may improve understanding of the underlying pathophysiology, particularly with regard to the extent this is shared or distinct with underlying dementia. Together, these have the potential for development of better risk stratification tools and perioperative interventions. 200 patients over the age of 70 scheduled for surgery with Methods: routine spinal anaesthetic will be recruited from UK hospitals. Their cognitive and functional baseline status will be assessed pre-operatively by telephone. Time-matched CSF and blood samples will be taken at the time of surgery and analysed for known biomarkers of neurodegeneration and neuroinflammation. Patients will be assessed daily for delirium until hospital 1,2 2,3 4 5

neuroinflammation. Patients will be assessed daily for delirium until hospital discharge and will have regular cognitive follow-up for two years. Primary outcomes will be change in modified Telephone Interview for Cognitive Status (TICS-m) score at 12 months and rate of change of TICS-m score. Delirium severity, duration and biomarker levels will be treated as exposures in a random effects linear regression models. PRIMED Risk has received regulatory approvals from Health Research Authority and London -South East Research Ethics Committee.
The main anticipated output from this study will be the Discussion: quantification of biomarkers of acute and chronic contributors to cognitive impairment after surgery. In addition, we aim to develop better risk prediction models for adverse cognitive outcomes.

Introduction
Delirium, a neuropsychiatric syndrome characterised by altered arousal, inattention and other cognitive deficits, is a common response to acute physiological stressors. The magnitude of the stress required to precipitate delirium is inversely proportional to the degree of prior brain vulnerability 1 . Delirium is therefore common in acutely ill older people, affecting at least 25% such inpatients at any time 2 , and will frequently complicate perioperative care 3 . The prevalence of postoperative delirium depends on type of surgery, particularly emergency procedures in populations with high frailty or comorbidities, but an overall postoperative incidence of 20% is a reasonable estimate 4 . Elective surgery carries a lower, but still significant risk (4% in a population-based cohort) 5 . Estimated prevalence varies substantially according to detection methods and diagnostic algorithms used 6 . Nonetheless, delirium is consistently associated with a range of adverse outcomes: increased length of stay, admission to critical care, institutionalisation and subsequent cognitive impairment and dementia risk 5,7,8 .
The pathophysiological mechanisms that underpin perioperative delirium and its sequelae have yet to be systematically studied. Certain pathways may account for the development of delirium itself, while others may drive longer-term cognitive outcomes such as progression of dementia 9 . The degree to which these interrelate, are separate or shared is largely unknown. One approach to investigate the neural basis of delirium has been to use cerebrospinal fluid (CSF) samples obtained through regional anaesthesia 10 .
Potential modifiable perioperative risk factors for delirium include depth of anaesthesia, intraoperative hypotension and cerebral oxygenation. These measures have been investigated separately and together in various surgical contexts but have not been extensively correlated with CSF biomarkers and longer-term neurocognitive outcomes 11 .

Aims
To understand mechanisms underlying perioperative delirium by assessing CSF and blood-based biomarkers of acute and chronic brain injury. Through pre-and post-operative follow-up, the cross-sectional and longitudinal significance of delirium and its biomarkers can be assessed. We will include intra-operative fronto-temporal processed EEG and functional near-infrared spectroscopy with perioperative delirium and subsequent cognitive decline. Together, we envisage these findings will lead to better ways of predicting delirium and long-term cognitive outcomes.

Recruit a cohort of individuals aged ≥70 years undergo-
ing elective and emergency surgery and where neuraxial anaesthesia/analgesia (spinal anaesthesia) is part of the anaesthetic technique.
2. Acquire paired CSF-blood samples to assess systemic inflammatory and neuroinflammatory markers of chronic neurodegeneration and acute brain injury.
3. Assess the impact of common neurophysiological insults in the perioperative period on the incidence of delirium and its sequalae.
4. Undertake serial measures of cognitive function, including delirium, with outcomes ascertained for at least two years.
PRIMED Risk will collaborate closely with investigators from Queen's University Belfast (QUB), which have access to an existing cohort of n=282 CSF samples collected from elective hip and knee replacements (PI: Cunningham) 12 . Our clinical assessment and CSF analysis protocols are closely aligned and we envisage replicating findings across both sets of samples where relevant.

Hypotheses
1. Biomarkers of acute and chronic brain injury are cumulatively associated with higher incidence of perioperative delirium.
2. Biomarkers of acute and chronic brain injury interact multiplicatively, associated with the fastest degree of cognitive decline.
3. Perioperative physiological stressors such as prolonged hypotension, excessive depth of anaesthesia and cerebral hypoxia are associated with clinical delirium and faster cognitive decline.

Recruitment
Adults aged ≥70 years undergoing surgery with planned spinal anaesthesia will be identified at the time of listing and approached either in person (e.g. at a pre-operative assessment clinic), or by letter. Emergency surgical patients will be identified from emergency theatre lists ( Figure 1). Patients will initially be recruited from University College London Hospitals (UCLH) and the Royal Free Hospital (RFH) but other sites may open through the NIHR Clinical Research Network. In addition to the experience of our QUB co-investigators, we have piloted the identification, consent and follow-up of potential participants across both sites using a range of approaches and have shown 25% recruitment to be feasible.
Planned patient groups include urological, colorectal, hepatic and gynae-oncological resections, major vascular surgery and emergency neck of femur fracture repair. Cardiothoracic and neurosurgical patients will be excluded because direct brain complications through routes other than delirium are more likely in these cases. Patients will also be excluded if they are aphasic or do not speak English sufficiently to undertake cognitive assessments, if they were not scheduled to receive spinal anaesthesia as standard of care, or were not expected to survive hospital stay.

Primary outcome
Modified Telephone Interview for Cognitive Status (TICS-m) score ascertained 12 months after index surgery 13,14 . Pre-operative assessment of baseline function Enrolled patients will be contacted by telephone prior to admission for surgery to complete a baseline neurocognitive and functional assessment. This includes assessment of general health, subjective memory complaints, hearing, vision, nutrition, continence, falls and mood. Pre-operative function will be assessed using the Barthel Index of Activities of Daily  . Emergency participants will also be recruited at admission. If such patients do not have delirium, the standard baseline assessment will be applied. For those with prevalent delirium, an additional assessment of the Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE) will be undertaken. This has shown to be a feasible method for obtaining a proxy for estimating a pre-existing dementia in patients with delirium 15 .
Peri-operative data collection Depth of anaesthesia (derived from processed fronto-temporal EEG) and cerebral oxygenation will be monitored with bispectral index (BIS) and near-infrared spectroscopy (NIRS) sensors respectively, attached to the patient prior to anaesthetic induction to monitor changes in these parameters during induction and intubation. Arterial catheters will routinely be used to monitor intra-operative and post-operative blood pressure.
Physiological parameters from the anaesthetic machine monitor will be contemporaneously recorded onto a research laptop at 1-minute intervals for the duration of the surgical procedure. UCLH uses an anaesthetic charting system integrated into the electronic health records system; additional physiological measurements will be extracted from these electronic anaesthetic records. Parameters recorded will include end-tidal carbon dioxide (etCO 2 ), minimum alveolar concentration (MAC) if using inhalational agents, invasive blood pressure, mean arterial pressure (MAP), cardiac output, arterial oxygen saturation (SpO 2 ), heart rate and core body temperature.
Case report forms will be used to collect patient information including demographics, body mass index, medical and drug history, functional capacity in metabolic equivalents (METs) and cardiopulmonary exercise test (CPET) data (where available as part of routine care), and baseline blood test results. Details of the surgical procedure and the delivered anaesthetic (including total intravenous anaesthesia doses if applicable) will also be captured on case report forms at the time of the operation.

Biological sample collection
Venous blood samples will be drawn from a newly inserted peripheral venous cannula into tubes for plasma (BD Vacutainer ® K2E (EDTA) 10.0mL) and serum (BD Vacutainer ® SST™ II Advance 8.5mL) prior to administration of anaesthetic or pre-medication agents. Samples will be centrifuged at 2000G for 10 minutes at 20°C, before dividing plasma and serum into 500µL aliquots and storing at -80°C for later analysis.
At the time of pre-operative spinal anaesthetic, up to 5mL of CSF will be collected into a sterile universal polypropylene container. It will be centrifuged at the same time as the blood samples to remove any cellular debris, and also stored in 500µL aliquots at -80°C for later analysis.
Post-operative assessments Patients will be assessed for delirium each weekday after surgery until hospital discharge using the Memorial Delirium Assessment Scale (MDAS), Observational Scale of Level of Arousal (OSLA), and Hierarchical Assessment of Balance and Mobility (HABAM).
Collateral history regarding cognition, mobility and sleep will be taken from a personal and/or healthcare professional informant.
Routinely collected physiological parameters (temperature, blood pressure, oxygen saturation, heart rate, respiratory rate) will be recorded, as will blood test results and any medication changes.
Post-operative morbidity will be recorded using the Post-Operative Morbidity Survey (POMS), a validated tool for quantifying post-surgical morbidity in 9 domains. This survey will be completed on post-operative days 1, 3, 5, 7, 10, 14 and subsequently every 7 days until hospital discharge. Post-surgical complications will be recorded and graded according to the Clavien-Dindo classification.
Long-term follow-up Following discharge from hospital, patients will be contacted by phone for repeat neurocognitive and functional assessment at regular intervals until two years after the procedure. These assessments will be the same as the pre-operative baseline questionnaire, including the TICS-m primary outcome measure.

Sample analysis
Samples will be analysed using established immunohistochemical methods for detection and quantification of biomarkers. Biomarkers will include, but are not limited to: Outcomes will be analysed using random-effects linear regression with TICS-m score measured two-monthly, adjusted for the exposures above. Between biomarker interactions will be estimated. Bonferroni correction will be performed to account for repeat biomarker testing. Stata 15.0 (StataCorp, Texas) will be used for all analyses.

Sample size calculation
The present literature on the CSF distribution of biomarkers in delirium is insufficient to provide clear estimates for power calculations. Therefore, we will estimate power for the whole study based on data from the QUB data. The optimal analysis would involve random effects ANOVA, where different distributions of biomarker profiles (factor 1) could be compared with delirium status (factor 2), where the random effects account for the inter-dependence of the biomarker levels. In the meantime, a conservative approach to sample size estimation would be to consider each category of biomarker as a binary parameter (present / absent). In this case, assuming α=0.05 with 4 degrees of freedom (e.g. inflammatory cytokines) and a 20% difference defining biomarker positivity, 298 individuals are needed to detect differences according to delirium status if β=0.8 or 395 individuals if β=0.9. A full statistical analysis plan will be published before the study is completed.
Though clinically relevant thresholds are not yet established for dementia, we are likely to be powered to detect even smaller effects given n=200 from PRIMED-risk with n=282 from QUB. Together, these representing the largest repository of perioperative delirium-CSF samples to date with unrivalled opportunities to assay a comprehensive biomarker battery with the power to detect subgroup effects (with/without dementia).

Dissemination of study results
On study completion, data will be accessible through the Dementias Platform UK repository. Study outputs will be disseminated through journal publications and conference presentations.

Study status
Study is currently open, with the first 20 (10%) participants recruited.

Discussion
The PRIMED Risk study represents an opportunity to understand the interacting effects of pre-morbid cognition, surgical and anaesthetic factors, and incident delirium on cognitive outcomes. It will also explore mechanisms underlying these relationships by assessing associated biomarkers of acute and chronic neuronal injury.
PRIMED Risk is related to the broader longitudinal Delirium and Population Health Informatics Cohort (DELPHIC) 16 . DELPHIC investigates the long-term impact of delirium by assessing cognition before, during and after hospitalisation in a whole population setting (urban sample in London, rural sample in Hereford). With identical clinical assessments, PRIMED Risk has the advantage of sampling from a wider geographical area while focusing on perioperative events and with more detailed biomarker evaluation than DELPHIC. Other ongoing studies with comparable methodological approaches include the SAGES and ASCRIBED cohorts 17,18 .
PRIMED Risk, together with existing data from QUB, sets out to address central questions on perioperative brain health. These concerns are of interest to a range of patient, hospital and public policy stakeholders. We anticipate our findings to contribute to local and national quality improvement work. Given the unmet need around delirium and older surgical patients, PRIMED Risk will systematically describe these and act as a driver for future service development. At the same time, we envisage novel contributions to the understanding of delirium and mechanisms underlying the range of cognitive impairments (including perioperative neurocognitive disorder) associated with surgical procedures 19 .

Data availability
No data are associated with this article postoperative Delirium].
The presented sample size calculation seems somehow arbitrary. We suggest to include a statistical review.

Minor concerns:
Here you present for the first time the recording of EEG and NIRS. Since you also present later on Aims: a Hypotheses including these measurements, please include this in the introduction and Abstract as well.

Methods:
You also wish to include Emergency participants, whereas your main patient cohort will be recruited by telephone call prior to admission. These different settings during neurocognitive setting might bias your data. It seems to me to be more reasonable to just focus on one patient group -hear patients you reach by telephone before admission.
Arterial catheters will be indicated in all patients?
Please indicate distinct time scales when you will undertake those "regular" follow-up telephone calls.
Sample analysis: here you present a bunch of possible biomarkers. Please give a more detailed overview within the introduction, about your hypotheses related to each biomarker you will examine.
Is the rationale for, and objectives of, the study clearly described? Yes

Are sufficient details of the methods provided to allow replication by others? No
Are the datasets clearly presented in a useable and accessible format? Yes No competing interests were disclosed.

Competing Interests:
We confirm that we have read this submission and believe that we have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however we have significant reservations, as outlined above.