Known unknowns: Examining the burden of neurocognitive impairment in the end-stage renal failure population

Improvements in life expectancy have driven a demographic transition and contributed to a rising prevalence of dementia worldwide. Most developed nations report a standardized population dementia rate of 5–8%.1 Despite a recognized minimum prevalence at least 3- to 5-fold that of the general population, the burden of neurocognitive disease in patients receiving maintenance dialysis is poorly understood. Clinically obvious cognitive deficits may impact dialysis provision across a number of aspects including behavioural concerns, reduced therapeutic engagement, ability to self-care, adherence to therapy and associated lifestyle restrictions. The consequence of impairment in these domains may impact the quality of dialysis delivered and ultimately, technique and patient survival. Less apparent but equally important effects of more subtle neurocognitive disease reflect insight and decision-making capacity, including that beyond the medical sphere, progressing toward loss of independence and transfer of the care burden to external carer supports. Timely recognition of cognitive deficits represents an opportunity for proactive patient-centred care, in particular, tailored communication and advanced care planning well before any progression into established dementia. Here, we review the current epidemiologic data, commonly applied screening methodologies and clinical implications associated with cognitive impairment in dialysis-dependent populations.

DIAGNOSTIC DEFINITIONS AND PATHOPHYSIOLOGY

Neurocognitive impairment (NCI) is a summative term covering persistent deficits from a range of potential cognitive domains occurring along a continuum of severity. A formal diagnosis of dementia traditionally sits at its most extreme. The latest iteration of the Diagnostic and Statistical Manual (DSM-V) has re-named the condition ‘major neurocognitive disorder’ (NCD),2 partly in an attempt to reduce the stigma burden of a dementia diagnosis and expand the category to disorders in those of younger age. Presently however, the term ‘dementia’ remains in common clinical and academic use. Dementia represents a persistent and progressive cognitive dysfunction from baseline characterized by deficits in one or more of the domains of learning and memory, language, complex attention, perceptual motor function, social cognition and executive function.

The most subtle and precursory form of NCI is Mild Cognitive Impairment (MCI) – a detectably reduced cortical function to a degree beyond ageing associated norms but short of fulfilling formal diagnostic criteria for dementia. While the course of MCI may be variable, including potential regression toward more normal cognitive testing outcomes over time, a diagnosis of MCI has been strongly associated with progression to dementia at a rate of 4–12% per year.3-6

In the general population, Alzheimer's type is the most common form of dementia accounting for approximately 60% of diagnoses, followed by vascular dementia in approximately 30%.7, 8 A singular diagnosis is, however, rarely straightforward due to substantial overlap in risk factors, clinical presentation and radiological findings between the two entities. There is general acknowledgement that the co-existence of either disease is both common and may adversely modify the typical trajectory of the other.9 Data from dialysis-dependent populations suggests a different pattern of epidemiology with substantially greater frequency of vascular dementia such that the prevalence rises to become equal to or greater than Alzheimer's diagnoses.10, 11

Observational studies have found independent associations between the presence of 25-hydroxyvitamin D deficiency, hyperparathyroidism, untreated anaemia, cardiovascular disease and prior stroke with dialysis-associated MCI.12-16 Associations with the parathyroid axis have been reinforced with postoperative improvements in cognitive and memory domains observed in patients undergoing surgery for secondary hyperparathyroidism,17 though only in small studies. It is hoped that prospective longitudinal studies such as the Brain IN Kidney disease (BRINK) cohort may provide deeper mechanistic insights in the future.18

At a structural level, paired brain MRI and cognitive testing in HD and non-dialysis dependent patients have shown significant reductions in grey matter volume, which correlates with lower cognitive testing scores, particularly in the executive domain.19, 20 The basis for excess grey matter atrophy and vascular-related NCI holds strong parallels with dialysis associated micro-circulatory threat to other organ systems. This cohort appears to represent a persistently vulnerable microcirculatory phenotype where the drivers of cumulative vascular disease extend beyond traditional atherothrombotic mechanisms of the general population to convey ‘renal-failure-specific’ risk.21 This may be exacerbated by the process of dialysis itself even in the setting of nominally ‘adequate’ renal replacement as described biochemically by small-solute clearance.

Non-traditional risk factors such as disordered calcium-phosphate metabolism, chronic volume overload, uraemia, oxidative stress, albuminuria, hyperuricaemia and sustained activation of the renin:angiotensin system are implicated in progression of vascular disease22-26 and may contribute to the progression of MCI. Cross-sectional analysis has shown the duration of dialysis therapy to be strongly associated with the severity of vascular disease,27 demonstrating that maintenance dialysis does little to ameliorate or reverse the evolution to a progressively more ‘abnormal’ cardiovascular state. Functional end-organ perfusion studies during haemodialysis (HD) have demonstrated that conventional HD treatment acutely creates an environment of repeated microcirculatory ischaemic injury, even in the absence of macroscopically demonstrable arteriopathy.28, 29 Regaining dialysis independence and (at least) partial restoration of physiology through renal transplantation may see domain-specific improvements in cognitive test scores, though typically not returning to levels matching general population controls.30

EPIDEMIOLOGY ACROSS DIALYSIS COHORTS

Against a background prevalence of approximately 5–8% in the general western population1 studies in dialysis cohorts tend to use point prevalence single-phase methodology, typically unadjusted for co-morbid disease or dialysis vintage, to indicate an approximately 3- to 5-fold increase in the frequency of NCI. Rather than an absolute association with dialysis this likely, in part, reflects a disease burden accumulated through the time-course of chronic kidney disease toward end-stage. In support of this hypothesis, linear increases in cross-sectional prevalence of NCI across falling levels of estimated GFR have been demonstrated.31, 32

A recent systematic review found evidence of significantly worse overall cognition in HD patients than the general population but was restricted in terms of further conclusions, largely due to marked inconsistency in cognitive assessment metrics.33 Unadjusted registry level analysis of diagnostic codes in the HD-DOPPS collective indicates a progressively incremental association with each decade in age beyond 59 years to a minimum prevalence of 18% by 90 years old.34 More targeted screening of prevalent HD cohorts in ambulatory and clinical trial environments has demonstrated a consistent relationship with age, however, much greater prevalence rates up to 30% in patients aged >65 years and just over half of patients dialysing beyond 85 years.

Deficits in the executive function domain (covering attention control, behavioural inhibition and working memory) appear as the dominant pattern of NCI.16, 35 Higher executive function scores are associated with lower mortality risk in a prevalent HD cohort followed for several years.36 It appears that both MMSE and executive function scores continue to drop off over time, a process hastened by advancing age, whereas the memory domain may stabilize or improve subtly with dialysis.36, 37 Stratification between a global reduction in cognition and specific detectable impairment in executive function preserves the relationship with advancing age but identifies the phenomenon much more frequently in younger HD patients such that those between 21 and 54 years have detectable NCI at a rate significantly greater than those over 65 years in the general population.38 Although data remain relatively sparse, small cohort studies in HD patients with ‘normal’ range MMSE scores >24 demonstrate significant reductions in executive function when compared to normative population data39, 40 indicating the incidence of subclinical NCI may be substantially higher than generally recognized.

THE IMPACT OF DIALYSIS MODALITY

Notwithstanding potential variation in dialysis prescription and delivery, the available data superficially suggests a slight excess of NCI in patients undergoing HD as opposed to peritoneal dialysis (PD). Consistent with extensive research comparing modalities across non-cognitive aspects, studies of NCI differences between modalities are confounded by reverse epidemiology, factors associated with initial modality selection, local transplantation practice and the nature of screening methodology applied. A retrospective USRDS and Medicare claims investigation found approximately half the cumulative incidence of ICD9 dementia-associated diagnostic codes in patients commencing PD as opposed to HD (1.0% Vs 2.7% and 3.9% Vs 7.3% by 1- and 3-years post-dialysis initiation, respectively).41 Through more specific screening in Hong-Kong patients commencing PD, the administration of the Montreal Cognitive Assessment (MOCA) tool during PD training identified NCI in 29%, with a 4-fold increase among those patients over 65 years. In addition to age, terminating education below secondary level and a history of hypertension or peripheral vascular disease was associated with the detection of NCI.42 Cross-sectional analysis comparing a formal neuropsychological test battery in English-speaking patients established on PD, HD or healthy volunteers found an odds ratio for NCI compared to control of 2.58 and 3.1 for PD and HD, respectively, against a control NCI rate significantly higher than commonly reported.43 Taiwanese national health-insurance analysis of clinician confirmed dementia diagnoses yielded a higher cumulative incidence amongst HD patients (crude HR 1.24 for HD vs. PD) across 8 years of followup. This modality effect was ameliorated following adjustment for age, sex, socioeconomic factor and urbanisation levels.44 In terms of prospective data, a single study has compared age and demographically matched HD and PD patients at two timepoints: dialysis initiation and 12 months post, finding a subtle improvement in cognitive function for the PD group.45 Mechanistic explanations for potential modality effect have not been studied and remain the subject of speculation.

SCREENING

The complexity of identifying a specific deficit across a spectrum of potential cognitive domains highlights the need for robust screening methodology, which appropriately examines all requisite areas. The most commonly applied tools are summarized in Table 1. Short-form diagnostic screens may be applied either in isolation or more effectively in combination to identify deficits across multiple areas of cognition. In the clinical dialysis context, the utility of screening tests are a balance of sensitivity and specificity against the time and specific resources required to administer. No specific screening tool covers all requisite cognitive domains nor have any been specifically validated in dialysis patients against a clinical diagnosis of NCI. Comparisons between studies have been confounded by the wide range of measures applied and the more recent observation that significant intra-individual variability may be seen in attention and executive function depending upon the timing of administration with respect to the dialysis cycle.46, 47

Despite a strong association with progressive age, screening strategies based on age alone may fail to capture a significant number of patients with detectable NCI, particularly executive function deficits in younger patients. Given the frequency of impairment in executive function in this population, screening strategies should incorporate specific coverage of this domain, implying a global screening strategy may be more appropriate if the intention is to intervene using a patient-centred supportive approach. There are no data or consensus to guide frequency of repeat screening in prevalent dialysis populations, though a pragmatic approach might be annual or semi-annual testing to balance natural history, service logistics, patient amenability and to mitigate survey fatigue and testing learning effect.

IMPLICATIONS AND PERSPECTIVES

In that the real prevalence of clinically significant NCI among patients receiving dialysis is likely underestimated as a function of systematic bias, the true magnitude of consequent implications for dialysis patients and services remains largely unrecognized. World Health Organization Global Burden of Disease estimates in the general population rank dementia in the top five contributors to disability adjusted life years burden and the second greatest contributor to years lived with disability in those aged over 60 years.48 The secondary impacts on family, carers and healthcare workers are particularly unknown, specifically at lower levels of NCI where impacts are difficult to quantify and the condition commonly unrecognised. The applicability of these observations to the co-morbid renal failure context is similarly challenging, though likely amplified where a chronic, time-consuming maintenance therapy and persistent illness state already see significantly reduced life expectancies with care requirements grossly higher than the general population.49, 50 Studies in non-renal failure clinical trial populations have observed those with executive dysfunction in particular are up to three times less likely to complete trial follow-up compared to patients with isolated memory or global cognitive deficits.51 Common manifestations of this phenomenon are typically subtle rather than outright resistive: cancelling appointments, multiple attempts to reschedule appointments and tendencies toward vague excuses52 – many of which are common behaviours observed in dialysis units. While a patient-centred approach to this problem frequent involves empathy, flexibility and tailored communication strategies the impact on adherence to therapy or clinical trial engagement are important.

In conventional centre-based HD cohorts, the presence of a dementia diagnosis has been significantly associated with the risk of hospitalization, all-cause mortality and forced decision-making around dialysis withdrawal.34, 53-55 Data from home-based dialysis therapies, where clinically obvious NCI is typically a barrier to entry and concerns regarding independence, technique and patient survival are magnified, is much more limited. A single 1-year prospective cohort study of 114 PD patients found those with confirmed NCI report significantly higher requirements for assistance with dialytic and day-to-day care, while demonstrating no difference in terms of PD-associated peritonitis rates.42 The risk of developing NCI once established on a home-based therapy, or the hypothesis that this dialysis choice might be in some way protective, is challenged by selection biases inherent to home-dialysis and has not been explored.

The natural trajectory of NCI in patients undergoing home dialysis overall is unknown, as is the relative interplay of dialysis modality and modification to dialysis prescription over time. In clinical trial patients undergoing HD, escalation of delivered dialysis dose yielded significant improvements in markers of small solute clearance and fluid volume control though there was no significant effect on the degree of NCI over 12 months as measured by 3MS or Trail-making B methodology.56 These data should be cautiously interpreted however as, in the absence of context and with a reasonable expectation of NCI progression drawn from general population studies,57 a relative lack of change might be speculatively inferred to represent a slowing in the rate of NCI decline.

Variation in reported prevalence, particularly the substantial under-reporting of coding record based analyses compared with formally screened patient samples indicates that NCI is neither well recognized nor documented in the provision of routine dialytic care. Without general awareness, validated screening strategies, acknowledgement of the benefits of diagnosis or resource for intervention, there appears no immediate catalyst for change. This would be consistent with known barriers to implementation of patient-centred practise in contemporary dialysis settings.58

In the absence of evidence-based and cost-effective treatments for dementia, consideration of the condition in terms of deficits facilitates opportunity for proactive resource engagement and support tailored to specific patient needs. Given the heterogeneity of presentations the clinical conceptualization of NCI might be more usefully inverted and considered in functional deficit terms rather than as a firm diagnostic label of itself. The detection of NCI in a dialysis patient might then be viewed optimistically as an opportunity to supportively enhance patient-centred care rather than through a prism of therapeutic nihilism.

Acknowledging that contemporary dementia management interventions neither provide cure nor significant likelihood of functional improvement, formalizing a diagnosis of NCI may aid nephrologists to better define goals of dialytic care and more suitably inform advanced care planning conversations. Clarity around specific cognitive deficits in particular, supports an evaluation of dialysis modality safety and may assist in-centre behavioural management day-to-day. If identified in the pre-dialysis phase, NCI represents an opportunity for discussion of nondialytic treatment, time-limited trials of dialysis with set timelines for ongoing re-evaluation59 or, in milder manifestations, assist supported decision-making, dialysis education and surrogate decision-maker engagement.

FUTURE RESEARCH PRIORITIES

While both common and a driver of adverse outcome, NCI in patients undergoing maintenance dialysis remains a deceptively unclear subject. Validation of appropriate screening tools for the dialysis environment and clear disease epidemiology specific to this population through uniform criteria are crucial first steps. Against this framework, a thorough understanding of the burden of NCI would encompass temporal variation in domain assessments across the longitudinal trajectory of time on dialysis. Given the statistical power and duration of observation required, there may be a role for registry-level contribution to achieve the scale necessary to build the knowledge foundation. Relationships between the effect of modality choice on the development and rate of progression and the reciprocal impact of NCI on the efficacy of dialysis delivered represent the next logical progression of study. Subsequent identification of potentially modifiable factors or interventions might allow arrest or reversal of decline. While presently the recognition of NCI in a dialysis patient tends to be opportunistic at the point of the obviously clinically apparent, the future is hopefully a path toward understanding and ultimately a drive to improvement in the dialysis experience for patients and caregivers.