INTRODUCTION
The prevalence of chronic kidney disease (CKD) has rapidly increased in the past decade, with current estimates suggesting that 1 in 10 people will develop the condition during their lifetimes (Chen & Harris, Reference Chen and Harris2015). Incidence is predicted to continue growing due to rising rates of related comorbid disorders which contribute to kidney failure such as obesity, hypertension, and diabetes (Tucker, Kingsley, Morton, Scanlan, & Dalbo, Reference Tucker, Kingsley, Morton, Scanlan and Dalbo2014). When kidney filtration rates are reduced, uremic toxins are not removed from the body and this build-up of toxins can negatively affect the central nervous system (Arnold, Issar, Krishnan, & Pussell, Reference Arnold, Issar, Krishnan and Pussell2016). This leads to a greater susceptibility to changes in cognitive functionality with decreased kidney function. Understanding cognitive dysfunction is critical, as it impacts overall functioning and quality of life in both mental (Fujino et al., Reference Fujino, Sumiyoshi, Sumiyoshi, Yasuda, Yamamori, Ohi and Imura2016; Tan, Thomas, & Rossell, Reference Tan, Thomas and Rossell2014) and physical disorders (Orbo et al., Reference Orbo, Aslaksen, Larsby, Schafer, Tande, Vangberg and Anke2015; Schuurs & Green, Reference Schuurs and Green2013), and specifically CKD (Seidel et al., Reference Seidel, Gronewold, Volsek, Todica, Kribben, Bruck and Hermann2014; Weiner & Seliger, Reference Weiner and Seliger2014).
Kidney disease is categorized into stages, which are determined by the degree of impaired kidney filtration rates. The most commonly used measure is the estimated glomerular filtration rate (eGFR), which is an estimate of how well the kidneys filter waste from the blood: early stage CKD consists of stages 1 (eGFR≥90 mL/min per 1.73 m2) and 2 (eGFR between 60 and 89 mL/min per 1.73 m2), moderate stage CKD consists of stages 3 (eGFR between 30 and 59 mL/min per 1.73 m2), and 4 (eGFR between 15 and 29 mL/min per 1.73 m2), and end stage CKD consists of stage 5 (eGFR of <15 mL/min per 1.73 m2) (Thomas, Kanso, & Sedor, Reference Thomas, Kanso and Sedor2008).
Cognitive Impairment in CKD
Cognitive impairment is defined as poorer cognitive function beyond that expected through normal aging and has been observed in both mental and physical disorders. It is a significant phenomenon in CKD specifically (Etgen, Chonchol, Forstl, & Sander, Reference Etgen, Chonchol, Forstl and Sander2012), affecting numerous patients and is exacerbated with worsening kidney function, independent of other confounding factors (Kurella Tamura et al., Reference Kurella Tamura, Wadley, Yaffe, McClure, Howard, Go and McClellan2008). Investigations of neurocognitive changes in end stage CKD have highlighted domains of cognition such as memory, concentration, and speed of processing (Berger et al., Reference Berger, Wu, Masson, Kelly, Duthie, Whiteley and Webster2016; Kaltsatou et al., Reference Kaltsatou, Grigoriou, Karatzaferi, Giannaki, Stefanidis and Sakkas2015). The vast majority of these studies have been conducted in older populations in line with the higher CKD prevalence and severity among older adults (Zhang & Rothenbacher, Reference Zhang and Rothenbacher2008); consequently, there is less of a focus on the earlier CKD stages.
This raises three issues for consideration. First, kidney disease is evident across the adult lifespan, affecting individuals from aged 18 onward. In a systematic review of studies totaling 65,181 adult participants, 5% of the population under the age of 30 had a CKD diagnosis, 7.2% were diagnosed in the range of 30–44 compared with 23.4% in those over 64 (Zhang & Rothenbacher, Reference Zhang and Rothenbacher2008). A common assertion is that kidney disease is relatively asymptomatic until the end stages (Levey et al., Reference Levey, Coresh, Balk, Kausz, Levin, Steffes and Eknoyan2003); however, there is evidence that deficits in executive and higher neurocognitive functions are already present in the earlier stages of renal impairment (Berger et al., Reference Berger, Wu, Masson, Kelly, Duthie, Whiteley and Webster2016; Sanchez-Roman et al., Reference Sanchez-Roman, Ostrosky-Solis, Morales-Buenrostro, Nogues-Vizcaino, Alberu and McClintock2011). Therefore, the characterization of neurocognitive change across the disease trajectory, particularly the profile of neurocognitive impairment in the earlier stages of CKD, is incomplete and requires better understanding.
The second issue is delineating the specific pattern of neurocognitive impairment in kidney disease. Studies to date have used a variety of different tasks to assess cognition in kidney disease with a focus on confirming neurocognitive impairment, but with less attention being paid to what aspects are impacted and at what disease-stage. A recent review (Berger et al., Reference Berger, Wu, Masson, Kelly, Duthie, Whiteley and Webster2016) has raised this issue, finding a range of neurocognitive deficits, with orientation, attention, and language being most impaired. A careful delineation of the neurocognitive profile in CKD, along with a mapping across illness stages, could help to advance the field, improve prognostic accuracy, and to identify areas of possible cognitive remediation.
Third, and arguably most importantly, the predominance of older populations in neurocognitive studies of CKD to date makes it difficult to determine whether reported findings of neurocognitive impairment are consequent upon the disease itself, or conflated with natural age-related cognitive changes. While cognitive change does vary by domain and between individuals across the life span, it is generally considered to significantly accelerate when an individual is in their 60s (Salthouse, Atkinson, & Berish, Reference Salthouse, Atkinson and Berish2003). Kidney filtration efficiency naturally falls with age and has been associated with impaired neurocognitive function (Khatri et al., Reference Khatri, Nickolas, Moon, Paik, Rundek, Elkind and Wright2009), and this would be exacerbated in healthy aging (Ziegler et al., Reference Ziegler, Piguet, Salat, Prince, Connally and Corkin2010). It is, therefore, important to understand how failing kidney function affects cognition before aging processes impact in order to characterize better the specific influence of CKD itself. Previous reviews in this field have not accounted for this (Berger et al., Reference Berger, Wu, Masson, Kelly, Duthie, Whiteley and Webster2016; Elias, Dore, & Davey, Reference Elias, Dore and Davey2013; Etgen et al., Reference Etgen, Chonchol, Forstl and Sander2012).
Presently, cognitive studies in elderly CKD populations have primarily revealed significant general cognitive deficits (Chang et al., Reference Chang, Lin, Tsai, Yang, Wang, Lin and Fuh2017; König et al., Reference König, Gollasch, Spira, Buchmann, Hopfenmüller, Steinhagen-Thiessen and Demuth2018; Romijn, van Marum, Emmelot-Vonk, Verhaar, & Koek, Reference Romijn, van Marum, Emmelot-Vonk, Verhaar and Koek2015). This is largely due to the wide use of the Mini-Mental State Examination (MMSE) in this group (Folstein, Folstein, & McHugh, Reference Folstein, Folstein and McHugh1975), which only provides a measure of general cognition. These deficits compound normal cognitive aging (Buchman et al., Reference Buchman, Tanne, Boyle, Shah, Leurgans and Bennett2009), are exacerbated with CKD progression (Romijn et al., Reference Romijn, van Marum, Emmelot-Vonk, Verhaar and Koek2015) and are related to an increased risk for mortality (Sharma et al., Reference Sharma, Yabes, Al Mawed, Wu, Stilley, Unruh and Jhamb2016) and comorbid conditions such as dementia (Bugnicourt, Godefroy, Chillon, Choukroun, & Massy, Reference Bugnicourt, Godefroy, Chillon, Choukroun and Massy2013; Weng et al., Reference Weng, Shu, Tang, Sheu, Tarng, Wu and Chuang2012).
CKD is also associated with other comorbid conditions, particularly for older individuals, such as stroke (Seliger, Gillen, Longstreth, Kestenbaum, & Stehman-Breen, Reference Seliger, Gillen, Longstreth, Kestenbaum and Stehman-Breen2003) and cardiovascular disease (Stenvinkel & Herzog, Reference Stenvinkel and Herzog2010). Risk for these comorbid conditions have been observed to rise with decreasing kidney function and these conditions have been invariably associated with cognitive deficits themselves. This adds to the complexity of specifying CKD-related cognitive dysfunction in older individuals.
To this end, this systematic review sought to address these three issues with the objectives of (1) specifying the areas of neurocognitive impairment in CKD; (2) characterizing neurocognitive impairment in CKD across all stages; and (3) examining neurocognitive impairment in CKD among individuals while minimizing the effects of age-related cognitive change (<65 years).
METHODS
Search Strategy
This review conforms to the criteria outlined in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) current statement (Moher, Liberati, Tetzlaff, Altman, & The PRISMA Group Reference Moher, Liberati, Tetzlaff and Altman2009). The electronic databases PubMed and Medline via Scopus were searched for articles up to December 2017 using the following Medical Subject Heading (MeSH) terms: “Chronic renal insufficiency”, “kidney diseases”, “renal disease”, “chronic kidney failure”, “cognition”, “cognitive dysfunction”, “executive function”, “executive dysfunction” in either the title, abstract or keywords. The initial search yielded 642 results. All relevant literature examining the association between CKD and cognitive task performance was considered in the search. A flow chart of the search strategy is presented in Figure 1. Within the final 11 studies, reference lists were searched manually to find any additional studies that were not found in the database output; this produced an additional 4 studies, making a total of 15 studies.
Fig. 1 Flowchart of search strategy, including reasons for exclusion and inclusion of studies.
Study Selection
Articles were included if they were peer-reviewed, published in English and included a clinical CKD sample undergoing hemodialysis who were not post-transplant. Articles were excluded based on demographic criteria if the samples had a history of stroke or neurological disorder, or consisted primarily of children, adolescents (<18 years) or the elderly (>65 years). However, studies comparing different age groups which encompassed our target age demographics were included, provided that the control groups were age-matched or age was controlled for in the analyses. In cases where only group mean ages were provided, studies were included if the mean age was below 65 years. Case studies, reviews, and studies without group comparisons were also excluded.
Abstracts were screened for relevance and any which focused on post transplantation or patients undergoing peritoneal dialysis were excluded as long-term kidney function would not be reliably represented via eGFR in such individuals. These conditions differ in their cognitive impact compared to haemodialysis, which is still the most common dialysis method. Studies where the connection between cognitive impairment and CKD were due to extraneous conditions and variables (e.g., anemia, hypertension, dementia) were also excluded. After reviewing abstracts, the full text of the remaining articles were reviewed.
RESULTS
Search Summary
The online search yielded 642 items. From the initial search six duplicate items were removed. The abstracts of the remaining 636 items were screened, and 581 were excluded, inter alia, due to irrelevant focus, incorrect demographics, non-human subjects, reviews and papers not written in English. Following this, 61 full-text articles were retrieved and assessed for eligibility. Among the full-text articles, 17 were inaccessible, 17 contained aging or pediatric populations, 9 did not focus on the association between the stages of CKD and cognitive impairment, 4 were reviews, 2 focused on peritoneal dialysis, and 1 was a study protocol. This resulted in 11 papers being included in the qualitative synthesis. A manual search of the reference lists from the 11 eligible studies yielded an additional 4 studies, totaling 15 articles included in the final qualitative synthesis. Two researchers (J.B. and E.J.T.) found 100% agreement for the studies that met inclusion criteria.
Study Characteristics
A descriptive summary of the articles included in this review is presented in Table 1. Four studies were conducted in North America, four in Africa, five in Asia, and one each in Europe and South America. In this review, 12 cross-sectional studies and one population based cohort design were included. Two studies did not specify their study design. Nine studies consisted of participants at more than one CKD stage.
Table 1 Summary of studies of cognitive change across different chronic kidney disease stages
Note. 3MS = Modified MMSE; CI = cognitive impairment; CKD = chronic kidney disease; COWA = Controlled Oral Word Association Test; CRPOK = Continuous Recognition Paradigm of Kimura; CVLT-II = California Verbal Learning Test – Second Edition; D-SPAN = Digit Span Test; D-SYM = Digit Symbol Test; ESKD = end stage kidney disease; HD = hemodialysis; MMSE = Mini Mental State Exam; NCT-A = Number Connection Test-A; RAVLT = Rey Auditory Verbal Learning Test; SDLT = Serial Digit Learning Tests; SDST = Serial Digit Learning Tests; SRTT = Simple Reaction Time Tests; TMT-B = Trail Making Test Part B.
In terms of comorbid conditions, seven studies (Dixit et al., Reference Dixit, Dhawan, Raizada, Yadav, Vaney and Kalra2013; Egbi, Ogunrin, & Oviasu, Reference Egbi, Ogunrin and Oviasu2015; Madan, Agarwal, Kalra, & Tandon, Reference Madan, Agarwal, Kalra and Tandon2007; Madan, Kalra, Agarwal, & Tandon, Reference Madan, Kalra, Agarwal and Tandon2007; Nasser, Shawki, El Shahawy, & Sany, Reference Nasser, Shawki, El Shahawy and Sany2012; Owolabi et al., Reference Owolabi, Abdu, Ibrahim, Owolabi, Nalado, Bappa and Taura2016; Pi et al., Reference Pi, Xu, Xu, Yang, Qu, Chen and Dong2016) excluded individuals with diabetes mellitus and/or cardiac risk or history. Five studies did not exclude these and reported rates of presentation (Hailpern, Melamed, Cohen, & Hostetter, Reference Hailpern, Melamed, Cohen and Hostetter2007; Owolabi et al., Reference Owolabi, Abdu, Ibrahim, Owolabi, Nalado, Bappa and Taura2016; Sanchez-Roman et al., Reference Sanchez-Roman, Ostrosky-Solis, Morales-Buenrostro, Nogues-Vizcaino, Alberu and McClintock2011; Thornton, Shapiro, Deria, Gelb, & Hill, Reference Thornton, Shapiro, Deria, Gelb and Hill2007; Tsai, Wang, & Fuh, Reference Tsai, Wang and Fuh2010), with three conducting analyses comparing them between the groups and all finding no significant differences. One other study reporting diabetes controlled for its presence in their analyses (Silverwood et al., Reference Silverwood, Richards, Pierce, Hardy, Sattar and Ferro2014). Several other studies (Bae & Park, Reference Bae and Park2008; Dixit et al., Reference Dixit, Dhawan, Raizada, Yadav, Vaney and Kalra2013; Egbi et al., Reference Egbi, Ogunrin and Oviasu2015; Madan, Agarwal, et al., Reference Madan, Agarwal, Kalra and Tandon2007; Madan, Kalra, et al., Reference Madan, Kalra, Agarwal and Tandon2007; Nasser et al., Reference Nasser, Shawki, El Shahawy and Sany2012; Owolabi et al., Reference Owolabi, Abdu, Ibrahim, Owolabi, Nalado, Bappa and Taura2016; Pi et al., Reference Pi, Xu, Xu, Yang, Qu, Chen and Dong2016; Sanchez-Roman et al., Reference Sanchez-Roman, Ostrosky-Solis, Morales-Buenrostro, Nogues-Vizcaino, Alberu and McClintock2011; Thornton et al., Reference Thornton, Shapiro, Deria, Gelb and Hill2007) also excluded participants with other comorbidities such as history of Parkinson’s disease and/or cerebrovascular diseases, hypertension, and head injuries. Two studies (Chen et al., Reference Chen, Qi, Kong, Wen, Liang, Zhang and Zhang2015; Gad, Ramzy, Abdelhamid, ElMassry, & Masoud, Reference Gad, Ramzy, Abdelhamid, ElMassry and Masoud2012) did not report any related conditions to CKD.
A total of 9304 participants were examined with a mean age varying from 31.8 to 63.24 years and an age range of 18 to 65 years. The results are discussed corresponding to the findings for each stage of CKD. Across studies, 46 individual cognitive tasks have been used, covering 13 domains of neurocognition (see Table 2). In presenting the results below, domains of difference are presented with the relevant cognitive tasks reported in brackets. Effect sizes of significant differences (Cohen’s d) were calculated and presented where permissible.
Table 2 Summary of the examined cognitive domains within the reviewed studies
Note. EF indicates domains that represent executive function cognitive processes.
Early Stage CKD (Stages 1–2)
Three studies examined patients with early stages of CKD. The third national health and nutrition examination survey (NHANES III) study (Hailpern et al., Reference Hailpern, Melamed, Cohen and Hostetter2007) included comparisons between an early and a moderate stage group, as defined by eGFR (mL/min per 1.73 m2) of ≥60 or <60. Early stage patients performed better than moderate stage patients (d = 0.25 to 0.9) in tasks associated with processing speed [Simple Reaction Time tests, Digit Symbol Substitution (D-SYM) and serial digit learning tests]. The second study investigated cognitive changes between early and moderate stage women, also defined by an eGFR (mL/min per 1.73 m2) of ≥60 or <60 (Tsai et al., Reference Tsai, Wang and Fuh2010).
A significant difference in scores (d = 0.31 to 0.52) was observed in verbal learning and working memory [the Rey Auditory Verbal Learning Test (RAVLT) delayed recall, backward digit span (DSB)] between early and moderate stage patients, with the latter performing worse. No significant differences were observed for long term verbal memory or processing speed (RAVLT recognition score, continuous recognition, categorical verbal fluency, Trail Making Test [TMT] A and B, forward digit span [DSF]). The final study examined cognitive function in stage 1–2 participants approaching retirement age (Silverwood et al., Reference Silverwood, Richards, Pierce, Hardy, Sattar and Ferro2014). A significant, negative, linear relationship (d = 0.14 to 0.15) was observed between cognitive performance and declining eGFR for verbal memory and processing speed (choice reaction time tasks).
Moderate Stage CKD (Stages 3–4)
Patients with moderate stage CKD were included in nine studies. Two of these studies have been discussed above (Hailpern et al., Reference Hailpern, Melamed, Cohen and Hostetter2007; Tsai et al., Reference Tsai, Wang and Fuh2010). Of the remaining seven studies, four compared the differences between two groups; a healthy cohort and a clinical group of patients with both stage 3 and 4 CKD inclusive (Gad et al., Reference Gad, Ramzy, Abdelhamid, ElMassry and Masoud2012; Nasser et al., Reference Nasser, Shawki, El Shahawy and Sany2012; Sanchez-Roman et al., Reference Sanchez-Roman, Ostrosky-Solis, Morales-Buenrostro, Nogues-Vizcaino, Alberu and McClintock2011), and a healthy cohort and a clinical group of patients with stage 3 to 5 CKD (Thornton et al., Reference Thornton, Shapiro, Deria, Gelb and Hill2007). Gad et al (Reference Gad, Ramzy, Abdelhamid, ElMassry and Masoud2012) found that a moderate CKD group performed significantly worse than the healthy control group (d = 2.44 to 3.45) in general cognition, processing speed, switching, logical and semantic memory, and initiation (MMSE, color TMT A and B, logical memory TMT A and B, COWAT).
In a separate study, Nasser et al. (Reference Nasser, Shawki, El Shahawy and Sany2012), reported moderate CKD participants to have significantly poorer scores (d = 3.34 to 6.33) than healthy controls in general cognition, processing speed, switching, and working memory (MMSE, TMT-B, D-SYM, and digit span). The third study (Thornton et al., Reference Thornton, Shapiro, Deria, Gelb and Hill2007) showed that moderate stage participants performed poorer on verbal learning, switching, and inhibition (California Verbal Learning Test - 2nd Edition (CVLT-II), TMT-B, Color-word Interference Test); however, only scores on the CVLT-II delayed recall trial were significantly impaired between groups (d = 0.55). Contrastingly, Sanchez-Roman et al. (Reference Sanchez-Roman, Ostrosky-Solis, Morales-Buenrostro, Nogues-Vizcaino, Alberu and McClintock2011) found that a moderate stage CKD group had no significant performance differences than a healthy cohort on attentional, executive function, and memory tasks.
Of the remaining studies that examined moderate CKD, two included a 3-group comparison of neuropsychological performance between stage 3, 4, and a control group (3-4-C) (Egbi et al., Reference Egbi, Ogunrin and Oviasu2015; Madan, Kalra, et al., Reference Madan, Kalra, Agarwal and Tandon2007). The first study found that there was no significant differences between groups in general cognition (MMSE) (Madan, Kalra, et al., Reference Madan, Kalra, Agarwal and Tandon2007). The second three-group comparison study used the six-item cognitive impairment test (6CIT), designed to examine cognitive impairment in memory, orientation and concentration (Egbi et al., Reference Egbi, Ogunrin and Oviasu2015). Six percent of the controls had cognitive impairment, compared to 24.0% of stage 3 patients (d = 0.08) and 41.6% of the stage 4 patients (d = 0.1).
End-Stage CKD (Stage 5)
Ten studies included in this review examined cognitive impairment in end stage CKD patients. Six of these studies examined an independent cohort of end stage CKD patients compared to a healthy control group (Bae & Park, Reference Bae and Park2008; Chen et al., Reference Chen, Qi, Kong, Wen, Liang, Zhang and Zhang2015; Dixit et al., Reference Dixit, Dhawan, Raizada, Yadav, Vaney and Kalra2013; Madan, Agarwal, et al., Reference Madan, Agarwal, Kalra and Tandon2007; Nasser et al., Reference Nasser, Shawki, El Shahawy and Sany2012; Owolabi et al., Reference Owolabi, Abdu, Ibrahim, Owolabi, Nalado, Bappa and Taura2016). A further three studies compared cognitive function across four separate groups, that is, healthy and stages 3, 4, and 5 (Egbi et al., Reference Egbi, Ogunrin and Oviasu2015; Madan, Kalra, et al., Reference Madan, Kalra, Agarwal and Tandon2007; Sanchez-Roman et al., Reference Sanchez-Roman, Ostrosky-Solis, Morales-Buenrostro, Nogues-Vizcaino, Alberu and McClintock2011). One study compared differences between an end stage CKD group on hemodialysis against a moderate end stage group who were not on dialysis (Pi et al., Reference Pi, Xu, Xu, Yang, Qu, Chen and Dong2016). The remaining study compared differences in cognitive function between a control group and an end stage CKD group, and has been discussed under the “moderate stage” subheading above (Nasser et al., Reference Nasser, Shawki, El Shahawy and Sany2012).
Three studies used only one measure of cognition to determine cognitive impairment. Two of the studies only used the MMSE as a measure of cognitive impairment, as the primary focus was electrophysiological monitoring (Madan, Agarwal, et al., Reference Madan, Agarwal, Kalra and Tandon2007; Madan, Kalra, et al., Reference Madan, Kalra, Agarwal and Tandon2007). In those studies, it was found that general cognition (MMSE) was reduced across all CKD stages compared to controls (Madan, Kalra, et al., Reference Madan, Kalra, Agarwal and Tandon2007). There was a minor reduction in MMSE scores between end stage CKD and control group (Madan, Agarwal, et al., Reference Madan, Agarwal, Kalra and Tandon2007), but these differences did not reach significance. The third study (Egbi et al., Reference Egbi, Ogunrin and Oviasu2015) found a significant difference (d = 0.26) in general cognition performance (6CIT) between the control group (6.00%), stage 3 (20.00%), 4 (33.71%), and end stage groups (46.15%).
Among the independent cohort studies that examined cognitive performance between end stage CKD and controls, the first (Chen et al., Reference Chen, Qi, Kong, Wen, Liang, Zhang and Zhang2015) found significant deficits in processing speed and visuospatial attention (Number Connection Test-A, D-SYM, Line-Tracing Test, Serial-Dotting Test) compared to the healthy control group (d = 0.56 to 1.58). The second study (Bae & Park, Reference Bae and Park2008) found that end stage CKD groups had significantly poorer processing speed and switching (TMT A and B) compared to controls (d = 2.04 to 2.64). The third study (Owolabi et al., Reference Owolabi, Abdu, Ibrahim, Owolabi, Nalado, Bappa and Taura2016) noted a significant reduction in performance for the end stage CKD group for overall auditory and visual processing speed, recognition memory, and attention (d = 0.88 to 2.46). The final study examined cognitive performance in end stage CKD patients before and after dialysis, compared to a control group (Dixit et al., Reference Dixit, Dhawan, Raizada, Yadav, Vaney and Kalra2013). For this review, only data in the pre-dialysis stage was considered, to account for the effects of hemodialysis on cognition, and to be consistent with the other reviewed studies. End stage CKD groups performed significantly poorer than controls for general cognition and attention [MMSE, Letter cancellation tasks (one and three letter), D-SYM] (d = 1.13 to 2.82).
In the study by Nasser et al. (Reference Nasser, Shawki, El Shahawy and Sany2012), comparing cognitive dysfunction between end stage CKD patients and controls, there was poorer cognitive performance between control and end stage CKD groups in general cognition, switching, and working memory (MMSE, TMT-B, D-SYM and digit span) (d = 2.26 to 3.83). Performance on these tasks was also significantly more impaired in the end stage CKD group compared to moderate CKD patients (d = 0.61 to 1.44), but with reduced effect size against the control comparison. Another study (Sanchez-Roman et al., Reference Sanchez-Roman, Ostrosky-Solis, Morales-Buenrostro, Nogues-Vizcaino, Alberu and McClintock2011) established significant differences between end stage CKD and healthy controls on measures of attention, executive function, and memory (d = 0.49 to 0.64). End stage CKD patients also had significantly poorer memory performance when compared to moderate stage CKD patients (d = 0.54).
The final study in this section contrasted the performance of end stage CKD patients against an un-dialyzed mixed stage 3–5 group (Pi et al., Reference Pi, Xu, Xu, Yang, Qu, Chen and Dong2016). The CKD group had significantly greater impairment compared to the stage 3–5 group on overall CI, executive function, immediate memory, delayed memory, and visuospatial performance (d = 0.41 to 1.17). No significant group differences were observed for language performance.
DISCUSSION
The current review examined cognitive changes across the lifespan in non-elderly adults with CKD, while minimizing the effects of natural age-related changes. The evidence suggests a significant deterioration in cognitive function across the illness course of CKD in line with existing conceptualizations. These observations are strengthened by a targeted focus on studies with a non-elderly adult population. From the 15 articles reviewed, significant differences were found between the patient and control groups in 13 studies (Bae & Park, Reference Bae and Park2008; Chen et al., Reference Chen, Qi, Kong, Wen, Liang, Zhang and Zhang2015; Dixit et al., Reference Dixit, Dhawan, Raizada, Yadav, Vaney and Kalra2013; Egbi et al., Reference Egbi, Ogunrin and Oviasu2015; Gad et al., Reference Gad, Ramzy, Abdelhamid, ElMassry and Masoud2012; Hailpern et al., Reference Hailpern, Melamed, Cohen and Hostetter2007; Nasser et al., Reference Nasser, Shawki, El Shahawy and Sany2012; Owolabi et al., Reference Owolabi, Abdu, Ibrahim, Owolabi, Nalado, Bappa and Taura2016; Pi et al., Reference Pi, Xu, Xu, Yang, Qu, Chen and Dong2016; Sanchez-Roman et al., Reference Sanchez-Roman, Ostrosky-Solis, Morales-Buenrostro, Nogues-Vizcaino, Alberu and McClintock2011; Silverwood et al., Reference Silverwood, Richards, Pierce, Hardy, Sattar and Ferro2014; Thornton et al., Reference Thornton, Shapiro, Deria, Gelb and Hill2007; Tsai et al., Reference Tsai, Wang and Fuh2010). Across all controlled studies, individuals with any stage of CKD scored lower on cognitive testing compared to controls. Overall cognitive performance diminished between each stage, from early to end. The pattern and extent of this deficit differed across cognitive domains.
Cognitive Impairment Across Disease Course
Early stage CKD
The progression from early to moderate CKD seems to be associated with significant decreases in speed of processing, response speed, attention, short-term memory, and set shifting ability. This is indicative that the bulk of cognitive deficits in the earlier stages of CKD comprise of the more basic abilities (e.g., speed of processing, attention). This is also aligned with findings for decreased mental sharpness as an early CKD symptom (Levey et al., Reference Levey, Coresh, Balk, Kausz, Levin, Steffes and Eknoyan2003), and contrasts with assumptions of early CKD being relatively asymptomatic. Over 90% of individuals with CKD are unaware that they have the condition (Australian Bureau of Statistics, 2013). Critically, evidence of a significant linear relationship between eGFR and cognitive performance (speed of processing, verbal memory), inclusive of early CKD stage participants (Silverwood et al., Reference Silverwood, Richards, Pierce, Hardy, Sattar and Ferro2014), improves our understanding of early symptoms or indicators of CKD. This has the potential to advance early detection and targeted treatments. Improved early detection has been shown to reduce the debilitation associated with declining kidney function by up to 50% (Johnson, Reference Johnson2004).
Moderate stage CKD
Moderate stage CKD patients were found to have significantly poorer speed of processing, verbal fluency, logical memory, recall memory, sequencing, short-term memory, orientation, and concentration compared to healthy comparison cohorts (Egbi et al., Reference Egbi, Ogunrin and Oviasu2015; Gad et al., Reference Gad, Ramzy, Abdelhamid, ElMassry and Masoud2012). Inhibition and switching deficits were also observed in this group, although these domains were not more severely impaired than in earlier stages of renal impairment (Thornton et al., Reference Thornton, Shapiro, Deria, Gelb and Hill2007). Cognitive performance was also substantially poorer in stage 4 patients compared to stage 3, with almost double the severity of memory and concentration deficits (Egbi et al., Reference Egbi, Ogunrin and Oviasu2015). These findings show that, as CKD progresses, memory and speed of processing continue to deteriorate. Additionally, difficulties in concentration and orientation begin to emerge. These latter abilities are considered more complex and higher-order in nature, demonstrating and confirming that cognitive faculties are degrading and the profile of impairment expands with poorer renal function.
End stage CKD
End stage CKD patients demonstrate significantly poorer attention, concentration, speed of processing, executive function, visuospatial abilities, delayed memory and immediate memory compared to controls (Bae & Park, Reference Bae and Park2008; Dixit et al., Reference Dixit, Dhawan, Raizada, Yadav, Vaney and Kalra2013; Nasser et al., Reference Nasser, Shawki, El Shahawy and Sany2012; Owolabi et al., Reference Owolabi, Abdu, Ibrahim, Owolabi, Nalado, Bappa and Taura2016; Pi et al., Reference Pi, Xu, Xu, Yang, Qu, Chen and Dong2016; Sanchez-Roman et al., Reference Sanchez-Roman, Ostrosky-Solis, Morales-Buenrostro, Nogues-Vizcaino, Alberu and McClintock2011). Additionally, most cognitive impairments that initially manifest in early or moderate stage CKD continued to exacerbate with progression into end stage CKD, with significant reductions noted in executive function, memory and global cognition measures (Nasser et al., Reference Nasser, Shawki, El Shahawy and Sany2012; Pi et al., Reference Pi, Xu, Xu, Yang, Qu, Chen and Dong2016). Not all cognitive domains are impaired to equal levels of severity, as language was found to be the least impacted area of cognitive impairment.
General cognition was poorer at stage 5 compared to stage 4, although this difference was less pronounced than between stage 3 and 4 (Egbi et al., Reference Egbi, Ogunrin and Oviasu2015). As found between the early and moderate stages, speed of processing and memory continued to decay with disease progression (Nasser et al., Reference Nasser, Shawki, El Shahawy and Sany2012), as do attention and executive function (Sanchez-Roman et al., Reference Sanchez-Roman, Ostrosky-Solis, Morales-Buenrostro, Nogues-Vizcaino, Alberu and McClintock2011) which further highlights the difference in impairment profile between CKD stages. With the recognized role of cognitive abilities in better functioning (Rispaud, Rose, & Kurtz, Reference Rispaud, Rose and Kurtz2016), this suggests potential pathways on intervention to improve patient quality of life, particularly at specific stages of the illness.
CKD-Related Versus Age-Related Cognitive Change
In the majority of previous work, the cognitive impairment observed in end stage CKD patients has potentially been confounded with natural age-related cognitive changes due to primarily elderly samples being assessed. The results of the current, more targeted, review are, however, aligned with those of previous reviews, that is, that individuals with CKD exhibit poorer cognitive performance compared to those without CKD (Etgen et al., Reference Etgen, Chonchol, Forstl and Sander2012), and that the extent of these differ between individual cognitive domains (Berger et al., Reference Berger, Wu, Masson, Kelly, Duthie, Whiteley and Webster2016).
Notably, not all previously identified cognitive deficits have been demonstrated in CKD samples under the age of 65. Language impairments were not observed in the studies reviewed here, suggesting that these may be conflated with age-related cognitive changes. That being said, language is a multi-faceted domain and further investigations of other language components are required to better understand this. Additionally, the severity of cognitive impairment observed here among younger patients could plausibly be exacerbated in patients above 65 years of age, although this has yet to be directly investigated. Consequently, more empirical work is needed to advance characterization of the cognitive effects of CKD.
Mechanisms of CKD-Related Cognitive Impairment
The mechanisms by which CKD leads to cognitive impairment are yet to be clearly defined, with a vast majority relating to a build-up of substances in the blood from inefficient kidney function and associated comorbid conditions. We briefly describe a few here and discuss how they might relate to the stages of impairment in CKD. Several vascular problems are associated with CKD, such as diabetes mellitus, hypertension, and cardiovascular disease (Schiffrin, Lipman, & Mann, Reference Schiffrin, Lipman and Mann2007), that may contribute to a higher risk for and severity of cognitive impairment. Poorer levels of sleep observed in CKD patients have also been implicated (Madero, Gul, & Sarnak, Reference Madero, Gul and Sarnak2008).
Cognitive impairment may also result from higher levels of inflammation, oxidative stress, anemia, and uremic toxins that rise with decreasing kidney function (McClellan et al., Reference McClellan, Aronoff, Bolton, Hood, Lorber, Tang and Leiserowitz2004; Oberg et al., Reference Oberg, McMenamin, Lucas, McMonagle, Morrow, Ikizler and Himmelfarb2004; Watanabe, Watanabe, & Nakayama, Reference Watanabe, Watanabe and Nakayama2014). With inflammation in particular, increased levels of interleukin-6, interleukin-1β, fibrinogen, and C-reactive protein in CKD patients have been associated with increased cognitive impairment (Kalsatou, Reference Kalsatou2016; Kurella Tamura et al., Reference Kurella Tamura, Tam, Vittinghoff, Raj, Sozio, Rosas and Yaffe2017). There is also evidence that the parathyroid hormone is linked to greater cognitive deficits (Roman et al., Reference Roman, Sosa, Pietrzak, Snyder, Thomas, Udelsman and Mayes2011).
The linear associations between these vascular and physiological problems and decreasing kidney function supports the progression of cognitive impairment observed in the stages of CKD. These could then plausibly be compounded by the subsequent emergence of the comorbid conditions, such as stroke and dementia, and lead to the typically most severe cognitive impairment profiles of later stage CKD (Berger et al., Reference Berger, Wu, Masson, Kelly, Duthie, Whiteley and Webster2016). As it stands, more specificity for these proposed mechanisms regarding the individual contributions to cognitive impairment in CKD is needed. This is critical given the recognized role of comorbidities such as cardiovascular infarctions and diabetes in poorer CKD outcomes (Anderson et al., Reference Anderson, Halter, Hazzard, Himmelfarb, Horne and Kaysen2009; Prichard, Reference Prichard2000). Improved understanding of these would help clarify the direct effects of kidney failure itself and its antecedents.
Additional studies of kidney failure in younger populations, accounting for general aging effects, would further specify these relationships. With greater clarity, remediation of these comorbid conditions thus potentially could improve prognoses for future CKD patients. Future studies should consider the collection of biological markers of the disease and comprehensive information on comorbidities alongside cognitive tests to permit such examinations and advance our collective understanding.
Measures of Cognitive Impairment
It is noteworthy that a variety of cognitive measures have been used in the extant literature, with little consistency across disease stages. Increased coordination and consistency in cognitive measures used would improve the generalizability and comparability of findings in the field. The MMSE was the only measure that has been used across all stages, displaying a minor continuous decrease, with end stage CKD groups reporting the lowest values (Madan, Kalra, et al., Reference Madan, Kalra, Agarwal and Tandon2007; Nasser et al., Reference Nasser, Shawki, El Shahawy and Sany2012). However, some inconsistencies were identified, and it should be noted that the MMSE also produced inconclusive results between different studies. Clinical participants in the study by Madan et al. (Reference Madan, Kalra, Agarwal and Tandon2007), although demonstrating significantly reduced performance compared to healthy controls, could still be considered within the normal cognitive range. Comparatively, Gad et al. (Reference Gad, Ramzy, Abdelhamid, ElMassry and Masoud2012) found that their moderate CKD group scores reflected mild cognitive impairment.
As a tool designed for broad use and not tailored specifically to CKD, the MMSE may have limited utility as a tool for assessing cognitive impairment in this population. The MMSE is less sensitive to variations in cognitive impairment levels, and so may have reduced efficacy for the early stages of CKD. In addition, it has no capacity to support the assessment of individual domains of cognitive function, an aspect of significant variation in CKD. The MMSE is also known to suffer from ceiling effects (Spencer et al., Reference Spencer, Wendell, Giggey, Katzel, Lefkowitz, Siegel and Waldstein2013). Reduced utility of the MMSE is evident in other medical cohorts as well (Ringdal et al., Reference Ringdal, Ringdal, Juliebo, Wyller, Hjermstad and Loge2011; Scazufca, Almeida, Vallada, Tasse, & Menezes, Reference Scazufca, Almeida, Vallada, Tasse and Menezes2009).
Alternative assessment tools that address some of these limitations include the widely recognized and established Repeatable Battery for the Assessment of Neurological Status (RBANS; Randolph, Tierney, Mohr, & Chase, Reference Randolph, Tierney, Mohr and Chase1998) and the Cambridge Neuropsychological Test Automated Battery [CANTAB (Cognitive assessment software], 2017]) that have more robust psychometric properties and permit the examination of multiple cognitive domains. The Montreal Cognitive Assessment (MoCA; Nasreddine et al., Reference Nasreddine, Phillips, Bédirian, Charbonneau, Whitehead, Collin and Chertkow2005) is another quick cognitive assessment tool that has demonstrated good utility in CKD populations (Tiffin-Richards et al., Reference Tiffin-Richards, Costa, Holschbach, Frank, Vassiliadou, Krüger and Reetz2014).
Despite the large number of measures used and cognitive domains investigated, the domain of executive function remains relatively less explored with only three studies thus far specifically investigating this. Executive functions are higher-order cognitive abilities, such as problem solving, planning and organization, that are more strongly related to real-world functioning (Cahn-Weiner, Boyle, & Malloy, Reference Cahn-Weiner, Boyle and Malloy2002). There is evidence that impairments in higher-order cognitive functions are related to deficits in more fundamental, or lower-order, cognitive abilities such as processing speed and attention (Neill & Rossell, Reference Neill and Rossell2013). As demonstrated in this review, these lower-order cognitive functions are impaired in CKD. It follows then that executive functions should be more thoroughly investigated in CKD to (1) extend the neurocognitive impairment profile of the disease, and (2) better understand how functional deficits manifest.
Limitations and Future Directions
This review carries several caveats. The number of included studies is small, largely due to the age-related exclusion criteria. However, accounting for age-related cognitive change is a necessary step to clarifying CKD-related cognitive change. Future studies should, therefore, adopt this criterion and expand the breadth of work to improve the characterization of CKD-related cognitive impairment. The mean age cutoff used here maintains the possibility that some individuals above 65 years old are in the included studies. However, it was considered practical for minimizing the impact of older age on cognitive performance while sufficiently maximizing the number of representative papers included. Non-English language and peritoneal dialysis studies were also excluded, although these resulted in just 16 omissions, far fewer than the 79 excluded for age.
The generalizability of some study findings here are limited by their design, such as when moderate and end stage CKD patients are grouped and considered together. The severity of cognitive impairment varies between moderate and end stage CKD patients, and future research should examine CKD groups independently to more effectively evaluate cognitive changes. The development of a consensus cognitive assessment approach would provide added benefit, particularly through the inclusion of at least two tasks per cognitive domain examined to improve the validity and sensitivity of assessment.
Furthermore, future studies should include comparisons against healthy control groups, particularly in studies of the earlier CKD stages to improve characterization and interpretability. This will assist in the recognition of cognitive symptoms that can be suitable as early markers for CKD diagnosis. Longitudinal studies that track individuals from the early stage to end stage CKD would also be beneficial for improving current understanding of CKD-related cognitive change.
The current review also included studies that controlled for comorbid conditions (e.g., diabetes, cardiovascular disease). Given high comorbidity rates for these conditions in CKD (Collins et al., Reference Collins, Li, Gilbertson, Liu, Chen and Herzog2003; Whaley-Connell et al., Reference Whaley-Connell, Sowers, McFarlane, Norris, Chen, Li and Collins2008), the reduced generalizability of the findings to more general CKD groups should be noted. It is, however, appropriate for our stated aim of examining how decreasing renal function specifically affects cognition.
CONCLUSIONS
This review is the first to examine the impact of CKD on cognition across the disease course while considering age-related cognitive changes. The evidence supports changes in cognitive impairment across the stages of CKD, independent of age, for both lower- and higher-order cognitive abilities. Speed of processing, response speed, attention and short-term memory, and set shifting abilities were the first cognitive domains to be affected in early stages, which are not entirely asymptomatic. Executive function, concentration, sequencing, orientation, working memory changes emerged in moderate stages, and verbal fluency, visuospatial abilities, and further memory impairment progressed in end stage CKD. Cognitive impairment in CKD is thus broad, transcends stages, and may be a useful marker, in conjunction with other measures, for early identification of CKD.
ACKNOWLEDGMENTS
The authors report no conflicts of interest in the present study. E.J.T. is funded by an NHMRC Peter Doherty Biomedical Fellowship (GNT1142424).
