Neurocognitive Deficits May Not Resolve Following Pediatric Kidney Transplantation

Abstract

Background

Pediatric chronic kidney disease (CKD) patients are at risk for cognitive deficits with worsening disease progression. A few, low-powered, cross-sectional studies suggest that cognitive deficits may improve following kidney transplantation. We sought to assess cognitive performance in relationship to kidney transplantation and CKD-specific medical variables in a sample of pediatric CKD patients who provided cross-sectional and longitudinal observations.

Methods

A retrospective chart review was conducted in patients who completed pre- and/or post-transplant neurocognitive testing at the University of Iowa from 2015–2021. Cognitive outcomes were investigated with developmentally appropriate, standardized measures. Mixed linear models estimated the impact of transplant status on cognitive function (z-scores). Subsequent post-hoc t-tests on change scores were limited to patients who had provided pre- and post-transplant assessments.

Results

38 patients underwent cognitive assessments: 10 had both pre- and post-transplant cognitive assessments, 11 had pre-transplant assessments only, and 17 had post-transplant data only. Post-transplant status was associated with significantly lower full-scale IQ and slower processing speed compared to pre-transplant status (estimate= −0.32, 95% confidence interval [CI]= −0.52: −0.12; estimate= −0.86, CI= −1.17: −0.55, respectively). Post-hoc analyses confirmed results from the mixed models (FSIQ change score= −0.34, 95% CI=−0.56: −0.12; processing speed change score= −0.98, CI= −1.28: −0.68). Finally, being ≥80 months old at transplant was associated with substantially lower FSIQ compared to being <80 months (estimate= −1.25, 95% CI= −1.94: −0.56).

Conclusions

Our results highlight the importance of monitoring cognitive function following pediatric kidney transplant and identify older transplant age as a risk factor for cognitive deficits.

Introduction

Over half of all children with congenital causes of chronic kidney disease (CKD) will eventually require kidney transplantation [1]. Cognitive deficits, ranging from mild to severe, are known to emerge in parallel to pediatric CKD progression – even prior to the need for dialysis or kidney transplantation – and are likely to worsen with the progression of disease severity and the duration of disease [2, 3]. There is variability in the full-scale intelligence quotient (FSIQ), with notable IQ deficits in advanced CKD and end-stage kidney disease (ESKD), accompanied by cognitive deficits in executive function (EF), memory, attention, and academic achievement [4–7]. While most of the medical sequelae associated with advanced CKD and/or dialysis (e.g., anemia, hypertension, and metabolic acidosis) are expected to resolve after kidney transplantation, the cognitive sequelae associated with CKD do not always resolve [8, 9]. Although typically subtle, these cognitive deficits may interfere with the patient’s ability to adhere to the complex medical regimens necessary for successful kidney transplantation, underscoring the importance of addressing cognitive sequelae in CKD.

There is a paucity of modern data speaking to the effect of pediatric kidney transplantation on cognition. This gap is due, in part, to the relatively small number of pediatric kidney transplant surgeries each year, with approximately 800 pediatric kidney transplants performed in any given year in the United States [10, 11]. The current literature on this topic is therefore hampered by small sample sizes from a single center, absence of data on the effect of transplant immunosuppression, and/or by the lack of longitudinal cognitive data comparing pre- and post-transplantation cognition (See Supplementary Table 1 for a summary of the relevant literature). Existing studies have produced conflicting results, with some work suggesting poor cognitive performance in pediatric CKD patients regardless of transplant status [5, 8, 9, 12, 13], whereas other studies indicate that cognitive function may improve – but does not fully normalize – among children who received a kidney transplant compared to those receiving dialysis [7, 14–16].

Given the identified gaps, we sought to 1) characterize cognitive performance in relationship to kidney transplantation, leveraging a mixed model framework to utilize all available data and estimate change in cognition pre- versus post-transplant; and (2) determine the relationship between cognition and disease-specific factors after kidney transplantation, including dialysis duration, age at transplantation, and immunosuppression regimen.

Materials and Methods

Study population

This study is a retrospective review of the available clinical data collected between 2015–2020 at our pediatric kidney transplant program. Inclusion criteria for potential study participants included: 1) first kidney transplant listing prior to age 18; 2) kidney transplantation performed at the University of Iowa Stead Family Children’s Hospital; and 3) available neurocognitive evaluation(s) performed as a standard of care by a clinical psychologist. Exclusion criteria included: 1) presence of previously diagnosed moderate to severe intellectual disability or diagnoses(s) which might adversely impact cognition (e.g., autism spectrum disorder, severe cognitive delays, and Trisomy 21); 2) a history of chronic and/or severe non-renal disease requiring treatment(s) which could negatively affect cognition (e.g., cancer/chemotherapy); 3) active treatment for kidney transplant rejection at the time of post-transplant neurocognitive assessment; and 4) age less than 3 years at the time of neurocognitive evaluation (see Figure 1 for a consort diagram).

Figure 1.

CONSORT diagram of sample.

The study was approved by the Institutional Review Board (IRB) at the University of Iowa (201906744). Consent was waived due to the retrospective nature of this study.

Demographic and disease-related variables

Patient characteristics (i.e., sex, race, age at evaluation and age at transplant) and relevant kidney disease variables were abstracted from electronic medical records. Disease-related variables included cumulative dialysis duration in days (date dialysis was initiated to kidney transplant), age at transplant expressed as a binary variable (<80 months vs. ≥80 months), disease duration at transplant in days (calculated from diagnosis to date of transplant), disease duration total in days (calculated as date of diagnosis to date of neurocognitive assessment), estimated glomerular filtration rate (eGFR, mL/min/1.73m²), prematurity (≥37 weeks gestational age vs. <37 weeks gestational age), disease type (non-glomerular, glomerular, “other” including unknown or genetic etiologies of ESKD) and cumulative exposure to glucocorticoids following transplantation expressed as a categorical variable (none, 1–30 days, >30 days). eGFR was calculated from routine clinic and laboratory data of height (cm) and serum creatinine (mg/dL) closest to the date of assessment using the modified Schwartz equation [17]. Note that the cutoff for the age-at-transplant categories were informed by the bimodal distribution of the data (Supplementary Figure 1).

Neurocognitive assessment

Neurocognitive assessment is offered and performed as part of the standard of care for children at the University of Iowa Stead Family Children’s Hospital prior to and following a kidney transplant. Pre-transplant neurocognitive assessment is recommended for all patients who are in the kidney transplant listing process. Following recipient transplant listing, a consultation/referral for neurocognitive assessment is placed with our pediatric psychology program and testing occurs on average within 3 months of this consult order. If kidney transplantation has not occurred within 1 year of listing, then consults placed as part of the listing process – including neurocognitive evaluation – would be updated as part of transplant center policy. Referral for post-transplant assessment typically occurs after 1-year post-kidney transplantation when the patient is medically stable.

Cognitive assessments were administered by a licensed clinical psychologist (T.W.) and included a broad range of cognitive domains, such as general intelligence, executive function, short-term memory, and attention. Data abstraction was limited to Wechsler Intelligence scales whereby age-based normative data (i.e., WPPSI-IV for ages 3 – 5 years old; WISC-IV or WISC-V for ages 7 – 15 years old; WAIS-IV for ages 16 years and older) was compiled for scaled subtest scores and FSIQ to obtain age-appropriate, standardized estimates of cognitive performance.

Given that subtest completion varied with participant age, an average scaled subtest category score was calculated for each participant to maximize use of available data and minimize variability in subtests administered between participants. Calculated domain average scores included verbal, nonverbal, short-term memory, and processing speed scores. The subtests included in each domain are listed in Supplementary Table 2. Subtest scores were not available for some patients if a patient was unable to complete testing. Reasons for incomplete testing could include patient- and/or clinic-based factors (for example: inattention or scheduling conflicts, respectively). Observations were excluded if the child did not complete enough subtests to calculate an average composite score. All scores were converted to z-scores (mean = 0 ± 1) using the psychometric conversion table to allow for interpretation data across measures and scoring systems.

Note that neurocognitive ability of children aged 3 years or younger was measured by the Bayley-2 or Bayley-3 at our center. Since there are no comparable scores between the Bayley and the Wechsler scales, observations from children <3 years were excluded from the analyses.

Statistical analysis

The first analytical aim was to determine the impact of kidney transplant status (pre- vs. post-transplant) on neurocognitive outcomes. Other relevant predictor variables were included in multivariable models if they were associated with neurocognitive outcomes at p<0.05 at the univariate level. Mixed models with random effects for participants were constructed to account for between-participant variability in number and timing of assessments. Corrections for multiple comparisons utilizing the false discovery rate (FDR) were performed for mixed linear models. Confirmatory post-hoc analyses were limited to individuals who had repeated assessments, where the change in score from pre- to post was compared to 0 using one-sample t-tests.

The second aim was to explore patient- and disease-related risk factors for poor neurocognitive performance in patients who had received a kidney transplant. Similar to aim 1, potentially relevant variables were first identified through univariate analyses. Variables that were associated with neurocognitive performance at p<0.05 were included in mixed linear models that included random effects for participants. As with the first aim, FDR was used to account for multiple comparisons in the multivariable models. All statistical analyses were conducted in R version 4.0.0.

Results

Sample

The final sample included 38 pediatric kidney transplant patients who provided 50 total observations (cognitive assessments). The pre- and post- assessment group included 10 individuals, of which 9 individuals provided two observations, and 1 individual provided three observations (21 observations total; Table 1). Of those who provided post-transplant data only, 16 completed one assessment, and 1 individual completed two post-transplant assessments (18 observations). In total, 11 participants completed one single assessment pre-transplant.

Table 1.

Individuals, assessments, and age at evaluation.

	Pre-Transplant Data Only	Post-Transplant Data Only	Pre- and Post-Transplant Data		Total
			Pre-Transplant	Post-Transplant
Unique participants: N	11	17	10		38
Number of assessments	11	18	11	10	50
Age at assessment, months: Mean (SD)	123.36 (72.14)	115.33 (58.92)	117.00 (33.89)	143.5 (40.18)

Demographics and relevant clinical information for the unique participants are summarized in Table 2. Of the 38 unique patients, 25 (65.8%) were male and 19 (50%) had developmental CKD due to congenital anomalies of the kidneys and urinary tract. Among those with post-transplant status and neurocognitive testing, median age at transplant was 46 months (range=12–190 months) and 24 (88.9%) required dialysis prior to transplantation. Two patients (5.3%) were diagnosed with ADHD prior to transplantation, and none were prescribed medication for ADHD prior to transplantation. For patients who provided pre- and post-transplant neurocognitive assessments, the median time between pre-transplant cognitive assessment and transplantation was 5.9 months (range 0.6 to 15.8 months). Furthermore, the median time since transplantation at post-transplant cognitive assessment was 29.6 months (range 0.5 to 160.0 months).

Table 2.

Sample demographics (unique individuals; baseline).

		Overall (n=38)
Sex, n (%)	Male	25 (65.8%)
	Female	13 (34.2%)
Race, n (%)	White	28 (73.7%)
	African American	2 (5.3%)
	Hispanic	3 (7.9%)
	Asian	2 (5.3%)
	Multiracial	3 (7.9%)
Gestational Age, n (%)	Term (>37 weeks)	23 (60.5%)
	Pre-term (<37 weeks)	15 (39.5%)
Disease Etiology, n (%)	Non-glomerular	19 (50.0%)
	Glomerular	11 (28.9%)
	Other	8 (21.1%)
* Age at transplant, months	Mean (SD)	75.8 (56.6)
	Median [Min, Max]	6.0 [12.0, 190]
* Dialysis Prior to Transplantation, n (%)	Yes	24 (88.9%)
	No	3 (11.1%)
* Total Dialysis Duration, days	Mean (SD)	42.6 (47.0)
	Median [Min, Max]	26.7 [0, 191]
Time Since Transplant, months	Pre-transplant Assessment to Transplant, Median [Min, Max]	5.9 [0.6, 15.8]
	Post-transplant to Assessment, Median [Min, Max]	29.6 [0.5, 160.0]
ADHD Diagnosis Prior to Transplant, n (%)	Yes	2 (5.3%)
	No	36 (94.7%)

^*Variable was calculated only for persons who had a transplant date and post-transplant cognitive assessment.

Transplant status, disease type, and neurocognitive outcomes

Participants had significantly lower z-scores relative to the normative mean across all outcome variables, regardless of transplant status (Supplementary Table 3). In univariate analyses, transplant status (pre- vs. post-) was associated with neurocognitive outcomes (all p<0.05; Supplementary Table 4), except verbal scores (p=0.809). Univariate analyses evaluating differences in performance by disease type approached significance for verbal scores (p=0.07) and short-term memory scores (p=0.007; Supplementary Table 4).

Following univariate model results, transplant status and disease type were included in subsequent multivariable models. Figure 2 shows the association between transplant status and neurocognitive function derived from the multivariable models. Transplant status (pre- vs. post-) was associated with neurocognitive function, whereby the post-transplant group scored significantly lower than the pre-transplant group on FSIQ (Estimate= −0.32, 95% CI −0.52: −0.12, FDR=0.0241) and processing speed (Estimate= −0.86, 95% CI −1.17: −0.55, FDR=0.00151) (Figure 2A). By contrast, transplant status was not associated with verbal score (Estimate= 0.08, 95% CI −0.19: 0.35), nonverbal score (Estimate= −0.24, 95% CI −0.53: 0.06), or short-term memory score (Estimate= −0.32, 95% CI −0.66: 0.02; Figure 2A). Our findings were confirmed in post-hoc analyses evaluating the change in scores from pre- to post-transplantation in individuals who had repeated assessments (Figure 2B): FSIQ, processing speed, and non-verbal scores decreased significantly from pre- to post-transplant (FSIQ change: −0.34, 95% CI −0.56: −0.12, p=0.00802; processing speed change: −0.98, 95% CI −1.28: −0.68, p<0.001; nonverbal change: −0.26, 95% CI −0.50: −0.02, p=.036). However, the mean changes in verbal scores (mean change: 0.06, 95% CI −0.35: 0.48) and short-term memory scores (mean change: −0.42, 95% CI −0.93: 0.10) were not significantly different from 0.

Figure 2.

Impact of transplant status on neurocognitive outcomes. Panel A shows the estimates for transplant status derived from multivariable, mixed linear models. Panel B scores the change score from 8 individuals with repeated assessments. The vertical line marks 0, i.e., no significant association. The number of observations per assessment domain are noted.

Supplementary Table 5 displays estimates for disease type derived from multivariable models. For verbal- and short-term memory scores, the ‘other’ disease group performed more poorly than the glomerular and non-glomerular groups (verbal scores: glomerular vs. other difference: 0.956, t_(31.2)=2.459, p=0.0197, FDR=0.10; short term memory scores: glomerular vs. other difference: 1.375, t_(32.8) =2.981, p=0.00538, FDR=0.03; Supplementary Table 5 and Supplementary Figure 2).

Risk factors for cognitive deficits post-transplant

Univariate analyses identified age at transplant as the only variable that was associated with cognitive performance among post-transplant patients (Supplementary Table 6). As shown in Figure 3, relative to those who were younger than 80 months at time of transplant, those who were 80 months of age or older at transplant had lower scores for FSIQ (Estimate= −1.25, 95% CI −1.94: −0.56, p=0.00177; FDR=0.009), verbal IQ (Estimate= −0.84, 95% CI −1.35: −0.33, p=0.00391; FDR=0.020), nonverbal IQ (Estimate= −0.89, 95% CI −1.53: −0.26, p=0.0110; FDR=0.055), short-term memory (Estimate= −0.79, 95% CI −1.55: −0.04, p=0.0511; FDR=0.255), and processing speed (Estimate= −1.16, 95% CI −1.87: −0.45, p=0.00447; FDR=0.022).

Figure 3.

Impact of age at transplant (<80 months vs. ≥80 months) on neurocognitive outcomes. The estimates (z-scores) were mixed linear models that were adjusted for random effects of participants. The vertical line marks 0, i.e., no significant association. The number of individual test observations for each domain are noted.

Discussion

We provide evidence that cognitive function may not normalize following pediatric kidney transplantation. Notably, we demonstrated significant reductions in FSIQ and processing speed following transplant. In our post-transplanted sample, impairments in processing speed evidently drove lower scores in FSIQ. Furthermore, older age at transplant (≥80 months) may increase the risk for sustained cognitive difficulties post-transplant. Conversely, glucocorticoid exposure, disease etiology, and/or dialysis duration appear to have limited impact on cognitive outcome in this modest-sized sample of pediatric patients.

Increased risk of cognitive impairment in children with CKD – even prior to dialysis or transplantation – has been well established [18]. Attention regulation and other executive functions appear to be particularly affected in pre-transplant CKD [18–22]. In line with this, our study provides evidence for striking deficits in processing speed. Processing speed is the ability to identify, discriminate, and integrate visual and verbal information for decision-making; thus, this construct is dependent on components of executive function, such as attention regulation, for optimal task performance [23, 24]. Previous work demonstrated that tasks of executive function are impaired in up to 35% of children with early, pre-dialytic CKD [20]. These deficits are more pronounced with lower kidney function (e.g., disease progression) [19, 20, 25], longer duration of kidney disease [19, 21, 22], metabolic acidosis [26, 27], and hypertension [28]. Executive functions are supported by areas of the brain that mature into early adulthood [29–31]; thus, it is conceivable that these areas may be vulnerable to the detrimental multisystem medical sequelae associated with compromised kidney function. In contrast to our findings, Mendley and Zelko (1999) performed longitudinal pre-/post-transplant cognitive assessment of nine children and found significant improvement in processing speed and attention after transplantation. This contrast may be attributable to differences in processing speed assessment and improvements in care for children with ESKD/transplant since the mid to late 1990s to current practice.

Much less is known about the persistence of cognitive deficits following pediatric kidney transplantation. Our current understanding is largely informed by studies that were published prior to the year 2010 and were hindered by small sample sizes with lack of longitudinal data (see Supplementary Table 1 for a numerical summary of pediatric kidney transplant studies on this topic). Using a mixed models approach, we are able to statistically extrapolate change in cognition relative to kidney transplantation. Our results suggest that cognitive deficits may not resolve following pediatric kidney transplantation and may even worsen despite successful kidney transplantation. There is MRI data to support the idea that the presence of pediatric CKD may confer risk for concomitant brain structural and white matter differences in comparison to healthy peers [32, 33]. Additionally, these neurodevelopmental differences are associated with age-related increases of neurofilament light chain (NfL), a plasma biomarker of neuroinjury, among pre-dialysis/pre-transplant pediatric CKD patients relative to healthy peers [34]. NfL is a marker of axonal health which is increased in the presence of hypertension, traumatic brain injury, and/or older adult age. Our previously reported data suggests that pediatric CKD patients may experience subclinical neuroaxonal injury well before development of advanced CKD/ESKD requiring kidney transplantation [34]. At present, no data exists to evaluate the trajectory of NfL after pediatric kidney transplantation; however, it is conceivable that transplantation may not fully resolve the neuroinjury profile present well before development of ESKD. Neuroinjury and neuroimaging data from children with pre-transplant CKD are consistent with the observed association between age at diagnosis and cognitive deficits in post-transplant patients. In our cohort, older age at transplant emerged as the single most important factor associated with persistent cognitive deficit following pediatric kidney transplant. This finding is consistent with historical reports in the literature [9, 14, 35, 36], and speculatively, this may support a protective effect of early transplantation on neurodevelopment (e.g., neuroplasticity).

Disease etiology was also associated with cognitive function following kidney transplant, although these effects were not as robust as those observed for age at transplant. Children with ESKD etiology due to “other” causes – such as unknown or genetic etiologies – generally perform more poorly compared to children with glomerular or non-glomerular causes. This may reflect a developmental predisposition to both kidney disease and associated neurodevelopmental risk among this subset of children requiring kidney transplantation. Future longitudinal studies should investigate whether children with “other” etiology are truly at increased risk. Furthermore, the inclusion of genetic testing within screening for neurocognitive assessments for pediatric transplant patients may provide valuable insight to increased genetic risk for cognitive deficits.

We also sought to explore patient- and disease-related risk factors for poor neurocognitive performance following transplantation. Published research suggests glucocorticoid exposure [37, 38] and longer pre-transplant dialysis and disease duration may be risk factors for cognitive deficits post-transplant [9, 14, 36]; however, we found no impact of post-transplant glucocorticoid exposure on cognition. Unfortunately, due to the limitations of sample size, we were not able to evaluate the effect of other immunosuppression regimens – such as calcineurin inhibitors - on cognitive performance; however, previously published data suggests that high doses of calcineurin inhibitors may confer a negative effect on the brain [40–42]. Further evaluation specific to this class of immunosuppressants is warranted. We similarly found no effect of dialysis duration or modality (e.g., hemodialysis vs peritoneal dialysis) on post-transplant cognition. Despite known associations of cognitive performance and socioeconomic status (SES) [43], we were unable to extract reliable covariates of SES from medical records due to the retrospective nature of our study.

Several limitations to this work warrant consideration. First, pediatric CKD is relatively rare [1]; as a result, the number of repeated assessments in this single-center sample is modest. Although this study is among one of the largest samples with pre- and post-transplant cognitive data within the last decade, the results of the study are more susceptible to greater variability due to the few children with repeated assessments (n=10). However, we addressed this limitation by leveraging robust statistical methods to account for variability in timing and number of assessments, allowing us to make meaningful inferences about the impact of transplant status on neurocognitive outcomes. Furthermore, this retrospective cohort study was conducted at a single institution and therefore, only represents one institution’s experience with neurocognitive assessments in pediatric kidney transplant. Multi-center studies are likely required to mitigate issues regarding sample size and selection bias. Second, participants may have only pre- or post-transplant cognitive assessment(s) for a variety of reasons. Reasons for pre-transplant only assessment included family declining further testing and scheduling conflicts with the COVID-19 pandemic. Similarly, reasons for post-transplant assessment only included young age at transplant (<3 years of age) which required a different neurocognitive assessment battery that is not directly comparable to the Wechsler measures, transfer to our pediatric nephrology clinic following transplantation, and new or continued cognitive concerns which required follow-up. Third, our study may also be limited by potential observation bias, since the clinician who administrated the cognitive tests was not blind to the patients’ health history. This bias was mitigated by administration of standardized evaluations. Finally, some post-transplant recipients may have had multiple post-transplant assessments out of response to parental and/or academic concerns, which could have biased our findings. This concern was addressed using post-hoc analyses, which were limited to patients with repeated observations.

Despite these limitations, this work provides the largest contemporary and comprehensive evaluation of cognitive outcomes in relationship to pediatric kidney transplant. A key implication of our study is that cognitive function may not fully normalize following pediatric kidney transplant. Indeed, performance appeared to decrease in several cognitive domains among pediatric CKD patients after kidney transplantation, particularly among children who were older at the time of transplant. Subtle neurocognitive difficulties are known to impede academic success and future employment opportunities as children age, as well as negatively impact quality of life [44], underscoring the importance of addressing neurocognitive sequelae of pediatric CKD. The findings of this study emphasize that the relationship between kidney function and cognition is complex. Previous work suggests that several medical variables appear to impact cognition among children with CKD/ESKD (e.g., hypertension, metabolic acidosis, and anemia) [19, 27, 28]. We did not observe any effect of these CKD-associated sequelae within our sample; however, this may speak to the limitations of a modest, single-center study.

Future research on this topic is required with attention to larger, longitudinal studies to more fully evaluate the impact and severity of such medical variables on cognition, especially in relationship to transplantation. Perhaps most notably, it is critical to understand whether the observed cognitive deficits predict risk for nonadherence to post-transplantation treatment regimens – such as immunosuppression –with the goal of improving allograft survival through maximizing patients’ adherence to their medication instructions. The complex relationship that may exist between medication adherence and possible persistent cognitive deficits should be considered by current and future physicians as children transition into adult clinics. Longitudinal studies examining cognition relative to other types of solid organ transplants should be considered for future research to determine whether non-normalization of cognition post-transplant is unique to kidney transplantation. Lastly, although standardized cognitive assessment provides an understanding of potential cognitive deficits following transplantation, quantitative neuroimaging research in combination with cognitive assessment is required to elucidate the neurobiological underpinnings for the cognitive deficits experienced by pediatric kidney transplant recipients.

Abbreviations:

CKD

chronic kidney disease

CAKUT

congenital anomalies of the kidneys and urinary tract

FSIQ

full-scale IQ

ESKD

end-stage kidney disease

eGFR

estimated glomerular filtration rate

WPPSI-IV

Wechsler Preschool and Primary Scale of Intelligence, Fourth Edition

WISC

Wechsler Intelligence Scale for Children, Fourth and Fifth Editions

WAIS

Wechsler Adult Intelligence Scale, Fourth Edition

CI

confidence interval

Data availability statement

Data may be shared upon reasonable request.