Predictors of cognitive dysfunction in hereditary transthyretin amyloidosis with liver transplant

Abstract Background Cognitive dysfunction is part of the broad spectrum of clinical manifestations in older untreated hereditary transthyretin amyloidosis patients with peripheral polyneuropathy. Objective The objective of this study is to systematically explore cognitive dysfunction in ATTRV30M amyloidosis patients whose disease course was modified by liver transplant (LT). Methods A series of 269 carriers of TTRVal30Met mutation treated with LT underwent a neuropsychological assessment. Clinical charts were reviewed to identify focal neurological episodes (FNEs), cognitive complaints and laboratory results. Chi-square and Mann–Whitney tests explored potential predictors of cognitive dysfunction. Results Cognitive dysfunction was identified in 35 patients (13%)−14 (5%) had mild and 21 (8%) had moderate dysfunction. In comparison to normal cognition, both mild and moderate cognitive dysfunction patients had older age, higher mPND score and elevated NT-proBNP and Cystatin C values. Mild cognitive dysfunction was associated with longer disease duration and history of FNEs, whereas moderate dysfunction was related to older age at disease onset and more cognitive complaints and depression symptoms. Conclusions Consistent with the natural history of the disease, older age and higher severity of the disease are significantly associated and potentially predictors of cognitive dysfunction in ATTRV30M patients treated with LT. The level of cognitive dysfunction may depend on some clinical variables.


Introduction
The transthyretin (TTR)-related amyloidoses are a group of clinically heterogeneous diseases. One of the main predominant phenotypes is a sensorimotor and autonomic neuropathy, hereditary ATTR (ATTRv) amyloidosis, frequently referred to as familial amyloid polyneuropathy (FAP). The two non-neuropathic forms are the cardiac amyloidoses (hereditary amyloid cardiomyopathy and wild-type cardiac ATTR amyloidosis) and the leptomeningeal amyloidosis [1]. These phenotypes are associated with specific TTR pathogenic variants, though there is considerable overlap [2]. ATTRv amyloidosis was first recognized by Corino de Andrade, in Portugal [3]. TTRVal30Met is the most common mutation associated with ATTRv amyloidosis. It typically presents as a progressive, axonal length dependent, sensorimotor and autonomic neuropathy with mixed involvement of other organs such as the eyes, the heart, the kidneys and/or the central nervous system (CNS) [4]. However, clinical manifestations of ATTRv amyloidosis may vary with age at disease onset and geographic origin [5].
In the past decade, CNS clinical manifestations [ie focal neurological episodes (FNE) and cognitive dysfunction] have been increasingly recognized in treated and untreated TTRVal30Met patients [6][7][8][9][10][11][12][13][14]. FNE (ie transient ischaemic attack-like episodes, stroke, aura-like episodes and epileptic seizures) have been documented in ATTRV30M patients treated with liver transplant (LT). These FNE are associated with longer disease duration [7,9,14]. Data from Portuguese patients also suggest that FNEs are more frequent in males and in patients with renal dysfunction [7]. Moderate cognitive dysfunction has been found in ATTRV30M untreated patients with older age (ie !50 years) and renal dysfunction, with minimal effects of longer disease duration and higher polyneuropathy disability [12]. The current knowledge regarding cognitive dysfunction and its predictors in ATTRV30M patients treated with LT is rather scarce.
This study aimed to explore higher brain functions in a large cohort of ATTRV30M amyloidosis patients treated with LT and to identify demographic and potential clinical predictors of cognitive dysfunction.

Materials and methods
Subjects ATTRV30M patients treated with LT and followed at Unidade Corino de Andrade (Centro Hospitalar Universit ario do Porto, Portugal), without other CNS disorders or serious chronic or acute medical illnesses that could interfere with cognition or mood, with at least 4 years of education, and who were willing to undergo a neuropsychological assessment were considered eligible to participate in the study. Eligible patients were included consecutively between 2013 and 2020. Two hundred and seventy-nine patients (which corresponds to 60% of the total number of ATTRV30M patients treated with LT and followed at Unidade Corino de Andrade during that time period) participated in the study. Ten patients were a posteriori excluded from the analyses because their cognitive assessment protocol was incomplete. The final study sample consisted of 269 ATTRV30M patients treated with LT (date of transplantation between 1993 and 2018). All participants provided their written informed consent before entering the study, which was approved by the institutional review board and was in accordance with the Declaration of Helsinki (ref. 123/CES/06).

Neurological assessment
All participants underwent a full neurological examination contemporaneous to the neuropsychological assessment during their regular follow-up visits. The modified polyneuropathy disability score (mPND score) was calculated. mPND score was converted into a numeric scale (ie 0-0, 1-1, 2-2, 3a-3, 3 b-3.5 and 4-4). Onset of the disease was defined by a neurologist specialized in the disease, as the occurrence of the first progressive characteristic symptoms, coincident with an abnormal neurological or neurophysiological observation.

Neuropsychological assessment
To determine the cognitive status of each patient, a battery of neuropsychological tests, previously standardized to the Portuguese population, were applied to explore overall cognitive functioning and three cognitive domains, namely, memory, language and attention/executive functions. The protocol included the Dementia Rating Scale-2 (DRS-2), the Auditory Verbal Learning Test (AVLT), the Verbal Fluency Test (Semantic and Phonemic) and the Trail Making Test (TMT; Part A and Part B). Test results were adjusted to the patients' sex, age and/or education according to the existing regression-based norms. Performance on each measure was then classified as normal (ie >10th percentile of the norms) or deficient (ie 5th-10th percentile or <5th percentile of the norms). Whenever a patient was unable to complete TMT Part B within the time limit or exceeded the limit of errors of the test, the performance was classified as <5th percentile. Patients were considered to have cognitive dysfunction when performance was classified as deficient in DRS-2 and/ or in !2/3 cognitive domains. Depending on the level of deficit, cognitive dysfunction was classified as mild (between the 5th and the 10th percentiles of the norms) or moderate (<5th percentile of the norms).
The Hospital Anxiety and Depression Scale was also applied to screen for anxiety and depression. For both subscales, the cut-off was !11.

Review of clinical and laboratorial records
Patients' clinical records were reviewed to identify history of post-LT FNEs attributed to CNS dysfunction and cognitive complaints, potential reversible causes of dementia (eg hypothyroidism, vitamin B12 deficiency and syphilis) and analytical results (ie N terminal pro brain natriuretic peptide -NT-proBNP and Cystatin C) contemporaneous to the neurological and neuropsychological examinations. The reference values used to classify abnormal NT-proBNP and Cystatin C values were respectively !227 pg/ml and !1.11 mg/l. Cardiac exams (i.e. electrocardiogram and/or Holter) and neuroimaging exams [ie brain computed tomography (CT) or magnetic resonance imaging (MRI)] of patients with cognitive dysfunction contemporaneous to the neurological and neuropsychological assessment were reviewed whenever available.

Statistical analysis
Descriptive statistics were used to summarize demographic and clinical characteristics. Chi-square (or Fisher's exact, when appropriate) and Mann-Whitney tests were applied for group comparisons.

Sample characteristics
A total of 269 patients were included in the study (Table 1). Fifty-five percent were male and the median (minimum-maximum): age was 45 years old (28-75), education was 7 years (4-21), age at disease onset was 30 years old (19-57), disease duration was 14 years (4-30), disease duration post-LT duration was 10 years (0-40) and mPND score was 2 (0-4). Atrial flutter or atrial fibrillation was identified in 13 patients (5.2%). Thirty-five patients (13%) had cognitive dysfunction and the remaining 234 patients (87%) had normal cognition. The cognitive dysfunction group was divided into two subgroups according to the level of dysfunction: 14 (5%) had mild cognitive dysfunction and 21 (8%) had moderate cognitive dysfunction. The frequency of deficits on each measure per each cognitive status group is presented in Supplementary Table S1.
Of the 35 patients with cognitive dysfunction, 22 (ie 8 mild and 14 moderate) had contemporaneous neuroimaging exams (ie 21 CT and 1 MRI). Seven patients (ie 3 mild and 4 moderate) with cognitive dysfunction had normal scans (31.8%). One patient with moderate cognitive dysfunction had leptomeningeal abnormal reinforcement (on the MRI scan), characteristic of cerebral amyloidosis angiopathy.
Other non-specific imaging abnormalities were identified in CT scans: enlargement of brain cerebrospinal fluid spaces suggestive of cerebral atrophy in 8 patients (ie 4 mild and 4 moderate), subcortical and periventricular white matter hypodensities and/or lacunar infarcts in 6 patients (ie 1 mild and 5 moderate) and/or hyperdensities suggestive of vascular calcifications in 2 patients (both moderate). None had territorial infarcts or lobar haemorrhages.

Late disease onset
Among the subgroup of patients with !50 years at disease onset (n ¼ 9), 4/9 patients had moderate cognitive Data are presented as frequencies (%) and medians (25th-75th percentiles). Chi-square (or Fisher's exact) and Mann-Whitney tests were used for group comparisons.

Focal neurological episodes
As presented in Table 3 Among patients with cognitive dysfunction, history of paroxystic atrial fibrillation and atrial flutter was identified respectively in one mild and two moderate cognitive dysfunction patients. These three patients were between 51 and 59 years old at the time of the neuropsychological assessment.

Discussion
The neuropsychological evaluation of 269 ATTRV30M patients treated with LT confirmed that cognitive dysfunction is part of the broad spectrum of clinical manifestations in hereditary TTR amyloidosis with peripheral polyneuropathy. The frequency of cognitive dysfunction in this cohort was 13% (5% mild and 8% moderate). The frequency of moderate cognitive dysfunction (ie performance <5th percentile in multiple cognitive domains) observed in this cohort of patients treated with LT is close to the frequency of 9% recorded in our previous study of untreated ATTRV30M patients [12]. Also, similar to previous studies [11,12], learning and memory and executive functions were the most affected cognitive domains. The present study further explored cognitive dysfunction in ATTRV30M patients, by characterizing a group of patients with milder deficits (ie performances between 5th and 10th percentiles in multiple cognitive domains).
Older age at assessment, lower education and higher severity of peripheral polyneuropathy (mPND score) were associated with cognitive dysfunction in ATTRV30M patients treated with LT, regardless of the level (ie mild or moderate). These findings are consistent with the results of our previous study in untreated ATTRV30M patients [12]. High NT-proBNP and Cystatin C values were also associated with cognitive dysfunction, irrespective of the level of dysfunction, suggesting that cognitive dysfunction in ATTRV30M patients is related to more cardiac and/or renal dysfunction. The development of cognitive dysfunction and systemic disease symptoms (eg polyneuropathy, cardiac and kidney dysfunction) may share common underlying mechanisms. Though, in the case of LT patients, it cannot be exclusively attributed to ATTRV30M disease. The observed effects of age and education cannot be easily explained by normal variations in cognitive performance related to these demographic variables, because cognitive deficits were identified at the individual level, based on large national normative data and took into account each patient's sex, age and education.
The effects of age at disease onset, disease duration, history of FNEs and cognitive complaints and current depression varied with level of cognitive dysfunction. Older age at disease onset, cognitive complaints and depression were related to moderate cognitive dysfunction. This pattern of associations is consistent with the results from the untreated ATTRV30M cohort, which also explored cognitive dysfunction at the moderate level [12]. The present study provides further evidence that age is strongly related to CNS phenotype expression of ATTRV30M amyloidosis. Older age at disease onset appears to accelerate cognitive decline in both untreated and treated with LT. The age-related cognitive dysfunction in patients with ATTRV30M amyloidosis may reflect the complex interaction between TTR, Ab and Tau pathology [15][16][17]. Other factors may modulate cognitive dysfunction in late onset of ATTRV30M amyloidosis, such as lower education, reflecting the importance of cognitive reserve and the involvement of other organs, namely the kidney and the heart [12,18]. Though, cardiac rhythmic disorders (ie atrial flutter or atrial fibrillation), which are important risk factors for ischaemic cerebrovascular disease [19], were not significantly associated with cognitive dysfunction or FNEs in our cohort of ATTRV30M amyloidosis patients treated with LT. Noteworthy, no difference was found between patients with moderate cognitive dysfunction vs. normal cognition regarding disease duration and history of FNEs. These set of findings suggest that moderate cognitive dysfunction occurs relatively early in the course of the disease (ie around the time of onset) in ATTRV30M amyloidosis patients with late disease onset. Longitudinal studies are necessary to elucidate whether cognitive deterioration in late onset has a cascadic (ie deterioration takes place in a relatively short period of time and not as a consequence of chronicity) or a progressive course. Notably, our previous study showed that older asymptomatic carriers are not at increased risk of developing cognitive dysfunction [12].
In comparison to the normal cognition group, patients with mild cognitive dysfunction had longer disease duration and more FNEs, despite having similar age at disease onset. The mild cognitive dysfunction subgroup was also younger at disease onset than the moderate cognitive dysfunction subgroup, even though they had similar age at assessment. In agreement with a previous study [7], the prevalence of FNEs in our series of 269 ATTRV30M patients treated with LT was 31%. Mild cognitive dysfunction in ATTRV30M amyloidosis appears to be closely related to both longer disease duration and FNEs. The literature has provided evidence that the occurrence of FNEs in ATTRV30M patients treated with LT increases with longer disease duration [7,9,14]. The results of our research appear to complement these studies, which did not explore cognition. It is reasonable to speculate that FNEs and mild cognitive dysfunction may share a common pathophysiological basis and even be 'two sides of the same coin' and point to an important contribution of disease duration to the occurrence of mild cognitive dysfunction.
The present study did not confirm that male sex and renal dysfunction are risk factors for FNEs in ATTRV30M patients treated with LT. Though, the use of Cystatin C as the single indicator of renal function and the presence of missing data limits the interpretation of these results. In our cohort, higher NT-proBNP values were recorded in patients with FNEs. Further studies are necessary to understand whether this reflects a global progression of the disease or the presence of common pathophysiological mechanisms in different target organs. The neuroimaging investigation of patients with cognitive dysfunction revealed mostly non-specific abnormalities in CT scans. MRI exams were not available for all but one case because patients had implanted cardiac pacemakers not compatible with MRI. In the reviewed exams, no distinct pattern was found for mild vs. moderate cognitive dysfunction. Also, no evidence of major cerebrovascular disease from common aetiologies was identified. Though, only structural imaging, mostly CT scans, was available and not all patients had contemporaneous neuroimaging exams, therefore limiting the interpretation of these results. Cranial MRI could have provided valuable clues to distinguish ATTR amyloidosis from other aetiologies for CNS involvement. Amyloid PET imaging could have also provided additional information, important to better understand the pathogenesis of cognitive dysfunction [9]. Regarding limitations of the study, the small number of late onset ATTRV30M patients treated with LT in our cohort, precluded further exploration of the effects of age and age at disease onset on cognitive dysfunction. Neither anxiety nor depression appear to be major contributor to cognitive dysfunction in ATTRV30M patients treated with LT, though only a screening tool was used to explore these psychopathological symptoms. The reduced number of cases also limited the complexity of the statistical analysis, and we chose to focus on bivariate associations. Another limitation of this study is the lack of biochemical and pathological data to further elucidate the pathogenesis of the neuropsychological syndrome described.

Conclusions
This is the largest study of cognitive dysfunction in ATTRV30M patients treated with LT. The study findings confirm that cognitive dysfunction is related to the severity of polyneuropathy. Older age at disease onset is statistically associate and potentially an important predictor of moderate cognitive dysfunction, whereas longer disease duration is necessary for the occurrence of mild cognitive dysfunction. A longitudinal approach is necessary to understand the pattern of progression of cognitive dysfunction in ATTRV30M patients treated with LT or under newer therapeutic strategies for ATTR amyloidosis.