--- Insert the Prisma Flow Chart Diagram about here ---
3.1 Neuro-psychological evaluation (Table 1 summarizes the studies and the outcome measures included in the quantitative meta-analysis)
On meta-analysis, eighteen studies evaluated, with standardized tests, the different cognitive domains in individuals with genetically confirmed FRDA compared to control participants devoid of neurological or psychiatric diseases. These studies, summarized in Table 1 were included in the quantitative meta-analysis. Six studies were included for qualitative review despite the lack of controls due to large sample size of FRDA patients,29,30 longitudinal follow-up,15,19 practical composite outcome measure test31 or structural correlation.32
--- Insert Table 1 about here ---
While standardized, the neuro-psychological test battery varied across the different studies but allowed data pooling, from at least two studies, for most outcome measures. Significant differences were found in all cognitive domains for individuals with FRDA compared to control participants. (Table 2, summarizes the pooled analysis)
3.1.1 Cognitive screening
Cognitive screening was realized in thirteen out of the eighteen studies. Three studies used the MOCA,22,33,34 three studies used the MMSE,16,18,20 two studies used the Symbol Digit Modality Test (SDMT),35,36 two studies evaluated IQ37,38 and one study assessed verbal IQ.14 Only results from MOCA and MMSE could be pooled as one study corrected SDMT score for ataxic and dysarthric impairments using patients’ PATA rate test and nine hole pegboard test score36 and IQ scores were not evaluated using the same method in the three studies. On pooled analysis, FRDA patients displayed lower MOCA and MMSE scores with large effect size (23.9±3.4 vs 26.9±2, p<0.0001, d: 1.2 and 28.6±1.4 vs 29.3±0.98, p=0.0005, d: 1.4 respectively).
3.1.2 Verbal fluencies
Seven studies assessed phonemic and semantic fluencies,14,16,18,22,33−35 of which two22,33 corrected the results using patients’ PATA rate test and nine hole pegboard test score using the methods described in Sacca et al.35 The raw values from Sacca et al,35 were pooled with the uncorrected studies. Two studies assessed action verbal fluencies16,18 and one used a distinct verbal fluency from the aforementioned studies.37 Verbal fluencies in all modalities were significatively poorer in FRDA patients compared to healthy controls with large effect size that were only marginally reduced by correction by PATA rate and nine hole pegboard test scores.
Language was evaluated in four studies. The Hayling sentence completion task was used in two studies,39,40 the Boston naming test in one37 and a Multiple choice vocabulary test task in another.34 Only in the Multiple choice vocabulary test task did individuals with FRDA performed similarly to controls.34
Sixteen studies assessed attention and/or executive functions. Eight studies used the Reitan’s41 Trail making Test (TMT),20,22,33,36,39,40,42,43 with patients’ PATA rate test and nine hole pegboard test scores correction in three.22,33,36 Six studies used the method developed by Golden44 to calculate the Stroop interference score,20,33,34,39,40,43 three studies used distinct and different declination of the Stroop test.14,22,37 Simon’s task incongruent reaction time was reported in two studies.43,45 Attentive matrices test was used in two studies,33,37 as was the Tower of London test.14,37 Wisconsin card sorting18 and every day attention tests46 were evaluated in two distinct studies. Pooled analysis disclosed lower performances in the TMT associated to a large effect size that grew larger after correction for ataxic symptoms. Individuals with FRDA showed pooled impaired performance on the Stroop interference score with a medium effect size.
3.1.5 Attention and working memory
Seven studies relied on the digital span to assess attention and working memory. Three studies used the digital span task included in the Wechsler adult intelligence scale, third version (WAIS-III) that starts with a sequence of two digits and ends with a sequence of nine digits,18,22,40 two studies used the task included in the revised Wechsler adult intelligence scale (WAIS-R) that starts with a sequence of three digits and ends with a sequence of eight digits,14,37 and two did not specify the version used.33,34 One study used the Paced Auditory Serial Addition Task (PASAT).34 Pooled analysis disclosed significant differences in digital span forward (DSF) in both WAIS versions and backward (DSB) only in WAIS-III version. There, individuals with FRDA displayed lower performance compared to controls, associated to a medium (DSF) or high effect size (DSB).
3.1.6 Memory and learning
Six studies assessed memory and learning skills. The evaluation included the 10/36 Spatial recall test (SPART) in four studies,18,22,33,36 the California verbal learning test (CVLT) in two,18,34 the Verbal learning and retention memory test in one14 and the logical memory test in one study.18 Only SPART performances were lower in individuals with FRDA on pooled analysis but with a large effect size.
Four studies included visuospatial skill tests. Those tests consisted in the Rey Auditory Verbal Learning Test (RAVLT) in two studies,22,33 segment length discrimination (SLD) in two studies,33,36 Raven Colored Progressive Matrice and mental rotation in one study,33 Incomplete letters and position discrimination tests in one,14 and Judgement line orientation, Facial recognition test and Block design in one study.18 FRDA patients performed worse than controls with large size effect for the SLD and medium size effect for the RAVLT.
3.1.8 Emotion recognition and social cognitive abilities.
Emotion recognition and social cognitive abilities were evaluated in three studies using different outcome measures including the Social Cognitive and Emotional Assessment and Ekman facial expression recognition Test 30; the Faux-pas test (n=22)34 and the Geneva Emotion Recognition Test (n=20).47 Individuals with FRDA compared to control participants, were less efficient in the Faux-pas test (n=22)34 and Geneva emotion recognition test (n=20).47
--- Insert Table 2 about here ---
3..2 Correlations between cognitive function and clinical parameters
Data from thirteen studies were identified as appropriate to include in potential correlations between neuropsychological test results and clinical parameters such as GAA1 (allele with smaller GAA repeat size), age of symptom onset (ASO), disease duration (DD) and clinical scales (Friedreich Ataxia Rating Scale (FARS), Scale for the Assessment and Rating of Ataxia (SARA)).
Ciancarelli et al. found no correlation between memory and phonemic verbal fluency test results and disease duration (n=24).19 Corben et al. reported a significant negative correlation between ASO and the incongruency effect in a Simon task in two consecutive studies from their group (n=13; n= 12) but no other significant correlations with GAA1, DD and the FARS score. 43,48 In addition a later work did not identify a correlation between ASO, DD, GAA1 or the FARS score with TMT and Stroop inferences scores (n=43).39 Dogan et al. found a significant correlation between impaired phonemic fluency performance and longer DD but no correlation between tests of memory, attention, executive and social cognition and clinical parameters (n=22).21 Similarly in a large European cohort (n=592), a correlation between DD and phonemic fluency was identified.49 Klopper et al. disclosed significant correlations between subtests of every day attention and GAA1 and the FARS score but not with DD nor ASO (n=16).46 Mantovan et al. described poorer Stroop and Tower of London performances in individuals with longer DD, however did not find similar correlations with memory, language and calculation tests. Moreover, GAA1 was not correlated to any neuropsychological measure while clinical scores like the FARS or SARA were not reported (n=13).37 Sayah et al., found correlations between measures of attention and the SARA, DD and GAA1 (n=46).30
Two longitudinal studies looked at the evolution of neuropsychological test results over time. Shishegar et al. found worsening performances for TMT (B-A) in individuals with FRDA over 24 months, but no degradation in working memory or executive function. In addition no correlation between clinical parameters and neuropsychological test results were identified (n=21).40 In the EFACTS cohort, over a two-year follow-up, no significant decline was found in verbal fluency.50 Yet, with longer follow-up, in a study where individuals with FRDA were assessed on average with an eight year separation, Hernandez-Torres et al. (n=39) found worse performances for the Stroop interference task, phonemic fluency tests and processing speed, while memory or visuospatial skills remained stable (n=29).15
Finally, two studies correlated composite scores based on combined neuropsychological test results. Nachbauer et al. found correlation with an executive score that included attention, executive, verbal fluency and visuospatial items and GAA1, ASO and the SARA score (n=29).14 Naeije et al. described a tight correlation between the cerebellar cognitive affective syndrome (CCAS) scale score and the SARA score (n=19).31
3.3 Correlations between cognitive performances and structural parameters
Six studies sought correlations between imaging measures and cognitive performances.
Ahklagi et al. (n=12), in a magnetic resonance imaging (MRI) and tractography study exploring reaction time and Simon (incongruence) effect in individuals with FRDA and controls found that, the mean and radial diffusivity of the dentato-rubral tract was positively correlated with choice reaction time, congruent reaction time, incongruent reaction time and Simon effect reaction time and negatively with the larger GAA1.48 Cocozza et al. (n=19), in a voxel-based morphometry and volumetric MRI study, found a direct correlation between cerebellar Lobule IX volume and impaired visuo-spatial functions but no correlations between structural parameters and executive and memory test results36. A former study by the same authors found no significant correlation between functional MRI (fMRI) resting state connectivity (rsFC) study (n=24) and language, memory, executive and visuospatial tests.33
Dogan et al. (n=22), in a fMRI and diffusion tensor imaging study combining language, memory, attention, executive and social cognition tests only showed in, post hoc correlations, a significant negative association between right cerebellar Crus I and left Brodmann area 44 and left insula functional activities for phonemic fluency execution in individuals with FRDA.34 Harding et al. (n=29), using a verbal n-back working memory task in fMRI, disclosed that task-related activation in the right dentate nucleus was significantly associated with a composite clinical index score based on ASO, the FARS score, DD and GAA1. This effect was most pronounced with respect to the FARS score and GAA1.32 In the longitudinal assessment of the same cohort, Shishegar et al. (n=21), found no significant correlations between longitudinal change in neurocognitive measures and change in brain activation over time.40