Professional Documents
Culture Documents
Meta-Analysis of Social Cognition in Amyotrophic Lateral Sclerosis
Meta-Analysis of Social Cognition in Amyotrophic Lateral Sclerosis
ScienceDirect
Journal homepage: www.elsevier.com/locate/cortex
Research report
Article history: Amyotrophic lateral sclerosis (ALS) is associated with executive dysfunction and behav-
Received 3 August 2016 ioural impairment. Recent studies suggested that social cognitive deficits might also be a
Reviewed 4 October 2016 prominent feature of ALS. Current meta-analysis aimed to summarize available evidence
Revised 15 October 2016 for deficits in social cognition including theory of mind (ToM) and emotion recognition in
Accepted 16 November 2016 ALS. In this meta-analysis of 15 studies, facial emotion recognition and ToM performances
Action editor Brad Dickerson of 389 patients with ALS and 471 healthy controls were compared. ALS was associated with
Published online 5 December 2016 significant impairments with medium effect sizes in ToM (d ¼ .65) and facial emotion
recognition (d ¼ .69). Among individual emotions recognition of disgust and surprise were
Keywords: particularly impaired. Deficits in perspective taking (d ¼ .73) aspects of ToM (ToM-PT) was
Theory of mind more pronounced in comparison to decoding (d ¼ .28) aspects of ToM (ToM-decoding). The
Social cognition severity of social cognitive impairment was similar to level of executive dysfunction and
Emotion recognition there was a significant relationship between social cognition and executive dysfunction.
Cognitive Deficits in social cognition are part of the cognitive phenotype of ALS.
Amyotrophic lateral sclerosis © 2016 Elsevier Ltd. All rights reserved.
behavioural impairment (ALSbi) and, in a small proportion, FTD 2009). A literature search was conducted using the databases
(ALS-FTD) (Strong et al., 2009). Pubmed, PsycINFO, ProQuest and Scopus to identify the rele-
Social cognitive impairment, in addition to dysexecutive vant studies (January 1990eJuly 2016) using the combination
syndrome, is one of the hallmark features of frontotemporal of keywords as follows: (ToM or emotion recognition or social
dysfunction syndromes (Bertoux, O'Callaghan, Dubois, & cognition) and “amyotrophic lateral sclerosis”. Reference lists
Hornberger, 2016; Bora, Velakoulis, & Walterfang, 2016; Bora, of published reports were also reviewed for additional studies.
Walterfang, & Velakoulis, 2015). Therefore, it might be ex- Inclusion criteria were studies that: (1) Compared ToM or
pected that deficits in social cognition can be a feature of ALS. The facial emotion recognition performances of nondemented
recognition of affect from perceptual cues and theory of mind ALS and healthy controls; (2) reported sufficient data to
(ToM), the ability to attribute mental states (feelings, beliefs, in- calculate the effect size and standard error of the social
tentions, and desires) to others and understand and predict cognition measure. Studies investigating emotion recognition
others' behaviour based on their mental states, are the most abilities with methods other than facial recognition were
commonly studied domains of social cognition. ToM is a het- excluded as only a few studies investigated ALS using such
erogeneous construct including abilities to infer others' mental methods (i.e., vocal). Two studies were excluded as they
states perspective taking (ToM-PT) or decoding (ToM-decoding) included a mixed sample of patients with and without de-
based on emotional/perceptual cues to infer others' thoughts, mentia (Table 1s in supplement). Savage et al. (2014) included
beliefs, emotions (Bora & Ko € se, 2016; Sabbagh, 2004; Sabbagh, pure ALS and FTD-ALS groups, FTD-ALS sample of this study
Moulson, & Harkness, 2004). A review of studies investigating was excluded from the current meta-analysis.
emotion recognition abilities suggested that results of the
studies were mixed but recognition of anger, sadness and disgust 2.2. Social cognition measures
might be impaired in ALS (Sedda, 2014). Recently, a number of
studies have also investigated ToM in ALS (Burke et al., 2016; Studies have investigated facial emotion recognition with a
Watermeyer et al., 2015). The outcomes of these studies were variety of methods (different sets of images of faces to label:
inconsistent, especially for ToM-decoding. A number of studies most commonly Ekman) (Ekman & Friesen, 1976). Different
have not found significant differences between ALS and control ToM tasks have been utilized across studies, most commonly
groups in ToM-decoding tasks (Cavallo et al., 2011; Watermeyer Reading the Mind in the Eyes Task (RMET); other tasks have
et al., 2015). The findings of these studies were also inconsis- included faux pas recognition (the task involves recognizing
tent regarding whether impaired performance in ToM-PT in ALS faux pas in series of short stories), judgement of preference
reflects specific deficits in social cognition or general cognitive test (JPT), Happe cartoons and stories, and The Awareness of
deficits (Cavallo et al., 2011; Gibbons et al., 2007). Currently, it is Social Inference Test (TASIT) sarcasm (Baron-Cohen,
not known whether recognition of some emotions or different Campbell, Karmiloff-Smith, Grant, & Walker, 1995; Baron-
aspects of ToM are differentially impaired in ALS. The severity of Cohen, Wheelwright, Hill, Raste, & Plumb, 2001). In RMET,
social cognitive impairment in comparison to other neuropsy- individuals are instructed to look at a series of photographs of
chological deficits in ALS is also unclear. just the eye region of the face, and picking which of four words
Some of the inconsistent findings of studies investigating best describe what the person in the photo is thinking or
social cognition in ALS might be related to low statistical feeling. RMET was the only measure for ToM-decoding (Baron-
power of individuals studies as most of the available studies Cohen et al., 2001; Sabbagh et al., 2004). All other ToM tasks
have small sample sizes. A meta-analysis can be helpful to were measuring ToM-PT (also called mental state reasoning
increase statistical power and estimate the effect size for (Sabbagh, 2004)).
different aspects of social cognitive deficits in ALS. There has
been only one attempt to summarize social cognitive deficits 2.3. Statistical analyses
in ALS using meta-analytical methods, the meta-analysis of
Beeldman et al. (2016) included a preliminary analysis of a For studies that reported more than one ToM task, pooled
small number of studies (k ¼ 5) including social cognitive effect sizes (Cohen d) and standard error values were calcu-
measures. Also, the social cognition measure in the meta- lated. A pooled effect size of social cognition based on ToM
analysis of Beeldman et al. (2016) included a combination of and/or total emotion recognition score was calculated for each
facial emotion recognition, prosody and ToM. The primary study. Individual tasks analyses for RMET (ToM-decoding)
goal of this meta-analysis was to quantify the magnitude of were also possible. A pooled effect size of ToM-PT was also
social cognitive deficits, to examine if some aspects of calculated by averaging effect size of ToM tasks other than
emotion recognition and ToM are relatively more impaired in RMET. In addition to the total emotion labelling score, sepa-
ALS than others, and to explore the relationship between so- rate effect sizes for six basic emotions (anger, fear, disgust,
cial cognitive impairment and other variables in ALS. sadness, happiness, surprise) were also calculated. In 10
studies that co-administered cognitive assessments along
with social cognitive tasks, a pooled effect size for executive
2. Methods dysfunction was calculated for each study. As a measure of
executive of functions, these studies included one or several
2.1. Study selection of Brixton Spatial Anticipation test, Hayling Sentence
completion test, Stroop interference, trail making B test, ver-
We followed PRISMA guidelines in conducting this meta- bal fluency tests and frontal assessment battery. Five studies
analysis (Moher, Liberati, Tetzlaff, Altman, & Group PRISMA, reported correlational coefficient for the relationship between
c o r t e x 8 8 ( 2 0 1 7 ) 1 e7 3
social cognition and executive functions, we also conducted a (Table 1). Egger's tests, trim and fill method (Table 1) and in-
meta-analysis of correlational effect sizes (r) including these spection of funnel plots provided evidence of publication bias
studies. for social cognition and ToM. Estimated effect sizes with trim
Meta-analyses were performed using packages in the R and fill method for ToM (d ¼ .47) and social cognition (d ¼ .49)
environment (OpenMetaAnalyst, Metafor) (Viechtbauer, 2010; were smaller.
Wallace et al., 2012). Effect sizes were weighted using the in- In the analysis of individual emotions, ALS patients
verse variance method. A random effects model (DerSimo- significantly underperformed controls in disgust (d ¼ .65,
nianeLaird estimate) was used. Homogeneity of the CI ¼ .29e1.02), surprise (d ¼ .65, CI ¼ .28e1.01) and sadness
distribution of weighted effect sizes was tested with the Q and (d ¼ .50, CI ¼ .13e.86) (Table 1). There were no significant
I2 tests. Tau-squared (t2), an estimate of between-study vari- deficits for recognition of anger, fear and happiness in ALS
ance, was used as a measure of heterogeneity in the random (Table 1). Among ToM domains, ToM-PT was particularly
effects model. Publication bias was assessed by inspection of impaired in ALS (d ¼ .73, CI ¼ .53e.93) (Fig. 3). There was also a
funnel plots, Egger's test and trim and fill method. Egger's test significant but a very mild impairment in ToM-decoding as
relies on the theory that small studies with significant rather measured with RMET (d ¼ .28, CI ¼ .04e.52) (Table 1). Distri-
than negative findings in studies with small sample sizes bution of effects sizes for ToM-PT, RMET, JPT and six basic
would be more likely to be reported while large-scale studies emotions were homogenous (Table 1).
would be more likely to be published regardless of the sig-
nificance of the findings. The significance threshold of Egger's 3.1. Effect of confounding variables
test was defined as p ¼ .1 due to small numbers of studies
included in some analyses. In two studies, depression ratings were higher in ALS patients
Subgroup analyses were conducted for depressive symp- compared to controls. However, the severity of social cogni-
toms (No difference vs significantly increased in ALS) and ALS tive impairment in ALS in these studies in were not signifi-
severity based on ALS Functional Rating Scale-revised form cantly different compared to other studies (d ¼ .83,
(ALSFRS-R; Cedarbaum et al., 1999) [Mild vs Moderate (cut-off CI ¼ .39e1.28 vs d ¼ .62, CI ¼ .46e.77, Qbet ¼ 1.1, p ¼ .29). Also,
score ¼ 37)]. A subgroup analysis of social cognition and ex- there was no significant effect of study level severity of ALS
ecutive dysfunction was conducted in 10 studies investigating symptoms (mild vs moderate on ALSFRS-R) on the social
both social cognition and other cognitive functions. The Qbet cognitive impairment in patients (d ¼ .67 vs .79, Qbet ¼ .45,
test was used to statistically compare subgroups in these an- p ¼ .50). There was a non-significant tendency for studies with
alyses. Meta-regression analyses were conducted for gender a higher percentage of ALS patients with bulbar onset
(ratio of males in ALS group) and bulbar onset (ratio of patients reporting more severe ToM impairment in ALS (Z ¼ 1.75,
with bulbar onset) with a random effects model were using p ¼ .08). Gender had no significant effect on between-group
the restricted-information maximum likelihood method with differences in social cognition. Study-level meta-regression
a significance level set at p < .05. The effect of bulbar onset on analysis found a significant relationship between severity of
social cognitive impairment was investigated as previous deficits in social cognition and executive functions (Z ¼ 1.95,
studies have associated bulbar-onset with more cognitive p ¼ .05, k ¼ 10). Also, a meta-analysis of correlational data
deficits in ALS. investigating the relationship between social cognition and
executive functions confirmed that social cognitive abilities
are associated with executive functions (r ¼ .52, CI ¼ .37e.67,
3. Results p < .001, k ¼ 5) (Fig. 1s in the supplement).
IdenƟficaƟon
Unique records idenƟfied through AddiƟonal records idenƟfied
searching keywords through reference lists
(n =44 ) (n =3 )
Excluded
(n =17 )
Included demented ALS=2
Studies included in
quanƟtaƟve synthesis
(meta-analysis)
(n =15)
Included
Table 1 e Mean weighted effect sizes for differences between patients with ALS and healthy controls on social cognition
tasks.
Test Study N ALS HC d 95% CI Z p Q Q (p) t2 Bias (p) I2 (%) Estimated Cohen d (CI)
Trim and fill
Social cognition 15 389 471 .64 .50e.78 8.8 <.001 13.8 .47 0 <.01 0 .49
ToM 11 311 339 .65 .47e.84 6.9 <.001 12.6 .25 .02 <.01 21 .47
- ToM-PT 11 311 339 .73 .53e.93 7.2 <.001 14.1 .17 .03 <.01 29
- RMET 5 166 170 .28 .04e.52 2.3 .02 14.1 .12 .01 .09 14
Facial emotion 5 92 152 .69 .42e.97 5.0 <.001 1.1 .89 .14 .41 0 No change
- Anger 3 56 74 .33 .18e.84 1.3 .20 3.9 .14 .10 .24 49
- Fear 3 56 74 .29 .08e.65 1.6 .12 .80 .67 0 .49 0
- Sad 3 56 74 .50 .13e.86 2.7 .007 0.5 .77 0 .16 0
- Happy 3 56 74 .13 .30e.57 0.6 .55 3.1 .21 .05 .36 36
- Disgust 3 56 74 .65 .29e1.02 3.5 <.001 0.4 .86 0 .71 0
- Surprise 3 56 74 .65 .28e1.01 3.5 <.001 1.1 .57 0 .20 20
ALS ¼ amyotrophic lateral sclerosis, HC ¼ healthy controls, d ¼ Cohen's d, ToM ¼ theory of mind, CI ¼ confidence interval.
behavioural impairment not reaching the threshold for de- (2009) are inadequate for the classification of nondemented
mentia. However, it is also important to note that executive ALS patients with mild behavioural and cognitive impair-
dysfunction and social cognitive deficits can be only partially ments. Including social cognitive and other cognitive deficits
overlapping in ALS. The data-driven approach of Consonni, in the definition of ALSci should be considered.
, Dalla Bella, and Cappa (2016) showed that social
Catricala Another important subject is the neural substrates of social
cognitive impairment co-exists with ALSci and ALSbi in most cognitive deficits in ALS. Previous studies found a relationship
of the subjects, there was a subgroup of ALS patients char- between cognitive deficits and neuroanatomical abnormal-
acterized by social cognitive impairment and non-executive ities including cortical atrophy and basal ganglia pathology in
dysfunction (language and episodic memory) but not execu- nondemented ALS (Machts et al., 2015; Mioshi et al., 2013). It is
tive dysfunction. Therefore, consensus criteria of Strong et al. likely that social cognitive impairment is also related to more
c o r t e x 8 8 ( 2 0 1 7 ) 1 e7 5
Fig. 2 e Forest plot for facial emotion recognition differences between ALS and healthy controls: CD ¼ Cohen d; Diamond
shape ¼ average effect size and its confidence interval based on random effects model.
Fig. 3 e Forest plot for ToM-PT differences between ALS and healthy controls.
widespread brain imaging abnormalities in ALS. Only three 2015), which are spared in most patients with ALS. However,
studies which are also included in the current meta-analysis ALS-FTD is associated with more extensive brain atrophy and
investigated structural and functional brain imaging corre- deficits in recognition of fear and anger were robust in a FTD-
lates of social cognitive impairment in ALS. The findings of ALS study (Savage et al., 2014). It might be hypothesized that
these studies suggested that volumetric reductions and deficits in ToM-decoding and recognition of some emotions
reduced functional activity in medial and inferior prefrontal can be a marker of a small group of individuals who are at risk
cortex and white matter tracts connecting these regions to of developing FTD-ALS at follow-up or who have C9ORF72
other parts of the brain are associated with social cognitive expansions. However, longitudinal studies are needed to test
impairment in ALS (Carluer et al., 2015; Cerami et al., 2014; this hypothesis.
Crespi et al., 2014). Social cognitive deficits observed in this meta-analysis
The severity of social cognitive deficits was much less were not significantly related to ALS severity as measured
pronounced and more circumscribed than FTD (Bora et al., with ALSFRS-R. However, total ALSFRS-R score might not be a
2015, 2016). FTD is associated with broad deficits in all as- sensitive measure to show the relationship between symptom
pects of ToM and emotion recognition. However, cognitive severity and social cognitive deficits in ALS. Two patients
profile of ALS included more specific deficits in ToM-PT and might differ significantly in their bulbar and spinal scores but
recognition of disgust and surprise. Abnormalities in the might have the same ALSFRS-R total score. Unfortunately,
frontal lobe and its insular and subcortical connections can nearly none of the included studies reported separate sub-
explain deficits in ToM-PT and in these some emotions. Basal scale scores of ALSFRS-R. The site of onset of ALS might be
ganglia and insula have important roles for recognition of related to social cognitive deficits in ALS. There was no suffi-
disgust. In recognition of surprised faces inverse coupling cient data to investigate the effect of bulbar onset on social
between ventromedial prefrontal cortex and amygdala is cognition quantitatively. However, one of the studies included
important (Kim, Somerville, Johnstone, Alexander, & Whalen, (19) suggested that ToM-decoding deficits might be evident
2003). Ventromedial prefrontal cortex dysfunction in ALS can only in bulbar-onset ALS.
lead to impaired positive interpretation and misidentification Another consideration was the relationship between
of surprised faces as fear. More widespread atrophy involving neuropsychiatric symptoms and social cognition in ALS.
temporal lobes in FTD can explain pronounced deficits in Apathy is a common finding in ALS (Lillo et al., 2011) and it is
decoding mental states which are best identified from eyes not known if apathy is related to social cognitive impairment
(anger, fear and complex emotions) (Bassili, 1979; Mason et al., in ALS. Another neuropsychiatric condition in ALS is
6 c o r t e x 8 8 ( 2 0 1 7 ) 1 e7
study. Amyotrophic Lateral Sclerpsis & Frontotemporal Murphy, J. M., Henry, R. G., Langmore, S., Kramer, J. H.,
Degeneration, 15, 21e29. Miller, B. L., & Lomen-Hoerth, C. (2007). Continuum of frontal
Consonni, M., Catricala , E., Dalla Bella, E., & Cappa, S. F. (2016). lobe impairment in amyotrophic lateral sclerosis. Archives of
Beyond the consensus criteria: Multiple cognitive profiles in Neurology, 64, 530e534.
amyotrophic lateral sclerosis? Cortex, 81, 162e167. Phukan, J., Elamin, M., Bede, P., Jordan, N., Gallagher, L., Byrne, S.,
Crespi, C., Cerami, C., Dodich, A., Canessa, N., Arpone, M., et al. (2012). The syndrome of cognitive impairment in
Iannaccone, S., et al. (2014). Microstructural white matter amyotrophic lateral sclerosis: A population-based study.
correlates of emotion recognition impairment in amyotrophic Journal of Neurology Neurosurgery Psychiatry, 83, 102e108.
lateral sclerosis. Cortex, 53, 1e8. Ringholz, G. M., Appel, S. H., Bradshaw, M., Cooke, N. A.,
Cui, B., Cui, L., Gao, J., Liu, M., Li, X., Liu, C., et al. (2015). Cognitive Mosnik, D. M., & Schulz, P. E. (2005). Prevalence and patterns
impairment in Chinese patients with sporadic amyotrophic of cognitive impairment in sporadic ALS. Neurology, 65,
lateral sclerosis. PLoS One, 10, e0137921. 586e590.
Ekman, P., & Friesen, W. (1976). Pictures of facial affect. Palo Alto, Sabbagh, M. A. (2004). Understanding orbitofrontal contributions
CA: Consulting Psychologists Press 1976. to theory-of-mind reasoning: Implications for autism. Brain
Gibbons, Z. C., Snowden, J. S., Thompson, J. C., Happe , F., and Cognition, 55, 209e219.
Richardson, A., & Neary, D. (2007). Inferring thought and Sabbagh, M. A., Moulson, M. C., & Harkness, K. L. (2004). Neural
action in motor neurone disease. Neuropsychologia, 45, correlates of mental state decoding in human adults: An
1196e1207. event-related potential study. Journal of Cognitive Neuroscience,
Kim, H., Somerville, L. H., Johnstone, T., Alexander, A. L., & 16, 415e426.
Whalen, P. J. (2003). Inverse amygdala and medial prefrontal Savage, S. A., Lillo, P., Kumfor, F., Kiernan, M. C., Piguet, O., &
cortex responses to surprised faces. NeuroReport, 14, Hodges, J. R. (2014). Emotion processing deficits distinguish
2317e2322. pure amyotrophic lateral sclerosis from frontotemporal
Machts, J., Loewe, K., Kaufmann, J., Jakubiczka, S., Abdulla, S., dementia. Amyotrophic Lateral Sclerosis & Frontotemporal
Petri, S., et al. (2015). Basal ganglia pathology in ALS is Degeneration, 15, 39e46.
associated with neuropsychological deficits. Neurology, 85, Sedda, A. (2014). Disorders of emotional processing in
1301e1309. amyotrophic lateral sclerosis. Current Opinion in Neurology, 27,
Mason, S. L., Zhang, J., Begeti, F., Guzman, N. V., Lazar, A. S., 659e665.
Rowe, J. B., et al. (2015). The role of the amygdala during Shen, D., Cui, L., Fang, J., Cui, B., Li, D., & Tai, H. (2016). Voxel-wise
emotional processing in Huntington's disease: From pre- meta-analysis of gray matter changes in amyotrophic lateral
manifest to late stage disease. Neuropsychologia, 70, 80e89. sclerosis. Frontiers Aging Neuroscience, 30(8), 64.
Mioshi, E., Caga, J., Lillo, P., Hsieh, S., Ramsey, E., Devenney, E., Strong, M. J., Grace, G. M., Freedman, M., Lomen-Hoerth, C.,
et al. (2014). Neuropsychiatric changes precede classic motor Woolley, S., Goldstein, L. H., et al. (2009). Consensus criteria for
symptoms in ALS and do not affect survival. Neurology, 82, the diagnosis of frontotemporal cognitive and behavioural
149e155. syndromes in amyotrophic lateral sclerosis. Amyotrophic
Mioshi, E., Lillo, P., Yew, B., Hsieh, S., Savage, S., Hodges, J. R., Lateral Sclerosis, 10, 131e146.
et al. (2013). Cortical atrophy in ALS is critically associated Swinnen, B., & Robberecht, W. (2014). The phenotypic variability
with neuropsychiatric and cognitive changes. Neurology, 80, of amyotrophic lateral sclerosis. Nature Reviews Neurology, 10,
1117e1123. 661e670.
Moher, D., Liberati, A., Tetzlaff, J., Altman, D. G., & Group PRISMA. Viechtbauer, W. (2010). Conducting meta-analyses in R with the
(2009). Preferred reporting items for systematic reviews and metafor package. Journal of Statistical Softwares, 36, 1e48.
meta-analyses: The PRISMA statement. British Medical Journal, Wallace, B. C., Dahabreh, I. J., Trikalinos, T. A., Lau, J., Trow, P., &
339, b2535. Schmid, C. H. (2012). Closing the gap between methodologists
Montuschi, A., Iazzolino, B., Calvo, A., Moglia, C., Lopiano, L., and end-users: R as a computational back-end. Journal of
Restagno, G., et al. (2015). Cognitive correlates in amyotrophic Statistical Softwares, 49, 1e15.
lateral sclerosis: A population-based study in Italy. Journal of Watermeyer, T. J., Brown, R. G., Sidle, K. C., Oliver, D. J., Allen, C.,
Neurology, Neurosurgery and Psychiatry, 86, 168e173. Karlsson, J., et al. (2015). Executive dysfunction predicts social
Murphy, J., Factor-Litvak, P., Goetz, R., Lomen-Hoerth, C., cognition impairment in amyotrophic lateral sclerosis. Journal
Nagy, P. L., Hupf, J., et al. (2016). Cognitive-behavioral of Neurology, 262, 1681e1690.
screening reveals prevalent impairment in a large multicenter
ALS cohort. Neurology, 86, 813e820.