Loss of “insight” into behavioral changes in ALS: Differences across cognitive profiles

Abstract Objective Behavioral impairment occurs in amyotrophic lateral sclerosis (ALS) and ALS–fronto‐temporal dementia (ALS‐FTD). It has been proposed that ALS patients without FTD retain an awareness of their behavioral impairment while ALS‐FTD patients lose this awareness (referred to as retention vs. loss of “insight”). Loss of insight has not yet been studied across the entire ALS‐FTD spectrum; our study addresses this gap by including patients with all the ALS cognitive‐behavioral profiles. Methods Eighty‐three ALS patients (and their informants) took part in this bicentric study involving two German recruitment sites. Patients and informants completed the Frontal Systems Behavior Scale covering the domains of apathy, disinhibition, and executive dysfunctioning. Patients were classified into five groups according to the Strong and Rascovsky criteria: cognitively unimpaired (ALSni), cognitively impaired without dementia (ALSci), behaviorally impaired (ALSbi), a combination of behaviorally and cognitively impaired (ALScbi), and ALS‐FTD. We applied Bayesian two‐way ANOVA to test whether there were subgroup differences regarding insight into their behavioral decline. Results All patient subgroups experienced behavioral decline (Bayes factor > 3). Only ALS‐FTD patients lost insight into disinhibition and executive dysfunctioning. ALSbi patients exhibited worse insight than ALSni and ALSci patients (Bayes factor > 10). Evidence regarding the ALScbi patients was inconclusive. Higher IQ was associated with worse insight (Bayes factor > 3). Conclusions Our findings provide solid support for the notion that ALS patients without dementia experience behavioral decline regardless of their cognitive‐behavioral profile and retain different levels of insight into this decline. The inverse association of premorbid verbal intelligence with insight was unexpected, leaving room for further investigation.


INTRODUCTION
As a multisystemic disorder, amyotrophic lateral sclerosis (ALS) may impair cognition and behavior alongside voluntary motor control (Benbrika et al., 2019) and forms a disease spectrum with fronto-temporal dementia (ALS-FTSD, Strong et al. 2017). Cognitive-behavioral impairment is present in approximately 50% of ALS-FTSD patients; this amounts to fully-blown FTD in up to 15% of all ALS-FTSD patients (Montuschi et al., 2015;Ringholz et al., 2005). Characteristically, cognitive impairment entails executive and/or language deficits (Beeldman et al., 2016;Benbrika et al., 2019). The most frequent symptoms of behavioral impairment on the ALS-FTD spectrum are perseveration (40% of patients), apathy (29%), and disinhibition (26%, . Perseveration includes behavioral repetition without flexibly adjusting to external circumstances; apathy encompasses loss of interest and drive; inadequate reactions in social settings can indicate disinhibition. Behavioral impairment may be accompanied by loss of insight into disease-related impairment in up to 25% of ALS-FTD patients . A consensus on what constitutes insight or loss thereof is lacking (Evers et al., 2007). Our study focused on loss of insight into disease-related behavioral subclinical changes and clinical impairment, specifically when a patient views their own behavior differently from their primary caregiver .

Such loss of impairment insight has been documented in bvFTD and
ALS-FTD (Beeldman et al., 2018;Griffin et al., 2016;Ichikawa et al., 2008;Saxon et al., 2017;.  reported that while ALS-FTD patients viewed themselves as behaviorally impaired, they still estimated the degree of their impairment as less severe than their caregivers. ALS patients without FTD experienced mild behavioral abnormalities but retained insight (Terada et al., 2011;. Insight in ALS is a crucial research area, 's findings have influenced the revised Strong criteria (Strong et al., 2017), the creation of the dimension apathy scale (Radakovic et al., 2016) and the ALS-cognitive behavioural screen . Current evidence only points to differences in insight between ALS-FTD and ALS without FTD. The most recent Strong criteria profile ALS-FTSD patients into the following subgroups: ALS without cognitive impairment (ALSni), ALS with cognitive impairment (ALSci), ALS with behavioral impairment (ALSbi), ALS with cognitive and behavioral impairment (ALScbi), and ALS-FTD (Strong et al., 2017). Differences in insight between these profiles have not yet been investigated. A further gap in insight research in ALS is the absence of studies investigating the effect of premorbid intelligence. Retained insight with higher levels of verbal intelligence has been documented in other neurodegenerative disorders (Spitznagel & Tremont, 2005).
Our study aimed to address these gaps by profiling patients based on the Strong criteria (Strong et al., 2017), and exploring the effect of intelligence on insight into behavioral impairment. We hypothesized that ALS-FTD and behaviorally impaired patients would self-rate as behaviorally impaired while losing insight into their impairments' magnitude compared with their caregivers. We also expected these patient groups to exhibit greater loss of insight into their behavioral decline than ALS patients without behavioral impairments.
There were two demographic differences ( patients were considered behaviorally impaired. Our data provided evidence that there was no effect of familial relation between informant and patient on change ratings for apathy (BF 01 = 6.20), disinhibition (BF 01 = 5.87), executive dysfunction (BF 01 = 5.59), and overall change (BF 01 = 6.20).

Premorbid IQ
Participants underwent thorough neuropsychological examination, including an estimate of verbal-crystallized intelligence (Kasper et al., 2015;Kasper et al., 2016;Machts et al., 2014). Estimates of premorbid verbal IQ were obtained after disease onset by asking the participants to identify correct target words among distractor pseudo-words. The number of correctly identified target words was converted to an IQ estimate based on published norms (Schmidt & Metzler, 1992). ALS patients' performance on this specific test has been shown to remain intact (Lange et al., 2016;Neudert et al., 2001;Temp et al., 2021), and independent from physical disability (Osmanovic et al., 2020;Temp et al., 2021).
Our primary analyses were analyses of covariance with cognitivebehavioral profile according to the Strong criteria as the betweensubjects independent variable, IQ as the covariate, and T Insight in apathy, disinhibition, executive dysfunction, and overall behavioral problems as outcomes. As we had hypothesized the cognitive subgroups of ALS to differ in insight, we aimed to support the most likely alternative hypothesis instead of rejecting the null hypothesis. To this end, we employed Bayes factor hypothesis testing which permitted us to quantify support in favor of our hypothesis by comparing it to the null model (Wagenmakers, 2007;Wagenmakers et al., 2018;Wagenmakers et al., 2018). The data were analyzed in Jeffreys' Amazing We applied the following evidence categories (Wagenmakers et al., 2018): a BF above 3 provides "moderate evidence", a BF above 10 provides "strong evidence", a BF above 30 provides "very strong evidence", and a BF above 100 provides "extreme evidence" in favor of the best model.

RESULTS
First, we ascertained the presence of subclinical behavioral changes or clinically relevant behavioral impairment across patient groups before exploring the influence of IQ, and the differences in insight between patient groups.

Behavioral impairment: Premorbid versus current
In Figure 1, clinically relevant impairments (T ≥ 65) are depicted within the red bands and statistically significant increases in behavioral abnormalities are signified by solid lines.

Current self-ratings
Clinically relevant impairments were self-reported by the ALScbi group across the apathy and total behavioral change domains (Figure 1a,d); the ALSni, ALSci, ALSbi, and ALS-FTD groups did not self-report any impairments.
Statistically relevant subclinical increases in behavioral abnormality were self-reported by all patient groups across the apathy and total behavioral change domains (Figure 1a

Current informant ratings
Clinically relevant impairments were reported by the informants of the ALSbi, ALScbi, and ALS-FTD patients across the apathy and

Insight
Next, we explored whether the effect of premorbid IQ on insight

DISCUSSION
We investigated the research questions of behavioral impairment, loss of "insight" and the influence of verbal intelligence in ALS-FTSD by conceptually replicating previous methodologies (Grossman et al., 2007;Spitznagel & Tremont, 2005;Spitznagel et al., 2006;. We measured behavior with instruments identical to those of Grossman et al. (2007) and , and we estimated verbal intelligence similar to Spitznagel and Tremont (2005) and Spitznagel et al. (2006). The original methodological contributions of this study lie in its strict profiling according to the most recent Strong criteria (Strong et al., 2017), in addition to its exploration of associa-tions between intelligence and "insight" in ALS. Furthermore, our conceptualization of "insight" as failure to recognize subclinical behavioral changes or clinically relevant behavioral impairments which are present or absent, reduced the risk of biasing our analyses towards the behaviorally impaired patient groups.

Behavioral impairment
All Strong profile subgroups experienced statistically significant increases in behavioral abnormalities (Figure 1) in accordance with the literature . In contrast to the results of , the ALS-FTD patients in our study did not self-report any clinically relevant impairments. Whereas the ALSbi patients only self-reported a clinical impairment in the apathy domain, the ALScbi patients self-reported clinical impairments in apathy, executive dysfunctioning, and overall behavior. According to their informants, the ALScbi patients only exhibited apathy impairments prior to motor symptom onset. This is concordant with previous findings (Grossman et al., 2007; without differentiating between Strong profiles of ALS. No patient group exhibited disinhibition (Figure 1b).
Clinically relevant levels of disinhibition in ALS without FTD are sub-ject to debate in the literature:  report that 75% of ALS patients experience no or only mild symptoms of disinhibition, while Grossman et al. (2007) report that 29% exhibit impairment in this domain. In general, our results suggest that noticeable increases in behavioral abnormalities occur even in behaviorally unimpaired ALS patient groups. The impact of these changes on the patients and their families' lives remains to be studied.

Insight between cognitive-behavioral strong profiles
Our ALSni and ALSci groups retained insight into their mild behavioral abnormalities, as expected, based on previous findings (Terada et al., 2011;. Detailed distinctions between ALSbi and ALScbi patients, however, had been lacking. Our findings suggest that both groups retain insight clinically. Unexpectedly, our ALS-FTD patients retained insight into their apathy impairment, meeting the clinical criterion for loss of insight (T Iinsight ≤ −20) only in the disinhibition and executive domains, where they did not consider themselves impaired (Figures 1 and 3). This clinically relevant loss of insight occurred only in our ALS-FTD group, congruent with previous literature (Beeldman et al., 2018;Saxon et al., 2017;.
Regardless of clinically significant impairment in insight, we had expected the ALS-FTD ALScbi and ALSbi patients to experience stronger loss of insight than the ALSci and ALSni patients. There was strong evidence of those between-group differences in our data regarding all individual domains but not the total score. An increased loss of insight emerged primarily in the ALSbi and ALS-FTD patients.
However, evidence regarding the ALS-FTD patients was inconclusive in the domains of apathy and executive dysfunction. There was also insufficient evidence to facilitate conclusions regarding the insight of ALScbi patients. Our results thus establish an absence of evidence, though not of effects. This absence may be explained by our small ALScbi and -FTD groups (n = 8, and n = 4, respectively); a larger sample would be beneficial to provide more conclusive evidence.
We replicated the loss of insight documented by  in ALS-FTD but not all the differences from ALSni patients. Conceptual replications such as ours yield mixed results in approximately 10% of attempts, while 4% fail and 86% succeed (Makel et al., 2012).
In addition to the small sample sizes, several methodological aspects may explain our mixed replication results. The internal coherence of our Bayesian modeling techniques ensures that there is no sample size below which our inferences could be viewed as untrustworthy (Wagenmakers et al., 2018), and our ALS-FTD sample size (n = 4) was identical to that of . Nevertheless, the instances of inconclusive evidence in our data may be due to our sample size, and/or the heterogeneity of FTD. Rates of loss of insight in ALS-FTD patients have been contrastive between a larger sample size taken from a metaanalysis by  and the smaller sample sizes of both our study and that of Saxon et al. (2017): the former group observed loss of insight in only 25% of 170 ALS-FTD patients whereas 75% was detected out of four ALS-FTD patients in our study and 88% was described in 56 ALS-FTD patients by Saxon et al. (2017), again suggesting that studies investigating loss of insight with larger sample sizes would be of value to reduce the risk of bias. This heterogeneity across studies suggests that the exclusion of loss of insight from the Rascovsky criteria for bvFTD (Rascovsky et al., 2011) was an improvement in sensitivity over the Neary criteria, which had included loss of insight (Neary et al., 1998).

Insight and premorbid verbal intelligence
Paradoxically, higher verbal intelligence was associated with worse insight into apathy and overall behavioral issues (Figure 2a with higher intelligence experienced a lesser loss of insight has previously been described in other neurodegenerative diseases (Spitznagel & Tremont, 2005;Spitznagel et al., 2006). Three possible mechanisms may underlie our paradoxical findings. The first mechanism concerns the patients themselves: individual bvFTD patients may experience a loss of insight to different degrees in that they may perceive their disease-related decline as a threat to themselves and their lives (Griffin et al., 2016). This perceived threat may be more pronounced in those with a higher premorbid ability, resulting in a more vicarious denial of behavioral abnormalities and thus reduced insight. The second mechanism concerns the patient-carer relationship. Given that married couples are often of similar cognitive and intelligence levels (Caillot-Ranjeva et al., 2021;Mascie-Taylor, 1989;Watson et al., 2004), our more intelligent participants would have similarly intelligent spouses who, in turn, may potentially have rated them more harshly than the spouses of less intelligent patients owing to their related critical thinking abilities. Third, our results may indicate either that the patients or their family members were unable to express degrees of behavioral abnormality adequately or that we, at present, lack the sufficient measurement tools. These three mechanisms are not mutually exclusive.
One limitation of this work was the very small number of ALS-FTD patients who were still capable of self-rating; it is conceivable that a loss of insight would have been more apparent in those who could no longer self-rate. The study design of comparing self-ratings to informant ratings by default introduces bias into our sample as patients who were cognitively unable to comprehend and complete the FrSBe were excluded from the analysis. Conclusive evidence was only provided in the subgroups with n ≥ 12. Possibly, the low sample size is at the root of the absent evidence in the smaller subgroups. However, where effects are small or heterogeneous, even large samples may produce absence of evidence.
Though the FrSBe is considered the gold standard for measuring behavioral impairments, it may overestimate them in ALS as some behavioral symptoms may indeed be related to motor impairment(s) . German translations of alternative, ALSspecific behavioral instruments would certainly be desirable . These results would benefit from future research replicating our findings in larger cohorts, especially in ALS-FTD and ALScbi patients. Recent efforts to establish a cognitive reserve in ALS have been successful (Canosa et al., 2020;Consonni et al., 2020;Costello et al., 2021;Temp et al., 2021), and future efforts could investigate that concept more specifically in relation to behavior and insight. For the sake of clarification, the paradox of lower levels of insight at higher levels of intelligence deserves future investigation; perhaps a more holistic approach to intelligence including the family raters would produce different results. Furthermore, the relationship between behavioral changes and caregiver burden and its implications for "insight" requires future investigation.
In conclusion, behavioral decline occurred in ALS patients regard-

DATA AVAILABILITY STATEMENT
A CSV file and the HTML results transcript from JASP are available on the Open Science Framework: https://osf.io/r7jz2/.