Developmental Changes of Autistic Symptoms, ADHD Symptoms, and Attentional Performance in Children and Adolescents with Autism Spectrum Disorder

This study followed up ADHD/autistic symptoms and attentional performance in children/adolescents with ASD and typically developing ones (TD) over 5–7 years. The participants were stratified by age at baseline into child (< 12 years) and adolescent (12–19 years) groups. ADHD symptoms, especially hyperactivity, and attentional functions significantly improved during follow-up, more in children than in adolescents, in both ASD and TD. Significantly more omission errors and perseverations were noted in ASD than TD through the follow-up. Children with ASD had more improvement in reaction time while adolescents with ASD had less improvement in commission errors and detectability than TD. No correlation of attentional functions and ADHD symptoms in ASD implied different neural mechanisms of ADHD symptoms between ASD and ADHD.

Despite frequently co-occurring, autistic and ADHDrelated symptoms might have different developmental trajectories (St. Pourcain et al., 2011). Autistic symptoms (Robinson et al., 2011) and ASD diagnosis (Woolfenden et al., 2012) were relatively stable over time, while ADHD symptoms, especially hyperactivity-impulsivity, often declined significantly with age (Faraone et al., 2006;. Recent data showed that ADHD and autistic symptoms overlapped more frequently during childhood/adolescence than adulthood in the general population (Hartman et al., 2016), and the presence of persistent ADHD symptoms was related to persistent social-communication impairment during adolescence, but not vice versa (St. Pourcain et al., 2011). Little is known about the relationship of developmental changes in autistic symptoms and ADHDrelated symptoms in ASD (Leitner, 2014). More data are required to give us a better understanding of whether the development of general and specific ADHD symptoms is correlated to development of autistic symptoms in children and adolescents with ASD; and whether the development of ADHD symptoms in ASD is similar to or deviant from that in typically developing (TD) children and adolescents.
According to the model of Posner and Rafal (1987), the basic attentional process includes tonic alertness, phasic alertness, selective attention, and sustained attention (Valdez, 2019). Tonic alertness is the ability to respond to any stimulus in the environment, which is intrinsic arousal providing a cognitive tone for executive control (Posner, 2008). Phasic alertness is the rapid attention change, the ability to respond to a brief event after a warning signal (Posner, 2008). Selective attention is the ability to respond to different stimuli and to filter out less relevant stimuli. Sustained attention refers to the capacity to maintain continuous responses during a period of time (minutes-hours). Attentional functions are frequently measured by the Continuous Performance Test . Conners' Continuous Performance Test (CCPT) has a high signal-to-noise ratio design compared to traditional CPT and is more sensitive to commission errors, an index representing response inhibition (Egeland & Kovalik-Gran, 2010b). In addition, based on signal detection theory, CCPT incorporates some measures into d′ (one' ability to discriminate signals from noises) and β (one's internal threshold to respond, a lower threshold leading to more responses). A faster reaction time (RT) represents one's more direct discrimination of targets  and also better tonic alertness (Sturm & Willmes, 2001;Tales et al., 2002;Tucha et al., 2011). Increased RT variability means one's decreased consistency in responding, high RT standard error of inter-stimuli changes (RTSEISI) refers to a low vigilance (attentional capacity to keep alert when less stimulated), and high RT standard errors of block change (RTSEBC) represents impaired sustained attention (attentional capacity to maintain alert as time goes by).
Impairments of attentional functions were persistently found in ADHD (Rommelse et al., 2011). High omission and commission errors, increased variability/standard errors of RT, lower signal detectability, and higher β (lower tendency to respond) in CCPT were correlated to increased ADHD symptoms in children without symptom domain specificity . Additionally, increases of RTSEISI and RTSEBC in CCPT were noted in ADHD (Egeland & Kovalik-Gran, 2010a;Ord et al., 2020). Children and adolescents with ASD were also found to have problems in focus and sustained attention (Rommelse et al., 2011), including more commission and omission errors, slower RT, higher variability of RT, and decreased signal detectability than TD in CCPT or Test of Variables of Attention (Chien et al., 2015;Lundervold et al., 2016;Swaab-Barneveld et al., 2000). These attention deficits were largely overlapped with what were found in ADHD. In studies of head-to-head comparison of CCPT performance, ASD and ADHD were generally comparable in deficits of most CCPT indices, despite a more rapid decline of vigilance over time in ADHD than in ASD (Rommelse et al., 2011;Swaab-Barneveld et al., 2000). While some found that low consistency of CCPT in ASD might be explained by cooccurring with ADHD (Hwang-Gu et al., 2019;Lundervold et al., 2016), others found no additional effect of ADHD on the impaired sustained attention of TOVA (Nyden et al., 2010) and CCPT (Ng et al., 2019) of ASD. Dimensionally, our previous study found weak correlations between CCPT indices and ADHD symptoms in children and adolescents with ASD and suspected a different mechanism underpinning ADHD symptoms in ASD from ADHD (Chien et al., 2014). Whether ADHD symptoms in individuals with ASD have the same neural mechanisms as those in ADHD individuals requires more evidence.
Longitudinal studies demonstrated significant developmental changes (improvements but not necessarily normalizations) of executive functions in ADHD (Coghill et al., 2014; from childhood to adolescence. Existing data concerning developmental changes of executive functions in ASD were almost all cross-sectional age-stratified analyses (Demetriou et al., 2018;Hwang-Gu et al., 2019). One recent meta-analysis of 235 studies concluded that overall executive deficits (including concept formation, mental flexibility, fluency, planning, working memory, and response inhibition) in ASD remained stable across the life span without subdomain differences. The effect sizes of executive deficits were smaller in adults than children or adolescents with ASD, implying either a maturation effect or more compensatory strategies used in adults with ASD (Demetriou et al., 2018). A few agestratified cross-sectional studies comparing the executive functions (including working memory, planning, set-shifting, and response inhibition) between ASD and TD showed differential patterns between the child and adolescent groups Geurts et al., 2014). Longitudinal data of the developmental changes of attentional functions in ASD are lacking. Whether the developmental trajectory of the attentional process of ASD is parallel to that of TD, implying a maturational lag as found in ADHD (El-Sayed et al., 2003;Shaw et al., 2007) or is deviant from TD, is of great interest.
The primary aim of our study was to explore the longitudinal changes of autistic symptoms, ADHD-related symptoms, and attentional functions in children and adolescents with ASD to their adolescence and young adulthood, respectively, and to compare these changes between ASD and TD as well. Our secondary aim was to test whether the attentional functions were correlated to the autistic and ADHDrelated symptoms cross-sectionally at Time 1 and Time 2, and whether their developmental changes were correlated to one another longitudinally.

Participants and Procedures
The Research Ethics Committee of National Taiwan University Hospital approved this longitudinal study at Time 1 (approval number, 200903062R; ClinicalTrials.gov number, NCT00916851) and Time 2 (approval number, 201403109RINC; ClinicalTrials.gov number, NCT02233348). All the recruitment and data collection started after the child assent and written informed consent were obtained by each participant and his/her parents, respectively.
At Time 1, we recruited participants with the clinical diagnosis of ASD by senior child psychiatrists from the Children's Mental Health Center of National Taiwan University Hospital and by the corresponding author if referred by other hospitals or schools in the northern Taiwan. The diagnosis was based on the criteria of autistic disorder and Asperger's disorder defined in Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) (American Psychiatric Association, 1994). The TD participants were recruited at schools in the same districts of participants with ASD through the referrals by school teachers and principals. All the TD participants were clinically evaluated to ensure that they and their siblings did not have ASD traits or diagnoses. The clinical diagnosis of ASD of all the participants with ASD was further confirmed by using the Chinese version of the Autism Diagnostic Interview-Revised (ADI-R) (Gau et al., 2011). In addition, individuals with ASD, TD, and their parents received interviews using the Chinese version of the Kiddie epidemiologic version of the schedule for affective disorders and schizophrenia (K-SADS-E) (Gau et al., 2005) to confirm their current and previous co-occurring psychiatric diagnoses.
At Time 2, after providing assent and written informed consent again, all the participants and their parents were reassessed with the K-SADS-E interviews for DSM-5 diagnoses (American Psychiatric Association, 2013;Chen et al., 2017). Participants with ASD also received Autism Diagnostic Observation Schedule (ADOS) (Gau et al., 2011;Lo et al., 2019), and their parents reported on the ADI-R interview again. The corresponding author reviewed all the clinical evaluations and ADI-R and confirmed that all the ASD participants met the DSM-5 diagnostic criteria for ASD.
All the participants were assessed for the intelligence by the Wechsler Intelligence Scale for Children-Fourth Edition (Wechsler, 2003) if age < 17 and Wechsler Adult Intelligence Scale, Fourth Edition (Wechsler, 2008) if age ≥ 17 years old at Time 1, and attention profiles by CCPT at both Time 1 and Time 2. Parents also reported autistic symptoms on the Social Responsiveness Scale (SRS)  and ADHD/oppositional defiant disorder (ODD) symptoms on the Swanson, Nolan, and Pelham Rating Scale (SNAP-IV) (Gau et al., 2008) at Time 1 and Time 2. The participants were excluded from the study if their full-scale intelligence quotient (FIQ) was lower than 60 at Time 1 or if they did not complete the CCPT assessment either at Time 1 or Time 2.
A total of 245 participants with ASD and 194 TD had K-SADS-E interviews, SRS, SNAP-IV, and CCPT assessments at both time points. Figure 1 illustrates the flowchart of data selection at two time points. After deleting the subjects who had outliers in any CCPT index [4 standard deviations (SD) from the group mean] at Time 1, there were 221 participants with ASD and 182 TD (Supplementary Table S1-1). Demographic data of the subjects who had any CCPT outliers (outlier subjects) and those who did not (non-outlier subjects) are presented in Table S1-2. There were no differences in age, sex, and duration of follow-up between outlier subjects and non-outlier subjects. The outlier subjects in ASD or TD groups had lower FIQ and PIQ than non-outlier subjects. The outlier subjects had higher hyperactivity/impulsivity [F = 5.54, p = 0.02, degree of freedom (df) = 244] and more  Fig. 1 A flowchart of study design and data selection process. M males, F females, ASD autism spectrum disorder, TD typically developing controls, K-SADS-E Kiddie epidemiologic version of the schedule for affective disorders and schizophrenia, SNAP-IV Swanson, Nolan, and Pelham Rating Scale, SRS Social Responsiveness Scale, CCPT Conners' Continuous Performance Test. *Outliers: Any CCPT index 4 standard deviations away from group (stratified by sex and ASD diagnosis) mean. # Time 1 age ± 1 years and Time 2 age ± 2 years; case number of ASD vs. TD = 1 ~ 2: 1. $ Case number of ASD vs. TD = 1 ~ 2: 1. & n = 10, ASD = 4, TD = 6, all males severe social communication (F = 4.11, p = 0.04, df = 244) and social awareness problems (F = 6.39, p = 0.01, df = 244) than non-outlier subjects in the ASD group and higher ODD symptoms (F = 6.39, p = 0.01, df = 193) than non-outlier subjects in the TD group. Due to unequal distributions of age and sex between ASD and TD, we first matched ages of ASD and TD by age ± 1 year at Time 1 and age ± 1-2 years at Time 2, which generated the sample with a similar age distribution between the ASD and TD groups (Supplementary Table S1-3). Then we matched sex and deleted the subjects aged 20 years or older at Time 1, and the final matched sample for analysis was 154 and 106 in the ASD and the TD groups, respectively (Supplementary Table S1-4). The average durations of followup of the final sample [in years (SD)] were 6.88 (1.66) for the ASD group and 5.39 (1.62) for the TD group, respectively.
We further divided the final matched sample into the child (aged 6-11 years) and adolescent (aged 12-19 years) groups based on the ages at Time 1 to explore whether there were differences in developmental changes across childhood and adolescence. Finally, 106 children with ASD, 67 TD children, 48 adolescents with ASD, and 39 TD adolescents were included for data analysis (Table 1). While the whole final matched sample showed no statistically significant difference in age at Time 1 (supplementary Table S1-4), there was a significant age difference between ASD and TD for the child group.

Psychiatric Diagnoses of ASD, ADHD and Other Psychiatric Disorders
The ADI-R (Lord et al., 1994), a standardized, comprehensive, semi-structured, and investigator-based interview Table 1 Demographic data ASD autism spectrum disorder, TD typing developing controls, SD standard deviation # Group comparison adjusting for age in the child group § Medication use: psychotropic medications treating ADHD, tics, irritability, anxiety, or other psychopathology N (%) or mean (SD) Child group # Adolescent group ASD (n = 106) TD (n = 67) Statistics χ 2 or F (p) ASD (n = 48) TD (n = 39) Statistics χ 2 or F (p) of caregivers, is used to diagnose ASD. The ADOS (Lord et al., 2000) uses the investigator-based procedure to assist clinical diagnosis of ASD, including four modules: (1) Preverbal Children, (2) Children Who Have Phrase Speech, (3) Fluent Children, and (4) Fluent Older Adolescents and Adults. The Chinese versions of ADI-R and ADOS, which were approved by the Western Psychiatry Service in June 2007 and April 2008, respectively, have been widely used in research in the ethnic Chinese population (Fan et al., 2021;Lo et al., 2019;Tung et al., 2021). The Chinese versions of the K-SADS-E for DSM-IV (Gau et al., 2005) and DSM-5 (Chen et al., 2017) have been validated and widely used in our previous clinical [e.g., (Gau et al., 2010;Hearne et al., 2019;Shang et al., 2019Shang et al., , 2020] and epidemiological [e.g., (Chen et al., 2019a, b;Gau et al., 2005)] studies for diagnosing ADHD and other psychiatric disorders. The details of instrument preparation and interview training of the three instruments have been described elsewhere (Chen et al., 2017(Chen et al., , 2019aGau et al., 2005).

Social Responsiveness Scale (SRS)
The SRS, a 65-item scale, measures autistic symptoms quantitatively in the natural social settings for children and adolescents aged 4-18 years in the general population for the past 6 months. The items are rated on a 4-point Likert scale ranging from "0" (not true) to "3" (almost always true). The Taiwan Autism Study Group conducted two-way translations of this scale considering the culture-relevant expressions and then the Chinese version of SRS was approved by Dr. Constantino in 2008. The exploratory and confirmatory factor analyses revealed a four-factor structure (social communication, stereotyped behaviors/interest, social awareness, and social emotion) after 5 unclassifiable items were deleted . The Chinese version of SRS demonstrates good internal consistency (Cronbach's alpha, 0.94-0.95), test-retest reliability (intra-class correlations 0.75-0.85), and convergent validity . The higher scores indicate more severe autistic symptoms. It has been widely used in the community [e.g., (Chen et al., 2020;Tsai et al., 2019)] and clinical [e.g., (Chien et al., 2017a, b;Hsiao et al., 2013;Lo et al., 2019;Tung et al., 2021)] studies in Taiwan.

The Chinese Version of the Swanson, Nolan, and Pelham Rating Scale, version IV (SNAP-IV)
The SNAP-IV scale was developed to rate the ADHD (item 1-18) and oppositional defiant (item 19-26) symptoms based on the DSM-IV diagnostic criteria (American Psychiatric Association, 1994). It was rated on a 4-point Likert scale: 0 for "not at all" to 3 for "very much." The psychometric properties and norm of the Chinese version SNAP-IV were established (Gau et al., 2008(Gau et al., , 2009, and it has been widely used in many clinical [e.g., (Chiang et al., , 2020Shang et al., 2015)] and community [e.g., (Chen et al., 2017(Chen et al., , 2019a] studies in Taiwan.

The Conners' Continuous Performance Test for Windows II (CCPT-II)
The CCPT is a 14-min computerized task that assesses the attentional functions, using the non-X type go/no-go paradigm (Hasson & Fine, 2012). The signal-to-noise ratio is high, including 90% target stimuli and 10% non-target stimuli in 360 trials. The whole task consists of six blocks and three sub-blocks. Each sub-block has 20 trials. The participants are asked to tap the space tab in all letters presented on the computer screen, except letter "X", and each letter appears on the screen for 250 ms in a random sequence. Three different inter-stimulus intervals (ISI), i.e., 1, 2, 4 s, are randomly assigned to the sub-blocks of each block. CCPT-II generates 12 indices: (1) omission, the number of non-responses to the targets; (2) commissions, the number of responses to non-targets; (3) hit reaction time (RT), the response time of correct responses; (4) hit reaction time standard error (RTSE); (5) variability, intra-individual variability of RT; (6) detectability, the ability to discriminate between targets and non-targets and the higher the score indicating the better performance; (7) response style (β), a tradeoff between speed and accuracy and a higher score indicating a conservative style; (8)

Statistical Analysis
We used SAS 9.4 (SAS Institute Inc, Cary NC, USA) for statistical analysis. The whole sample was divided into the ASD and the TD groups stratified by age. We compared continuous and categorical variables between ASD and TD for the child and the adolescent groups at Time 1 using the general linear model and Chi-squared test (or Fisher's exact test if needed). The paired t-test was used to compare the differences between Time 1 and Time 2 in each subdomains of autistic and ADHD symptoms, and CCPT measures. The mixed model was used to compare repeated measures, estimate slope over time, and group*time interactions after adjusting for the duration of follow-up. Pearson's correlations were calculated for the relationship between autistic, ADHD symptoms, and CCPT measures. A significant level was preselected as p < 0.05 in regressions and p < 0.01 in Pearson's correlation analysis due to multiple comparisons. Cohen's d was used to calculate the effect size. We also conducted the sensitivity analysis using the data without matching (Supplementary Tables S1-1, S2-1, S5-1, S5-2), only matching for age (Supplementary Tables S1-3, S2-2, S6-1, S6-2), and matching for age and sex (without age stratification) (Supplementary Tables S1-4, S2-3, S7-1, S7-2). Table 1 presents the demographic data. Participants with ASD were older than TD participants at Time 2 in both child and adolescent groups. There was a longer followup duration in ASD than TD. The IQ profiles were lower in ASD than TD. Paternal educational level was lower in children with ASD than TD children without differences in maternal educational level. No significant ASD-TD difference in paternal/maternal educational levels in the adolescent groups. Children and adolescents with ASD were more likely to have co-occurring psychiatric disorders, mainly ADHD and anxiety disorders (Table 2), and more likely to use psychotropic medications than TD (Table 1).

Autistic, ADHD/ODD Symptoms
The autistic (SRS) and ADHD/ODD symptoms (SNAP-IV) were significantly more severe in ASD than TD at Time 1 and Time 2 ( Fig. 2A-G). Autistic symptoms were relatively stable over time for the four groups. With time, we found a modest improvement in social communication (Cohen's d = − 0.37) and decreased stereotyped behaviors (Cohen's d = − 0.41) in the adolescent ASD group. There were significant reductions of ADHD/ODD symptoms over time in almost all dimensions for the four groups. There were no significant changes of inattentive symptoms in the child TD group and ODD symptoms in the adolescent ASD group. The only significant group* time interaction was noted in hyperactivity/impulsivity, with a greater magnitude of symptom reduction in children with ASD than TD children (p < 0.001, Fig. 2F). There was no other group*time interaction of autistic symptoms and ADHD/ODD symptoms in children or adolescents.

Attentional Performance Assessed by CCPT
The results of CCPT are shown in Fig. 3A-J (see Supplementary Tables S3-1 and Table S3-2 for details of analytical results of ASD-TD group comparisons for the child and adolescent groups, respectively, and Tables S4-1, S4-2 for the slope of changes and group*time interactions). Children and adolescents with ASD performed significantly worse than their TD counterparts in omission and perseveration at Time 1 and Time 2. Children at Time 1 and adolescents at Time 2 showed significant ASD-TD differences in RT. There were significantly more commission errors of ASD than TD in children at Time 2 and in adolescents at Time 1 and Time 2. Children with ASD   had significantly larger RTSE than TD at both time points without significant ASD-TD group differences in adolescents. Children with ASD at Time 1 and Time 2 and adolescents with ASD at Time 1 had significantly higher variability than their TD counterparts. There was a significant ASD-TD difference in response style at Time 1 in the child group. We found a significantly higher RTSEISI in ASD than TD in children only at Time 2 but not found in the adolescent group (Fig. 3J). There was no group difference in RTSEBC, regardless of age groups and time (Fig. 3I). Children with or without ASD had significant improvements in almost all indices in CCPT over time (Fig. 3, Supplementary Table S3-1) with the following exceptions. No improvement was noted in RTSEBC in children with ASD and in response style in TD. Adolescents with ASD had significant improvements in commission errors, RTSE, and perseveration, while TD adolescents significantly improved in omission, commission errors, variability, and detectability over time (Fig. 3, Supplementary  Table S3-2).
Group*time interactions were noted in RT and response style in children (ASD had a greater magnitude of improvement than TD, Fig. 3C, G, Table S4-1) and in commission and detectability in adolescents (ASD had a smaller magnitude of improvement than TD) (Fig. 3B, F, Table S4-2).

Correlations of CCPT and Autistic/ADHD-Related Symptoms
There were high correlations between autistic symptoms and ADHD/ODD symptoms in ASD and TD at Time 1 and Time 2 (Table 3). Time 1 SRS and SNAP sub-scores significantly predicted corresponding Time 2 SRS and SNAP sub-scores (all p < 0.01, data not shown) in ASD but not in TD. While separating ASD and TD for correlation analysis, neither autistic (SRS) nor ADHD/ODD symptoms (SNAP-IV) were correlated to the CCPT performance in ASD and TD at Time 1 or Time 2 (all p > 0.05, data not shown).
In children, ADHD/ODD symptom changes were correlated to changes of social communication impairment and stereotyped behaviors assessed by SRS in ASD or TD (Table 3). In adolescents, only the changes of ODD symptoms were associated with changes of social communication impairment assessed by the SRS in ASD and TD (Table 3). No significant correlations between the changes of CCPT and changes of ADHD/ODD/autistic symptoms were found (all p > 0.05, data not shown).
The associations of omission, RT, RTSE, variability, and response style with FIQ were found only in children with ASD (data not shown, all p < 0.01) but not in adolescents with ASD or TD. FIQ was not correlated with ADHD symptoms or autistic symptoms of SRS at Time 1 or Time 2 in all four groups (data not shown, all p > 0.05).

Main Findings
This work was one of the few longitudinal studies reporting developmental changes of attention performance assessed by CCPT and autistic/ADHD/ODD symptoms for the children and adolescents with and without ASD. We found that ADHD-related symptoms decreased and CCPT performance improved with time from childhood to young adulthood, especially during childhood to adolescence. Generally speaking, autistic symptoms persisted over time. Yet, there were modest improvements in social communication and stereotyped behavior in adolescents with ASD. Individuals with ASD still had significantly more severe autistic/ADHD/ODD symptoms than TD in late adolescence/young adulthood. Children with ASD had significantly more RT and response style improvements than TD, and adolescents with ASD had significantly less improvement in RT and detectability than TD. Despite the improvements, there were significant ASD-TD differences in omission, commission, detectability, and perseveration, regardless of age groups. Our study did not find any linear relationship between changes in CCPT and ADHD/ODD or autistic symptoms.

Developmental Changes of Autistic/ADHD-Related Symptoms and their Correlations
Consistent with previous reports, our study also showed that autistic symptoms were relatively stable from childhood to adolescence (Robinson et al., 2011;St. Pourcain et al., 2011). On the other hand, we found that adolescents with ASD had decreased social communication impairment and stereotyped behaviors/interests with time from early adolescence to late adolescence/young adulthood. Our finding corresponds to other studies which showed reducing severity of autistic symptoms in late adolescence or adulthood (Magiati et al., 2014) and even up to midadulthood in individuals without intellectual disabilities (Howlin et al., 2013). Baseline autistic symptoms were highly predictive of later autistic symptoms in ASD, during childhood or adolescence, which also echoed the stability of ASD trait (Howlin et al., 2013;McGovern & Sigman, 2005;Szatmari et al., 2009). Intelligence or cognitive functions were well-known predictors of adaptation functions/adult outcomes in ASD (Hartman et al., 2016;Magiati et al., 2014). Nevertheless, in our study, neither IQ nor attentional functions was associated with autistic symptoms at Time 2 or improvement of autistic symptoms. Different outcome measures might explain the inconsistent findings, and the relatively high IQ of our participants might also influence the predictive effect of IQ (Howlin et al., 2013;Magiati et al., 2014;Szatmari et al., 2009). Not surprisingly, ADHD symptoms in ASD declined over time, which was similar to developmental changes in ADHD alone and TD .
The autistic symptoms were significantly correlated to ADHD/ODD symptoms, as shown in many previous studies in ASD (Leitner, 2014) or the general population (St. Pourcain et al., 2011). Although autistic symptoms were largely persistent, the changes in autistic symptoms were significantly correlated to ADHD/ODD symptoms changes, especially from childhood to adolescence. Significant correlations were found between changes in ODD symptoms and changes in social communication from adolescence to young adulthood. The high rate of co-occurrence of autistic and ADHD/ODD symptoms and high correlations of developmental changes of these symptoms implied that general factors might contribute to these disorders occurrence and development of symptoms (Stergiakouli et al., 2017). On the other hand, symptoms of one disorder might influence the outcome of the other disorders interactively. For example, inattention and hyperactivity might hinder the development of social communication skills, and poor social communication skills might further contribute to oppositional attitude and increased irritability (ODD symptoms). A recent longitudinal study echoed our findings, showing that reduced inattentive symptoms at follow-up in ASD were correlated to improved social skills (Zachor & Ben-Itzchak, 2020).

Developmental Changes in Attentional Performance
Children and adolescents with ASD and TD counterparts demonstrated significant improvements in their CCPT performance, similar to what were found in the communitybased study . Across childhood and adolescence, individuals with/without ASD both have improved alertness and consistency (shorter and less variable RT), improved selective attention and response inhibition (fewer omission and commission errors), increased detectability (d′), higher tendency to response (lower β) and improved sustained attention (RTSEBC) . Especially, children with/without ASD had significantly greater magnitudes of improvements than adolescents, which might reflect a typical developmental trajectory in the attentional functions through childhood and adolescence  and also could be the ceiling effects of CCPT through adolescence to adulthood (Lin et al., 1999). We specifically found that children with ASD showed a greater magnitude of decreased RT and decreased β (increased tendency to respond) than TD children over time. The decreased RT (only in the correct responses) in CCPT implies improved tonic alertness and early information processing speed (Nigg, 2005). Our study demonstrated a catchup development of tonic alertness/early information processing speed in ASD from childhood to adolescence, followed by a stationary developmental course to young adulthood in ASD and TD. There was conflictual data concerning the processing speed of ASD: some reported a slow processing speed (Travers et al., 2014), and others stated that the slow processing speed was due to inattention in ASD (Kramer et al., 2020). Since we did not find a correlation between inattentive symptoms and CCPT RT, the deviant development of tonic alertness and processing speed in ASD was not only due to the co-existence of inattentive symptoms in ASD.
The significant increase of response tendency (decreased β) over time was only noted in the child ASD group but not in the other three groups, which showed no change of response style over time. Conflictual data existed concerning the normal developmental changes of response style of CCPT. One reported a developmental growth toward a tendency to respond through childhood to adolescence  while others did not (Lin et al., 1999;Miranda et al., 2008). According to the signal detection theory, β refers to the likelihood of hitting the signal trials over the likelihood of hitting the noise trials. A higher β referred to a higher internal threshold and a tendency to respond conservatively to avoid commission errors (Stanislaw & Todorov, 1999). A significant decrease in β, in conjugation with a significant decrease in RT might further indicate that children with ASD had catch-up in tonic alertness and assertive responding. However, the signal discrimination and consistency of response in ASD were still impaired throughout childhood. Furthermore, adolescents with ASD showed less improvement in commission errors and signal detectability than TD, implying that signal discrimination and response inhibition were trait markers in ASD throughout childhood and adolescence, not just a developmental delay.

Correlations Between ADHD Symptoms and Attentional Performance in ASD
Out of our expectation, CCPT performance was not significantly correlated to ADHD symptoms (and autistic symptoms, either) in ASD or TD dimensionally, since this attention performance was known to be substantially correlated to ADHD symptoms in ADHD . Also, the developmental change of attention performance was not associated with autistic and ADHD symptoms changes in ASD in a linear pattern. However, these ASD and ADHD symptoms and attentional impairments tended to aggregate together in a categorical pattern. Based on the finding of significant correlations between inattention and social impairments at an early age (as early as 1.5 years), inattention might be the early common problem of ASD and ADHD (Visser et al., 2016). However, it was unclear whether the overlapped inattention phenotype in ASD and ADHD was based on the same underlying cognitive/neural mechanisms (Visser et al., 2016). Besides, the inattentive symptoms observed in ASD might not be completely the same as in ADHD, considering the underlying contexts (Visser et al., 2016). A recent study revealed differential correlations of sensory processing and inattentive symptoms in ASD and ADHD (Dellapiazza et al., 2021). Our study's lack of correlation between ADHD symptoms in ASD and attentional functions also implied different neural mechanisms underpinning attentional problems in ASD and ADHD.
On the other hand, as previous studies stated (Lundervold et al., 2016;Munkvold et al., 2014), our study also showed that the performance of CCPT was significantly correlated to the general intelligence in children with ASD but not so prominent in adolescents with ASD and TD. Thus, although more evidence was required, different executive processes, such as attentional problems, inhibitory control, and intelligence, might share a common genetic basis (Polderman et al., 2009), at least during childhood.

Limitations
Our study should be interpreted with some limitations. First, our participants were recruited in clinical settings in the metropolitan region, so our results could not be generalized to the general population and rural areas. Second, autistic symptoms in this study depend on the subjective questionnaires of SRS, but not the objective measures, such as ADOS, and might be suffered from the bias of different social expectations of participants' parents. Third, we did not control for the diagnosis of ADHD and could not evaluate the influence of ADHD categorically. The underlying mechanisms of ADHD symptoms in individuals with ASD might not be entirely the same as those without ASD [e.g., response to stimulants (Leitner, 2014) and brain imaging results (Christakou et al., 2013)]. This study aimed to observe the developmental changes of dimensional ADHDrelated symptoms and attentional function in the scope of the whole sample of ASD instead of the influences of ADHD diagnosis. Fourth, some of our participants had ever taken ADHD medication, which might influence the results of CCPT performance. However, we found that medication use was correlated to higher inattentive symptoms in children with ASD, which implied a self-selected bias in medication use. We asked the participants to discontinue medication 24 h before the evaluation and found no correlation between CCPT performance and medication use. Fifth, we did not collect detailed information on the participants' nonpharmacological interventions or service uses. Therefore, we could not evaluate the influences of these interventions on the developmental changes of autistic/ADHD symptoms or attention performance. Sixth, children and adolescents with ASD had a longer duration of follow-up. We tried to resolve this problem by matching age and sex and controlling for follow-up duration in the group*time interaction analyses. Seventh, only a small number of girls were in our study, and therefore the results could not be generalized to girls' conditions. Further studies with more precise age-matched and equal sex samples would be required. Eighth, the interaction between social communication and ADHD symptoms is not under the scope of this paper but will be the next step of our future research with a larger sample size.

Conclusion
ADHD-related symptoms and attentional functions improved but remained impaired in children and adolescents with ASD when they entered late adolescence and young adulthood, while autistic symptoms were relatively stable. Social communication impairment was significantly correlated to both ADHD and ODD symptoms in children and only ODD symptoms in adolescents. Children with ASD showed greater improvement in tonic alertness/processing speed and assertive responding than TD over time, while adolescents with ASD had less improvement in response inhibition and signal detectability than TD over time. There might be a deviant developmental course in the laboratory attentional functions in ASD as compared to TD. Developmental changes of laboratory attentional functions were not linearly correlated to developmental changes of ADHDrelated and autistic symptoms in ASD, although they often clustered together. ADHD symptoms in ASD might have different underlying neural mechanisms from those in ADHD. Further longitudinal studies, including brain imaging and genetic research, and clinical intervention studies of the effects of managing ADHD-related symptoms on the developmental changes of ASD symptoms, will help elucidate the mechanisms underlying behavioral symptoms of these two frequently co-occurring neurodevelopmental disorders.