Our research considered a large population of Italian patients to identify the ASD early signs and took into account some correlations among variables, which have never been reported.
Sample and age at onset of symptoms
On account of the extreme etiological variability in ASD, we considered both idiopathic and non-idiopathic cases diagnosed according to DSM-IV-TR criteria [
22]. With respect to age at onset of ASD, we found that 41.9% of the cases presented the first symptoms between 7 and 12 months and 27.6% between 13 and 24 months. Notably, there were no significant differences for this variable in relation to the diagnosis according to DSM-IV-TR (see the comparison between AD and PDDNOS), etiology (see the comparison between idiopathic and non-idiopathic cases), early onset epilepsy, and IQ/DQ level. These data confirm that regardless of the etiology or ASD type and the comorbidity with ID and early onset epilepsy, the age at onset may shift more frequently between 7 and 24 months with a peak around 7 and 12 months, even though it is possible to recognize an earlier onset between 0 and 6 months in 21.9% of the cases. In the literature, the data on the age at onset of early signs is extremely variable. The similar age at onset regarding the different ASD types and the IQ/DQ levels were reported also in the review of Daniels and Mandell [
24].
Prospective studies are consistent with retrospective studies in finding that for many ASD children, symptoms emerge gradually over the first 18 months or so of life.
We did not consider variables such as higher socioeconomic level, greater parental concern, healthcare, and education, which may influence the age at earlier ASD diagnosis in our study but we wish to underscore the importance of promptly recognizing early signs.
The lack of significant differences for age at onset between AD and PDDNOS indirectly confirms what DSM-5 now takes into account: ASD represents a disorder without the need to consider the subgroups previously reported in DSM-IV-TR.
Early ASD signs and age at onset in the whole sample according to the five categories of clinical features
Social interaction and relationships (93.3%) and language (92.4%) are the categories of early signs most represented in our sample, which confirms the literature data [
20].
Delay in spoken language (intended as both verbal production and verbal comprehension) represents one of the most common (even though not specific) symptoms prompting initial medical consultation for a possible ASD diagnosis.
Categories of stereotyped behavior and activities, motor skills and regulation (feeding/sleep disorders) recurred less frequently (78.1% vs 57.1% vs 43.8/32.4%, respectively); nevertheless, they are equally important. Although repetitive behaviors and interests have been reported in ASD children, little data exist on how these symptoms manifest in early development. In our sample, the most represented subcategories of stereotyped behaviors and activities were stereotypies (45.7%), psychomotor agitation/aggression (44.8%), and atypical games (41.9%). The average age at onset of stereotypies and psychomotor agitation were 24 and 19.9 months, respectively. Hypoactivity recurred in 16.2% of the cases, in 70.6% of which it was evident during the first 24 months of life. Little information is available from retrospective studies, but the literature reports that repetitive behaviors do not appear until after the first birthday [
21].
Regarding motor skills, although the subgroup gross motor developmental delay recurred in 43.8% of the cases and it was the most present one, we want to underscore the importance of also finding such conditions as absence of sucking reflex, neonatal hypotonia, hypotonia in the first 3 years of life, fine motor developmental delay, and tonic dialogue disorder early on.
Finally, with regard to the regulation category, feeding problems and sleep disorders may represent a warning early sign suggesting an ASD as it has also been reported in the literature so far [
14‐
17].
In conclusion, our study offers new data on account of the methods we have used (systematic search for correlations among early clinical features and later clinical-instrumental findings) and the consistent sample size we have considered. It is not easy to compare our results with the literature data concerning retrospective studies for methods utilized and sample sizes considered [
21,
25]. Larsen and collaborators reported six symptoms differentiating ASD children from typically developing children; this retrospective study examined children who attended day care centers at 12–24 months of age. Our data partly overlap with this study, excluding joint attention impairment which is less frequent in our sample [
26]. This discrepancy is perhaps due to the different data source, namely parents in our study, day care personnel in Larsen and collaborators.
Early signs, age and mode at onset of ASD considering all variables
Our results enrich the literature on the subject with new data; in fact, for the first time, many variables have been considered comparatively.
Regarding the onset age, as we have mentioned above, it is interesting to underscore that there were no significant differences between AD and PDDNOS, idiopathic versus non-idiopathic cases, among IQ/DQ levels, and when epilepsy started before 3 years of age. Once an early diagnosis of ASD is made, it is necessary to do some exams for the etiologic diagnosis for example to distinguish idiopathic and non-idiopathic cases. The identification of a specific condition underlying ASD, as already reported by our group, could be useful to provide medical treatment and to advise family on the ASD recurrence risk [
27].
In our sample, AD starts with a developmental delay or regression rather than stagnation (
p = 0.047). Motor skill disorders were more frequent in patients with age at onset between 0 and 6 and 7–12 months. This finding confirms literature data regarding motor delays and motor impairments that are quite common in ASD [
28].
In our study, 60.4% of the patients with delay and 59.3% of the cases with regression as onset modes presented an AD, while only 33.3% of the subjects with stagnation had an AD, the difference was significant (p = 0.047). These data, not yet reported in literature, are not easy to explain even if we may suppose that a delay or regression represent something serious more evident in cases with AD rather than in PDDNOS.
Regarding idiopathic and non-idiopathic cases, and patients with and without early onset epilepsy, we found no differences regarding the onset mode. Note that the awake and sleep EEG recording, performed in all patients, both in those with epilepsy and in those without epileptic seizures, during the follow-up excluded in all cases a condition of ESES, which is implicated in the appearance of a heterogeneous spectrum of developmental disorders [
29,
30]. However, awake and sleep EEG was not available at the onset of ASD. Only for IQ/DQ, it is relevant to point out that the cases without ID were affected by a stagnation of development at onset more frequently than a delay or a regression (
p = 0.037). Patients with a severe/profound ID, instead, more frequently had, at onset, a delay or a regression of development (
p = 0.016). These data are important and suggest that the cognitive level may affect the mode at onset.
With regard to the early signs and their categories, it is interesting to note that language signs were less frequent in cases with regression (81.5%) than in cases with other onset modes (p = 0.046); while motor skill disorders as early signs prevailed in cases with a delay at onset (72.9%) (p = 0.0068791). Feeding problems were more frequent in cases with delay and stagnation of development at onset than in cases with regression (p = 0.031).
All these data, never reported in the literature so far, may contribute to recognizing and tracing an early trend of the pathology.
Age at onset and onset mode seemed to be somewhat correlated with each other: age at onset between 0 and 6 months was significantly more frequent in the cases showing delay and stagnation of development at onset (
p = 0.0054580), whereas age at onset between 13 and 24 months prevailed in the cases with regression of development at onset (result near to significance:
p = 0.057). Age at onset between 25 and 36 months prevailed significantly (
p = 0.040) in cases with stagnation and regression at onset, while the two cases with age at onset between 37 and 51 months showed a regression of development (
p = 0.053). These last two cases did not have a childhood disintegrative disorder according to DSM-IV-TR [
22].
This study presents a few limitations, the most important of which is its retrospective nature. In addition, it was carried out when DSM-IV-TR considered ASD as subdivided into different categories. The sample that we examined may represent a selected population, perhaps more clinically compromised coming to an Italian university centre for Autism. This study did not consider a control population. The sample was examined during a period of 10 years and 6 months, during which the literature increasingly reported data concerning the diagnosis, etiopathogenesis and evolution of autism, leading to significant progress in instrumental examinations: see, for example, the increasingly widespread use of array CGH - Comparative Genomic Hybridization, and secondly of exome sequencing, in the diagnostic work-up of these patients, in view of the important genetic component in the etiology of ASD, also for the purpose of a genetic counseling for the family.
More importantly, this study has several significant strengths. It considers a large sample coming from Italy. In this respect, we wish to underscore the importance of considering the location to improve knowledge of regional variations in age at diagnosis. It consists in an examination of data that were always collected in a consistent way, by the same group of child neuropsychiatrists at the Autism Centre of the University of Bologna, who used the same anamnesis checklist for all the variables considered regarding the early signs. The variables analyzed are completely new. A similar systematic evaluation of data is not reported in the literature and our results contribute to a better definition of the onset mode in relation to the categories of early symptoms, level of IQ/DQ, and presence of early onset epilepsy.
Autism, ID, and epilepsy are highly comorbid, which suggests shared etiologies [
7,
31]. In most cases ASD etiology is still a mystery. Recently, Casanova and collaborators studied ID with a known molecular origin by considering ID with high, modest or absent risk of autism or epilepsy, and reported that ID with high rates of ASD comorbidity is present with a particularly homogenous genetic profile. They also reported that genes with high penetrance for syndromic and non-syndromic ASD are localized in the nucleus and are connected in transcription regulation [
32]. A correlation among these results and early ASD symptoms is unknown.