Developmental regression in children: Current and future directions

Developmental regression describes when a child loses previously established skills, such as the ability to speak words and is most recognised in neurodevelopmental conditions including Autism; Developmental Epileptic Encephalopathies, such as Landau Kleffner syndrome, and genetic conditions such as Rett syndrome and Phelan McDermid syndrome. Although studies have reported developmental regression for over 100 years, there remain significant knowledge gaps within and between conditions that feature developmental regression. The certainty of evidence from earlier work has been limited by condition-specific studies, retrospective methodology, and inconsistency in the definitions and measures used for classification. Given prior limitations in the field, there is a paucity of knowledge about neurocognitive mechanisms, trajectories and outcomes for children with developmental regression, and their families. Here we provide a comprehensive overview, synthesise key definitions, clinical measures, and aetiological clues associated with developmental regression and discuss impacts on caregiver physical and mental health to clarify challenges and highlight future directions in the field.


1.
Background: what is developmental regression?Developmental regression describes a child losing their ability to use previously established skills.For example, a child who was regularly saying single words and then stops speaking, or a child who was using gestures and making regular eye contact loses social responsiveness.Developmental regression may be an early feature of a neurodevelopmental condition, such as Autism Spectrum Disorder (Autism), a monogenetic condition, Developmental Epileptic Encephalopathy (DEE), or much more uncommonly be part of (alongside abnormal neurological signs) the onset of a progressive deteriorating condition causing childhood dementia (CD) or progressive intellectual and neurological deterioration (PIND).Consequently, there is a need for time-critical medical review to identify or exclude an underlying disorder and initiate best practice management.
Neurodegenerative conditions are beyond the scope of this paper, but for more information, the British Paediatric Surveillance Unit has collected data on PIND conditions since 1997 (Verity et al., 2021).Regression that may also be associated with nutritional deficiencies or developmental decline post brain insult or injury (On et al., 2022), such causes are also beyond the scope of this paper.Rather, this update focuses on neurodevelopmental conditions such as Autism and Childhood Disintegrative Disorder (CDD); genetic conditions such as Rett (RTT) and Phelan McDermid Syndrome (PMS); and DEE conditions, including Landau Kleffner Syndrome (LKS) and Continuous Slow Wave Sleep disorder (CSWS), all of which may present with developmental regression.
Despite interest and research focusing on regression spanning decades, there remain significant knowledge gaps about the causes, presentations, trajectories, and outcomes of children with developmental regression.Various mechanisms for developmental regression have been proposed including potential associations with pro-inflammatory immune responses (Vargas et al., 2005), and familial risks with associations between developmental regression and a family history of autoimmune conditions such as Diabetes Mellitus type 1 and auto-immune thyroiditis (Scott et al., 2017), celiac disease and rheumatoid arthritis (Thurm et al., 2014).Furthermore brain plasticity impacts autism onset and is dependent on developmental processes like synaptic pruning (Thomas et al., 2011), with superimposed influences of phenomena such as epileptiform activity or infection (Tuchman & Rapin, 1997), and these neuro-cortical process may be influential in progressive and regressive development.
The difficulties in understanding causes of developmental regression imposed by such biological complexity are worsened as methodological limitations, including lack of an agreed definition ( van Karnebeek & Stockler, 2012;Zhang et al., 2019;Zwaigenbaum, 2019a) and inconsistent use of measurement tools, as well as retrospective and conditionspecific study designs.Such limitations reduce the certainty of available onset and life course evidence and application to clinical care.At the same time, there has been relatively little focus on the impact of developmental regression on caregivers and other family members.What literature does exist suggests that caregivers who observe skill loss in their children experience more stress, anxiety, and emotional loss compared to parents of children without loss (e.g., Ellis et al., 2022).Given the traumatic nature of witnessing developmental regression and the uncertain developmental trajectories of affected children, clinical supports should include family-centered care to ensure that caregivers' mental and emotional needs are addressed, as well as the child's needs.
Here, we aim to review transdiagnostic processes involving developmental regression across a range of neurodevelopmental conditions, with a focus on neurocognitive mechanisms.We suggest future directions for clinical translational endeavours and global collaborative initiatives.

2.
Different terms, definitions and measures A clear, detailed and concise definition remains elusive within and between conditions that feature developmental regression.Table 1 provides examples of definitions used in studies with a focus on developmental regression across a number of conditions.As shown, there are similarities and differences in key features which impact upon participant inclusion and the characteristics of children assessed.This, in turn, will create between study variation in findings, hampering emergence of a clear understanding of prevalence, cause, prognosis or effective interventions for studies exploring these important clinical factors and outcomes.
A commonly used definition in studies with a focus on regression in autistic children is a loss of spoken language following an initial word acquisition phase of 3e5 words, however there are also reported proto word losses that occur before the 3e5-word stage that may be missed if a definition is reliant on loss of 3e5 words.For example, Baird et al., (2006) used a "definite language regression" (loss of 5 words used communicatively for 3 months before loss) and a "lower level" (stasis or loss of words or babble where the child had not reached the 5-word stage or there was regression in skills other than language) definition.Interestingly the different definitions did not appear to impact outcome results, however authors acknowledge limitations of a relatively small sample size and reliance on retrospective methodologies.Richler et al., (2006) used a broader definition than specified in the Autism Diagnostic Interview Revised (ADI-R), defining word loss as the spontaneous use of at least 3 meaningful words (other than mama and dada) daily for at least 1 month then then stopped using all words for at least 1 months prior to 36 months.They reported higher impairment scores for social reciprocity in the autistic children with regression compared to autistic children without regression.When Goin-Kochel et al., (2014) compared prevalence of regression in a cohort of autistic children using the ADI-R with a loss supplement form to detect subtle skill loss, an additional 11.7% of children with subthreshold language loss were identified.Furthermore, many studies focusing on developmental regression use study-unique definitions with variations in key features, such as age of onset of skills loss, duration of loss, and types of lost skills and inconsistently consider early developmental abilities and longer term outcomes or trajectories (Bernabei et al., 2007;Jones & Campbell, 2010).
For other conditions such as Rett syndrome even when the definition for the condition is agreed there is variation in how regression within each condition is defined.For example, in Rett syndrome there are core and supportive consensus criteria (Neul, 2010) that include regression, but how regression is defined is variable as shown in Table 1.
Similarly for Phelan McDermid syndrome (PMS) developmental regression is recognised but variably described and defined as shown in Table 1.
There are different tools used to measure developmental regression, and they are inconsistently used in research and rarely used routinely in clinical care.Table 2  c o r t e x 1 6 9 ( 2 0 2 3 ) 5 e1 7 specific measures and illustrates the similarities and differences between measures.Importantly, some measures gain information about pre-loss abilities, timing (onset of losses and duration of loss) and specify lost development domains or skills.The Developmental and Neurobehavioural Regression questionnaire (DANR) (Frye et al., 2020), and Regression Supplementation Form (RSF) (Goldberg et al., 2003) also record whether there was a 'trigger' or concurrent event at the time of the developmental regression.

lists regression
Concerns that reliance on the Autism Diagnostic Interview-Revised (ADI-R), which is being used more widely than any other tool to capture regression and autistic behaviours, may miss more subtle skill losses (Goin-Kochel et al., 2014;Jones et al., 2015;Prescott & Ellis Weismer, 2021), has led to some studies expanding the ADI-R to use less stringent criteria, asking additional questions, and/or supplementing with other measures, such as home videos to capture skills data.To the authors knowledge there have not been any direct comparisons or critiques of regression specific measures (e.g.RVI, RSF or DANR) to compare sensitivity for detecting developmental regression in children with and without autism.Soorya et al., (2013) studied regression in children with Phelan-McDermid Syndrome and found 28% (N ¼ 9) of children had loss of skills based on psychiatric evaluation but none of these children were coded as having experienced regression when the ADI-R definition was used for regression because none of the participants had reached the 5-word phase of language development.
There has been interest in capturing data about skills loss and gains using home videos of children since earlier work in the 2000's (Werner, 2003;Maestro, 2006).A recent systematic review of home video use for early development and regression in autistic children (Kaufman, Cox, Veronica, Williams, & Ure, 2022) found that, of the 12 included studies, each study used a different definition for regression, and none explained a rationale for their definition choice.The studies used a heterogenous mix of methods to measure regression, including quantifying regression as decreasing the frequency of behaviours per length of home video (e.g.Maestro et al., 1999), or the onset of ASD-related symptoms (Maestro et al., 2006).Studies differed in their reporting of key features, for example, whether they described age of onset, types of skills studied, and the longitudinal progression of regression of skills in autistic children.Data about early development before regression was inconsistently asked or reported.Findings from studies that did consider pre-loss abilities were variable with some studies reporting advanced abilities before regression (Ozonoff et al., 2011;Werner & Dawson, 2005), some reporting delayed skills prior to regression (Bernabei, Cerquiglini, Cortesi, & D'Ardia, 2007;Maestro et al., 2006) and others reporting typical pre-regression development (Osterling et al., 2002).Meta analysis of results was not possible due to the

Incidence and prevalence of developmental regression
Inconsistency in terms, definitions and measures used has hindered precise estimates of the incidence and prevalence of developmental regression.Within specific conditions, prevalence and incidence statistics vary depending on the definition and data collected.Table 3 reports best-prevalence estimates and phenotypic data for the disorders commonly associated with developmental regression.Fig. 1 represents key features and typical age reported for onset of regression for historically described conditions featuring developmental regression.These are based on a synthesis of the literature.
Barger and colleagues' (Barger et al., 2013) meta-analysis of regression in autistic children reported that at least 30% lose some skills, mostly in speech and language.An update of Barger et al.'s review (Tan et al., 2021) included 75 studies with 33 000 autistic participants who regressed and also reported a pooled proportion of autistic regression of 30%.High heterogeneity and varied prevalence were identified according to the type of skills lost, with higher proportions regressing in language/social skills, and adaptive functioning outcomes years later.

Developmental regression in autistic children
Most descriptions of autistic children who experience developmental regression report skill loss between 18 and 24 months of age (Thompson et al., 2019) with an average age of 20e22 months (Tan et al., 2021).There are also reports of loss of skills at later (preschool and school) ages (Boterberg et al. 2019).
Thomas Heller first reported regression of abilities in typically developing children aged 3 and 4 years in 1908.This sudden and dramatic loss of skills was originally referred to as dementia infantalis.More than 100 years later, descriptions remain limited to studies that focus on speech and language loss (Thurm et al., 2014), and largely relies on retrospective caregiver reports and sometimes video data, as described above.Retrospective data has risks of forward telescoping and recall bias, compounded by the other methodological inconsistencies, described above (Goldberg, 2008;Luyster et al., 2005;Pearson, Charman, Happ e, Bolton, & McEwen, 2018).
Autistic children typically regress in spoken language and/ or social and communication abilities.Although some have c o r t e x 1 6 9 ( 2 0 2 3 ) 5 e1 7 questioned the existence of regression in autism, recent prospective studies of children with autism report that most autistic children are predicted to experience some degree of loss (Ozonoff & Iosif, 2019).Autistic children who experience developmental regression demonstrate lower cognitive abilities compared to autistic children without regression (Martin-Borreguero et al., 2021).Reports indicate an increased prevalence of intellectual disability (Oslejskova et al., 2008;Thompson et al., 2019;Tuchman & Rapin, 1997), lower mean performance on receptive and expressive language outcomes (Bernabei et al., 2007), and sleep disorders (Giannotti et al., 2008).However, the outcomes for autistic children who experience developmental regression is highly varied, making individualised support and planning difficult.
Childhood disintegrative disorder (CDD) is a rare diagnosis, with prevalence estimates ranging from 1.1 to 9.2 per 100,000 (Mehra et al., 2019).CDD is characterised by the loss of developmental and adaptive skills, in addition to the presence of autistic features, in a child with apparently typical development for at least the first 2 years of life (American Psychiatric & American Psychiatric Association, 2000).CDD and autistic disorder were separate categories of PDD under DSM versions III to IV-TR.However, in DSM 5, these separate categories have been subsumed under the single diagnostic label of ASD.
A recent systematic review by Mehra and colleagues (Mehra et al., 2019) found that children with CDD and autistic children experiencing regression share many core-diagnostic and extra-diagnostic features, including social-communication challenges and repetitive behaviours.Additionally, the authors reported that outcomes related to the extent of core autistic features and some adaptive skill abilities were comparable in CDD and autism with regression.
However, Mehra and colleagues (Mehra et al., 2019) also highlighted key differences in the nature of regression between the two conditions.These differences suggest that developmental regression in CDD may be considered distinct from regression in autism.First, loss of skills affects a broader set of domains in CDD compared to autism, including adaptive skills (for example, the loss of bladder/bowel control and self-help skills), motor skills and emotional and behavioural regulation.In support, Matson and Mahan (2009) reported that children with CDD tended to have a higher rate of psychiatric symptoms and higher support needs.Second, while the onset of regression in CDD most commonly occurs gradually over a period of 6e12 months (Ellis et al., 2021), the frequency of sudden onset regression is higher in CDD compared to autism.Mehra and colleagues reported that up to 40% of CDD cases had regression over 8 weeks or less compared to 5% in autism with regression.Third, the period of regression in CDD is more commonly associated with fear and anxiety than in autism (Mehra et al., 2019).Finally, the mean age at regression is later in CDD compared to autism: 38 months in CDD (Mehra et al., 2019), compared to 19.8 months in autism (Tan et al., 2021).
In sum, autistic children who experience regression have greater support needs, irrespective of whether their diagnostic label is autism spectrum disorder, autism, or CDD.
Different approaches to improving our understanding of developmental regression in autism, including identifying subtypes with the hope of identifying biomarkers and biological cause(s), have been suggested (Nordahl et al., 2011).To date, the yield from subgroup approaches has yet to inform care.Autistic children who have regressed have an increased family history of autoimmune disease (Scott et al., 2017;Valicenti-McDermott et al., 2008) and neuropsychiatry diagnoses (Zhang et al., 2012) compared to those who have not experienced regression.More recently, genetic variants have been identified in sample of 134 individuals with autism and regression (Yin et al., 2020).This study lacked parent samples but was a large sample for studies of developmental regression in autistic children.It is expected to be the beginning of the application of current genetic techniques to further our understanding of causes and biological processes that result in developmental regression in autistic children.There have been some efforts to identify if distinct mechanisms underly regression in autism compared to regression in CDD.Although some authors have reported altered immunological markers and brain networks in CDD compared to autism with regression, findings are limited and remain of uncertain clinical significance (Connolly et al., 2006;Gupta et al., 2017).

Regression associated with seizures and epileptiform activity
Developmental regression in epilepsy is uncommon but is seen in some epilepsy syndromes, classified under the umbrella term of Developmental and Epileptic Encephalopathies (DEE).DEE encompass developmental conditions with epileptic seizures and conditions where the epileptic activity itself may contribute to cognitive and behavioural impairments beyond those expected from the underlying aetiology alone.In most DEEs, however, altered development is attributed to the underlying neurobiology rather than seizure burden or epileptiform activity (Syrbe, 2022).Children with DEEs typically display severe phenotypes, characterised by childhood-onset seizures, marked electrographic activity, and altered neurodevelopment (Syrbe, 2022) and account for under 10% of childhood onset epilepsies (Camfield & Camfield, 2019).Some examples include Otahara syndrome, West Syndrome, Landau Kleffner Syndrome (LKS), and developmental and epileptic encephalopathy with spike-wave activation in sleep (DEE-SWAS, a term that now also encompasses LKS).With nextgeneration genetic sequencing, the genetic underpinnings of DEE are becoming better understood (Syrbe, 2022).
Quantifying developmental regression in epilepsy has similar challenges to those described above, including a lack of consensus on the definition.For example, some authors include a "failure to develop skills or a decline in developmental trajectory" as regression (Camfield & Camfield, 2019) despite neither involving a loss of skill.
Developmental regression is relatively well-characterised in LKS.Children with LKS exhibit typical development until 3e8 years of age, when they begin to lose language function (with or without co-occurring focal seizures).Affected children typically present with auditory verbal agnosia (the inability to understand the spoken word, with a preserved ability to understand non-verbal sounds) (Murugesan et al., 2020).Many also develop expressive language difficulties in addition to behavioural (e.g., attention deficit hyperactivity disorder, aggression), psychiatric and cognitive impairments.Longitudinal studies have shown that less than a fifth of children recover lost language skills (Rossi et al., 1999).Epileptiform processes in LKS have been shown to originate in the language cortex of the dominant temporal lobe, with secondary spread to the contralateral cortex (Maquet et al., 1990).The EEG abnormality is markedly enhanced by sleep deprivation and during sleep, and it is typical to see continuous bilateral spike-and-wave in slow sleep at some time during the condition.Although mutations in GRIN2A are highly correlated with LKS, not all children with LKS have an identified aetiology (Stefanatos, 2008).
The domains affected by regression in DEEs depend on the epilepsy syndrome.Regression may be global, as in electrical status epilepticus during slow sleep (ESES) or Continuous Spike and Wave during Slow sleep (CSWS).Alternatively, regression may predominantly affect one domain, such as language in LKS.That said, these conditions are considered to exist on the same spectrum: an individual child may show some features of more than one of these syndromes or transition from one to another over time (Camfield & Camfield, 2019).Recent classifications cluster these conditions together as developmental and epileptic encephalopathy with spike-wave activation in sleep.
Epileptic encephalopathies are DEEs where seizures are thought to have a causative role in regression.An example is West Syndrome, an epileptic encephalopathy characterized by the onset of epileptic spasms typically commencing in the first year of life, and associated with global developmental impairment.Evidence for seizures causing regression includes a study of 107 patients where a delay in commencing anti-epileptic medications of over 2 months led to outcomes of lower adaptive skills at age 4 years (O'Callaghan et al., 2011).Additionally, regression in West Syndrome may be improved with decreased seizure frequency, for example, after epilepsy surgery (Asarnow et al., 1997), but there is limited high-quality evidence to support this (Camfield & Camfield, 2019).
Epilepsy is also associated with developmental regression in neurodevelopmental conditions such as autism.A metaanalysis of 29 studies of epilepsy, autism and regression found that epilepsy is 59% more likely to occur in autistic people who experience regression than those who do not (Barger et al., 2017).
Finally, there has been much research on the high prevalence (6e61%) of epileptiform discharges in autistic people without seizures (Spence & Schneider, 2009).In the absence of seizures in an individual, epileptiform discharges are considered a non-specific marker of altered brain function rather than markers of altered seizure threshold (Spence & Schneider, 2009;Tuchman & Rapin, 1997).Large studies have shown a greater prevalence of epileptiform activity in autism with regression (19%) compared to autism without regression (Tuchman & Rapin, 1997), suggesting altered brain function in children with regression.Other autism studies have found no difference in rates of epileptiform activity between regression and non-regression groups but were considerably smaller (Baird et al., 2006;Canitano et al., 2005;Giannotti et al., 2008).

Regression associated with genetic conditions
Rett syndrome (RTT) is an X-linked genetic condition caused mainly by loss of function mutations in the MECP2 gene.RTT has an estimated prevalence of 1:10 000 females (Einspieler & Marschik, 2019) and core diagnostic criteria describe a period of motor and behavioural regression (Neul, 2010), typically involving progressive functional decline over several months to years (Zoghbi, 2016) (Table 3).Communication skills and purposeful hand movements may evolve into stereotyped hand movements, social withdrawal, and behavioural regression.Deceleration of head growth is typical and onset of regression tends to occur between 12 and 19 months, with earlier (Fehr et al., 2011) and later onset of regression reported (Urbanowicz et al., 2015), which may indicate different loss of function mutations.Developmental regression is thought to result from altered neuronal function (Pohodich & Zoghbi, 2015;Zoghbi, 2016).
Phelan McDermid, or 22q13 deletion syndrome, is an autosomal dominant disorder caused by changes to the SHANK3 gene on chromosome 22.Newborn presentations include hypotonia, feeding challenges, and crying, followed by slow developmental progression, with preschool global developmental delays commonly recognised.Neurologic and motor regression are often reported, sometimes followed by losses in speech, self-help, play, and social skills (Table 3).Developmental regression of language, motor and or behavioural abilities can occur in 43e65% (Reierson et al., 2017;Brignell, Gu, Holm, Carigg, Morgan., 2021) of children with Phelan McDermid.The onset of regression occurs around 6 years of age (Reierson et al., 2017).
Goin-Kochel and colleagues (Goin-Kochel et al., 2017) studied de novo likely gene disrupting mutations in about 2500 autistic children, 300 of whom experienced regression.Mutations affecting genes that code for postsynaptic density proteins, which play a key role in activity dependent synaptic plasticity, were associated with an increased prevalence of regression.In contrast, mutations affecting genes that code for proteins expressed in the embryo were associated with a reduced regression rate.One interpretation is that, in the case of a postsynaptic density mutation, initial socialcommunication skills can develop but are lost when the necessary experience-dependent synaptic plasticity cannot occur.On the other hand, in embryonically expressed mutations, foundational social-communication skills cannot develop in the first place, making regression less salient.This is supported by computational modelling (Thomas et al., 2016), concluding that regression may be associated with over-aggressive synaptic pruning.
It is not possible to list all genetic conditions associated with developmental regression in this summary, however, it is acknowledged that genetic discovery through next generation sequencing will expand the list of conditions of childhood that feature developmental regression and may assist our understanding of biological processes.

7.
Influence of child developmental regression on caregiver perceptions and mental health Parents expect their children to build skills progressively over time according to typical developmental milestones.Observing a regression or skill loss is alarming.While there is an abundance of literature demonstrating increased stress levels and poorer health outcomes among parents of children with developmental disorders in general (Barroso et al., 2018;Masefield et al., 2020), few studies have specifically focused on the mental and emotional experiences of caregivers observing regression in their children.In a recent report of 12 parents of children with CDD, most caregivers shared that their child's neurodevelopmental decline caused them stress and anxiety (75%), and was traumatic for them (58%), and half that it made family life more challenging (Ellis et al., 2022).Moreover, they shared feelings of loss, sadness, and confusion over the significant changes they observed in their children, with some hoping for a return of their child's pre-loss personality.Similarly, parents of children with RTT have reported increased levels of stress and decreased marital satisfaction compared to clinical and control samples (Cianfaglione et al., 2015;Perry et al., 1992), as well as elevated depression symptoms and poorer health-based quality of life (Sarajlija et al., 2013).One report of 396 parents of individuals with RTT found that caregivers who assumed the bulk of care for the affected individual reported higher levels of family dysfunction compared to those who shared caregiving responsibilities more equitably (Lamb et al., 2016).Less effective family functioning was also associated with having more than one affected child, a later onset of symptoms in the child, decreased parental self-efficacy, and poorer parental coping styles or strategies.
Similarly, in a study of more than 8,000 parents of autistic children, those who indicated that their child lost skills (42%) perceived more negative consequences, less personal understanding, more negative feelings toward, and associated a higher number of symptoms with their child's autism compared to parents whose children did not lose skills (52%; Goin-Kochel, 2019).Those who witnessed regression reported greater levels of parental distress and lower levels of familybased and social support in their adjustment to their child's diagnosis.It seems that witnessing regression as the onset of a neurodevelopmental condition, as opposed to observing developmental delays from the outset, is particularly traumatic because parents have the time to observe and know their children on an expected developmental path.The outcomes following skill loss also are uncertain.Still, they likely represent a significant long-term shift in trajectory, leaving families to adjust expectations rapidly.
In addition to the mental and emotional toll on caregivers, regression can influence parents' cognitions about their child's diagnosis.In one study that examined the phenomenon of skill loss among 39 parents of autistic children, researchers noted that parents of children who regressed (47.5% of the sample) were more likely to offer both an aetiological explanation for and express feelings of guilt about their child's diagnosis compared to parents of children who had not regressed (Davidovitch et al., 2000).In similar studies examining parents' beliefs about causes of autism relative to symptom onset, parents who reported a regression in their child were significantly more likely to endorse environmental mechanisms (for example, vaccines, pollutiondfactors otherwise external to the child) as causal, while parents not reporting a regression more frequently endorsed biological mechanisms (for example, genetics, brain structuredfactors otherwise internal to the child) (Goin-Kochel et al., 2015;Goin-Kochel & Myers, 2005).In a recent large-scale study of more than 16,000 c o r t e x 1 6 9 ( 2 0 2 3 ) 5 e1 7 parents of autistic children, regression and lower levels of both language skills and adaptive functioning predicted parental endorsement of vaccines as a cause for their child's autism (Fombonne, Goin-Kochel, & O'Roak, 2020).Understandably, parents who witness skill loss may perceive that a temporal insult has changed the trajectory of their child's development.For this reason, understanding the influence of skill loss on parents' cognitions about their child's diagnosis is valuable because it may change health-related decisions that they make on behalf of the affected child (for example, which treatments to pursue) and other family members.

Conclusions
Given Developmental regression must be recognised, and children and their families supported in a timely, holistic way and in a manner that allows for individualised and personalised support.Consistency in care relies on an agreed definition and methodology for measuring skills loss in children.Considering the breadth of known and emerging conditions that feature developmental regression and the heterogeneity in onset and clinical course, a logical inclusive approach would be to consider developmental regression using the RDOC framework.For example, we demonstrated the overlap with epileptic genetic developmental encephalopathies which may be linked to altered excitatory-inhibitory imbalance.Developmental remodelling of neural circuits may be a feature of optimisation (associated with grey matter reduction and cortical thinning).Since the immune system appears to be implicated in synaptic remodelling (Sekar et al., 2016), this would explain why familial immune problems are associated with increased risk.Collaborative research and detailed phenotyping will advance knowledge about the biological neurocognitive processes, clinical course and outcomes with the potential to improve child and family outcomes through earlier recognition and support.

9.
Future directions i Definition The development of an international special interest group formed within the International Society for Autism Research has initiated a Delphi procedure to ask experts with lived experience and within the fields of developmental paediatrics, neurology, psychiatry and other related subspecialities, to reach an agreed definition for developmental regression across a breadth of disorders.It is anticipated that this work will be completed in 2023.The Delphi procedure will also help delineate whether increasing divergence from typically developing trajectories counts as regression; or whether reduction in absolute performance over time is required.An agreed definition will allow consistency in neurobiological studies, and consistency with regards outcome measures and intervention trials important for children and their families.It will also enhance merging datasets and international collaborations.
ii Data collection and collaboration Use of agreed prospective data collection, core objective measures and standardised use of home videos for regression in all neurodevelopmental conditions.Consenting for data sharing after regulatory board approval, will be important to allow new collaborations with multidisciplinary colleagues offering different technical expertise (such as artificial intelligence analysis methods).
iii Understanding the biological processes Basic science research to assist with genomic, metabolomic and proteomic understanding of the biological underpinnings of developmental regression, for the purpose of developing preventative strategies and precision therapies.Developmental regression may be a common causal pathway of different underlying mechanisms.By understanding pathway targets which cut across different developmental phenotypes, pre-clinical studies and patient-informed clinical trials can be applicable across multiple conditions with regression.iv Best practice transdiagnostic processes Phenotyping children with developmental regression using the RDOC approach whereby regression is an early dimensional symptom.This will allow a trans-diagnostic approach across conditions with regression as a presentation, to better delineate psychological constructs of cognition and neural circuitry across a range of neurodevelopmental conditions.This approach may also identify levels important to stratify for in treatment trials where several causes of developmental regression may have one functional pathology with shared c o r t e x 1 6 9 ( 2 0 2 3 ) 5 e1 7 behavioural outcomes.Or that different causes due to multiple pathologies, may have different behavioural outcomes but with shared similarities.
v Family stress The impact of regression on family mental and emotional health outcomes must be considered and family-based approaches to the care of all family members offered when neurodevelopmental diagnoses are made.This is important as will emphasise the growing importance of co-designing and co-producing intervention trials which are important to children and their families.

Funding
Dr Kirsten Furley is a PhD student receiving stipend support from the University of Monash, Melbourne, to complete a translational PhD in loss of skills in children.

Declaration of competing interest
None.

Fig. 1 e
Fig. 1 e Developmental regression onset and key features by condition.Green Regression of developmental skills Blue Other typical features may be identified before onset of regression

Table 1 e
Examples of definitions used in studies focusing on developmental regression in autistic children.

Table 2 e
Measures used to collect data about developmental regression.

Table 3 e
Prevalence and characteristics of some specific conditions associated with developmental regression.