Genetic Etiologies and Risk Factors for Regressive Autism and Childhood Disintegrative Disorder: A Scoping Review

Heritability of Regression

Results are conflicting. In a review, ⁴ 2 studies reported only a moderate heritability of regression. Of these, one ¹⁵ included 71 monozygotic (MZ) and 126 dizygotic (DZ) twin pairs and heritability was assessed using 3 intra-class correlations for language regression, social skills and broader regression within each twin group. In the MZ group all 3 correlations were low (around 0.3). Correlations for the DZ group were very low (0.019 or less). A higher level of evidence was provided by others ¹⁶ using data from 179 sibling pairs with autism. The concordance of autism and autistic regression was only 19%, not significantly higher than that expected with independence. Authors concluded that their results did not support separate genetic influences on autism and regression. ¹⁶ Contrastingly, the review ⁴ also reported 2 results of a study ¹⁷ providing stronger evidence of the concordance of autism and autistic regression. First, of 27 DZ twins with autism, the concordance of autism and loss of any skills was 67%. Second, of 25 MZ twin pairs with autism, only 2 were discordant for regression. ¹⁷ These disparate results suggest that environment is playing a role as well as genetics.

Autosomal variants

Higher-level evidence associated with CDD was provided by a multi-disciplinary study. ⁷ Using whole exome sequencing of 17 probands with CDD, 1 or more rare gene variants, not present in typically developing (TD) siblings, were found in 16. Along with 3 other autosomal variants, a de novo (DN), missense single-nucleotide variant (SNV) of BBS5 was found in female MZ twins and a BBS9 SNV in a male proband carrying another autosomal variant. Syndromic intellectual disability (ID) had previously been associated with variants of BBS. ⁷

Weaker evidence of variants associated with CDD was found in a case study. ¹⁸ Six former patients, each afflicted with a fatal neurodegenerative disease of infant onset, were compared to the female proband with CDD. All 6 of the former patients were homozygous or compound heterozygous for a BSCL2 variant and most died of pulmonary infection at 7 or 8 years. The female proband regressed after 3 years and, as an adult, developed atypical Parkinson disease. She also died of pulmonary infection. Her relevant genotype was likely compound heterozygosity for 2 variants affecting the splicing of exon 7 in BSCL2 ¹⁸ (Table 3).

Researchers ¹⁹ provided higher-level evidence pertaining to broader RA. To start, a literature search identified 47 genes associated with broader RA. These genes were then sequenced in a cohort of 134 individuals with broader RA to identify variants in the group. Two unrelated male probands each had the same DN SNV in GRIN2A and a third male carried another GRIN2A variant. These variants, occurring in 3 of a cohort of 134, prompted the authors ¹⁹ to propose that they might be causal of autistic regression in these males. Also in the cohort was a female proband with a novel, missense SNV in UBE3A. Her regression began at 18 months and associated symptoms were epilepsy, ID, quadriparesis and balance issues. Another was a female with a missense, hemizygous SNV in SLC9A6. She had seizures from 11 months and autistic regression at 2 years. Another 2 males each carried a variant in the same gene. The first carried a novel, heterozygous missense SNV in EHMT1, along with an X-linked SNV. In the second, heterozygous missense variants were identified in EHMT1 and MFSD8. At 8 years, clinicians noted the boy's severe cognitive and language delay and proposed that his regression may have been due to a partial loss of function in the pathways of EHMT1 and MFSD8. Persons with autism have previously been found to carry DN variants (DNVs) in GRIN2A and EHMT1. Variants in EHMT1 have not been associated with broader RA before and only homozygous variants in MFSD8 have been previously associated with broader RA ¹⁹ (Table 3).

Two case studies^20,21 provided preliminary evidence connecting HEXA variants to broader RA. Each proband had Tay Sachs Disease (TSD). In the first, ²⁰ a girl with a homozygous SNV in HEXA, had CDD and refractory epilepsy from 3 years. In the second, ²¹ a boy with a homozygous but different SNV in HEXA had CDD and refractory epilepsy from 4 years (Table 3). Another case study ²² described 7 members of 2 consanguineous, but unrelated, families. In each, 1 member had a different HNMT missense SNV, along with severe ID, speech delay, and regression after 5 years. The condition was named ‘mental retardation, autosomal recessive 51’ (MRT51). The study focus was a male with MRT51 and homozygous for an HNMT variant. He exhibited aggression, speech delay, autistic traits and gastro-intestinal (GI) issues. Treatment with hydroxyzine, antihistamine and a histamine-restricted diet led to sustained and significant improvement ²² (Table 3).

Preliminary evidence of 2 case studies^23,24 linked a PRRT2 variant and RA. The first ²³ described a boy with a heterozygous PRRT2I deletion and an autosomal deletion. He had MRT51 with autistic regression at 1 year. The second ²⁴ described a girl with benign familial infantile seizures [BIS (familial)] and a heterozygous, frameshift PRRT2 duplication. She developed typically until exhibiting autistic regression at 15 months. Her family had variable phenotypic expression of the PRRT2 mutation. Mother, brother and proband were all carriers, with the mother being asymptomatic, the brother having BIS (familial) and the proband, RA ²⁴ (Table 3).

Chromosomal variants

Higher level evidence connected gene variants in particular chromosomal regions to broader RA. ²⁵ Using data from the Autism Genetic Resource Exchange Database, researchers defined an affected relative pair (ARP) as 2 probands with autism who were first- or second-degree relatives. Linkage analysis investigated autism susceptibility loci pertaining to broader RA in 34 ARPs with regression, and 151 without regression. Results suggested that changes in genes on chromosomes 7 and 21 rendered regression more likely ²⁵ (Table 3).

Further preliminary evidence associated chromosome 7 and RA. ²⁶ The primary interest was a boy with post-natal growth retardation, 2 febrile convulsions, mild hypoglycemia and autistic regression at 18 months. A maternal uniparental disomy of chromosome 7 (mUPD7) was identified, and the boy was subsequently diagnosed with Silver-Russell syndrome. The authors ²⁶ proposed that his RA was due to combined genetic and environmental risk factors such as an mUPD7 and hypoglycemic events (Table 3).

Sex-linked variants

An observational study ²⁷ identified a male and female with the same SNV in PLXNB2. The male suffered febrile seizures and autistic regression at 36 months whereas the female regressed at 14 months. Authors suggested that this SNV played a role in these regressions. Two other probands carried different SCN1A variants. The first was a heterozygous, novel, splice site variant in a female who regressed from 6 months and with febrile seizures and convulsions from 3 months. The second was also a heterozygous, novel variant in a male. He presented with brachycephaly, language delay and autistic regression at 2 years. In the same study, ²⁷ 2 homozygous SNVs in SCN2A and MTHFR were located in a male proband with motor impairment and autistic regression at 11 months. ²⁷ Others ⁴ reported that around a third of carriers of SCN2A variants had autism. Regression was documented in 8.5% of cases (Table 3).

Deletions and point mutations in SHANK3 have been implicated in RA with a research group ⁷ reporting a homozygous, missense variant in SHANK3 in a female with CDD. Further, but lower-level evidence of a variant associated with CDD and SHANK3 was found in a case study ²⁸ focussing on a girl with CDD carrying an interstitial SHANK3 deletion. Authors considered that this variant was associated with her disorder. A review ⁴ associated SHANK3 with RA. Those with large deletions in SHANK3 had a 28% to 43% regression rate (RR), whereas those with point mutations had a 65% RR. Associated phenotypes of SHANK3 included autism, ID and schizophrenia ⁴ (Table 3).

A C-C study ¹² provided higher level evidence of an association of a single nucleotide polymorphism (SNP) with broader RA. The frequency of 2 single SNPs in SNAP-25 was investigated in children with broader RA (n = 58) and EOA (n = 69). In the broader RA group, the rs363039 G allele and GG genotype were significantly less frequent than in the EOA group. On the other hand, in the broader RA group, the rs1051312 T allele and TT genotype were significantly more frequent than in the EOA group. The authors decided that these SNPs might be useful biomarkers for RA (Table 3).

Along with an autosomal variant, researchers ⁷ found that a male with CDD had a DN, missense TRRAP SNV, previously reported in a person with autism (Table 3). There were also reports of a DN, missense SNV in NAV2 in a male with CDD whilst another male with CDD had a homozygous insertion in NOP9, along with 2 X-linked variants. In a third male with CDD, ⁷ a DN SNV in ZNF236 was identified. Other studies were cited which reported autism in 2 carriers of missense SNVs in NAV2. ⁷ A fourth male with CDD, with 4 variants on the X and another on the Y chromosome, was heterozygous for an SNV in CLUL1. The same study reported that female MZ twins carried a DN deletion in the 3´UTR region of OGDHL which had previously been associated with neurodegeneration. These girls also carried a DN SNV in NSD1 and a hemizygous SNV in TUBGCP5. Both SNVs had been identified in probands with autism and ID before. ⁷ Apart from 6 hemizygous SNVs, a male with CDD had a homozygous deletion in PRKCSH. ⁷ Along with a second autosomal SNV, a female proband ⁷ carried a homozygous missense SNV in RAD51C. In addition to 3 hemizygous mutations, 2 homozygous deletions in KIAA2018 and a homozygous, missense SNV in BRIP1, were identified in a male with CDD. ⁷ Nonsense, missense DNVs in KIAA2018 have been found in individuals with autism before. ⁷ Along with a hemizygous variant, another male with CDD had a homozygous variant in TCTEX1D2 which had been previously associated with schizophrenia ⁷ (Table 3).

A male with CDD carried a homozygous SNV in DNMT3B, along with 2 DN, autosomal variants and 2 X-linked SNVs. This DNMT3B variant had formerly been reported in persons with syndromic ID. ⁷ Along with a hemizygous, missense SNV, a second male with CDD carried a homozygous, missense SNV in DAP3 whilst a homozygous, missense SNV in ADAMTS18 was identified in a third male with CDD ⁷ (Table 3).

Preliminary evidence associating a gene variant to CDD was provided by a case study ²⁹ assessing a female proband with autistic regression from 3 years, epilepsy and macrocephaly. Genetic sequencing identified a heterozygous, DN, missense PAK1 SNV. Authors ²⁹ suggested a link between CDD and the dominant PAK1 SNV. Similar preliminary evidence was provided by others ²³ where SEZ6L2 and QPRT deletions were identified in a male with BIS and RA. Variants of SEZ6L2 have been reported in persons with autism and QPRT variants in persons with neurodegeneration ²³ (Table 3).

A review ⁴ described variants associated with autistic regression. A DN microdeletion in CHD2 was associated with broader RA in 5 of 6 carriers. The resultant phenotype was early-onset epileptic encephalopathy followed by cognitive loss. A CHD8 variant had a 40% RR in carriers. Carrier traits included macrocephaly, GI symptoms and distinctive facial features. With an RR of 43%, 60% of carriers of heterozygous KMT5B variants had febrile seizures. Five of 8 (63%) carriers of biallelic CNTNAP2 variants whereas only 1 of 7 (14%) carriers of heterozygous variants in CNTNAP2 exhibited regression. Of 17 carriers with a deletion in NRXN1, only 3 (18%) exhibited regression. There was also reported regression in 2 probands with DEAF1 variants, 1 in a carrier of a PTEN variant and 1 in a carrier of a TRIP12 variant. The same review reported that from 41% (7 of 17) to 80% (4 of 5) of carriers of SYNGAP1 variants were in probands with RA. Contrastingly, ADNP variants were associated with regression in only 20% of 59 carriers. There was reported regression in a carrier of an ANKRD11 variant and a carrier of an ASH1L) DNV. Regression occurred in 2 carriers of GRIN2B variants and authors ⁴ considered that these GRIN2B variants were associated with autism (Table 3).

A case series study ³⁰ provided weaker evidence associating an SGSH variant with CDD. Here, 2 brothers, homozygous for a novel variant in SGSH, experienced autistic regression from 4 years (Table 3).

Sex-linked variants

An observational study ²⁷ identified a novel missense SNV in MECP2 in a 7-year-old girl with microcephaly, seizures from 6 months and subsequent RA. Her regression was attributed to the SNV but it was unclear as to whether her diagnosis was Rett syndrome (RTT). Preliminary evidence of a case study ³¹ described atypical RTT in a girl with a nonsense SNV in MECP2. She had epilepsy from 6 months, followed by impeded development, regression, microcephaly and craniofacial dysmorphia. A review ⁴ contained 2 studies reporting on MECP2 duplication syndrome. One ³² comprised 31 probands with RTT and of these, 12 (39%) experienced autistic regression after an initial seizure. The second ³³ comprised 17 probands, mostly with autism. Language regression was reported in 8 (47%), and regression in other areas in 7 (41%). See Table 3.

Two males with CDD ⁷ carried 2 different hemizygous, missense SNVs in SUPT20HL2. One also had an autosomal, DN SNV and the other, an autosomal, homozygous deletion, and 5 hemizygous, missense, X-linked SNVs (Table 3).

Beta propeller protein–associated neurodegeneration (BPAN) is a single gene disorder involving neurodegeneration due to iron accumulation in the brain. A case study ³⁴ provided preliminary evidence connecting WDR45 to RA. It featured a 5-year-old girl with autistic regression from 1 year, seizures, intracranial calcification and iron accumulation in the brain. Her diagnosis of BPAN was further supported by a heterozygous, nonsense, DN mutation in WDR45 which had previously been associated with BPAN. ³⁴ Another case study ³⁵ provided preliminary evidence. It focused on a 4-year-old boy with GI issues, autistic regression from 33 months and carrying a hemizygous, 2-basepair deletion in exon 6 of TMLHE. Testing indicated a carnitine deficiency, and with carnitine supplementation, his functioning improved (Table 3).

Missense SNVs in NRK, TBC1D8B, NKRF and SAGE1 were identified in a boy with CDD. ⁷ In addition to 2 autosomal variants, another male carried hemizygous, missense SNVs in PDK3, ARSF, ALAS2, Cxorf57 and ALG13, and a deletion in STARD8. Another male with CDD carried a missense SNV in MTMR8. Authors ⁷ reported that SNVs in NRK, NKRF, Cxorf57, ALG13 and STARD8 had been identified in people with ID, SNVs in TBC1D8B in individuals with autism and ID, and variants in SAGE1I, MTMR8 and ARSF in individuals with autism and individuals with ID. ⁷

The same study ⁷ also identified missense, hemizygous SNVs in CDR1 and FAM50A in a boy with CDD whilst another carried a hemizygous missense SNV in CNGA2. Along with an autosomal variant, another male with CDD carried a hemizygous missense SNV in USP26. In addition to an autosomal variant, another carried an X-linked missense SNV in ENOX2 whilst another, along with an autosomal variant, carried a hemizygous, missense, DN SNV in ZXDA and a hemizygous, missense SNV in SPANXN2. Variants in these genes have previously been associated with ID. ⁷ Another male with CDD was heterozygous for an autosomal mutation, carried hemizygous missense SNVs in BCOR, BEX2, SRPK3 and VSIG1, and a fifth on the Y-chromosome in USP9Y. Authors ⁷ reported that previous studies had found ID in probands with variants in BCOR, BEX2, SRPK3 and VSIG1, with autism also identified in 2 with a splice site variant in VSIG1 and in 1 with a hemizygous, missense SNV in BEX2 (Table 3).

A study ²⁷ described a female with a missense SNV in CDKL5 and considered this a likely cause of her RA from 1 year. Another had an X-linked, missense SNV in PCDH19. Symptoms included autistic regression from 16 months, epilepsy and a Dravet-like syndrome. Authors proposed a disruption of shared pathways of RA and the Dravet-like syndrome had led to her regression ²⁷ (Table 3).

More complex Inheritance

A C-C study ⁶ provided higher level evidence linking genetics to broader RA. Authors wrote that more than 200 genes associated with autism were in the metabolic pathway of the amyloid precursor protein. Secreted amyloid precursor protein–a (sAPPa) is produced when amyloid precursor protein is cleaved by a secretase during non-amyloidogenic processing. This can result in amyloid plaques, found in the brains of some autistic children. Moreover, high levels of sAPPa can impede brain development during important periods. Plasma levels of sAPPa were significantly elevated in children with broader RA compared to those with EOA and TD children. Authors ⁶ suggested that higher sAPPa levels may be a useful biomarker for broader RA (Table 3).

A review ⁴ described a study, using family data, to estimate the involvement of likely gene-disrupting (LGD) DNVs in RA vs EOA. Authors focused on genes encoding for fragile X mental retardation protein, diseases of Mendelian inheritance, chromatin modification, postsynaptic density proteins and proteins expressed embryonically. The LGD DNVs encoding for postsynaptic density proteins were more frequent in the broader RA group but small numbers limited any conclusion. ⁴

Pathogenicity

Established methods ⁷ assessed variant pathogenicity in the 4 areas of brain expression, conservation, intolerance to functional variation and likelihood of being damaging. Of 40 candidate genes, 14 (NRK, TBC1D8B, TRRAP, NAV2, OGDHL, ZNF236, PRKCSH, MTMR8, BCOR, SRPK3, USP9Y, KIAA2018, CXorf57 and ALG13) met each criterion and were described as ‘potentially pathogenic’. Others ²⁷ used the criteria of the American College of Medical Genetics and Genomics to assess variant pathogenicity. ³⁶ Sixteen variants of 12 genes met 1 criterion of pathogenicity and were classified as variants of uncertain significance. Both GRIN2A variants, both SCN1A variants, along with SCN2A, MTHFR, PCDH19, UBE3A and EHMT1 (first variant described in Table 3) variants were classified as ‘moderately pathogenic’ (PM2). Both PLXNB2 variants and both SLC9A6 variants, along with CDKL5, MFSD8I and EHMT1 (second variant described in Table 3) variants, were classified as ‘supporting pathogenic’ (PP3), while the MECP2 variant was deemed ‘strongly pathogenic’ (PS1). See Table 3. Contrastingly, a third group ²² described an HNMT variant as pathogenic based on trio-based exome sequencing (Table 3).

Epigenetic associations, toxicant exposures, and gene expression

A C-C study ¹² provided stronger evidence of epigenetics in relation to broader RA. An allele of SNAP-25, a target of microRNA, was significantly more common in probands with broader RA than in TD or EOA probands. The findings of an animal study ¹⁴ suggested that in combination, timed toxicant exposure after birth and a particular genotype resulted in social regression in mice. Researchers investigated whether mice with a GSTM1 deletion were more sensitive to sodium valproate exposure postnatally than wild-type mice. The genetically altered mice were injected with sodium valproate or saline on post-natal day 14. Mice exposed to sodium valproate had increased neuronal losses and over time, sodium valproate exposure resulted in changed brain chemistry and less sociability. ¹⁴

Lesser preliminary evidence was provided by a C-C study using autistic rat models. ³⁷ Authors aimed to explore aberrant social behavior and anxiety in relation to broader RA and EOA after exposure to valproic acid. To emulate EOA, 5 pups of 5 dams were administered equal amounts of valproic acid on embryonic day 12 (Model 1). To emulate broader RA, 5 pups of 5 dams were administered equal amounts of valproic acid on post-natal day 14 (Model 2). Pups from each model exhibited aberrant sociability and anxiety. Microarray analysis compared the gene expression profiles of Model 1 and Model 2 pups. In both models, Neu2 and Mt2A changed expression in the same direction. Researchers concluded that with exposure to neurotoxins at different developmental stages, gene expression changes rather than gene changes may have a stronger effect on sociability and anxiety. ³⁷ A case study ³⁸ supplied preliminary evidence relating to exposure to neurotoxins and CDD. A 14-year-old vegetarian male patient had typical development until 6½ years. Thereon, he was fearful, with fading speech, echolalia and emergent autistic behaviors and an eventual diagnosis of CDD. Testing showed low serum vitamin B₁₂, raised plasma homocysteine levels, and panhypoperfusion of the brain. After vitamin B₁₂, anti-oxidant and other supplementations, improvement was noted. Clinicians reported that reduced vitamin B₁₂ levels may have caused the neurotoxicity and regression. ³⁸

A study ⁷ found that their 40 candidate genes for CDD were more highly expressed in non-neocortical regions than neocortical regions and that their expression profile was similar to that of an independent cohort with autistic regression. Further, expression was increased during the developmental periods of 1-6 years and 6-11 years, which is when probands’ symptoms emerged. They then compared the median expression levels of non-synonymous and synonymous variants in non-neocortical and neocortical regions in probands from the 4 groups of CDD, TD siblings, RA and EOA. With 1 exception, the expression of the candidate genes for CDD was distinctive and qualitatively different from the profiles of other gene sets. The exception was in the non-synonymous variants of probands with RA. This was despite probands with CDD and probands with RA having only a NAV2 variant in common. Extending the analysis to other genes sets showed that the gene expression profile of non-synonymous variants in probands with CDD and probands with RA were both distinctive and qualitatively different from the other gene set profiles. ⁷

A C-C study ³⁹ provided low-level evidence regarding gene expression and broader RA. Associations between broader RA and GI issues were explored in 4 groups. A case group comprised 25 children with broader RA from 12 months and comorbid GI issues. Three control groups were children with: Crohn disease, no autism; Ulcerative colitis, no autism; and Neither inflammatory bowel disease nor autism. For all children, transcriptome profiling of GI mucosal biopsy tissue–enabled gene expression for each group to be quantified. Results suggested a unique gene expression profile associated with inflammatory bowel disease in children with broader RA and GI issues. ³⁹

Summary

The reported concordance rates of autism and broader RA were inconsistent. In broader RA, 86 gene variants, 3 chromosomal regions, and 1 metabolic pathway were described. Of the genes, 33 were X-linked, 1 Y-linked, and 52 autosomal. Apart from the X-chromosome, the most common variant locations were chromosome 16 (7) and chromosomes 2 and 11 (6 each). Only 10 variants were novel. Hemizygous (vs homozygous or heterozygous) was the most stated variant zygosity. Less common references to zygosity were oligogenic heterozygosity, for 4 variants, and compound heterozygosity, for 1. At the DNA level, there were 57 SNVs, 15 deletions, 3 frameshift variants, 4 variants at splice sites, 3 duplications, 1 insertion, and 3 CNVs or associated with CNVs. In terms of protein production, there were 38 missense and 4 nonsense variants, an LGD variant and a loss of function variant. With respect to origin, 19 variants were DN and 2 were of maternal origin (Table 3).

Treatments improving the functioning of probands with broader RA were described in conjunction with variants of HNMT and TMLHE. Frequently associated phenotypes of broader RA were epilepsy and macrocephaly. Conditions most often related to variants in those with broader RA were ID, autism and schizophrenia. Most genes were cited only once, with exceptions being MECP2 and SHANK3, each cited in 3 studies, and EHMT1, HEXA, PLXNB2, PRRT2, SCN1A, SCN2A and SLC9A6, each cited in 2 (Table 3). Of the described variants, 58 were in probands with CDD and 37 in probands with broader RA but not necessarily CDD.

Higher levels of sAPPa in children with broader RA may become a useful biomarker. ⁶ Two studies^7,21 assessed the pathogenicity of variants associated with broader RA using established criteria. Of variants assessed, about a third met 1 of more of the criteria for pathogenicity.

Differential frequency of 2 SNAP-25 alleles in probands with broader RA from those with EOA suggested epigenetic involvement in broader RA. ¹² A case study ³⁸ and 2 animal C-C studies^14,37 associated toxicant exposure to broader RA. In the case study, a boy with low serum vitamin B₁₂ and raised plasma homocysteine levels was diagnosed with CDD. In the first animal study, rats exposed to valproic acid postnatally developed impaired sociability, and in the second, mice with a GSTM1 deletion exposed to sodium valproate postnatally developed impaired sociability.

Candidate genes for CDD were differentially expressed in non-neocortical regions from neocortical regions, and non-synonymous variants of these genes were expressed in a similar way to those of probands with RA but not probands with EOA. ⁷ Unique gene expression was identified in children with broader RA and GI issues compared with other groups. ³⁹

Broader regressive autism

Higher level evidence of differences in brain chemistry pertaining to the immune system in those with broader RA, compared to those with EOA, were reported by a C-C study. ⁴⁰ Proinflammatory cytokines, plasminogen activator inhibitor-1 (PAI-1), and resistin levels were compared in children with broader RA, with and without developmental disability and children with EOA. The mean level of proinflammatory cytokines did not differ between the combined broader RA and EOA group. However, resistin and PAI-1 levels were higher in the broader RA with DD group than in the other 2 groups. ⁴⁰

The BDNF gene encodes brain-derived neurotrophic factor (BDNF) produced by activated brain microglial cells. These cells share features with peripheral macrophages, suggesting that BDNF production has an important role in the immune system. A C-C study ⁴¹ produced higher level evidence of a connection between the immune system and broader RA. Among the groups of neonates compared were those who developed broader RA and those who were later diagnosed with EOA. Comparing blood samples of the broader RA and EOA groups found increased BDNF levels (though not significantly) in the EOA group. Authors hypothesized that BDNF levels may be associated with autism development and altered for those with broader RA. ⁴¹ Other studies^42,43 provided higher level evidence of an immune connection with broader RA. In the first, ⁴³ medical charts of children with autism were reviewed. Autoimmune conditions in first- and second-degree relatives were recorded according to whether the child had broader RA (n = 33) or EOA (n = 206). Autoimmune thyroid disease and familial type 1 diabetes were significantly more common in broader RA families, suggesting that those with broader RA were more prone to immune issues than those with EOA. ⁴³ In the second, ⁴² haplotypes of the human leukocyte antigen (HLA) system, in 33 children with broader RA and 98 with EOA, were examined. Of those with broader RA, 62 (43%) had HLA-DPA1*01-DPB1*04 (DPA1) vs 14 (63%) with EOA, suggesting that DPA1 offers some protection from regression.

Regression in Down syndrome

Autistic regression can occur in people with Down syndrome and mostly during adolescence or early adulthood. ⁴⁴ Down syndrome disintegrative disorder is one label for the resultant condition. A review ⁴⁵ reported that most of those with DSDD have language regression and sometimes depression, social withdrawal, anxiety, psychoses and newly acquired insomnia. One study in the review investigated 4 patients with DSDD where treatment with benzodiazepines and electro-convulsive therapy resulted in recovery to baseline. ⁴⁵

Mild evidence of autoimmune associations in people with DSDD came from a case series study ⁴⁴ of 4 probands with DSDD, each carrying various auto-antibodies. After intravenous immunoglobulin treatment, all improved significantly, implying that regression in DSDD may be mediated by autoimmunity. ⁴⁴ A C-C study ⁴⁶ provided higher level evidence of a relationship between thyroid autoimmunity and DSDD. Compared with age-matched controls, 10 of 11 probands had higher concentrations of thyroperoxidase antibodies.

Summary

First, compared to probands with EOA and broader RA without DD, probands with broader RA and DD had higher resistin and PAI-1 levels. Second, compared to probands with EOA, probands with broader RA had lower BDNF levels as neonates and different allelic ratios in an HLA polymorphism than those with EOA. Third, there were higher proportions of autoimmune thyroid disease and familial type 1 diabetes in families of those with broader RA compared to those with EOA. Lastly, those with DSDD were more prone to autoimmune disorders than those with only Down syndrome.