Subclinical neuropsychiatric trait variation in parents of children with autism spectrum disorder: a cohort study

Abstract

Background

Many neuropsychiatric conditions share overlapping features underpinned by shared genetics. In this study, we examined the co-aggregation of sub-threshold variation in neuropsychiatric phenotypes in biological parents of autistic children to provide insights into the intergenerational transmission of genetic liability for autism.

Methods

Autistic, neuropsychiatric (i.e., anxiety, depression, ADHD), and cognitive traits were characterized in biological parents of children enrolled in a longitudinal developmental study: 189 families of autistic children and 100 families with no autistic children were included. Families were further characterized as having only one (simplex) or more than one (multiplex) autistic child. Maternal and paternal traits were compared across groups using analysis of variance. Between-parent and within-parent correlations across trait domains examined patterns of familial trait aggregation and assessed shared versus unique contributions to the inheritance of autism. Logistic regression assessed the predictive utility of parental traits for simplex vs. multiplex group membership.

Results

Mean levels of paternal autistic traits (F(2, 224) = 5.67, FDR-adjusted p-value (q) = 0.013) and maternal anxious (F(2, 262) = 11.14, q < 0.001) and depressive (F(2, 262) = 7.08, q = 0.005) traits differed between groups. Post-hoc tests revealed elevated autistic traits in multiplex fathers (q = 0.009) and elevated anxious (q < 0.001) and depressive (q = 0.004) traits in multiplex mothers compared to parents of non-autistic children; simplex parents did not differ from either of the other groups. Parental traits jointly accounted for 7.9% of autism recurrence liability in multiplex families. Elevations in autistic traits co-occur with elevations in neuropsychiatric traits in simplex parents (0.35 ≤ r ≤ 0.43) but were uncorrelated in multiplex parents.

Limitations

Findings from this study may not generalize to more heterogeneous samples with differing racial, ethnic, and socioeconomic demographics. Exclusionary criteria for the original study may result in families with a lower trait burden for psychiatric disorders than the broader population of parents of autistic children.

Conclusions

Results highlight unique patterns of autistic, anxious, and depressive trait presentation and aggregation in multiplex parents. Findings call for a transdiagnostic approach to quantifying genetic liability in families with autistic children.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13229-026-00720-5.

Introduction

Autism is highly heritable [1, 2] and attributable to both common polygenic variation [3–5] and rare genetic variation that act additively to influence phenotype [6–8]. Younger siblings of autistic children are approximately 7.5 times more likely to be diagnosed with autism spectrum disorder (ASD) than children in the general population [9]. Autism recurrence in families has been traced to additive genetic effects, where polygenic profiles related to autism are over-transmitted from non-autistic parents to their autistic children [6]. This is consistent with the liability threshold model [10] whereby autism-related genetic variation contributes additively to a continuum of behavioral traits spanning typical, to subclinical, to clinical variation in the population [11]. The additive genetic effect is further evidenced by quantitative trait studies demonstrating that families with more than one autistic child (multiplex) exhibit greater likelihood of recurrence [9, 12] that is related to the aggregation of higher levels of subclinical autistic traits in parents (the broad autism phenotype, or BAP [13–16]) and higher levels of autistic traits in diagnosed older siblings (“probands”) [17]. Subclinical BAP features have been found to be positively correlated among non-autistic parents of autistic children [14, 16, 18–20] indicative of assortative mating [18] whereby individuals with similar phenotypes tend to pair up in reproduction. Assortative mating for BAP features is hypothesized to result in the compounding of heritable autistic features which surpasses the diagnostic threshold from one generation (non-autistic parents) to the next (autistic child) [18].

While the BAP has been well studied in parents of children with autism, it remains unknown whether parental autistic traits co-aggregate with other neuropsychiatric phenotypes that could additively influence liability for autism in offspring. There is a critical need to index multiple neuropsychiatric and neurodevelopmental phenotypes in the study of autism transmission [11, 18, 21], as there is growing evidence that autism has overlapping polygenic architecture with several neuropsychiatric disorders, including anxiety, attention-deficit/hyperactivity disorder (ADHD), depression, schizophrenia, and bipolar disorder [4, 22–25]. This shared genetic architecture mirrors the high rate of co-occurrence of these disorders, and their features, with autism [26, 27], as well as higher rates of these conditions among first- and second-degree family members of autistic individuals [14, 28–41].

In the current study, we characterized quantitative trait variation in autistic, ADHD, anxiety, and depression domains, as well as cognitive ability, in biological parents of autistic and non-autistic children. Our sample includes maternal and paternal neuropsychiatric trait data from a cohort of parents with one (simplex) or more than one (multiplex) autistic child ascertained as part of a prospective study, and a comparison sample of parents of typically developing children. Our aims were to: (1) describe and compare trait profiles for mothers and fathers across groups, (2) interrogate assortative mating via within-family (cross-parent) correlations in each trait domain, (3) examine within-parent correlations between autistic traits and other neuropsychiatric phenotypes to clarify patterns of co-aggregation in families, and (4) quantify the joint contribution of parental traits to autism recurrence in families. To our knowledge, this is the first study of its kind to examine the familial co-aggregation of subthreshold variation in genetically-related neuropsychiatric phenotypes in parents of autistic children, offering novel insights into genetic liability for autism. Our work expands on prior literature charactering the BAP and reporting higher rates of neuropsychiatric diagnoses in parents of autistic children by directly examining neuropsychiatric traits in a continuous manner, individually and jointly, within both individual parents and parental pairs. This work has important implications for understanding, and indexing, genetic liability for autism in families.

Method

Participants

Study data were collected as part of the multisite Infant Brain Imaging Study (IBIS) which followed younger full-biological siblings of autistic probands (“high likelihood”; HL) and a comparison sample of infants with at least one typically developing older sibling and no first- or second-degree family history of autism (“low likelihood”; LL) from infancy through school-age. Exclusion criteria included known genetic syndromes in the proband or infant, perinatal adversity, multiple gestation pregnancy, medical conditions affecting development, psychosis, schizophrenia or bipolar disorder in first-degree family members, and predominant language other than English spoken in the home (for details see Supplemental Online Materials). Parents provided informed consent for study participation. Data collection took place at four research sites: The University of North Carolina at Chapel Hill, the University of Washington, the Children’s Hospital of Philadelphia, and Washington University in St. Louis. Research protocols were approved by the Institutional Review Boards at each study site and parents provided informed consent for their children.

Infant siblings who met inclusion criteria were enrolled at 6 or 12 months of age and were followed longitudinally through school-age. Confirmation of proband autism diagnosis and determination of clinical best estimate diagnoses for enrolled siblings were completed as described in the eMethods. Primary analyses compare three groups: LL parents with no autistic children (n = 100), HL parents with only one autistic child (HL-Simplex; n = 131), and HL parents with more than one autistic child (HL-Multiplex; n = 58). We grouped the HL families for analyses as being multiplex if the infant enrolled in the study was classified as having an ASD outcome at any point in the study. HL-Simplex status was determined if the enrolled infant never met study criteria for an ASD outcome, and only one child in the family (the proband) had ASD, based on parent report. This approach to classification of multiplex families ensures a carefully characterized group, as ASD diagnoses for the autistic probands were confirmed by medical records and administration of the Autism Diagnostic Interview-Revised (ADI-R [42]) by trained study staff at the time of enrollment in the original IBIS study. HL families were excluded from the current study if the enrolled infant did not have an ASD outcome and parent report indicated that other children in the family (in addition to the proband) were autistic (diagnosed by community providers), or if any non-proband sibling autism status was unknown (n = 23 HL families excluded). This careful characterization of simplex and multiplex families ensures that diagnostic status and recurrence among included children in HL families was confirmed with study data.

Parent trait measures

Parent measures were collected at their child’s infant or school-age study visit using self- and informant-report questionnaires and a self-administered cognitive test. Informant reports were completed primarily by a spouse or partner (the enrolled child’s other biological parent). In cases where the enrolled child’s biological parents were separated, another person who knows the parent well (a new partner, sibling, parent, or close friend) completed the informant report. Autistic traits were measured via the Social Responsiveness Scale, Second Edition (SRS-2 [43]); our analyses utilized the SRS-2 total raw score [19, 44]. The adult forms from Achenbach System of Empirically Based Assessment (ASEBA [45]) were used to measure anxious, depressive, and ADHD traits; analyses used raw scores from the corresponding DSM-oriented subscale. We also included a measure of cognitive ability given associations between parental cognitive ability and heritability of autism [4, 46]. Parents completed a publisher-approved computerized version of the Shipley-2 assessment [47] and standard scores from Composite B of the Shipley-2, comprising the Vocabulary and Block Design subtests, were examined.

Statistical methods

Key demographic variables were described and compared using a chi-squared test. Maternal and paternal SRS-2 total raw scores, ASEBA DSM-oriented subscale raw scores, and standard Shipley-2 composite scores were compared across groups using an analysis of variance (ANOVA) model or analysis of covariance (ANCOVA) when covariates were present. A Tukey post-hoc test was performed to determine significant pairwise group differences. FDR correction was implemented to adjust p-values to account for multiple comparisons [48], reported below as q-values. All FDR adjustments were performed separately within each analytic family, with ANOVA models corrected within the set of ANOVA outcomes and correlation tests corrected within the set of pairwise correlations, to ensure appropriate control of type I error for each analysis type. For the FDR correction in each correlation matrix, FDR was applied only to the off-diagonal p-values derived from the correlation matrix to ensure appropriate control of multiple comparisons across unique pairwise associations while excluding redundant self-correlations on the diagonal. Power analyses were conducted and determined that our sample is adequately powered to detect small to medium effects (Cohen’s f 0.10 to 0.25 for between-group differences and moderate correlation values (r ≥ 0.475) within our smallest group of HL-Multiplex participants. Thus, while we are well powered to examine group differences in parental traits, we may be underpowered to detect small but meaningful parental trait correlations in the HL-Multiplex group. For further information on power analyses, see Supplemental Online Materials.

For each parent trait measure used in the current study, there was a primary and secondary report type (self-report or informant-report), with the secondary report substituted for missing primary reports to address missing questionnaire data (i.e., differing availability of self and informant reports for each parent) and increase statistical power. For measures of anxious, ADHD, and depressive domains from the ASEBA forms, self-report was the primary report type and informant report was the secondary report type used in analyses because they are largely internalizing symptoms that may be best rated by the person experiencing them [49, 50]. For the measure of autistic traits (SRS-2), the informant report was the primary report type and self-report was the secondary report type, as self-reports of autistic traits are underestimated in interrater agreement and sensitivity and specificity analyses [43, 51]. The Shipley-2 is self-administered, and only a single test score is available for either parent. To confirm that our method of substitution of was acceptable, we evaluated concordance of self- and informant-report by measure and group and conducted sensitivity analyses including score type (primary vs. secondary) as a covariate in the ANCOVA model; results of which are shown in eTables 3–4 in the Online Supplement.

To account for potential associations between parental neuropsychiatric traits and the mental health challenges of raising a child with a disability, we conducted sensitivity analyses adding the child’s Adaptive Behavior Composite (ABC) of the Vineland Adaptive Behavior Scales, Third Edition, Interview Form (VABS [52]) to our model. Prior work has shown that parental stress was correlated with adaptive functioning in their autistic children [53]. The VABS was administered in interview format with the child’s parent.

To assess whether trait scores were significantly related within parent dyads, indicative of assortative mating, Pearson correlations were calculated between mother and father scores for each trait. To assess the degree to which traits co-aggregated in parents, we computed within-parent Pearson correlations for each by group. FDR-correction was applied to adjust for multiple comparisons. Fisher’s Z transformation was applied to depict normally distributed z-values as reported in eTable 8 in the Supplement. Finally, logistic regression predicting HL-Simplex vs. HL-Multiplex group membership was used to evaluate the contribution of autistic and neuropsychiatric traits in predicting recurrence risk in families. Child’s sex and study site were included as covariates in the model. Nagelkerke’s R-square was used to estimate the proportions of variance explained by the model.

Results

Participant demographics and data availability

Participant demographics for each group are presented in Table 1. Chi-squared tests detected group differences in male-to-female ratio, ABC scores in children, and parental separation rates, as expected given higher autism rates in males, lower adaptive behavior scores in autistic samples, and higher rates of parental separation among families of children with autism [54]. eTable 1 reports data availability (after replacement) for each measure and by group. Availability of both the SRS-2 and ASEBA forms for mothers and fathers was high (≥ 77%) across groups (eFigure 1). Cross-informant agreement for parental traits yielded significant correlations on the order of 0.34 − 0.58 (eTable 3). These results are aligned with past reports of cross-informant agreement on the ASEBA [45] and SRS [20] and support our method of replacement for missing forms. Parent score distributions by trait and group are depicted in Fig. 1 and largely represent subclinical variation. Fathers’ self-report for the SRS had the highest proportion of scores in the clinically significant range (7.5% in the mild range, 3.8% in the moderate range).

Table 1.

Demographics of the enrolled child

	HL-Multiplex	HL-Simplex	LL	Overall	p-value
	(N = 58)	(N = 131)	(N = 100)	(N = 289)
Sex
Male	45 (78%)	72 (55%)	59 (59%)	176 (61%)	0.012
Female	13 (22%)	59 (45%)	41 (41%)	113 (39%)
Child Race
White	46 (79%)	107 (82%)	74 (74%)	227 (79%)	0.76
Black/African American	2 (3%)	3 (2%)	3 (3%)	8 (3%)
Asian	0 (0%)	2 (2%)	0 (0%)	2 (1%)
More than one race	8 (14%)	14 (11%)	8 (8%)	30 (10%)
Missing	2 (3.4%)	5 (3.8%)	15 (15%)	22 (7.6%)
Child Ethnicity
Hispanic	1 (2%)	9 (7%)	3 (3%)	13 (4%)	0.24
Non-Hispanic	54 (93%)	117 (89%)	82 (82%)	253 (88%)
Missing	3 (5.2%)	5 (3.8%)	15 (15%)	23 (8.0%)
Highest Parental Education Level
Some high school	1 (2%)	2 (2%)	0 (0%)	3 (1%)	0.16
High school	0 (0%)	4 (3%)	0 (0%)	4 (1%)
Some college/2-year degree	9 (16%)	12 (9%)	9 (9%)	30 (10%)
College degree	20 (34%)	37 (28%)	36 (36%)	93 (32%)
Some graduate level	3 (5%)	2 (2%)	3 (3%)	8 (3%)
Graduate degree	24 (41%)	72 (55%)	52 (52%)	148 (51%)
Not answered	1 (2%)	0 (0%)	0 (0%)	1 (0%)
Missing	0 (0%)	2 (1.5%)	0 (0%)	2 (0.7%)
Parents Living Together
Yes	48 (83%)	106 (81%)	94 (94%)	248 (86%)	0.04
No	7 (12%)	20 (15%)	6 (6%)	33 (11%)
Not answered	3 (5%)	5 (4%)	0 (0%)	8 (3%)
VABS Adaptive Behavior Composite
Mean (SD)	81 (16)	98 (11)	100 (11)	96 (14)	< 0.001
Missing	2 (3.4%)	9 (6.9%)	4 (4%)	15 (5.2%)

Fig. 1.

Parent trait score ranges by group. Parent trait score ranges by parent and by group. Panels a-e depict maternal score ranges and panels f-j depict paternal score ranges. Red represents the HL-Multiplex group, green represents the HL-Simplex group, and blue represents the LL group. Panels depicting the Shipley instrument (a, f) plot the Composite B standard score. Panels depicting the SRS instrument (b, g) plot the total raw score and include cutoff lines indicating scores that reach the “Mild”, “Moderate”, and “Severe” thresholds as indicated on the SRS Profile Sheets. Panels depicting the ASEBA DSM-oriented scales for ADHD Problems (c, h), Anxiety Problems (d, i), and Depressive Problems (e, j) plot the raw score for each subscale

Parental trait comparisons

Mothers’ anxiety (F(2, 262) = 11.14, q < 0.001) and depressive problem scores (F(2, 262) = 7.08, q = 0.005) captured by the ASEBA differed across groups (Table 2). Pairwise post-hoc comparisons revealed elevated anxious (q < 0.001) and depressive problem scores (q = 0.004) in HL-Multiplex mothers compared to LL mothers. Father’s SRS scores differed across groups (F(2, 224) = 5.67, q = 0.013), such that scores were elevated in HL-Multiplex fathers compared to LL fathers (q = 0.009). No group differences were observed in cognition or ADHD traits for mothers or fathers. Pairwise comparisons did not differentiate HL-Simplex from either of the other groups. Adjusting for the child’s adaptive behavior level and parental score type in analyses resulted in similar findings to those outlined above (eTable 4).

Table 2.

Group comparisons of parental cognitive, autistic, and neuropsychiatric traits

	HL-Multiplex (a)	HL-Simplex (b)	LL (c)
	(N = 58)	(N = 131)	(N = 100)	Overall Comparison			Groupwise Comparisons1
				F-statistic	p-value	q
Maternal Traits
Shipley	110 (12)	110 (12)	120 (9.8)	0.76	0.469	0.521
SRS Total	36 (20)	33 (20)	28 (20)	3.21	0.042	0.105
ADHD Problems	5.7 (4.2)	5.1 (4.6)	4.2 (3.7)	2.27	0.105	0.165
Anxiety Problems	5.7 (3.1)	4.6 (2.6)	3.6 (2.4)	11.14	< 0.001	< 0.001***	a > c ***
Depressive Problems	7.0 (5.0)	5.5 (4.3)	4.2 (4.1)	7.08	0.001	0.005**	a > c **
Paternal Traits
Shipley	120 (12)	120 (11)	120 (12)	0.35	0.709	0.709
SRS Total	37 (26)	29 (23)	23 (19)	5.67	0.004	0.013*	a > c**
ADHD Problems	6.3 (5.3)	5.8 (4.8)	4.9 (4.5)	1.6	0.204	0.255
Anxiety Problems	4.1 (2.9)	3.8 (2.8)	3.2 (2.5)	2.23	0.110	0.165
Depressive Problems	5.3 (4.2)	4.7 (4.3)	3.9 (3.4)	2.18	0.115	0.165

¹Direct group comparisons conducted via Tukey post-hoc tests. FDR corrected p-value (q) for group difference indicated by asterisks (* < 0.05, ** < 0.01, *** < 0.001); for full table of groupwise comparisons see eTable 7

Between- and within-parent trait correlations

Autistic (r = 0.51, q < 0.001), depressive (r = 0.31, q = 0.007), ADHD (r = 0.24, q = 0.03), and cognitive (r = 0.36, q = 0.03) traits were significantly correlated in LL parent dyads (Fig. 2, eTable 5). Autistic traits were correlated in HL-Simplex parent dyads (r = 0.30, q = 0.013), but not HL-Multiplex dyads. Scatterplots indicate that the lack of association in HL-Multiplex dyads is driven by discordant elevation in SRS in one parent (typically the father), but not the other (eFigure 2). No other traits were significantly correlated among parent dyads of either HL group.

Fig. 2.

Trait correlations by group. Matrices depict Pearson correlations between parents (top left of each panel) and within parents (i.e., cross-trait; top right for fathers, bottom left for mothers) by trait and by group: (a) HL-Multiplex, (b) HL-Simplex, and (c) LL. Cells with a black border indicate between-parent trait correlations for identical trait domains. Darker shades indicate stronger correlation values. Asterisks indicate significant FDR-adjusted p-values (q) < 0.05

Autistic traits were positively correlated with anxious, depressive, and ADHD traits in both LL and HL-Simplex mothers (0.35 ≤ r ≤ 0.46, all q ≤ 0.001). Autistic traits were correlated with depressive traits in LL (r = 0.47, q < 0.001) and HL-Simplex (r = 0.36, q = 0.001) fathers, as well as anxious traits in LL fathers (r = 0.28, q = 0.02), and ADHD traits in HL-Simplex (r = 0.35, q = 0.001) fathers. There were no significant correlations between autistic traits and other neuropsychiatric traits in HL-Multiplex parents. HL-Multiplex mothers exhibited strong correlations between anxious and depressive traits (r = 0.78, q < 0.0001) (eTable 6).

Parent traits and autism recurrence

Parent traits jointly account for 7.9% of the variance in autism recurrence in HL families, with the strongest predictors being maternal depressive problem scores (3.3%) and paternal SRS score (2.0%) according to the logistic regression. Sex of the child predicted 6.2% of the variance (Table 3).

Table 3.

Contributions of parental traits to explaining ASD recurrence in HL-Multiplex vs. HL-Simplex families

Variable	R2
Total	0.179
Sex	0.062
Site	0.038
Mother SRS Total	0.001
Father SRS Total	0.020
Mother ADHD Problems	0.007
Father ADHD Problems	0.001
Mother Anxiety Problems	0.004
Father Anxiety Problems	0.005
Mother Depressive Problems	0.033
Father Depressive Problems	0.008

Discussion

Our study is, to our knowledge, the first of its kind to examine quantitative variation in continuous measures of autistic, anxious, depressive, ADHD, and cognitive traits of biological parents of autistic and non-autistic children. Group comparisons revealed higher subclinical levels of autistic, anxious, and depressive traits in multiplex autism families, and no group differences in ADHD-related traits or cognitive ability. Patterns of assortative mating and within-parent, cross-trait correlations differed by trait and by group. The findings of the current study highlights the need for a transdiagnostic approach to studying and indexing genetic liability for autism in families.

Fathers of HL-Multiplex families had elevated levels of autistic traits compared to LL fathers of non-autistic children, aligning with reports of elevated levels of autistic traits in fathers of autistic children [19, 55]. Multiplex and simplex fathers had comparable trait levels, as previously reported [16, 55]. We found evidence of assortative mating for autistic traits for HL-Simplex and LL families, but not HL-Multiplex families, which were more frequently characterized by discordant elevations in a single parent (typically the father), consistent with prior reports of BAP features [56] and SRS scores [19]. This could indicate that when both parents have elevated ASD traits, they may be less likely to have children or participate in research [16, 56]. Between-parent correlations in the LL sample (r = 0.51) align with prior literature reporting moderate correlations in SRS scores between parents in epidemiologically ascertained samples of children born around the same time as those in our study (ICC of 0.45) [57]. Between-parent SRS score correlations in the current study and in Forsen and colleagues’ study [57] demonstrate an observed increase in between-parent correlations of SRS scores from previous generations (r = 0.30) [18], suggestive of increases in assortative mating for autistic traits in the general population over time which could underlie increasing prevalence of autism in the US population [57]. Larger multi-generation, population-level datasets could be useful in interrogating this further [58].

We found that anxious and depressive traits were elevated, and highly correlated, in HL-Multiplex mothers compared to LL mothers. These findings remained after accounting for adaptive behavior of the child enrolled in our study, suggesting that elevations in anxious and depressive traits are not entirely attributable to providing support for autistic children who have lower adaptive skills in our sample. Prior work characterizing the BAP has similarly reported elevated levels of depression and anxiety in parents of autistic children when compared to parents of children with Down syndrome who face similar experiences of raising children with a disability [39]. Our findings extend work from other infant sibling and epidemiological studies reporting higher rates of depression and anxiety diagnoses in first-degree family members of autistic children [28–31, 40] to demonstrate that subclinical variation in symptomology is present among multiplex mothers as well.

Interestingly, paternal SRS scores did not correlate with maternal anxiety or depression in HL-Multiplex families, mirroring the unique (though small), additive contributions to the prediction of autism recurrence in our model. These findings suggest that these traits represent separable (or only minimally overlapping) components of genetic liability that act additively to increase the likelihood of autism in offspring, as has been proposed in prior family studies demonstrating ‘complementary pairing’ of BAP traits across mother-father dyads [59]. This aligns with a large population-based study that found that an individual’s ASD polygenic score and their family’s psychiatric history are non-overlapping indices of risk [60]. It is also possible that polygenic liability for autism may manifest differently in unaffected mothers and fathers: for fathers this may be expressed in elevated autistic traits, while in mothers this may present as elevations in anxious and depressive traits. Furthermore, anxious and depressive trait elevations could represent a unique presentation of autistic traits for female adults, as well as autistic trait masking in females [61]. Either interpretation calls for future work characterizing the genetic architecture of autism to also index inherited liability for anxiety and depression.

We found that for both mothers and fathers, autistic traits and the other neuropsychiatric traits we measured tended to travel together (i.e., were significantly correlated) in LL an HL-Simplex parents, in line with shared genetic underpinnings across traits [22]. While we found evidence of assortative mating for autistic, depressive, ADHD, and cognitive traits in the LL sample, only autistic traits were significantly correlated among HL-Simplex parents. Patterns of co-aggregation of autistic and neuropsychiatric traits within parents were less evident in the HL-Multiplex sample, as were correlations between pairs that would be indicative of assortative mating. This could suggest distinct patterns of parental paring (e.g., the complementary pairing of BAP features [59]) and trait presentations in parents of multiple children with ASD that deserves further study in larger samples that also include genomic data. It could also be an artifact of the relatively small sample size of the HL-Multiplex group in this study.

No differences were observed in parental cognitive ability across groups, mirroring prior work reporting no broad-scale differences in cognition between parents with and without autistic children [38], though differences in specific cognitive processing domains related to executive function have been reported [34, 38]. We also found no differences in ADHD traits across groups, which was surprising given prior work suggesting that ADHD and autistic traits correlate in parents of autistic children [62], and given high rates of ADHD among first-degree relatives of autistic children [63]. This may be a limitation of the measures used in this study and future work should consider tools designed to better capture variation in ADHD- and attention-related traits.

Limitations

While our study has many strengths, including a deeply phenotyped cohort of biological parents with children followed from infancy and assessed and diagnosed by clinicians who are experts in autism, certain factors may limit the generalizability of findings and should be addressed in future work. First, findings from this study may not generalize to more heterogeneous samples with differing racial, ethnic, and socioeconomic demographics (See Table 1). Importantly, families were excluded from the original study if either parent was diagnosed with intellectual disability, schizophrenia, bipolar disorder, or psychosis; while some parents have subsequently reported diagnoses during the course of the study, this sample is comprised of mothers and fathers who may carry a lower trait burden than expected among the broader population of parents of autistic children. While there is compelling evidence to conclude that elevated subthreshold autistic and neuropsychiatric traits in HL-Multiplex families are largely attributable to genetic transmission [4, 22, 24], it is possible that parenting stress could be a contributing factor to the elevations in psychiatric traits (i.e., anxiety and depression in mothers of autistic children) [64]. This, however, is not likely to be the main driver of our results as prior work suggests that neuropsychiatric conditions in parents of autistic children are present prior to the environmental impact of raising an autistic child [29, 65]. Further, elevations in parental psychiatric traits in families with autistic children in the current study remained after correcting for the enrolled child’s adaptive ability, suggesting that increased support needs for the child, and by extension parenting stress [53], do not wholly account for parental psychiatric phenotypes. Future molecular genetic investigations incorporating polygenic scores for ASD and other neuropsychiatric conditions, rare variant burden information, and family-based sequencing will be essential to parsing genetic and environmental effects and building additional lines of evidence to support the findings of this study. The ASEBA was used in the current study to measure neuropsychiatric traits across ADHD, anxious, and depressive domains. Future studies employing more targeted quantitative trait scales for each of these domains could be useful in further characterizing the findings in our study. Finally, our power analysis indicated that limited sample size for the HL-Multiplex group could have precluded the ability to detect more subtle, but meaningful, cross-trait and between-parent correlations.

Conclusions

In our cohort study, we report elevations in subclinical, continuous measures of autistic, anxious, and depressive traits in biological parents of multiplex autism families, calling for a transdiagnostic approach to quantifying genetic liability in families with autistic children. Our findings highlight the power of parental phenotypes for explaining autism liability in offspring and have important implications for future work aiming to improve early identification of autism recurrence in families. For example, parental traits could be considered in combination with behavioral data [66], neuroimaging data [67, 68], or genetic data [60, 69, 70] as a means to accurately identify infants at the highest likelihood for autism among HL families who already have one diagnosed child. This is essential for designing, testing, and implementing feasible, cost-effective early interventions [71] that may one day have broader applications to other infant groups.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

ABC

Adaptive Behavior Composite

ADHD

Attention–deficit/hyperactivity disorder

ASD

Autism spectrum disorder

ASEBA

Achenbach System of Empirically Based Assessment

BAP

Broad Autism Phenotype

DSM

Diagnostic and Statistical Manual of Mental Disorders

HL

High likelihood

IBIS

Infant Brain Imaging Study

LL

Low likelihood

SRS

Social Responsiveness Scale, second edition

Author contributions

LJM contributed to data acquisition, data analysis, interpretation of data, and drafted and revised the manuscript. KX contributed to data analysis, interpretation of data, and revision of the manuscript. JW contributed to data acquisition and data analysis. CML contributed to data acquisition and revision of the manuscript. TGB contributed to data acquisition and data analysis. CL, CAB, JTE, and MDS contributed to data interpretation and revision of the manuscript. SRD, AME, and HCH contributed to study design. NM, JP, RTS, TSJ, and LZ contributed to study design and revision of the manuscript. JNC contributed to study design and manuscript conceptualization, drafting and revisions. JP led the study and contributed to data interpretation and manuscript revisions. JBG contributed to study supervision and conceptualization, data interpretation, and manuscript drafting and revision. JP and JBG contributed equally to this work. All authors read and approved the final manuscript.

Funding

Funding for the research reported in the current study includes: K01-MH122779 (Girault), K23HD112507 (Burrows), R01-HD055741 (Piven), P30-HD003110 (Piven), MH118362–02S1 (Shen), and the Foundation of Hope (Girault).

Data availability

The deidentified data analyzed during the current study will be stored in the NIMH Data Archive, (https://dx.doi.org/10.15154/m84b-k079) and (https://dx.doi.org/10.15154/9hk1-q442), and will be made accessible to researchers whose proposed use of the data has been approved for a specified purpose after approval of a proposal.

Declarations

Ethics approval and consent to participate

Research protocols were conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Boards at each study site. Parents provided informed consent for their children to participate.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.