The promise of precision medicine lies in data diversity. More than the sheer size of biomedical data, it is the layering of multiple data modalities, offering complementary perspectives, that is thought to enable the identification of patient subgroups with shared pathophysiology. In the present study, we use autism to test this notion. By combining healthcare claims, electronic health records, familial whole-exome sequences and neurodevelopmental gene expression patterns, we identified a subgroup of patients with dyslipidemia-associated autism.
Subscribe to Journal
Get full journal access for 1 year
only $4.92 per issue
All prices are NET prices.
VAT will be added later in the checkout.
Tax calculation will be finalised during checkout.
Rent or Buy article
Get time limited or full article access on ReadCube.
All prices are NET prices.
Familial WES datasets can be obtained from https://ndar.nih.gov as Collections 1918, 2004 and 2042. The human neurodevelopmental transcriptome dataset is available at http://www.brainspan.org/api/v2/well_known_file_download/267666524. Functional annotations can be obtained from ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2go.gz and https://www.gsea-msigdb.org/gsea/downloads.jsp. EHRs and healthcare claims data used in the present study are not publicly available due to patient privacy concerns. Mouse phenotypes are available at http://www.informatics.jax.org/downloads/reports/MGI_GenePheno.rpt.
The code used in the present study is available at https://github.com/yuanluo/autism_precision_medicine.
National Research Council (US) Committee ona Framework for Developing a New Taxonomy of Disease. Toward Precision Medicine: Building a Knowledge Network for Biomedical Research and a New Taxonomy of Disease (National Academies Press, 2011).
Maenner, M. J. et al. Prevalence of autism spectrum disorder among children aged 8 years—autism and developmental disabilities monitoring network, 11 sites, United States, 2016. MMWR Surveill. Summ. 69, 1–12 (2020).
de la Torre-Ubieta, L., Won, H., Stein, J. L. & Geschwind, D. H. Advancing the understanding of autism disease mechanisms through genetics. Nat. Med. 22, 345–361 (2016).
Lord, C. et al. Autism spectrum disorder. Nat. Rev. Dis. Prim. 6, 5 (2020).
Li, J. et al. Integrated systems analysis reveals a molecular network underlying autism spectrum disorders. Mol. Syst. Biol. 10, 774 (2014).
Parikshak, N. N. et al. Integrative functional genomic analyses implicate specific molecular pathways and circuits in autism. Cell 155, 1008–1021 (2013).
Herbst, R. S. et al. Predictive correlates of response to the anti-PD-L1 antibody MPDL3280A in cancer patients. Nature 515, 563–567 (2014).
Chapman, P. B. et al. Improved survival with vemurafenib in melanoma with BRAF V600E mutation. N. Engl. J. Med. 364, 2507–2516 (2011).
Parsons, D. W. et al. An integrated genomic analysis of human glioblastoma multiforme. Science 321, 1807–1812 (2008).
Shi, L., Zhang, Z. & Su, B. Sex biased gene expression profiling of human brains at major developmental stages. Sci. Rep. 6, 21181 (2016).
Werling, D. M., Parikshak, N. N. & Geschwind, D. H. Gene expression in human brain implicates sexually dimorphic pathways in autism spectrum disorders. Nat. Commun. 7, 10717 (2016).
Jung, H. et al. Sexually dimorphic behavior, neuronal activity, and gene expression in Chd8-mutant mice. Nat. Neurosci. 21, 1218–1228 (2018).
Grissom, N. M. et al. Male-specific deficits in natural reward learning in a mouse model of neurodevelopmental disorders. Mol. Psychiatry 23, 544–555 (2018).
Rosenson, R. Measurement of blood lipids and lipoproteins. in UpToDate (ed. Post, T. W.) https://www.uptodate.com (accessed 22 January 2018).
Coleman, D. M., Adams, J. B., Anderson, A. L. & Frye, R. E. Rating of the effectiveness of 26 psychiatric and seizure medications for autism spectrum disorder: results of a national survey. J. Child Adolesc. Psychopharmacol. 29, 107–123 (2019).
Sikora, D. M., Pettit-Kekel, K., Penfield, J., Merkens, L. S. & Steiner, R. D. The near universal presence of autism spectrum disorders in children with Smith–Lemli–Opitz syndrome. Am. J. Med. Genet. A 140, 1511–1518 (2006).
Tierney, E. et al. Behavior phenotype in the RSH/Smith–Lemli–Opitz syndrome. Am. J. Med. Genet. 98, 191–200 (2001).
Gong, H. et al. Lipoprotein lipase (LPL) is associated with neurite pathology and its levels are markedly reduced in the dentate gyrus of Alzheimer’s disease brains. J. Histochem. Cytochem. 61, 857–868 (2013).
Beffert, U., Stolt, P. C. & Herz, J. Functions of lipoprotein receptors in neurons. J. Lipid Res. 45, 403–409 (2004).
Kysenius, K., Muggalla, P., Matlik, K., Arumae, U. & Huttunen, H. J. PCSK9 regulates neuronal apoptosis by adjusting ApoER2 levels and signaling. Cell. Mol. Life Sci. 69, 1903–1916 (2012).
David, M. M. et al. Comorbid analysis of genes associated with autism spectrum disorders reveals differential evolutionary constraints. PLoS ONE 11, e0157937 (2016).
Buchovecky, C. M. et al. A suppressor screen in Mecp2 mutant mice implicates cholesterol metabolism in Rett syndrome. Nat. Genet. 45, 1013–1020 (2013).
Kyle, S. M., Saha, P. K., Brown, H. M., Chan, L. C. & Justice, M. J. MeCP2 co-ordinates liver lipid metabolism with the NCoR1/HDAC3 corepressor complex. Hum. Mol. Genet. 25, 3029–3041 (2016).
American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, 5th edn (American Psychiatric Association, 2013).
Wong, C. T., Wais, J. & Crawford, D. A. Prenatal exposure to common environmental factors affects brain lipids and increases risk of developing autism spectrum disorders. Eur. J. Neurosci. 42, 2742–2760 (2015).
Wong, C. T. et al. Prostaglandin E2 promotes neural proliferation and differentiation and regulates Wnt target gene expression. J. Neurosci. Res. 94, 759–775 (2016).
El-Ansary, A. & Al-Ayadhi, L. Lipid mediators in plasma of autism spectrum disorders. Lipids Health Dis. 11, 160 (2012).
Kim, E. K., Neggers, Y. H., Shin, C. S., Kim, E. & Kim, E. M. Alterations in lipid profile of autistic boys: a case control study. Nutr. Res. 30, 255–260 (2010).
Tierney, E. et al. Abnormalities of cholesterol metabolism in autism spectrum disorders. Am. J. Med Genet. B Neuropsychiatr. Genet. 141B, 666–668 (2006).
Kang, H. J. et al. Spatio-temporal transcriptome of the human brain. Nature 478, 483–489 (2011).
Zhang, M. et al. Axonogenesis is coordinated by neuron-specific alternative splicing programming and splicing regulator PTBP2. Neuron 101, 690–706 e610 (2019).
Su, C. H., D., D. & Tarn, W. Y. Alternative splicing in neurogenesis and brain development. Front. Mol. Biosci. 5, 12 (2018).
Everitt, B. S. The Cambridge Dictionary of Statistics (Cambridge Univ. Press, 2006).
Csardi, G. & Nepusz, T. The igraph software package for complex network research. Inter Journal, Complex Systems, 16951704 (2006).
Rosvall, M. & Bergstrom, C. T. Maps of random walks on complex networks reveal community structure. Proc. Natl Acad. Sci. USA 105, 1118–1123 (2008).
Cleveland, W. S. & Devlin, S. J. Locally weighted regression: an approach to regression analysis by local fitting. J. Am. Stat. Assoc. 83, 596–610 (1988).
Ritchie, M. E. et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 43, e47–e47 (2015).
Benjamini, Y. & Hochberg, Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J. R. Stat. Soc. Ser. B Methodol. 57, 289–300 (1995).
DePristo, M. A. et al. A framework for variation discovery and genotyping using next-generation DNA sequencing data. Nat. Genet. 43, 491–498 (2011).
Krumm, N. et al. Excess of rare, inherited truncating mutations in autism. Nat. Genet. 47, 582–588 (2015).
Noreen, E. W. Computer-Intensive Methods for Testing Hypotheses: An Introduction (Wiley, 1989).
Neale, B., Ferreira, M. & Medland, S. Statistical Genetics (Taylor & Francis Group, 2012).
Lawrence, M. et al. Software for computing and annotating genomic ranges. PLoS Comput. Biol. 9, e1003118 (2013).
Fisher, R. A. Statistical Methods For Research Workers (Cosmo Publications, 1925).
Dunn, O. J. Multiple comparisons among means. J. Am. Stat. Assoc. 56, 52–64 (1961).
Subramanian, A. et al. Gene Set Enrichment Analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl Acad. Sci. USA 102, 15545–15550 (2005).
Bateni, M. et al. Affinity clustering: hierarchical clustering at scale. Adv. Neural Inf. Process. Syst. 2017, 6864–6874 (2017).
Kohane, I. S., Churchill, S. E. & Murphy, S. N. A translational engine at the national scale: informatics for integrating biology and the bedside. J. Am. Med. Inform. Assoc. 19, 181–185 (2012).
Medicode. ICD-9-CM: International Classification of Diseases, 9th Revision, Clinical Modification (Medicode, 1996).
World Health Organization. ICD-10: International Statistical Classification of Diseases and Related Health Problems (World Health Organization, 2004).
Denny, J. C. et al. Systematic comparison of phenome-wide association study of electronic medical record data and genome-wide association study data. Nat. Biotechnol. 31, 1102–1110 (2013).
Data analyzed in this manuscript reside in the National Institutes for Health (NIH)-supported NIMH Data Archive’s NDAR as Collection nos. 1918, 2004 and 2042. We thank all the families at the participating Simons Simplex Collection sites, as well as the principal investigators (A. Beaudet, R. Bernier, J. Constantino, E. Cook, E. Fombonne, D. Geschwind, R. Goin-Kochel, E. Hanson, D. Grice, A. Klin, D. Ledbetter, C. Lord, C. Martin, D. Martin, R. Maxim, J. Miles, O. Ousley, K. Pelphrey, B. Peterson, J. Piggot, C. Saulnier, M. State, W. Stone, J. Sutcliffe, C. Walsh, Z. Warren and E. Wijsman). We thank SFARI Base for access to their phenotypic data. Approved researchers can obtain the Simons Simplex Collection population dataset described in the present study (https://base.sfari.org/ordering/phenotype/sfari-phenotype/download?code=11) by applying at https://base.sfari.org. We thank J. Eichler, D. Margulies and members of the Kohane lab for fruitful discussions. We thank somersault18:24 (www.somersault1824.com) for illustrations. Y.L. was supported by the US National Institutes of Health (1R21LM012618 and 5UL1TR001422). A.E., P.S. and I.S.K. were supported by the National Institute of Mental Health (P50MH106933). A.E. was supported by the Israeli Ministry of Science and Technology (grant no. 17708) and by the PrecisionLink Initiative at BCH. N.P. received funding support from Aetna Life Insurance Co. P.A. was supported by the US National Institutes of Health (U01HG007530, OT3OD025466, OT3HL142480, U54HG007963, 1U01TR002623-01 and 1U54HD090255-01).
The authors declare no competing interests.
Peer review information Kate Gao was the primary editor on this article and managed its editorial process and peer review in collaboration with the rest of the editorial team.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Data modalities integrated in the present study.
In addition to examining the relation between ASD and dyslipidemia in the entire cohorts, we also restricted our analyses to individuals with no prescription records of drugs commonly prescribed in ASD which are known to alter lipid levels, namely atypical antipsychotics, anticonvulsants, and antidiabetics. Error bars indicate the 95% CIs for the proportions. a, Rates of dyslipidemia diagnoses in individuals with ASD (red) and individuals with no ASD diagnosis (cyan), stratified by drug use (entire cohort OR = 1.93, 95% CI = (1.88, 1.99), Fisher’s exact two-sided P < 1 × 10−323, n = 6,621,118 individuals; individuals not taking atypical antipsychotics, anticonvulsants, or antidiabetics OR = 1.73, 95% CI = (1.67, 1.79), Fisher’s exact two-sided P = 1.11 × 10−201, n = 6,488,315). b, Rates of ASD diagnoses in individuals with dyslipidemia (red) and individuals with no dyslipidemia diagnosis (cyan), stratified by drug use (effect sizes as in a). c, Fraction of individuals with abnormal fasting LDL levels out of individuals with ASD and at least one fasting LDL test result (red), and individuals with no ASD diagnosis and at least one fasting LDL test result (cyan), stratified by drug use (entire cohort OR = 1.48, 95% CI = (1.36, 1.61), Fisher’s exact two-sided P = 1.06 × 10−20, n = 48,775 individuals; individuals not taking atypical antipsychotics, anticonvulsants, or antidiabetics OR = 1.48, 95% CI = (1.27, 1.73), Fisher’s exact two-sided P = 6.16 × 10−7, n = 34,751 individuals). d, Fraction of individuals with ASD out of individuals with abnormal fasting LDL (red), and individuals with all fasting LDL test results within the reference range (cyan), stratified by drug use (effect sizes as in c). e-f, Same as c-d but for fasting total cholesterol (TC). Entire cohort OR = 1.69, 95% CI = (1.49, 1.92), Fisher’s exact two-sided P = 7.14 × 10−15, n = 43,650 individuals; individuals not taking atypical antipsychotics, anticonvulsants, or antidiabetics OR = 1.77, 95% CI = (1.36, 2.27), Fisher’s exact two-sided P = 2.00 × 10−5, n = 31,690 individuals. g-h, Same as c-d but for fasting triglycerides (TG). Entire cohort OR = 1.33, 95% CI = (1.20, 1.46), Fisher’s exact two-sided P = 1 .73 × 10−8, n = 47,650 individuals; individuals not taking atypical antipsychotics, anticonvulsants, or antidiabetics OR = 1.33, 95% CI = (1.10, 1.60), Fisher’s exact two-sided P = 2.99 × 10−3, n = 39,165 individuals.
Extended Data Fig. 3 Enrichment of dyslipidemia diagnoses in parents of children with ASD (maternal OR = 1.16, 95% CI = (1.12, 1.20), Fisher’s exact two-sided P = 5.28 × 10−18; paternal OR = 1.13, 95% CI = (1.09, 1.16), Fisher’s exact two-sided P = 1.92 × 10−14; n = 38,846 families vs. repeatedly resampled matched controls from a total of 34,003,107 individuals).
a, Association between maternal dyslipidemia and having a child with ASD. Shown is a forest plot detailing diagnosis-specific ORs by circles and their 95% CIs by horizontal lines. b, Association between paternal dyslipidemia and having a child with ASD. A diagnosis-specific forest plot is shown as in (a).
A forest plot depicts the association estimates for ASD-related clinical characteristics more common in individuals with ASD and dyslipidemia, as compared to individuals with ASD and no dyslipidemia (n = 80,714 individuals). ORs and their 95% CIs are shown by circles and horizontal lines, respectively.
Extended Data Fig. 5 Phenotypic clustering of ASD (blue), dyslipidemia (orange), and SLOS (red) mouse models.
Hierarchical clustering of 1,315 phenotypes measured in ASD (n = 34), dyslipidemia (n = 10), and SLOS (n = 1) mouse models identified four clusters. Three clusters (shown on top) include both dyslipidemia and ASD mice, with shared phenotypes such as seizures, abnormal synapse morphology, abnormal learning, abnormal brain size, and abnormal coordination. The fourth cluster (bottom) is ASD-specific. Thus, some ASD models are more similar to dyslipidemia models than to other ASD mice.
About this article
Cite this article
Luo, Y., Eran, A., Palmer, N. et al. A multidimensional precision medicine approach identifies an autism subtype characterized by dyslipidemia. Nat Med 26, 1375–1379 (2020). https://doi.org/10.1038/s41591-020-1007-0