Lynn Petukhova, [email protected]
Assistant ProfessorDepartments of Dermatology and Epidemiology
Columbia University
Alopecia Areata Research SummitNovember 15, 2016
Precision Medicine
“And that’s the promise of precision medicine -- delivering the right treatments, at the right time, every time to the right person.”
- President Barack Obama, January 30, 2015
Precise definitions of diseaseimprove patient outcomes and decrease healthcare costs.
How do we define disease mechanisms?
NIH Roadmap for Precision Medicine
Adapted from Francis Collins, ASHG, 2015
Large cohorts of engaged patients
Genomic datainforms on
biology
mHealth datainforms on
environment & behavior
EHR datainforms on
natural history and clinical trajectories
New therapeutic interventions and companion diagnostics
Large cohorts of engaged patientsLarge cohorts of engaged patients
Precise definitions of diseaseimprove patient outcomes and decrease healthcare costs.
Precision Medicine
“And that’s the promise of precision medicine -- delivering the right treatments, at the right time, every time to the right person.”
- President Barack Obama, January 30, 2015
Precise definitions of diseaseidentify patient subtypes and mechanistic links among diseases.
Disease subtypes and mechanistic links
disease subtypes
mechanistic link
Precisely defined disease mechanismsare at the crux of precision medicine.
alopecia areata rheumatoid arthritis
JAK-STATsignaling
HFvulnerability
Costimulatory
pathway
clinical implications
Disease subtypes and mechanistic links
Framework for Disease Comorbidities in Precision Medicine
Increased riskfor
Diagnosis B
A cohort of comorbid patients will be enriched for the shared disease mechanism.
Population riskfor
Diagnosis B
How do we identify comorbidities?
Epidemiological Studies of Comorbidities
Rzhetsky et al., 2007
Eaton et al., 2007
1. Ask the patient (lack power)2. Look in medical records (confounding)
Type 1 diabetes
Rheumatoid arthritis
Psoriasis
Systemic lupus erythromytosis
Multiple sclerosis
Goiter (hypothyroidism)
Myositis
Type 1 diabetes
Rheumatoid arthritis
Psoriasis
Vitiligo
Thyrotoxicosis/thyroiditis
Systemic lupus erythromytosis
Inflammatory bowel disease
ICD co-occurrence
autoimmune rheumatoid arthritis
multiple sclerosis
systemic lupus
erythematosus
type 1 diabetes
psoriasis
inflammatory hypersensitivity angiitis
allergic rhinitis
goiter
infection susceptibility Hepatitis C
meningococcus
streptococcus
tuberculosis
virus
CNS viral disease
helicobacter pilori
mumps
Hepatitis B
pertussis
neoplasm benign neoplasms
carcinoma in situ
neurofibromatosis
metabolic
disorders of lipid
metabolism
type 2 diabetes
cholelithiasis
AA metabolism (aromatic)
neuropsychiatric depression
migraine
epilepsy
bipolar
attention deficit
EHR studies of ICD co-occurrences at CUMC
Rhetsky et al, PNAS, 2007
GWAS
Biological Validation with PheWAS
PheWAS (requires a cohort with genetic data linked to EHR)
Leverage Public Databases linking EHR to Genome Data
Outcome• Groups are defined by disease status (case or control)
Exposure• Obtain genotypes
Statistical test
• Test for allele frequency differences between disease groups
• Identify Risk alleles
Outcome• Groups are defined by allele status at risk SNPs (risk or protective allele)
Exposure• Obtain all phenotypes in EHR
Statistical test
• Test for ICD frequency differences between allele groups
• Identify comorbid condition with a biological basis
Genetic Studies reveal comorbidities
11,410,409 SNPs imputed in our meta-analysis cohort,
revealing 16,848 associated SNPs across 14 GWAS loci
Biological Validation with PheWAS
https://phewascatalog.org/phewasDenny JC et al. Nat Biotechnol. 2013 Dec;31(12):1102-10
AA SNPS implicated 275 conditions, including autoimmune, inflammatory, cancers, cardiometabolic, and anxiety disorders.
ICD co-occurrence Phewas
autoimmune rheumatoid arthritis ✔
multiple sclerosis
systemic lupus erythematosus ✔
type 1 diabetes ✔
psoriasis ✔
inflammatory hypersensitivity angiitis
allergic rhinitis ✔
goiter
infection susceptibility Hepatitis C
meningococcus
streptococcus
tuberculosis
virus
CNS viral disease ✔
helicobacter pilori
mumps
Hepatitis B
pertussis ✔
neoplasm benign neoplasms ✔
carcinoma in situ
neurofibromatosis
metabolic disorders of lipid metabolism ✔
type 2 diabetes ✔
cholelithiasis ✔
AA metabolism (aromatic)
neuropsychiatric depression
migraine
epilepsy
bipolar
attention deficit
https://www.jax.org/strain/000659
Mouse Phenotyping
Alopecia areata mouse model (C3H/HeJ)
C3H/HeJ Phenotyping Results
autoimmune prone to colitis
prone to IgA nephropathy (exaggerated IgA responses)
infection susceptibility lethal infection by Gram-negative bacteria (defective lipopolysaccharide response; TLR4-LPS-d)
increased susceptibility to viral infection
abnormal T-helper 2 physiology
abnormal macrophage function
neoplasm high incidence of hepatomas
low incidence of mammary tumors
metabolic resistent to diet-induced atherosclerosis
high lipid levels; high cholesterol
elevated heme oxygenase
decreased circulating alanine transaminase level
neuropsychiatric absence seizures (Gria4spkw)
attenuated responses to tactile and thermal stimulation
retinal degeneration (100% prevalent; Pdebrd1)
abnormal glial cell apoptosis
prone to anxiety and impulsivity
disruptions in social behavior
[a characteristic of depression, autism, bipolar and schizophrenia]
Dermatological Diagnoses at Columbia University
Data on 22,291 patients
17,575 only a single diagnosis
Alopecia,
unspecified AA
benign
neoplasm keratosis acne
Other
disorders of
skin and
subcutaneous
tissue
Alopecia, unspecified 2369 1701 136 227 124 331 71
AA 752 136 503 53 29 104 16
benign neoplasm 7871 227 53 4838 1172 1816 263
keratosis 5787 124 29 1172 3668 1133 57
acne 9424 331 104 1816 1133 6139 445
Other disorders of skin and subcutaneous tissue 1412 71 16 263 57 445 726
Lipid Panel Patient Counts Cholesterol Total HDL Cholesterol LDL Cholesterol Triglyceride
AA 275 187.28 56.16 108.65 116.82
acne 5864 176.55 53.03 100.25 118.64
Benign neoplasm of skin 1318 185.28 54.48 106.12 124.48
keratosis 7844 171.86 52.90 94.06 127.74
Other disorders of skin and subcutaneous tissue 1049 186.45 52.54 107.64 129.73
Unspecified alopecia 1704 183.75 54.94 105.00 120.09
Grand Total 178.10 53.49 100.28 123.88
Preliminary Results
EHR data of Human Alopecia Areata Patients
Alopecia areata may include disease manifestations in cells and tissues other than hair follicle and immune system.
Disease mechanisms may contribute to dysregulation in lipid metabolism.
Biological basis to psychosocial conditions frequently reported by patients
Wrap up
Conclusions
Future directions NAAF funded study of comorbidities in National Alopecia
Areata Registry. Updated the questionaire to validate existing data.
Characterize the distribution of risk alleles for possibly comorbid conditions in the GWAS cohort.
Leverage EHR cohorts linked to genetic data to further pursue investigation of biological validation
Rheumatology
Joan Bathon
Gastroenerology
Ben Lebohwl
Govind Bhagat
Ali Jabbari
Jane Cerise
Annemieke de Jong
Zhengpeng Dai
Stephanie Erjavec
Alexa Abdelaziz
Claire Higgins
Muhammad Wajid
Sivan Harel
Yutaka Shimomura
Tarek Yamany
Esther Drill
Mazen Kurban
Hynumi Kim
Katie Fantauzzo
Courtney Luke
Rita Cabral
Gina DeStefano
Ming Zhang
Hazi Lam
Department of Dermatology
Angela M. Christiano
Julian MacKay-Wiggan
Neuropsychiatric Epidemiology
Ruth Ottman
Sharon S. Schwartz
Biomedical Informatics
Chunhua Weng
George Hripcsak
Neurology
Claire S. Riley
Cardiology
Alan Tall