Edt

Genetic Influences on RA: Susceptibility, Severity, and Response to Treatment

By: S. Louis Bridges, Jr., MD, PhDProfessor of Medicine and MicrobiologyInterim Director, Division of Clinical Immunology and RheumatologyUniversity of Alabama at BirminghamDirector, UAB RA ClinicBirmingham, AlabamaDecember 5, 2008

Dr. Bridges has no financial disclosures to make.

Outline Introduction Genetics of Susceptibility to rheumatoid arthritis (RA)

o HLA-DRB1 RA in African-Americans

o Epidemiologyo HLA-DRB1 and Genetic Admixture

Non HLA Genes Associated with Susceptibility to RA Influences on Radiographic Severity of RA Personalized Medicine in RA

o Pharmacogenetics The Future of Genetics Research in RA

IntroductionPrevalence of RA in various populationsThe prevalence of RA is anywhere from 0.5 percent to 1 percent in most populations. There are a few populations that have a high prevalence of RA such as the Pima Indians and Chippewa Indians (5 percent to 7 percent). There isn’t much data on African-Americans or black Africans. There are some hints but not really strong data.

Pathogenesis of RAThe target joint in RA is the synovium, the thin membrane that surrounds the joint. In RA, it's infiltrated with TMB cells, becomes hyperplastic and erodes into the joint, forming an inflammatory tissue called pannus, which is shown below invading into the cartilage.

Patterns of lymphocytic infiltrates in RA synoviaThere can be different patterns of lymphocytic infiltrates in RA synovia, including those that look like germinal center-type structures within the inflammatory RA tissue.

A theoretical model: pathogenesis of RAThe causes of RA are unknown. Before 1938, it was called chronic infectious arthritis because of its resemblance to an infection. However, something causes the disease and at some point in the disease it's likely that perpetuating factors take over different from the susceptibility-initiating factors, making it difficult to “turn off” the disease. Modern drugs are often more successful at that.

Genetics of susceptibility to rheumatoid arthritis Genetic and environmental contributions to complex disordersRA is not a monogenic disease, it's a complex disease––many variants of small effect contribute to disease risk, along with many environmental factors. There are multiple genes that influence RA, and a large degree of environmental influence; either from infections, smoking, pollution or other things that are difficult to measure.

SusceptibilityThere is familial aggregation of RA. The concordance rate of monozygotic twins studies is anywhere from 30 percent to 50 percent. Genes are estimated to account for 15 percent to 40 percent of total disease risk and in Caucasians a third of the genetic component in RA is attributed to the HLA-DRB1 locus.

Association of the B-cell Alloantigen DRw4 with RA(Stastny, P. New Engl J Med 1978:298;869–871) 80 white patients with erosive, rheumatoid-factor-positive rheumatoid arthritis. The B-cell alloantigen HLA-DRw4 occurred in:

o 70% of 54 RA patientso 28% of 68 normal controls (P <0.00001)

The results indicate that RA in whites is associated with genes of the HLA-D region and that immunogenetic factors linked to HLA have a role in its pathogenesis.

Location and organization of the HLA complex

The HLA locus is shown in the figure above, on chromosome 6. HLA-DRB1 is located here among the class III genes you can see TNF-alpha and lymphotoxin here. This region is replete with genes that have immune function. And in fact, it's the most gene dense region in the entire genome. So these are immune-related genes. Each dash represents a gene that has possibly an immune function.

Structure of HLA - class I and class II moleculesCD40-positive T cells express class II MHC. An antigen-presenting cell ingests the antigen, and it's cleaved into peptide fragments from 6 to 20 amino acids. It's stuck in a groove and expressed on the cell surface. Then a T cell comes along, recognizes this through its T-cell receptor and if there are post-stimulatory molecules, the T-cell cascade ensues.

HLA-DRB1 shared epitope in RA Third Hypervariable Region Sequences of DRB1 Chains Associated With RAThere are 9 or 10 HLA-DRB1 alleles associated with RA. Gregersen and Winchester published a seminal paper that looked at the entire amino acid sequence evolved at the time (200 or so), HLA-DRB1 alleles and the possibility of a common pattern for association with RA. They found a unique sequence encoded by glutamine, lysine arginine and alanine. These amino acids are located at 70 through 74, which theoretically might affect what peptide sits in that antigen. After 20 some years of analyzing, it’s still not entirely understood why this shared epitope is associated with RA, but it has been consistently shown in many populations.

Citrullination (delimination) of peptidylarginine by peptidylarginine deiminaseThere are relatively specific autoantibodies to cyclic citrullinated peptide (CCP)––they're fairly specific for RA. Citrulline is an arginine residue that has undergone hydrolysis to create a citrulline residue. The amino group is changed and this can occur in a variety of proteins - vimentine, polagrine and others. And why it is that this is specific for RA is really unknown. But the presence of CCP antibodies is relatively specific for RA.

Relative risk of developing RASubjects Exposed to Different Combinations of Smoking and HLA–DR SE AllelesKlareskog and colleagues observed that the HLA-DRB1 association with RA is restricted primarily to CCP positive RA. In their study, they also found a risk of smoking. The relative risk of RA increases with the shared epitope and also increases with smoking in patients with CCP antibody positive RA but not in CCP negative RA.

Case–Control analysis for 2,221 SNPs spanning 10.7 Mb across the extended MHC region

Chromosome position is shown on the x axis; the y axis shows the –log P-value.(a) Case–control association statistics by Amitage trend test for anti-CCP positive RA vs control group. (b) Case–control association statistics by Amitage trend test for the anti-CCP-negative RA vs control group. The association of RA is very strong where the HLA-DRB1 locus is.

RA in African-AmericansEpidemiology of RA in African-Americans

The prevalence of RA in Caucasian populations in Europe and North American ranges from 0.5 to 1.0%. Surprisingly little is known about the incidence and prevalence of RA in African-Americans.

Epidemiology of RA in Africans Several small studies suggest that RA is rare in black African populations. In a rural population in remote South Africa, there were no cases of RA among 543 subjects comprising 97% of

the population. A survey of 3 hospitals serving that area identified only 14 cases of definite RA (of a population of ~520,000)

for a prevalence of 0.0026%. (Brighton et al. J.Rheumatol. 15:405, 1988.) In a population screening survey of 2,000 inhabitants of 2 rural townships in Nigeria, no documented cases of

RA were found. Simultaneous monitoring of the health clinic serving the townships for 4 months also failed to reveal a case of

RA. (Silman et al. J.Rheumatol. 20:618, 1993.) In a study of 15,834 black new patients seen in a 12-month period in a clinic in Lesotho, only 39 (0.2%) met

criteria for RA. (Moolenburgh et al. Ann.Rheum.Dis. 43:40, 1984.) In Quibdo, Colombia a stratified sampling of all African Colombians aged > 18 with arthralgias was performed. Questionnaires, radiographs of hands and feet, and serologic testing were performed. 18 individuals fulfilled ACR criteria for RA, a period prevalence of 0.01% (95% CI 0.008-0.02). (Anaya et al.

Semin.Arthritis Rheum. 31:191, 2001.) In an adjusted denominator population of 1,046 black Caribbeans and 997 whites living in Manchester, UK, the

cumulative prevalence of RA was 2.9/1000 in black Caribbeans and 8/1000 in whites. This represents prevalence in Black-Caribbeans of 0.36 times that found in whites. (MacGregor et al.

Ann.Rheum.Dis. 53:293, 1994.)

Some information about the prevalence of RA in black African ancestry is known, but not a lot. A lot of African-Americans with RA are seen in clinics and when rheumatologists are asked, "Do you not a prevalence difference?"; the anecdotal answer is, "no."

People to People Ambassador ProgramSouth Africa is a country of 48 million people with 40 rheumatologists. There are about 500 patients on biologic agents for RA in the entire country. They see RA quite a bit. Some of the earlier studies that didn't find it, so either the incidence has changed or the methodology is better. Anecdotally, the incidence of RA is less common in black Africans than it is in the white African population, but that's not quantitated or published anywhere.

Unanswered questions about RA in African-Americans What is the role of HLA DRB1 alleles containing the shared epitope in susceptibility to RA and its radiographic

severity? Are genetic polymorphisms associated with RA in other ethnic groups (e.g. PTPN22 and PADI4) associated

with RA susceptibility? Is there novel susceptibility or protective loci in African-Americans? What non-genetic factors (e.g. smoking), influence susceptibility to RA?

Consortium for the longitudinal evaluation of African-Americans with early RA (CLEAR) The CLEAR registry was started in 2000 and successfully renewed in 2006 for a total of 11 years of funding. The goal is to have 1,000 African-American RA patients by the end of the study: about 350 with early RA and about 650 with longstanding RA; and about 500 controls. There will be clinical data, radiographic data, serum and DNA, as well as RNA and DEXA on some of them.

Baseline characteristics of African-American RA subjects in the CLEAR registryCharacteristic

Onset age, mean (± SD) 51.2 ± 13.3 years

Disease duration, mean (± SD) 13.5 ± 7.2 months

Gender (female) 81%

Positive rheumatoid factor 73%

Positive anti-CCP antibody 62%

Use of methotrexate at baseline visit 59.3%

Use of other DMARD at baseline visit 22.0%

Current cigarette smoking at baseline visit 33.9%

HLA-DRB1HLA-DRB1 alleles containing the shared epitope are found in 60-70% of Caucasians with RA versus about 40% of controls. The role of the HLA-DRB1 SE in African-Americans with RA is unclear.

There is an association of shared epitope alleles with RA in African-Americans. There may be protective alleles or it may be that certain other alleles are deleted in order to enrich for the shared epitope. Forty-three percent of African-Americans with RA had at least one shared epitope allele versus 25 percent of the controls and the P value was very high.

Looking at the 60 percent to 40 percent calculation (Caucasians versus control) and the 43 percent to the 25 percent calculation (African-Americans versus control), the P value is very significant in both, but baseline allele frequency of shared epitope alleles is very different. There is an association, but what is the absolute contribution? The absolute contribution of HLA-DRB1 alleles containing the SE to susceptibility to RA in African-Americans appears to be less than in Caucasians.

Is the association of the HLA DRB1 shared epitope the result of genetic admixture of European ancestry into the African-American population? Typological race does not exist; there is no “pure Caucasian” or a “pure black African”. Race is complex; it's composed of ethnic heritage, biogeographical ancestry and interaction.

Ancestry informative markers are single nucleotide polymorphism (SNPs) with large (>30 percent) differences in minor allele frequencies minor between two populations under study (e.g. Africans and Europeans). An example of using ancestry informative markers is taken from a paper that focused on looking for pharmacogenetic markers in methotrexate. When the MTHFR gene was sequenced in 20 African-Americans and 20 Caucasians a polymorphism was found. The minor allele frequency, the C allele, was near zero in normal Caucasians, but it was above 30 in normal African-Americans. This is what is referred to as an ancestry informative marker.

There was a difference in admixture based on shared epitope status. Among patients who had no shared epitope alleles, the mean European ancestry was 15 percent, with 1 shared epitope, it was 16 percent and with 2 shared epitopes, it was 20 percent. Looking at presence versus absence, this difference was statistically significant.

Does genetic admixture increase the risk for RA in African-Americans?The genetic risk––a larger proportion––would be from the shared epitope. When you have genetic admixture there would be an intermediate frequency, at the general population of African-Americans, that's higher of the shared epitope alleles and whatever environmental triggers exist, occur in those that have a shared epitope and they are more likely to get the disease.

Non-MHC genetic associations in RAAssociation studies support a role for several non-MHC genetic associations in RA patients of European or Asian ancestry: PADI4, SLC22A4, RUNX1 – the enzymes that catalyzes the citrullination reaction that’s been associated with

RA in Japanese. PTPN22 – associated with RA in the Caucasian population. STAT4 – associated with RA in the Asian and Caucasian population. CTLA4 – has been somewhat controversial in the Caucasian population

PTPN22 SNP and autoimmune diseasesPTPN22 is a phosphatase expressed predominantly in T cells. Interestingly, it's associated with type-1 diabetes, lupus, Grave's disease and other autoimmune disease, in addition to RA. One of the emerging themes in genetic associations and autoimmune disease is the concept of “cassettes” of autoimmune disease genes that are common to different autoimmune disease.

Frequency of the PTPN22 risk allele in discovery and replication samples and other populationsPopulation No. Subjects Risk Allele (T) Frequency

Discovery Controls (Caucasian) 475 0.088

Replication Controls (Caucasian) 926 0.087

Additional Caucasians 560 0.084

Mexican Americans 99 0.035

African-Americans 409 0.024

Africans 21 0

Han Chinese 100 0

T-cell regulation by CD28 and CTLA4Cytotoxic T-lymphocyte associated protein 4 (CTLA4, CD152) is a negative regulator of T-cell activation. As the T-cell activation signal propagates due to costimulatory B7 molecule (CD80, CD86) binding of CD28, the cell surface expression of CTLA4 increases, thereby competing for binding with CD28. CTLA4 also delivers an inhibitory signal to prevent further clonal expansion of the T-cell and can inhibit osteoclast formation.

CTLA4 in our African-American populationSome studies show an association of CTLA4 with RA in Caucasians, some show an absence of an association, some show it as positive but at a low odds-ratio only within the CCP positive. Using ancestry information markers, a minor allele frequency can be identified that is zero in the Asian/Caucasian population.

There is an association of RA in African-Americans with rs231778, which is intronic in intron II. Using the CLEAR I longitudinal patients, looking at AA versus AG versus GG at rs231778–it was associated with absence of RA. It was more common in controls than in patients with a p value of 10 -8. If we looked at our cross-sectional 10-14 and combined it was 10-26. The statistics are strong, but what is the biology? It could be that this intronic SNP has something to do with gene expression somewhere else. It’s not clear if this protective affect is real, and if so, how it fits biologically.

Conclusions A positive association was found with rs231778 and RA in African-Americans (p = 4.5 x 10-26). This SNP is only polymorphic in African populations. This association was detected in two collections of samples: CLEAR longitudinal and CLEAR cross-sectional

registries. This protective effect (OR = 0.19) may help explain the purported lower prevalence of RA among black

Africans, but the biologic function of this intronic SNP remains unknown. Our finding provides an example of an ethnic-specific genetic contributory factor to a complex autoimmune

disease.

Genome-wide association studiesThe genome wide association studies to find an odds ratio of 1.2 with many thousands of patients. These studies take a large number of people and do a million polymorphisms, ie. 650 on the illumina chip and up to the 1 million SNP chip.

In RA, some of the main regions that have popped up are the C5-TRAF1, the 6q23, which has a gene called TNFAIP3 and the 4q27. This is when we need to do functional genomics. There's a lot of controversy of these genes are found, but it hasn’t led to big advances in diagnosis, prognosis or treatment. The affects are real, but the genetics and the heritability are not always the same.

There are no genome-wide association studies in RA. Because of this, the African-American RA network has been proposed to get an additional 800 patients with CCP antibody positive. So I'm going to skip through some of this.

Chi-square test statistics of 744 case-control samples matched 1:1 by HLA-DRB1 genotype

The table above shows very strong links with equilibrium across this region–the polymorphisms are inherited together. One of the ways to get rid of that was to see if there were other risks. This paper took 744 patients and controls and matched them for the HLA-DRB1 alleles on both chromosomes. If there was a patient who was 0401 1001, they were matched with a control who was 0401 1001. Looking at the chi-squared values across the region, risks of TNF are shown. There may be additional genes in this area that are important as well.

Radiographic progression in 108 patients from the CLEAR registryErosion ScoreNo. pts (%)

JSN ScoreNo. pts (%)

Total ScoreNo. pts (%)

Non-Progressors No BL Damage

62 (57.4%) 70 (64.8%) 54 (50.0%)

Non-Progressors BL Damage

16 (14.8%) 10 (9.3%) 13 (12.0%)

Progressors BL Damage

30 (27.8%) 28 (25.9%) 41 (38.0%)

Potential markers of radiographic severity in RATreatment factors such as DMARD use; clinical factors such as age; and environmental factors such as smoking are all important in determining radiographic severity in RA.

Personalized medicine in RA: In the pipeline or a pipe dream?What is personalized medicine? It's the aim to develop a rational way to test or identify which drug is best for a patient.

The figure to the left is a patient that presented in September of 2005 when she was 20 years old. Her arthritis began four months post-partum. She had a family history, systemic symptoms and a lot of active synovitis. Her rheumatoid factor and CCP were negative. Her x-ray film showed that she had disease for approximately six months.

Treatment options include traditional DMARDS, corticosteroids and biological agents. The cost of these different options is quite different––the yearly cost of methotrexate (approximately $215) versus a TNF inhibitor (approximately $12000) is quite substantial.

The concept of personalized medicine is not unique to rheumatology or rheumatic disease. A New England Journal of Medicine study looked at a drug and a biologic agent that targets the epidermal growth factor receptor. Investigators looked at non-small cell lung cancer patients with a horrible prognosis. There was a subset of people that had terminal non-small cell lung cancer that got better. It tended to be patients of Japanese ancestry with a polymorphism in the EGF receptor.

Factors in choosing DMARDs in RA Results of clinical trials

(randomized/observational)o Efficacyo Side effects

Route of administration Reimbursement/cost Patient preference

Physician preference Complete response in 100% of patients No side effects Easily administered Nonperishable Long-acting Low cost

Etanercept versus Methotrexate (ERA Trial) Some work done on a cohort (Bathon JM, et al. New Engl J Med. 2000;343:1586-1593) compared etanercept to methotrexate. This work was done before the genome association era, and it showed how to target treatment based on the pathology or the pathway of the disease.

Interactions among TNF, LTA, TNF receptors, and TNF inhibitorsTNF is a cytokine, but it's not a typical cytokine. It's expressed as a trimer on the cell surface; it’s not released as other cytokines, TNF- converting enzymes cleave it off. Those same enzymes can also cleave off TNF receptors. Infliximab comes in and binds TNF; Etanercept comes in and binds TNF and lymphotoxin; and Humira also binds TNF.

SNPs and microsatellites analyzed in the Immunex/Amgen ERA trialA paper published in 2004 with Mike Seldin and Lindsey Criswell looked at the shared epitope, TNF microsatellite and a few SNPs in the TNF and lymphotoxin region. The findings showed if there were two shared epitope alleles, a patient was more likely to respond to Etanercept than if there were one or zero shared epitope alleles. Subsequently, there's been lots of controversy about which TNF SNPs associate with good response to TNF inhibitors.

Methotrexate pathwayMethotrexate is a polyglutamated drug––it goes into the cell. There are 2 – 10 glutamate residues that are stuck on it and it stays in the cell because it's very large. That's why the half-life of an intercellular molecule of methotrexate is long: it stays in the cell. The folate and adenosine pathway may influence methotrexate toxicity.

Treatment of early aggressive RA (TEAR) trialThe TEAR trial has been collecting samples for about five years. The trial basically studies patients that are given: Methotrexate and Enbrel; Methotrexate and sulfasalazine and Plaquenil; Methotrexate for six months and if the DAS is greater than 3.2 at 6 months, Etanercept is added and/or they go

to triple therapy; Methotrexate for six months and if the DAS is less than 3.2 at 6 months, sulfasalazine and Plaquenil are added

and/or they go to triple therapy.Everyone in this study is on placebo for the drugs that they're not assigned.

Challenges in Pharmacogenomics

Challenge Potential Approaches

Establishing that drug responses are heritable ·Twin studies; Family studies·Linkage between drug response and genomic loci in cell lines or animal models

Defining candidate genes ·Pharmacokinetic·Pharmacodynamic -Drug targets -Biological milieu in which drugs act·Disease genes and pathways

·Whole genome approaches

Defining drug responses ·Biomarkers -Surrogate -“Hard” end points

Data management, including uniform representation of phenotypic data

·Improved Bioinformatics·Centralized, web-accessible public database related genetic variants and drug responses

Reproducibility ·Replication Sets·Large study populations

Statistical analysis of associations ·New statistical methods, including consideration of haplotypes

Interrogating very large numbers of polymorphisms in large numbers of patients

·New platforms (e.g. chip- or bead-based)

Moving to clinical practice ·Reproducible study results·Cost-effectiveness·Health care provider education

Reasons for optimism for future personalized medicine in RA Explosion of Genetic Data

o Human Genome Projecto International HapMap Project

Rapidly Improving Methods for Proteomics, Genotyping with Faster Results at Lower Cost New Statistical Methods to Analyze Huge Datasets Targeted Therapies Allow More Focused Insights into Mechanisms of Action Physicians and Third Party Payers Will Take Notice if Highly Reliable Predictive Algorithms Emerge Potential Collaboration of Pharmaceutical Companies The future of genomics research The NIH views genomics research as a very high priority. They are embracing a new paradigm and adopting a vision that emphasizes "3Ps": prediction of pathogenesis; personalized, precise medicine; and preemption of disease before it occurs.

The Genetic Information Non-Discrimination Act of 2008 is very important because it says it enables physicians to use particular polymorphisms to predict RA severity and bad drug response, making it illegal for insurance companies to cease coverage or charge higher premiums when the more expensive drugs are prescribed.

The Arthritis Foundation Treatment Efficacy and Toxicity in RA Database TETRADTETRAD is an initiative through the Arthritis Foundation to create a large cohort that can be used to analyze results.

The proposed phase-I is to get it peer reviewed and establish the infrastructure. The proposed phase-II includes collaboration with other organizations (ACR REF, RA Alliance, etc) to co-fundraise/co-market and have an independent advisory board that will oversee and add sites.

There would ultimately be a TETRAD advisory board, which would not be the TETRAD investigators. It would be an external review board that would establish the research priorities, review the study requests, monitor the progress, oversee the legal and regulatory affairs and help identify funding. The TETRAD would not be a research study; it would be a collection of data and samples that would allow any investigators in the world to explore or replicate their findings.

What questions can be answered using TETRAD?

1. Are there baseline (prior to treatment) markers (SNPs, serum proteins, etc) that can predict efficacy levels of different therapeutic agents? Are there longitudinal changes in these or other markers?

2. Can similar markers be identified for toxicity or adverse events?3. Based on treatment responses to different drugs (anti-TNF, anti-B cell, anti-T cell, etc), can patients with RA be

divided into different subsets in which different pathogenetic pathways are predominant?

Future genetics research in RA and other chronic complex diseasesWhere is genetics and RA going? Copy Number Variants Whole Genome Sequencing

o Complete coverage of genome (1,000 Genomes Project) Epigenomics

o Epigenetic processes control normal growth and development o Understanding how and when epigenetic processes control genes during different stages of development

and throughout life. Modifications to DNA-associated histones Diet and exposure to environmental chemicals Expression Quantitative Trait Loci (QTL)

o Genetic Variants can Affect Gene Expression at Distant Loci