Arthritis & Allied Conditions
15th Edition

Chapter 28
Genetic Basis of Rheumatic Disease
John B. Harley
Amr H. Sawalha
The enormous complexity of biologic systems renders the understanding of disease states very difficult. In many situations the critical abnormalities that initiate these processes are not known. The basic understanding of the etiology of disease has a profound and practical influence on the development of treatment and preventive strategies.
Etiology derives from two sources: genetics and environment. The genetic differences between individuals in a species lead to structural differences. These in turn change the probability that a genetic difference will “cause” or explain the etiology of the observed phenotype of interest, for our purposes, a rheumatic disease. Often genetic causes interact with or require other genes or environmental factors to be fully expressed.
This chapter explores how rapidly improving modern genetic strategies and newly available genomic resources are being used to discover the genetic components of the etiology of rheumatic disorders. The environmental component of etiology for individual disorders is discussed in chapters elsewhere in this text.
The available conceptual and technical methods provide not only the opportunity for discovery, but also set the limits for what is possible to discover. Genetic discovery in human diseases now relies heavily on improved DNA technology, more incisive applied mathematical methods, and better infrastructure describing and accessing the human genome. Each of these areas is in the throes of fundamental and very rapidly paradigm-shifting changes. Cumulatively, they are making previously impenetrable problems in the rheumatic diseases accessible to genetic solution. Indeed, the new boundaries of what might be possible are not well defined. Disappointment, through failure to meet current expectations, seems inevitable, as do unexpected successes through brilliant and paradigm-challenging experiments.
This is an exciting time to work in human genetics, with the high drama of enormous effort and human enterprise pursuing the unfulfilled promise and hope offered by a rapidly changing discipline. Consequently, subsequent volumes of this text are likely to describe the genetic components of most rheumatic diseases with the kind of detail that is well beyond our present understanding.
In the previous era of human genetics, only two decades ago, the underlying genetics were understood by inferring genotypes from phenotypes. A phenotype is an observed characteristic (or set of characteristics) in an individual, whereas a genotype is the heritable property responsible for that phenotype. Today, a genotype is the actual DNA sequence itself, and when the DNA is different among the typed individuals, the genotype becomes a description of the DNA variation. The previous generation of geneticists would find polymorphic differences in proteins or body appearance or capabilities (e.g., tongue rolling), and then infer probabilities for the possible genotypes at each locus. This would then be used to explain the observed marker phenotype. The phenotype of interest was evaluated relative to these genotypic probabilities.
For example, the first genetic linkage in humans was published in 1935, when Penrose showed that red hair color was linked to blood type A and B using affected pairs of siblings (1). The genotypic difference for red hair is complicated, but he relied only on the inference obtained from the presence or absence of the phenotypes. The ABO blood type phenotype sufficiently limits the possible genotypes that may be present at this second locus, and thereby provides the analytic capability to discover linkage. Relative to previous generations of rheumatic disease geneticists, we now enjoy the enormous advantage of directly determining genotypes at the nucleic acid sequence level of DNA (or RNA). This profoundly simplifying advance has greatly increased our ability to identify genetic relationships.
Of course, the genomic infrastructural resources being made available are also extremely important. Genetic maps containing sequenced genes, microsatellites, single nucleotide polymorphisms (SNPs), and disease linkages are rapidly
P.610

becoming more and more dense. Indeed, in early 2003 it was announced that the sequence of the human genome was complete. Those trying to use the currently available human genome sequence might disagree; however, the remaining ambiguities, gaps, and errors are being addressed and progressively eliminated.
Advances in applied mathematics have been as important for genetic progress as have advances in DNA manipulation, computer technology, and genomics. These methods further increase experimental productivity and make otherwise inaccessible genetic effects discoverable. Indeed, experimental design is dictated by the analytic methods available.
There are two standard strategies now used for disease-associated gene identification. The first is to test candidate genes for association. Here, a statistically-significant relationship is usually sought between a phenotype and an allele (also called a polymorphism) at a single locus. Rheumatic disease investigators have been exploring candidate genes for association for many decades. The many histocompatibility [human leukocyte antigen (HLA)] associations now known for the inflammatory disorders are among the best known examples of genetic association. In the next few years, the putative candidate gene associations are expected to substantially increase as the capacity to evaluate the entire genome for genetic association with a particular disease becomes a practical possibility.
The second strategy for disease-associated gene identification begins with genetic linkage. Once linkage has been established, efforts are made to confine the area of interest to the smallest genomic region possible. The region is then explored for candidate genes containing the polymorphisms responsible for the observed genetic linkage. This approach ultimately employs genetic association, but uses linkage to increase the prior probability of there being genetic association in a candidate region of the genome.
SEGREGATION
How does one tell if a particular phenotype represents a genetic problem amenable to solution? There are a few general rules. Identifying a genetic model explaining the mode of inheritance is powerful and virtually assures that there will be a successful outcome in the effort to find the responsible genes. The lower the genetic concordance between identical twins, the less likely that a genetic explanation for the phenotype will be found. A high identical twin concordance also helps, but may also be explained by a shared environment.
The distribution of diseases in the population varies according to the mechanisms by which they are generated. Some disorders appear to occur in families following the known rules of inheritance. These reflect the source of DNA as somatic DNA (Mendelian), sex-linked (X chromosome), patrilineal (Y chromosome), and matrilineal (mitochondrial DNA). Indeed, DNA is the repository for genetic information. Its variations, also called polymorphisms or alleles, at a gene (here meaning a locus) change the risk for a phenotype. The word gene is ambiguous, which leads to confusion. Gene is sometimes used to mean an allele and in other contexts is used to mean a locus. The ambiguity has historical roots, since the word gene preceded both the more specific use and meaning of allele and locus.
There are, for example, families whose members have a risk for calcium pyrophosphate deposition disease and the accompanying arthritis of pseudogout transmitted in an autosomal-dominant pattern of inheritance. Some disorders appear to be associated with the DNA of the mitochondria and are inherited only through the maternal lineage. Myopathies, such as myoclonic epilepsy with ragged red fibers, are examples of this form of inheritance.
Other diseases appear to have more complicated genetics. Although they appear to have a genetic component, the mechanism of inheritance responsible for the observed relationships is not apparent. Indeed, for many of the more common rheumatic disorders, such as rheumatoid arthritis (RA), systemic lupus erythematosus (SLE), osteoarthritis, and osteoporosis, it is widely suspected that many different genes are making variable contributions toward each of these phenotypes. These are commonly referred to as complex genetic disorders.
The effort to understand the pattern of disease phenotype inheritance is called segregation analysis. Knowing the pattern of disease inheritance is important. This can help determine which approach is more likely to be successful in finding the genes responsible.
Some diseases segregate as if they have a “founder effect.” This means that the disease phenotype should theoretically be able to be traced to a single DNA variation in a single progenitor individual (Fig. 28.1). A founder effect means that the affected individuals share an original affected ancestor whose critical piece of DNA, which causes the disorder, is shared among subsequent progeny affecteds. The process of establishing and identifying the responsible DNA variation in the presence of a founder effect is different and, in some ways, simpler than in situations where no founder effect is present. Huntington chorea is perhaps the most classic example. Virtually all of the thousands of individuals at risk for Huntington chorea appear to be the descendant of a single individual (2). Hemochromatosis, by virtue of being concentrated in one ethnic group where an HLA association had been identified, had long been suspected to have a founder effect (3). This prediction was used to great advantage in the successful effort to identify the responsible gene.
FIG. 28.1. Individual founder chromosome. A founder effect classically begins with the mutation of a single chromosome (*). Crossovers replace the chromosomal DNA from the founder (white) with DNA not from the founder chromosome (black) in affected individuals of subsequent generations. Crossovers continue to reduce the size of the region from the founder. Although subsequent affected individuals share the mutation (*) conferring risk for the phenotype and the surrounding DNA, the size of this shared founder DNA region varies among affected progeny. In addition, other founder DNA (white) will tend to be randomly distributed across the chromosome and will be present, on average, as a function of the number of generations and of the population structure (inbreeding).
Neils Risch developed a useful measure now commonly applied to the segregation analysis of complex diseases. The rate of phenotype concordance between a proband and his or her relatives versus the phenotype occurrence in the general population is used to assess the apparent capacity of a particular study to reveal underlying genetic effects (4). The rate of concordance of siblings of the proband divided by the rate of disease in the general population is such a
P.611

measure and is referred to as the λs. The higher the λs, the greater the potential genetic contribution to the phenotype. Although one does not know in advance how complicated the genetics may be, this provides a general benchmark for how difficult the genetic answers may be to find. In addition, when the rate of concordance decreases by more than 3 for each degree of relationship, the genetics of the phenotype are predicted to be complex (4).
GENOTYPING
Genomic DNA is usually used at two levels: first, to establish genetic linkage by using genotypic markers, and second, to identify the responsible gene by testing for association. The latter finds the genotypes responsible for the observed phenotype from the evaluation of candidate genes. Candidate gene polymorphisms are often directly evaluated, and surprisingly, often positive results are found without any evidence for linkage having been previously collected. Identifying the genes and understanding their biology is the great challenge. If association studies will be eventually performed, directly evaluating promising candidate genes has the potential to greatly accelerate progress.
There are two types of genetic markers now ordinarily used for genotyping: microsatellites and SNPs (Fig. 28.2). Microsatellites are also called variable number tandem repeats because a sequence of two, three, or four nucleotides is repeated a variable number of times. The polymorphism or allele for that marker is based on the number of times the short sequence is repeated. There are estimated to be more than 50,000 microsatellites in the human genome.
FIG. 28.2. Microsatellites and single nucleotide polymorphisms (SNPs). A: Polymorphic microsatellite compares a pair of alleles found on two individual’s chromosomes, one with three repeats of the trinucleotide sequence and the other with four repeats. B: The single nucleotide polymorphism difference found at nucleotide position 394 of the FcγRIIA nucleotide sequence on chromosome 1q23.1. This polymorphism leads to a change within the amino acid at position 131 [histidine is changed to arginine (H131R)], which changes the affinity of IgG2 for this Fc receptor.
Polymorphisms are usually identified after expansion using the polymerase chain reaction (Fig. 28.3). This requires the preparation of primers complementary to the DNA sequence flanking the variably repeated element and geometric expansion of the DNA through repetitious primer extension and DNA denaturation. The procedure has been
P.612

semiautomated, making possible the collection of thousands of genotypes a day (Fig. 28.4).
FIG. 28.3. The polymerase chain reaction geometrically expands specific DNA. Complementary oligonucleotide DNA primers locate the 3′ end of the DNA to be expanded. The intervening DNA is replicated. With each elevation of temperature (to 94°C) the DNA is denatured, allowing the primers to again locate the 3′ end of the target DNA for another round of DNA synthesis when the temperature is reduced. Repeating this cycle progressively produces larger and larger quantities of the target DNA sequence.
FIG. 28.4. Genotyping a pedigree multiplex for systemic lupus erythematosus. Top: A three-generation African-American pedigree is presented. A horizontal line connects spouses. The progeny of a mating divide horizontal lines with vertical lines. Bottom: The pedigree drawing has been redrawn so that the genotyping results at D12S1042 are given in the image from the polyacrylamide gel electrophoresis presented below. The alleles vary by the number of three-base repeats present (135 to 153 indicates the number of nucleotide bases in the DNA fragment). The larger alleles move relatively more slowly in the electric field and are therefore higher in the gel.
Single nucleotide polymorphisms are expected to supplant the microsatellites because there are so many more of them and their rate of mutation is so much lower. There is a relatively common SNP every 500 or fewer bases of the human genome, leading to the expectation that there would be on the order of 7,000,000 SNPs with which to perform human genetic analyses. In addition, DNA chip technologies evaluate tens or hundreds of thousands of these polymorphisms simultaneously and rapidly. The SNP Consortium has, to date, described 1.8 million SNPs. Many laboratories around the world joined this effort to identify SNPs and to provide them to a public database. This has been an extraordinary example of the power of cooperating through the Internet. The reader is invited to go to snp.cshl.org/ and look up their favorite gene, locus, or chromosome location.
A current controversy focuses on how useful SNPs will prove to be. The mutation rate is not linear along the chromosome. Therefore, situations are expected in which the more recent and subpopulation variable SNPs would be the most useful for this approach, but they are generally less prevalent than the evolutionarily older SNPs. A recent study preliminarily reported to have failed to locate the well-known genetic defect for sickle cell anemia using this approach (5). Nevertheless, there is much hope that well-characterized SNPs, along with a detailed and accurate map of their location, will help solve a number of presently intractable problems. A central role for the SNPs in locating genetic effects is anticipated, but some work remains in order to understand their capabilities and to build the resources and databases needed to best exploit this promising approach.
Discovery of genetic effects by directly testing genes is intrinsically inefficient. There are on the order of 25,000 human genes, and an unknown number of polymorphisms. Any present effort to find disease genes in this way is bound to be incomplete with existing technology. On the other hand, the technical capacity to evaluate more than 100,000 SNPs routinely from a single individual should be available as this text is going to press. Such capability will revolutionize genetic analysis.
CROSSOVERS
Of course, the goal is to find the disease-associated genes and then to explain their role in pathophysiology of the phenotype.
P.613

Finding the genes is made possible by the 37 crossovers (usually one or more per somatic chromosome) that occur every meiosis (the process by which the haploid DNA is generated for the egg and sperm) (Fig. 28.5). The haploid chromosomal DNA of the gamete then becomes a mosaic construction of the diploid chromosomal DNA from the parent. In each generation, the DNA is shuffled a little, since in the entire human genome there are about 37 crossovers, by definition one on average every 100 centiMorgans. Slowly, over the generations, each DNA base pair is randomized with its neighbors and provides the measure of genetic distance used in linkage and association studies to identify regions containing genes that cause observed effects.
FIG. 28.5. Crossover. One chromosome is depicted for each of three founders (I1, I2, and II2) in a three-generation pedigree. The chromosomes of the progeny become mosaic constructions of the founders, as indicated.
LINKAGE AND ASSOCIATION
A phenotype being linked to a genetic marker usually means that a statistically-unexpected relationship is found between either the presence or absence of the phenotype in the family and the sharing of alleles at the marker locus within families. In contrast, most tests of allelic association evaluate relationships between the phenotype in members of a population (i.e., between families) and the particular alleles at a given marker. Thinking of this from another perspective, the tests for linkage usually rely on crossovers within pedigrees and are tests for involvement of a locus; they do not identify particular alleles. On the other hand, allelic association does identify a particular allele and also relies on crossovers within a population (i.e., between pedigrees).
The more crossovers there are between affecteds, the smaller the average piece of DNA that can be statistically related to genetic risk (Figs. 28.1 and 28.5). Most strong linkage effects are detected over 20 to 30 megabases of DNA, whereas tests of association usually operate over 0.1 megabases or less. There are some exceptions to this, such as the HLA region, in which genetic association effects are detected over 3.5 megabases. The region of the genome over which association can be detected is said to be in linkage disequilibrium.
Some populations have a much lower number of crossovers between affecteds. For example, in selected isolated populations, an association may be detected over a greater genomic distance. Also, when populations with different haplotype frequencies have been recently admixed, then association may be found over relatively large genetic distances. Such effects are predicted to be present in African-Americans.
Consequently, linkage is ordinarily used to identify genetic effects in a region, whereas association is usually used to narrow such a region. Indeed, association is sometimes used to find the disease gene itself. Associations of HLA-B27 with ankylosing spondylitis, HLA-DR4 with RA, and FcγRIIA alleles (histidine vs. arginine at amino acid 131) with lupus nephritis in African-Americans are but a few of the hundreds of genetic associations that have been found in the rheumatic diseases (6).
ASCERTAINMENT
Appropriate study designs greatly improve the prospect of a successful outcome of a genetic project, as they do in all science. What can be discovered is usually dictated by the study design, and study designs begin with the enrollment of the first subject. A critical question is how to collect the pedigrees and pedigree materials. The data collection, methods of analysis, and possible results are dramatically different when cohorts of cases and controls are collected, as opposed to, for example, multiplex, multigenerational pedigrees.
FOUNDERS AND ISOLATED POPULATIONS
Case-control designs are appropriate for some situations, particularly isolated populations and within populations with a founder effect for the phenotype of interest. Isolated populations offer special opportunities. The genetic variation in such groups is different from that in the general population. Usually, the number of possible genetic explanations is fewer, and hence, often easier to find. By evaluating three benign cholestatic jaundice patients (a known autosomal-recessive disorder) and their parents at 256 markers, Houwen and colleagues established a linkage on chromosome 18 (7). This is also the strategy used in the comprehensive genetic evaluation of the Icelandic population, which has recently become a commercial enterprise and is the focus of much discussion and controversy (8).
Tan et al. (10) and Arnett et al. (9) also used the isolated population strategy to study scleroderma in the Choctaw Indians. They showed two separate genomic regions, one containing the fibrillin gene and the other containing the HLA genes, to be genetic risk factors in these patients. The hope, of course, is that these same genes will be important in scleroderma patients from the general population. But,
P.614

even if they are not, the genes responsible for the observed effects are still likely to be important in helping understand the pathogenesis of the disorder. Opportunities presented by isolated populations should not be ignored.
A founder effect, as mentioned previously, is the presence of a phenotype due to a common ancestral mutation. The founder is the first to pass the genotype to his or her offspring. The founder need not have the phenotype, only the mutation, and then only in his or her germline. An increased risk for the phenotype can be traced back through the generations (if the records existed) to the founder. How much of the neighboring genome is in linkage disequilibrium with the genotype conferring disease risk depends on how many generations ago the founder introduced the genotype. The more distant in the past, the smaller the region over which linkage disequilibrium will exist (Fig. 28.1). When founder effects are present, the gene can often be discovered by association studies, as was done for hemochromatosis (11).
TRANSMISSION DISEQUILIBRIUM
A popular technique for assessing association is the transmission disequilibrium test (TDT) (12). In its simplest form, the affected and either one of the parents is considered a unit, a case, and a matched control unit (Fig. 28.6), but only when that parent is heterozygous and the inheritance pattern is unambiguous. The test evaluates the distortion in the ratio of allele transmission to allele nontransmission from parent to progeny. The TDT is not susceptible to artifacts arising from population stratification because the control (which here is the nontransmitted allele) comes from the same parent as the case (the transmitted allele) and, hence, is taken from the same population.
FIG. 28.6. The transmission disequilibrium test. Four pedigrees are shown in which the “D” and “d” alleles are shown for each pedigree member. In the first pedigree the parent-progeny unit from the father shows that D is transmitted and d is not transmitted. These are tallied, much like they are in the classic case with matched control design. Here X2 ∼ [(b - c)2/(b + c)] = 5, which with one degree of freedom leads to p < 0.05. The odds ratio = b/c. Note the homozygous parents (either DD or dd) do not contribute. df, degree of freedom.
AFFECTED SIBLING PAIRS
In situations where there is strong evidence for a genetic effect, but the mode of inheritance is not known, affected sibling pairs have been collected. Although there are many variations on the strategy for ascertaining affected sibling pairs, the most straightforward is the collection of the affected sibling pairs and their parents. At markers where both parents are heterozygous for different alleles, the test for linkage is based on the expected sharing of no alleles (0) and of two alleles (2) at a marker locus (Fig. 28.7). Given that a marker locus is fully linked to the disease gene, a concordant affected sibling pair would be expected to share no alleles 0% of the time and 2 alleles 100% of the time. The extent to which the sibling pairs differ from these expected values measures the probability of linkage at the locus.
FIG. 28.7. Affected sibling pairs. The most informative situation is when both parents are differently heterozygous, which is the case for each pedigree presented. Here, all affected pairs share two alleles, providing some support for linkage (p < 0.05). To determine if there is genetic association, the allele frequencies of the cases are usually compared with controls. IBD, identical by descent; df, degree of freedom.
Usually hundreds, if not thousands, of sibling pairs are needed to establish linkage in situations where the genetics are complex and the genetic effects at any given locus are thought to be relatively small. An often unstated assumption in the application of this method is that no other process is operating to select for or against particular alleles. For example, if an allele at the marker locus is recessive lethal, then no affected would be found who is homozygous for this allele at this locus. The distribution of allele sharing
P.615

would be distorted from the expected for a reason other than linkage.
There are, of course, many variations of the affected sibling pair method presented above. One variation is the inclusion of unaffected siblings and other relative pairs. Another is using a quantitative, rather than qualitative, trait. Yet another is including only the siblings with the most extreme phenotype on a scale of measurement. A recent variation to this method is to evaluate genetic interactions as well as other possible covariates. All of the above-mentioned extensions increase the power to detect linkage and are, therefore, of great interest to geneticists. Several other mathematical accommodations and advances have been made, and many useful software programs are available but are beyond the scope of this discussion. However, the reader should be aware that many good choices are available once affected sibling pairs and their relatives have been ascertained.
EXTENDED MULTIPLEX PEDIGREES
Ascertainment of large pedigrees with several affecteds is very powerful. Clearly, the larger the pedigree is, the greater the potential to find several affecteds. This is the classic ascertainment strategy for simple or single-gene disorders when the segregation analysis supports an autosomal-dominant, autosomal-recessive, or X-linked mode of inheritance.
The classic method of analyzing these data is based on the work of Newton Morton, developed almost a half century ago (13). The likelihood ratio of the odds for linkage divided by the odds of no linkage is calculated by estimating the odds for linkage as a function of the recombination fraction (Fig. 28.8). The recombination fraction is a measure of distance from the marker in a single generation. The maximum evidence for linkage over the recombination interval of 0 to 0.5 is usually reported as the logarithm of the odds (LOD) score or the log of the likelihood ratio. For example, LOD scores of greater than 4 support linkage of specific regions of the genome in two hereditary forms of pseudogout (14,15). As the distance from the marker increases, the likelihood that there has been a crossover increases. A recombination fraction of 0.5 by this method is the equivalent of no linkage. Any recombination fraction less than this and convincingly present is consistent with linkage. Of course, small recombination fractions
P.616

produce the strongest evidence for linkage. A 1% recombination fraction is equivalent to 1 centiMorgan and is approximately 1 megabase of DNA.
FIG. 28.8. Maximum likelihood ratio test for linkage. Consider a two-generation pedigree (A) when the phenotype is known to segregate as an autosomal dominant disorder. B: The LOD [log10 (likelihood of linkage/likelihood of no linkage)] as a function of the recombination fraction (θ). C: The plot of the LOD score as a function of θ. The maximum LOD score (0.8) is at θ = 0.125, some genetic distance from the marker.
As might be evident from this discussion, for the classic method of linkage by the likelihood ratio test, any error in pedigree relationship or genotyping will appear as an error in the estimate of the recombination fraction. The procedure maximzes the likelihood across the range of penetrance values possible and across genetic heterogeneity. This allows for refinement of the inheritance mechanism and increased power to detect linkage. Penetrance is the probability of having the phenotype, given the genotype. For example, the penetrance for a strict autosomal-dominant disorder is 100% for an individual heterozygous or homozygous for the responsible allele. Genetic heterogeneity is a measure of the proportion of the pedigrees linked at the locus of interest. Finally, with consideration of the genotypic data at neighboring loci simultaneously, LOD scores can be calculated at each interesting position using a specified model of inheritance. These multipoint LOD scores are the current standard for this kind of genetic analysis.
In recent years, other algorithms have been developed for extended multiplex pedigrees. Of these, the inheritance distribution pattern approach, based on the work of Leonid Kruglyak and colleagues, is widely used (16). This and other relative pair methods have been extended from the affected sibling pair approach to accommodate extended pedigrees.
GENOME SCAN FOR LINKAGE
Genome-wide scans are being conducted for most of the major disease problems thought to have a genetic component. This means that the entire collection of pedigrees, whether sibling pairs or multiplex families, is genotyped for markers at roughly equal spacing throughout the human genome, usually 120 to 400 microsatellites. This approach has led to evidence for additional genes (beyond HLA-B27) for the development of ankylosing spondylitis and is now being applied to many rheumatic disease problems with a genetic component (17).
SIGNIFICANT LOD SCORES AND p VALUES
The issue of deciding what level of significance is meaningful is not simple. Indeed, entire scientific disciplines have been organized around this issue. The critical decision is made when a linkage or association is sufficiently significant to imply or even identify a gene. Thankfully, there are some general rules to follow. First, one should know whether an important finding has been confirmed, since this greatly increases the confidence that a finding is correct. Confirmation by repetition is also a basic tenet of the scientific method. However, failure of other investigators to confirm a finding does not necessarily mean that a finding is not correct. This depends greatly on the context of both the original study and its replicate. In genetic situations where segregation analysis has established a simple genetic mechanism, failure to confirm a linkage usually means that there is a second gene in the pedigree or population being evaluated. In genetically complex diseases, pedigree collections are unintentionally enriched for some linkages and depleted for others, due to random variation in the population. Different genes will then be detected as a result of selecting different samples from the different populations. Thus, positive confirmation, rather than negative, is the most convincing, although there are many examples of initially convincing linkages that are now thought to have been false-positive linkage signals.
Most researchers express the significance of their results as either a probability (p value) or a likelihood (LOD score). The p value is the probability that the observation will occur by chance. The LOD score, as explained above, is the log10 of the likelihood that the marker is linked to a disease gene divided by the likelihood that the marker is not linked. LOD scores of greater than about 3.3 are estimated to be found by chance only once in every 20 genome scans using marker data that are not linked to a disease gene (18). Therefore, an LOD of about 3.3 is widely accepted as a threshold for significance in a genome scan, especially in complex genetics diseases. At times the expression of a p value as an LOD score is desired. This can be done by assuming a normal distribution, then determining the Z corresponding to the respective p value, and approximating χ2 = ∼Z2 and LOD = ∼(χ2 / 4.6). Using this method, the LOD of 3.3 is approximately equivalent to p < 0.00002.
CANDIDATE GENES
There are three important sources for candidate genes, or genes that have the potential to explain the known biology of the disease in question. First, work with the biology of the phenotype may raise the suspicion that a particular gene is related. This can be the result of brilliant thinking, as in the example of apolipoprotein E and Alzheimer disease (19). It can also be the result of a complete accident, as in the relationship of HLA-B27 with ankylosing spondylitis. The latter association was found because ankylosing spondylitis patients were selected as controls for an inquiry into an association of gout with HLA (20). Second, linkage results may restrict the genes being considered to a relatively small region of the genome. Genes in this region are considered as candidates in an order of prior probability chosen by the investigator. Third, linkages or genes found in other species, usually from the mouse, may specifically suggest genomic regions or specific genes in humans.
This discussion would not be complete without addressing this third source in more detail. Mouse genetics in particular has played a substantial role in the discovery of genes important in human disease. Our capacity to manipulate
P.617

the genetic composition of the mouse far exceeds what can be done in humans and is an approach proven to greatly accelerate progress.
Murine knockouts are animals with a dysfunctional gene at a particular locus. Transgenic mice have a gene or set of genes installed from another DNA source (e.g., human gene). Breeding programs manipulate genetic composition. Although these experiments often fail to produce the expected result, they offer profoundly important insights into the biology of the system being explored.
A few simple examples will serve to illustrate what is now possible. First, a transgenic mouse is available containing much of the human DNA for the variable regions of the immunoglobulin locus (21). These animals are presumed to have an enhanced capacity to imitate the human immune response. Second, a human transgenic rat has begun to suggest a role for the HLA-B27 allele in the inflammatory response (22). Third, breeding experiments have shown that the MRL lpr/lpr mouse defect is in the fas gene, which is responsible for lupus nephritis in this mouse. A large body of literature on apoptosis and autoimmunity has been spawned from this observation (23).
Ward Wakeland and his colleagues have been isolating the genes disposing to lupus in the NZM mouse [a fixed interstrain cross from the classic (NZB × NZW)F1 model of lupus]. They have evidence for at least eight genes (24,25). Some of these genetic effects are very close to one another, whereas others are on separate chromosomes. Even genes operating to suppress disease are suggested. The mouse offers the prospect of identifying these genes quickly, but gene identification even in the mouse can be problematic. The murine SLE locus sle1 has been found to have at least four distinct components, located very close to one another on mouse chromosome 1 (26). Distinguishing the causative gene from among these haplotypes composed of multiple polymorphisms in neighboring, ancestrally-duplicated genes (i.e., sle1) presents a daunting level of complexity. The situation is much more complicated in humans, which should give pause to anyone who thinks that genetic solutions will rapidly follow the advancing technology.
Mouse genetics offers another advantage. Mouse genes and human genes are syntenic, meaning that the gene order on the chromosomes in the two species is similar. In humans and mice, they are astonishingly similar (over 95% identical). When a linkage is found in the mouse, it is likely to be in the same relative position in humans. This is a strategy for exploring candidate linkages that has been successfully applied. Once a gene is known in the mouse, the homologue can be directly explored in humans.
THE STATUS QUO
The methods used throughout this chapter have been used to identify rheumatic disease genes. With the phenotype as the bait, the new technologies and resources are aggressively being applied to go fishing for the linkages and genes that will explain etiology.
Familial Mediterranean Fever
Familial Mediterranean fever (FMF) is an autosomal-recessive autoimmune disease characterized by recurrent episodes of fever, serositis, arthritis, and skin rash. Pras et al. mapped the gene causing FMF to the short arm of chromosome 16 (27). Subsequently, the gene (MEFV) was cloned by two independent groups in 1997 (28,29). MEFV, predominantly expressed on myeloid cells, encodes for a 781–amino acid protein, pyrin or marenostrin, and appears to be up-regulated during myeloid differentiation (30). The MEFV gene has 10 exons. At least 29 mutations have been reported, mostly involving exons 10 and 2. The most common mutations are M694V, V726A, M694I, and M680I on exon 10 and E148Q on exon 2 (31). About 22% to 67% of the patients evaluated have the M694A mutation and 7% to 35% have V726A. M694A means that the DNA code has changed from the codon for methionone to the codon for alanine at amino acid position 694 of the pyrin protein molecule. Other mutations have been reported on exons 1, 3, 5, and 9. The severity of the disease and the risk for amyloidosis appear to vary among these different mutations. The homozygous state for the M964V mutation is indeed associated with a more severe disease and higher risk for amyloidosis (32,33). More recently, it has been shown that male sex, coupled with articular manifestations, is associated with a fourfold increase in the risk for amyloidosis among patients who are homozygous for the M964V mutation (34).
Pyrin consists of four functional subunits, which in order are the PYRIN domain (N-terminal), the B-box zinc-finger domain, a coiled-coil domain, and the B30.2 domain (C-terminal). The PYRIN domain is shared by a family of related proteins involved in apoptosis and inflammation (35). Interaction among these various PYRIN domain–containing proteins appears to be mediated via the PYRIN domain. In most FMF patients, the relevant mutations involve the C-terminal part of the molecule; thus, a full functional PYRIN domain is usually retained. In a recent study, macrophages from mice expressing a truncated pyrin molecule, yet retaining the PYRIN domain, produce elevated levels of interleukin (IL)-1β (36). In addition, macrophages derived from pyrin-truncated mice demonstrate impaired apoptosis (36).
Rheumatoid Arthritis
The familial clustering of RA (λs = 5), and a higher concordance in monozygotic twins compared to dizygotic twins support the role of genetic factors in the development of RA. The most consistent genetic association reported is
P.618

with the HLA class II locus. Indeed, association with several HLA-DRB1 alleles has been confirmed, including the association with HLA-DRB1*0401 and HLA-DRB1*0404 (37,38). Other confirmed HLA associations include HLA-DRB1 *0101, *0102, *0405, *0408, *1001, and *1402. The association between HLA-DRB1 alleles and disease severity and the presence of extraarticular manifestations has also been reported. In one study, all patients with nodular disease carried the genotype HLA-DRB1*04/04 (39). HLA-DRB1*0401 is associated with a higher frequency of bronchiectasis in RA patients (40).
Genome-wide scans performed in affected sibling pair families have confirmed a linkage on chromosome 6p21–23 (41,42,43). Other reported linkages include 1p36, Xq27 (44), 6q22(42), and 4q22–24 (43). Additional non-HLA genetic associations have also been reported. A linkage to the corticotropin-releasing hormone gene at chromosome 8q13 has been reported and replicated in a second group of RA sibling pair families, but the effect was not sufficient to be considered an established linkage (45). Similarly, linkage to chromosome 17q22 has been confirmed in affected sibling pair analysis (46). In addition, numerous other non-HLA associations have been reported and replicated, including insulin-dependent diabetes mellitus susceptibility locus IDDM5 (47,48) and tumor necrosis factor (TNF) polymorphisms (49,50,51,52).
Osteoarthritis
A genetic component is clearly involved in susceptibility to developing osteoarthritis (OA), as suggested by twin studies, sibling risk, and familial aggregation. The disease has a λs = 2.32 in a study of severe OA requiring joint replacement (53). Some suspect that the role of genetics in the development of OA is more important in women than in men (54). In addition, the locus-specific genetic susceptibility varies by articular location of the OA. Several linkages have been suggested by sibling pair analysis and by the rare OA pedigrees, including effects on chromosomes 2q, 4q, 6, 7p, 11q, 16p, and Xcen. Replicated linkage effects include chromosomes 2q and 4q (55). The effect on 4q12–21.2 produced a LOD = 3.9 in female sibling pairs with hip disease (56). In a fine mapping study of the chromosome 6 effect, the susceptibility locus was mapped between 70.5 and 81.9 centiMorgans from the 6p telomere in sibling pairs concordant for total hip replacement (57). Interestingly, stratification analysis suggested that this effect is completely accounted for by female total hip replacement families (maximum LOD = 4.6 at the marker D6S1573) (57). A recent genome-wide linkage analysis for hand OA revealed evidence for convincing linkages on chromosome 4q, 3p, and 2p (58). The maximum LOD score on chromosome 2 coincides with a gene encoding a noncollagenous cartilage extracellular matrix protein, matrilin-3. Further analysis revealed a novel missense mutation in the matrilin-3 encoding gene, MATN3, that was responsible for the observed linkage effect (58).
Several genetic association studies reported association between OA and polymorphisms in the type II collagen gene COL2A1, which constitutes a major protein in articular cartilage (59,60,61). The COL2A1 gene is located on chromosome 12q12–13.1, very close to the vitamin D receptor (VDR) gene. The association between OA and the VDR gene has been repeatedly reported as well (62,63,64,65). In one study, the COL2A1 gene was associated with joint narrowing and the VDR gene with osteophyte formation, both in patients with knee OA (66).
Osteoporosis
Osteoporosis is a major health problem that results in significant morbidity and mortality. Several environmental factors are clearly involved in the development of osteoporosis; however, the role of genetics is also central. This is supported by twin studies and familial aggregation. The disease is characterized by loss of bone mass, as identified by bone mineral density (BMD) studies. Quantitative trait loci (QTL) are chromosomal regions that contain genes regulating quantitative traits. A QTL approach has been used to identify genes or effects that regulate bone mineral density. Several genome-wide linkage studies have been performed to identify loci that are linked to bone mineral density. Evidence for established linkage, as defined by an LOD ≥ 3.3, has been reported on chromosomes 1p36 (67,68), 1q21 (69), and 11q12–13 (70). Other loci with evidence of suggested linkage include 2p23 (67), 4q33 (67), 2p21 (71), 5q33–35 (69), 6p11–12 (69), 7p22 (72), 12q24 (72), 13q33–34 (72), and 10q26 (72). Among the numerous linkages reported, only 1p36, 4q, and 13q have been replicated (67,68,72).
The effect on 11q12–13 is linked to rare bone diseases, including osteoporosis-pseudoglioma and high bone mass syndrome. Finer mapping and sequencing of this locus identified the gene responsible for this effect to be low-density lipoprotein receptor–related protein 5 (LRP5). Activating mutations in this gene are responsible for high bone mass syndrome, while on the other hand, inactivating mutations cause osteoporosis-pseudoglioma (73). A number of candidate genes have been studied in relation to BMD. Associations have been reported for VDR (74), collagen type I gene (COL1A1) (75,76), estrogen receptor gene (ESR1) (77,78), and parathyroid hormone receptor gene (PTHR1) (79).
Ankylosing Spondylitis
Ankylosing spondylitis is a potentially disabling disease with a major genetic component, suggested initially by familial aggregation. This was confirmed by the discovery of the strong association between the disease and HLA-B27, now confirmed well over 100 times. However, twin studies
P.619

indicate that other genetic components are also involved, since the concordance rate in HLA-B27-positive dizygotic twins is lower than that in monozygotic twins (23% vs. 63%) (80).
Other HLA genes reportedly involved in the susceptibility to ankylosing spondylitis are HLA-B60 (81) and HLA-DRB1 (82). Non-HLA genes identified by the candidate gene approach reveal a potential association with TNF-α (83,84,85), IL-1 receptor antagonist (IL-1ra) (86,87), and cytochrome P450 2D6 (CYP2D6) gene polymorphisms (88,89). Genome-wide linkage studies have shown a significant effect on chromosome 16q (LOD = 4.7) (90). This effect on 16q exceeds the threshold for established linkage and has been confirmed (90,91). Other chromosomal effects suggested by genome-wide scans include loci on chromosomes 1p, 2q, 6p, 9q, 10q, and 19q (90).
Psoriatic Arthritis
About 15% to 30% of patients with psoriasis will develop psoriatic arthritis. Both psoriasis and, hence, psoriatic arthritis appear to have excessive paternal transmission (92). This means that progeny are more likely to develop psoriasis or psoriatic arthritis if the father has psoriasis or psoriatic arthritis. A strong association was repeatedly reported between psoriatic arthritis and HLA-Cw*0602 (93). This HLA allele is also a recognized susceptibility locus for psoriasis and is, indeed, associated with earlier onset of the disease (94). MICA-A9 triplet repeat polymorphism, corresponding to the MICA-002 allele, is associated with susceptibility to psoriatic arthritis, independent of the genetic effect of the nearby HLA-Cw*0602 (95). The reported association with TNF-α polymorphisms has been inconsistent and, perhaps, explained by linkage disequilibrium with HLA-Cw*0602.
A recent genome-wide linkage study identified a susceptibility locus for psoriatic arthritis on chromosome 16q. When conditioned on paternal transmission, this effect gave an LOD score = 4.19, which exceeds the threshold for established linkage (96). In addition, an effect on chromosome 17q has been separately reported for both psoriasis and psoriatic arthritis (97,98).
Calcium Pyrophosphate Deposition Disease
The deposition of calcium-containing crystals within the articular cartilage induces an arthritis known as pseudogout. Calcium pyrophosphate crystal deposition increases with age and is associated with elevated pyrophosphate in the joints (99). Although mostly sporadic, rare forms of familial pseudogout with autosomal-dominant inheritance have been observed. Those have been linked to susceptibility loci at chromosome 8q (CCAL1) (100) and 5p (CCAL2) (101,102). It was recently described that families with the CCAL2 effect have mutations in the human homologue of the mouse progressive ankylosis gene (ANKH) (103,104), which has previously been shown to be involved in pyrophosphate regulation and joint calcification.
Paget’s Disease
Islands of increased bone turnover, resulting in disorganized bone growth, is characteristic of Paget’s disease. This common disease often segregates in an autosomal-dominant manner with incomplete penetrance, although it has substantial genetic heterogeneity (105). Linkage studies identified a disease susceptibility gene on chromosome 6p21.3 (PDB1) (106) and 18q21–22 (PDB2) (107,108). The effect on chromosome 18q21–22 has been previously linked to a rare bone dysplastic disease, known as familial expansile osteolysis (109). However, unlike the case in familial expansile osteolysis, mutations in the receptor activator of the nuclear factor κB gene (RANK) mapped to the PDB2 region (110) are not observed in classic Paget’s disease (111). Interestingly, evidence of an osteosarcoma tumor suppressor gene (on chromosome 18q region) linked to Paget’s disease has been reported (112). This might provide an explanation for the higher frequency of osteosarcoma in Paget’s disease patients. Additional susceptibility loci for Paget’s disease have been reported on chromosome 5q35-qter (PDB3) and 5q31 (PDB4) (113,114,115).
Systemic Lupus Erythematosus
SLE is a complex disease of controversial etiology. Both environmental and genetic factors contribute to its pathogenesis. Evidence for genetic contribution comes from twin studies and familial aggregation (λs = 10–20) (116,117,118,119). The known associated genetic polymorphisms and the linkages reported using genome-wide scans of multiplex pedigrees and affected sibling pair families further support a genetic component.
Studies of histocompatibility molecules have been performed in various SLE populations (120,121,122,123,124,125,126,127,128,129,130,131,132). The DR2 and DR3 alleles at HLA-DR have the most consistent association with SLE. As is true for all human histocompatibility associations, the pathophysiology and mechanistic details that explain this association are not known.
SLE is also associated with polymorphisms of at least FcγRIIA (133,134) and FcγRIIIA (135,136), and perhaps other Fc receptor genes. In both cases, the allele that binds immunoglobulin G (IgG) subtypes with lower affinity is associated with increased risk for SLE. These alleles are thought to have decreased capacity to clear immune complexes composed of the specific low-affinity IgG subtypes. These genes are neighbors on the human genome at chromosome 1q22–24, in a region also known to have genetic linkage with SLE. These two Fc receptors are separated by only 20 kb of genomic DNA and are in linkage disequilibrium.
P.620

Recently, alleles of PDCD-1, a T-lymphocyte transcription factor, have been implicated as being responsible for the linkage at 2q37 (137), first found in Nordic pedigrees multiplex for SLE. Further confirmation of this result is widely anticipated.
Several other genes are thought to increase susceptibility to the development of SLE without producing consistently convincing evidence of genetic association. These include polymorphisms involving IL-10, FAS (138,139) and FASL (140), the mannose-binding lectin genes (141,142,143), Bcl-2 (144), CTLA4 (145), the T-cell receptors (146), prolactin (147), TNF-α (148,149), and TNF receptors (150).
Genome-wide linkage studies in pedigrees multiplex for SLE have revealed some potentially important genetic polymorphisms. At least five linkages have been established and confirmed, including those at 1q23, 2q37, 4p16, 6p21, and 16q13, along with many others that have not yet produced such convincing results (151). Of these linkages, convincing associations potentially explaining them are found at FcγRIIIA (1q23) (152), PDCD-1 (2q37) (137), and HLA-DR (6q21) (153). Many other genetic effects (>15) have been established and await confirmation (151). Progress in the human genome project and improved methods of analysis are anticipated to provide much more rapid progress toward a more complete genetic understanding than has been possible in the past.
Sjögren’s Syndrome
Primary Sjögren’s syndrome is a complicated polygenic disorder with many genes predicted to be interacting with environmental factors. Several HLA class II genes, in particular HLA-DRB1, DRB3, DQA1, and DQB1, have been associated with Sjögren’s syndrome and the production of anti-Ro and anti-La autoantibodies (154). Using DNA methodologies, confirmed alleles include HLA-DRB1*0301, -DRB1*1501, -DQA1*0103, -DQA1*0501, -DQB1*0201, and -DQB1*0601 (Fig. 28.9).
FIG. 28.9. The human leukocyte antigen (HLA) region on the short arm of chromosome 6 showing the HLA alleles associated with Sjögren’s syndrome. Confirmed associations are in italics. Modified with permision from Sawalha et al. (154).
Numerous non-HLA genetic associations have been described, including polymorphisms of IL-10 (155), IL-1Ra (156), IL-6 (157), Ro52 (158), TAP2 (159), GSTM1 (160), MBL (161,162), and FAS genes (163). However, confirmation is generally lacking with the non-HLA associations that have been reported for primary Sjögren’s syndrome.
Hemochromatosis
Hemochromatosis is a common inherited disorder with an autosomal-recessive pattern of inheritance. The disease is characterized by iron overload and manifested by liver cirrhosis, symmetric OA, skin discoloration, cardiomyopathy, hypogonadism, diabetes, weakness, and lethargy. An environmental component is thought to have a major role in hemochromatosis because of alcohol abuse.
Epidemiologic data for hemochromatosis suggest a founder effect. There is a relatively high prevalence of affecteds among Northern Europeans, and the estimated gene frequency has been found to be over 10%, with 0.5% of the population being homozygous. For more than two decades, an association of HLA-A3 within Northern Europeans with hemochromatosis has been known (164). This means that the gene responsible for hemochromatosis tends to be found in haplotypes containing HLA-A3, or alternatively, that HLA-A3 and the hemochromatosis gene are in linkage disequilibrium in people with Northern European ancestors. Consequently, a direct approach was taken seeking to
P.621

identify an increase in association between disease and genetic markers while moving farther away from HLA-A3.
This effort was a spectacular success and has identified the hereditary hemochromatosis gene, now designated HFE, which appears to be mutated in most hemochromatosis patients (165). The postulated founder effect is supported by a single amino acid coding change in the HFE gene, resulting in tyrosine being substituted for cysteine at amino acid position 282 (C282Y). The vast majority of hemochromatosis patients (>90%) are associated with being homozygous for C282Y (166). In addition, other mutations of HFE have been described, including H63D and S65C (167). The S65C mutation is associated with mild iron overload in individuals who are also heterozygous for the C282Y or the H63D mutations (168,169). Contrary to the initial impression, only a minority of the C282Y homozygotes develop clinical features of hemochromatosis (170). Indeed, in one study the penetrance of the C282Y homozygous genotype has been estimated to be 1% (171), providing more evidence consistent with the involvement of environmental or other genetic factors in the susceptibility to hemochromatosis and being more than 10-fold lower than estimates of the penetrance that had been made only a few years ago, before the knowledge of the specific mutation involved had been applied in population-based studies. New evidence for the involvement of other genes, including various iron absorption and metabolism genes, is therefore not unexpected (170).
Defining the genotype that confers risk provides a new perspective from which to evaluate related phenotypes. For example, nine pedigrees with juvenile hemochromatosis have been studied by a genome scan. A maximum LOD score of 5.16 was found on chromosome 1q (at D1S2344) (172). The affecteds in these pedigrees, having a younger age of onset (second and third decades of life) as well as showing no linkage to the region on chromosome 6p containing HFE, establish juvenile hemochromatosis (type 2) as a distinct phenotype and, therefore, presumably a distinct pathogenesis. Type 3 hemochromatosis is caused by a mutation in a transferrin receptor gene (TFR2) and is an autosomal-recessive disease (173,174,175). Types 4 and 5 are both autosomal dominant and are caused by mutations in the ferroprotein 1 gene, which encodes for an intestinal iron transport molecule, and the H-ferritin gene, which encodes for the H subunit of the ferritin molecule (176,177,178,179,180).
Summary
This section is entitled “The Status Quo,” which is misleading because of rapid progress in the genetic characterization of disease. New technologies promise to accelerate further the pace of discovery; indeed, the new genetic approaches discussed in the early part of the chapter promise to revolutionize our conception of how these diseases occur. The challenge is to understand more than the gene, to identify it, and then to place the gene in its pathophysiologic context so that the mechanism of disease is fully explained and understood. In some situations this has occurred quickly (e.g., familial Mediterranean fever or hemochromatosis). In others, many decades have passed finding us still waiting for the critical insight that explains the essential role of the allele in the disease process (e.g., ankylosing spondylitis and HLA-B27). To be more than curiosities or only diagnostic tests, the genetic advances must be accompanied by a deeper understanding of the biology of these illnesses.
THE FUTURE
As in other disciplines, situations in which the genetic model is known and the pedigree material is sufficient are likely to yield genes that contribute to the observed phenotype. Knowing the genetic solution is perhaps the easier part of the answer. We hope that explaining the genetics will lead to a solution to the more difficult problem of understanding the biology of the disease phenotypes.
In the more common disorders, where the genetic mechanisms are not known, genetic progress is more likely to be episodic and difficult. We are likely to repeat the experience of psychiatric and affective disorders where the difficulty of confirming alleged linkages threatens the entire effort of identifying genes.
For type 1 diabetes, approximately 2,000 pedigrees are available, and consequently, a variety of effects have been identified (181). These results would have been convincing, save for the failed effort to confirm their presence in a large American consortium (182). Although the differences have generated much speculation, the reasons for the discrepancies are, even now, not completely known. Understanding these issues is important for everyone interested in the genetics of complex diseases, particularly because type 1 diabetes has commanded many more resources than have been dedicated to any of the complex genetic rheumatic diseases.
No doubt, our progress in explaining the genetics of rheumatic diseases will repeat at least part of the history of affective disorders and diabetes. The major lesson would appear to be that important progress in the genetic characterization is now possible, even for complex disorders. On the other hand, we should not be lulled into complacency or false security. Rather, some genetic rheumatic disease problems are likely to be virtually intractable, even given the incredible technologic advances in progress. Nevertheless, as long as research continues genes will be identified, as a result of incredible persistence, almost unbelievable luck, or both. We can then focus on the biology of the identified genes.
There are only about 25,000 human genes. One day we will have the capacity to assess inexpensively all of them in a large number of individuals simultaneously and to efficiently evaluate all of the data thereby produced. Finding the responsible genes may then be less difficult than it is today.
P.622

But even when such an incredible capacity is commonly available, understanding the biology of the genes will remain a significant impediment requiring a profound understanding and insight. Actually, we are not helped much by finding the gene if our conceptual framework for it remains outside the context of its biologic action. Consequently, several decades will be required to understand the biology of many of the genetic effects now being discovered. One only needs to consider a few examples. HLA-B27 has been powerfully associated with ankylosing spondylitis and known for more than 30 years (20). Yet we do not understand the biology at a sufficiently fundamental level to appreciate why the association of HLA-B27 with ankylosing spondylitis exists. The molecular basis for sickle cell anemia has been known for an even longer time, but only now are we beginning to develop therapies capable of influencing the level of normal hemoglobin in these patients. Thus, finding genes in many situations is now possible, but not easy. Once done, however, the much more difficult task of explaining the biology will provide a daunting challenge for many years, and perhaps generations to come.
ACKNOWLEDGMENTS
We appreciate the support of the National Institutes of Health (Grants AI42460, AR12253, AI24717, RR15577, AI31584, AR01005, AR048940, and AR049084) and the U.S. Department of Veterans Affairs for our work.
REFERENCES
1. Penrose LS. The detection of autosomal linkage in data which consist of pairs of brothers and sisters of unspecified parentage. Ann Eugen 1935;6:133–138.
2. Almquist E, Andrew S, Theilmann J, et al. Geographical distribution of haplotypes in Swedish families with Hunington’s disease. Hum Genet 1994;94:124–128.
3. Simon M, Bourel M, Fauchet R, et al. Association of HLA-A3 and HLA-B14 antigens with idiopathic haemochromatosis. Gut 1976;17: 332–334.
4. Risch N. Linkage strategies for genetically complex traits. I. Multilocus models. Am J Hum Genet 1990;46:222–8.
5. Pennisi E. A closer look at SNPs suggest difficulties. Science 1998; 281:1787–1789.
6. Salmon JE, Millard S, Schachter LA, et al. Fc gamma RIIA alleles are heritable risk factors for lupus nephritis in African-Americans. J Clin Invest 1996;97:1348–1354.
7. Houwen RHJ, Baharloo S, Blankenship K, et al. Genome screening by searching for shared segments: mapping a gene for benign recurrent intrahepatic cholestasis. Nat Genet 1994;8:380–386.
8. Stefansson K. Health care and privacy. An interview with Kari Stefansson, founder and CEO of decode Genetics in Reykjavik, Iceland. EMBO Rep 2001;2:964–967.
9. Arnett FC, Howard RF, Tan F, et al. Increased prevalence of systemic sclerosis in a native American tribe in Oklahoma: association with an Amerindian HLA haplotype. Arthritis Rheum 1996;39:1362–1370.
10. Tan FK, Stivers DN, Foster MW, et al. Association of microsatellite markers near the fibrillin gene on human chromosome 15q with scleroderma in a native American population. Arthritis Rheum 1998;41: 1729–1737.
11. Feder IN, Gnirke A, Thomas W, et al. A novel MHC class I–like gene is mutated in patients with hereditary haemochromatosis. Nat Genet 1996;13:399–408.
12. Spielman RS, McGinnis RE, Ewens WJ. Transmission test for linkage disequilibrium: the insulin gene region and insulin-dependent diabetes mellitus (IDDM). Am J Hum Genet 1993;52:506–516.
13. Morton NE. Sequential tests for the detection of linkage. Am J Hum Genet 1955;7:277–318.
14. Baldwin CT, Farrer LA, Adair R, et al. Anderson L. Linkage of early-onset osteoarthritis and chondrocalcinosis to human chromosome 8q. Am J Hum Genet 1995;56:692–697.
15. Hughes AE, McGibbon, Woodward E, et al. Localisation of a gene for chondrocalcinosis to chromosome 5p. Hum Mol Genet 1995;4: 1225–1228.
16. Krugylak L, Daly MJ, Reeve-Daly MP, et al. Parametric and nonparametric linkage analysis: a unified multipoint approach. Am J Hum Genet 1996;58:1347–1363.
17. Brown MA, Pile KD, Kennedy LG, et al. A genome-wide screen for susceptibility loci in ankylosing spondylitis. Arthritis Rheum 1998;41: 588–595.
18. Kruglyak L. Thresholds and sample sizes. Nat Genet 1996;14:132–133.
19. Corder EH, Saunders AM, Strittmatter WJ, et al. Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer’s disease in late onset families. Science 1993;261:921–923.
20. Brewerton DA, Hart FD, Nicholls A, et al. Ankylosing spondylitis and HL-A27. Lancet 1973;1:904–907.
21. Fishwild DM, O’Donnell SL, Bengoechea T, et al. High-avidity human IgG kappa monoclonal antibodies from a novel strain of minilocus transgenic mice. Nat Biotechnol 1996;14:845–851.
22. Khare SD, Bull MJ, Hanson J, et al. Spontaneous inflammatory disease in HLA-B27 transgenic mice is independent of MHC class II molecules: a direct role for B27 heavy chains and not B27-derived peptides. J Immunol 1998;160:101–106.
23. Watanabe-Fukunaga R, Brannan CI, Copeland NG, et al. Lymphoproliferation disorder in mice explained by defects in Fas antigen that mediates apoptosis. Nature 1992;26:314–317.
24. Morel L, Wakeland EK. Susceptibility to lupus nephritis in the NZB/W model system. Curr Opin Immunol 1998;10:718–725.
25. Mohan C, Morel L, Yang P, et al. Genetic dissection of lupus pathogenesis: a recipe for nephrophilic autoantibodies. J Clin Invest 1999; 103:1685–1695.
26. Morel L, Blenman KR, Croker BP, et al. The major murine systemic lupus erythematosus susceptibility locus, Sle1, is a cluster of functionally related genes. Proc Natl Acad Sci U S A 2001;98:1787–1792.
27. Pras E, Aksentijevich I, Gruberg L, et al. Mapping of a gene causing familial Mediterranean fever to the short arm of chromosome 16. N Engl J Med 1992;326:1509–1513.
28. The French FMF Consortium. A candidate gene for familial Mediterranean fever. Nat Genet 1997;17:25–31.
29. The International FMF Consortium. Ancient missense mutations in a new member of the RoRet gene family are likely to cause familial Mediterranean fever. Cell 1997;90:797–807.
30. Centola M, Wood G, Frucht DM, et al. The gene for familial Mediterranean fever, MEFV, is expressed in early leukocyte development and is regulated in response to inflammatory mediators. Blood 2000;95: 3223–3231.
31. Hull KM, Shoham N, Chae JJ, et al. The expanding spectrum of systemic autoinflammatory disorders and their rheumatic manifestations. Curr Opin Rheumatol 2003;15:61–69.
32. Livneh A, Langevitz P, Shinar Y, et al. MEFV mutation analysis in patients suffering from amyloidosis of familial Mediterranean fever. Amyloid 1999;6:1–6.
33. Shohat M, Magal N, Shohat T, et al. Phenotype-genotype correlation in familial Mediterranean fever: evidence for an association between Met694Val and amyloidosis. Eur J Hum Genet 1999;7:287–292.
34. Gershoni-Baruch R, Brik R, Lidar M, et al. Male sex coupled with articular manifestations cause a 4-fold increase in susceptibility to amyloidosis in patients with familial Mediterranean fever homozygous for the M694V-MEFV mutation. J Rheumatol 2003;30:308–312.
35. Kastner DL, O’Shea JJ. A fever gene comes in from the cold. An editorial summarizing recent findings on the PYRIN family of proteins. Nat Genet 2001;29:241–242.
36. Chae JJ, Komarow HD, Cheng J, et al. Targeted disruption of pyrin, the FMF protein, causes heightened sensitivity to endotoxin and a defect in macrophage apoptosis. Mol Cell 2003;11:591–604.
37. Stastny P. Association of the B-cell alloantigen DRw4 with rheumatoid arthritis. N Engl J Med 1978;298:869–871.
P.623

38. Nepom GT, Seyfried CE, Holbeck SL, et al. Identification of HLA-Dw14 genes in DR4+, rheumatoid arthritis. Lancet 1986;1002–1005.
39. Weyand CM, Hicok KC, Conn DL, et al. The influence of HLA-DRB1 genes on disease severity in rheumatoid arthritis. Ann Intern Med 1992;117:801–806.
40. Toussirot E, Despaux J, Wendling D. Increased frequency of HLA-DRB1 *0401 in patients with RA and bronchiectasis. Ann Rheum Dis 2000;59:1002–1003.
41. Cornelis F, Faure S, Martinez M, et al. New susceptibility locus for rheumatoid arthritis suggested by a genome-wide linkage study. Proc Natl Acad Sci U S A 1998;95:10746–10750.
42. MacKay K, Eyre S, Myerscough A, et al. Whole-genome linkage analysis of rheumatoid arthritis susceptibility loci in 252 affected sibling pairs in the United Kingdom. Arthritis Rheum 2002;46:632–639.
43. Jawaheer D, Seldin MF, Amos CI, et al. A genomewide screen in multiplex rheumatoid arthritis families suggests genetic overlap with other autoimmune diseases. Am J Hum Genet 2001;68:927–936.
44. Shiozawa S, Hayashi S, Tsukamoto Y, et al. Identification of the gene loci that predispose to rheumatoid arthritis. Int Immunol 1998;10: 1891–1895.
45. Fife MS, Fisher SA, John S, et al. Multipoint linkage analysis of a candidate gene locus in rheumatoid arthritis demonstrates significant evidence of linkage and association with the corticotropin-releasing hormone genomic region. Arthritis Rheum 2000;43:1673–1678.
46. Barton A, Eyre S, Myerscough A, et al. High resolution linkage and association mapping identifies a novel rheumatoid arthritis susceptibility locus homologous to one linked to two rat models of inflammatory arthritis. Hum Mol Genet 2001;10:1901–1906.
47. Myerscough A, John S, Barrett JH, et al. Linkage of rheumatoid arthritis to insulin-dependent diabetes mellitus loci. Arthritis Rheum 2000; 43:2771–2775.
48. Cheng SF, Lum RF, Peden E, et al. Additional support for an autoimmune susceptibility gene(s) at 6q24–27. Arthritis Rheum 2001;44 (suppl):104.
49. Van Krugten MV, Huizinga TW, Kaijzel EL, et al. Association of the TNF +489 polymorphism with susceptibility and radiographic damage in rheumatoid arthritis. Genes Immun 1999;1:91–96.
50. Castro F, Acevedo E, Ciusani E, et al. Tumor necrosis factor microsatellites and HLA-DRB1*, HLA-DQA1*, and HLA-DQB1* alleles in Peruvian patients with rheumatoid arthritis. Ann Rheum Dis 2001;60: 791–795.
51. Waldron-Lynch F, Adams C, Amos C, et al. Tumor necrosis factor 59 promoter single nucleotide polymorphisms influence susceptibility to rheumatoid arthritis (RA) in immunogenetically defined multiplex RA families. Genes Immun 2001;2:82–87.
52. Meyer JM, Han J, Moxley G. Tumor necrosis factor markers show sex-influenced association with rheumatoid arthritis. Arthritis Rheum 2001;44:286–295.
53. Chitnavis J, Sinsheimer JS, Clipsham K. Genetic influences in end-stage osteoarthritis: sibling risks of hip and knee replacement for idiopathic osteoarthritis. J Bone Joint Surg Br 1997;79:660–664.
54. Kaprio J, Kujala UM, Peltonen L, et al. Genetic liability to osteoarthritis may be greater in women than men. BMJ 1996;313:232.
55. Loughlin J. Genetic epidemiology of primary osteoarthritis. Curr Opin Rheumatol 2001;13:111–116.
56. Loughlin J, Mustafa Z, Irven C, et al. Stratification analysis of an osteoarthritis genome screen: suggestive linkage to chromosomes 4, 6 and 16. Am J Hum Genet 1999;65:1795–1798.
57. Loughlin J, Mustafa Z, Dowling B, et al. Finer linkage mapping of a primary hip osteoarthritis susceptibility locus on chromosome 6. Eur J Hum Genet 2002;10:562–568.
58. Stefànsson SE, Jónsson H, Ingvarsson T, et al. Genomewide scan for hand osteoarthritis: a novel mutation in matrilin-3. Am J Hum Genet 2003;72:1448–1459.
59. Hull R, Pope FM. Osteoarthritis and cartilage collagen genes. Lancet 1989;1:1337–1338.
60. Meulenbelt I, Bijkerk C, De Wildt SCM, et al. Haplotype analysis of three polymorphisms of the COL2A1 gene and associations with generalized radiologic osteoarthritis. Ann Hum Genet 1999;63:393–400.
61. Loughlin J, Irven C, Athanasou N, et al. Differential allelic expression of the type II collagen gene (COL2A1) in osteoarthritic cartilage. Am J Hum Genet 1995;56:1186–1193.
62. Keen RW, Hart DJ, Lanchbury JS, et al. Association of early osteoarthritis of the knee with a TaqI polymorphism of the vitamin D receptor gene. Arthritis Rheum 1997;40:1444–1449.
63. Uitterlinden AG, Burger H, Huang Q, et al. Vitamin D receptor genotype is associated with radiographic osteoarthritis at the knee. J Clin Invest 1997;100:259–263.
64. Jones G, White C, Sambrook P, et al. Allelic variation in the vitamin D receptor, lifestyle factors and lumbar spinal degenerative disease. Ann Rheum Dis 1998;57:94–99.
65. Videman T, Leppävuori J, Kaprio J, et al. Intragenic polymorphisms of the vitamin D receptor gene associated with intervertebral disc degeneration. Spine 1998;23:2477–2485.
66. Uitterlinden AG, Burger H, van Duijn CM, et al. Adjacent genes, for COL2A1 and the vitamin D receptor, are associated with separate features of radiographic osteoarthritis of the knee. Arthritis Rheum 2000;43:1456–1464.
67. Devoto M, Shimoya K, Caminis J, et al. First-stage autosomal genome screen in extended pedigrees suggests genes predisposing to low bone mineral density on chromosomes 1p, 2p and 4q. Eur J Hum Genet 1998;6:151–157.
68. Karasik D, Myers RH, Hannan MT, et al. Mapping of quantitative ultrasound of the calcaneus bone to chromosome 1 by genome-wide linkage analysis. Osteoporos Int 2002;70:457–462.
69. Koller DL, Econs MJ, Morin PA, et al. Genome screen for QTLs contributing to normal variation in bone mineral density and osteoporosis. J Clin Endocrinol Metab 2000;85:3116–3120.
70. Koller DL, Rodriguez LA, Christian JC, et al. Linkage of a QTL contributing to normal variation in bone mineral density to chromosome 11q12–13. J Bone Miner Res 1999;13:1903–1908.
71. Niu T, Chen C, Cordell H, et al. A genome-wide scan for loci linked to forearm bone mineral density. Hum Genet 1999;104:226–233.
72. Deng HW, Xu FH, Huang QY, et al. A whole-gemone linkage scan suggests several genomic regions potentially containing quantitative trait loci for osteoporosis. J Clin Endocrinol Metab 2002;87:5151–5159.
73. Ralston SH. Genetic control of susceptibility to osteoporosis. J Clin Endocrinol Metab 2002;87:2460–2466.
74. Morrison NA, Qi JC, Tokita A, et al. Prediction of bone density from vitamin D receptor alleles. Nature 1994;367:284–287.
75. Grant SF, Reid DM, Blake G, et al. Reduced bone density and osteoporosis associated with a polymorphic Sp1 binding site in the collagen type I [alpha] 1 gene. Nat Genet 1996;14:203–205.
76. Garnero P, Borel O, Grant SF, et al. Collagen I [alpha]1 Sp1 polymorphism, bone mass, and bone turnover in healthy French premenopausal women: the OFELY study. J Bone Miner Res 1998;13:813–817.
77. Sano M, Inoue S, Hosoi T, et al. Association of estrogen receptor dinucleotide repeat polymorphism with osteroporosis. Biochem Biophys Res Commun 1995;217:378–383.
78. Kobayashi S, Inoue S, Hosoi T, et al. Association of bone mineral density with polymorphism of the estrogen receptor gene. J Bone Miner Res 1996;11:306–311.
79. Duncan EL, Brown MA, Sinsheimer J, et al. Suggestive linkage of the parathyroid receptor type 1 to osteoporosis. J Bone Miner Res 1999; 14:1993–1999.
80. Brown MA, Kennedy LG, MacGregor AJ, et al. Susceptibility to ankylosing spondylitis in twins: the role of genes, HLA, and the environment. Arthritis Rheum 1997;40:1823–1828.
81. Robinson WP, van der LS, Khan MA, et al. HLA-Bw60 increases susceptibility to ankylosing spondylitis in HLA-B27+ patients. Arthritis Rheum 1989;32:1135–1141.
82. Brown MA, Kennedy LG, Darke C, et al. The effect of HLA-DR genes on susceptibility to and severity of ankylosing spondylitis. Arthritis Rheum 1998;41:460–465.
83. Hohler T, Schaper T, Schneider PM, et al. Association of different tumor necrosis factor alpha promoter allele frequencies with ankylosing spondylitis in HLA-B27 positive individuals. Arthritis Rheum 1998;41:1489–1492.
84. McGarry F, Walker R, Sturrock R, et al. The 2308.1 polymorphism in the promoter region of the tumor necrosis factor gene is associated with ankylosing spondylitis independent of HLA-B27. J Rheumatol 1999;26:1110–1116.
85. Milicic A, Lindheimer F, Laval S, et al. Interethnic studies of TNF polymorphisms confirm the likely presence of a second MHC susceptibility locus in ankylosing spondylitis. Genes Immun 2000;1:418–422.
86. McGarry F, Neilly J, Anderson N, et al. A polymorphism within the interleukin 1 receptor antagonist (IL-1Ra) gene is associated with ankylosing spondylitis. Rheumatology 2001;40:1359–1364.
P.624

87. van der Paardt M, Crusius JB, Garcia-Gonzalez MA, et al. Interleukin-1beta and interleukin-1 receptor antagonist gene polymorphisms in ankylosing spondylitis. Rheumatology 2002;41:1419–1423.
88. Beyeler C, Armstrong M, Bird HA, et al. Relationship between genotype for the cytochrome P450 CYP2D6 and susceptibility to ankylosing spondylitis and rheumatoid arthritis. Ann Rheum Dis 1996;55: 66–68.
89. Brown MA, Edwards S, Hoyle E, et al. Polymorphisms of the CYP2D6 gene increase susceptibility to ankylosing spondylitis. Hum Mol Genet 2000;9:1563–1566.
90. Laval SH, Timms A, Edwards S, et al. Whole-genome screening in ankylosing spondylitis: evidence of non-MHC genetic-susceptibility loci. Am J Hum Genet 2001;68:918–926.
91. Brown MA, Pile KD, Kennedy LG, et al. A genome-wide screen for susceptibility loci in ankylosing spondylitis. Arthritis Rheum 1998; 41:588–595.
92. Rahman P, Gladman DD, Schentag CT, et al. Excessive paternal transmission in psoriatic arthritis. Arthritis Rheum 1999;42:1228–1231.
93. Höhler T, Märker-Hermann E. Psoriatic. Arthritis: clinical aspects, genetics, and the role of T cells. Curr Opin Rheumatol 2001;13: 273–279.
94. Al-Heresh AM, Proctor J, Jones SM, et al. Tumor necrosis factor-alpha polymorphism and the HLA-Cw*0602 allele in psoriatic arthritis. Rheumatology 2002;41:525–530.
95. Gonzalez S, Martinez-Borra J, Torre-Alonso JC, et al. The MICA-A9 triplet repeat polymorphism in the transmembrane region confers additional susceptibility to the development of psoriatic arthritis and is independent of the association of Cw*0602 in psoriasis. Arthritis Rheum 1999;42:1010–1016.
96. Karason A, Gudjonsson JE, Upmanyu R, et al. A susceptibility gene for psoriatic arthritis maps to chromosome 16q: evidence for imprinting. Am J Hum Genet 2003;72:125–131.
97. Tomfohrde J, Silverman A, Barnes R, et al. Gene for familial psoriasis susceptibility mapped to the distal end of human chromosome 17q. Science 1994;264:1141–1145.
98. Samuelsson L, Enlund F, Torinsson A, et al. A genome-wide search for genes predisposing to familial psoriasis by using a stratification approach. Hum Genet 1999;105:523–529.
99. Halverson PB, Derfus BA. Calcium crystal-induced inflammation. Curr Opin Rheum 2001;13:221–224.
100. Baldwin CT, Farrer LA, Adair R, et al. Linkage of early-onset osteoarthritis and chondrocalcinosis to human chromosome 8q. Am J Hum Genet 1995;56:692–697.
101. Hughes AE, McGibbon D, Woodward E, et al. Localization of a gene for chondrocalcinosis to chromosome 5p. Hum Mol Genet 1995;4: 1225–1228.
102. Andrew LJ, Brancolini V, de la Pena LS, et al. Refinement of the chromosome 5p locus for familial calcium pyrophosphate dehydrate deposition disease. Am J Hum Genet 1999;64:136–145.
103. Pendleton A, Johnson MD, Hughes A, et al. Mutations in ANKH cause chondrocalcinosis. Am J Hum Genet 2002;71:933–940.
104. Williams CJ, Zhang Y, Timms A, et al. Autosomal dominant familial calcium pyrophosphate dehydrate deposition disease is caused by mutation in the transmembrane protein ANKH. Am J Hum Genet 2002;71:985–991.
105. Nance MA, Nuttall FQ, Econs MJ, et al. Heterogeneity in Paget disease of the bone. Am J Hum Genet 2000;92:303–307.
106. Tilyard MW, Gardner RJ, Milligan L, et al. A probable linkage between familial Paget’s disease and the HLA loci. Aust N Z J Med 1982;12:498–500.
107. Cody JD, Singer FR, Roodman GD, et al. Genetic linkage of Paget disease of the bone to chromosome 18q. Am J Hum Genet 1997;61: 1117–1122.
108. Haslam SI, Van Hul W, Morales-Piga A. Paget’s disease of bone: evidence for a susceptibility locus on chromosome 18q and for genetic heterogeneity. J Bone Miner Res 1998;13:911–917.
109. Hughes AE, Shearman AM, Weber JL, et al. Genetic linkage of familial expansile osteolysis to chromosome 18q. Hum Mol Genet 1994;3:359–361.
110. Hughes AE, Ralston SH, Marken J, et al. Mutations in TNFRSF11A, affecting the signal peptide of RANK, cause familial expansile osteolysis. Nat Genet 2000;24:45–48.
111. Kormas N, Kennerson M, Hooper A, et al. Does the TNFRSFIIA insertion mutation occur in familial Paget’s disease of bone in Australia? Bone Suppl 2000;27:25.
112. Nellissery MJ, Padalecki SS, Brkanac Z, et al. Evidence for a novel osteosarcoma tumor-suppressor gene in the chromosome 18 region genetically linked with Paget disease of bone. Am J Hum Genet 1998;63:817–824.
113. Laurin N, Brown JP, Lemainque A, et al. Paget disease of bone: mapping of two loci at 5q35-qter and 5q31. Am J Hum Genet 2001;69: 528–543.
114. Hocking LJ, Herbert CA, Nicholls RK, et al. Genomewide search in familial Paget disease of bone shows evidence of genetic heterogeneity with candidate loci on chromosomes 2q36, 10p13, and 5q35. Am J Hum Genet 2001;69:1055–1061.
115. Laurin N, Brown JP, Morissette J, et al. Recurrent mutation of the gene encoding sequestosome 1 (SQSTM1/p62) in Paget disease of bone. Am J Hum Genet 2002;70:1582–1588.
116. Hochberg MC. The application of genetic epidemiology to systemic lupus erythematosus. J Rheumatol 1987;14:867–869.
117. Sestak AL, Shaver TS, Moser KL, et al. Familial aggregation of lupus and autoimmunity in an unusual multiplex pedigree. J Rheumatol 1999;26:1495–1499.
118. Vyse TJ, Todd JA. Genetic analysis of autoimmune disease. Cell 1996;85:311–318.
119. Deapen D, Escalante A, Weinrib L, et al. A revised estimate of twin concordance in systemic lupus erythematosus. Arthritis Rheum 1992; 35:311–318.
120. Ahearn JM, Provost TT, Dorsch Ca, et al. Interrelationships of HLA-DR, MB, and MT phenotypes, autoantibody expression, and clinical features in systemic lupus erythematosus. Arthritis Rheum 1982;25:1031–1040.
121. Celada A, Barras C, Benzonana G, et al. Increased frequency of HLA-DRw3 in systemic lupus erythematosus. Tissue Antigens 1980; 15:283–288.
122. Gladman KK, Urowitz MB, Darlinkton GA. Disease expression and class II HLA antigens in systemic lupus erythematosus. Lupus 1999; 8:466–470.
123. Hashimoto H, Tsuda H, Matsumoto T, et al. HLA antigens associated with systemic lupus erythematosus in Japan. J Rheumatol 1985; 12:919–923.
124. Kachru RB, Sequeira W, Mittal KK, et al. A significant increase of HLA-DR3 and DR2 in systemic lupus erythematosus among blacks. J Rheumatol 1984;11:471–474.
125. Kawai T, Katoh K, Tani K, et al. HLA antigens in Japanese patients with central nervous system lupus. Tissue Antigens 1990;35:45–46.
126. Mehra NK, Pande I, Taneja V, et al. Major histocompatibility complex genes and susceptibility to systemic lupus erythematosus in northern India. Lupus 1993;2:313–314.
127. Reinertsen JL, Klippel JH, Johnson AH, et al. B-lymphocyte alloantigens associated with systemic lupus erythematosus. N Engl J Med 1978;299:515–518.
128. Reinharz D, Tiercy JM, Mach B, et al. Absence of DRw15/3 and DRw15/7 heterozygotes in Caucasian patients with systemic lupus erythematosus. Tissue Antigens 1991;37:10–15.
129. Reveille JD, Barger BO, Hodge TW. HLA-DR2-DRB1 allele frequencies in DR2-positive black Americans with and without systemic lupus erythematosus. Tissue Antigens 1991;38:178–180.
130. Scherak O, Smolen JS, Mayr WR. HLA-DRw3 and systemic lupus erythematosus. Arthritis Rheum 1980;23:954–957.
131. So AK, Fielder AH, Warner CA, et al. DNA polymorphism of major histocompatibility complex class II and class III genes in systemic lupus erythematosus. Tissue Antigens 1990;35:144–147.
132. Yen JH, Chen CJ, Tsai WC, et al. HLA-DMA and HLA-DMB genotyping in patients with systemic lupus erythematosus. J Rheumatol 1999;26:1930–1933.
133. Duits AJ, Bootsma H, Derksen RH, et al. Skewed distribution of IgG Fc receptor IIa (CD32) polymorphism is associated with renal disease in systemic lupus erythematosus patients. Arthritis Rheum 1995; 39:1832–1836.
134. Salmon JE, Millard S, Schachter LA, et al. Fc gamma RIIA alleles are heritable risk factors for lupus nephritis in African Americans. J Clin Invest 1996;97:1348–1354.
P.625

135. Wu J, Edberg JC, Redecha PB, et al. A novel polymorphism of Fc gamma RIIIa (CD16) alters receptor function and predisposes to autoimmune disease. J Clin Invest 1997;100:1059–1070.
136. Salmon JE, Ng S, Yoo DH, et al. Altered distribution of Fc gamma receptor IIIA alleles in a cohort of Korean patients with lupus nephritis. Arthritis Rheum 1999;42:818–819.
137. Prokunina L, Castillejo-Lopez C, Oberg F, et al. A regulatory polymorphism in PDCD1 is associated with susceptibility to systemic lupus erythematosus in humans. Nat Genet 2002;32:666–669.
138. Horiuchi T, Nishizaka H, Yasunaga S, et al. Association of Fas/ APO-1 gene polymorphism with systemic lupus erythematosus in Japanese. Rheumatology 1999;38:516–520.
139. Lee YH, Kim YR, Ji JD, et al. Fas promoter 2670 polymorphism is associated with development of anti-RNP antibodies in systemic lupus erythematosus. J Rheumatol 2001;28:2008–2011.
140. Wu J, Wilson J, He J, et al. Fas ligand mutation in a patient with systemic lupus erythematosus and lymphoproliferative disease. J Clin Invest 1996;98:1107–1113.
141. Garred P, Madsen HO, Halberg P, et al. Mannose-binding lectin polymorphisms and susceptibility to infection in systemic lupus erythematosus. Arthritis Rheum 1999;42:2145–2152.
142. Davies EJ, Teh LS, Ordi-Ros J, et al. A dysfunctional allele of the mannose binding protein gene associates with systemic lupus erythematosus in a Spanish population. J Rheumatol 1997;24:485–488.
143. Sullivan KE, Wooten C, Goldman D, et al. Mannose-binding protein polymorphism in black patients with systemic lupus erythematosus. Arthritis Rheum 1996;39:2046–2051.
144. Mehrian R, Quismorio FP, Strassmann G, et al. Synergistic effect between IL-10 and bcl-2 genotypes in determining susceptibility to systemic lupus erythematosus. Arthritis Rheum 1998;41:596–602.
145. Ahmed S, Ihara K, Kanemitsu S, et al. Association of CTLA-4 but not CD28 gene polymorphisms with systemic lupus erythematosus in the Japanese population. Rheumatology 2001;40:662–667.
146. Tebib JG, Alcocer-Varela J, Alarcon-Segovia D, et al. Association between a T cell receptor restriction fragment length polymorphism and systemic lupus erythematosus. J Clin Invest 1990;86:1961–1967.
147. Stevens A, Ray D, Alansari A, et al. Characterization of a prolactin gene polymorphism and its associations with systemic lupus erythematosus. Arthritis Rheum 2001;44:2358–2366.
148. Wilson AG, Gordon C, diGiovine FS, et al. A genetic association between systemic lupus erythematosus and tumor necrosis factor alpha. Eur J Immunol 1994;24:191–195.
149. Zuniga J, Vargas-Alarcon G, Hernandez-Pacheco G, et al. Tumor necrosis factor-alpha promoter polymorphisms in Mexican patients with systemic lupus erythematosus (SLE). Genes Immun 2001;2:363–366.
150. Komata T, Tsuchiya N, Matsushita M, et al. Association of tumor necrosis factor receptor 2 (TNFR2) polymorphism with susceptibility to systemic lupus erythematosus. Tissue Antigens 1999;53:527–533.
151. Kelly JA, Moser KL, Harley JB. The genetics of systemic lupus erythematosus: putting the pieces together. Genes Immun 2002;3(suppl 1):71–85.
152. Edberg JC, Langefeld CD, Wu J, et al. Genetic linkage and association of Fcg receptor IIIA (CDE16A) on chromosome 1q23 with human systemic lupus erythematosus. Arthritis Rheum 2002;46:2132–2140.
153. Graham RR, Ortmann WA, Langefeld CD, et al. Visualizing human leukocyte antigen class II risk haplotypes in human systemic lupus erythematosus. Am J Hum Genet 2002;71:543–553.
154. Sawalha AH, Potts R, Schmid WR, et al. The genetics of primary Sjögren’s syndrome. Curr Rheumatol Rep 2003;5:324–332.
155. Hulkkonen J, Pertovaara M, Antonen J, et al. Genetic association between interleukin-10 promotor region polymorphisms and primary syndrome. Arthritis Rheum 2001;44:176–179.
156. Perrier S, Coussediere C, Dubost JJ, et al. IL-1 receptor antagonist (IL-1RA) gene polymorphism in Sjögren’s syndrome and rheumatoid arthritis. Clin Immunol Immunopathol 1998;87:309–313.
157. Hulkkonen J, Pertovaara M, Antonen J, et al. Elevated interleukin-6 plasma levels are regulated by the promoter region polymorphism of the IL6 gene in primary Sjögren’s syndrome and correlate with the clinical manifestations of the disease. J Rheumatol 2001;40:656–661.
158. Nakken B, Jonsson R, Bolstad AI. Polymorphisms of the Ro52 gene associated with anti-Ro 52kd autoantibodies in patients with primary Sjögren’s syndrome. Arthritis Rheum 2001;44:638–646.
159. Kumigai S, Kanagawa S, Morinobu A, et al. Association of a new allele of the TAP2 gene, TAP2*Bky2 (Val577), with susceptibility to Sjögren’s syndrome. Arthritis Rheum 1997;40:1685–1692.
160. Morinobu A, Kanagawa S, Koshiba M, et al. Association of the glutathione S-transferase M1 homozygous null genotype with susceptibility to Sjögren’s syndrome in Japanese individuals. Arthritis Rheum 1999;42:2612–2615.
161. Wang ZY, Morinobu A, Kanagawa S, et al. Polymorphisms of the mannose binding lectin gene in patients with Sjögren’s syndrome. Ann Rheum Dis 2001;60:483–486.
162. Tsutsumi A, Sasaki K, Wakamiya N, et al. Mannose-binding lectin gene: polymorphisms in Japanese patients with systemic lupus erythematosus, rheumatoid arthritis and Sjögren’s syndrome. Genes Immun 2001;2:99–104.
163. Bolstad AI, Wargelius A, Nakken B, et al: Fas and Fas Ligand gene polymorphisms in primary Sjögren’s syndrome. J Rheumatol 2000; 27:2397–2405.
164. Simon M, Bourel M, Fauchet R, et al. Association of HLA-A3 and HLA-BI4 antigens with idiopathic haemochromatosis. Gut 1976;17: 332–334.
165. Feder JN, Gnirke A, Thomas W, et al. A novel MHC class I–like gene is mutated in patients with hereditary hemochromatosis. Nat Genet 1996;13:399–408.
166. Mura C, Raguenes O, Ferec C. HFE mutations analysis in 711 hemochromatosis probands: evidence for S65C implication in mild form of hemochromatosis. Blood 1999;93:2502–2505.
167. Pointon JJ, Wallace D, Merryweather-Clarke AT, et al. Uncommon mutations and polymorphisms in the hemochromatosis gene. Genet Test 2000;4:151–161.
168. Mura C, Raguenes O, Ferec C. HFE mutation analysis in 711 hemochromatosis probands: evidence for S65C implication in mild form of haemochromatosis. Blood 1999;93:2502–2505.
169. Wallace DF, Walker AP, Pietrangelo A, et al. Frequency of the S65C mutation of HFE and iron overload in 309 subjects heterozygous for C282Y. J Hepatol 2002;36:474–479.
170. Bomford A. Genetics of haemochromatosis. Lancet 2002;360:1673–1681.
171. Beutler E, Felitti VJ, Koziol JA, et al. Penetrance of 845G A (C282Y) HFE hereditary hemochromatosis mutation in the USA. Lancet 2002; 359:211–218.
172. Roetto A, Totaro A, Cazzola M, et al. Juvenile hemochromatosis locus maps to chromosome lq. Am J Hum Genet 1999;64:1388–1393.
173. Camaschella C, Roetto A, Cali A, et al. The gene TFR2 is mutated in a new type of haemochromatosis mapping to 7q22. Nat Genet 2000; 25:14–15.
174. Roetto A, Totaro A, Piperno A, et al. New mutations inactivating transferrin receptor 2 in hemochromatosis type 3. Blood 2001;97: 2555–2560.
175. Mattman A, Huntsman D, Lockitch G, et al. Transferrin receptor 2 (TfR2) and HFE mutational analysis in non-C282Y iron overload: identification of a novel TfR2 mutation. Blood 2002;100:1075–1077.
176. Njajou OT, Vaessen N, Joosse M, et al. A mutation in SLC11A3 is associated with autosomal dominant hemochromatosis. Nat Genet 2001;28:213–214.
177. Montosi G, Donovan A, Totaro A, et al. Autosomal dominant hemochromatosis is associated with a mutation in the ferroportin (SLC11A3) gene. J Clin Invest 2001;108:619–623.
178. Devalia V, Carter K, Walker AP, et al. Autosomal dominant reticuloendothelial iron overload associated with a three base pair deletion in the Ferroportin 1 gene (SLC11A3). Blood 2002;100:695–697.
179. Wallace DF, Pederson P, Dixon JL, et al. Novel mutation in ferroportin 1 is associated with autosomal dominant hemochromatosis. Blood 2002;100:692–694.
180. Kato J, Fujikawa K, Kanda M, et al. A Mutation in the iron-responsive element of H ferritin mRNA, causing autosomal dominant iron overload. Am J Hum Genet 2001;69:191–117.
181. Todd JA, Farrall M. Panning for gold: Genome-wide scanning for linkage in type 1 diabetes. Hum Mol Genet 1996;5:1443–1448.
182. Concannon P, Gogolin-Ewens KJ, Hinds DA, et al. A second generation screen of the human genome for susceptibility to insulin-dependent diabetes mellitus. Nat Genet 1998;19:292–296.