A perennial discussion among neurologists revolves around whether multiple sclerosis (MS) is primarily a degenerative disease with secondary inflammation or an autoimmune disease in which the immune system turns against the body. This question has now been answered through investigation of the genetics of MS and the sequencing of the human genome, all of which implicate the immune system (see A complex code). In 2007, the first interrogation of the whole human genome in patients with MS clearly identified two gene regions, or haplotypes, that are clearly associated with risk of developing the disease. In 2011, an international effort examining 9,772 cases of MS collected by 23 research groups working in 15 different countries expanded the number of gene regions identified to more than 50. Moreover, a recent meta-analysis of 14,802 cases and 26,703 controls from about 15 separate studies found 113 additional regions associated with disease risk that had not been previously observed. Each haplotype individually has only a small effect on the risk of developing MS; however, together they account for almost one-half the genetic risk of disease.

We have long known that autoimmune diseases cluster in families, and patients frequently have more than one. We have now evaluated the extent of this sharing of genetic risk across seven immune-mediated diseases and found that MS has many of the same genetic risk factors as other autoimmune diseases. So, mechanistically, MS is primarily an immune — not a degenerative — disease, whatever the inciting event might be.

Understanding the function of these genetic variants could provide insights into how they contribute to the inflammation of the central nervous system. Although individual genetic variants only weakly predict the risk of developing MS, they nevertheless exert major biological effects, such as dictating protein-expression levels or splicing. One particular genetic variant near the CD58 gene, for example, has only a mild effect on developing MS (with an odds ratio for risk of 1.2), yet it is strongly associated with decreases in the expression of CD58, an adhesion molecule necessary for T cells to interact with macrophages and other antigen-presenting cells.

Identifying genetic variants for the highly complex phenotype of autoimmune-disease risk requires thousands of subjects. But because genetic variation can be directly linked to biological effects, significantly smaller cohorts are required to elucidate genotype-to-phenotype function (provided the investigations are conducted in subjects who carry the risk alleles but do not have inflammatory disease or MS, and are not receiving immune-modulating therapy, as these will confuse the outcomes).

One of the most striking observations relates to CD4 T cells, which are central to immune regulation. Genetically deleting the FOXP3 transcription factor, which is critical for their function, results in severe autoimmune diseases in animal models or, if it occurs naturally in humans, in a rare autoimmune disease known as immunodysregulation, polyendocrinopathy, enteropathy and X?linked (IPEX) syndrome. Defects in CD4 cells are common to autoimmune type 1 diabetes and MS; in both, secretion of the inflammatory cytokine interferon g renders the CD4 cells unable to prevent autoimmune responses. In fact, several genetic variants associated with both MS and autoimmune diabetes — including Helios, Ikaros, CD58, and interleukin-2 and interleukin-7 receptors — are linked with regulatory T-cell function.

Assessing these genetic results is not straightforward. If we view the cup as half empty, we have merely found some genetic factors that have individually weak effects on MS risk. These factors are neither necessary nor sufficient to cause MS — they cumulatively explain only 40% of heritable risk, and are therefore of limited predictive use. But maybe the cup is half full. After all, we have identified more than 100 genetic regions associated with disease risk that fall along distinct immunological pathways. This knowledge might allow us to identify the molecular underpinnings of risk, and find new targets for diagnosis and therapy. For example, anti-tumour-necrosis factor (TNF) therapies have dramatic clinical effect in some patients with rheumatoid arthritis (RA) and inflammatory bowel disease (IBD), but can exacerbate disease in patients with MS. However, many different genetic variants in the TNF pathway have been associated with disease risk in patients with these diseases. Experiments are now in progress to determine how these genetic variants determine both in vitro responses to TNF and in vivo responses to anti-TNF treatments. Perhaps patients with RA and IBD who do not respond to anti-TNF treatments have the genotypes associated with MS.

Perhaps most importantly, the common genetic background among the different autoimmune diseases suggests a new way to view these conditions. There is a wide spectrum of therapeutic responses to immune manipulations in patients with autoimmune diseases. So, rather than defining this class of diseases by clinical phenotype, perhaps we should regroup them by genetic variations and biological phenotype. There is no reason to assume that all the autoimmune diseases traditionally classified around one organ system will necessarily use the same immune mechanisms. For example, there is potential genetic evidence implicating both the pathways involving T-helper 1 (TH1) and TH17 immune cells in MS, but which is predominant might differ from one MS patient to the next.

We have traditionally defined and treated autoimmune diseases by classic gross pathology and clinical symptoms — inflammation in the target organ ultimately defined the disease. Using genomic approaches, we have begun to consider human autoimmune diseases with respect to their molecular pathology. Defining autoimmune diseases around the genetic and biological pathways leading to disease risk, rather than simply the target organ, might spark a new era of rational therapeutic manipulation.