Helping to understand studies examining genetic susceptibility to sepsis

Unravelling the pathophysiology of sepsis and septic shock has been one of the major advances in medicine of the last 20 or so years. Infection leads to an inflammatory response in a programmed sequence that ensures limitation or eradication of the infection, together with survival of the host. Why then, do some patients die of overwhelming infection, while others succumb to what appears to be an over-exuberant inflammatory response? The answer must lie, to some degree, in our genetically programmed response to infection.

Some of the first insights in the genetics of the inflammatory response to infection came from experiments conducted in 1965 when C3H/HeJ mice were first identified. These mice were found to be refractory to all of the biological effects of endotoxin (LPS) [1]. Their resistance was traced to a spontaneous mutation on mouse chromosome 4. This suggested that a single pathway for LPS response must operate in mice, and by implication, other mammals. Other mutations in this gene (known as Lps) have since been discovered. Further understanding of how LPS triggered the inflammatory response to gram-negative infection might offer insight into the most powerful pro-inflammatory pathway known. Positional cloning of the area in mouse chromosome 4 found that the locus for the relevant mutations conferring refractoriness to LPS contained the Toll-like receptor (TLR) 4 gene [2]. For the first time, a receptor (or coreceptor) for LPS was found, which contained a signalling domain that could cause translocation of nuclear factor-κB (NF-κB) and thus activate mammalian cells. Whilst being resistant to the effects of LPS may appear to be an advantage, it was subsequently found that such mice were hypersusceptible to other organisms, particularly those which cause prolonged intracellular infection (such as Leishmania, Mycobacteria and Salmonella) [3]. This suggested that whatever its noxious effects, the mechanisms for LPS signalling could initiate a response leading to rapid initiation of the host immune response and thus clearance of an otherwise potentially fatal infection.

Following these findings, it has been discovered that cytokines such as tumour necrosis factor-α (TNF-α), Interleukin 1β (IL-1β), Interleukin 6 (IL6), Gamma Interferon (IFN-γ) and Interleukin 10 (IL10), together with other mediators such as clotting factors, fibrinolytic pathway factors and other pro-and anti-inflammatory mediators, are released or up-regulated following infection [4].

It appears to be the degree and duration of these proinflammatory and anti-inflammatory responses, and the balance of these responses, that determines the outcome of infection.

More recently it has become clear that the release or upregulation of many of the above factors are under genetic control. Although our understanding of the genetic basis of resistance or susceptibility to infection is in its infancy, it is clear that it is determined by a huge number of genes, and therefore patterns of inheritance will be complex [5]. The involvement of many different genes means that the contribution of a single genetic locus may be relatively small, and may depend on other genetic loci (a phenomenon known as epistasis), as well as environmental factors. In addition, there are obvious nongenetic reasons why different patients with the same infection have differing outcomes. For example, appropriateness of antimicrobial therapy, and health-care delivery are obvious prognostic factors.

Despite these difficulties, there is a growing body of disease associations with specific candidate gene polymorphisms that have been described in recent years.

Candidate gene analysis is conceptually the simplest approach to a complex disease trait such as sepsis. The understanding of the pathophysiology of sepsis has allowed the identification of a large number of cytokines and mediators for which the genetic control of variability has been, or is being sought.

In this issue of Clinical and Experimental Immunology, Arnalich et al. have investigated the influence of a polymorphism in the IL-1 receptor antagonist (IL-1Ra) gene (IL-1RN*) on the outcome of severe sepsis [6]. The authors found that there was a significant association between homozygosity for allele 2 of the IL-1RN*gene (IL-1RN*2/2) and mortality. In addition, they investigated the functional significance by correlating genotype with total production of IL-1Ra protein in stimulated PBMCs. They found that the IL-1RN*2/2 genotype was strongly associated with decreased production of Il-1Ra from PBMCs in patients with severe sepsis. There was no effect on production of IL-1β with this or any other IL-1RN* genotype. As the authors state in their conclusion, they only had 13 patients with the IL-1RN*2/2 genotype in their study, and their results conflict with previously published data which showed that carriage of allele 2 was related to a higher in vivo secretion of IL-1Ra in healthy white Australians.

There have been many reports of associations between candidate gene polymorphisms and infectious disease susceptibility and severity. These will doubtless proliferate as the human genome project bears fruit. There will be huge implications for the genetic analysis of many human diseases, but there are several pitfalls to this approach [7].

Most of the reported associations between genetic polymorphism and disease have been population based, inasmuch as they compare gene frequency in diseased individuals with those of a control group recruited form the same or a similar population, as in the study by Arnalich et al. If the results of such studies are to be meaningful, there are certain absolute requirements:

The disease under scrutiny must be strictly defined. This may be easy for certain discrete disease entities, such as leprosy or malaria, but is not so straight forward for critical illness and ‘sepsis’, where disease manifestations can be varied, and underlying aetiologies may be manifold.

The control group must be representative of the background population. For logistic reasons the control group is often selected from patients attending the hospital for other reasons, but it is essential to determine whether they are representative of the population as a whole. The most important confounder of population–based association studies is ethnic admixture. For example, if a certain ethnic group is likely to develop a disease for sociocultural reasons, a random sample of cases with the disease will contain a higher proportion of this ethnic group than is found in the general population and a spurious ‘genetic association’ may be found which simply reflects the ethnic difference rather than true disease susceptibility factors. Despite attempts to ensure ethnic matching, this may be difficult to achieve with certainty. This has lead to the growing use of intra-familial association studies, where the distribution of genotypes amongst index cases is compared from that predicted from their parental genotypes using statistical techniques such as the transmission disequilibrium test [8]. This method excludes the possibility of ethnic mismatching and should be statistically more robust than a well-conducted case-control study.

As the number of candidate genes and polymorphisms under investigation grows, there is an increasing requirement for a large sample size in order to permit multiple comparisons: a P-value 0·05 may be of little significance if there are 100 polymorphisms that are being assessed in the same study. With the human genome project reaching fruition it is likely that it will shortly be possible to screen for association with polymorphic markers in each known human gene. Therefore sample sizes of around 2000 patients are likely to be needed to validate genetic associations that give a doubling or halving of relative risk [9].

An authentic genetic association, validated in several well controlled population or family based studies of adequate sample sizes in well-defined disease entities is not all that is required to be sure that disease is caused by the polymorphism under investigation. It indicates that the genetic marker is close to a disease susceptibility locus, but it does not prove that this polymorphism is the underlying cause, as it may be in linkage disequilibrium with a functional polymorphism located in a neighbouring region of the same chromosome. To localize the causative polymorphism with certainty, it is necessary to build up a pattern of disease-associated haplotypes, which consist of clusters of linked polymorphisms that share the disease association. To identify the causative polymorphism from these haplotypes can be extremely difficult without access to a large study population in whom haplotypic linkages have been broken down by extensive genetic recombination.

There is no doubt that IL-1β is implicated in the pathophysiology of sepsis and septic shock. However, the absolute contribution of IL-1β and IL-1Ra to disease severity and outcome remains unclear [4]. There have been large studies of the use of IL-1Ra in patients with sepsis and septic shock, which have not had any significant effect on mortality [10].

The possibility of understanding the genetic contribution to response to immunomodulatory agents remains one of the most exciting prospects of the unravelling of the human genome. In the near future, patients with critical illness will be able to be genotyped within a few hours of admission to determine the genetic basis of their inflammatory response and have their treatment tailored accordingly. Until this time, studies to identify disease susceptibility genes should be in well-defined disease entities, in ethnically matched populations, and of a large enough sample size to allow multiple comparisons and mapping of multiple markers.