Robert K. Semple, MB, PhD, Metabolic Research Laboratories, Institute of Metabolic Science, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 0QQ, UK. Tel.: +44 1223 769 035; Fax: +44 1223 330 598; E-mail: firstname.lastname@example.org
The complex organization and regulation of the human hypothalamic–pituitary–gonadal axis render it susceptible to dysfunction in the face of a variety of genetic insults, leading to different degrees of hypogonadotrophic hypogonadism (HH). Although the genetic basis of some HH was recognized more than 60 years ago the first specific pathogenic defect, in the KAL1 gene, was only identified within the last 20 years. In the past decade, the rate of genetic discovery has dramatically accelerated, with defects in more than 10 genes now associated with HH. Several themes have emerged as the genetic basis of HH has gradually been uncovered, including the association of some genes such as FGFR1, FGF8, PROK2 and PROKR2, both with HH in association with hyposmia/anosmia (Kallmann syndrome) and with normosmic HH, thus blurring the clinical distinction between ontogenic and purely functional defects in the axis. Many examples of digenic inheritance of HH have also been reported, sometimes producing variable reproductive and accessory phenotypes within a family with non-Mendelian inheritance patterns. In strictly normosmic HH, human genetics has made a particularly dramatic impact in the past 6 years through homozygosity mapping in consanguineous families, first through identification of a key role for kisspeptin in triggering GnRH release, and very recently through demonstration of a critical role for neurokinin B in normal sexual maturation. This review summarises current understanding of the genetic architecture of HH, as well as its diagnostic and mechanistic implications.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
The hypothalamic–pituitary–gonadal (HPG) axis subserves a critical function in permitting sexual reproduction, both through the regulation of gametogenesis and the secretion of sex hormones which confer secondary sexual characteristics. It also differs in some key respects from the other four endocrine axes that use the anterior pituitary gland as a relay point. First, GnRH neurones, the point of integration and common hypothalamic conduit for regulatory inputs to the axis, originate outside the central nervous system (CNS) in the olfactory placode before migrating to the mediobasal hypothalamus. Second, the axis is unusually dynamic, with major fluctuations in activity during development, in response to environmental stresses, and from day to day in adult life. Thus, in primates at least, HPG activity shows a characteristic ‘on-off-on’ pattern of activity, being active in utero and in early postnatal life before being repressed through obscure central mechanisms until puberty. Reproductive function is also suppressed centrally during severe stress or malnutrition. In both situations, this serves to avoid the energetic toll of maintaining reproductive readiness and attempting to sustain a pregnancy in the context of insufficient energy stores or body size. Furthermore, the nature of the endocrine interplay with the end organs is complex, at least in adult females, featuring both negative and positive feedback by sex steroids sequentially during the ovarian cycle.
On one hand, such sophisticated regulation of the axis is necessary to sustain cyclical fecundity while being able to switch efficiently to noncyclical endocrine activity in the early weeks of pregnancy. On the other, this ontogenic and regulatory complexity means that the HPG axis is particularly vulnerable to dysfunction in the face of genetic insults, which present as hypogonadotrophic hypogonadism (HH). The genetic defect underlying HH can be identified in some patients; however, collective experience suggests that in up to 70% the putative mutation(s) remain unidentified. This presents both a diagnostic challenge and a great scientific opportunity: while the inherent complexity of the system makes confident identification of candidate genes difficult, nonhypothesis-based genetic techniques such as homozygosity mapping in consanguineous families with multiple affected members have the power to identify novel mechanisms in human HPG function. This is invaluable as regulation of HPG function in commonly used model species such as mice differs in key respects from that in humans.1 Indeed, the recent history of investigation of the HPG axis is peppered with examples of major mechanistic insights coming from human genetic studies of rare patients with familial HH.
Hypogonadotrophic hypogonadism may result from disruption of the complex process of GnRH neuronal migration during development, disruption of signals required for GnRH neurone survival, disruption of GnRH neuronal connections or perturbaztion of GnRH secretion or action despite anatomically normal GnRH neurones. One key clinical discriminator of developmental defects as opposed to purely functional defects is the co-existence of HH with hyposmia/anosmia. Although this clean distinction has become blurred, as will be discussed, it remains a clinically useful basis for classification.
More than 150 years ago, Aureliano Maestre de San Juan described an adult male with testes of prepubertal size and absent olfactory bulbs.2 In the 1940s, Franz Kallmann added to this by documenting hypogonadism co-segregating with anosmia in two families,3 convincingly establishing a genetic basis for the condition, while de Morsier later added neuropathological detail.4 The discovery in the early 1970s of GnRH5,6 led quickly to the demonstration that the hypogonadism of Kallmann Syndrome (KS) – as the combination of HH and anosmia has come to be labelled in the English speaking world – was central in origin.7
Kallmann Syndrome has a prevalence of around 1 in 8000, and is five times more common in men than women. X-linked recessive, autosomal dominant (AD) and autosomal recessive (AR) patterns of inheritance are observed; however, many cases are sporadic or do not appear to show a Mendelian inheritance pattern. The first underlying genetic defect was identified 47 years after Kallmann’s report, since when there has been a dramatic increase in the rate of identification of further KS genes (Table 1).
Table 1. Genetic defects causing hypogonadotrophic hypogonadism, and their modes of discovery
KAL1. Study of a patient with KS and a deletion on the X chromosome led to the identification nearly 20 years ago of the KAL1 gene, encoding an extracellular glycoprotein called anosmin 1, as the site of mutations in X-linked KS.8–10 Anosmin 1 associates with the cell membrane via heparin sulphate proteoglycans (HSPGs) and has multiple protein interaction domains suitable for a role in facilitating cell surface signalling. Importantly, post mortem study of a 19-week-male foetus with a deletion including KAL1 revealed that GnRH neurones had differentiated from the olfactory placode and had begun to migrate before being arrested in a tangle of olfactory neurones at the point of entry into the CNS.11 A total of 10–20% of males with KS carry KAL1 mutations or intragenic microdeletions,12–14 and most pathogenic mutations entirely abrogate protein function. Interestingly, the KAL1 locus on the X chromosome appears partially to escape somatic inactivation,9 meaning that men have lower KAL1 expression than women. It has been suggested that this lowers the threshold for clinical expression of mutations in other functionally related genes in men, and that this may account for the strong male preponderance of KS.15 The KS phenotype produced by KAL1 mutation seems both more severe and less variable than that seen with other known molecular defects,14,16 and only one case has been reported to date with normosmic IHH (nIHH) and a convincing KAL1 mutation.17
Accessory clinical features that are particularly prevalent in KAL1-associated KS include mirror movements or synkinesia, which occurs in up to 75% of patients and unilateral renal agenesis, which occurs in around 30% of patients (reviewed in 18).
FGFR1, FGF8. The next major genetic advance in KS came in 2003, also beginning with patients with overlapping chromosomal deletions. This narrowed the search to 3 genes, of which FGFR1, encoding the type 1 fibroblast growth factor receptor, was the most likely candidate. On sequencing 129 patients with KS, 10% were found to have loss-of-function mutations,19 and although precise figures vary, other studies collectively support this prevalence of FGFR1 mutations in KS.20,21 Interestingly, FGFR1 requires HSPG as a coreceptor, and anosmin 1, which is also HSPG-associated, is likely to play a role in facilitating FGFR1 signalling.15
Further study has revealed a spectrum of phenotypes associated with FGFR1 mutations, and has refuted the notion that anosmia/hyposmia may reliably discriminate ontogenic from purely functional defects in GnRH release: loss of FGFR1 function has been confirmed to produce reproductive abnormalities ranging from severe AD KS through AD, fully penetrant nIHH to delayed puberty,20–24 with combinations of these syndromes or their formes frustes, including isolated anosmia/hyposmia, being observed in affected families.23
Discovery of a mutated receptor in endocrine genetics leads logically to genetic examination of the gene encoding the endogenous ligand in the same condition. However the complexity of FGFR1 signalling, which involves various co-receptors and around 11 different FGF ligands, delayed this until 2008, when several lines of evidence implicated FGF8 as the relevant endogenous ligand.25 Sequencing of FGF8 in 461 patients with HH revealed heterozygous loss-of-function mutations in six patients. As for FGFR1, these mutations were not confined to KS. Indeed, while two were identified in familial KS, one was found in familial nIHH, two in sporadic nIHH, and one in the rare and milder adult onset nIHH,25 with either AD and AR inheritance seen in familial cases.
Although there are no reliable pathognomonic accessory features of FGF8/FGFR1 loss of function in most patients, cleft palate is found in up to 30% of patients (reviewed in reference19) while much rarer cartilage abnormalities in either ear or nose, and some digital anomalies, have been specific to FGFR1 defects in reports to date.18
PROKR2, PROK2. The PROK2 gene encodes prokinetecin 2, an 81 amino acid peptide that signals via the G protein-coupled product of the PROKR2 gene. This ligand-receptor pair were implicated initially in diverse functions,26–29 but it was the observation that prok230,31 or prokr2 knockout mice had defective development of the olfactory bulbs and failed migration of GnRH neurones32 that made the human genes strong candidates for KS. In the first genetic study reported, predicted loss-of-function mutations in PROKR2 or PROK2 were found in 9% of KS patients, most of which were heterozygous, however homozygous and compound heterozygous mutations were also described.33 Interestingly, this and subsequent reports showed patients with PROK2 or PROKR2 defects to have considerable phenotypic variability,31,34,35 with 6% of KS patients and also 3% of nIHH patients affected in one study.35 A variety of accessory features including fibrous dysplasia, sleep disorder, severe obesity, synkinesia, and epilepsy have been noted in patients with PROK2 or PROKR2 mutations, but none of these are either highly prevalent or diagnostic.35
CHD7. CHARGE syndrome (Colobomata, Heart Anomalies, choanal Atresia, Retardation, Genital and Ear anomalies) includes IHH and hyposmia as part of a wider spectrum of abnormality. Although mostly sporadic, AD familial cases have been reported, and in 2004 causative mutations were found in CHD7,36 which encodes a chromatin-remodelling factor, such mutations subsequently being found to account for about 60% of cases. Based on the hypothesis that KS and nIHH may be ‘formes frustes’ of CHARGE syndrome, CHD7 was screened in nearly 200 patients with KS or nIHH, and heterozygous, likely loss-of-function, mutations were identified in three KS and four nIHH patients.37 A subsequent study suggested that affected patients may be identifiable clinically due to the occurrence of mild features of CHARGE syndrome, including particularly deafness and hypoplasia of the semicircular canals, in conjunction with HH.38
NELF. The gene encoding the nasal embryonic LHRH factor, NELF, was identified as a strong candidate gene for KS based on its characterization as an axon guidance factor in murine GnRH neurones. However, although rare sequence variants have been reported in KS,39,40 no functional studies have been reported.
Phenotypic and genetic complexity in Kallmann syndrome
A recurring theme of these studies has been the substantial variation in clinical expression of the same genetic defect in families of patients with KS. Thus, even where the proband is investigated due to full blown anosmia, HH and absent puberty, other family members have not uncommonly been found to have less severe HH manifesting as delayed puberty or even as the rare condition of acquired HH in adulthood.41–43 Furthermore family members may not only have only partial hyposmia rather than anosmia, but they may even have no discernible impairment of smell, thus being classified as nIHH. This emphasizes that the distinction between KS and nIHH is not entirely dichotomous, and that a careful family history may be necessary to tease out whether the underlying defect in a patient with IHH is likely to be developmental or functional.
This type of phenotypic heterogeneity may in principle be ascribed to environmental factors, some of which may be exerted by epigenetic effects on gene expression, however a second explanation is the co-existence within families of defects in two or more different genes which interact functionally. In KS/HH this idea has received strong and growing support since the first report of digenicity in HH40 with the identification of multiple families with mutations in more than one of the KS genes described above (Table 2). This means that many family histories of KS patients fail to conform obviously to Mendelian inheritance patterns, a phenomenon now increasingly recognised in other classic ‘Mendelian’ genetic diseases.44
Table 2. Examples of digenic interactions in hypogonadotrophic hypogonadism
Such genetic interactions commonly mirror functional interactions among encoded proteins, and although many questions remain to be answered about the precise biology of KS/HH genes, well reviewed evidence supports the contention that the gene products are involved in axonal migration and survival of GnRH neurones during their journey from the olfactory placode to the mediobasal hypothalamus, with many of them interacting physically at the cell membrane to form part of the axon guidance and survival machinery.15
Defects in HPG function
For all the phenotypic variability seen in the KS spectrum of disorders, there is little evidence that KS genes play any role in functional regulation of an anatomically normal HPG axis. Because modulation of inputs into normal GnRH neurones is of most relevance to therapeutic manipulation of the HPG axis, there is intense interest in identifying the genes involved in pure nIHH, and in this area human genetics has been particularly fruitful.
The HPG axis in humans is fully active for the first few months of life before entering a state of quiescence until puberty.1 Classic studies of agonadal humans45 and gonadectomized rhesus monkeys46 provide strong evidence that this ‘juvenile pause’ involves central suppression of the GnRH pulse generator. Furthermore, the GnRH content of prepubertal primate hypothalami is similar to that of mature adults,47 and tiny doses of glutamate agonists, which are potent GnRH secretagogues, can prematurely reactivate the axis.48 These findings suggest that prepubertal suppression of HPG function represents functional stasis rather than immature neural development. Nevertheless how this is exerted, and how it is relieved at puberty, remain among the great unanswered questions in human endocrinology. It is this question which may most directly be illuminated by identification of the genetic defects that disrupt the process (Table 1).
Known genetic defects in HPG function
GNRH1/GNRHR. GnRH, acting via the GnRH receptor, has been called the ‘pilot light’ of the HPG axis,49 and has long been the most obvious candidate gene to produce a purely functional defect in its activity. However although genetic defects in the GNRHR gene, encoding the GnRH receptor, were identified more than 10 years ago,50,51 it was only during the writing of this article that convincing genetic defects in the GNRH1 gene itself were reported for the first time.52,53 Both GNRH1 and GNRHR defects produce AR isolated nIHH without developmental defects such as hyposmia,54,55 and in large series GNRHR mutations have been suggested to account for about 40–50% of familial AR nIHH, and around 17% of sporadic nIHH.54 Early evidence suggests that GNRH1 mutations may also produce AD HH,53 but further support for this is awaited.
Considerable variability in the clinical expression of GnRHR mutations within families has been described, particularly where these lead to only partial loss of function,54,56 and in one case this has been attributed to interaction with a mutation in FGFR1,40 establishing that combinations of developmental and functional defects may produce different phenotypes. While physiological study of patients with partially inactivating GNRHR mutations has refined understanding of differential actions of GnRH on gonadal function in vivo,56,57 the robust and well established role of GnRH as the ‘pilot light’ of the HPG axis prior to finding of GNRHR mutations meant that insights into upstream neuroendocrine control of the HPG axis afforded by the discovery of GNRHR mutations were limited.
By contrast, two discoveries in the past 6 years have led to unexpected leaps in understanding of regulation of GnRH secretion, and significantly, both of these have come from genetic approaches which did not rely on predefined candidate genes, illustrating the power of such ‘non-hypothesis-based’ genetics to yield novel mechanistic insights.
KISS1R. The first of these insights came in 2003 with nearly simultaneous discovery, using linkage mapping in familial nIHH, of loss-of-function mutations in a G protein-coupled receptor, GPR54.58,59 GPR54 had previously been shown to be the receptor for small peptides derived from the KISS1 gene60 (leading to its recent redesignation as KISS1R), which, in turn, had been identified in a screen for putative metastasis suppressors, but it had not previously been implicated in HPG function. Allied to this Gpr54 knockout mice were discovered at the same time to be a faithful phenocopy of the human condition,59 with nearly absent sexual maturation despite neuroanatomically normal GnRH neurones and normal hypothalamic GnRH content.
Following those seminal reports of human KISS1R mutations it was established in short order that kisspeptin acts as a potent GnRH secretagogue across many species including rodents,61,62 sheep,63 monkeys64 and humans.65 Furthermore multiple independent mouse knockouts of Kiss1r and Kiss1 have been reported, all of which recapitulate the human nIHH phenotype, confirming KISS1/KISS1R to be robust proximal regulators of GnRH release.66 Increased expression of KISS1, and increased connectivity between KISS1 and GNRH neurones have been suggested to be one of the effector mechanisms at puberty,64,67 while downregulation of KISS1 has been suggested to mediate some of the HPG suppression seen in severe nutritional deprivation.68 KISS1 neurones also are highly responsive to oestrogen, and have been implicated in both negative and positive central feedback of sex steroids to GnRH production.69,70
In the past 6 years further reports of human loss-of-function mutations, each associated with AR nIHH have emerged, however it has become clear that KISS1R mutations are an exceedingly rare cause of HH, with only 19 individuals from six unrelated families so far reported to have proven loss-of-function mutations,58,59,71–73 and uncharacterized rare KISS1R variants found in <3% of patients with nIHH in another large screening study.74 No pathogenic KISS1 mutations have been reported to date. Individuals with nIHH due to KISS1R mutations have been shown to have severely reduced LH pulse amplitude, but approximately normal pulse frequency.59,73 Successful pregnancy with normal delivery after exogenous induction of fertility has been reported in a patient with KISS1R mutations, suggesting that KISS1R function is not critical for placental function despite high levels of placental expression.75
On the basis of these findings, KISS1, acting via the KISS1R, is now firmly established as an evolutionarily conserved, potent stimulator of GnRH release that is necessary for normal sexual maturation. Not quite so clear yet, however, is whether KISS1/KISS1R represents the regulated input to GnRH neurones which determines the timing of puberty. It has been shown that KISS1 is upregulated peripubertally in the hypothalamus in both primates64 and rodents,67,68 albeit with differences in the intrahypothalamic distribution of KISS1 expression,67,76 that exogenous kisspeptin can induce precocious sexual maturation in mice, and that central kisspeptin can maintain normal adult HPG function after priming of prepubertal rhesus monkeys.64 Finally, a gain-of-function KISS1R mutation has been associated with precocious puberty in a single patient.77 However while upregulation of KISS1 may be part of the proximal mechanism involved in upregulation of GnRH secretion peripubertally, the strikingly preserved ability of kisspeptin potently to stimulate GnRH release in rodents as well as humans challenges the contention that changes in kisspeptin tone may explain the juvenile pause and its relief: studies of gonadectomised rodents78–80 and primates46 suggest that only primates exhibit true centrally-mediated suppression of GnRH secretion in the prepubertal period, and it might be expected that any single key trigger of this suppression would also show divergence in its effects between primates and rodents.
TAC3/TACR3. The major and novel insights into HPG regulation initiated by human genetic study of KISS1R led us recently to reapply the strategy of studying nIHH in rare consanguineous families with multiple affected members, concentrating on families without mutations in known HH-related genes. This led to identification of three different loss-of-function mutations in TACR3, the gene encoding the neurokinin 3 receptor (NK3R), and one in TAC3, the gene encoding neurokinin B (NKB), its endogenous ligand.81,82 In these initial studies, defects in either TACR3 or TAC3 were found in 11 patients from five out of 10 families studied, but in none of 50 sporadic cases.81,82 Although follow-up studies are awaited, these findings prove that intact function of the NKB/NK3R system is required for normal HPG activation at puberty. The presence of micropenis and cryptorchidism in male patients with TACR3 mutations signifies that intact NKB/NK3R function is required also for normal fetal gonadotrophin secretion.
Neurokinin B belongs to the tachykinins, a phylogenetically ancient family of secreted peptides including substance P and neurokinin A.83 The NK3 receptor, mainly expressed in the CNS, is the most selective of the tachykinin receptors, with highly preferential binding and activation by NKB.84,85 The process of teasing out the precise role of NKB/NK3R in regulating HPG action is just beginning, but the past 6 years of investigation of KISS1/KISS1R provides a valuable paradigm. Already several key areas of similarity and difference are apparent.
Kisspeptin vs. Neurokinin B
Both Kisspeptin and neurokinin B are excitatory neuropeptides that act on target cells via Gq-coupled receptors. Both receptors are expressed on GnRH neurones,70 and both peptides are co-expressed by neurones in the arcuate nucleus of the hypothalamus that project to GnRH neurones.70 Furthermore several lines of evidence suggest that expression of both peptides in the arcuate is downregulated by oestrogen.70 However although it is tempting to infer that kisspeptin and neurokinin B thus play closely similar roles in relaying feedback from sex steroids to GnRH production, other observations suggest major differences in the their actions: while kisspeptin is a potent GnRH secretagogue in all species reported, central administration of a potent NK3R agonist failed to stimulate GnRH release in rats.85 Moreover, while Kiss1r knockout mice are faithful models of the human nIHH phenotype, Tacr3 knockout mice have grossly normal fertility (86and Dr Jeffrey Stock, Pfizer, personal communication), though detailed study of the reproductive function of these animals is awaited.
These discrepancies should inform and motivate future studies, and suggest that NKB may play a role in HPG activation which is at least partly distinct from that of kisspeptin. Indeed, although it remains likely that loss of NKB action does lead to defective GnRH production from anatomically normal GnRH neurones, the blurring between KS and nIHH suggested by studies of ‘KS’ genes is a reminder that the lack of hyposmia in the patients identified to date does not absolutely rule out a role of NKB/NK3R in the survival or precise localisation of GnRH neurones. Furthermore, although NKB and kisspeptin are co-expressed in the arcuate nucleus, NKB has a wider pattern of CNS expression than KISS1,87 suggesting that the arcuate may not be the sole, or even the most important, site at which NKB exerts its influence on reproductive function. Rapid progress in answering these questions is to be expected over the next few years.
Implications for genetic diagnosis of HH
Treatment of HH aims first to correct the endocrine deficit of sex hormones, and second, where desired, to achieve fertility. Both these aims can generally be successfully achieved with available therapies, and so in that sense making a molecular diagnosis in patients with HH is not critical. Furthermore, while recent genetic advances have greatly improved understanding of the basic biology of the HPG axis in humans, they pose new challenges for clinical genetic testing and counselling. Molecular genetic diagnosis in highly penetrant Mendelian disorders permits confident explanation and counselling of affected patients and appropriate screening of family members, however the variable penetrance of many ‘HH’ mutations, and frequent deviation from Mendelian inheritance due to oligogenic interactions complicates prognostication and hence genetic counselling. This needs to be considered carefully for each patient and family prior to embarking upon genetic testing, especially as the people in whom a genetic diagnosis may be most valuable are prepubertal female relatives of index cases. In this group there is no clinical sign of endocrine disorder even in those who subsequently prove to have full blown nIHH, and yet in these patients forewarning of failed puberty may permit earlier treatment and thus mitigation of psychological distress. In the future, it is to be anticipated that microarray-based approaches to parallel sequencing of many disease genes simultaneously may be applied diagnostically to HH, accelerating genetic diagnosis and allowing rapid identification of oligogenic networks of pathogenic mutations. This will aid accurate genetic counseling and avoid problems arising from sequential identification of several mutations during prolonged screening of individual genes. Accredited genetic testing is currently only available for some HH-related genes, however, in part because of the recent rapid progress in genetic research, but a notional algorithm for future clinical and genetic evaluation of patients with HH once testing is more widely available is shown in Fig. 1.
The major insights into the basic biology of the HPG axis that have come from human genetics makes clear that strong consideration should be given to enrolling patients in research studies. As well as offering them the possibility of finding an explanation for their endocrine disease, identification of specific genetic defects will permit dedicated search for and treatment of associated clinical features, and may prove to be one of the best ways in the long term of identifying new targets for intervention in manipulating human HPG function for therapeutic benefit.
Despite the major genetic advances in understanding HH over the past 20 years, a pathogenic genetic defect can only currently be found in around 30% of cases (Table 3), suggesting that many mutations remain to be discovered which may give new insights into the organisation of the neurocircuitry regulating GnRH secretion. Gathering rare families with nIHH will form a key part of the strategy to gain these insights. More generally, as increasing understanding is gained of binary inputs into GnRH neurones, the more challenging task of understanding how they are integrated into a complex pulsatile system will become of increasing importance. Only once this is achieved will full insight into the obscure mechanism which reawakens the GnRH pulse generating machinery from its prepubertal suspended animation, be achievable.
Table 3. Approximate prevalence of known genetic defects in hypogonadotrophic hypogonadism
Normosmic Hypogonadotrophic Hypogonadism
*Small studies reported to date found mutations in TACR3 in 40% and in TAC3 in 10% of familial cases of nIHH, but further reports are awaited. †May usually be identified by accessory features including semicircular canal hypoplasia. Note that the prevalence of genetic defects has consistently proved to be significantly higher in familial than in sporadic disease.
Unknown HH (KS + nIHH)
Robert K. Semple is supported by the Wellcome Trust (Intermediate Clinical Fellowship 080952/Z/06/Z). A. Kemal Topaloglu is supported by the Scientific and Technological Research Council of Turkey (TÜBİTAK): The Support Programme for Scientific and Technological Research Projects (1001), project 106S276; and from the Cukurova University Scientific Research Projects Support Unit, Grant TF-2006-BAP20.