HLA-DEPENDENT TNF SECRETORY RESPONSE MAY PROVIDE AN IMMUNOGENETIC LINK BETWEEN PRE-ECLAMPSIA AND TYPE 1 DIABETES MELLITUS

Tumour necrosis factor (TNF) may be relevant to the pathogenesis of both pre-eclampsia and type I diabetes, and there is evidence than human TNFa responses to stimuli are HLA-DR dependent. To test the hypothesis that pre-eclampsia and diabetes may share a common immunogenetic susceptibility, 92 pre-eclampsia patients were compared with 264 general population controls. The relative frequencies of individual HLA-DR antigens in pre-eclamptics were found to correlate with reported relative TNFa responses for those antigens. Moreover, putative high responder HLA-DR I, DR3 and DR4 alleles were significantly (p<O.OO I) more frequent in pre-eclampsia patients (79%) than in controls (59 %). This hypothesis could explain the weak association between pre-eclampsia and diabetes and may help resolve the apparently conflicting literature on HLA in pre-eclampsia.


INTRODUCTION
Studies on pre-eclampsia have thrown up many conflicting ideas.There is undoubtedly a genetic component, but its nature us unclear (Cooper et aI. , 1993).Data on HLA relationships have been particularly inconsistent (Cooper et aI., 1993).Our own finding of an association between pre-eclampsia and sharing of HLA-DR4 between fetus and mother (Kilpatrick et aI., 1990) has never been confirmed or refuted, but others have failed to confirm an association with maternal HLA-DR4 or to implicate linkage disequilibrium between maternal HLA-DR4 and the putative pre-eclampsia susceptibility gene (Wilton et al., 1991;Hayward et al., 1990) , Nevertheless, the association with maternal HLA-DR4 is the only HLA association that has been reported from two independent laboratories (Cooper et aI., 1993;Simon et al. , 1988).
Diabetics are at increased risk of pre-eclampsia (Garner et al., 1990), and conversely, women with pre-eclampsia are more likely than normotensive women to experience type I diabetes in later life (Dahlquist and Kallen, 1992).lt is well established that HLA-DR4 is positively associated with insulin dependent diabetes mellitus, and it is possible these clinically unrelated diseases share a common immunogenetic susceptibility marker.
A totally separate line of investigation could also link pre-eclampsia with diabetes.The oxidative stress hypothesis attributes a central role to tumour necrosis factor (TNF) in the pathogenesis of pre-eclampsia (Stark, 1993).Human TNF secretion in response to lipopolysaccharide is HLA-Class II dependent (Santamaria et aI., 1989).Pociot et al. (1993) have investigated the relationship between TNF release and individual HLA-DR alleles.HLA-DR I, 3 and 4 individuals were found to be high responders, while HLA-DR2 and 5 individuals were low responders.Pociot et al. suggested that TNF plays a role in the pathogenesis of type I diabetes and is linked to the HLA associations of that disease.I have therefore re-analysed HLA-DR data for pre-eclampsia patients and controls to test the hypothesis that pre-eclampsia may also be associated with the DR alleles linked to a high TNF response.

METHODS
HLA-DR typing was previously carried out on 92 unrelated women whose pregnancies were complicated by proteinuric pre-eclampsia meeting the criteria for gestational proteinuric hypertension (Kilpatrick et at., 1990).Control data representing the general population were obtained from 132 women and their husbands after normal pregnancies (Jazwinska et aI., 1987).Data on TNFu secretion as a function of HLA-DR type was taken from Pociot et al. (1993).
Regression analysis was performed by Student's t-test. 2 x 2 analyses were conducted by the x 2 test and Fisher's exact test.

RESULTS
The relative TNFu secretory capacity calculated from Pociot et at.(1993) closely resembles the relative frequency of HLA-DR alleles in pre-eclamptic patients (Table 1).The DR 5,6 and 8 antigens have been grouped together because serological typing with the antisera available at the time did not distinguish accurately between those specificities, and all three were associated with a below-average TNFu response .Thus analysed, there was a striking correlation (r=O.78) which approached statistical significance (p<0.07).
When the HLA data were analysed to compare the proportion of women with preeclampsia (n=92) with HLA DR 1 or 3 or 4 (79 %) to the corresponding proportion in the general population (59% of 264 controls), the difference was highly significant (x 2 = 12.6; p<O.OO 1).Even when all patients and controls possessing HLA-DR4 were excluded from the analysis, the putative high TNFu responders were still significantly overrepresented (Table 2).

DISCUSSION
These results clearly confirm an association between pre-eclampsia and the group of HLA-DR specificities found in association with high TNFu responders.This analysis and interpretation relies on the data of Pociot and coworkers, but other groups have also found DR3 to be associated with high, and DR2 to be associated with low, TNFu responders (Bendtzen et aI., 1988;Jacob et aI., 1990;Peces et at. , 1995).
This relationship not only supports a role for TNF in the pathogenesis of preeclampsia, but may help explain some of the confusion in the literature of HLA and preeclampsia.It is not a single HLA-DR specificity that would be over-represented in the disease, but a group of antigens.which might occur in varying proportions in different series of patients.The putative HLA-DR4 association in particular has been called into question, and therefore the data were analysed separately after the withdrawal of all DR-  Although there i, 3n une\plained association between pre-eclampsia and type I diabetes.mo't Ji aberic, do not e\pe rience pre-eclampsia during pregnancy and only a small proporti on o t pre-e c lJ.rnp ~i a p3ti ents subsequently develop diabetes.These circumstances sugge 'l th aI pre-eclampsia and diabetes are distinct diseases with separate causes, ye t \\ith some relc\ant com mon fac tor linking the two.A tendency to produce a high level of T\,F in response to (presumably different) stimuli could be the common factor.
We found no singl e DR4-bearing haplotype to be over-represented in pre-eclampsia (Liston and Kilpatri ck. 1991 ) and others have failed to find any linkage between preeclampsia and the maternal HLA-DR region (Wilton eta!., 1991;Hayworth eta!., 1992).If the suggestion made here be true .it is not necessary to invoke HLA-DR as being linked to a disease susc eptibility gene.Instead, HLA-DR4 (and other DR~ alleles) would function as immune re spon se genes, modulating the TNFa response to (unspecified) stimulation.Such HLA associations would presumably be in addition to, and independ-ent of, other disease susceptibility genes .HLA would be part of a more complex genetic susceptibility, which would also include a fetal contribution (Liston and Kilpatrick, 1991).It might be worthwhile for other workers with HLA data in pre-eclampsia to reexamine their HLA-DR data in the light of this analysis .