Article Text

Association of HLA type with pseudoexfoliation of the lens capsule
  1. G SESHUBABU
  1. D II/20, Dhanvantari Nagar
  2. JIPMER, Pondicherry 605 006, India
    1. J S FITzSIMON,
    2. A MULVIHILL,
    3. S KENNEDY,
    4. A FINCH,
    5. L M T COLLUM,
    6. P EUSTACE
    1. Mater Misericordiae Hospital, Eccles Street, Dublin 7, Ireland

      Statistics from Altmetric.com

      Editor,—I read with interest the recent article by FitzSimon et al.1 The finding of association of particular HLA types in pseudoexfoliation assumes importance in view of reports of possible cross reaction between antigens of infectious agents and particular HLA types in disorders such as ankylosing spondylosis.2 Genetic predisposition in association with infectious disease in the aetiology of anterior uveitis is another example.3 Hence, the following points need attention in drawing any conclusions from the study.

      In apprehension of bias towards null hypothesis, sampling bias has led to systematic differences between study populations and general populations, posing a threat to external validity. This could prevent generalisation in the population under study. Deliberate introduction of selection bias between cases and controls also resulted in systematic differences between comparison groups posing a threat to internal validity and thus invalidating the conclusions of the study with regard to the Irish population. No control measures were employed for confounding variables such as solar radiation; the association between HLA type and solar radiation thus confounding one another. Though the type of solar radiation is the same in both cases and controls, the duration of solar radiation is not the same in both groups.

      By selecting age-matched controls without pseudoexfoliation and performing ocular examination, correlation could have been more acceptable. It is preferable to risk bias towards a null hypothesis rather than introducing sampling or selection bias in a controlled study. In an earlier study both sampling and selection biases were avoided.4

      The authors have mentioned multifactorial inheritance of pseudoexfoliation. No hereditary pattern has been clearly established so far.5 Rather, association of HLA typing with multifactorial environmental influences, including solar radiation, and probably an association with infectious disease is more plausible. Finally, the reason for quoting a prevalence of 1.33% in the general population over 50 years on p 402 and in the population over 40 years on p 404 is not clear.

      A carefully planned case-control study or, preferably, a prospective cohort study could establish a cause and effect relation of HLA status and pseudoexfoliation.

      References

      Reply

      Editor,—We thank Dr Seshubabu for his interest in our work and would like to make the following points.

      In all, 128 cases with pseudoexfoliation (PXF) were identified on ophthalmic examination and the control group was 11 805 blood donors. While we did not specifically ask patients about solar or outdoor exposure, we believe our large numbers and representatives from both sexes and rural and town populations in each group reduced the risk of sampling a subset of the population. We believe that possible differences in the case and control populations are not valid in the small, uniform, and racially homogeneous population of this island.

      Solar radiation may be a factor in the pathogenesis of PXF but it is difficult to explain the different frequencies in populations of the world. It is possible that solar radiation in an individual with a certain HLA profile may lead to PXF. Likewise, specific HLA antigens may confer disease susceptibility and infection could lead to development of PXF, but this is speculative only. An HLA association would be an integral part of the pathogenesis.

      In the earlier study1 age-matched controls were used but the study numbers were much smaller. Age-matched controls in such a study would have been ideal but this was an opportunistic study in a genetically relatively isolated population.

      A population based study in Ireland documented the prevalence of PXF as 1.33% in cases over 40 years old; therefore some controls may have had undetected PXF but, as stated, this biases towards the null hypothesis.

      We agree that prospective cohort studies would clarify the cause and effect relation between PXF and HLA and emphasise that these should be in different populations in the world where PXF is a relatively frequent finding.

      References

      View Abstract

      Request permissions

      If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.