Neurology -- Correspondence for Maher et al., 58 (1) 79-84

Correspondence published:

Reply to Letter to the Editor: Richard H Myers (3 May 2002)

Epidemiologic study of 203 sibling pairs with Parkinson’s disease: The GenePD study: Raul de la Fuente-Fernandez (3 May 2002)

Reply to Letter to the Editor 3 May 2002

Richard H Myers
Boston University School of Medicine
Send Correspondence to journal:
Re: Reply to Letter to the Editor

rmyers{at}bu.edu Richard H Myers

We concur with Dr. Fuente-Fernández that an assessment for familial patterns of age at onset should include both living and deceased affected relatives. We recently completed such a study in 948 randomly ascertained PD cases drawn from clinics at Boston University, the University of Medicine and Dentistry of New Jersey and the University of Virginia. [2] In formal segregation analysis, we found compelling evidence for a major gene influencing age at onset in Parkinson's disease. In this sample, the intraclass correlation for age at onset in 41 PD affected sibling pairs was r=0.48, which is only modestly below the r=0.56 figure reported by Maher et al. [1] These findings support strong familial similarity for onset age in PD.
The objective of our analysis was to contrast the similarity in age at onset with similarity in year at onset, and was not intended as a formal assessment of familial patterns of onset age. That the sample is drawn from recent diagnoses, will have little bearing on the ages at onset of the participants. Indeed, any bias in the design will serve to inflate the similarity in the year of onset, because the year of onset is restricted to past events among living cases. This ascertainment may bias toward relatively recent events, increasing the similarity for year of onset, or a bias toward the null hypothesis.
Our recent genome scan and linkage analysis for age at onset in PD [3] reveals that the PARK3 locus, and three other loci, may influence age dependent penetrance. The etiology of PD remains elusive but studies of onset age provide evidence for possible genetic influences in the expression of the illness and may provide insights into its cause.
References:
1. Maher NE, Golbe LI, Lazzarini AM, et al. Epidemiologic study of 203 sibling pairs with Parkinson's disease: the GenePD study. Neurology 2002;58:79-84.
2. Maher NE, Currie LJ, Lazzarini AM, et al. Segregation analysis of Parkinson disease revealing evidence for a major causative gene. Am J Med Genet 2002;109:191-197.
3. DeStefano AL, Lew MF, Golbe LI, et al. Park3 influences age at onset in Parkinson disease: A genome scan in the GenePD study. Am J Hum Genet 2002; 70:1089-1095.

Epidemiologic study of 203 sibling pairs with Parkinson’s disease: The GenePD study 3 May 2002

Raul de la Fuente-Fernandez
University of British Columbia Vancouver Canada
Send Correspondence to journal:
Re: Epidemiologic study of 203 sibling pairs with Parkinson’s disease: The GenePD study

raulff{at}interchange.ubc.ca Raul de la Fuente-Fernandez

I read with interest the article by Maher et al. on sibling pairs with Parkinson's disease (PD). [1] They found that the intraclass correlation of ages at onset of PD between sibling pairs (r = 0.56) was larger than the correlation between the corresponding years of onset (r = 0.29). The authors concluded that this observation supports a genetic component for PD. While this may well be the case, there are some statistical problems, mostly associated with the study design, that preclude drawing such a conclusion from this study.
First, this is a cross-sectional study (data collection occurred between September 1997 and July 2000), which forces the distribution of year of onset to be highly skewed. In other words, cases of recent diagnosis (say between 1990 and 2000) will tend to predominate, particularly in a degenerative disorder such as PD, and the distribution of year of onset will be truncated on the right, with a maximum value of 2000 (see figure in Ref 1). This limitation inherent in the study design will, by itself, tend to decrease the correlation between siblings' years of onset.
Second, the study focuses on families with two affected living siblings (sibling pairs in which one affected sibling was deceased at the time of the study were excluded from the analysis). This introduces another problem when applying correlation analysis. Ages (current ages at the time of the study) between siblings will tend to be highly correlated (i.e. two siblings will tend to be more alike in age than two people taken at random from the general population). In addition, since any cross-sectional study on PD will tend to include a great number of cases of recent diagnosis (see above), current age and age at onset are also likely to be highly correlated with each other. For example, the Pearson's correlation coefficient between current age and age of onset in another epidemiological cross-sectional study on 299 PD cases [2] was found to be 0.8. The combined effect of these two factors will certainly tend to inflate the correlation of ages at onset between sibling pairs (i.e., much of the correlation will be due to the effect of current age).
Correlation analysis is a powerful and effective tool to address a great number of research questions regarding the relationship between variables. However, its use (misuse) can sometimes lead to erroneous conclusions. [3]
References
1. Maher NE, Golbe LI, Lazzarini AM, et al. Epidemiologic study of 203 sibling pairs with Parkinson's disease: the GenePD study. Neurology 2002;58:79-84.
2. de la Fuente-Fernández R. Maternal effect on Parkinson's disease. Ann Neurol 2000;48:782-787.
3. Altman DG. Use and misuse of correlation. In: Altman DG, ed. Practical statistics for medical research. London: Chapman & Hall, 1991:282-285.