Colon cancer genetics / edited by Patrick M. Lynch and Henry T. Lynch.

Date:
[1985], ©1985
Catalogue details

Licence: Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0)

Credit: Colon cancer genetics / edited by Patrick M. Lynch and Henry T. Lynch. Source: Wellcome Collection.

    44/344 (page 24)
    24 COLON CANCER GENETICS unaffected as simplex. If one or more of the siblings of the index case are affected, the sibship is said to be multiplex. This is the nomenclature of Morton. Ascertainment is complete when the relative proportions of various multiplex sibships to each other and to simplex sibships are consistent with the truncated binomial distribution determined by the sibship size (N) and the segregation ratio (risk) being estimated. In Figure 2-1, the ratio of simplex to multiplex is 6 to 1 corresponding ioN=2 and a segregation ratio of 1 in 4. Special methods of analysis such as those of Hogben^'' or the sib method of Weinberg''^ can be used to give the correct segregation ratio when we are sure that ascertainment is complete. Note that complete ascertainment does not depend on finding all cases in the population (although that would do it) but rather on determining whether the various types of sibships exist in the correct proportions. More commonly, multiplex sibships tend to be over-represented with respect to simplex sibships and do not bear the same relationships to each other and the simplex sibships as they would if ascertainment was complete. In extreme cases, multiplex sibships are present in proportions dependent on the truncated binomial distribution and the number of affected persons in the sibship. Under these conditions, ascertainment is incomplete, and the propositus method of Weinberg^' or the sib method of Fisher^^ must be used. Usually, however, ascertainment is neither complete nor incomplete, lying somewhere between the two. Consequently, yet another genetic parameter must be introduced: the ascertainment probabiUty, defined as the chance that an affected person will become an index case. Under conditions of complete ascertainment this quantity is tending to 1 (certainty), and under incomplete ascertainment it is tending to 0, making multiplex sibships more likely to be ascertained because they contain more affected members. In this situation, methods such as Morton's should be used, but a body of data for colorectal cancer suitable for these powerful analytic techniques has prob¬ ably never been collected. If ascertainment was known to be incomplete, the estimation of the seg¬ regation ratio ([u + (У2) v]^) would be obtained by omitting the index case and finding the proportion affected in the remainder. However, the assump¬ tion about incomplete ascertainment implies that sibships with two affected members will be detected with twice their appropriate frequency in the pop¬ ulation, and sibships with three affected members will have three times their frequency. But this situation is known to occur in human genetics only rarely when affected siblings coexist in their affected state. As the cases of colo¬ rectal cancer in sibships are observed not to coexist, except in rare instances, it is difficult to imagine that anything approaching incomplete ascertain-