Twinning causative origin investigated by Sartwell's biometrical method

Sartwell's biometrical model is useful to test potential causal factors at the origin of biological phenomena of unknown etiology. It postulates a lognormal distribution of the incubation time. The method has been applied in the past to infectious and neoplastic diseases and, only lately, to genetic diseases. We report here an application to unlike-sex and like-sex twins in an attempt to infer causal origin from a family of potentially related factors. Two hypotheses are tested: the maternal prenatal origin of unlike-sex twins, and the menopausal basis of like-sex twins. The hypotheses are respectively grounded in the presumed genetic origin of DZ twinning, and in the ovarian-dysfunction origin of MZ twinning. The study is population-based; it proceeds from a matched control group and considers various time windows (induction periods) of possible etiologic significance-namely, age at onset of twins, the time window from the first birth to the twin birth, and the time window from the twin birth to the last birth. The study of lognormality has been carried out with the Box-Cox transformations, Fisher's cumulants method, and the normal probability plots. Results show clearly that unlike-sex (presumably DZ) twins have their causal origin in the maternal prenatal period as age of onset is definitely lognormal. Median induction time is 31 ± 6 years. This is interpreted as a multiplicative age-related causal process involving hypophyseal hormones. The menopausal origin of like-sex twins, a mixture of MZ and DZ, is less clear-cut, as the fit of the distribution of the time window until last birth to a lognormal model is less than perfect. In spite of the heterogeneity of the like-sex twin group, there is evidence that some cases of MZ twins may be menopause-related. But this complex question needs more study.