47XXY also known as Klinefelter’s Syndrome is a numerical chromosome variation, characterised by the presence of one or more extra X chromosomes. It occurs in about 1.5 per 1000 of the male population. The clinical syndrome was initially described in 1942, and the chromosome constitution was discovered in 1959. Characteristically,the patients have hypogonadism and elevated gonadotropin levels, and various other hormonal and physical abnormalities occur. There has been limited information about long-term mortality risks, however, because of the lack of large cohort (follow-up) studies. The only such published studies have been a cohort of 466 men from a Scottish register, later extended to two other centres with a total of 695 men, and a cohort of 781 men from Denmark. To enable more detailed analyses, based on much larger numbers, we assembled a cohort of cases of Klinefelter syndrome diagnosed in Britain for as long as records are held by the cytogenetics centres in the country and followed up the cohort for mortality, for periods of up to 40 years.
We extracted identification and diagnostic data about all patients with Klinefelter syndrome diagnosed as far back as records had been retained (1959 at earliest), from each cytogenetics laboratory in Britain except two small ones. Appropriate ethics committee agreement was obtained for this. Data for the comparatively rare 46,XX male variant of Klinefelter syndrome were not included in the study because these data have particular potential to include recording and transcription errors from normal males and females. Patients who were recorded as being karyotyped because of cancer were excluded from the study. Identification data about the cohort members were sent to the National Health Service Central Registers (NHSCRs) for England and Wales and for Scotland, which holds records of all NHS patients in their countries and are therefore virtually complete population registers. The registers hold data on deaths, emigrations, and other exits, and therefore the cohort members were flagged on these registers to obtain information on mortality and other losses to follow-up. We were sent death certificates for those who had died. These were coded to the underlying cause of death, using the International Classification of Diseases (ICD) revision employed in Britain at the time of death: ICD7 from 1958–1967, ICD8 from 1968–1978, ICD9 from 1979–1999 in Scotland and 1979–2000 in England and Wales, and ICD10 from 2000 in Scotland and from 2001 in England and Wales. The coding was largely undertaken by the national death coding office, and for the remainder by the authors following the national coding procedures. We then bridge-coded between ICD revisions to give the ICD9 categories shown in the tables. We made an exception for deaths coded under ICD rules to Klinefelter syndrome, which we recorded to the underlying cause that would have applied if Klinefelter had not been written on the certificate; this was done because the purpose of the study was to compare causes of death in Klinefelter patients with the general population, rather than to discover how often certifiers believed Klinefelter to be the cause of other fatal diseases.
For each man in the cohort, we computed person-years of follow-up by 5-yr age group, calendar year, and country (England and Wales vs. Scotland), beginning from the date of cytogenetic diagnosis and ending at June 30, 2003, or the 85th birthday, date of death, or other loss to follow-up, whichever was earliest. Follow-up was censored at age 85 because at older ages than this, national (i.e. expected) mortality rates are not available by 5-yr age group, and the certified cause of death is often inaccurate. We calculated expected cause-specific mortality in the cohort by multiplying the age-, calendar years, and country-specific person-years at risk in the cohort by the corresponding national mortality rates for men. Standardized mortality ratios (SMRs) were then calculated as the ratio of observed to expected deaths, and 95% confidence intervals (CIs) for the SMRs were calculated assuming a Poisson distribution (8). Tests for trend and for the difference between SMRs were conducted as described by Breslow and Day (8). Significance tests were two-sided. Absolute excess risks were calculated by subtracting the expected from the observed numbers of deaths and dividing by person-years at risk.
We subdivided the subjects for analysis by the number of sex chromosomes, whether mosaicism was present, and if so, the constitution of the non-Klinefelter component. Where information was available for mosaics on the numbers of cells diagnosed with each mosaic component, we designated the subject as mosaic only if more than one cell had been counted with each component. We did not have direct information for the study subjects on whether mosaicism was congenital or acquired, but as a rough proxy for this [because the prevalence of acquired mosaicism rises with age (9)], we conducted separate analyses for mosaics diagnosed before age 45 yr and those diagnosed at older ages.
To assess, as far as possible, whether bias might account for certain of the results, we conducted several subanalyses of risks in subdivisions by birth year, risks omitting follow-up and deaths in the early years after cytogenetic diagnosis, and risks omitting cohort members recorded by the Medical Research Council Human Genetics Unit.