Editor—Incidence of thyroid carcinoma in Iceland is high1 and papillary carcinoma comprises 80% of all thyroid malignancies in the country.2 It is well known that medullary thyroid carcinoma aggregates in families and the genetic component in the aetiology has been established.3 Familial occurrence of non-medullary thyroid carcinoma is rare, but it has been increasingly recognised in recent years.4-6 There has been increasing awareness that this may constitute a distinct disease subgroup rather than chance occurrence, as reflected by the review of case series by Loh.7 Familial papillary microcarcinoma has also been suggested as a new clinical entity.8 9 In an earlier paper, we reported on the occurrence of papillary thyroid carcinoma (373 probands) in first degree relatives of all patients diagnosed between 1955 and 1984 in Iceland.10 This study showed a non-significant trend to familiality. In the present study, the population includes second and third degree relatives in addition to first degree relatives and we now include all non-medullary thyroid carcinoma patients.

Information on thyroid cancer cases was obtained from the Icelandic Cancer Registry. All cases (n=712) of non-medullary thyroid carcinoma diagnosed in Iceland between 1955 and 1994 form the basis of this study. This includes 147 thyroid cancers found incidentally at necropsy and 565 thyroid cancers diagnosed clinically. The information on first, second, and third degree relatives was obtained from the Genetical Committee of the University of Iceland. This committee has a computerised family tracing resource based on the 1910 census and birth records for 1840-1910. This resource is linked with the national register and death records for later periods. Permission was obtained from the Data Protection Commission to link files from the Cancer Registry with records from the Genetical Committee. The 712 families contributed to the man year calculations on which expected numbers are based. For nine probands of foreign origin, there were no relatives in the study population.

All cases were histologically verified. The histopathology was re-examined by one of the authors (JGJ). Included in the file were all differentiated thyroid cancers. These include papillary and follicular thyroid cancers, as well as anaplastic cancers, as these tumours are derived from the follicular epithelial cells. Thyroid carcinomas showing Hürtle cell differentiation were classified either as follicular or papillary histological type. Medullary cancers and lymphomas of the thyroid were excluded from the study.

The study risk period started in January 1955 or at birth (for persons born after 1 January 1955) and ended on 31 December 1997 or at death if death occurred earlier. Incidence rates of thyroid cancer stratified according to gender, five year age intervals, and five year calendar intervals from 1955 to 1997 were used. Incidence rates of thyroid cancer before 1955 were not available.

Years at risk stratified in the same manner for each degree of relatedness were calculated using the program PERSON-YEARS (PYRS) from the World Health Organization.11 Expected numbers of cases were calculated as the sums of products of incidence and years at risk. Relative risk (RR) was calculated as the ratio between observed and expected numbers. Pooled relative risk was calculated as the weighted sum of relative risk over the degrees of relatedness 1, 2, and 3 with the weights 1, 1/2, and 1/4, respectively; p values less than 0.05 were considered significant.

Table 1 summarises the study population. Altogether, there were 57 391 relatives in the pedigrees. Of these, 49 949 (25 396 males and 24 553 females) were alive on or after 1 January 1955, thus contributing to the man year calculations. Table 1 describes the distribution of probands and relatives. There were 712 thyroid cancer cases as probands. Among relatives, thyroid cancer occurred 249 times.

Table 2 classifies the probands according to family history, histology, and sex. There were 184 probands with one or more relatives affected. In six pedigrees, three affected relatives were found. There is only a small difference in histology between familial and sporadic cases. Papillary cancer is somewhat more common in the familial group (81%) than in the sporadic group (76%) in males. This difference was not found among females.

Table 3A shows the relative risk of thyroid cancer for relatives of all thyroid cancer patients. Figures are shown separately for first, second, and third degree relatives and for sex, as well as for all relatives pooled. For all probands, the risk ratio was 3.83 for male relatives and 2.08 for female relatives. For first degree relatives only, the respective risk ratios were 4.10 and 1.93. All these risk ratios are significant. In table 3B and C, the risk ratios for male and female probands are shown separately. For families of male probands, the pooled risk ratio was 6.52 for male relatives and 2.55 for female relatives. For families of female probands, the pooled risk ratio was 2.92 for male relatives and 2.02 for females. All these ratios are higher for male relatives than for female relatives and highest for male relatives of male probands.

Calculations were done separately on the 565 thyroid cancers diagnosed clinically (results not shown). Relatives of these patients (150 males and 415 females) also had a significantly increased risk of developing thyroid cancer, although the risks were somewhat lower than for the total cohort. This shows that relatives of patients with non-medullary thyroid cancer are at significantly increased risk of developing thyroid cancer. The risk is greater in first degree relatives in each sex compared to second and third degree relatives, as shown in table 3. The risk is highest in male relatives of male probands (6.52) and lowest in female relatives of female probands (2.02). As shown in table1, the disease is more than two times more common in females than males. This finding of greater excess risk of males related to males supports the suggestion that the greater risk in females in the general population in based on either female specific risk factors or greater susceptibility of females to external causes of non-medullary thyroid cancer. Thus, in males, a greater proportion of cancer risk is the result of familial factors than in females.

This study covers the whole Icelandic population, which is relatively small with only around a quarter of a million inhabitants. Because the nation has been rather isolated for centuries, people may be more interrelated than in other countries, which could result in a higher risk of familiality. In a review of case series,7 the previous study from Iceland10 showed the highest prevalence among 15 case series. There the prevalence was defined as the number of patients with a family history over the total number of patients with non-medullary thyroid cancers in the respective series.

The aetiology of thyroid cancer is mostly unknown with the exception of radiation. In a previous investigation concerning the effect of radiation used for benign conditions in children, we followed a cohort irradiated between 1930 and 1950. We found three cases of thyroid cancer, none of whom had a relative with thyroid cancer.12Therefore, radiation has not contributed to excess familial risk in our study.

Microcarcinomas of the thyroid gland are mostly of papillary histological type and have been thought to be a low risk category of tumour.13 Recently, reports have been published showing familiality of non-medullary thyroid microcarcinomas.8 9Our study shows that when taking into account cases found incidentally at necropsy (which are mainly papillary microcarcinomas), we see a stronger risk for familiality. A block from the thyroid gland is taken routinely at all necropsies in Iceland and microscopically evaluated.

In a study on forensic necropsies specifically looking for incidental carcinoma of the thyroid gland in Iceland, the frequency was 7.5% in males and 5.1% in females.14 This male to female ratio of incidental tumours is in sharp contrast to the male to female ratio of clinically detected cases. The finding that incidentally found papillary thyroid carcinomas have stronger familiality than clinically diagnosed cases may explain this observation.

Further studies are needed on the molecular genetics of non-medullary thyroid cancer, as well as on the prevalence of neoplasms at other sites and survival of familial cases.