Cancer mortality in relatives of desmoid sarcoma patients

Summary Desmoid sarcoma patients from the U.S. and Canada, diagnosed under the age o! 16 years and referred to the University of Texas, M.D. Anderson Hospital - MDAH - between 1944 and 1975 were surveyed. Family information was collected for grandparents, parents, parental siblings, proband offspring and siblirtgs. Expected mortality was calculated by applying age-race-sex specific U.S. mortality rates to the person-years at risk. Standardized mortality ratio's - SMR-were calculated. Among 429 relatives o! 26 desmoid sarcoma probands no overali cancer excess was found, observed/expected = 13/247. The risk among parents, the ones more informative for genetic analysis were also not significant, obsetved/expected = 2/1.83. An updated follow-up of incidence data of the study population and of the general population is suggested in the years to come, and in doing so, the inclusion of all familly members history should pull the results to a number closer to rea lity.

the deveiopment of DT [27]. There is only one report suggesting that an inherited defect in growth regulation of the connective tissue is the underlying cause of DT [18]. References of histologically similar tumors or aggressive fibromatosis however, support the concept of a hereditary inherited tendency to fibrobiast proliferation [3, 5, 14, 15, 19, 37, 51, 57-591. Furthermore, studies of soft tissue sarcomas, a category which includes DT, have given reason to suspect such a relation [1,8,26,30,38,40,41,43,46,48,52,56,59]. if an inherited or mutant gene is involved in the etiology of malignant neopiasm and particuiarly of soft tissue sarcomas, a clustering pattern should be observed within families. We wished to test the association between DT and cancer mortality among amily members of DT cases.

Materiais and methods
Ali DT cases referred to the MDAH from 1944 to 1975 under 16 years of age at the time of onset were compiled. Patients from outside the United States and Canada, and adopted patients were not included. Twenty-six eligible cases were identified.
Ali parents of eligible patients were contacted by ietter and asked to participate in an interview and provide information on lhe current health status of lhe patient family members. The questionnaire included date of birth, death and occurrence of tumors for lhe proband and his/her párents, siblings, half-siblings, aunts, uncles, grandparents and offspring. Cousins and other relatives were excluded due to lhe difficulty in ascertaining this information on ali lhe families involved in this study. Relatives whose country of origin and permanent residence were outside lhe U.S. and Canada were excluded from lhe analysis.
The underlying cause of death was coded according to lhe Eighth Revision of lhe international Classification of Diseases -iCD-8- [34]. When a cancer death was reported in any of lhe relatives, death certificates along with existing medical records were obtained. Faiiure to verify lhe cause of death caused lhe reported cancer to be dropped from lhe analysis.
Person-years at risk were determined from lhe entry date -1944 -to lhe date of death or date of study termination -1980. Age-sex-year-specific U.S. cancer mortality rates for lhe period 1925-1979 were applied to generate lhe expected cancer death rates for each DT relative. Mortality rates from 1925 were used to compute lhe expected deaths for those entering lhe study prior to 1925. Those rates were considered lhe best available cancer data for that period. The SMR then, is lhe ratio of lhe observed number of deaths to lhe expected number of deaths. The 95% confidence intervais -Cifor lhe risk were determined by lhe assumption of a Poisson distribution for lhe numerator observed number of deathsand a constant denominatorexpected number of deaths [6, 321. We computed lhe attained significance leveis -P-valuesfor lhe risks by taking lhe expected number of deaths as lhe Poisson parameter and assessing lhe probability of lhe Poisson distribution tail area defined by lhe observed number of deaths [32]. We computed tests of significance between risks by using normal distribution approximations considering lhe relative risk as a Poisson variate divided by a constant.

Results
A significant deficit of overall cancer mortality was observed, O/E = 13/24.7. Olher studies have reported similar deficits of deaths in lhe overail run [41]. No deficit of cancer deaths among parents of lhe cases however, was observed, O/E = 2/1/83 95% Cl = 0.12 -3.92.
We had a total number of 53 reported cancers among DT relatives of which 19 were non-confirmed cancers. Fourteen referred to cancer deaths and 5 referred to cases of cancer. Of lhe 14 non-confirmed cancer deaths 10 referred to cancers occurring in cousins and other relatives.
Of lhe 35 cancer deaths reported, 25 occurred in first or second degree relatives, those relatives for which information was collected syslematicaliy for every kindred. Of those 4, not confirmed deaths and one cancer death of a mother half-sibiing were excluded from lhe analysis making 20 eligible dealhs. From those 20 elegible cancer deaths, oniy 13 were considered in this study and refer to lhe cancer deaths that had death certiticates stating cancer as cause of lhe death. The decisions taken here might seem unreasonable for that we perpetuate lhe innacurracies in lhe report of cause of death in death certificates, despite lhe fact that we have availabie lhe pathology reports of lhe seven cases lett out. Those innacuracies however, are thought to be also present in lhe general population rates used to generate our expected rates.
The most frequenl cancer site among DT relatives was cancer of lhe lung, five cases or deaths, followed by three cases or deaths of stomach cancer, three 01 cancer of lhe prostate, two of lhe bladder, one of cancer of lhe sigmoid, one lhe rectum, one of lhe cecum, one of lhe pancreas among other sites. Six families had no cancer history and 18 first une relatives have had benign fibromatosis reported.

Discussion
Since three was no excess cancer mortaiity we did not pursue lhe anaiysis any further and could say that there is no indication in this data set of a genetic causation for lhe desmoid tumors.
The sample size -26 kindredsmay be too smali to have included lhe congenital form of DT, which is, genetically very interesting. The extreme rarily of desmoid tumor mostly contributed for this, given that only 26 cases were accumulated in a fairly large period of lime, 31 years. Genetic mutations on lhe other hand, are rarely detected even in eslablished known hereditary tumors. An excess of cancer risk wouid lhen be expected only in a few kindreds. Therefore, to increase lhe chance of detecting a cancer risk, additional kindreds is needed and highly recommended. The fact that not ali reported deaths could be verified plus lhe exclusion of deaths of cousins and other distant relatives, added wilh lhe above comments, may have influenced lhe deflated resuits seen here. Finaily, lhe need for a more detailed family history such as information about polyposis coli, palmar and plantar fibromatosis and thickening of lhe tendons among family members of DT cases and lhe frequency of those neopiastic processes in lhe general population, appear to be first priority in future research.