Bone cancer osteosarcoma

OSTEOSARCOMA is one of the 3 main subtypes of bone cancer. It arises most often from the growing ends of long bones. Data from the Surveillance, Epidemiology, and End-Results (SEER) program of the National Cancer Institute indicate that osteosarcoma has a bimodal age distribution. (1,2) The first peak appears between the ages of 5 yr and 20 yr for both males and females, and a second peak is seen among males over the age of 65 yr. Males have a higher incidence (per 100,000) than females among both whites and blacks (i.e., 0.35 in white males and 0.37 in black males; 0.25 in white females and 0.29 in black females). (2)

The results of epidemiological studies have shown an association between exposure to high doses of radium and bone cancers? The exposures occurred among patients treated with radium for other cancers, and among radium-dial painters who adjusted the point on their brush with their lips. (3,4)

There is concern about the risk of osteosarcoma from low levels of radium in drinking water. According to data from the Wisconsin Department of Natural Resources (DNR), the highest levels of radium in drinking water in the state are found in water drawn from 2 rock formations: (1) the deep sandstone of the state's eastern quarter, and (2) the crystalline granite rock of north central Wisconsin. (5) Elevated radium levels typically occur in public water supplies. Generally, private wells are not drilled into the deeper geologic formations that contain higher concentrations of radium.

In 3 North American studies investigators assessed osteosarcoma risk in humans in relation to radium in drinking water. The Midwest Environmental Health Study was an early study that examined osteosarcoma deaths in communities with different levels of radium-226 in drinking water. (6) In that study, elevated levels of radium-226 were defined as levels exceeding 3.0 picocuries per liter (pCi/l; 1 pCi/l = 0.037 Bq/l). A total of 111 communities in Illinois and Iowa were identified as having elevated radium levels in their drinking water during the study period 1950-1962. The researchers reported that, compared with the control group, the exposed population had higher osteosarcoma mortality rates in 6 of the 9 age groups, and significantly higher rates in the 20-29 yr and 60-69 yr age groups.

Two more-recent reports have described epidemiologic studies of the association of radium in drinking water and the occurrence of osteosarcoma. Moss et al. conducted a case-control study in Wisconsin and compared 167 individuals with osteosarcoma diagnosed between 1979 and 1989. (7) The control group contained 989 patients diagnosed with other cancers during the same period as for the cases. The investigators reported a weak association that was not statistically significant (odds ratio [OR] = 1.5; 95% confidence interval [Cl] = 0.8, 2.8) between osteosarcoma and water-borne radium exposure. In that study, water-borne radium exposure was defined as residence in a county with a water system containing gross et-radiation of 9 pCi/l or higher. All residences in other counties were coded as low-radium residences.

Finkelstein and Kreiger conducted a case-control study, similar to the study by Moss et al., (7) of bone cancer in youths living in Ontario, Canada, and they also used patients with other cancers as controls. (8) Those investigators reported an association between the risk of osteosarcoma and birthplace "exposures" to even relatively low levels of radium in drinking water, but at a p = .10 level (OR = 1.77; 90% Cl = 1.03, 3.00). No significant association was found between osteosarcoma and lifetime exposure to radium in drinking water (OR = 1.33; 90% Cl = 0.76, 2.24).

We conducted a population-based case-control study in Wisconsin to examine the association between radium in drinking water and incidence of osteosarcoma.

Method

Study population

Cases. The case group comprised Wisconsin residents who had an initial diagnosis of osteosarcoma reported to the state Cancer Reporting System between 1980 and 1997.

Controls. For each case, we selected 10 controls from the general population of Wisconsin. The universe of potentially eligible controls was constructed from the 1980 and 1990 decennial federal census data on the basis of age, sex, and ZIP code of residence. For cases diagnosed between 1979 and 1984, we used the 1980 census data, and for the cases diagnosed between 1985 and 1997, we used the 1990 census data. The following method was used on each census. On the basis of age, sex, and number of persons residing in each ZIP code, as specified in each census, we generated a population database with 1 record per person in Wisconsin. Each population database had approximately 5 million records. Using Stata (9) statistical software, we extracted 10 random controls per case. This method guaranteed equal probability for every subject in the Wisconsin population.

ZIP codes refer to the place where mail is received, which, in most cases, is a residence. We were unable to use 1 case and 2 controls in our analyses because the subjects' ZIP codes did not match a ZIP code for which we had ecological (census-derived) data. One control's ZIP code was used for a Minnesota border community; the other control ZIP code and the case ZIP code were both post office box ZIP codes, which could not be clearly assigned to another ZIP code. The loss of these subjects resulted in a total of 319 cases and 3,198 controls.

Data sources

Radium data. The Wisconsin DNR requires periodic testing of public water supplies to screen for potentially elevated levels of radium. The level of gross [alpha]-radiation activity is used as a screening test for elevated radium levels. If the gross [alpha] level is less than 3 pCi/l, no testing for specific radium isotopes is required. If the gross [alpha] level is greater than or equal to 3.0, testing for radium isotopes (radium-226 and radium-228) is required, as is more frequent sampling.

A data set with Wisconsin public water utilities' gross [alpha], radium-226, and radium-228 levels from 13 September 1978 through 11 August 1999 was obtained from the Wisconsin DNR. The data set included sample location, ZIP code, sample date, analyte measured, measurement units, and level of analyte found. Subsequent use of the word "radium" in this article refers to the combined total of radium-226 and radium-228, unless otherwise noted. We used ZIP codes in this study to link radium level, residence of cases and controls, and ecological census data. To ensure that the correct radium level was assigned to each Wisconsin ZIP code, we conducted telephone surveys of the following public water utilities to obtain lists of the ZIP codes each utility serviced: (a) utilities servicing cities with populations greater than 50,000, (b) utilities with radium levels higher than 20 pCi/l, and (c) utilities located near ZIP codes that do not have their own utility service.

We used a federal government website--http://venus. census.gov/cdrom/lookup (19 January 2001)--to verify that selected ZIP code areas, that did not have public water utilities and were not serviced by nearby public utilities, were rural, and, therefore, most likely obtained their water from private wells. In these cases, negligible exposure to radium in drinking water was assumed. Investigative samples in areas of the water system not representative of water delivered to the public (e.g., backwash, test pumps) were excluded. Nonrepresentative samples were determined by a DNR official. Overall, 138 investigative samples from 35 water supplies were excluded.

Public water supplies can be classified as municipal or other than municipal. Other-than-municipal water supplies are those operated by apartment owners, condominium associations, subdivisions, neighborhood water trusts, mobile home parks, nursing homes, and health-care centers. Radium measurement records were coded as being from either municipal or other-than-municipal sources.

Cancer data. The Wisconsin Cancer Reporting System is a population-based registry that collects, manages, and analyzes cancer incidence and mortality data on Wisconsin residents. (10) Out-of-state cancer registries provide reports on Wisconsin residents diagnosed in their states to the Wisconsin registry under data exchange agreements with the Wisconsin Cancer Reporting System. Registry records are also matched annually to the Wisconsin resident death file so that cases not reported by the regular reporting process are identified. During the time period of our study, we estimated data completeness to range from 93% to 95%. (11,12) The Wisconsin Cancer Reporting System offers statewide training sessions to reporting facilities. The training focuses on increasing complete case finding, timely reporting, and compliance with the cancer reporting statute and administrative rule.