Association of population mixing and incidence of acute lymphocytic leukemia in children and young adults in the State of Texas

Although accelerated population mixing (PM) has been associated with increased incidence of acute lymphocytic leukemia (ALL), the most common cancer in children, the mechanisms underlying this association are not known. To determine whether genetic mixing or environmentally transmitted factors are responsible for the association of PM and ALL, we conducted an ecological study of incidence of ALL in children & young adults (0–21 years old) among 253 of 254 counties in the State of Texas (USA) and with leukemia and surrogate indicators of genetic and environmental population mixing. The study used ALL incidence data from the Texas Cancer Registry for 2000 and 2009, and county population statistics from the US Census Bureau for 2000 and 2010. Surrogates of genetic mixing included proportion of multiracial households, ratio of Hispanics to non-Hispanics, and racial diversity index. Surrogates for environmental population mixing included total population density, ratio of foreign to native-born residents, and proportion of households with ≥5 persons. We assessed confounding or interaction with median income of households in the county (as a marker for socio-economic status) and with proportion of urban population in the county. Poisson multivariable regression was used to compare incidence rates of ALL with categorized PM indicators. The regression model included time change, the raw PM variable, and a PM-by-time interaction term. Incidence rate ratios, 95% confidence intervals, and p-values were calculated. We found a widely variable degree of on-going population mixing among the counties of Texas. A higher proportion of multiracial households and higher ratio of Hispanics to non-Hispanics in a county were associated with 52%–81% higher incidence rate of ALL after adjusting for time change. This association was not confounded or modified by median income. Interestingly, the “risk” of ALL associated with multiracial households waned over time—suggesting that the risk of ALL is affected by dynamic population mixing rather than static genetic heterogeneity. None of the surrogates of environmental mixing were associated with incidence of ALL. There was no association of ALL incidence with urbanization. These results are consistent with our hypothesis that the leukemogenic agent implicated in population mixing is probably genetically transmitted rather than an exogenous infection. The results also explain why cosmopolitan urban populations that do not have significant ongoing PM may not have higher ALL incidence. To follow up on these findings, we are currently testing whether transmission of endogenous retroviruses may account for PM and ALL incidence in children