Significant Recent Declines In Adult Leukemia Incidence Rates In the United States

Abstract 873

Each year in the United States, approximately 45,000 individuals are newly diagnosed with leukemia and 22,000 will die of the disease. Due to this poor survival, leukemia ranks fifth in person years of life lost among specific cancers. Little is known about causes, although exposure to solvents, radiation, pesticides and, to a modest extent, cigarette smoke has been implicated for some subtypes. The last comprehensive report of leukemia trends covered the period 1973–1998 [Xie Y et al, Cancer 2003]. Evaluation of recent leukemia incidence trends could provide important new etiologic insights.

Using Surveillance, Epidemiology and End Results (SEER) Program data, we analyzed leukemia incidence trends in U.S. adults (≥ 20 years of age) by age, leukemia subtype (acute myeloid (AML), acute lymphoid (ALL), chronic myeloid (CML), chronic lymphoid (CLL)) sex, race, and ethnicity for the period 1987–2007. Frequencies, age-adjusted incidence rates (IR, per million), and trends were calculated along with annual percent change (APC) and corresponding 95% confidence intervals (CI). Joinpoint analyses were used to detect any significant directional changes in IRs over the period.

Of 43,970 newly diagnosed cases identified, IRs increased with age and were consistently higher in males than females for all four subtypes. The highest IRs occurred for CLL (54.4), followed by AML (38.3), CML (20.6) and ALL (7.0). With regard to trends, IRs for CLL (APC -0.5; CI: -0.9, -0.1) and CML (APC -1.2; CI: -1.6, -0.8) declined over the time period; declines were observed in males and females, and by race and ethnicity. Male(M):Female(F) IR ratios remained relatively constant at approximately 2.0 and 1.7, respectively. For ALL, IRs decreased in males (APC -0.9; CI: -1.9, 0.2) but slightly increased in females (APC 0.4; CI: -1.0, 1.7), which was most notable in Hispanics (APC 4.0; CI: 1.2, 6.8). In contrast to CML and CLL, the overall M:F rate ratio for ALL decreased, although it did not reach statistical significance (p=0.08). For AML, IRs increased significantly for males (APC 1.0; CI: 0.3,1.6) and females (APC 1.7; CI: 0.7, 2.7) from 1987–2000 and 1987–2001, respectively. However, since then, AML IRs for males have been significantly decreasing by 4.2% per year (CI: -6.4, -2.1), while IRs for females have been decreasing by 1.6% per year (CI: -4.1, 0.9). Across the entire time period 1987–2007, there was a statistically significant negative trend (p=0.002) in the M:F IR ratio for AML.

Decreasing IRs across many leukemias since 1987 are unlikely to reflect changes in screening or diagnostic coding practices. Instead, these observations may reflect temporal changes in etiologically relevant environmental exposures. Of note, the prevalence of cigarette smoking in the population has decreased and occupational safety practices (e.g., reducing solvent/radiation/pesticide exposure) have improved over the last several decades, which could contribute to the gradual decreases in some IRs observed. In contrast, the rapid and significant decrease noted for AML since 2000, especially following a significant increase, was striking and deserved additional scrutiny. We further consulted with our cancer registry colleagues to determine whether the introduction of myelodysplastic syndrome (MDS) as a new malignancy in SEER in 2001 could be influencing recent AML trends given the (apparently) coincidental overlap in time periods. Of note, approximately one third of MDS patients subsequently develop AML. We learned that AML following an MDS diagnosis from 2001–2009 was not reportable to SEER and therefore not counted. We are not aware that this has been documented in the literature. However, beginning for 2010 diagnoses, SEER changed this practice such that AML following MDS will be captured as a second malignancy. Based on these changes in AML surveillance, it will especially be important to monitor future trends for this malignancy. Overall, this study demonstrates the value of in-depth analyses of SEER cancer IRs and trends; analyses may reveal patterns of clinical and/or etiological importance, or, in the instance of AML, unpublished coding rule changes.