Estimating Therapy-Related Myeloid Neoplasm Risks in the US

Background: Therapy-related myeloid neoplasms (t-MN) have traditionally been a dreaded but rare complication arising from the treatment of other cancers. With rising numbers of long-term cancer survivors, there is growing concern for increasing rates of t-MN. We provide estimates of the relative risks of developing these secondary cancers after various 1^st cancers.

Methods: We used the Surveillance Epidemiology and End Results (SEER) data 1973-2011 and the R package SEERaBomb to examine 33 primary cancer risks of subsequent MN defined as acute myeloid leukemia (AML), chronic myeloid leukemia (CML), chronic myelomonocytic leukemia (CMML), or myelodysplastic syndromes (MDS). First time cancer survivor person-years at risk (PY) for developing a second cancer were computed based on age-at-diagnosis of the first cancer, survival time, and age at diagnosis of any second cancer. Individual PY age intervals were stacked using R objects of class GRanges and the function coverage() of the Bioconductor package GenomicRanges. This yielded PY that we multiplied into age specific incidence rates computed using all t-MN regardless of being 1^st, 2^nd or later cancers. This yielded cases expected under a null hypothesis of t-MN being independent of prior cancers. Relative risks (RR) were then computed as observed/expected (O/E) cases and ordered by lower limit ([O - 2sqrt(O)]/E) distances from 1. PY and t-MN were restricted to those arriving after 1/1/2006 to avoid MDS incidence transients that arose immediately after its introduction into SEER in 2001.

Results: 2,361 white females (Table 1), 3,415 white males (Table 2), 239 African American (AA) females (Table 3), and 309 AA males (Table 4), had t-MN diagnosed after 1/1/2006 that were known to have occurred at least 1 year after the 1^st cancer. In these tables/results: RR after non-Hodgkin lymphoma (NHL), Hodgkin's disease (HD), and multiple myeloma (MM) was high, save MM in AA males; high risks after acute lymphocytic leukemia (ALL) and brain cancer were not significant for 1st cancers diagnosed >15 years of age, so these are due to childhood 1^st cancers; and high risks of 2^nd MN after a 1^st are largely due to progressions being classified as 2^nd cancers.

Conclusion: t-MN with histories of NHL, HD, and MM may be enriched for truly treatment-induced cases. Mechanistic studies of t-MN should focus on such patients.

*RR ~95% confidence intervals are given in parentheses