Projected Effects of Radiation-Induced Cancers on Life Expectancy in Patients Undergoing CT Surveillance for Limited-Stage Hodgkin Lymphoma: A Markov Model.

OBJECTIVE

Patients with limited-stage Hodgkin lymphoma (HL) undergo frequent posttreatment surveillance CT examinations, raising concerns about the cumulative magnitude of radiation exposure. The purpose of this study was to project radiation-induced cancer risks relative to competing risks of HL and account for the differential timing of each.

MATERIALS AND METHODS

We adapted a previously developed Markov model to project lifetime mortality risks and life expectancy losses due to HL versus radiation-induced cancers in HL patients undergoing surveillance CT. In the base case, we modeled 35-year-old men and women undergoing seven CT examinations of the chest, abdomen, and pelvis over 5 years. Radiation-induced cancer risks and deaths for 17 organ systems were modeled using an organ-specific approach, accounting for specific anatomy exposed at CT. Cohorts of 20-, 50-, and 65-year-old men and women were evaluated in secondary analyses. Markov chain Monte Carlo methods were used to estimate the uncertainty of radiation risk projections.

RESULTS

For 35-year-old adults, we projected 3324/100,000 (men) and 3345/100,000 (women) deaths from recurrent lymphoma and 245/100,000 (men, 95% uncertainty interval [UI]: 121-369) and 317/100,000 (women, 95% UI: 202-432) radiation-induced cancer deaths. Discrepancies in life expectancy losses between HL (428 days in men, 482 days in women) and radiation-induced cancers (11.6 days in men, [95% UI: 5.7-17.5], 15.6 days in women [95% UI: 9.8-21.4]) were proportionately greater because of the delayed timing of radiation-induced cancers relative to recurrent HL. Deaths and life expectancy losses from radiation-induced cancers were highest in the youngest cohorts.

CONCLUSION

Given the low rate of radiation-induced cancer deaths associated with CT surveillance, modest CT benefits would justify its use in patients with limited-stage HL.