KEY POINTS
- Investigators retrospectively analyzed 44 patients with stage I lung cancer who had at least 10 years of follow-up after SBRT delivered between 2001 and 2014. Median clinical follow-up was 143 months, or nearly 12 years.
- Most tumors were peripheral (89%), median diameter was 22 mm, and 95% of patients had been considered medically operable. The most common schedules were 48 Gy/4 fractions (43%) and 50 Gy/4 fractions (34%).
- 39/44 patients (89%) remained free of grade ≥3 toxicity, but several unusual adverse events became clinically significant only after a decade — beyond the time horizon of most SBRT studies.
- Progressive bronchial stenosis and obstruction led in one patient to recurrent infection, atelectasis, abscess and ultimately grade 5 empyema 13 years after treatment. Another patient developed an enlarging FDG-avid postobstructive inflammatory lesion that initially mimicked recurrence.
- Among 28 tumors within 10 mm of the chest wall, 8 developed progressive mass-like chest-wall abnormalities associated with rib fracture, edema and calcification. One lesion continued enlarging for 13 years and was surgically removed because of concern for radiation-induced sarcoma, but pathology showed no malignancy.
- Two patients with tumors adjacent to cardiac structures developed grade 3 late pericardial abnormalities, including progressive pericardial thickening and cardiac dysfunction more than a decade after SBRT.
- Four local failures occurred at 41, 69, 73 and 89 months; three were therefore more than five years after treatment, but no local recurrence occurred beyond 10 years. Five second primary lung cancers and five extrapulmonary malignancies occurred, with no radiation-induced malignancy identified in the treated field.
CLINICAL TAKEAWAY
For long-term survivors after lung SBRT, five years may not capture the full toxicity story. Rare airway, chest-wall and cardiac changes can continue evolving for more than a decade and can mimic recurrent cancer, but this highly selected 44-patient survivor cohort cannot be used to estimate how common these events are in the overall SBRT population.