Marginal Failures with Stereotactic Body Radiation Therapy(SBRT) for Early Stage Lung Cancer: Room for Improvement?

Abstract

SBRT is an effective treatment for early stage non-small cell lung cancer (NSCLC). Prior series distinguish local from regional/distant recurrences, but detailed patterns of local failure are not well-described. We evaluate the location of local failures to provide further insight for treatment planning. Patients treated with SBRT for T1-2N0 NSCLC from 2007-2016 at a single institution were included. Prescribed dose was 48-54Gy in 3 to 5 fractions and planning target volume (PTV) margins were generally 5mm radially and 8mm superior/inferior. Site of first failure was classified as local, regional (hilar/mediastinal lymph nodes) or metastatic. For patients with a component of local failure, post-treatment diagnostic scans at time of recurrence were fused to the initial planning scan using rigid registration to lung features proximal to recurrence. The recurrence volume was delineated and compared to the treatment PTV. For each local failure, percent overlap of the PTV and recurrence volume was calculated. Distance from centroid (geometric center) of recurrence to PTV edge was measured in 3 planes and average distance was calculated. Local failures were categorized as: in-field failure (≥80% overlap or centroid >1cm inside PTV), marginal failure (20-80% overlap and centroid <1 cm inside/outside PTV), or involved lobe failure (overlap <20% or centroid >1cm outside PTV). Parametric statistics compared characteristics of recurrences. A total of 217 patients were treated with robotic radiosurgery (94%) or linac-based (6%) SBRT. 83% were T1 tumors, 69% upper lobe tumors, and 74% biopsy-proven malignancy. With a median follow-up of 20 months (range 0-98), 38% had progression of disease. 16% (N=34) had a component of local progression at time of first failure: 18 local only, 8 locoregional, 8 metastatic. 60% were pathologically confirmed. Of the local failures, 9 were classified as in-field, 15 marginal and 10 involved lobe. For in-field, marginal, and involved lobe failures there were differences in mean PTV/recurrence overlap (93% vs 43% vs 2%) and distance from centroid to PTV margin (1.39cm vs 0.56cm vs 1.78cm). For all local failures, PTV D95 was at least prescription dose. Marginal failures developed earlier than other local failures (mean 15 vs 25 months, p<0.05). Upper and lower lobe tumors had similar rates of local failure (16% vs 18%); however, 8 of the 11 local failures seen in lower lobe tumors were marginal failures compared to only 7 of 23 local failures in upper lobe tumors. Local progression is seen in roughly 16% of patients following SBRT. Marginal failures represent nearly half of these events, and appear to be of particular concern for lower lobe tumors. These findings highlight an opportunity to further evaluate margin size, motion management, and accuracy of dose calculations at the PTV margin to further improve local control in lung SBRT.