Does Local Failure Following Oligometastasis-Directed SBRT Predict for Widespread Progression and Overall Survival?

Abstract

<h3>Purpose/Objective(s)</h3> The potential of oligometastasis (OM)-directed therapy to translate to survival benefit has been shown in randomized trials, but typically in small or histology-specific cohorts (i.e., prostate or NSCLC). We investigated the impact of local control (LC) on widespread progression (WSP) and overall survival (OS) in a large, multi-institutional cohort of patients across heterogeneous histologies treated with SBRT to OMs. <h3>Materials/Methods</h3> Patients treated at 6 international centers from 2007 – 2016 were retrospectively reviewed and included if they had 1) five or fewer lesions, 2) no history of brain metastases, 3) curative treatment for primary disease, and 4) SBRT to all known OMs at initial presentation. The relationship between LC at certain time points and OS and WSP (>5 active/untreated lesions) were explored using Cox and Fine-Gray regression models. The relative strength of association between LC and dosimetric predictors was analyzed with competing risk regression using death as a competing risk. <h3>Results</h3> In total, 1033 patients treated to 1700 OMs were analyzed. The median follow up was 24.1 mo (range: 0.3-104.7). Most patients (57.1%) had solitary OM, 24.0% had 2 OMs, 10.3% had 3 OMs, and 8.6% had 4-5 OMs. The most common histologies were NSCLC (25.2%), colorectal (22.8%), prostate (12.8%), and breast (8.1%). The median prescription dose was 48 Gy (range: 10-60) in a median of 5 fractions (range: 1-15). Median OS was 44.2 mo. Cumulative incidence of local failure at 6m, 1y, 2y and 3y was 3.3%, 10.0%, 19.6% and 23.9% respectively. As detailed below, patients who fail locally in any or all SBRT-targeted OMs have significantly increased risk of WSP and death compared to those with no failure. These risks may be greater with shorter duration of LC. Patients who failed in only a subset of treated OMs and those who failed in all OMs did not have significantly different WSP and OS outcomes. Minimum BED to the GTV or ITV was most predictive of LC when compared to prescription dose, PTV Dmin, and PTV Dmax, across a range of simulated a/b ratios. <h3>Conclusion</h3> This large multi-institutional series demonstrates that duration of LC in SBRT-targeted OMs translates strongly to WSP and OS outcomes in an all or nothing manner, where failure in a subset of treated OMs appears no different in risk compared to failing in all OMs. This highlights the importance of maximizing LC to each OM in this treatment paradigm, where GTV or ITV Dmin may be a key dosimetric parameter.