What is the predictive value of RECIST criteria following stereotactic lung radiation?

• During long-term follow-up, two-thirds of patients who underwent lung SABR will meet criteria for progressive disease by RECIST. • Less than 20% of these patients who meet RECIST criteria for progressive disease will have local recurrence and, therefore, the positive predictive value of RECIST in this setting is less than 20%. • These results provide context for clinicians following patients after lung SABR and advise caution in interpreting studies that use PFS as an endpoint after lung SABR. Response Evaluation Criteria in Solid Tumors (RECIST) is commonly used to assess response to anti-cancer therapies. However, its application after lung stereotactic ablative radiotherapy (SABR) is complicated by radiation-induced lung changes. This study assesses the frequency of progressive disease (PD) by RECIST following lung SABR and correlates this with actual treatment outcomes as determined by longitudinal follow-up. We reviewed patients treated with lung SABR for primary lung tumors or oligometastases between 2010 and 2015. Patients were treated with SABR doses of 54–60 Gy in 3–8 fractions. All follow-up scans were assessed and the treated lesion was serially measured over time, with the maximum diameter on axial CT slices used for RECIST calculations. Lesions demonstrating PD by RECIST criteria were identified and subsequently followed for long-term outcomes. The final 'gold-standard' assessment of response was based on size changes after PD and, as available, positron emission tomography scan and/or biopsy. Eighty-eight lesions met inclusion criteria. Seventy-five were lung primaries and thirteen were lung metastases. Median follow-up was 52 months (interquartile range: 33–68). Two-thirds (66 %, 58/88) of treated lesions met RECIST criteria for PD; however, local recurrence was only confirmed in 16 % (9/58) of cases. Most lesions that triggered PD by RECIST (47/58, 81 %) were ultimately found not to represent recurrence, while a minority (2/58, 3 %) had an uncertain response. The positive predictive value [PPV] of a RECIST defined PD event was 0.16. If PD was triggered within 12-months post-treatment, PPV was 0.08, compared to 0.21 for lesions triggering PD after 12-months. Using RECIST criteria, two-thirds of patients treated with lung SABR met criteria for PD. However, only a minority had recurrence, leading to a poor PPV of RECIST. This highlights the limitations of RECIST in this setting and provides context for physicians when interpreting post-lung SABR imaging. [ABSTRACT FROM AUTHOR]