Oligometastatic Tumor Size as a Predictor of Distant Failure in Synchronous Oligometastatic NSCLC

Abstract

<h3>Purpose/Objective(s)</h3> Oligometastatic disease (OMD) is typically defined by a threshold number of metastases. Local ablative treatment of oligometastases is intended in part to prevent the development of new metastases elsewhere. It is unclear whether the size of metastases is associated with the risk of developing new metastasis. We hypothesized that in non-small cell lung cancer (NSCLC) treated with complete local therapy of synchronous OMD, size of total synchronous oligometastatic disease burden is a predictor of freedom from new metastasis (FFNM), independent from the total number of metastases. <h3>Materials/Methods</h3> Retrospective review was undertaken for metastatic NSCLC patients treated at a single cancer center from 2010-2019. Patients were included if they presented with synchronous OMD, defined as 1-5 distant metastases at time of diagnosis, and underwent definitive local therapy to all sites of disease. FFNM was calculated using the Kaplan-Meier method. The diameter of each metastasis was extracted from radiology reports, and used to calculate the sum of the diameters of all metastases (DSum). As an alternative metric of size, volume of each metastasis was estimated from its diameter by approximation as a sphere. Patients with incomplete data were excluded from analysis. Variables that were associated with FFNM on univariate analyses were included in multivariable Cox regression analyses. <h3>Results</h3> Overall, 84 NSCLC patients with synchronous OMD underwent local therapy to all disease sites, whether radiation, surgery, or both. Of these patients, 79 had complete data and were included in analyses. Number of metastases at treatment varied from 1-4 (1 met n=51, 2 mets n=15, 3 mets n=8, 4 mets n=5). DSum varied from 0.4 cm to 10.8 cm (IQR 1.5 cm – 3.4 cm). Univariate analyses showed FFNM was most strongly associated with DSum, number of metastases, N2-3 stage, and brain involvement, whereas FFNM was not associated with estimated total volume of metastatic disease. A multivariable Cox regression model showed a non-statistically significant trend towards an association between DSum and FFNM (HR 1.13, 95% CI 0.99-1.29, p=0.07). There was an association between brain involvement and FFNM (HR 2.00, 95% CI 1.03-3.86, p=0.04), but there was no association noted between number of metastases (p=0.47) or N2-3 stage (p=0.21) with FFNM. <h3>Conclusion</h3> The cumulative size of NSCLC oligometastases (as assessed by summed diameters) but not number of metastases trended towards an association with new metastasis. This radiographic data is easily retrievable from the medical record. Future studies of oligometastatic treatment may benefit by investigating total tumor size in addition to number of lesions.