The variance of lymphatic vascular invasion (LVI) diagnosis and its impact on overall survival(OS) independent of node stage in patients undergoing surgical resection on NSCLC.

Abstract

e21060 Background: The criteria for diagnosis of lymphatic vascular invasion(LVI) have not been standardized. Our investigation uses the National Cancer Database(NCDB) to assess the impact of this factor on survival(OS) and whether there are differences in the diagnosis of LVI based upon institution type and regional location in patients undergoing definitive resection. Methods: The NCDB was queried from the years 2010-2014 to find a patient population who underwent (bi)lobectomy with at least ten lymph nodes examined. Multivariable analysis was used to find factors associated with diagnosis of LVI, and the impact of LVI on OS. Propensity score matching(PSM) was used to adjust for bias in diagnosing LVI while testing the impact of LVI on OS. Results: 18,057 patients were eligible for our study with a median follow-up of (36.1 months). LVI status was determined in 91.8%. 19.1% of surgical specimens were found to be LVI positive. 2,323 patients had positive nodes with 50.8% of specimens having LVI, while 14.0% of specimens with negative nodes had LVI. Academic medical centers(AC); Medical centers associated with populations > 1,000,000(1M); and Mid-Atlantic(MA) region had higher rates of LVI(all p values < 0.0001, AC 22.7% vs 16.5%, OR = 1.49; 1M 21.1% vs 16.8%, OR = 1.32; MA 27.6% vs 17.1%, OR = 1.85), and higher rates of LVI associated with positive nodes (all p values < 0.0001, AC 34.6% vs 7.9%, OR = 6.14; 1M 36.4% vs 7.7%, OR = 6.82; MA 31.0% vs 6.2%, OR = 6.76). LVI was most frequently diagnosed in the MA region and least frequently found in the Mountain location( 27.6% vs 12.2%, OR = 2.73). LVI was associated with a significant decrement in OS that was independent of institution type, regional population, and institution location. PSM demonstrated that LVI was associated with a decrement in OS to the same degree per each nodal stage(N0,N1,N2) (p < 0.0001, HRs = 1.21 N0, 1.26 N1, 1.18 N2). Conclusions: LVI diagnosis and its association with positive nodes varies based upon hospital location/type and population. LVI was associated with a decrement in OS that was independent of N-Stage. LVI must be standardized and considered as a prognostic factor for staging cancer patients.