Lymphovascular space invasion does not predict vaginal relapses in stage I endometrioid adenocarcinoma of the endometrium

This study was conducted to determine whether, in a pure population of patients with International Federation of Gynecology and Obstetrics stage I endometrioid endometrial (S1EE) carcinoma that is confined to the uterus and without lymph node metastases, the presence of lymphovascular space invasion (LVSI) is positively associated with vaginal relapses. Pathologic reports for all S1EE diagnosed in a hysterectomy specimen during a 9-year period (1997-2005) were reviewed. Cases with LVSI were selected and immunohistochemical staining for CD34, factor VIII-related antigen and pancytokeratin were performed on the relevant slides for confirmation. Various established prognostic/predictive clinicopathologic parameters were documented for the whole cohort and were correlated with the presence or absence of LVSI. One-way analysis of variance with Bonferroni correction and Fisher exact/chi(2) tests were used in the respective comparisons of continuous and categoric variables among the different groups. A total of 345 patients were diagnosed with S1EE during this period. The mean patient age for the cohort was 61.9 (+/-12.2) years (range, 28-89 years) and median follow-up was 80 months. Among these 345 patients, LVSI was present in 52 (15%), representing 1 (0.8%) of 121 stage 1A, 23 (14.5%) of 159 stage 1B, and 28 (43%) of 65 stage 1C cases (P < .001). Pelvic and/or para-aortic lymph node dissection was performed in 216 cases, and none had positive lymph nodes. We noted that LVSI correlated significantly with nuclear grade, with LVSI being present in 48% (10/21), 18% (32/175), and 5% (7/140) of cases with nuclear grades 3, 2, and 1 cases, respectively (P < .001). It was also more likely to be present in cases with architectural grades 2 and 3 (35/105) than grade 1 (16/234) (P < .01). Vaginal recurrences occurred in 2 (3.8%) of 52 patients with LVSI and 8 (2.7%) of 293 patients without LVSI (P = .65). In addition, on univariate analysis, LVSI did not correlate significantly with patient age, body mass index (using 50-year and 30 body mass index thresholds, respectively), or diabetic status. We conclude that LVSI is seen in approximately 15% of S1EE, and correlates with established prognostic parameters such as the level of myometrial invasion (substage), nuclear grade, and architectural grade. However, LVSI does not denote an increased likelihood of vaginal recurrences in S1EE. The presence of LVSI should not, in of itself, alter clinical staging, or trigger unnecessary therapeutic intervention.