Significance of lymphovascular space invasion in epithelial ovarian cancer

While the prognostic significance of lymphovascular space invasion (LVSI) is well established in endometrial and cervical cancer, its role in ovarian cancer is not fully understood. First, a training cohort was conducted to explore whether the presence and quantity of LVSI within the ovarian tumor correlated with nodal metastasis and survival (n = 127). Next, the results of the training cohort were applied to a different study population (validation cohort, n = 93). In both cohorts, histopathology slides of epithelial ovarian cancer cases that underwent primary cytoreductive surgery including pelvic and/or aortic lymphadenectomy were examined. In a post hoc analysis, the significance of LVSI was evaluated in apparent stage I cases (n = 53). In the training cohort, the majority of patients had advanced-stage disease (82.7%). LVSI was observed in 79.5% of cases, and nodal metastasis was the strongest variable associated with the presence of LVSI (odds ratio [OR]: 7.99, 95% confidence interval [CI]: 1.98–32.1, P = 0.003) in multivariate analysis. The presence of LVSI correlated with a worsened progression-free survival on multivariate analysis (hazard ratio [HR]: 2.06, 95% CI: 1.01–4.24, P = 0.048). The significance of the presence of LVSI was reproduced in the validation cohort (majority, early stage 61.3%). In apparent stage I cases, the presence of LVSI was associated with a high negative predictive value for nodal metastasis (100%, likelihood ratio, P = 0.034) and with worsened progression-free survival (HR: 5.16, 95% CI: 1.00–26.6, P = 0.028). The presence of LVSI is an independent predictive indicator of nodal metastasis and is associated with worse clinical outcome of patients with epithelial ovarian cancer.


Introduction
Ovarian cancer remains a deadly disease and the most common cause of death among gynecologic malignancies. In 2012, over 22,300 women in the United States are estimated to be diagnosed with ovarian cancer, and over 15,500 will die of this disease [1]. The majority of ovarian cancer patients were present with advanced-stage disease, and cytoreductive surgery remains a mainstay in management [2]. Cytoreductive surgery for ovarian cancer includes total abdominal hysterectomy, bilateral salpingooophorectomy, pelvic and para-aortic lymphadenectomy,

Cancer Medicine
Open Access and omentectomy. Information obtained from the surgical specimen is useful for determining prognosis including histology, grade, and the extent of disease spread. However, given the imperfect predictive nature of these factors, additional markers are needed.
Lymphovascular space invasion (LVSI) is defined as the presence of tumor cells inside the capillary lumens of either the lymphatic or the microvascular drainage system within the primary tumor. The significance of LVSI has been extensively studied in other types of gynecologic malignancies such as endometrial [3][4][5][6][7], cervical [8][9][10][11][12][13], and vulvar cancer [14][15][16]. In each, the presence of LVSI in the tumor is associated with increased risk of disease spread (especially nodal metastases), increased chance of disease recurrence, and decreased survival outcomes. In contrast, there has been little investigation on the impact of LVSI in epithelial ovarian cancer [17,18], and the role of LVSI in the outcome of women with ovarian cancer remains unclear. The aim of this study is to evaluate the impact of the presence of LVSI within the ovarian malignancy on clinical variables and survival outcomes in women with ovarian cancer.

Training set cohort
After Institutional Review Board (IRB) approval was obtained in Mercy Medical Center in Baltimore, a previously established ovarian cancer database was utilized for this study [19]. Inclusion criteria included cases with epithelial ovarian cancer that underwent primary cytoreductive surgery including pelvic and/or aortic lymphadenectomy between January 1995 and January 2009. Cases with metastatic disease from sites other than ovarian primary, synchronous cancer types, and tumors of low malignant potential were excluded from the study. Variables abstracted from the medical records were patient demographics and survival outcomes after surgery.
The histopathology slides of the cases that met inclusion criteria were examined. The total number of slides stained with hematoxylin and eosin, the number of slides that contained ovarian tumor, and the number of slides with ovarian tumor that contained foci of LVSI were recorded for each case. The number of foci of LVSI was manually counted and the average number of foci of LVSI per slide was determined per case, and classified into "none" for no LVSI, and "low" (1-33 percentile), "moderate" (34-66 percentile), and "high" ( 67 percentile) among LVSI presenting tumors. For instance, if five slides contained LVSI within the ovarian tumor for a total of 10 foci, the average number of LVSI foci per slide was classified as 2 for that case. The gynecologic pathologist who evaluated the slides in this cohort was completely blinded to the clinical information. The total number of lymph nodes examined and the number of lymph nodes with tumor metastasis were also recorded. The presence and quantity of LVSI were then correlated with clinical variables, nodal metastasis, and survival outcome.

Validation cohort
By utilizing the results of the training set cohort, an additional cohort was conducted and examined using a previously collected database for ovarian cancer in the participating institutions (Gynecologic Oncology Group in Osaka, Japan). IRB approval was obtained at each site. This study group was chosen to demonstrate whether the training set results were reproducible in a different population. Similar to the training set cohort, the histopathology slides of women with epithelial ovarian cancer who underwent primary cytoreductive surgery including pelvic and/or aortic lymphadenectomy were examined by different gynecologic pathologists who were blinded to the results of the training set cohort and to clinical information. Pulled slides were examined manually and the quantity of LVSI was scored as none, low, moderate, and high as defined by the training set cohort.

Definition
Detailed description for the definition of LVSI is shown in Supplemental Methods. Among serous histology, grade 2 and 3 tumors were grouped as high-grade serous carcinoma analyzed as an independent group, whereas grade 1 tumors were grouped as low-grade serous carcinoma based on the recent accumulating data [20,21]. In this two-tier grading system, it adequately correlates to conventional FIGO (the International Federation of Gynecology and Obstetrics) grading system and further provides valuable clinical outcome for ovarian cancer patients when compared with FIGO grading system demonstrating high-grade serous carcinoma as the distinct ovarian cancer subtype [20]. The significance of the presence of LVSI within the ovarian tumor was evaluated in apparent stage I disease across the two cohorts defined as ovarian tumor grossly confined to the ovary. The date of progression was determined by clinical examination, imaging studies, and/or CA-125 levels. Progression-free survival was defined as the time interval from the date of primary cytoreductive surgery to the date of documented first recurrence or progression of disease. If there was no recurrence, progression-free survival was determined as the date of last follow-up. Overall survival was defined as the interval between the primary cytoreductive surgery and the date of death or last follow-up.

Statistical analysis
Continuous variables were assessed for normality (Kolmogorov-Smirnov test) and expressed as appropriate (mean with SD or median with range). Student's t-test or Mann-Whitney U-test was performed for continuous variable as appropriate. Categorical variables were evaluated with Fisher's exact test or chi-square test as appropriate, expressed with odds ratio (OR) and 95% confidence interval (CI). Risk factor of LVSI was evaluated with logistic regression test, and multivariate logistic regression test was further performed among significant variables in univariate analysis. Receiver-operator characteristic (ROC) curve analysis was performed to identify the risk factors for nodal metastasis expressed with area under the curve (AUC), and the cutoff analysis was performed to maximize the risk of nodal metastasis. Sensitivity, specificity, positive and negative predictive values, and accuracy of nodal metastasis were determined with the results of LVSI status. For survival data analysis, to determine the significance of variables for the survival outcomes for progression-free survival and overall survival, univariate (log-rank) and multivariate (Cox proportional hazard regression test) analyses were performed as appropriate. Survival curves were constructed with the Kaplan-Meier method. P-values of less than 0.05 were considered statistically significant (all, two-tailed). The Statistical Package for Social Science software (SPSS, version 12.0, IL) was used for all analyses.

Discussion
The key findings of our study are that the presence of LVSI within the ovarian tumor is an independent predictive factor of both nodal metastasis and survival in women with ovarian cancer. If LVSI was present, the quantity of LVSI did not add further prognostic information, but it did impact the likelihood of nodal metastases. Our results add new information to the management of ovarian cancer exhibiting LVSI. Several key areas in the study deserve special mention. In a view of systematic literature review using public searching engine PubMed and MEDLINE between 1955 and March 2012 with entry keywords of "ovarian cancer" and "lymphovascular space invasion," there is little data evaluating the prognostic significance of the presence of LVSI in ovarian cancer [17,18,[22][23][24]. These limited numbers of prior studies showed mixed results and were hampered by either small sample, lack of quantification of LVSI, or lack of a validation set. The summary of literature review is provided in Table S1. Collectively, the size, number, and quality of prior studies investigating the prognostic significance of LVSI in ovarian cancer were limited, and the impact of the presence of LVSI in ovarian cancer had not been clearly delineated.
Our study is the first to have a formal review of LVSI by a pathologist of all samples, presenting quantitative data, defining the role of LVSI, and validation in a separate independent cohort in epithelial ovarian cancer. The results demonstrated the strong link between the presence of LVSI and the likelihood of nodal metastases in women with ovarian cancer in two disparate study groups. In each, the quantity of LVSI present, as defined by maximum number of foci per slide, further correlated with likelihood of nodal metastases. When apparent stage I cases were combined in a post hoc analysis, the absence of LVSI within the ovarian tumor was an excellent predictor of negative nodal status. There are two possible clinical implications to these results. The first is that if surgical staging was incomplete in that lymphadenectomy was not performed, the lack of LVSI within the ovarian tumor gives reassurance that the nodes are uninvolved. Second, women with advanced-stage ovarian cancer whose tumors contain LVSI have a worsened progression-free survival and may benefit from consolidation therapy after completion of front-line chemotherapy. As vascular endothelial growth factor (VEGF) pathway is strongly associated with increased LVSI, targeting the VEGF axis with anti-VEGF inhibitor may be an attractive approach in ovarian tumor expressing LVSI [25][26][27].
A strength of our study is that the sample population is homogeneous in which only primary epithelial ovarian cancer cases who underwent primary surgery were included. Also, this is one of the largest studies evaluating the significance of the presence of LVSI in ovarian cancer. Furthermore, we demonstrated the durability of the impact of LVSI in two disparate cohorts. Potential weaknesses of the study are that it is retrospective in nature and thus confounding factors might have been missed, and that the sample size of women with apparent stage I disease is relatively small. In addition, only 50.4% cases of training set cohort have the information of aortic lymph nodes, and the number of lymph nodes sampled in our study was fewer than the reported literature. Therefore, nodal number may not be enough to evaluate the status of lymph node metastasis in apparent stage I ovarian cancer. Another limitation is that evaluation of the presence of LVSI in our study is based on hematoxylin and eosin staining but not on immunohistochemical analysis; however, it is the former that is widely used to determine the presence of LVSI. Prospective studies will be useful to help to confirm our findings, particularly in terms of the lack of nodal metastases seen in apparent stage I disease when LVSI is absent.
In conclusion, the presence of LVSI is an independent predictive indicator of nodal metastasis and is associated with a worse progression-free survival in ovarian cancer. Standardization of evaluation and scoring of LVSI would potentially yield important information that might help guide management. Further prospective investigation on the impact of the presence and quantity of LVSI in women with epithelial ovarian cancer is warranted.