Characterizing Lymphovascular Invasion in Pediatric and Adolescent Malignant Ovarian Nongerminomatous Germ Cell Tumors: A Report from the Children’s Oncology Group

Abstract

Background:

Lymphovascular invasion (LVI) has been identified as a poor prognostic factor for a variety of tumors; however, its significance in malignant ovarian germ cell tumors (MOGCT) in pediatric and adolescent patients is not well described. We aim to clarify the significance of LVI in the subset of patients with nongerminomatous MOGCT.

Methods:

Records of patients 0–20 years of age with MOGCT enrolled on Children’s Oncology Group study AGCT0132 were reviewed. Patients with documented presence or absence of LVI in either institutional or central review pathology reports were included.

Results:

Of 130 patients with MOGCTs, 83 patients had of the presence or absence of LVI documented in their pathology report. 42/83 patients (50.6%) were found to have LVI present. The estimated odds of having LVI was higher in patients with stage II and III disease, 11 years and older and with the presence of choriocarcinoma. Event-free survival (EFS) and overall survival (OS) remained high in patients with LVI. Approximately 50% of patients with a documented LVI status in either institutional pathology report or central review were found to have LVI.

Conclusions:

The presence of LVI was higher in tumors with adverse risk factors including higher stage and age greater than 11 years. While LVI was not associated with EFS or OS in the intermediate risk group, further work is necessary to determine the effect of LVI on long-term disease-free survival. We, therefore, recommend routinely incorporating LVI status into institutional pathology reports for pediatric and adolescent patients with MOGCT.

Level of Evidence:

III.

1. Introduction

The histopathologic evaluation of malignant ovarian germ cell tumors (MOGCTs) is important for identifying risk factors that adversely affect patient outcomes. Unlike their testicular counterparts, the evaluation for and reporting of lymphovascular invasion (LVI) in MOGCTs is not a standard practice by pathologists. LVI (the presence of tumor cells within blood vessels or lymphatic channels) is a long established indicator of possible metastatic disease and has been identified as a poor prognostic factor in a wide variety of malignant neoplasms such as breast carcinoma, colorectal carcinoma, gastric carcinoma, etc. [1–4]. LVI as a poor prognostic factor is well established in adult testicular germ cell tumors (GCT) [5,6]. In fact, poor prognostic factors leading to relapse in stage 1 nonseminomatous testicular germ cell tumors are LVI, the absence of yolk sac tumor and the presence of embryonal carcinoma, with LVI as the strongest predictor of relapse [7–9]. The importance of LVI coupled with the need to unify international staging systems resulted in the development of the International Germ Cell Consensus Classification (IGCCC) system which incorporated LVI and established a clinical risk stratification system of metastatic testicular GCTs [10].

Refining a robust and reproducible risk stratification scheme for pediatric and adolescent patients with MOGCTs remains a critical objective to accurately identify patients at high risk for disease recurrence that will benefit from intensive systemic therapy. Conversely, this risk stratification scheme should also identify a low risk group requiring minimal treatment, thereby minimizing therapy-related toxicity, unwanted morbidities and potentially long term sequelae including the development of secondary malignancies. In 2015, Frazier et al. proposed a revised risk classification for pediatric patients with extracranial GCTs based on a model using the cure characteristics of patient age, tumor site and clinical stage. This cure model identified age ≥11 years, ovarian stage IV disease and extragonadal stage III to IV disease as significant predictors of worse long-term disease-free survival [11]. Similarly, in 2015 Meisel et al. proposed a risk stratification system based on the Memorial Sloan Kettering Cancer Center experience with MOGCTs which was modeled after the IGCCC system for metastatic testicular GCTs, thereby establishing it as the modified IGCCC risk classification system. Patients were assigned to good, intermediate and poor risk categories based on clinical FIGO stage, tumor histology, preoperative IGCCC group and pre-chemotherapy IGCCC group [12]. Unfortunately, with this modified IGCCC system, not all patients were able to be assigned to a risk category.

Despite the aforementioned studies, we continue to seek a comprehensive risk stratification system for MOGCTs with clearly defined statistically significant prognostic factors. Historically, various international cooperative trials have established such factors as older age, higher disease stage, elevated serum alpha fetoprotein (AFP) level and extragonadal site as poor prognostic factors [11]. Given our knowledge of the strong predictive value of LVI in a variety of tumor types including adult testicular germ cell tumors, we seek to explore the significance of LVI in MOGCTs. We hypothesize that LVI is a poor prognostic factor in MOGCTs.

2. Materials and methods

2.1. Patient identification

We conducted a secondary, retrospective analysis using data from the Children’s Oncology Group (COG) study AGCT0132, a phase III clinical trial conducted from November 2003 to July 2011. To be eligible for the trial, patients aged 0–20 years must have had an extracranial malignant germ cell tumor containing yolk sac tumor (YST), embryonal carcinoma, choriocarcinoma, or a mixture of these histologies with or without a component of teratoma (mature or immature). Patients with pure dysgerminoma or immature teratoma were excluded. Details of AGCT0132 have been previously published [13]. Of note, patients with ovarian tumors were only included if they had stage I, II, or III disease; patients with stage IV were considered high-risk and were not eligible for this trial. The current analysis only includes patients with a primary MOGCT with mention of LVI status in the pathology report, regardless of the presence or absence of LVI. The median follow-up time for the intermediate risk patients was 4.2 years (interquartile range 2.8–5.8 years).

2.2. Pathology and laboratory analysis

The presence of LVI was determined by the primary institution pathology report and/or from the expert COG central review pathologist. Central pathology review consists of a secondary review of diagnostic histology slides by a pathologist with expertise evaluating GCTs. The pathology report from the primary institution was reviewed as well as the report from the central review pathologist. Per central review, the presence of clustered neoplastic cells within a single blood vessel or lymphatic channel constitutes LVI. One tumor cell is not sufficient, as an artifact of slide cutting cannot be excluded. Additional factors that support the histologic presence of LVI (but are not required) include neoplastic cells attached to the wall of the blood vessel or red blood cells amongst the cluster of neoplastic cells.

Patients were classified as LVI positive if reports from either the primary institution or central review indicated the presence of LVI. Patients were classified as LVI negative if both the institution and central pathology review indicated the absence of LVI. If one report did not identify LVI and the other report did not comment on LVI status, the patient was deemed to be LVI negative. Patients were omitted if LVI was not mentioned in either report; however, a summary of patient characteristics in this cohort are presented to better understand whether these patients that lacked LVI designation differed from those with LVI status available. As per protocol, serum AFP levels were obtained preoperatively when available, or as close to the procedure as possible when not available. The components of the tumor histology (yolk sac tumor, embryonal carcinoma, choriocarcinoma, teratoma (mature and/or immature) and dysgerminoma) were obtained from central review reports, when available, or secondarily from the primary institution report.

2.3. Statistical analysis

Patient characteristics are described using the number and percentage of patients in each category. To investigate whether each of the a priori defined characteristics was associated with the presence of LVI, we fit unadjusted and adjusted logistic regression models [14]. The adjusted logistic regression model included age, stage, AFP value, and two binary variables indicating whether or not choriocarcinoma and embryonal carcinoma were identified. The estimated odds ratios and corresponding unadjusted 95% confidence intervals are presented. Likelihood ratio tests were performed with a significance level of α = 0.05 to test the association between each variable in the adjusted model and the presence of LVI. Because the work presented here is considered exploratory, we did not adjust for multiple testing.

Kaplan–Meier plots were generated to explore the association between LVI and event-free survival (EFS) and overall survival (OS). Time-to-event for EFS was defined as the time from enrollment until the first occurrence of relapse, second malignant neoplasm, or death and time for OS was defined as the time from enrollment until time to death. Patients who did not experience an event were censored at the last time they were seen during the trial. Because patients in the low-risk stratum (stage I) were treated with active surveillance and patients in the intermediate risk stratum (stage II or III) received chemotherapy, we looked at EFS and OS for both strata separately. Due to the low number of patients in the low-risk stratum (n = 11), we do not present these results. We conducted log-rank tests with a significance level of α = 0.05 to test the association between LVI and EFS and OS [15].

3. Results

One hundred and thirty ovarian patients from AGCT0132 are included in the current manuscript. Of these, 47 patients (36.2%) did not have any mention of the presence of absence of LVI and were therefore excluded from additional analyses. Table 1 presents the patient characteristics for these patients stratified by LVI status, including those with unknown LVI status. The percentage of stage I MOGCTs was higher among patients with an unknown LVI status compared to those with known LVI status. The percentage of patients with choriocarcinoma, embryonal carcinoma and dysgerminoma was higher among those with a known LVI status compared to those with an unknown LVI status. A higher percentage of patients with teratoma (both mature and immature) was observed among patients with an unknown LVI status.

Table 1.

Baseline patient characteristics stratified by Lymphovascular Invasion (LVI) classification. AFP (Alpha fetoprotein).

	LVI+	LVI −	Total (known LVI status)	Unknown LVI status
N	42	41	83	47
Age
Less than 11 years	9 (21.4%)	14 (34.1%)	23 (27.7%)	14 (29.8%)
11 years or older	33 (78.6%)	27 (65.9%)	60 (72.3%)	33 (70.2%)
AFP
1,0000 ng/ml or less	11 (26.2%)	12 (29.3%)	23 (27.7%)	19 (40.4%)
>1000 ng/ml	31 (73.8%)	29 (70.7%)	60 (72.3%)	28 (59.6%)
Stage
I	4 (9.5%)	7 (17.1%)	11 (13.3%)	12 (25.5%)
II	11 (26.2%)	19 (46.3%)	30 (36.1%)	16 (34.0%)
III	27 (64.3%)	15 (36.6%)	42 (50.6%)	19 (40.4%)
Histology
Choriocarcinoma	12 (28.6%)	7 (17.1%)	19 (22.9%)	3 (6.4%)
Embryonal carcinoma	10 (23.8%)	8 (19.5%)	18 (21.7%)	4 (8.5%)
Immature Teratoma	7 (16.7%)	19 (46.3%)	26 (31.3%)	22 (46.8%)
Mature Teratoma	11 (26.2%)	13 (31.7%)	24 (28.9%)	20 (42.6%)
Dysgerminoma	22 (52.4%)	9 (22.0%)	31 (37.3%)	9 (19.1%)
Yolk sac tumor	41 (97.6%)	39 (95.1%)	80 (96.4%)	45 (95.7%)

Table 2 presents the estimated odds ratios (OR) for the unadjusted and adjusted models. Comparing patients with similar age, AFP levels, and histology, we estimate the odds of LVI are 1.12 (95% CI: 0.26, 4.86) times higher for patients with stage II tumors compared to patients with stage I tumors, and 4.39 (95% CI: 0.99, 19.39) times higher for patients with stage III tumors compared to patients with stage I tumors. The odds of LVI were also estimated to be higher for patients 11 years or older compared to patients less than 11 years of age (estimated OR: 3.45, 95% CI: 1.10, 10.86) and for patients with choriocarcinoma histology compared to those without choriocarcinoma (estimated OR: 2.96, 95% CI: 0.64, 13.79). Only age and stage were found to be statistically significant based on the likelihood ratio tests.

Table 2.

Estimated odds ratios (OR) and corresponding 95% confidence intervals (CI) for adjusted and unadjusted logistic regression models with LVI as the outcome. The p-values correspond to the likelihood ratio test comparing the full adjusted model and the reduced models, with each variableremoved.

	Unadjusted Estimated OR (95% CI)	Adjusted Estimated OR (95% CI)	p-value
Age
>11 years	Reference	Reference	0.03
≥ 11 years	1.90 (0.71, 5.06)	3.45 (1.10, 10.86)
Stage
I	Reference	Reference	0.02
II	1.01 (0.24, 4.26)	1.12 (0.26, 4.86)
III	3.15 (0.79, 12.54)	4.39 (0.99, 19.39)
Alpha fetoprotein ( > 1000 vs. ≤ 1000 ng/ml)	1.17 (0.45, 3.05)	0.88 (0.30, 2.61)	0.81
Choriocarcinoma (yes vs. no)	1.94 (0.68, 5.57)	2.96 (0.64, 13.79)	0.15
Embryonal (yes vs. no)	1.29 (0.45, 3.68)	0.81 (0.18, 3.69)	0.79

Figure 1 presents the Kaplan–Meier plots for OS and EFS of intermediate risk (stage II and III) patients (n = 72). Using the log-rank test, we found that LVI was not associated with EFS (p-value = 0.9) or OS in intermediate risk patients (p-value = 0.1).

Fig. 1.

Kaplan–Meier plots of event-free survival (A) and overall survival (B) for intermediate risk (stage II and III) patients. The number of patients at risk and the number of cumulative events at selected time-points are presented below the plots. Estimated hazard ratios (HR) and 95% confidence intervals are also presented.

Figure 2 demonstrates a histologic example of LVI.

Fig. 2.

A. Ovarian dysgerminoma (Hematoxylin and eosin stain, 10X). B. High power view of dysgerminoma cells within a blood vessel (Hematoxylin and eosin stain, 20X).

4. Discussion

LVI is a poor prognostic factor in a variety of tumors but has not been explored in pediatric and adolescent patients with MOGCT. Given ongoing efforts to establish a robust risk stratification system, lack of agreement regarding the prognostically significant predictive factors and the strong data establishing LVI as a poor prognostic factor in testicular GCTs, we sought to characterize LVI in nongerminomatous MOGCTs. We report 83 pediatric and adolescent patients with MOGCTs evaluated for LVI by histopathologic assessment. The presence of LVI was more likely to be reported in patients with higher stage. Among those with known LVI status, approximately 50% were determined to have LVI present. The presence of LVI was associated with higher stage, older age, and presence of choriocarcinoma. LVI was not found to be associated with EFS or OS in patients in the intermediate risk stratum, although it should be noted that there were very few events in this study, particularly OS events.

Risk stratification in pediatric and adolescent patients with MOGCTs is critical to identify high risk groups that would benefit from intensified treatment, while minimizing systemic therapy in low risk groups. Despite a 94% 4-year overall survival rate, Billmire et al. reported a 52% 4-year EFS in stage I nongerminomatous MOGCTs managed with surgery and observation and stressed the importance of identifying prognostic variables to assist in predicting patients with an increased risk of recurrent disease [16].

A study of MOGCT using the SEER database identified dysgerminoma, young age, low grade, and surgery to be associated with improved survival [17]. In an international, multivariable analysis of 519 patients with extracranial GCT, Frazier et al. reported extragonadal site, higher stage, and age >11 years as adverse prognostic indicators and associated with worse long-term disease-free survival. Serum AFP was not found to be a significant prognostic factor [11]. Similarly, in the current study, the odds of having LVI was higher in patients 11 years of age and older as well as in those with higher stage (II and III) and serum AFP was not found to be a significant prognostic factor. The current study did not find an association between LVI and EFS or OS in intermediate risk patients. Given the small number of patients and events, as well as the inconsistency with reporting LVI in this study, larger cohort studies are needed to determine if LVI is a risk factor affecting outcome in MOGCT. We plan to conduct such an analysis when the current COG study of intermediate risk GCTs, AGCT1531, is completed.

LVI has been demonstrated to be associated with poor outcomes in a wide variety of tumors. For example, the presence of LVI has been correlated with lymph node and distant metastatic disease in primary breast cancer [18]. Additionally, in a recent systematic review, Kang et al. reports LVI as a risk factor in small rectal neuroendocrine tumors [19]. In a study looking at 3292 women with cervical carcinoma, LVI was identified to be a risk factor for ovarian metastatic disease [20]. However, the extent that LVI is incorporated into risk stratification and/or should alter treatment options is less clear. Blakely et al. reported LVI to be associated with lymph node involvement in patients with appendiceal neuroendocrine tumors and discussed consideration of further surgical intervention in patients with LVI [21]. Comparably, LVI has been shown to be an independent prognostic factor for nodal recurrence and lower overall survival in patients with stage 1 endometrial adenocarcinoma. The authors encourage consideration of adjuvant therapy in these stage 1 patients with LVI [22].

While the significance of LVI has not been established in MOGCTs, the presence of LVI has been demonstrated to be a poor prognostic factor in other types of ovarian surface epithelial neoplasms in adults. Lorenzini et al. recently reported 852 patients within the FRANCOYGYN research group and identified LVI as an independent predictive factor for worse overall and recurrence free survival, and this held true at both early and advanced stages. Furthermore, they reported that LVI was associated with residual disease after surgery and a lower response to chemotherapy. The authors recommend LVI be included in routine pathology review and conclude that LVI may affect treatment options [3]. Similarly, Matsuo et al. described patients with epithelial stage I ovarian tumors and found LVI to be associated with histology and stage, as well as an increased risk of hematogenous and lymphatic metastasis [23]. Comparably, Delvallee et al. reported on LVI in clear cell carcinoma, mucinous, low grade serous and low grade endometrioid ovarian cancer and identified LVI as a poor prognostic indicator for both OS and EFS. The authors go on to discuss the possibility of adjuvant chemotherapy in this subset of patients [24]. Darmus et al. found LVI as an independent predictor of lymph node metastases in low-grade serous ovarian tumors sand discussed the potential of reoperative surgery for lymphadenectomy in those patients who have LVI [25]. The current study expands the knowledge of LVI to now include nongerminomatous MOGCT. While our findings identify LVI to correlate with factors associated with a worse long-term disease-free survival, further work is necessary to determine whether LVI has a similar prognostic effect in ovarian GCT as seen in other ovarian tumors.

LVI in testicular GCT has been well-demonstrated to be an independent risk factor and predictor for metastatic disease with increased risk of relapse [7,26–28]. In pediatric and adolescent boys with stage 1 malignant testicular GCT, Rescorla et al. identified patients with LVI had a lower event-free survival than those without LVI (62 vs 84%); however, the numbers are small in this report and further work is necessary to clarify the role of LVI in pediatric testicular GCT patients [29]. A recent systematic review and meta-analysis again demonstrated LVI as the strongest risk factor for relapse in clinical stage 1 nonseminomatous GCT [30]. Both national and international guidelines now incorporate LVI in risk-adapted management guidelines of adult testicular cancer [8,9,26,27,31,32]. For example, Avulova et al. recommends consideration of retroperitoneal lymph node dissection or adjuvant chemotherapy rather than surveillance after resection in patients with LVI presence and high risk of recurrence after orchiectomy [33]. The current study initiates the inquiry of LVI in MOGCT; however, further data are necessary to determine whether the same prognostic implications of LVI in patients with testicular GCT can be expanded to patients with MOGCT.

Due to challenges with identifying LVI, Yossepowitch et al. sought to determine clinicopathological correlates for adult patients with testicular GCT with an aim to assist pathologists in identifying LVI. The authors conclude that providing pathologists with information on pre–orchiectomy tumor marker levels and, possibly, testicular pain at presentation may improve their diagnostic accuracy [34].

Unlike the testis, the pathologic and clinical staging of MOGCTs does not take into account the presence or absence of lymphovascular invasion. Therefore, it is not a required component of the pathology report. In the current study, LVI status was not documented by either the institutional pathologist or central review for 47 patients. It is unclear if LVI was not searched for or just not commented upon in these patients. There are two general options for uniform synoptic reporting for pathology reports in cancer patients, the College of American Pathologists (CAP) and the International Collaboration on Cancer Reporting. Although each provides a template for ovarian tumor reporting, the templates were designed primarily for epithelial based tumors as these are the most common ovarian tumors in adult women; they do not include MOGCTs. The CAP guidelines are intended for all malignant ovarian tumor histologies, but do not include assessment for LVI. To further clarify the role of LVI in these patients, we recommend including the status of LVI in synoptic pathology reports for all ovarian tumors.

There are several limitations to this study. First, this is a retrospective study with a subset of patients excluded given LVI was not commented upon. Since the presence or absence of LVI was not routinely documented, only patients with LVI status known were included, which may have led to a biased sample. Furthermore, in some cases, there were discordant results between central pathology review and institutional reports and results were deemed positive for LVI if one of the two reports identified the presence of LVI. In addition, given the inconsistent documentation of LVI and the low number of events, we could not report on EFS and OS in the low risk (stage I) group and results in the intermediate risk group should be further investigated. Given these limitations, we aim to further explore the role of LVI as a possible poor prognostic risk factor in ongoing prospective studies within COG.

5. Conclusion

In pediatric and adolescent patients with nongerminomatous MOGCTs, 50% of patients with a documented status of LVI were found to have LVI present. The presence of LVI was more likely to be found in older patients and in those with higher stage, both of which have been identified as adverse prognostic factors [11]. LVI was not associated with EFS or OS in intermediate risk (stage II or III) patients. However, given the inconsistences of reporting LVI, future prospective studies are necessary to further explore LVI as a possible risk factor for recurrent disease and outcome as well as its possible role in treatment algorithms. We therefore recommend routinely incorporating LVI into standardized pathology reports for pediatric and adolescent patients with MOGCT.

Abbreviations:

AFP

alpha fetoprotein

CAP

College of American Pathologists

COG

Children’s oncology group

EFS

event-free survival

GCT

germ cell tumor

IGCCC

International Germ Cell Consensus Classification

LVI

lymphovascular invasion

MOGCT

malignant ovarian germ cell tumor

OS

overall survival

YST

yolk sac tumor