Lymphovascular space invasion as a predictive factor for lymph node metastases and survival in endometrioid endometrial cancer – a Swedish Gynecologic Cancer Group (SweGCG) study

Abstract

Background: The aim of this study is to evaluate the impact of lymphovascular space invasion (LVSI) on the risk of lymph node metastases and survival in endometrioid endometrial adenocarcinoma.

Material and methods: As regard the study design, this is a cohort study based on prospectively recorded data. Patients with endometrioid endometrial adenocarcinoma registered in the Swedish Quality Registry for Gynecologic Cancer 2010–2017 with FIGO stages I–III and verified nodal status were identified (n = 1587). LVSI together with established risk factors, namely DNA ploidy, FIGO grade, myometrial invasion and age, were included in multivariable regression analyses with lymph node metastases as the dependent variable. Associations between the risk factors and overall and relative survival were included in multivariable models. Estimates of risk ratios (RR), hazard ratios (HR), excess mortality rate ratios (EMR), and 95% confidence intervals (95% CI) were calculated.

Results: The presence of LVSI presented the strongest association with lymph node metastases (RR = 5.46, CI 3.69–8.07, p < .001) followed by deep myometrial invasion (RR = 1.64, CI 1.13–2.37). In the multivariable survival analyses, LVSI (EMR = 7.69, CI 2.03–29.10,) and non-diploidy (EMR = 3.23, CI 1.25–8.41) were associated with decreased relative survival. In sub-analyses including only patients with complete para-aortic and pelvic lymphadenectomy and negative lymph nodes (n = 404), only LVSI (HR = 2.50, CI 1.05–5.98) was associated with a worsened overall survival.

Conclusion: This large nationwide study identified LVSI as the strongest independent risk factor for lymph node metastases and decreased survival in patients with endometrioid adenocarcinomas. Moreover, decreased overall survival was also seen in patients with LVSI-positive tumors and negative lymph nodes, indicating that hematogenous dissemination might also be important.

Introduction

Several factors have been identified as predictors of lymph node metastases, recurrence and survival in patients with early-stage endometrioid endometrial adenocarcinoma (EC). These factors include age, FIGO grade, cervical stromal invasion, depth of myometrial invasion (MI), and lymphovascular space invasion (LVSI) [1–10]. Currently, treatment for presumed early-stage EC involves surgery with or without lymphadenectomy followed by chemotherapy and/or radiation therapy based upon the estimated risk for recurrence according to above-mentioned high-risk factors [11].

LVSI has repeatedly been shown to be an independent negative prognostic factor for recurrence and survival [5,10,12,13] and, in some studies, even for patients without lymph node metastases [9,14]. In the ESMO-ESGO-ESTRO Consensus guidelines from 2016, patients with LVSI-positive endometrioid tumors (grades 1 and 2) that have not undergone lymph node staging are categorized as being in the high-intermediate risk group, and adjuvant external beam radiotherapy is recommended to decrease the risk of pelvic recurrence [11]. However, previous studies on LVSI are mostly single or multicenter studies, and the results cannot without caution be applicable at a national level. The aim of this study is to evaluate the impact of LVSI on the risk for lymph node metastasis and to assess its value as a prognostic factor for EC, using data from the Swedish Quality Registry for Gynecologic Cancer (SQRGC). The material is unique as it is population-based and obtained nationwide, and illustrates the usefulness of LVSI as a parameter on a national level.

Methods

The Swedish Quality Registry for Gynecologic cancer

All residents of Sweden are allocated a personal identification number facilitating the operation of official registries and research. Reporting to the Swedish Cancer Registry, founded in 1958, is mandatory for both clinicians and pathologists. The Swedish Cancer Registry has over 95% coverage of all malignant tumors compared to the Swedish Hospital Discharge register, and in 99% of cases the morphology is verified [15]. However, clinical data including treatment and follow-up are not registered. Hence, the SQRGC was established in 2008. Reporting to the SQRGC is performed prospectively by all hospitals and clinics in Sweden. The registration is web-based and includes information on patient and tumor characteristics, details of treatments and outcome, as well as clinical follow-up data for 5 years. SQRGC continuously receives data on date of death from the Population Registry and additionally the cause of death is retrieved from the Cause of Death Registry where 96% of individuals have a specific underlying cause of death recorded [16]. The registration for uterine malignancies in SQRGC started in 2010. Details of SQRGC have been described previously [17].

Considering the number of reported patients with malignancy of the uterus, the SQRGC shows high coverage (96%) compared with the Swedish Cancer Registry [17]. The validity of the recorded data in the SQRGC has previously been assessed, 268 patients were randomly selected and the agreement between the review of the patient records and the registered data was between 72% and 98% for 12 core variables, with the largest differences for dates [17]. For the variables analyzed in this study the concordance was 87% (n = 173) for FIGO stage, 80% (n = 179) for lymph and/or blood vessel invasion, and 96% (n = 80) for lymph node metastases.

Lymphovascular space invasion

Data on the presence of LVSI are part of the standardized pathology report for EC in the guidelines of Swedish Society of Pathology. Positive LVSI is defined as obvious lymphovascular space invasion identified in routine, hematoxylin and eosin staining [18] (in Swedish). There is no requirement for foci to be confirmed with immunohistochemistry, only that the pathologist making the diagnosis judges the focus to be sufficiently clear to be diagnostic.

Study population

Women with endometrioid adenocarcinoma of the corpus uteri registered in SQRGC with date of diagnosis between 1 January 2010 and 31 December 2017 were eligible for the study. Inclusion criteria were the following: endometrioid histology, FIGO stages I–III, and that information regarding lymph node status and treatment was available. The following data were collected from SQRGC: age, histology, FIGO stage, FIGO grade, degree of MI, presence of LVSI (from the pathology results of the hysterectomy specimen) DNA ploidy status (from the preoperative specimen or hysterectomy specimen) number and status of pelvic or para-aortic lymph nodes removed, and postoperative treatment.

In the statistical analysis, the total cohort was subdivided into two study cohorts. Study cohort 1 included patients with pelvic ± para-aortic sampling or lymphadenectomy with information about lymph node pathology and study cohort 2 included patients regarded as adequately staged according to the Swedish National Guidelines (≥5 para-aortic nodes and ≥10 pelvic nodes removed) with no lymph node metastases (Figure 1).

Figure 1. Flow chart on study cohorts. A patient can fullfil several exclusion criteria which is why total excluded is lower than the sum of exclusion criteria.

The Swedish National Guidelines for endometrial cancer

Most of the patients in the study were treated according to the Swedish National Guidelines for Endometrial cancer from 2011 (the guidelines were updated in 2017) [19]. In the guidelines, preoperative high-risk was defined as non-endometrioid histology (serous, clear cell carcinoma or carcinosarcoma), endometrioid adenocarcinoma FIGO grade 3, or non-diploid tumors. In high-risk tumors, a lymphadenectomy of the pelvic and para-aortic regions (up to the left renal vein) was recommended in addition to hysterectomy and salpingo-oophorectomy. Deep myometrial infiltration was not included in the national guidelines until the revised version of 2017.

Postoperatively, patients with high-risk histology and negative nodes were recommended chemotherapy ± brachytherapy and those with positive nodes or no lymphadenectomy were offered chemotherapy + external beam radiotherapy. LVSI was not included as a high-risk parameter in either the first or in the revised versions.

Statistics

The study has a retrospective cohort design based on prospectively recorded register data. The Generalized Linear Model with log-binomial regression was used to evaluate the association between risk factors and lymph node metastases. Estimates of risk ratios (RR) and 95% confidence intervals (95% CI) were calculated. Royston–Parmar flexible parametric models (module stpm2) [20] was used to explore associations between risk factors and overall survival (OS) and relative survival (RS) in multivariable models. Differences in overall survival was evaluated with hazard ratios (HR) while differences in relative survival was evaluated with excess mortality rate ratios (EMR). The proportionality assumption was tested with estat phtest. No evidence for violation of the proportionality assumption was found. Life table analyses of RS were computed with the strs module [21]. A 5% significance level (two-sided tests) was used in all analyses. The STATA Statistical Software release 13 (StataCorp, College Station, TX) was used.

The study was approved by the Regional Ethics Committee in Gothenburg (D.nr. 814-15).

Results

The flow chart illustrates the selection for the two study cohorts and remaining subjects for the subsequent analyses. From the 11,689 women with endometrial malignancies in the registry, 1587 patients with endometrioid histology and pelvic ± para-aortic sampling or lymphadenectomy with information about lymph node pathology were selected for study cohort 1, and 595 patients with adequate lymphadenectomy and negative lymph nodes were selected for study cohort 2 (Figure 1). Patients with undefined values of any variable included in the multivariable analyses were excluded; hence, 959 patients remained for multivariable analysis in study population 1 and 404 in study population 2. A patient can fulfill several exclusion criteria which is why total excluded is lower than the sum of exclusion criteria. Table 1 presents the clinical and tumor characteristics of the selected study cohorts. Analysis of LVSI was not done/not evaluable for 28% of cases in cohort 1 and for 20% in cohort 2. The median numbers of harvested lymph nodes were 26 pelvic and 12 para-aortic in cohort 2.

Table 1. Patient, tumor, and treatment characteristics according to study population.

	All uterine malignancies (n = 11,689)	Study population 1^a (n = 1587)	Study population^b 2 (n = 595)
Median age (range)	70 (24–105)	67 (31–89)	67 (36–83)
Median number of pelvic lymph nodes (range)	–	20 (0–74)	26 (10–72)
Median number of para-aortic lymph nodes (range)	–	5 (0–51)	12 (5–51)
FIGO grade (endometrioid adenocarcinoma, n = 9395)
Grade 1	3619 (39%)	471 (30%)	141 (24%)
Grade 2	2848 (30%)	548 (35%)	187 (31%)
Grade 3	1336 (14%)	497 (31%)	227 (38%)
Not evaluable/missing	1592 (17%)	71 (4.5%)	40 (6.8%)
Surgical FIGO stage
IA	5641 (48%)	754 (48%)	307 (52%)
IB	2269 (19%)	370 (23%)	178 (30%)
I substage not specified	23 (0.2%)	0	0
II	780 (6.7%)	175 (11%)	78 (13%)
III	1159 (10%)	288 (18%)	32 (5.4%)
IV	558 (4.8%)	–	–
Not evaluable/not done	1259 (11%)	0	0
Myometrial invasion (MI)
None	949 (8.2%)	115 (7.3%)	60 (10%)
<50%	4834 (41%)	806 (51%)	297 (50%)
≥50%	3112 (27%)	651 (41%)	237 (40%)
Not evaluable/not done	2794 (24%)	15 (1.0%)	1 (0.17%)
DNA ploidy
Ploidy not reported	1753 (15%)	4 (0.25%)	3 (0.50%)
Diploid	5773 (49%)	813 (51%)	269 (45%)
Non-diploid	2323 (20%)	614 (39%)	280 (47%)
Not evaluable/not done	1840 (16%)	156 (9.9%)	43 (7.2%)
Lymphovascular space invasion (LVSI)
LVSI not reported	1761 (15%)	2 (0.13%)	0
LVSI not present	5264 (45%)	828 (52%)	377 (63%)
LVSI present	1409 (12%)	308 (19%)	97 (16%)
Not evaluable/not done	3255 (28%)	449 (29%)	121 (20%)
Treatment
Treatment not reported	1732 (15%)	7 (0.33%)	1 (0.17%)
Surgery only	5687 (49%)	864 (41%)	340 (57%)
Surgery + adjuvant treatment	3275 (28%)	1206 (58%)	252 (42%)
Other	995 (8.5%)	23 (1.1%)	2 (0.34%)

^a Endometrial endometrioid carcinomas with any lymph node staging/lymphadenectomy.

^b Endometrial endometrioid carcinomas with adequate lymphadenectomy according to the Swedish National Guidelines (≥5 para-aortic nodes and ≥10 pelvic nodes removed) with no lymph node metastases.

LVSI and risk for lymph node metastases

The results of the multivariable model for risk of lymph node metastases in cohort 1 are presented in Table 2. The presence of LVSI presented the strongest association with lymph node metastases (RR = 5.46, CI 3.69–8.07) followed by deep MI (RR = 1.64, CI 1.13–2.37). Neither DNA non-diploidy nor FIGO grade 3 showed any significant association with risk of lymph node metastases.

Table 2. Multivariable analyses of risk factors for lymph node metastases in endometrial endometrioid carcinomas with any lymph node staging (study cohort 1, n = 959).

	LNM/total LN^a	RR (95% CI)
LVSI
Not present	38/707 (5.4%)	Reference
Present	84/252 (33%)	5.46 (3.69–8.07)
Myometrial invasion
<50%	38/542 (7.0%)	Reference
≥50%	84/417 (20%)	1.64 (1.13–2.37)
FIGO-grade
Grades 1 and 2	68/618 (11%)	Reference
Grade 3	54/341 (16%)	0.88 (0.64–1.22)
DNA-ploidy
Diploid	69/555 (12%)	Reference
Non-diploid	53/404 (13%)	0.98 (0.71–1.34)
Age (continuous)^b	112/959 (12%)	0.98 (0.97–1.00)

^a LNM: number of women with lymph node metastasis; Total LN = number of women who had lymph nodes removed.

^b The effect of being one year older at diagnosis.

In the multivariable survival analysis of cohort 1, LVSI and non-diploidy were associated with decreased RS (LVSI: EMR = 7.69, CI 2.03–29.10 and non-diploidy: EMR = 3.23, CI 1.25–8.41) and OS (Table 3).

Table 3. Multivariable analyses of risk factors for overall and relative survival in endometrioid endometrial carcinomas with any lymph node staging (study cohort 1, n = 959).

	Deaths/total	Overall survival HR (95% CI)	Relative survival EMR (95% CI)
LVSI
Not present	42/707 (5.9%)	Reference	Reference
Present	45/252 (18%)	3.18 (1.93–5.24)	7.69 (2.03–29.10)
Myometrial invasion
<50%	34/542 (6.2%)	Reference	Reference
≥50%	53/417 (13%)	1.56 (0.97–2.52)	2.52 (0.86–7.36)
FIGO-grade
Grades 1 and 2	50/618 (8.1%)	Reference	Reference
Grade 3	37/341 (9.1%)	0.88 (0.55–1.41)	0.96 (0.42–2.21)
DNA-ploidy
Diploid	35/555 (6.3%)	Reference	Reference
Non-diploid	52/404 (13%)	1.94 (1.24–3.04)	3.23 (1.25–8.41)
Adjuvant treatment
No	31/480 (6.5%)	Reference	Reference
Yes	56/479 (12%)	0.77 (0.47–1.27)	1.11 (0.22–5.53)
Age (continuous)^a	87/959 (9.1%)	1.05 (1.03–1.08)	1.02 (0.97–1.07)

^a The effect of being one year older at diagnosis.

LVSI as a prognostic factor

In study cohort 2, only LVSI (HR = 2.50, CI 1.05–5.98) was associated with decreased overall survival (Table 4). The data was not sufficient to perform a relative survival analysis.

Table 4. Multivariable analyses of risk factors for overall survival in endometrial endometrioid carcinomas with adequate lymphadenectomy and no lymph node metastases (study cohort 2, n = 404).

	Deaths/total	HR (95% CI)
LVSI
Not present	17/318 (5.4%)	Reference
Present	11/86 (13%)	2.50 (1.05–5.98)
Myometrial invasion
<50%	12/230 (5.2%)	Reference
≥50%	16/174 (9.2%)	1.47 (0.63–3.44)
FIGO-grade
Grades 1 and 2	15/224 (6.7%)	Reference
Grade 3	13/180 (7.2%)	0.92 (0.41–2.07)
DNA-ploidy
Diploid	11/213 (5.2%)	Reference
Non-diploid	17/191 (8.9%)	1.42 (0.64–3.17)
Adjuvant treatment
No	13/227 (5.7%)	Reference
Yes	15/177 (8.5%)	0.76 (0.32–1.82)
Age (continuous)^a	28/404 (6.9%)	1.07 (1.02–1.13)

^a The effect of being 1 year older at diagnosis.

In patients with lymph node metastases (n = 357, 111 events), the strongest and only significant prognostic factor in the multivariable analysis including LVSI, MI, ploidy, and age (n = 213, 61 events) was LVSI positivity with an EMR of 5.56 (CI 1.62–19.05). The 5-year relative survival in this group was 93% (CI 79–100%) for LVSI-negative compared with 53% (CI 43–63%) for LVSI-positive tumors (univariable analysis) (data not shown).

When separately analyzing a low-risk group (FIGO grades 1 and 2, no deep MI and diploid tumors (n = 3342, 286 events)), the HR for overall survival for LVSI-positive tumors was 1.69 (95% CI 1.00–2.86) in a univariable analysis (n = 2,575, 215 events). In this group, the 5-year overall survival was 91% (CI 89–92%) for LVSI-negative and 88% (79–93%) for LVSI-positive tumors and the corresponding relative survival was 99.9% (CI 98–101%) and 98.6% (CI 88.6–105%) respectively (data not shown).

Discussion

The results from this population-based, nationwide study identified lymphovascular space invasion as the strongest risk factor for lymph node metastases and poor prognosis in patients with endometrioid endometrial cancer.

Our results are consistent with previous findings that a high rate of lymph node metastases [12,22,23] and decreased survival is found in patients with LVSI-positive tumors [12,13,24,25]. Furthermore, we show that LVSI is associated with a poorer prognosis even in adequately staged patients with negative nodes.

The common definition of LVSI is the presence of tumor cells in a space lined by endothelial cells outside the immediate invasive tumor border, but the exact definition and the impact of the extent of the LVSI are debated. In the pooled analyses of the PORTEC 1 and 2 trials including stage I endometrial carcinomas (129 patients), substantial LVSI (in contrast to no or only focal LVSI) remained the strongest independent prognostic factor for pelvic recurrence, distant metastasis and overall survival (HR 2.0, CI 1.3–3.1). External beam radiotherapy reduced the risk of pelvic regional recurrence (HR 0.3, CI 0.1–0.8) [5]. In accordance with these results, the ESMO-ESGO-ESTRO Consensus Guidelines from 2016 recommend adjuvant external beam radiotherapy for patients with LVSI-positive (regardless the depth of invasion), stage I endometrioid tumors (grades 1 and 2) to decrease the risk of pelvic recurrence [11]. In the guidelines of the Swedish Society of Pathology [18], LVSI is defined as obvious lymphovascular invasion in routine staining which probably is comparable to the definition of substantial LVSI in the PORTEC study. In a study by Matseo et al. the importance of the depth and extent of LVSI was pointed out [26]. It has been suggested that the blood vessel invasion rather than the lymph vessel invasion is responsible for the increased risk of recurrence and death, probably due to increased risk for hematogenous metastases [27]. This is in line with our results demonstrating decreased overall survival also in patients without lymph node metastases.

Most previous reports are single institution studies, but here we demonstrate that the analysis of LVSI is feasible on a national level with comparable results.

Other strengths of this study are in particular the large sample size and that it is based on a nationwide population. The health care system is quite uniform all over Sweden and free of charge to all citizens living in Sweden, reimbursed by the public social security system. There are some limitations that should be considered. The Swedish guidelines for endometrial cancer published in 2011 included recommendations for lymph node staging and the procedure was gradually implemented in the various regions in Sweden. This is illustrated by the fact that only around 30% of endometrioid grade 3 tumors actually had a lymphadenectomy performed during the study period, and in approximately 30% of cases LVSI was not analyzed. In cohort 2 (patients with negative lymph nodes), a potential bias is that malignant lymph nodes might occasionally have been missed during the lymphadenectomy, leading to incorrect staging.

An argument against the usefulness of LVSI is that it is not reliable in preoperative biopsies. However, in the era of sentinel node biopsy, the previous debate about preoperative low/high-risk may no longer be relevant. Instead, data on LVSI can be used as a complement to other factors in the postoperative management. The presence of LVSI can emphasize the need for adjuvant therapy in borderline cases, for example when finding isolated tumor cells in lymph nodes or if the sentinel node procedure was unsuccessful for some reason. Prognostic factors are also of great importance in identifying candidates for randomized clinical studies in order to optimize the number of events.

Conclusion

This large nationwide study identifies LVSI as the strongest independent risk factor for lymph node metastases and decreased survival in patients with endometrioid adenocarcinomas. Moreover, decreased overall survival was seen in patients with negative lymph nodes, indicating that hematogenous dissemination might be important in these tumors.

Abbreviations
CI	=	confidence interval
EC	=	endometrial cancer
EMR	=	excessive mortality rate ratio
FIGO	=	The International Federation of Gynecology and Obstetrics
HR	=	hazard ratio
LVSI	=	lymphovascular space invasion
MI	=	myometrial invasion
OR	=	overall survival
RR	=	risk ratio
RS	=	relative survival
SQRGC	=	Swedish Quality Registry of Gynecologic Cancer

Disclosure statement

All authors declared no conflicts of interest.

Additional information

Funding

This research was funded by the Swedish Cancer Society.