Is Substantial Lymphovascular Space Invasion Prognostic in Patients With Pathologically Lymph Node-Negative Endometrial Cancer?

Abstract

Purpose:

Lymphovascular space invasion (LVSI) predicts for higher rates of recurrence and increased mortality in endomerial cancer. Using 3-tier LVSI scoring, a PORTEC-1 and -2 trials analysis demonstrated that substantial LVSI was associated with worse locoregional (LR-DFS) and distant metastasis disease-free survival (DM-DFS), and these patients possibly benefited from external beam radiation therapy (EBRT). Furthermore, LVSI is a predictor for lymph node (LN) involvement, but the significance of substantial LVSI is unknown in patients with a pathologically negative LN assessment. We aimed to evaluate clinical outcomes of these patients in relation to the 3-tier LVSI scoring system.

Methods and Materials:

We performed a single-institutional retrospective review of patients with stage I endometrioid-type endometrial cancer who underwent surgical staging with pathologically negative LN evaluation from 2017 to 2019 with 3-tier LVSI scoring (none, focal, or substantial). Clinical outcomes (LR-DFS, DM-DFS, and overall survival) were analyzed using the Kaplan-Meier method.

Results:

A total of 335 patients with pathologically LN-negative stage I endometrioid-type endometrial carcinoma were identified. Substantial LVSI was present in 17.6% of patients; 39.7% of patients received adjuvant vaginal brachytherapy and 6.9% of patients received EBRT. Adjuvant radiation treatment varied by LVSI status. In patients with focal LVSI, 81.0% received vaginal brachytherapy. Among patients with substantial LVSI, 57.9% received vaginal brachytherapy alone, and 31.6% of patients received EBRT. The 2-year LR-DFS rates were 92.5%, 98.0%, and 91.4% for no LVSI, focal LVSI, and substantial LVSI, respectively. The 2-year DM-DFS rates were 95.5%, 93.3%, and 93.8% for no LVSI, focal LVSI, and substantial LVSI, respectively.

Conclusions:

In our institutional study, patients with pathologically LN-negative stage I endometrial cancer with substantial LVSI had similar rates of LR-DFS and DM-DFS compared with patients with none or focal LVSI. These findings highlight the need for multi-institutional studies to validate the prognostic value of substantial LVSI in this patient population.

Introduction

In the United States, approximately 66,000 women are diagnosed with endometrial cancer yearly, with the majority of these patients presenting with localized disease.¹ Standard of care for these patients consists of initial surgical staging followed by risk-adapted therapy with most patients receiving observation. The proper selection of patients at high risk of local and distant recurrence that would benefit from adjuvant therapy has been ongoing (GOG-99, PORTEC-1, PORTEC-2).^2–4 Patients’ risk of recurrence is currently stratified by multiple known clinical and pathologic factors including age, myometrial invasion, International Federation of Gynecology and Obstetrics (FIGO) grade, and lymphovascular space invasion (LVSI). A pooled analysis of PORTEC-1 and -2 examined different LVSI scoring systems and demonstrated that a 3-tier scoring system (none, focal, or substantial) predicted for pelvic regional recurrence, distant metastasis, and overall survival, and patients with substantial LVSI potentially benefited from external beam radiation therapy (EBRT).⁵ These data influenced the accruing PORTEC-4a trial, which is currently randomizing patients with substantial LVSI into the high-risk arm receiving EBRT.⁶ Multiple retrospective series have since demonstrated substantial LVSI is predictive for worse clinical outcomes.^7,8 Substantial LVSI is correlated with lymph node (LN) involvement.⁹ There is currently a knowledge gap about whether substantial LVSI is a surrogate predictor of LN involvement or an independent prognostic factor for worse clinical outcomes. In this study, patients with stage I endometrial cancer with pathologically negative LN evaluation at time of surgical staging were examined regarding the relationship of none, focal, or substantial LVSI to locoregional disease-free survival (LR-DFS), distant metastasis disease-free survival (DM-DFS), and overall survival (OS).

Methods and Materials

This study reports the clinical outcomes of a subset of patients with stage I endometrial cancer whose clinical and pathologic factors were previously reported.⁹ After institutional review board approval, we conducted a retrospective review for patients with stage I endometrioid-type endometrial cancer who underwent surgical staging from July 2017 to September 2019. Inclusion criteria consisted of patients with stage I pT1a-b endometrioid-type endometrial cancer with negative LN assessment. Patients with isolated tumor cells (ITC) on LN evaluation were included to be consistent with current FIGO staging guidelines. Exclusion criteria included patients with mixed histology tumors, synchronous tumors, no LN assessment at time of surgery, or neoadjuvant therapy. LN assessment was performed using standard sentinel LN mapping technique as described in NCCN guidelines with isocyanine green followed by dissection.^10–12 Side specific pelvic lymphadenectomy was performed in the event of failed mapping. Any clinically enlarged or palpably abnormal nodes were removed regardless of mapping. All sentinel nodes were subjected to evaluation with ultrastaging.^13,14 After completion of surgery and adjuvant radiation therapy, patient follow-up schedule typically consisted of a physical examination every 3 months for the first 2 years and every 6 months for years 3 to 5. Imaging was reserved for examination findings and/or symptoms concerning for recurrence. Patient charts were retrospectively reviewed to collect clinicopathologic and treatment characteristics. LVSI was defined according to the PORTEC 3-tier system⁵ by institutional pathologists specializing in gynecologic pathology. Clinical outcomes were defined by time from surgery to event and analyzed using the Kaplan-Meier method with log-rank test.¹⁵ Cox regression was used for univariate analysis for predictors of LR-DFS and DM-DFS. The χ² test was used to assess for association between extent of LVSI and adjuvant radiation. A P value <.05 was considered statistically significant. All statistics were performed with SPSS, version 27. Kaplan-Meier curves were generated using GraphPad Prism 8.

Results

Patient characteristics

The clinical and pathologic factors of patients with clinically uterine-confined endometrioid-type endometrial cancer who underwent surgical staging have been published previously.⁹ This analysis presents the clinical outcomes of patients with stage I endometrial cancer who underwent LN evaluation and had pathologically LN negative disease. During the accruing time, there were a total of 438 patients with stage I endometrioid-type endometrial cancer. Of these 438 patients, 335 patients (76.5%) had a negative LN evaluation in which 65.7% had sentinel LN biopsy and 34.3% had unilateral or bilateral LN dissection. The clinical, pathologic, and adjuvant characteristics of these 335 patients are presented in Table 1. The median age was 63.6 years (interquartile range, 58.3–70.7 years), and myometrial invasion ≥50% was present in 32.5% of patients. In this cohort, 63% of tumors were FIGO grade 1, 29.9% were FIGO grade 2, and only 7.2% were FIGO grade 3. Focal LVSI was present in 19.4% (n = 65) of patients, and substantial LVSI was present in 17.6% (n = 59). The majority (50.4%) of patients received no radiation after surgical staging, 39.7% of patients received vaginal brachytherapy (VBT) alone, and 6.9% of patients received EBRT with or without VBT. The majority of patients (95.8%) received no chemotherapy.

Table 1.

Characteristics of patients with node-negative endometrial cancer (n = 335)

Characteristic	Patients, n (%)
Age (y), median (IQR)	63.6 (58.3–70.7)
FIGO grade
1	211 (63.0)
2	100 (29.9)
3	24 (7.2)
MMI
<50%	226 (67.5)
≥50%	109 (32.5)
Extent of LVSI
None	211 (63.0)
Focal	65 (19.4)
Substantial	59 (17.6)
Adjuvant radiation therapy
No radiation therapy	169 (50.4)
Vaginal brachytherapy only	133 (39.7)
EBRT	23 (6.9)
Unavailable	10 (3.0)
Adjuvant chemotherapy
No chemotherapy	321 (95.8)
Chemotherapy	4 (1.2)
Unavailable	10 (3.0)

Abbreviations: EBRT = external beam radiation therapy; FIGO = International Federation of Gynecology and Obstetrics; IQR = interquartile range; LVSI = lymphovascular space invasion; MMI = myometrial invasion.

Adjuvant radiation

We next examined the relationship between LVSI and adjuvant radiation. Most patients with no LVSI had no adjuvant radiation (75.1%), and most patients with focal LVSI received vaginal brachytherapy alone (81.0%). In patients with substantial LVSI, 57.9% of patients received vaginal brachytherapy alone and 31.6% of patients received EBRT with and without vaginal brachytherapy (P < .001; Table 2). Our institutional practice during this time was to observe patients with stage IA grade 1–2 disease and offer vaginal brachytherapy for patients with stage IA grade 3 or stage IB grade 1–2 disease, and we would consider EBRT with brachytherapy boost in patients with substantial LVSI with >3 vessels involved or anyone with stage IB grade 3 disease.

Table 2.

Extent of LVSI and administration of adjuvant radiation for patients with node-negative endometroid-type endometrial cancer

	Adjuvant radiation, n (%)
	None n = 205	Vaginal BT n = 63	EBRT ± vaginal BT n = 57	P value
LVSI				<.001
None	154 (75.1)	49 (23.9)	2 (1.0)
Focal	9 (14.3)	51 (81.0)	3 (4.8)
Substantial	6 (10.5)	33 (57.9)	18 (31.6)

Abbreviations: BT = brachytherapy; EBRT = external beam radiation therapy; LVSI = lymphovascular space invasion.

Clinical outcomes

With a median follow-up of 25.8 months (interquartile range, 19.2–34.7 months), LR-DFS, DM-DFS, and OS were 93.3%, 94.6%, and 97.8% for the entire cohort, respectively. For no LVSI, focal LVSI, and substantial LVSI, the 2-year LR-DFS rates were 92.5%, 98.0%, and 91.4%, respectively (P = .3451, Fig. 1). For no LVSI, focal LVSI, and substantial LVSI, the 2-year DM-DFS rates were 95.5%, 93.3%, and 93.8%, respectively (P = .9732, Fig. 2). The 2-year rates for OS were 97.6%, 98.1%, and 98.2% for no LVSI, focal LVSI, and substantial LVSI, P = .879 (data not shown). Univariate analysis was performed on known prognostic factors and did not demonstrate association with LR-DFS, DM-DFS, or OS (Table 3 for LR-DFS and DM-DFS). There were 10 patients with isolated tumor cells (ITC) and no recurrences among this cohort, but this small number of patients does not provide meaningful data in this subset.

Fig. 1.

LR-DFS for node-negative patients with none, focal, and substantial LVSI. Abbreviations: LR-DFS = locoregional disease-free survival; LVSI = lymphovascular space invasion.

Fig. 2.

DM-DFS for node-negative patients with none, focal, and substantial LVSI. Abbreviations: DM-DFS = distant metastasi disease-free survival; LVSI = lymphovascular space invasion.

Table 3.

Univariate analysis of predictors for LR-DFS and DM-DFS (n = 319)

Univariate analysis for LR-DFS
Characteristic	LR-DFS (−), n (%)	LR-DFS (+), n (%)	HR for LR-DFS (+) (95% CI)	P value
FIGO grade				.880
1	194 (95.1)	10 (4.9)	1 (reference)
2	87 (94.6)	5 (5.4)	0.711 (0.16–3.25)
3	21 (91.3)	2 (8.7)	0.659 (0.13–3.4)	.548
MMI
<50%	204 (95.3)	10 (4.7)	1 (reference)
≥50%	98 (93.3)	7 (6.7)	0.744 (0.28–1.96)
Age (y)				.649
≤60	92 (93.9)	6 (6.1)	1 (reference)
>60	210 (95.0)	11 (5.0)	0.794 (0.29–2.15)
LVSI				.401
None	188 (94.0)	12 (6.0)	1 (reference)
Focal	61 (98.4)	1 (1.6)	0.262 (0.03–2.02)
Substantial	53 (93.0)	4 (7.0)	1.145 (0.37–3.55)
Adjuvant radiation				.603
None	151 (94.4)	9 (5.6)	1 (reference)
Vaginal BT	122 (95.3)	6 (4.7)	0.688 (0.24–1.93)
EBRT	19 (90.5)	2 (9.5)	1.472 (0.32–6.82)
Adjuvant				.724
chemotherapy
None	288 (94.4)	17 (5.6)	1 (reference)
Received	4 (100)	0 (0)	0.049 (0.00–9.7 × 105)
Univariate analysis for DM-DFS
Characteristic	DM-DFS (−), n (%)	DM-DFS (+), n (%)	HR for DM-DFS (+) (95% CI)	P value
FIGO grade				.993
1	194 (95.1)	10 (4.9)	1 (reference)
2	87 (94.6)	5 (5.4)	0.938 (0.32–2.75)
3	23 (100)	0 (0)	0.000 (0.00-N/A)
MMI				.538
<50%	203 (94.9)	11 (5.1)	1 (reference)
≥50%	101 (96.2)	4 (3.8)	0.698 (0.22–2.19)
Age (y)				.776
≤60	94 (95.9)	4 (4.1)	1 (reference)
>60	210 (95.0)	11 (5.0)	1.181 (0.38–3.71)
LVSI				.973
None	191 (95.5)	9 (4.5)	1 (reference)
Focal	59 (95.2)	3 (4.8)	1.059 (0.29–3.92)
Substantial	54 (94.7)	3 (5.3)	1.167 (0.32–4.31))
Adjuvant radiation				.620
None	151 (94.4)	9 (5.6)	1 (reference)
Vaginal BT	123 (96.1)	5 (3.9)	0.581 (0.20–1.74)
EBRT	20 (95.2)	1 (4.8)	0.723 (0.09–5.71)
Adjuvant				.741
chemotherapy
None	290 (95.1)	15 (4.9)	1 (reference)
Received	4 (100)	0 (0)	0.049 (0.00–2.9 × 106)

Abbreviations: BT = brachytherapy; CI = confidence interval; DM-DFS = distant metastasis disease-free survival; FIGO = International Federation of Gynecology and Obstetrics; HR = hazard ratio; LR-DFS = locoregional disease-free survival; LVSI = lymphovascular space invasion; MMI = myometrial invasion; N/A = not applicable.

To investigate the interaction of radiation in patients with substantial LVSI, we compared patients who received no radiation or vaginal brachytherapy alone to those that received EBRT. When comparing patients receiving no radiation or vaginal brachytherapy versus patients receiving EBRT, the 2-year LR-DFS was 93.0% versus 87.8% (P = .473), 2-year DM-DFS was 93.5% and 94.1% (P = .990), and 2-year OS was 98.4% and 100% (P = .522), respectively.

Discussion

The results from our single institution study demonstrated that patients with stage I endometrioid-type endometrial cancer with pathologically negative LNs and substantial LVSI have similar rates of LR-DFS, DM-DFS, and OS compared with patients with none or focal LVSI. In our cohort, patients with substantial LVSI were more likely to receive EBRT compared with patients with none or focal LVSI. However, there was no difference in DM-DFS when comparing these 2 groups (Table E1), which contrasts with the pooled PORTEC analysis.⁵ The relatively high rates of LR-DFS, DM-DFS, and OS in this patient population compared with PORTEC is likely partially due to stage migration from patients with LN metastases being excluded. These results highlight the difficulty in making clinical decisions about adjuvant therapy in this specific patient population with substantial LVSI and negative LN evaluation.

In the pooled PORTEC-1 and -2 analysis, patients with substantial LVSI had a 5-year pelvic regional recurrence risk of 15.3%, compared with 1.7% for patients with no LVSI and 2.5% for patients with focal LVSI. Patients with substantial LVSI appeared to benefit from EBRT with 5-year pelvic regional recurrence risk of 4.3% compared with 27.1% to 30.7% in patients receiving no adjuvant EBRT or vaginal brachytherapy.⁵ PORTEC-1 and -2 were completed without routine LN dissection and only suspicious LNs were removed at the time of surgery.^3,4 The PORTEC-1 and -2 findings have been confirmed in multiple retrospective studies. In patients who met PORTEC-2 high intermediate risk inclusion criteria treated with surgery followed by vaginal brachytherapy alone, substantial LVSI was shown to be a predictor of pelvic and distant failure, with 4-year disease-free survival for patients with substantial LVSI of 73% compared with 89% and 94% for those with focal and no LVSI, respectively.⁷ Notably, only 34% of these patients had LN assessment. In a subset analysis of 111 patients with nodal assessment, there was only a 3% rate of pelvic failure and no factors were found to be predictive for pelvic failure including 3-tier LVSI scoring.⁷ In contrast, in the 214 patients who did not have nodal assessment substantial LVSI was found to be predictive for pelvic failure.⁷ In another study using the Danish Gynecological Cancer Database, patients meeting the ESMO-ESGO-ESTRO 2016 high risk endometrial carcinoma definition were reviewed with respect to the 3-tier LVSI scoring system.⁸ Substantial LVSI was present in 6.2% of these patients and 62% of patients had lymphadenectomy. In stage I/II patients, substantial LVSI was a risk factor for overall recurrence, pelvic/para-aortic LN recurrence, and distant LN recurrence, but not for local recurrence or distant metastasis.⁸ The subgroup of patients who had stage I/II disease with pathologically negative LN had similar rates of recurrence irrespective of LVSI status.⁸ In our previous publication, among all patients with pT1a-b disease, 3.5% had LN involvement, whereas among patients with substantial LVSI, 10.6% of patients had LN involvement. This highlights the correlation between substantial LVSI and LN involvement and may indicate significant implications for patients who did not have LN evaluation.

The retrospective nature of this work and the fact that these patients were treated at a single institution should be noted. The initial cohort consisted of all patients with stage I disease identified after complete surgical staging between 2017 and 2019. Although the majority had LN assessment, we cannot exclude the possibility of patient factors confounding who received LN evaluation. Nonetheless, our results correlate well with previous publications on LN evaluation. It should be noted that the 2-year follow-up presented in this work represents intermediate follow-up for patients with early-stage low-grade endometroid endometrial cancer, and delayed recurrence may still occur. Also, the relative low rate of recurrence in this population highlights the importance of multi-institutional studies in this area to examine this question with more patients. We acknowledge the differences in adjuvant radiation treatment modalities (vaginal brachytherapy versus EBRT with or without vaginal brachytherapy) in these patients with substantial LVSI, which was our practice pattern at that time.

Our study represents the largest retrospective study of patients with stage I endometrial cancer with LN evaluation with regards to substantial LVSI. In conjunction with other studies examining substantial LVSI, it highlights the discrepancy in prognostic value of substantial LVSI, specifically in the presence or absence of LN evaluation. PORTEC-4a randomized patients with substantial LVSI to EBRT, has now finished accrual, and we now wait for data to mature.⁶ At our institution, we saw a ~4 times higher rate of substantial LVSI than in the PORTEC experience.⁹ Robotic-assisted surgeries also makes pathology review more difficult in determining LVSI because these surgeries have higher rates of procedural artifact (ie, crush artifact), which complicates the interpretation of LVSI. These factors would have resulted in a significant increase in patients recommended to receive EBRT. There is now also a consensus for the definition of substantial LVSI, requiring 4 or more vessels in at least 1 hematoxylin and eosin slide to be involved.¹⁶ These new recommendations will potentially result in more consistent reporting among different institutions. Our current practice pattern of offering EBRT to patients with LVSI involving 4 or more vessels is consistent with PORTEC-4a, although the effect of adjuvant EBRT in the setting of nodal assessment is yet to be ascertained.

Conclusions

In patients with stage I endometrioid-type endometrial cancer who have pathologically LN negative disease, we noted similar clinical outcomes between patients with no, focal, or substantial LVSI. These results likely demonstrate that there is a prognostic difference between patients with substantial LVSI who did and did not have LN evaluation and stresses the importance of LN assessment in patients with substantial LVSI to guide treatment and predict prognosis. Overall, our data highlight the need for multi-institutional studies examining this population.