Is it useful to detect lymphovascular invasion in lymph node-positive patients with primary operable breast cancer?




Lymphovascular invasion (LVI) is a widely recognized prognostic factor in lymph node-negative breast cancers. However, there are only limited and controversial data about its prognostic significance in lymph node-positive patients.


Among 931 patients operated on and monitored at the authors' institution for an invasive breast carcinoma between 1989 and 1992, all 374 lymph node-positive breast cancers entered the study (median follow-up, 126 months).


LVI was present in 46% of tumors and was associated with age ≤40 years (P = .02), high histological grade (P = .01), and negative estrogen receptor status (P = .032), but not with tumor size, number of involved lymph nodes, or HER-2/neu status. LVI was an independent prognostic factor for distant metastases (P = .002). Furthermore, in HER-2/neu–negative/hormone receptor-positive (n = 287) tumors, the number of independent prognostic factors (LVI, age, histological grade, number of involved lymph nodes, and tumor size) was associated with a 5-years metastasis-free survival ranging from 100% if no factors (n = 25) to 89% ± 2% if 1 or 2 factors (n = 186) and 67% ± 6 if 3, 4, or 5 factors (n = 76) were present (P < .001).


LVI is an independent prognostic factor in lymph node-positive breast cancer and merits further prospective investigations as a decision tool in the adjuvant chemotherapy setting. Cancer 2010. © 2010 American Cancer Society.

One of the more important challenges in the management of early stage breast cancer patients is determining those who could benefit from adjuvant chemotherapy and for whom a chemotherapy regimen could provide the best outcome in terms of risk/benefit and cost/efficacy ratios. The increasing knowledge of breast cancer molecular biology and the perfecting of large-scale molecular tools have led to the development of genomic predictors of relapse such as OncotypeDx recurrence score and the 70-gene prognosis signature, which are being prospectively tested in 2 large randomized trials (TAYLOR and MINDACT). At the present time, for several worldwide clinicians, the decision regarding whether to use adjuvant chemotherapy in breast cancer patients is still mainly based on 5 traditional prognostic factors: lymph node status, age, tumor size, estrogen receptor (ER) status, and histological grade. Although the prognostic value of lymphovascular invasion (LVI) has been recognized in lymph node-negative patients with breast cancer,1 there are only limited and controversial data about its prognostic significance in lymph node-positive patients.2-5 We report here the longest-term analysis of the prognostic value of LVI in a large cohort of patients with lymph node-positive breast cancer.



Between 1989 and 1992, 931 patients with primary operable invasive breast carcinoma were operated on and monitored at our institution. All these patients had been prospectively included by senior pathologists (I.M., G.M., I.S.) in our clinical, biological and histological database (65 pathological items for noncancerous lesions and 181 for cancers), and a prognostic tissue microarray block was built (G.M.).6 Ninety patients were excluded from the study: 76 with a previous (n = 59) or a concurrent contralateral (n = 17) breast carcinoma and 14 with too many missing data. There were 467 lymph node-negative and 374 lymph node-positive patients. The clinicopathological data and treatments of the 374 lymph node-positive patients included are summarized in Tables 1 and 2. Patients were treated with surgery by either modified radical mastectomy (101 cases) or local tumor resection (273 cases), with axillary lymph node dissection followed by postoperative breast irradiation (288 cases). Adjuvant therapy with chemotherapy and/or hormone therapy was decided according to lymph node status, hormone receptor results, and menopausal status of the patients. At that time, adjuvant chemotherapy (anthracycline based) was standard only for nonmenopausal women, and tamoxifen was usually prescribed to hormone receptor-positive (estrogen- and/or progesterone-positive) postmenopausal women or women with chemotherapy-induced amenorrhea. Of course, no patient received adjuvant trastuzumab. All patients were observed every 3 months for 2 years, twice a year for the next year, and then once a year.

Table 1. Clinicopathological Data of Patients Included
Patient CharacteristicNo. (%)LVI Absent, n=203, %LVI Present, n=171, %P
  • LVI indicates lymphovascular invasion; mSBR, modified Scarff-Bloom-Richardson; ER, estogen receptor; PR, progesterone receptor.

  • a

    Assessed on tissue microarray. ER and PR: <10% of positive tumor cells = negative; ≥10% of positive tumor cells = positive. Mib1: <20% of positive cells = low; ≥20% = high. Score on Herceptest: 10%, intensity 1-3; 0% and 1 = no overexpression; ≥10% and 2 or 3 = overexpression.

Age, y    
 ≤4034 (9)3565.02
 >40340 (91)5644 
Histological tumor size, mm    
 ≤1029 (8)7228.55
 10-20215 (57)5347
 >20124 (33)5248
 Not specified6 (2)8317
Mitotic count    
 Score 1100 (27)6337.12
 Score 2141 (38)5149
 Score 3133 (36)5149
mSBR grade    
 I76 (20)6733.01
 II181 (48)5545
 III117 (31)4555
Lymph node status    
 1 N+163 (44)5842.22
 2 N+76 (20)5545
 3 N+38 (10)5842
 ≥4 N+97 (26)4555
Estrogen receptorsa    
 Negative90 (24)4456.032
 Positive284 (76)5743
Progesterone receptorsa    
 Negative134 (36)5050.184
 Positive238 (64)5743
Hormonal receptorsa    
 ER and PR negative69 (18)4555.056
 ER and/or PR positive305 (82)5644
HER-2 overexpressiona    
 <10% or intensity 1329 (88)5644.06
 ≥10% and intensity 2 or 344 (12)4159
 Not specified1 (<1)1000 
Ki-67/Mib1 proliferation indexa    
 <20%254 (68)5743.08
 ≥20%118 (32)4753 
 Not specified2 (1)1000 
Table 2. Locoregional and Systemic Adjuvant Therapies
TherapyNo.LVI Absent, n=203, No. (%)LVI Present, n=171, No. (%)P
  1. LVI indicates lymphovascular invasion; CT, chemotherapy; HR, hormone receptor.

Locoregional treatment   .53
 Mastectomy alone8647 (23)39 (23) 
 Mastectomy + irradiation156 (3)9 (5) 
 Lumpectomy + irradiation273150 (74)123 (72) 
Adjuvant systemic treatment, n=374    
 CT ± endocrine treatment126/37454/203 (27)72/171 (42).001
 Endocrine treatment ±CT [HR+, n = 304]215/304129/172 (75)86/132 (65).06
 None35/37417/203 (8)18/171 (11).47
Non menopausal women, n=129    
 Chemotherapy, n=129108/12948/54 (89)60/75 (80).17
 Endocrine treatment [HR+, n=95]26/958/41 (20)18/54 (33).13
 None5/1291 (2)4 (5)
Menopausal women, n=245    
 Chemotherapy, n=24518/2456/149 (4)12/96 (12).01
 Endocrine treatment [HR+, n=209]189/209121/131 (92)68/78 (87).22
 None30/24516 (7)14 (8).91

Tissue Sample and Histological Examination

After fixation in Hollande-Bouin, the tumor was macroscopically sectioned, and each slice was put in as many separate cassettes as necessary and paraffin-embedded. In total, ≥3 blocks per tumor were obtained in 369 cases (99.2%) (median number, 3). Each block was examined on 1 hematoxylin and eosin (H & E) section/slide. The diameter of the tumor (histological size) was measured on fresh specimens and on H & E slides. A modified Scarff-Bloom-Richardson grading system grade was used.7 The mean number of lymph nodes examined per case was 14.75 (range, 1-31). Each lymph node was macroscopically cut entirely into 1- to 1.5-mm-thick slices perpendicular to the long axis (1-9 slices; mean, 4). All slices from a single lymph node were placed together in as many numbered cassettes as necessary and were embedded in paraffin. The number of cassettes (paraffin blocks) required to analyze each entire lymph node ranged from 1 (90% of the cases) to 3. Each block was examined on 1 H & E slide.

Hormone receptor status was originally determined on a fraction of fresh tumor by the dextran-coated charcoal (DCC) method, with cutoff values of 10 and 15 fmol/mg protein for ER and progesterone receptor, respectively. Tumor size, mitotic count, modified Scarff-Bloom-Richardson grade, LVI, and lymph node status were systematically and prospectively entered into a computer database for each tumor. To achieve the analysis of prognostic factors, an additional immunohistochemical (IHC) analysis was retrospectively performed on a tissue microarray.6 Four tissue cores, 0.6 mm in diameter, were obtained from each donor block and mounted in a recipient block using a tissue microarray instrument according to the manufacturer's instructions (Alphelys, Plaisir, France). Tissue sections were incubated for 60 minutes with ER antibody (DakoCytomation, Trappes, France; 1D5; dilution, 1:100), progesterone receptor (PgR) antibody (DakoCytomation; PgR636; dilution 1:100), Mib1 antibody (DakoCytomation; Mib1; dilution 1:100), and HER-2 antibody (DakoCytomation; DA485; dilution 1:2000). ER and PgR expressions were evaluated semiquantitatively, with cutoff values of 10% positive invasive tumor cells. For Mib1, a cutoff of 20% positive tumor cells, corresponding to the 75th percentile in the series, was chosen. HER-2 expression was evaluated semiquantitatively according to a standard protocol (Herceptest, DakoCytomation). Considering hormone receptor status, a concordance between the 2 methods of analysis (DCC and IHC) was observed in 92% of cases, with potential impact on adjuvant treatment decision for only 7 postmenopausal patients.

Definition of Lymphovascular Invasion

LVI was prospectively assessed on original H & E-stained slides as defined by Rosen and Oberman8; LVI was defined as carcinoma cells present within a definite endothelial-lined space, at a distance from the tumor (Fig. 1), in the breast surrounding invasive carcinoma, that is, separated from the tumor by normal breast tissue and often next to a small artery or vein.1, 8, 9 When H & E LVI was suspected in the vicinity of the tumor, differential diagnosis with an artifact (retraction) was difficult, and such equivocal cases were recorded as negative. In our study, only unequivocal H & E LVI was prospectively recorded as positive without distinguishing blood vessels from lymphatics. Although H & E emboli were not counted, they were single in most cases.

Figure 1.

Lymphovascular invasion is shown: (Top) H & E; original magnification, ×200; (Bottom) H & E; original magnification, ×400.

Statistical Analysis

The correlations between LVI and patients' characteristics were studied with the chi-square and Wilcoxon tests. Metastasis-free survival rates were calculated according to the Kaplan-Meier method from the date of surgery to the occurrence of distant metastasis. Overall survival (OS) was calculated from the date of surgery to death or last-known-alive date. All causes of death were considered as events. We compared metastasis-free survival and OS between the patients with and without vascular invasion using the log-rank test (SPSS v14, SPSS Inc., Chicago, Ill). Multivariate analysis was performed stepwise using the Cox regression model. Analyses were performed in the whole group and in the HER-2–negative tumors according to the hormone receptor status.

For multivariate analysis, 366 of 374 and 324 of 329 medical records were available in the entire and in the HER-2–negative groups, respectively.



The median follow-up was 150 months (range, 136.7-155). The total number of distant metastases with or without locoregional recurrences was 143 (38%). At the time of analysis, 115 (31%) of 374 patients had died from breast cancer; OS at 5 and 10 years were 84.2% and 67.7%, respectively; metastasis-free survival was 77.3% and 64.8% at 5 and 10 years, respectively.

Correlation of Lymphovascular Invasion With Clinical and Histological Criteria

LVI was found in 171 (46%) of 374 patients. It was significantly more frequent in patients ≤40 years old (P = .02) and in high modified Scarff-Bloom-Richardson grade (P = .01) or negative ER (P = .03) tumors. No correlation was found with tumor size, mitotic count, proliferation index, or number of involved lymph nodes (Table 1). Considering treatments, patients with LVI+ tumors more frequently received adjuvant chemotherapy (42% vs 27%, P = .04). Locoregional and adjuvant endocrine treatments were not significantly different between the 2 groups (Table 2).

Prognostic Value of Lymphovascular Invasion for Metastasis-Free Survival

At 10 years, both metastasis-free survival and OS were shorter in patients with LVI: 56.1% versus 74.4% (P = 3×10−4) for metastasis-free survival (Fig. 2, Top) and 58.5% versus 76.4% (P = 2×10−5) for OS (Fig. 2, Bottom).

Figure 2.

(Top) Metastasis-free survival curves are shown according to presence (+) or absence (−) of lymphovascular invasion (LVI). (Bottom) Overall survival curves according to presence or absence of LVI are shown.

A worse metastasis-free survival was also associated with young age (<40 years), tumor size (>20 mm), mitotic count (score of 2 or 3), proliferation index (≥20%), modified Scarff-Bloom-Richardson grade (II or III), lymph node status, hormone receptor status (both estrogen and progesterone negative), and HER-2 status (positive) (Table 3).

Table 3. Univariate Analysis of MFS and Clinicopathological Criteria
Patient CharacteristicNo.MFS 5 Years, %MFS 10 Years, %P
  • MFS indicates metastasis-free survival; LVI, lymphovascular invasion; mSBR, modified Scarff-Bloom-Richardson.

  • a

    Assessed on tissue microarray. Estrogen receptor and progesterone receptor: <10% of positive tumor cells = negative; ≥10% of positive tumor cells = positive. Mib1: <20% of positive cells = low; ≥20% = high. Score on Herceptest: 10%, intensity 1-3; 0% and 1 = no over expression; ≥10% and 2 or 3 = overexpression.

Lymphovascular invasion    
 ≤40 years3455.941.21×10−3
 >40 years34079.768.5 
Histological tumor size    
 ≤10 mm (pT1a-b)2910089.71×10−5
 10-20 mm (pT1c)21580.972.6 
 >20 mm (pT2-3)12467.750.8 
Mitotic count    
 Score 110093.985.98×10−6
 Score 214173.861.0 
 Score 313369.256.4 
mSBR grade    
Nodal status    
 1 N+16381.575.58×10−9
 2 N+7689.576.3 
 3 N+3876.360.5 
 ≥4 N+9761.944.3 
Estrogen receptorsa    
Progesterone receptorsa    
Hormonal receptorsa    
 Estrogen and progesterone6949.346.43×10−6
 Estrogen and/or progesterone+30483.970.7 
HER-2 overexpressiona    
 <10% or intensity 132981.569.65×10−7
 ≥10% and intensity 2 or 34447.740.9 
Ki-67/Mib1 proliferation indexa    

By multivariate analysis, LVI was an independent significant predictor of distant recurrence in the whole group of patients (n = 366) (relative risk [RR], 1.70; 95% confidence interval [CI], 1.21-2.40; P = .002), as were high number of involved lymph nodes (≥3; RR, 2.61%; 95% CI, 1.86-3.66; P = 3×10−8), large tumor size (>20 mm; RR, 2.01%; 95% CI, 1.44-2.82; P = 5×10−5), HER-2/neu overexpression (RR, 2.35; 95% CI, 1.53-3.61; P = 9×10−5), and high proliferative index (≥20%; RR, 1.73; 95% CI, 1.23-2.44; P = .002).

Furthermore, LVI was also an independent predictor of poor outcome in the HER-2/neu–negative and in the HER-2/neu–negative/hormone receptor-positive subgroups of patients (RR, 1.82; 95% CI, 1.24-2.67; P = .002 and RR, 1.71; 95% CI, 1.12-2.62; P = .013, respectively). When including systemic treatment in the final multivariate analysis, a significant value was observed for both chemotherapy and endocrine treatment. However, the impact of others factors, especially LVI, was not significantly modified (data not shown).

In the HER-2/neu–negative/hormone receptor-positive group (n = 287), age ≤40 years, high Scarff-Bloom-Richardson grade (2 or 3), high number of involved lymph nodes (≥3), and large tumor size (>20 mm) were also found to be independent predictors of poor outcome (Table 4). Furthermore, in this subgroup of patients, the 5-year metastasis-free survival was correlated (log-rank test, P = 10−12 Fig. 3) with the number of predictors of poor outcome: 100% if none (n = 25), 89% ± 2% if 1 or 2 (n = 186), and 67% ± 6 if 3, 4, 5 factors (n = 76). Only 1 late (after 103 months) distant recurrence was observed among the 25 patients with none of these prognostic factors (Table 5).

Figure 3.

Metastasis-free survival is shown in the 287 patients with HER-2/neu–negative/hormone receptor-positive tumors according the presence of unfavorable factors (≥3 involved lymph nodes, modified Scarff-Bloom-Richardson grade 3 or 2, age ≤40 years, tumor size >20 mm, and the presence of lymphovascular invasion).

Table 4. Metastasis-Free Survival: Multivariate Analysis in the HER-2–Negative/Hormone Receptor-Positive Subgroup (n=287)
Patient CharacteristicRelative Risk95% CIP
  1. CI indicates confidence interval; mSBR, modified Scarff-Bloom-Richardson; LVI, lymphovascular invasion.

≥3 involved lymph nodes2.731.80-4.152×10−6
mSBR grade III2.621.27-5.389×10−3
Age ≤40 years at time of diagnosis2.411.24-4.669×10−3
Tumor size >20 mm1.731.13-3.64.01
mSBR grade II2.021.08-3.81.03
Table 5. Systemic Adjuvant Treatment of the 3 Subgroups of Patients With HER-2–Negative/Hormone Receptor-Positive Tumors (n=287) According to the Number of Significant Prognostic Factors
Factor01 or 23, 4, or 5Total
  1. LVI indicates lymphovascular invasion; LVI, lymphovascular invasion; MFS, metastasis-free survival; CI, confidence interval.

Number of patients2518676287
Number of LVI + (%)061 (33%)60 (79%)121 (42%)
5-year MFS, 95% CI100%89%±2%67±6%84±2%
No systemic treatment2 (8.0%)6 (3.2%)1 (1.3%)9 (3.1%)
Endocrine treatment19 (76.0%)127 (68.3%)47 (61.8%)193 (67.2%)
Chemotherapy4 (16.0%)34 (18.3%)23 (30.3%)71 (24.7%)
Both chemotherapy and endocrine treatment09 (4.8%)5 (6.6%)14 (4.9%)

No prognostic analysis was performed in the small (n = 69) group of patients with hormone receptor-negative tumors.


In this retrospective analysis, LVI prognosis was assessed, as LVI has been prospectively registered in our database at the time of surgery. By using H & E, LVI was observed in 46% of cases, with a frequency similar to other series of lymph node-positive tumors.2-5 Recently, it has been shown that immunostaining, especially with D2-40, may be useful to detect missed H & E LVI.10, 11 In a previous work on lymph node-negative tumors, we also observed that D2-40 increases the diagnostic sensitivity of LVI by a factor of 1.4, and strengthens the prognostic impact of this factor.9 As only peritumoral LVI has prognostic significance, this analysis cannot be easily performed on tissue microarray blocks. It may be anticipated that such complementary analysis would also give similar results in lymph node-positive tumors.

With a relative risk of 1.71 having a P value of .01 in multivariate analysis, our study clearly demonstrates that LVI adds prognostic information in patients with lymph node-positive breast cancer. Previously, 3 other studies also found an adverse prognostic significance of LVI in lymph node-positive tumors, with a slightly less marked hormone receptor.2-4 These results contrast with those reported by Colleoni et al, showing a statistically significant impact of LVI in terms of OS and disease-free survival (DFS) in the population of patients with lymph node-negative disease but not in the population of patients with lymph node-positive cancer.5 This discordance may be explained by the shorter follow-up in the study of Colleoni (3.8 years for DFS and 4.3 years for OS) and by a restriction of the analysis to patients with 1 to 3 metastatic lymph nodes. Furthermore, in this analysis, DFS of all lymph node-positive tumors looked surprisingly good and identical to patients with lymph node-negative, LVI-negative tumors, whereas prognosis of lymph node-negative status with extensive LVI tumors was worse. Neither was significant prognostic value observed for grade. Recently, in a large Danish cohort of breast cancer, LVI prognosis was assessed in 2 complementary subgroups defined by risk of systemic relapse. A significant prognostic significance for DFS and OS was observed only in the high-risk group defined by 1 of the following factors: age <35, grade II (ductal) or III, size >20 mm, or for negative hormone receptor and positive lymph nodes. For those patients, the negative prognostic impact of LVI for DFS and OS seemed to be identical in patients who were lymph node negative, positive in 1 to 3 lymph nodes, and positive in >3 lymph nodes.12

Colleoni et al suggested that it may be of importance to score the extent of LVI.5 However, in their analysis, LVI scoring gave supplementary information only within the subgroup of patients with negative sentinel lymph node biopsy. Whereas an unfavorable value was observed for extensive LVI, no trend effect was observed according to the LVI score as described by the Milan team. Unfortunately, as LVI was not prospectively quantified in our database, this factor could not be quantitatively assessed. Only 1 peritumoral vascular emboli was observed in most cases.

In addition to LVI, other significant independent pathological factors in our study are tumor size, modified Scarff-Bloom-Richardson grade, and lymph node involvement, all 3 factors constituting the Nottingham Prognostic Index. As LVI appears to be partially independent from these factors, our results suggest that this factor should be taken into account in the Nottingham Prognostic Index, as it has been previously suggested for lymph node-negative breast cancer.13 Conversely, the correlation that we and others observed5, 14 between LVI and phenotypic characteristics such as histological grade, absence of estrogen expression, and HER-2/neu status but also age (with a cutoff of 40 years, but also when analyzed in quintiles; data not shown) suggests that LVI may be integral to the tumor genotype rather than acquired with tumor progression.

We then assessed whether these prognostic factors may help to identify a subset of patients for whom systemic adjuvant chemotherapy could be avoided despite lymph node involvement. This analysis was restricted to the large subgroup of 287 patients with HER-2/neu–negative/hormone receptor-positive tumors who benefit from endocrine treatments and for whom chemotherapy may be contraindicated because of less chemosensitivity.15 Multivariate analysis distinguished 5 unfavorable independent factors that were combined for identification of different subgroups of patients according to the number of unfavorable factors (Table 5). Only 1 late relapse was observed in the subgroup of 25 patients without any unfavorable factor, and 5-year metastasis-free survival was 91% ± 3% for the 94 patients with 1 factor (data not shown). These results should be considered with caution, as 16% and 20% of these patients had received adjuvant chemotherapy, respectively (Table 5). Conversely, about 30% did not receive actual standard endocrine treatment, although their tumors were hormone receptor-positive. However, these results suggest that LVI and other prognostic factors may identify some patients for whom chemotherapy may be contraindicated despite lymph node involvement. This could be all the more interesting in elderly or frail patients and/or in patients with poor tumor chemosensitivity predictive factors, that is, with high hormone receptor positivity and low proliferation.

In conclusion, this analysis reinforces the importance of pathologic analysis in the prognostic evaluation of breast cancer, especially those factors that cannot be analyzed by molecular analyses, such as size, lymph node status, and LVI. Conversely, it seems that molecular analyses such as Oncotype DX and Mammaprint 70-gene signature may be more effective for evaluation of tumor aggressiveness, especially proliferation, and 2 retrospective large analyses have recently underlined that classical pathological and molecular diagnoses include different and complementary information.16, 17 In our opinion, these techniques should be combined rather than set against each other, to improve evaluation of prognostic and predictive factors for a more efficient, cost-effective, and convenient clinical decision-making process in the adjuvant setting.18 For these reasons, tumor size, lymph node status, and LVI will remain cornerstones of the description of a breast carcinoma.


The authors made no disclosures.