The prognostic significance of lymphovascular invasion in invasive breast carcinoma




Although lymphovascular invasion (LVI) has been associated with a poor outcome in patients with breast cancer, it is not included in most internationally recognized staging systems, including the American Joint Committee on Cancer tumor, lymph node, metastasis (TNM) classification. This is mainly because it remains unclear whether the presence of LVI is an independent, high-risk criterion in clinically relevant staging subgroups.


The current study was based on a large and well characterized consecutive series of patients who had operable (pathologic T1 [pT1]-pT2, pathologic N0 [pN0]-pN3, M0) breast cancer (3812 informative cases) who were treated according to standard protocols at a single institution and who had long-term follow-up to assess the prognostic value of definite LVI in clinically and molecularly relevant staging subgroups.


LVI was strongly associated with both breast cancer-specific survival (BCSS) and distant metastasis-free survival (DMFS) in the entire series and in different subgroups. Multivariate analyses identified LVI as an independent predictor of both BCSS and DMFS in patients with operable breast cancer overall; in the TNM clinical subgroups pT1a-pT1c/pN0 and pT2/pN0; and in the molecular classes estrogen receptor (ER)-positive, ER-negative, human epidermal growth factor 2 [HER2]-negative, and triple-negative. In patients who had lymph node-negative tumors, LVI could be used as a high-risk criterion providing survival disadvantage equivalent to that provided by 1 or 2 involved lymph nodes (pN0 to pN1) and to that provided by 1 size category (pT1 to pT2). The use of immunohistochemistry for detecting an endothelial-specific marker contributed to the prognostic significance of LVI when applied to routine LVI negative/possible cases.


LVI provided a strong predictor of outcome in patients with invasive breast cancer and should be incorporated into breast cancer staging systems. Cancer 2012. © 2011 American Cancer Society.


Lymphovascular invasion (LVI), which refers to the invasion of lymphatic spaces, blood vessels, or both in the peritumoral area by tumor emboli, is 1 of the critical steps in metastasis. The prognostic value of LVI in breast cancer was described first more than 4 decades ago.1 Subsequently, several independent studies have demonstrated a clear relation between LVI and outcome in patients with negative lymph node (LN) status,2-17 and, with some controversy, in patients with positive LN status.4, 15, 18 Routine assessment of LVI is now part of the minimum data set for breast cancer pathology reporting produced by the United Kingdom Royal College of Pathologists,19 the European Commission,20 and College of American Pathologists.21 It is endorsed by the World Health Organization,22 the St. Galen Consensus Conference, and the American Joint Committee on Cancer/International Union Against Cancer (AJCC/UICC) (seventh edition of the TNM classification) as a prognostic factor in patients with breast cancer.

Despite this, currently, LVI is not incorporated into most of internationally recognized staging systems, such as the AJCC/UICC TNM staging system,23 prognostic algorithms and indices (eg, Adjuvant!Online,24 Predict,25 and the Nottingham Prognostic Index [NPI]26), and guidelines (eg, the North Central Cancer Network breast cancer treatment guidelines27 and the American Society of Clinical Oncology breast guidelines), to determine the use of adjuvant chemotherapy. This can be explained by 1 or more of the following: 1) most studies of LVI have been small or only used univariate analysis; 2) variation with regard to distinguishing lymphatic invasion from blood vessel invasion4, 13, 28, 29 or consideration of the extent of LVI16; 3) the use of different methods for the detection of LVI, including pathologic routine assessment and immunohistochemistry8, 28, 30, 31; 4) the few studies that included large series of breast cancer either lacked details of the significance of LVI in the different clinically relevant subgroups, did not report its risk value compared with other well established prognostic markers, or reported a lack of significance.15 Subsequently, the use of LVI in clinical management decisions remains a matter of debate. Therefore, the objective of the current study was to assess the prognostic value of LVI in a large and well characterized series of patients with operable breast cancers (pathologic T1 [pT1]-pT2 tumors) who had long-term follow-up and detailed clinicopathologic, molecular, and outcome information. This study also focused on the prognostic information provided by LVI in the different staging and clinically relevant subgroups of breast cancer compared with the other well established prognostic factors.


The study population was derived from the Nottingham Tenovus Primary Breast Carcinoma Series of women aged ≤70 years who presented with primary, operable, invasive breast carcinomas (with tumors that measured <5 cm in greatest dimension on clinical/preoperative measurement and/or on operative histology; pT1 and pT2) between 1989 and 2004. Patient's clinical history and tumor characteristics were obtained from the database and included the method of referral (screening vs symptomatic); patient age; menopausal status; primary tumor size; histologic tumor type22, 32; histologic grade33, 34; LN status; LVI3; estrogen receptor (ER), progesterone receptor, and human epidermal growth factor receptor 2 (HER2) status; and Ki-67 index.35-37

This is a well characterized series of patients treated uniformly in a single institution. A 4-LN axillary sample was the standard procedure.38 LN-negative patients did not receive axillary radiotherapy. Patients who had from 1 to 3 positive LNs were offered radiotherapy, whereas those who had >3 positive LNs were offered axillary clearance. A small proportion of women (28%) received axillary clearance. Adjuvant systemic therapies were offered according to the Nottingham Prognostic Index (NPI)26 group and ER status. The NPI was calculated using the following formula: NPI = histologic grade1-3 + LN status (1-3, where 1 indicates negative; 2, 1-3 positive LNs; 3, ≥4 positive LNs) + (tumor size/cm × 0.2). No systemic therapy was offered to patients in the good prognostic groups (NPI ≤3.4).26 Patients in the moderate I group (NPI, 3.41-4.4) who had ER-positive tumors were offered hormone therapy. Patients in the moderate II (NPI, 4.41-5.4) and poor (NPI, >5.41) groups received hormone therapy for ER-positive tumors and cytotoxic therapy for ER-negative tumors. LVI was not used to guide systemic therapy decisions. Survival data, including length of survival, disease-free survival, and development of distant metastasis (DM), and local and regional recurrences, were maintained on a prospective basis. Patients were followed initially at 3-month intervals, then every 6 months, and annually thereafter for a median of 85 months (range, 5-220 months). At the time of death, the hospital notes were examined and deaths were identified as either “with/from breast cancer” or “without known breast cancer.” Breast cancer specific survival (BCSS) was defined as the interval between surgery and death from breast cancer, with death scored as an event, and patients who died from other causes or who remained alive were censored at the time of last follow-up. Distant metastasis-free survival (DMFS) was calculated from the date of first surgery; the first distant recurrence was scored as an event, and other patients were censored at the time of last follow-up or death. Local recurrence was defined as tumor arising in the treated breast or chest wall. Regional recurrence was defined as tumor arising in the regional LNs (ie, axillary or internal mammary).

Assessment of Lymphovascular Invasion

In this series, LVI was assessed in a consistent way using uniform methods of specimen fixation, sampling, and processing. Tumors were received fresh in the pathology laboratory and were incised immediately to ensure good fixation. LVI was assessed in the peritumoral tissue on hematoxylin and eosin (H&E)-stained sections. At least 3 different tumor sections were examined from each specimen. LVI was defined as carcinoma cells present within a definite, endothelial-lined space (lymphatic and/or blood vessel). Only definite LVI was regarded as positive. Possible LVI (tumor cell emboli observed in a space with the appearance of a vessel but without a recognizable endothelial lining) was scored as negative.2 No attempt was made to distinguish between blood vessels and lymphatics. Recorded data did not distinguish extensive LVI from nonextensive LVI.

In addition, data on immunohistochemical (IHC) detection of LVI using the lymphovascular endothelial-specific marker D2-40 were available in a series of 976 LN-negative cases. The detailed staining technique was described previously.28 Briefly, 1 representative wholemount section from each specimen was stained with podoplanin/D2-40 (polyclonal, 11-003; 1:100 dilution; AngioBio, Del Mar, Calif). Cases in which podoplanin/D2-40 reactivity was observed around tumor cell clusters were considered LVI IHC-positive.

Statistical Analysis

BCSS and DMFS curves were drawn using Kaplan-Meier estimates and were compared using log-rank tests. Survival rates are presented with their 95% confidence intervals (CIs). Multivariate analyses of DMFS and BCSS with stepwise variable selection were conducted using Cox proportional hazards regression models. The clinicopathologic variables were compared using contingency tables and chi-square tests. All P values < .01 were considered significant.


Of the whole series (4482 patients), 3812 patients had data available on LVI, and these comprised the basis of the current study. Of these patients, 1129 (30%) had definite LVI, whereas 2683 (70%) did not have definite LVI. The median tumor size was 1.7 cm (range, 0.2-5 cm). LNs were negative (pN0) in 2426 patients (64%), from 1 to 3 LNs were positive (pN1) in 1026 patients (27%), from 4 to 9 LNs were positive (pN2) in 249 patients (7%), and ≥10 LNs were positive (pN3) in 66 patients (2%). The mean and median numbers of LNs removed were as follow: 4.6 and 4 removed LNs in patients with pN0 tumors, 6 and 5 removed LNs in patients with pN1 tumors, 9 and 8 removed LNs in patients with pN2 tumors, and 21 and 19 LNs removed in patients with pN3 tumors. For the purpose of the current study, the pN2 and pN3 categories were combined together (315 patients) as pN2-pN3.

Table 1 summarizes the associations between LVI and other clinicopathologic variables. There was a positive association between definite LVI and symptomatic presentation, young premenopausal patients, larger primary tumor size, higher histologic grade, higher LN status (and absolute number of positive LNs; Z = 19; P < .0001), high Ki-67 index, negative ER status, and positive HER2 status.

Table 1. Association Between Lymphovascular Invasion and Other Clinicopathologic Variables
 No. of Patients (%)  
VariablesLVI NegativeLVI PositiveChi-Square StatisticP
  • Abbreviations: ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; LVI, lymphovascular invasion.

  • a

    Statistically significant.

  • b

    Pathologic lymph node status (pN) according to TNM classification was as follows: pN0, negative; pN1, 1 to 3 positive lymph nodes; pN2-pN3, ≥4 positive lymph nodes (pN2 and pN3 were combined).

  • c

    HER2 status was assessed using immunochemistry and chromogenic in-situ hybridization in borderline cases. Positivity for Ki-67 was defined as ≥10%. Recurrences developed in 979 patients (27%), including 276 (7%) who had local recurrences, 224 (6%) who had regional recurrences, and 736 (20%) who had distant metastasis. Thirty-two percent of patients who had definite LVI developed distant metastasis compared with 14% of patients who did not have definite LVI in their primary tumor. Twenty-seven percent of patients who had definite LVI died of breast cancer compared with 12% of patients who did not have definite LVI. Using the Nottingham histologic grading system, 19% of tumors were grade 1, 37% were grade 2, and 44% were grade 3. Estrogen receptor status was available in 3013 patients, and, of those, 2238 (74%) were positive. Hormone therapy was received by 1407 patients (39%), 678 patients (17%) received chemotherapy, and 168 patients (5%) received both hormone therapy and chemotherapy. Six hundred twenty-three patients (17%) died of breast cancer, 373 (10%) died of other causes, and 2743 (73%) either remained alive at the end of follow-up or were lost to follow-up.

Total2683 (70.4)1129 (29.6)  
Referral method    
 Screening1045 (82)229 (18)127<.001a
 Symptomatic1616 (64)898 (36)  
Patient age, y    
 ≤501007 (64)565 (36)51<.001a
 >501673 (75)563 (25)  
Menopausal status    
 Premenopausal724 (60)481 (40)90<.001a
 Postmenopausal1736 (76)561 (24)  
Operation type    
 Mastectomy1255 (61)797 (39)181<.001a
 Local excision1417 (81)327 (19)  
Primary tumor size, cm    
 ≤1592 (88)84 (12)215<.001a
 >1-21292 (73)475 (27)  
 2-5696 (56)536 (44)  
Histologic grade    
 1550 (88)72 (12)181<.001a
 2884 (73)322 (27)  
 3850 (60)577 (40)  
Pathologic lymph node statusb    
 pN01983 (82)443 (18)501<.001a
 pN1559 (55)467 (45)  
 pN2-pN399 (31)216 (69)  
ER status    
 Negative504 (65)271 (35)8.006a
 Positive1574 (70)664 (30)  
HER2 statusc    
 Negative1148 (70)481 (30)9.004a
 Positive156 (61)98 (39)  
Ki-67 indexb    
 Negative435 (79)117 (21)42<.001a
 Positive617 (63)365 (37)  
 No2448 (71)1005 (29)0.4.258
 Yes191 (69)85 (31)  
 No2503 (71)1002 (29)11.001a
 Yes136 (61)88 (39)  
Distant metastasis    
 No2275 (75)755 (25)151<.001a
 Yes383 (52)353 (48)  

Association With Patient Outcomes

In the whole series, definite LVI was associated with poorer BCSS and DMFS at both 5-year and 10-year follow-up (Table 2, Fig. 1). This association was maintained at 15-year and 20-year follow-up. When subgroup analyses were performed, definite LVI maintained its association with a poor outcome in the LN-negative (pN0) and LN-positive (≥1 positive LN) subgroups and in the different size (pT1 and pT2) subgroups (Table 2). LVI was associated with outcome in tumors of different histologic grade (grades 1, 2, and 3) in the ER-positive and ER-negative subgroups, in the HER2-negative and HER2-positive subgroups, and in the triple-negative (ER-negative, progesterone receptor-negative, and HER2-negative) molecular classes (P < .0001). The association between LVI and outcome was also maintained when patients were stratified according to local and systemic therapy. When the analysis was restricted to LN-negative tumors, LVI maintained its association with outcome in all molecular subgroups except in the HER2-positive subgroup.

Table 2. Association Between Lymphovascular Invasion and Patient Outcomes in the Whole Breast Cancer Series and in the Different Pathologic Subsets
Patient GroupaNo. of Patients at RiskProportion Surviving: LVI-Negative vs LVI-Positive, %HR (95%CI)PProportion Surviving: LVI-Negative vs LVI-Positive, %HR (95%CI)P
  • Abbreviations: BCSS, breast cancer-specific survival; CI, confidence interval; DMFS, distant metastasis-free survival; ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; HR, hazard ratio; LVI, lymphovascular invasion; pN, pathologic lymph node status; pT, pathologic tumor classification.

  • a

    The numbers of patients exposed to risk are shown at the beginning and the proportion of patients surviving are shown at the end of 5-year and 10-year periods in the LVI-positive versus LVI-negative subgroups. P values are <.001 unless indicated otherwise.

Whole series3757 2.5 (2.1-2.9)<.001 2.6 (2.2-2.9)<.001
 5-Y survival375792 vs 81  90 vs 73  
 10-Y survival216185 vs 66  82 vs 60  
pN0: Lymph node negative, n = 2362       
 pT1a,pT1: Tumors ≤1 cm573 5.1 (2.2-12.1)<.001 4.8 (2.4-9.8)<.001
 5-Y survival57399 vs 90  97 vs 85  
 10-Y survival30996 vs 82  94 vs 70  
T1c: Tumors >1 to 2 cm12062.0 (1.4-2.8) <.001 2.0 (1.4-2.7)<.001
 5-Y survival120694 vs 91  93 vs 83  
 10-Y survival76087 vs 75  85 vs 72  
T2: Tumors >2 to 5 cm583 1.6 (1.1-2.4).01 1.8 (1.3-2.6)<.001
 5-Y survival58388 vs 82  86 vs 79  
 10-Y survival35680 vs 71  78 vs 64  
pN1: Lymph node positive, n = 1017       
 pT1: Tumors ≤2 cm567 1.4 (0.9-2.2).04 1.5 (1.1-2.3).01
 5-Y survival56792 vs 88  88 vs 81  
 10-Y survival4581 vs 72  79 vs 65  
 T2: Tumors >2 to 5 cm450 1.3 (0.9-2.0).23 1.4 (0.9-2.1).13
 5-Y survival45086 vs 79  82 vs 70  
 10-Y survival24769 vs 63  63 vs 58  
pN2-pN3: Lymph node positive, n = 289       
 pT1: Tumors ≤2 cm102 1.6 (0.8-3.2).1388 vs 811.4 (0.8-2.6).09
 5-Y survival10271 vs 58     
 10-Y survival4551 vs 34  79 vs 65  
 T2: Tumors >2 to 5 cm187 1.7 (0.9-3.1).08 1.6 (0.9-2.6).51
 5-Y survival18777 vs 62  67 vs 52  
 10-Y survival24763 vs 43  62 vs 39  
ER subgroups, n = 2967       
 ER-positive2156 2.6 (2.1-3.3)<.001 2.7 (2.2-3.3)<.001
 5-Y survival 95 vs 87  93 vs 77  
 10-Y survival 88 vs 70  85 vs 63  
 ER-negative741 2.3 (1.8-3.0)<.001 2.4 (1.9-3.2)<.001
 5-Y survival 83 vs 62  81 vs 58  
 10-Y survival 74 vs 52  74 vs 48  
HER2 status subgroup, n = 1838       
 Positive249 2.3 (1.6-3.4)<.001 2.4 (1.7-3.5)<.001
 5-Y survival 76 vs 57  72 vs 49  
 10-Y survival 68 vs 43  68 vs 35  
 Negative1589 2.2 (1.8-2.7)<.001 2.3 (1.9-2.8)<.001
 5-Y survival 92 vs 80  89 vs 73  
 10-Y survival 83 vs 67  80 vs 59  
Figure 1.

This chart illustrates the association between lymphovascular invasion (LVI) (definite [+] and negative [−]) with (Top) breast cancer-specific survival and (Bottom) distant metastasis-free survival.

After adjusting for other prognostic indicators (including LN status, tumor size, and histologic grade), multivariate analyses were performed to determine whether LVI was an independent prognostic factor for BCSS and DMFS. This analysis indicated that LVI was an independent predictor of survival. The same results were obtained after the addition of patient age, ER status, and adjuvant systemic treatment to the previous model (Table 3). When the multivariate analysis was repeated with addition of the interaction between LVI and histologic grade, LN status, and tumor size (fitted as linear terms), the Pinteraction values were not significant (P = .96, P = .61, and P = .34 for histologic grade, LN status, and tumor size, respectively), indicating that the effects of LVI on patient outcomes were not affected significantly by grade, LN status, or tumor size.

Table 3. Multivariate Cox Regression Analysis of Factors Associated With Breast Cancer-Specific and Distant Metastasis-Free Survival in the Whole Series (N = 3757)
PredictoraHR (95% CI)PHR (95% CI)P
  • Abbreviations: BCSS, breast cancer-specific survival; CI, confidence interval; DMFS, distant metastasis-free survival; HR, hazard ratio.

  • a

    Lymph node status and histologic grade were coded as 1, 2, or 3; size was coded as 1 (≤2 cm) or 2 (>2-5 cm); and age was coded as 1 (≤50 years) and 2 (>50 years). Chemotherapy and hormone therapy were coded as 1 (not given) or 2 (given; any type). Estrogen receptor status was coded as 1 (negative) or 2 (positive; defined as 10% positivity).

  • b

    Statistically significant.

Lymphovascular invasion1.7 (1.4-2.0)<.0001b1.7 (1.4-2.1)<.0001b
Lymph node status1.9 (1.6-2.2)<.0001b2.0 (1.7-2.3)<.0001b
Histologic grade2.0 (1.7-2.4)<.0001b1.8 (1.5-2.1)<.0001b
Tumor size1.3 (1.1-1.6).0231.3 (1.1-1.6).003b
Patient age1.1 (0.8-1.4).4620.9 (0.8-1.2).740
Chemotherapy0.6 (0.5-0.9).003b0.6 (0.4-0.8)<.0001b
Hormone therapy0.8 (0.6-1.0).0580.8 (0.6-0.9).006b
Estrogen receptor status0.7 (0.6-0.9).005b0.8 (0.6-1.0).062

When multivariate analysis was restricted to LN-negative patients, LVI maintained its independent association with outcome. In the LN-negative/ER-negative subgroup (464 informative cases), multivariate analysis indicated that LVI was the only significant predictor of BCSS (P < .0001; hazard ratio [HR], 2.1; 95% CI, 1.5-1.3) and DMFS (P < .0001; HR, 2.4; 95% CI, 1.7-3.2), whereas histologic grade, tumor size, patient age, HER2 status, and systemic therapy were not all significant. Similar results were obtained in the triple-negative tumors (179 informative cases).

Lymphovascular Invasion in Relation to TNM Staging Categories

Patients were stratified according to TNM staging classification. This revealed that LVI was associated with both BCSS and DMFS (P < .0001) in the LN-negative (pN0) group when further stratified by tumor size (Table 2). Multivariate analyses after adjustment for other prognostic indicators and adjuvant systemic treatment were performed in each of these subgroups. Those analyses indicated that LVI and histologic grade were independents prognostic factors for both BCSS and DMFS in the pT1/N0 subgroup. In the pT2/N0 subgroup, LVI was the only significant prognostic variable (P = .001; HR, 1.5; 95% CI, 1.2-1.8).

Although univariate survival analysis revealed an association between LVI and outcome in the pT1/N1 subgroup, multivariate analysis indicated that this association was not independent of histologic grade. No significant associations between LVI and outcome were identified in the pT2/N1 or pT1-pT2/N2-N3 subgroups (P > .05).

Lymphovascular invasion, positive lymph node status, and tumor size

Because LN status is the most recognized prognostic factor for patients with operable breast cancer, we compared the prognostic significance of LVI with that of LN status. The association between combined LN categories and LVI subgroups indicated an almost complete overlap between the pN0 LVI-positive curve and the pN1 LVI-negative curve, indicating that definite LVI in LN-negative tumors provided prognostic value equivalent to that provided by the presence of 1 to 3 positive LNs (pN1) if the tumor was LVI-negative (Fig. 2). Similar associations were identified with DMFS (P < .0001). Multivariate analyses indicated not only that LVI was an independent predictor of both BCSS and DMFS but also that the HRs for definite LVI were higher than the HRs for the presence of 1 or 2 positive LNs compared with LN-negative tumors (Table 4).

Figure 2.

The association between combined lymphovascular invasion (LVI) (definite [+] and negative [−]), pathologic lymph node (pN) status (pN0, no positive lymph nodes; pN1, 1 to 3 positive lymph nodes; pN2-pN3, ≥4 positive lymph nodes), and breast cancer-specific survival (chi-square statistic, 397; 5 degrees of freedom; P < .0001) is illustrated. This graph shows pN0/LV− (1965 patients) at the top and pN1/LVI+ (455 patients), pN2-3/LVI− (94 patients), and pN3/LVI+ (207 patients) at the bottom. In the middle, there is overlap between pN0/LVI+ (441 patients; survival rates: 88% at 5 years, 75% at 10 years, and 61% at 15 years) and pN1/LVI− (551 patients; survival rates: 89% at 5 years, 76% at 10 years, and 67% at 15 years).

Table 4. Association Between Lymphovascular Invasion With Outcome Compared With 1, 2, and 3 Positive Lymph Nodes Versus Negative Lymph Node Status (n = 2248)
 HR [95% CI]
  1. Abbreviations: BCSS, breast cancer-specific survival; CI, confidence interval; DMFS, distant metastasis-free survival; HR, hazard ratio; LN, lymph node; LVI, lymphovascular invasion.

Definite LVI vs negative LVI1.8 [1.5-2.2]1.9 [1.6-2.3]
1 LN-positive case (n = 486) vs LN-negative cases1.4 [1.1-1.8]1.4 [1.2-1.8]
2 LN-positive cases (n = 209) vs LN-negative cases1.6 [1.1-2.2]1.4 [0.9-1.9]
3 LN-positive cases (n = 114) vs LN-negative cases2.6 [1.9-3.7]2.8 [2.0-3.8]

Similar to LN status, the association between combined tumor size (pT1 and pT2) and LVI subgroups in the LN-negative cohort indicated an almost complete overlap between the pT1 LVI-positive curve and the pT2 LVI-negative curve (Fig. 3), indicating that definite LVI in pT1 tumors had prognostic value equivalent to that of 1 change in tumor size (pT2) provided that the tumors was LVI-negative (Table 5).

Figure 3.

The association between combined lymphovascular invasion (LVI) (definite [+] and negative [−]), tumor size (pathologic tumor classification 1 [pT1] and pT2), and breast cancer-specific survival (chi-square statistic; 69; 3 degrees of freedom; P < .0001) in lymph node-negative patients is illustrated. Note the overlap between pT1 LVI+ (281 patients; 5-year survival rate, 91%; 10-year survival rate, 76%) and pT2 LVI− (425 patients; 5-year survival rate, 88%; 10-year survival rate, 80%). A similar association was obtained with distant metastasis-free survival (chi-square statistic, 75; 3 degrees of freedom; P < .0001).

Table 5. Association Between Lymphovascular Invasion and Outcome Compared With Change in Size Category (Pathologic T2 Tumors vs Pathologic T1 Tumors) in Lymph Node-Negative Patients
 HR [95% CI]a
  • Abbreviations: BCSS, breast cancer-specific survival; CI, confidence interval; DMFS, distant metastasis-free survival; HR, hazard ratio; LVI, lymphovascular invasion.

  • a

    HRs were derived from multivariate analyses.

Definite LVI vs negative LVI2.1 [1.6-2.6]2.2 [1.7-2.8]
pT2 (n = 791) vs pT1 (n = 1864) in the LN-negative cases1.8 [1.4-2.3]1.6 [1.3-2.1]

Lymphovascular invasion and locoregional recurrences

In this series, 276 patients (7.2%) developed local recurrences, and 224 patients (5.9%) developed regional recurrences during follow-up. Definite LVI was associated with the development of regional recurrence in the whole series and in the LN-negative subgroup. Multivariate analysis indicated that LVI was an independent predictor of regional recurrence in the LN-negative group (HR, 1.8; 95% CI, 1.3-2.7; P = .002); however, in the whole series, LVI was not independent of histologic grade or LN status. No association was observed between LVI and local recurrence in the whole series (P = .226) or when patients were stratified according to surgery type (mastectomy [P = .624] vs local excision [P = .064]), histologic grade, or systemic treatment subgroups (P > .05).

Lymphovascular invasion and immunohistochemistry

There was an association between LVI detected by IHC and LVI detected by routine H&E-stained sections (Table 6). A survival analysis of the LN-negative patients revealed an association between definite, immunohistochemically detected LVI and poorer outcome (Fig. 4). Multivariate analysis indicated that IHC-detected LVI and routine-detected LVI were associated with both BCSS and DMFS.

Table 6. Association Between Routine Lymphovascular Invasion (LVI) Detected in Hematoxylin and Eosin-Stained Sections and LVI Detected by Immunostaining for Endothelial Marker (D2-40) in a Subset of 976 Lymph Node-Negative Patients
  Routine H&E-Detected LVI: No. of Patients (%)  
IHC-Detected LVINo. of PatientsLVI NegativePossible LVILVI PositiveChi-Square StatisticP
  • Abbreviations: H&E, hematoxylin and eosin; IHC, immunohistochemistry.

  • a

    In this study, immunostaining was performed in 1 tumor section only without prior knowledge of the routine LVI status and without choice of the representative tumor section to correlate with routine LVI status. Routine LVI was assessed based on examination of H&E-stained tumor sections, including at least 4 sections, except for very small tumors, in which 1 to 3 sections may have been examined.

  • b

    Statistically significant.

Negative792591 (89)91 (80)110 (56)a108<.0001b
Positive18475 (11)22 (20)87 (44)  
Total 666 (82)113 (12)197 (20)  
Figure 4.

These charts illustrate the association between lymphovascular invasion (LVI) detected with immunohistochemistry (IHC) in the lymph node-negative, routine LVI-negative group (784 patients), including (Top) breast cancer-specific survival (hazard ratio, 2.3; 95% confidence interval, 1.5-3.5; chi-square statistic, 16; P < .0001) and (Bottom) distant metastasis-free survival (hazard ratio, 2.2; 95% CI, 1.5-3.4; chi-square statistic, 16; P < .0001).

In the LN-negative group, when patients who had definite, IHC-detected LVI were added to the patients who had definite, routine, H&E-detected LVI, an improved association with outcome was observed (HR, 2.2; 95% CI, 1.7-2.9; P < .0001) compared with patients who had routine, H&E-detected LVI alone (HR, 1.8; 95% CI, 1.4-2.6; P < .0001).


In the current study, we assessed the prognostic significance of LVI in a large series of patients who had operable breast cancer, who were treated uniformly in a single institution, and who had long-term follow-up. The proportion of tumors with definite LVI in this study was consistent with most previous studies of breast cancer (range, 25%-35% for the whole series4, 13, 14, 16; 18%-22% for the LN-negative subgroup2, 7, 9, 16 and for those with pT1 tumors39; and 45%-60% for those with LN-positive tumors13, 18, 40). Our results demonstrate that LVI is associated with other well established prognostic variables and with patient outcomes. Definite LVI is associated with a poorer outcome in terms of shorter BCSS and shorter DMFS. This prognostic value is independent of other prognostic variables, such as patient age, LN status, histologic grade, tumor size, ER status, and systemic therapy. LVI provides independent prognostic information not only in the whole series but also in the clinically relevant subgroups, in which decisions about systemic therapy need to be determined (pT1N0M0 and pT2N0M0). It is noteworthy that, in patients with LN-negative tumors, LVI is an independent, high-risk criterion that, by itself, can be considered sufficient to move patients into a high-risk group, because LVI indicates a risk equivalent to that indicated by the presence of 1 or 2 positive LNs (ie, prognostic value equivalent to a change from pN0 to pN1) and to that indicated by a change in 1 size category (from pT1 to pT2). Although, in the current study, patients typically underwent a 4-LN sampling procedure, this usually includes the sentinel LN; therefore, we expect that our findings may be extrapolated to patients staged by the currently commonplace sentinel LN biopsy.38

Because this study provides evidence that the presence of LVI has prognostic implications for LN-negative tumors (which makes them comparable to LN-positive tumors within pT classifications) and for the LN-negative group (which makes pT1 tumors comparable to pT2 tumors), we recommend revising the AJCC staging system to incorporate LVI into the staging of LN-negative breast cancer. This can be comparable to the changes introduced to the AJCC staging of melanoma when it was established that ulceration provided independent prognostic information.

LVI also provides independent prognostic information for patients with ER-negative tumors and triple-negative tumors. This is an important finding, because other well established clinicopathologic and molecular markers, including the recent multigene assays, have a limited prognostic value in these molecular classes. Most of these tumors are of high grade (grade 3)41, 42 and typically exhibit poor-prognosis gene signatures.43 Molecular classifiers, such as Oncotype DX, Genomic Grade Index,41 and Mammaprint, which are recommended as suitable for all breast cancer patients, have negligible discriminatory power in patients with ER-negative and triple-negative disease.41, 42

It has been reported that LVI should elevate the risk category only if it is extensive16, 44; however, our results demonstrate the prognostic value of LVI regardless of its extent, because our results did not distinguish between extensive LVI and nonextensive LVI. In a previous study, we demonstrated that the extent of LVI assessed using IHC markers had no impact on the association of LVI with outcome.28 In addition to the low frequency of cases defined as extensive LVI (6.5%16), the definition of extensive LVI may suffer from a considerable degree of subjectivity. However, Ejlertsen et al,15 in analyzing a series of patients with breast cancer from a population-based study, reported that LVI was not an independent prognostic marker in the low-risk group. However, in their study, the frequency of LVI, which was assessed locally in a large number of laboratories (n = 18), in the low-risk group was very low (1.6%; 54 of 3217 patients), and there appeared to be an underestimation of definite LVI in that group, which brings into question the validity of their results.14 The current study also provides evidence that, in a pragmatic approach, the application of immunostaining using a single endothelial marker on a single section of tumor with routinely diagnosed (based on H&E-stained sections) negative LVI or possible LVI can improve the sensitivity of LVI detection with a subsequent increase of its prognostic significance.

Although several studies have reported an association between LVI and the development of local recurrence,3, 5, 9, 10, 45, 46 those studies included old series before the use of adjuvant therapy and demonstrated high local recurrence rates (from 21%46 to 25%3, 39). The recent application of systemic therapies and the improvements in local management of breast cancer have significantly reduced the frequency of local recurrence47 (local recurrence rates in the current study were 6% after 5 years and 10% after 10 years). This may explain the lack of an association between LVI and local recurrence in the current study for both the mastectomy subgroup and the wide local excision subgroup.

The accuracy detecting LVI depends on high-quality tissue preservation. We have consistently applied and endorsed rigorous, optimized, and standardized methods for tissue handling, fixation, and preparation. The optimal method for assessing LVI in routine practice is the use of good-quality, promptly fixed and processed H&E-stained sections. Good fixation is vital to preserve the morphologic detail not only for reporting LVI but also for reporting other important prognostic variables, including histologic grade, hormone receptor status, and HER2 status.19, 20 From our experience and the results identified in the current study, we believe that LVI, when assessed in such a rigorous fashion, is an undervalued but powerful and valuable prognostic factor in breast cancer that can be assessed reproducibly and cost efficiently in routine clinical practice to yield consistent results complimentary to the assessment of other well established prognostic variables, such as LN status and tumor size.

In conclusion, our current results demonstrate that LVI has independent prognostic significance particularly in the low-risk pT1-pT2/N0 subgroup, in which LVI can be used as a high-risk criterion sufficient to move patients into a high-risk group with a risk equivalent to that indicated by the presence of 1 to 3 positive LNs (pN1) and to that indicated by a change in 1 tumor size category (from pT1 to pT2). LVI is an independent predictor of poor outcome in the LN-negative, ER-negative, and triple-negative molecular classes. LVI has both biologic and time-dependent staging characteristics. We recommend routine assessment of LVI using published criteria3, 19, 20, 22 linked with good practice in tumor tissue handling and preparation. IHC for endothelial-specific markers can improve the prognostic significance of LVI when used in LVI negative/possible cases. We advocate the incorporation of LVI status into breast cancer staging systems.


No specific funding was disclosed.


The authors made no disclosures.