Predicting the status of axillary sentinel lymph nodes in 4351 patients with invasive breast carcinoma treated in a single institution

Abstract

BACKGROUND

Reliable predictors of metastatic involvement would enable a better selection of candidate patients for sentinel lymph node biopsy (SLNB) and possibly allow identification of patients with such a low risk of axillary sentinel lymph node (SLN) involvement to be even spared SLNB.

METHODS

The authors evaluated 4351 consecutive patients surgically treated for breast carcinoma who also underwent SLNB. Clinicopathologic features significantly associated with SLN metastases by univariate analysis were included in a multivariate model.

RESULTS

By multivariate analysis, the prevalence of SLN metastases was associated directly with tumor size, multifocality, and with the occurrence of peritumoral vascular invasion (PVI; all P < 0.0001), and was associated inversely with a favorable histotype (P = 0.0007) and lack of progesterone receptors (P = 0.004). A predictive model based on the features more closely associated with SLN status documented that the patients with a favorable tumor type ≤ 1 cm in size and without PVI (n = 178 [4% of the population]) had the lowest risk of SLN metastases (9.5%) whereas patients with tumors > 2 cm and with PVI (n = 250 [5.7% of the population]) had the highest risk (77.2%) of SLN involvement.

CONCLUSIONS

Tumor size and PVI emerged as the most powerful independent predictors of SLN metastases. Although no combination of features identified patients with a < 9.5% risk of SLN metastases, the current data may be used to tailor the management of patients with breast carcinoma with the aim of minimizing as much as possible the diagnostic and therapeutic procedures, thus improving the quality of life of the patients without any adverse effect on their survival rates. Cancer 2005 © 2004 American Cancer Society

Axillary lymph node status has long been considered the most powerful predictor of survival for patients with breast carcinoma,1–3 and axillary lymph node dissection (ALND) has been the standard of care to estimate prognosis and to guide the choice of adjuvant treatment.

More recently, however, systemic adjuvant therapy has been offered to patients according to selected clinicopathologic features of the primary tumor and independently of the axillary lymph node status with an equivalent survival benefit.4, 5 Accordingly, ALND has become less relevant for treatment selection and the need for completion of ALND for all patients has been questioned.6 This is particularly true when considering the increasing prevalence of early, screen-detected breast carcinomas with an expected low risk of axillary lymph node involvement7 and the relevant morbidity associated with ALND.

Sentinel lymph node biopsy (SLNB) has emerged as a most powerful tool to probe axillary lymph node status and to spare completion of ALND in patients with uninvolved sentinel lymph nodes (SLN).8 Although SLNB is a minimally invasive and well tolerated surgical procedure, it yields metastasis-free SLN in 65–70% of patients, at the expense of more complex patient management, with the involvement of nuclear medicine specialists should radioactive tracing of the SLN be necessary, a prolonged surgical time, and extensive histopathologic examination of the lymph nodes.

The identification of reliable clinicopathologic predictors of the risk for SLN metastases would be effective in selecting patients for SLNB and sparing axillary lymph node surgery to those at the lowest risk of axillary lymph node involvement. Possible clinicopathologic predictors of axillary lymph node involvement in patients with breast carcinoma were studied quite extensively in the pre-SLNB era, with the identification of primary tumor palpability,9–11 size,6, 11–21 type10, 13 and grade,11–13 lymphovascular invasion,6, 10–12, 14, 16, 20, 21 receptor status,13, 22 young age,13, 14 and multifocality23 as features independently correlated with axillary lymph node metastases. None of these features, however—alone or combined with others in predictive models—has attained enough strength to be adopted for the selection of patients for whom completion of ALND may be avoided.

The population of candidate patients for SLNB is unique because, currently, they are being selected according to the relatively smallness (most often ≤ 3 cm in diameter) and unicentricity of the primary tumor, and to a clinically negative axilla. Furthermore, the extensive histopathologic examination of the SLN, with serial sectioning and the use of ancillary immunohistochemical techniques, increases the detection rate of lymph node involvement, with special reference to the identification of isolated tumor cells and micrometastases.

We determine whether the clinicopathologic parameters predictive of axillary lymph node involvement retain the same value in predicting the risk of SLN metastases in the cohort of patients undergoing SLNB, and whether they may be used to select patients at such a low risk of SLN involvement to be even spared SLNB.

We report the correlation between prospectively collected clinicopathologic features and SLN status in a large series of 4351 consecutive patients treated with conservative or ablative surgery for primary invasive breast carcinoma and with SLNB in a single institution during the period 1996–2003.

MATERIALS AND METHODS

Patients

From April 1996 to November 2003, 6371 patients with clinically lymph node-negative breast carcinoma underwent SLNB at the European Institute of Oncology (Milan, Italy). Of these patients, 4351 received surgery (either conservative or radical mastectomy) for primary invasive breast carcinoma at the same institution and represent the study population. The cohort comprised 4344 females and 7 males with a median age of 54 years (range, 24–85 years).

The first 371 patients were enrolled in a pilot study designed to evaluate the predictive power of SLNB and all were treated with SLNB and simultaneous complete ALND, as reported.24 Subsequently, a clinical trial was performed, whereby 516 patients were randomized to receive either SLNB and simultaneous ALND or SLNB followed by ALND only when the SLNB results were positive.25

Based on the results of these two studies, the remaining 3464 patients underwent SLNB as the standard treatment, and completion of ALND was performed selectively only in patients with metastatic SLN. SLNB as a standard treatment was offered to patients with cytologically or histologically verified unifocal or multifocal breast tumors ≤ 3 cm (measured clinically and/or by mammography, ecography or both, as appropriate) and with clinically uninvolved axillary lymph nodes. Patients did not receive SLNB if they had multicentric disease, inflammatory carcinoma, if they had received previous neoadjuvant treatments, and if they were pregnant or lactating.

The study protocols were approved by the ethics committee of the European Institute of Oncology, and informed consent was obtained from all patients.

Axillary Sentinel Lymph Node Localization and Examination

The SLNs were identified by injection of a radioactive (⁹⁹Tc-labelled) tracer the day before surgery, and were removed surgically with the aid of a hand-held gamma-ray detecting probe (Ethicon Endo-Surgery, Cincinnati, OH), as previously reported.24 The majority of patients received SLNB during conservative or ablative breast surgery. Only 480 patients underwent SLNB in an outpatient setting, a few days before breast surgery.

The SLN sections removed during breast surgery were frozen immediately and examined intraoperatively according to the previously described procedure.26 Briefly, the SLN sections were bisected along the major axis, and both moieties were embedded in OCT compound (Cellpath, Hemel Hempstead, England) cut surfaces up, before being snap-frozen in isopentane cooled with liquid nitrogen. Lymph nodes < 5 mm were embedded and frozen uncut.

Thirty pairs of adjacent frozen sections, 4 μm thick, were cut at 50-μm intervals. If residual tissue sections remainedt, additional pairs of sections were cut at 100-μm intervals until the lymph node was entirely sectioned. One section of each pair was stained with hematoxylin and eosin (H&E). Whenever needed, to ascertain the nature of atypical cells observed on the H&E-stained sections, the mirror sections were immunostained for cytokeratins using a rapid staining method (EPOS anticytokeratins/horseradish peroxidase; Dako A/S, Glostrup, Denmark) with the MNF116 monoclonal antibody. This procedure was indeed required in a limited subset of patients, accounting for 6% of those showing micrometastatic deposits in the SLN.

For patients subjected to SLNB before breast surgery, the lymph nodes were bisected fresh if they were > 5 μm and fixed in 10% neutral-buffered formalin for 6–8 hours, before being embedded in paraffin. Pairs of paraffin-embedded sections were cut at 50 and 100-μm intervals and stained as above.

Metastatic deposits were measured according to the three axis, and the largest size was recorded.27 The size of the metastatic nests in the plane of the tissue sections was measured histologically with an ocular micrometer, and the thickness was calculated according to the number of involved contiguous sections and to the sectioning interval between them. To avoid underestimation of the thickness of the metastases, the cutting intervals immediately preceding the first and following the last involved sections also were included. The recorded largest size corresponded to the maximum diameter in the plane of the section or to the thickness of the metastatic foci, whichever was larger. If multiple but distinct (i.e., separated by uninvolved tissue sections) metastases were identified in the same SLN, the size of the largest was recorded.

Examination of Primary Tumors and Nonsentinel Axillary Lymph Nodes

The breast surgical specimens were examined fresh, and the maximum tumor diameter was recorded. Histologically, the tumor specimens were classified according to the World Health Organization histologic classification of breast tumors,28 as modified by Rosen and Oberman.29 Tumor grade was evaluated using the recommendations of Elston and Ellis.30 The occurrence of peritumoral vascular invasion (PVI) was assessed according to the criteria of Rosen and Oberman.31

The expression of estrogen (ER) and progesterone receptors (PgR) and the tumor proliferative fraction were evaluated immunohistochemically as previously reported.32 Results were recorded as the percentage of cells showing definite nuclear immunoreactivity over ≥ 2000 neoplastic cells in 10 randomly selected high power fields (i.e., × 400) from sections retrieved from the periphery of the tumor. In the last 1982 patients, HER-2/neu overexpression also was investigated immunohistochemically, using a specific polyclonal antiserum (Dako A/S, working dilution 0.05 mol/L) and the Envision detection reagent (Dako A/S), according to the manufacturer's instructions. The staining results were recorded in a 4-tier scale, from 0 to ≥ 3, according to the percentage of immunoreactive cells, and to the intensity and completeness of membrane staining as recommendend by the Food and Drug Administration.33

Nonsentinel axillary lymph nodes were tagged by Berg level. They were bisected if > 5 mm, and processed routinely. Three to six H&E-stained sections per lymph node cut at 100–500-μm intervals were examined.

Statistical Analyses

Logistic regression was used to assess the association between the various clinicopathologic characteristics and the occurrence of SLN metastasis. The variables that were associated significantly with SLN involvement in univariate analysis were included in a multivariate model. The final logistic model was used to produce a predictive plot of SLN metastasis as a function of tumor size and other tumor characteristics. All analyses were performed using SAS software (SAS Institute, Cary, NC), and all statistical tests were two sided.

RESULTS

SLN metastases were detected in 1446 (33.2%) of the 4351 patients. In 564 patients, the SLN harbored only micrometastases (≤ 2 mm), whereas larger metastases were identified in the remaining 882 patients. The SLN was the only involved axillary lymph node(s) in 766 (57.3%) patients. In univariate analysis, the prevalence of SLN metastases was correlated significantly with several clinicopathologic parameters, including patient age, tumor size, type, grade and multifocality, occurrence of PVI, higher proliferative fraction, and lack of PgR (Tables 1 and 2).

Table 1. Prevalence of Sentinel Lymph Node Metastases According to Clinicopathologic Characteristics

Variable	Patients (n)	SLN positive (%)
SLN: sentinel lymph node; PVI: peritumoral vascular invasion; ER: estrogen receptor; PgR: progesterone receptor.
Gender
Females	4344	33.2
Males	7	28.6
Age (yrs)
0–35	150	44.7
36–45	768	39.2
46–55	1398	35.2
56–65	1266	29.5
≥65	769	27.7
Histology
Ductal	3509	35.9
Lobular	416	27.2
Medullary	9	0
Cribriform	207	17.9
Papillary	24	33.3
Mucinous	86	11.6
Apocrine	31	32.3
Metaplastic	4	25.0
Tubular	57	12.3
Other	8	0
pT
T1mic	42	9.5
T1a	186	8.1
T1b	971	19.6
T1c	2346	35.4
T2	772	49.9
T3–T4	34	61.8
Grade
G1	1142	22.7
G2	2087	37.2
G3	1068	38.0
PVI
Absent	3448	24.1
Present	903	68.1
Focality
Unifocal	3940	31.8
Multifocal	411	47.0
ER status
ER negative	519	30.8
ER positive	3814	33.6
PgR status
PgR negative	1090	29.0
PgR positive	3240	34.7
Ki67 expression
< 16%	2139	29.2
≥ 16%	2190	37.2
HER-2/Neu expression
0	1526	32.8
1	518	33.8
2	224	33.0
3	329	38.3

Table 2. Association between Clinicopathologic Factors and Occurrence of Sentinel Lymph Node Metastases in 4351 Patients with Breast Cancer

Variable	Univariate OR (95% CI)	P value	Multivariate OR (95% CI)a	P value
OR: odds ratio; CI: confidence interval; PVI: peritumoral vascular invasion; ER: estrogen receptor; PgR: progesterone receptor. a OR and 95% CI values were obtained from multivariate analysis with all variables significant in univariate analysis fitted simultaneously. b Favorable histology: medullary, cribriform, tubular, mucinous
Gender
Females	Reference
Males	0.80 (0.16–4.15)	0.794
Age (yrs)
0–35	Reference		Reference
36–45	0.80 (0.56–1.14)	0.211	0.95 (0.64–1.42)	0.816
46–55	0.67 (0.48–0.95)	0.022	1.02 (0.69–1.51)	0.907
56–65	0.52 (0.37–0.73)	0.0002	0.84 (0.57–1.24)	0.376
≥65	0.48 (0.33–0.68)	< 0.0001	0.74 (0.49–1.11)	0.146
Histology
Ductal	Reference		Reference
Lobular	0.67 (0.53–0.84)	0.0004	0.84 (0.65–1.07)	0.162
Favorable histologyb	0.32 (0.24–0.43)	< 0.0001	0.55 (0.39–0.78)	0.0007
Other	0.72 (0.42–1.24)	0.234	0.76 (0.42–1.40)	0.381
Tumor size (cm)
≤ 0.5	Reference		Reference
> 0.5–1	2.68 (1.63–4.39)	< 0.0001	2.34 (1.40–3.90)	0.0011
> 1–2	6.04 (3.75–9.73)	< 0.0001	4.03 (2.46–6.61)	< .0001
> 2	11.3 (6.90–18.4)	< 0.0001	7.62 (4.57–12.7)	< .0001
Grade
G1	Reference		Reference
G2	2.02 (1.71–2.38)	< 0.0001	1.27 (1.04–1.56)	0.021
G3	2.09 (1.74–2.52)	< 0.0001	1.06 (0.80–1.40)	0.689
PVI
Absent	Reference		Reference
Present	6.73 (5.73–7.89)	< 0.0001	5.26 (4.44–6.23)	< .0001
Focality
Unifocal	Reference		Reference
Multifocal	1.90 (1.55–2.33)	< 0.0001	1.78 (1.41–2.24)	< .0001
ER status
ER+	Reference		Reference
ER−	0.88 (0.72–1.08)	0.215	0.94 (0.70–1.26)	0.665
PgR status
PgR+	Reference		Reference
PgR−	0.77 (0.66–0.89)	0.0006	0.73 (0.59–0.90)	0.004
Ki-67 expression
< 16%	Reference		Reference
≥ 16%	1.48 (1.30–1.69)	< 0.0001	0.98 (0.82–1.17)	0.823
HER-2/Neu
0	Reference
1	1.04 (0.85–1.29)	0.690	Reference
2	1.01 (0.75–1.36)	0.951
3	1.27 (0.99–1.63)	0.058	1.01 (0.76–1.33)	0.971

According to patient age, the risk for SLN metastases steadily decreased from youngest to oldest, with patients ≤ 35 years having the highest risk of metastases (44.7%) and patients > 65 years having the lowest risk of metastases (27.7%). Instead, the rate of SLN metastases was correlated directly with the size of the primary invasive tumor: 8.3% of tumor specimens ≤ 5 mm in diameter had SLN metastases, whereas metastases were documented in 50.6% of invasive tumor specimens > 2 cm (P < 0.0001). The same strong positive correlation (P < 0.001) was obtained when the primary tumor specimens were classified according to the pT categories, i.e., the chance of SLN metastases steadily increased from pT1mic (9.5%) to pT2 and pT3 tumor specimens (49.9% and 65.6%, respectively). Histologic tumor type also correlated with the likelihood of SLN metastases, with invasive ductal carcinomas not otherwise specified (NOS) carrying a significantly higher risk of SLN metastasis than other histotypes, including lobular carcinomas (P = 0.0004) and the special types of ductal carcinoma associated with a better prognosis (cribriform, medullary, mucinous, and tubular), either considered as individual tumor types or as a whole group of favorable prognosis neoplasms (P < 0.0001).

Similarly, higher tumor grade (Grade 2 and Grade 3), proliferative fraction > 16% (16% representing the median value of Ki-67 immunostaining in the current series), multifocality, and histologic evidence of PVI all were associated significantly (P < 0.0001) with a higher prevalence of SLN metastases. Patients with tumors devoid of PgR had a lower risk (P = 0.0006) of SLN metastases, whereas HER-2/neu overexpression (≥ 3) was correlated only marginally with a higher likelihood of SLN involvement (P = 0.058).

The multivariate analysis with all variables significant in univariate analysis fitted simultaneously documented an independent direct association of SLN metastases with primary tumor > 1 cm (P < 0.0001), multifocality (P < 0.0001), and occurrence of PVI (P < 0.0001), whereas an inverse association was assessed with favorable histotypes (P = 0.0007) and lack of PgR immunoreactivity (P = 0.004).

PVI resulted in the single most powerful predictor of SLN metastases in the stepwise logistic regression analysis, with an odds ratio (OR) of 5.26 (95% confidence interval [CI], 4.44–6.23%).

To assess whether any of the variables was also predictive of the size of SLN metastasis, we evaluated them separately in the subsets of patients with micrometastasis in the SLN (n = 564), and in those with larger metastases (n = 882). The multivariate analysis failed to identify any parameter associated selectively with either micrometastatic or macrometastastic disease.

When patients were stratified according to a predictive model based on the features more closely correlated with SLN status (namely, tumor size and type, focality, and PVI), distinct risk categories were identifiable, with the lowest risk (9.5%) of SLN metastases for patients with favorable histology tumors ≤ 1 cm without PVI (n = 178 [4% of the patient population]) and the highest risk (77.2%) for patients with tumors > 2 cm and exhibiting PVI (n = 250 [5.7% of the patients]; Table 3). All the other combinations predicted an intermediate risk of SLN metastases between these two extremes. When SLN metastases are subclassified as macrometastases or as having minimal lymph node involvement (micrometastases and isolated tumor cells (ITC) only, according to the TNM classification of malignant tumors),34 the same model predicts a risk of SLN macrometastases < 10% for unifocal and multifocal tumors of ≤ 1 cm, without lymphovascular invasion (estimated risk ranging from 5.6% to 8.3%), and for tumors of favorable histology without lymphovascular invasion, irrespective of the size (estimated risk ranging from 5.6% to 7.5%; Table 4).

Table 3. Association between Selected Clinicopathological Factors and Micro/Macro-Metastases of the Sentinel Lymph Node(s)

Variable	Micrometastases (n = 564) OR (95% CI)	P value	Macrometastases (n = 882) OR (95% CI)a	P value
OR: odds ratio; 95% CI: 95% confidence interval. a OR and 95% CI were obtained from multivariate analysis also adjusted for age, tumour grade, ER and PgR status, Ki-67 and HER-2/neu expression. b Favorable histotype: medullary, cribriform, tubular, mucinous.
Histology
Ductal	Reference		Reference
Lobular	0.64 (0.44–0.94)	0.021	1.02 (0.76–1.37)	0.906
Favorable histotypeb	0.51 (0.32–0.83)	0.006	0.58 (0.37–0.90)	0.016
Other	0.98 (0.45–2.12)	0.953	0.62 (0.29–1.33)	0.220
Tumour size (cm)
≤ 0.5	Reference		Reference
> 0.5–1	2.30 (1.16–4.56)	0.017	2.29 (1.12–4.67)	0.023
> 1–2	3.35 (1.72–6.52)	0.0004	4.57 (2.29–9.11)	< .0001
> 2	6.51 (3.27–12.9)	< .0001	8.35 (4.12–16.9)	< .0001
PVI
Absent	Reference		Reference
Present	3.76 (3.00–4.71)	< .0001	6.39 (5.27–7.74)	< .0001
Focality
Unifocal	Reference		Reference
Multifocal	1.53 (1.11–2.10)	0.009	1.98 (1.52–2.59)	< .0001

Table 4. Prevalence of Sentinel Lymph Node Metastases According to Tumor Size and Other Pathological Characteristics of the Tumors

	All tumors			Tumors 1cm			Tumors 1–2cm			Tumors 2cm
	No. of patients	SLN positive (%)	Macro/micro	No. of patients	SLN positive (%)	Macro/micro	No. of patients	SLN positive (%)	Macro/micro	No. of patients	SLN positive (%)	Macro/micro
PVI: peritumoral vascular invasion; macro: macrometastases; micro: micrometastases; SLN: sentinel lymph node.
PVI positive	903	68.1	46.5/21.6	90	56.7	28.9/27.8	563	65.9	44.0/28.9	250	77.2	58.4/18.8
PVI negative
Ductal/lobular multifocal	271	37.6	24.7/12.9	72	18.1	8.3/9.7	146	38.4	25.3/13.0	53	62.3	45.3/17.0
PVI negative
Ductal/lobular unifocal	2790	24.3	13.3/11.0	862	14.7	7.2/7.5	1478	25.8	14.6/11.2	450	37.6	20.4/17.1
PVI negative Favorable histology	336	12.5	6.3/6.3	158	9.5	5.1/4.4	145	15.9	8.3/7.6	33	12.1	3.0/9.1

DISCUSSION

This is the largest single institutional study aimed at assessing the value of several clinicopathologic features of primary breast carcinomas in predicting the status of axillary SLN. From the evaluation of 4351 patients with invasive breast carcinoma treated in an 8-year period, 5 parameters emerged as independent predictors of the SLN status in multivariate analysis, namely, primary tumor size and type, multifocality, PVI, and PgR status. The age of the patient, tumor grade, and proliferative fraction (as assessed by Ki-67 immunostaining) did not retain in multivariate analysis the association with SLN status ascertained in univariate analysis.

Although several previous investigations already have documented the value of some of the above factors in predicting axillary lymph node metastases after completion of ALND,9–23 the current study population includes only patients with relatively small unicentric primary tumors (81.5% of the tumors were ≤ 2 cm in diameter) and with clinically negative axilla.

Tumor size emerged as the most powerful independent predictor (together with PVI) of SLN metastases. Compared with the smallest tumors (≤ 0.5 cm), the risk for SLN metastases was 2.3-fold greater for 0.5–1-cm tumors, and 4 or 7.6-fold greater for 1–2-cm tumors or for larger tumors, respectively. These figures are perfectly in line with previous findings on the likelihood of axillary lymph node involvement at completion of ALND according to tumor size,10–21 although the rate of detection of metastatic disease after ALND in our series was slightly higher than that reported earlier. This is conceivable when taking into account the more thorough histologic examination of the SLN resulting in a higher number of metastases detected.

PVI was associated with a 5.3-fold greater risk for SLN involvement, thus reinforcing that it is the most powerful predictor of axillary lymph node metastases.

Previous reports23 also have documented the predictive value of multifocality, although patients with multifocal and multicentric primary tumors were combined in the same risk category. Multifocality of the primary tumor per se was a significant independent predictor (OR, 1.78; 95% CI, 1.41–2.24%) of SLN metastases in the current series, which excluded patients with multicentric tumors.

The predictive role of sex hormone receptor status in previous investigations is controversial, with some studies reporting no value for both ER and PgR status,6, 14 and others pointing to lower risk of axillary lymph node metastases for tumors negative for either receptor13 or for PgR only.11, 22 In our series, lack of PgR was associated with a 27% reduction (95% CI, 10–41%) in the risk of SLN metastases. This finding is counterintuitive, but in line with the findings of Ravdin et al.,22 who reported that PgR concentrations were associated independently with an increased risk of axillary lymph node metastases.

Favorable histology of the primary tumor resulted in a 45% reduction (95% CI, 22–61%) of the likelihood of SLN metastases, confirming the data obtained in previous studies.6, 13 Contrary to previous findings,10, 12 patient age and tumor grade did not retain any significant association with SLN status in multivariate analysis.

Based on the parameters most closely correlated with SLN status (e.g., tumor size and type, multifocality, and PVI), we devised a predictive model, whereby patients were stratified in categories at a different risk of SLN metastases. Favorable histology tumors ≤ 1 cm without vascular invasion have the lowest (9.5%) risk, whereas tumors of any type, tumors > 2 cm, and tumors with vascular invasion have a > 77% risk of SLN involvement. None of the possible combination of factors is associated with a < 9.5% likelihood of metastases, contrary to the findings of Silverstein et al.,11 who reported a ≤ 6% risk of axillary lymph node metastases for tumors < 15 mm, tumors with low nuclear grade, and tumors without lymphovascular invasion, and of Chua et al.,6 who did not find any axillary lymph node metastasis in patients without palpable lymph nodes if the tumor was ≤ 5 mm, unifocal, and without lymphovascular invasion. The failure to identify a cohort of patients with a very low risk of SLN metastases in our study may be related to the more extensive histopathologic examination of the SLN on one side, and to the larger series of patients investigated.

The large series reported in the current study raises many questions related to the role of axillary lymph nodes in the overall strategy of breast carcinoma management. Axillary lymph node status, with or without metastases, has represented for a full century the single most important factor for evaluating patient prognosis and for selecting the most appropriate adjuvant treatments. The current investigation emphasizes the most powerful predictive features of the primary tumor related to the metastatic involvement of the axillary SLN in a definite subgroup of patient candidates for SLNB. Thus, if axillary lymph node metastases are the most relevant prognostic parameter, the correlated features of the primary tumor should be given the same clinical relevance. Many clinicopathologic characteristics that are still considered to be important independent prognostic factors should be reconsidered. Although some prognostic variables, like tumor size, PVI, and favorable histology, were correlated with axillary lymph node status in our series, other variables (e.g., age, tumor grade, ER status, and HER-2/neu expression) were not, at least in this selected population of patients with relatively small primary tumors and clinically uninvolved axilla.

Age, which is correlated with SLN metastases in univariate analysis, does not retain any significant correlation in multivariate analysis. Therefore, if young age, by itself, is not an adverse prognostic factor, it should not be considered as an indicator for more aggressive treatments.

Increasing tumor grade does not predict a higher risk for axillary lymph node metastases in multivariate analysis. In particular, Grade 3 tumors do not show any increased propensity to metastasize to the regional lymph nodes and perhaps to distant organs. Therefore, any possible overtreatment of patients on the basis of tumor grade only should be discouraged.

PgR status shows a surprising inverse correlation with axillary lymph node metastases, both in univariate and multivariate analyses. The possible biologic reasons for this association deserve further investigations.

One of the aims of our study was the possible identification of a clinicopathological profile of the primary tumors predictive of such a low risk of axillary involvement as to justify the avoidance of SLNB. Although this procedure is well tolerated by the patients, it still requires the administration of radioactive tracers in the breast, a number of scintigraphic examinations, a prolonged surgical procedure, and additional work for pathologists. Patients with tumors exhibiting PVI—and especially if the tumors are > 2 cm— are at the highest risk for axillary lymph node involvement, and an immediate ALND could be recommended without a preliminary SLNB. The information on PVI, however, is available only after the primary tumor has been removed whereas, n many cases, the indications for SLNB must be derived preoperatively.

Given that we were unable to find any constellation of predictive features that would identify patients at such a low (i.e., significantly < 10%) risk of SLN metastases to be safely spared SLNB, we continue to offer this staging procedure to all eligible patients.

Finally, the current series confirms that the particularly thorough examination of SLN increases the detection rate of micrometastases and isolated tumor cells that would have remained undetected at a routine histopathologic examination of ALND specimens. This will upstage a significant number of patients with all the subsequent therapeutic consequences. As there is uncertainty about the clinical implications of minimal lymph node involvement, the treatment policy of this subgroup of patients will require a redefinition. Specifically, it is still being debated whether minimal SLN involvement actually dictates completion of ALND. A clinical trial comprising patients with isolated tumor cells or micrometastases in the SLN randomized to receive ALND or follow-up is in progress at our institution.

In conclusion, the current data may be used to tailor the management of patients with breast carcinoma with the aim of minimizing as much as possible diagnostic and therapeutic procedures, thus improving the quality of life of the patients without any adverse effect on their survival rates.