Bringing doctors together

or

A free, private network for verified physicians

Leading voices

Dr. Jad M Abdelsattar, M.D, FACS

Breast Surgical Onc · University of Oxford

View

Dr. Lauren Kopicky

Breast Surgical Onc · Cleveland Clinic

View

Dr. Jennifer Plichta

Breast Surgical Onc · Duke University

View

Dr. Sarah Blair

Breast Surgical Onc · UCSD

View

Dr. Karen Goodwin

Breast Surgical Onc · UCSF

View

Dr. Christin Collins

Breast Surgical Onc · MetroHealth

View

Dr. Philip Schauer

Bariatric Surgery · PBRC

View

Dr. Todd Rosengart

Cardiac Surgery · BCM

View

Dr. Larry Kaiser

Thoracic Surgery · UPenn

View

Dr. Neri Cohen

Thoracic Surgery · XVIVO

View
  • Discover your community

  • Stay current

  • A profile that represents you

  • Find your people

About RoonPrivacyTerms

    Overview

    Management and Outcomes in Patients with Breast Cancer with 1-of-1 and 2-of-2 Positive Sentinel Nodes.

    Aug 23, 2025DOI: 10.1245/s10434-025-18097-9Journal ArticleASO

    Written by: Walker LyonsE McAuley FishRichard J BleicherCecilia ChangAllison A AggonAlycia L SoAndrea S PorpigliaAustin D Williams

    Find these authors on Roon

    AW
    Dr. Austin Williams
    Breast Surgical Onc · Fox Chase Cancer Center

    Patients and Methods

    We performed a retrospective analysis of the National Cancer Database (NCDB), a joint effort of the American College of Surgeons and the American Cancer Society. The NCDB contains deidentified patient-level data from more than 1500 Commission-on-Cancer accredited programs and captures approximately 70% of all malignancies in the United States.11,12From the NCDB breast cancer Participant User File, we identified female patients ≥ 18 years of age who underwent upfront surgery with SLNB for their first clinical tumor (cT)1–T3, N0, M0 breast cancer between 2018 and 2021 (Supplementary Fig. 1). Patients treated prior to 2018 were not included in the analyses because 2018 is the first year that the NCDB began recording specific SLN information such as the number of nodes removed and the number positive for metastasis. We excluded patients if they had clinically occult breast cancer, if their breast or axillary surgery was unknown, and if they had any form of neoadjuvant therapy. Finally, we excluded patients who had > 2 SLNs removed, and those with 2 SLNs retrieved with all being negative.Patients were then grouped on the basis of the number of SLNs positive/removed: 0/1, 1/1, 1/2, and 2/2. We compared demographic and clinicopathological characteristics and surgical approach between SLN groups. The proportion of patients who underwent ALND was calculated, and among these patients the total number of lymph nodes (LNs) examined and positive were compared. The use of ALND over time was analyzed for all SLN groups. We assessed the factors within the overall cohort associated with receipt of ALND, and among those undergoing ALND, the factors associated with finding additional +LNs.Subgroup analyses of surgical approach and adjuvant therapies were performed on the basis of receptor subtype. For patients with hormone receptor (HR) positive, human epidermal growth factor receptor 2 (HER2) negative breast cancer, we also analyzed the use and results of 21-gene recurrence score testing (Oncotype DX, Genomic Health) in the overall group and in groups stratified by age (≤ 50 and > 50) as a surrogate for menopausal status (as was done in the RxPONDER study when menopausal status was unknown).13Finally, we performed analyses of overall survival (OS). Unadjusted OS was compared among SLN group according to ALND receipt.Comparisons between groups were made using chi-squared test, Student’s t-test, and analysis of variance (ANOVA), as appropriate. Univariate and multivariable logistic regression models, adjusting for pertinent clinicopathologic and demographic features, were performed. The Kaplan–Meier method and log-rank tests were used to analyze OS. Statistical tests were two-sided, and a p-value < 0.05 was considered significant. All analyses were performed using SAS version 9.4 (SAS Institute, Inc., Cary, NC, USA). The study was deemed exempt by the Fox Chase Cancer Center Institutional Review Board.

    Results

    In total, 179,918 patients with cT1-3N0 breast cancer who underwent upfront surgery with SLNB and had < 2 SLNs removed were identified (Supplementary Fig. 1). After excluding 77,116 patients with 0/2 +SLNs, we were left with an analysis cohort of 102,802 patients. Patients were stratified by the number of +SLNs and total number of SLNs removed. Most patients (n = 79,106, 77%) had one negative LN removed (0/1) while 10% (n = 10,549) had 1/1 + SLNs, 10% (n = 10,068) had 1/2 +SLNs, and 3% (n = 3079) had 2/2 +SLNs (Table 1). Patients in the 0/1 +SLNs group were older and more likely to have cT1 tumors, ductal histology, and low-grade and absent lymphovascular invasion (allp < 0.001) when compared with the other SLN groups. Conversely, patients with 2/2 +SLNs were more likely to have cT3 tumors, lobular histology, high tumor grade, and presence of lymphovascular invasion (all p < 0.001). A lower proportion of patients with HR+/HER2− cancers had 0/1 +SLNs when compared with patients with HER2+ and triple negative breast cancers (TNBC) (76% versus 81% and 85%, respectively, p < 0.001). There was very little difference between the SLN groups based on race/ethnicity, Charlson–Deyo score, insurance status, and high school education.Table 1Clinicodemographic featuresOverallSentinel lymph node status (positive/removed)p0/11/11/22/2N102,80279,10610,54910,0683079Age at diagnosis, years [median (IQR)]63 (52–74)64 (53–75)61 (49–73)60 (48–72)60 (48–72)< 0.001Race/ethnicity< 0.001NH white80,219 (78.0)62,225 (78.7)8045 (76.3)7562 (75.1)2387 (77.5)NH Black8822 (8.6)6604 (8.4)1004 (9.5)942 (9.4)272 (8.8)Hispanic6077 (5.9)4476 (5.7)686 (6.5)710 (7.1)205 (6.7)Asian4998 (4.9)3768 (4.8)515 (4.9)572 (5.7)143 (4.6)Other/unknown2686 (2.6)2033 (2.6)299 (2.8)282 (2.8)72 (2.3)Charlson–Deyo score0.06083,576 (81.3)64,164 (81.1)8576 (81.3)8294 (82.4)2542 (82.6)113,232 (12.9)10,288 (13.0)1336 (12.7)1242 (12.3)366 (11.9)≥ 25994 (5.8)4654 (5.9)637 (6.0)532 (5.3)171 (5.6)Insurance status< 0.001Not insured992 (1.0)670 (0.9)133 (1.3)153 (1.5)36 (1.2)Private insurance49,445 (48.1)36,966 (46.7)5381 (51.0)5444 (54.1)1654 (53.7)Medicaid5503 (5.4)3867 (4.9)741 (7.0)682 (6.8)213 (6.9)Medicare44,881 (43.7)36,113 (45.7)4070 (38.6)3575 (35.5)1123 (36.5)Other government1165 (1.1)861 (1.1)135 (1.3)134 (1.3)35 (1.1)Unknown816 (0.8)629 (0.8)89 (0.8)80 (0.8)18 (0.6)No high school degree< 0.001≥ 15.3%13,799 (13.4)10,432 (13.2)1472 (14.0)1464 (14.5)431 (14.0)9.1–15.2%22,571 (22.0)17,244 (21.8)2396 (22.7)2218 (22.0)713 (23.2)5.0–9.0%26,629 (25.9)20,612 (26.1)2691 (25.5)2537 (25.2)789 (25.6)< 5.0%23,102 (22.5)17,986 (22.7)2241 (21.2)2229 (22.1)646 (21.0)Unknown16,701 (16.3)12,832 (16.2)1749 (16.6)1620 (16.1)500 (16.2)Median income0.24< $46,27710,862 (10.6)8268 (10.5)1161 (11.0)1086 (10.8)347 (11.3)$46,227–57,85616,815 (16.4)12,879 (16.3)1771 (16.8)1644 (16.3)521 (16.9)$57,857–74,06220,678 (20.1)16,002 (20.2)2050 (19.4)2027 (20.1)599 (19.5)≥ $74,06337,580 (36.6)28,994 (36.7)3799 (36.0)3678 (36.5)1109 (36.0)Unknown16,867 (16.4)12,963 (16.4)1768 (16.8)1633 (16.2)503 (16.3)Institution type< 0.001Community cancer center7880 (7.7)6073 (7.7)824 (7.8)768 (7.6)215 (7.0)Comprehensive community cancer program44,477 (43.3)34740 (43.9)4483 (42.5)3970 (39.4)1284 (41.7)Academic/research program27,191 (26.5)20,745 (26.2)2756 (26.1)2856 (28.4)834 (27.1)Integrated network cancer program20,926 (20.4)16,192 (20.5)2071 (19.6)2040 (20.3)623 (20.2)Unknown2328 (2.3)1356 (1.7)415 (3.9)434 (4.3)123 (4.0)Clinical tumor stage< 0.001cT182,072 (79.8)66,788 (84.4)6872 (65.1)6712 (66.7)1700 (55.2)cT219,223 (18.7)11,656 (14.7)3304 (31.3)3092 (30.7)1171 (38.0)cT31507 (1.5)662 (0.8)373 (3.5)264 (2.6)208 (6.8)Receptor subtype< 0.001HR+/HER2−90,147 (87.7)68,502 (86.6)9640 (91.4)9204 (91.4)2801 (91.0)HER2+4332 (4.2)3507 (4.4)371 (3.5)339 (3.4)115 (3.7)TNBC6203 (6.0)5287 (6.7)395 (3.7)401 (4.0)120 (3.9)Unknown2120 (2.1)1810 (2.3)143 (1.4)124 (1.2)43 (1.4)Tumor histology< 0.001Ductal89,202 (86.8)69,313 (87.6)8846 (83.9)8644 (85.9)2399 (77.9)Lobular12,830 (12.5)9136 (11.6)1653 (15.7)1374 (13.7)667 (21.7)Mixed/other/unknown770 (0.8)657 (0.8)50 (1.4)50 (0.5)13 (0.4)Tumor grade< 0.001Low28,931 (28.1)24,505 (31.0)1997 (18.9)1927 (19.1)502 (16.3)Intermediate52,118 (50.7)38,417 (48.6)6108 (57.9)5760 (57.2)1833 (59.5)High19,747 (19.2)14,386 (18.2)2332 (22.1)2307 (22.9)722 (23.5)Unknown2006 (2.0)1798 (2.3)112 (1.1)74 (0.7)22 (0.7)Lymphovascular invasion< 0.001Absent78,341 (76.2)65612 (82.9)1136 (10.8)5704 (56.7)1364 (44.3)Present13,832 (13.5)5401 (6.8)3752 (35.6)3284 (32.6)1395 (45.3)Unknown10,629 (10.4)8093 (10.2)5661 (53.7)1080 (10.7)320 (10.4)Data are expressed as n (%) unless otherwise specified.NH non-Hispanic, HR hormone receptor, HER2 human epidermal growth factor receptor 2, TNBC triple negative breast cancer*Results suppressed: NCDB does not permit aggregate results for cell sizes < 10Clinicodemographic featuresData are expressed as n (%) unless otherwise specified.NH non-Hispanic, HR hormone receptor, HER2 human epidermal growth factor receptor 2, TNBC triple negative breast cancer*Results suppressed: NCDB does not permit aggregate results for cell sizes < 10Overall, partial mastectomy was more commonly performed than mastectomy (75% versus 25%, Table 2), but mastectomy was more common in the 2/2 +SLN group, when compared with the other SLN groups (48% versus 0/1: 21%, 1/1: 37%, 1/2: 36%, p < 0.001). Similar to cT stage, pathologic tumor (pT) stage increased as the number of positive nodes increased (p < 0.001). Most patients had both cT1 (80%) and pT1 (73%) tumors. There was, however, a notable upstage rate in the 2/2 +SLN group, from 7% cT3 to 13% pT3 when compared with the other groups (0/1: 1–1%, 1/1: 4–6%, 1/2: 3–5%).Table 2Surgical approach and pathologic stageOverallSentinel lymph node statusp0/11/11/22/2N102,80279,10610,54910,0683079Breast surgeryPartial mastectomy77,529 (75.4)62,806 (79.4)6660 (63.1)6456 (64.1)1607 (52.2)< 0.001Mastectomy25,273 (24.6)16,300 (20.6)3889 (36.9)3612 (35.9)1472 (47.8)Pathologic tumor stage< 0.001pT084 (0.1)83 (0.10)pTis126 (0.1)124 (0.16)pT175,196 (73.2)63,020 (79.7)5472 (51.9)5463 (54.3)1241 (40.3)pT224,273 (23.6)14,433 (18.3)4330 (41.1)4082 (40.5)1428 (46.4)pT32436 (2.4)896 (1.1)684 (6.5)468 (4.7)388 (12.6)pT4123 (0.1)37 (0.05)38 (0.4)31 (0.3)17 (0.6)pTX188 (0.18)175 (0.22)*Unknown376 (0.37)338 (0.43)17 (0.16)16 (0.16)Axillary surgery< 0.001SLNB alone91,851 (89.4)74,035 (93.6)7814 (74.1)8190 (81.4)1812 (58.9)SLNB then ALND10,951 (10.7)5071 (6.4)2735 (25.9)1878 (18.7)1267 (41.2)Results of ALNDAdditional LNs removed6.6 ± 6.82.8 ± 3.510.0 ± 7.28.5 ± 7.111.2 ± 7.4< 0.001Additional LNs positive0.9 ± 2.70.2 ± 0.91.6 ± 3.70.6 ± 1.92.6 ± 4.2< 0.001Total LNs removed7.8 ± 7.03.8 ± 3.511.0 ± 7.210.5 ± 7.113.2 ± 7.4< 0.001Total LNs positive1.5 ± 3.00.2 ± 0.92.6 ± 3.71.6 ± 1.94.6 ± 4.2< 0.001Results of ALND< 0.001No additional positive nodes8268 (72.6)4716 (93.1)1629 (59.6)1372 (73.1)551 (43.6)Additional positive nodes2675 (24.4)351 (6.9)1104 (40.4)506 (26.9)714 (56.4)Pathologic nodal stage (overall)< 0.001pN078,484 (76.3)78,038 (98.7)177 (1.7)231 (2.3)38 (1.2)pN122,691 (22.1)627 (0.8)9863 (93.5)9699 (96.3)2502 (81.3)pN2867 (0.8)44 (0.1)339 (3.2)97 (1.0)387 (12.6)pN3336 (0.3)11 (0.01)152 (1.4)26 (0.3)147 (4.8)pNX85 (0.1)84 (0.1)**Unknown339 (0.3)302 (0.4)Pathologic nodal stage (SLNB alone)< 0.001pN073,754 (80.3)73367 (99.1)160 (2.1)198 (2.4)29 (1.6)pN117,664 (19.2)312 (0.4)7624 (97.6)7973 (97.4)1755 (96.9)pN2/359 (0.1)19 (0.3)*27 (1.5)pNX81 (0.1)Unknown293 (0.3)272 (0.4)10 (0.1)*Pathologic nodal stage (SLNB then ALND)< 0.001pN04730 (43.2)4671 (92.1)17 (0.6)33 (1.8)9 (0.7)pN15027 (45.9)315 (6.2)2239 (81.9)1726 (91.9)747 (59.0)pN2827 (7.6)41 (0.8)327 (12.0)91 (4.9)368 (29.0)pN3317 (2.9)10 (0.2)145 (5.3)23 (1.2)139 (11.0)pNX****Unknown*Data are expressed as n (%) unless otherwise specifiedNH non-Hispanic, HR hormone receptor, HER2 human epidermal growth factor receptor 2, TNBC triple negative breast cancer*Results suppressed: NCDB does not permit aggregate results for cell sizes < 10Surgical approach and pathologic stageData are expressed as n (%) unless otherwise specifiedNH non-Hispanic, HR hormone receptor, HER2 human epidermal growth factor receptor 2, TNBC triple negative breast cancer*Results suppressed: NCDB does not permit aggregate results for cell sizes < 10Overall, 11% of patients underwent a cALND either in the same operation or during a subsequent procedure. When examined over time, the rate of cALND decreased slightly over time (Fig. 1). As expected, cALND was most commonly performed in patients with 2/2 +SLNs (41% versus 0/1: 6%, 1/1: 26%, 1/2: 19%, p < 0.001). Among patients in whom cALND was performed, additional +LNs were found in 24% of patients. When examined by SLN group, additional +LNs were found in more than half of patients with 2/2 +SLNs (56%, 714/1267), in 40% (1104/2735) of patients with 1/1 +SLN, and in 27% (506/1878) of patients with 1/2 +SLNs. As expected, it was rare to find any +LNs when cALND was performed in patients with 0/1 +SLN (7%, 351/5071).Fig. 1Rates of ALND over time; ALND axillary lymph node dissection, SLNs sentinel lymph nodesRates of ALND over time; ALND axillary lymph node dissection, SLNs sentinel lymph nodesOverall, a mean of 6.6 ± 6.8 additional LNs were removed when cALND was performed with 0.9 ± 2.7 additional +LNs found. Interestingly, patients with 0/1 +SLNs had only a mean of 2.8 additional LNs removed at the time of cALND, which was lower than the other SLN groups (1/1: 10.0, 1/2: 8.5, 2/2: 11.2, p < 0.001). In line with other findings, the mean number of additional positive nodes found on cALND was highest among patients with 2/2 +SLNs (2.6 versus 0/1: 0.2, 1/1: 1.6, 1/2: 0.6, p < 0.001). Overall, 24% of patients upstaged from cN0 to ≥ pN1, something that was more common for patients who had cALND versus SLNB alone (57% versus 20%). Only 1% of patients were ultimately staged as pN2–3. Within the 2/2 +SLN group, 0.4% of patients who had SLNB alone had pN2–3 disease compared with 40% of patients who had cALND. This upstage in the SLNB alone group can be explained either by data entry error or a small proportion of patients with evidence of nodal involvement in basins outside the axilla on subsequent imaging.On univariate analysis, almost all clinicodemographic and pathologic features had some association with receipt of cALND (Table 3). Subsequent multivariable analysis adjusting for pertinent features found that the strongest independent predictor of cALND was SLN status. Specifically, having 2/2 +SLNs was the strongest predictor (OR 6.8; 95% CI 6.1–7.5), followed by 1/1 +SLNs (OR 3.8; 95% CI 3.5–4.1) and 1/2 +SLNs (OR 2.6; 95% CI 2.4–2.8) when compared with patients with 0/1 +SLNs. Other predictors of ALND were Black race (OR 1.2; 95% CI 1.1–1.3) and Hispanic ethnicity (OR 1.3; 95% CI 1.2–1.4) when compared with non-Hispanic white patients. Additional predictors of ALND included the highest quartile of educational attainment, treatment in a community cancer center, more advanced tumor stage, HER2+ and TNBC subtypes, presence of lymphovascular invasion, and mastectomy as the surgical approach (all p < 0.05). Age and tumor histology (ductal versus lobular) were not independent predictors of cALND.Table 3Univariate and multivariable analysis of factors predictive of ALND receiptUnivariateMultivariableOR95% CIpOR95% CIpAge at diagnosis (years)0.980.98–0.98< 0.0011.001.00 - 1.000.91Race/ethnicityNH white (ref.)––––––NH Black1.21.12–1.28< 0.0011.181.08–1.30< 0.001Hispanic1.461.36–1.58< 0.0011.281.15–1.41< 0.001Asian1.131.03–1.240.0010.990.88–1.110.84Charlson–Deyo score0 (ref.)––––––11.010.95–1.070.750.990.91–1.070.7120.900.81–1.010.080.880.76–1.030.11≥ 30.950.83–1.080.430.960.81–1.150.68Insurance statusNot insured (ref.)––––––Private insurance0.750.63–0.900.0020.910.72–1.160.45Medicaid0.940.78–1.140.540.920.72–1.190.55Medicare0.600.50–0.72< 0.0010.910.71–1.160.44Other government0.750.58–0.960.020.940.68–1.320.74No high school degree≥ 15.3% (ref.)––––––9.1–15.2%0.860.80–0.92< 0.0010.890.82–0.960.0055.0–9.0%0.810.76–0.87< 0.0010.860.79–0.94< 0.001< 5.0%0.730.68–0.78< 0.0010.790.72–0.87< 0.001Median income< $46,277 (ref.)––––––$46,227–57,8560.950.88–1.030.211.030.94–1.130.55$57,857–74,0620.940.87–1.010.091.101.00–1.210.06≥ $74,0630.860.80–0.92< 0.0011.050.95–1.160.31Institution typeCommunity cancer center (ref.)––––––Comprehensive community cancer program0.760.71–0.82< 0.0010.720.66–0.79< 0.001Academic/research program0.730.68–0.79< 0.0010.710.64–0.78< 0.001Integrated network cancer program0.700.65–0.76< 0.0010.660.59–0.73< 0.001Clinical tumor stagecT1 (ref.)––––––cT22.182.08–2.28< 0.0011.171.10–1.25< 0.001cT35.094.56–5.69< 0.0011.501.29–1.74< 0.001Receptor subtypeHR+/HER2− (ref.)––––––HER2+1.131.03–1.250.0011.080.95–1.230.22TNBC1.070.99–1.160.111.161.03–1.300.01Tumor histologyDuctal (ref.)––––––Lobular1.391.31–1.46< 0.0011.040.96–1.120.34Mixed/other1.100.86–1.400.450.850.60–1.210.37Tumor gradeLow (ref.)––––––Intermediate1.571.49–1.65< 0.0011.091.02–1.170.01High1.911.80–2.03< 0.0011.201.10–1.31< 0.001Lymphovascular invasionAbsent (ref.)––––––Present2.812.68 - 2.94< 0.0011.231.15 - 1.31< 0.001Breast surgeryPartial mastectomy (ref.)––––––Mastectomy4.474.29–4.65< 0.0013.493.30–3.68< 0.001SLN status0/1 (ref)––––––1/15.114.85–5.38< 0.0013.803.54–4.08< 0.0011/23.353.16–3.55< 0.0012.562.37–2.77< 0.0012/210.219.45–11.03< 0.0016.766.09–7.50< 0.001ALND axillary lymph node dissection, NH non-Hispanic, HR hormone receptor, HER2 human epidermal growth factor receptor 2, TNBC triple negative breast cancerUnivariate and multivariable analysis of factors predictive of ALND receiptALND axillary lymph node dissection, NH non-Hispanic, HR hormone receptor, HER2 human epidermal growth factor receptor 2, TNBC triple negative breast cancerAn additional regression model was created to assess factors predictive of finding additional +LNs among patients who underwent cALND. Univariate analysis showed that all tumor and treatment factors, younger age, and institution type were associated with additional +LNs (Table 4). On multivariable analysis, the strongest predictor of having additional +LNs nodes was again SLN status. Odds of finding additional +LNs were ten times higher among patients with 2/2 +SLNs than those with 0/1+ SLNs, six times higher in those with 1/1 +SLNs, and 3.5 times higher in the 1/2 +SLN group. Other independent predictors included presence of lymphovascular invasion (OR 2.2; 95% CI 1.9–2.5), cT2 (OR 1.4; 95% CI 1.3–1.6) and cT3 (OR 1.9; 95% CI 1.4–2.5) stage, mastectomy as the surgical approach (OR 1.6; 95% CI 1.4–1.8), lobular histology (OR 1.4; 95% CI 1.2–1.7), and intermediate (OR 1.3; 95% CI 1.1–1.5) and high (OR 1.3; 95% CI 1.1–1.7) tumor grade.Table 4Univariate and multivariable analysis of factors associated with finding additional positive nodes among those undergoing ALNDUnivariateMultivariableOR95% CIpOR95% CIpAge at diagnosis (years)0.990.99–0.99< 0.0011.000.99–1.010.88Race/ethnicityNH white (ref.)––––––NH Black0.940.81–1.090.411.090.88–1.360.42Hispanic0.930.79–1.100.390.790.62–1.010.06Asian0.890.73–1.090.250.710.54–0.940.02Charlson–Deyo score0 (ref)––––––11.000.88–1.140.990.960.79–1.160.6921.060.83–1.360.631.200.85–1.700.30≥ 30.890.65–1.210.440.980.64–1.490.91Insurance statusNot insured (ref.)––––––Private insurance1.060.72–1.560.751.380.80–2.370.25Medicaid1.240.82–1.870.301.770.99–3.160.05Medicare0.930.63–1.370.701.440.83–2.520.20Other government1.020.59–1.760.961.000.46–2.170.99No high school degree≥ 15.3% (ref)––––––9.1–15.2%1.010.87–1.160.920.980.81–1.190.845.0–9.0%0.950.82–1.090.440.980.80–1.210.86< 5.0%0.870.75–1.010.060.940.74–1.190.61Median income< $46,277 (ref.)––––––$46,227–57,8560.950.80–1.120.510.960.77–1.200.74$57,857–74,0620.920.78–1.070.280.930.74–1.160.50≥ $74,0630.860.74–0.990.040.840.66–1.060.13Institution typeCommunity cancer center (ref.)––––––Comprehensive community cancer program1.211.03–1.430.020.960.76–1.210.73Academic/research program1.291.08–1.530.000.960.75–1.220.74Integrated network cancer program1.241.04–1.490.020.930.72–1.200.57Clinical tumor stagecT1 (ref.)––––––cT22.462.24–2.70< 0.0011.431.25–1.64< 0.001cT35.064.19–6.10< 0.0011.911.41–2.47< 0.001Receptor subtypeHR+/HER2− (ref.)––––––HER2+0.890.72–1.090.261.020.76–1.380.88TNBC0.540.44–0.67< 0.0010.830.61–1.110.21Tumor histologyDuctal (ref.)-Lobular1.781.59–1.98< 0.0011.401.19–1.65< 0.001Mixed/other0.660.36–1.230.190.810.33–2.020.65Tumor gradeLow (ref.)––––––Intermediate2.021.78–2.30< 0.0011.261.05–1.510.02High2.081.80–2.41< 0.0011.341.08–1.670.01Lymphovascular invasionAbsent (ref)––––––Present4.363.95 - 4.81< 0.0012.171.90 - 2.47< 0.001Breast surgeryPartial mastectomy (ref.)––––––Mastectomy2.862.60–3.15< 0.0011.581.37–1.81< 0.001SLN status0.00–0.00–0/1 (ref)––––––1/19.117.97–10.40< 0.0016.015.02–7.20< 0.0011/24.964.27–5.75< 0.0013.492.86–4.25< 0.0012/217.4114.91–20.34< 0.00110.058.16–12.38< 0.001ALND axillary lymph node dissection, NH non-Hispanic, HR hormone receptor, HER2 human epidermal growth factor receptor 2, TNBC triple negative breast cancerUnivariate and multivariable analysis of factors associated with finding additional positive nodes among those undergoing ALNDALND axillary lymph node dissection, NH non-Hispanic, HR hormone receptor, HER2 human epidermal growth factor receptor 2, TNBC triple negative breast cancerAfter stratifying the cohort by receptor subtype, the use of adjuvant therapies was compared between pN+ patients undergoing SLNB alone and cALND in each SLN group. Among patients with HER2+ or TNBC, cALND was not associated with differences in the rates of adjuvant chemotherapy or radiation therapy for nearly all groups (Supplementary Tables 1, 2). The only observed difference was a lower rate of adjuvant radiation among patients who had cALND in both patient with HER2+ and TNBC (p = 0.02 and p = 0.003, respectively), though these differences did not persist after stratifying by breast surgical approach, likely reflecting the higher rate of mastectomy in the cALND group. The major differences in adjuvant therapy were seen in pN+ HR+/HER2− patients (Supplementary Table 3). Overall, adjuvant chemotherapy was most commonly given to patients with 2/2 +SLNs who underwent cALND (65%) and least in those with 1/1+ SLNs who had SLNB alone (28%), with significant differences noted when comparing SLN groups and axillary surgical approach (both p < 0.001).Since current clinical practice supports using the 21-gene recurrence score (RS) to guide systemic therapy only among pN+ patients who are post-menopausal, we created a cohort of the 16,288 HR+/HER2− patients > 50 years old for further analysis (Table 5). Within this group, RS testing was performed in 55% of patients and was less common among patients who had cALND and 2/2 +SLNs (both p < 0.001). The mean RS of the cohort was 16, with > 80% of patients in each group having RS ≤ 25. There was no difference in mean RS seen when comparing SLN group (p = 0.55) or by ALND receipt (p = 0.43). Despite this lack of difference in RS in node-positive HR+/HER2− patients > 50, those who underwent cALND had a higher rate of adjuvant chemotherapy in all SLN groups (1/1: 40% versus 23%; 1/2: 33% versus 22%; 2/2: 60% versus 35%; all p < 0.0001). When the cohort was stratified by pathologic nodal stage, this higher rate of adjuvant chemotherapy for cALND persisted in patients with pN1 disease (p < 0.001), but not those with pN2 disease (p = 0.69, data not shown). Similar to patients with HER2+ and TNBC, those who had cALND were less likely to undergo adjuvant radiotherapy (p < 0.001).Table 5Adjuvant therapies among pN+ patients > 50 with HR+/HER2− breast cancerSentinel lymph node statusp1p21/11/22/2SLNB onlycALNDSLNB onlycALNDSLNB onlycALNDn56591763564811141288816Recurrence scoreNot performed2297 (40.6)1017 (57.7)2217 (39.3)532 (47.8)641 (49.8)579 (71.0)< 0.001< 0.001Performed3362 (59.4)746 (42.3)3431 (60.8)582 (52.2)647 (50.2)237 (29.0)Mean ± SD16.2 ± 9.316.2 ± 9.216.0 ± 9.016.6 ± 10.115.8 ± 8.415.9 ± 9.10.550.43Score ≤ 252816 (83.76)633 (84.85)2908 (84.76)472 (81.10)567 (87.64)200 (84.39)0.110.09Score > 25436 (12.97)90 (12.06)401 (11.69)94 (16.15)61 (9.43)30 (12.66)Chemotherapy< 0.001< 0.001No/unknown4567 (80.7)1050 (59.6)4416 (78.2)742 (66.6)841 (65.3)326 (40.0)Yes1272 (22.5)713 (40.4)1232 (21.8)372 (33.4)447 (34.7)490 (60.1)Endocrine therapy0.12< 0.001No447 (7.9)182 (10.3)423 (7.5)96 (8.6)108 (8.4)76 (9.3)Yes5145 (90.9)1548 (87.8)5150 (91.2)992 (89.1)1152 (89.4)721 (88.4)Unknown67 (1.2)33 (1.9)75 (1.3)26 (2.3)28 (2.2)19 (2.3)Radiation: all patients< 0.001< 0.001No835 (14.8)600 (34.0)902 (16.0)386 (34.7)153 (11.9)163 (20.0)Yes4756 (84.0)1137 (64.5)4671 (82.7)707 (63.5)1106 (85.9)638 (78.2)Unknown68 (1.2)26 (1.5)75 (1.3)21 (1.9)29 (2.3)15 (1.8)Radiation: mastectomy patients< 0.0010.23No510 (42.4)533 (45.9)587 (44.1)347 (49.7)80 (22.2)137 (25.6)Yes671 (55.8)614 (52.8)721 (54.1)337 (48.3)267 (74.2)388 (72.4)Unknown21 (1.8)15 (1.3)24 (1.8)14 (2.0)13 (3.6)11 (2.1)Radiation: lumpectomy patients< 0.001< 0.001No325 (7.3)67 (11.2)315 (7.3)39 (9.4)73 (7.9)26 (9.3)Yes: breast only1730 (38.8)185 (30.8)1725 (40.0)149 (35.8)241 (26.0)49 (17.5)Yes: breast and LNs2146 (48.2)298 (49.6)2018 (46.8)193 (46.4)543 (58.5)180 (64.3)Yes: other181 (4.1)35 (5.8)182 (4.2)49 (5.3)Unknown75 (1.7)16 (2.7)76 (1.8)22 (2.4)Data are expressed as n (%) unless otherwise specified.pN pathological nodal stage, HR hormone receptor, HER2 human epidermal growth factor receptor 2, TNBC triple negative breast cancer, SLNB sentinel lymph node biopsy, cALND completion axillary lymph node dissection, LN lymph node,p*1 comparison between LN groups,p*2 comparison between no ALND versus ALND*Results suppressed: NCDB does not permit aggregate results for cell sizes < 10Adjuvant therapies among pN+ patients > 50 with HR+/HER2− breast cancerData are expressed as n (%) unless otherwise specified.pN pathological nodal stage, HR hormone receptor, HER2 human epidermal growth factor receptor 2, TNBC triple negative breast cancer, SLNB sentinel lymph node biopsy, cALND completion axillary lymph node dissection, LN lymph node,p*1 comparison between LN groups,p*2 comparison between no ALND versus ALND*Results suppressed: NCDB does not permit aggregate results for cell sizes < 10We compared unadjusted OS between patients who did and did not undergo cALND for each SLN group. With a median follow up of 35.4 months, 5-year OS was ≥ 90% in every group, and cALND was not associated with improved OS in any SLN group (Fig. 2). The only difference we observed in unadjusted OS was in the 1/1 +SLN group, where patients undergoing cALND had a worse OS (p < 0.001).Fig. 2Kaplan–Meier curves for overall survival comparing SLNB alone versus SLNB+ALND for each SLN group; A Kaplan–Meier curves for overall survival 0/1 +SLN, B Kaplan–Meier curves for overall survival 1/1 +SLN, C Kaplan–Meier curves for overall survival 1/2 +SLN, D Kaplan–Meier curves for overall survival 2/2 +SLN; SLNB sentinel lymphadenectomy, ALND axillary lymph node dissection, SLN sentinel lymph nodeKaplan–Meier curves for overall survival comparing SLNB alone versus SLNB+ALND for each SLN group; A Kaplan–Meier curves for overall survival 0/1 +SLN, B Kaplan–Meier curves for overall survival 1/1 +SLN, C Kaplan–Meier curves for overall survival 1/2 +SLN, D Kaplan–Meier curves for overall survival 2/2 +SLN; SLNB sentinel lymphadenectomy, ALND axillary lymph node dissection, SLN sentinel lymph node

    Discussion

    In this analysis of a large national database, we found that ALND is being performed at higher-than-expected rates in patients found to have ≤ 2 SLNs retrieved, where all are positive. This behavior flouts randomized evidence demonstrating the safety of ALND omission in this situation. This therapeutic excess may extend to overtreatment in the adjuvant setting for some patients without providing any survival benefit.To date, this is the first study that evaluates patients who fall into a poorly defined area of axillary management: those in whom ≤ 3 SLNs are retrieved and all are positive. Due to concerns for additional axillary disease in these patients, surgeons (and others) often question whether ALND should be performed for local control and/or to change adjuvant treatment recommendations if more disease is found. This is likely the reason we found that SLN status was the strongest predictor of cALND, surpassing any other clinicopathologic factor. The extent of the ALNDs being performed does need to be questioned given the low number of additional LNs, as most cALND should include > 10 additional LNs, especially in the non-neoadjuvant patient. Perhaps the extent of ALND performed was influenced by the surgeon’s expectation of finding additional disease. This contradicts data from Z0011 and AMAROS that demonstrated the safety of ALND omission despite additional axillary disease found when ALND was performed in 27% and 33% of patients, respectively. 5,6 Overall, our analysis demonstrated a similar proportion of additional axillary disease (24%) in patients who underwent cALND. Although many would consider the patients in our cohort to be at higher risk for harboring additional axillary disease than those enrolled in the randomized trials (many of whom likely had negative SLNs retrieved), these comparable findings suggest that applying the paradigms of cALND omission is reasonable. We did find that 40% of patients with 1/1 +SLN and more than half of patients in the 2/2 +SLN group had additional axillary disease, which is similar to the findings in the SENOMAC trial, where 31% of patients had additional axillary disease in patients with one +SLN and 51% of patients with two +SLNs.8 This does give some credence to the concern of surgeons about leaving behind clinically significant disease in these patients. Therefore, cALND should be considered on a case-by-case basis in these patients after a thorough multidisciplinary review of risks and benefits.Any decision to perform a cALND for the perceived benefit in improved local and distant control needs to be weighed against the risks that come with the operation. The increased morbidity of ALND, specifically the risk of lymphedema, but also of seroma formation, paresthesias, infection, and decreased shoulder mobility, should not be taken lightly. Lymphedema may be associated with decreased OS in patients with breast cancer.14 The incidence of arm lymphedema in women with breast cancer is approximately 16%, but is estimated to be four times higher in women who undergo ALND versus those who undergo SLNB alone.15 This has been a driving force behind continued studies on de-escalation of axillary treatment over the past 30 years. Additionally, Black race and Hispanic ethnicity have been identified as independent risk factors in the development of lymphedema.16 In our study, both Black race and Hispanic ethnicity were independent predictors of ALND, but not of finding additional positive nodes. Thus, these already-vulnerable patient populations are at a higher risk for surgical morbidity from ALND without deriving additional benefit, and surgeon education to reduce avoidable axillary overtreatment is crucial.Multidisciplinary teams may also advocate for cALND due to the perceived benefit that finding additional positive nodes would result in changing adjuvant therapy recommendations on the basis of the findings. While the surgical approach may change the adjuvant radiation therapy field planning and volume of nodal disease for some radiation oncologists, it is unlikely that pN1 versus ≥ pN2 disease changes whether adjuvant radiation is considered.17 This was confirmed in our analysis, as we found that ALND did not affect rates of adjuvant radiation in this study.Similarly, recommendations for adjuvant chemotherapy administration are unlikely to differ between pN1 and ≥ pN2 disease in patients with HER2+ and TNBC, and in pre-menopausal patients with HR+/HER2− disease. Our results confirm adherence to these treatment paradigms, with similar rates of adjuvant chemotherapy in these groups whether cALND was performed or not. While the receipt of adjuvant CDK4/6 inhibition is often based on the total number of positive nodes in patients with HR+/HER2− disease, we have previously demonstrated that cALND should not be routinely used to determine candidacy for this adjunctive therapy.18,19 One subgroup where pN upstaging could affect adjuvant chemotherapy decision-making is in post-menopausal patients with HR+/HER2− disease where ≥ pN2 patients are recommended adjuvant chemotherapy but pN1 patients are only recommended adjuvant chemotherapy if they are found to be at high risk based on genomic testing.17 We studied this specific cohort and found that despite similar recurrence scores, patients in the cALND group had higher rates of adjuvant chemotherapy use even among patients who remained stage pN1. This suggests that performance of ALND is associated with adjuvant chemotherapy overtreatment in this population.While this study and a recent study examining predictors of pN2 disease found that the strongest predictor of finding additional positive nodes with cALND was SLN status, there were several additional predictors of finding additional positive nodes that can be used as considerations when estimating its benefit in specific situations.20 These included lymphovascular invasion, cT2 and cT3 stage, mastectomy as the surgical approach, lobular histology, and intermediate and high tumor grade. These factors are similar to those used to create the additional non-SLN metastases nomogram from Memorial Sloan Kettering Cancer Center.21,22 Katz et al. found that increasing primary tumor size, histologic grade, number of +SLNs, and size of the largest SLN metastasis were the most predictive of upstaging to pN2 or greater, while an increasing number of uninvolved SLNs was protective against upstaging.23 Given the high likelihood of finding additional axillary disease and pN upstaging when 2/2 +SLNs are retrieved, it is suggested that in HR+/HER2− patients > 50 years old with RS ≤ 25, patients with 2/2 +SLNs be discussed in a multidisciplinary tumor board regarding the possible benefits of undergoing cALND. Unfortunately, due to the small number of patients, attempts at analyzing this cohort for changes in adjuvant treatments were underpowered.Local control with surgery and optimized adjuvant therapies are all employed toward the goal of improving disease-free survival and OS but need to be right sized for each patient and used only when there will be positive impact. From the evaluation in NSABP B-04 of radical mastectomy versus total mastectomy to more modern trials of axillary de-escalation, such as SINODAR-ONE and SENOMAC, there has been a consistent theme that no randomized control trial has ever shown a survival benefit from ALND. In line with these prospective studies, we found no evidence that cALND improved survival in this analysis and therefore should not be pursued in patients for whom this is the only goal of additional surgery.This study has several strengths and limitations. Its major strength is the use of a large, national database that permitted us to investigate current practice patterns in this patient population. Additionally, data regarding the number of sentinel and non-sentinel nodes that were removed and positive allowed us to analyze the incremental benefit of cALND when performed. The study is limited by its retrospective nature and lack of patient-specific decision-making data points. Specifically, ECE is missing and this may factor into some providers’ decision-making as it was part of the exclusion criteria in Z0011; although ECE was not excluded in the SINODAR-ONE and SENOMAC trials.7,8 In addition, the NCDB lacks cancer-specific outcomes such as disease-free survival along with rates of local, regional, and distance recurrence, which would be better endpoints when evaluating the efficacy of treatments. Additionally, we attempted to evaluate and compare OS in HR+/HER2− patients > 50 with pN1 versus ≥ pN2 disease who had a SLNB alone versus ALND, but due to small numbers in this cohort this analysis was underpowered. Therefore, this is something that should be further investigated. Lastly, due to SLNB-specific information not reported in NCDB until 2018, this study lacks long-term follow-up, as the median was only 35.4 months. Future work with longer follow-up and breast-cancer-specific outcomes is needed.

    Conclusions

    The criteria for ALND omission defined by both ACOSOG Z0011 and AMAROS can and should be followed for patients in whom fewer than 3 SLNs are retrieved. No further axillary surgery should be pursued in patients with 1 +SLN, but a detailed multidisciplinary discussion weighing the surgical morbidity and the potential benefits of local control and adjuvant therapy should be undertaken for patients with 2/2 +SLNs.

    Supplementary Information

    Below is the link to the electronic supplementary material.Supplementary file1 (DOCX 130 kb)Supplementary file1 (DOCX 130 kb)
    AWDr. Austin Williams

    Dr. Austin Williams

    Surgery · Breast Surgical Oncology
    Fox Chase Cancer Center
    austin