Written by: Mandeep KaurKatrina DimitroffJudy C BougheyLaura J EssermanChristina YauJulia TchouAstrid QuirarteMarie Catherine LeeMarissa M Howard-McNattKayla SwitallaHenry KuererCandice SauderLauren M PostlewaitCletus ArcieroRoshni RaoAnne WallaceChantal ReynaKamran AhmedLily GutnikNeil TaunkJane PerlmutterAngela DeMicheleDouglas YeeNola M HyltonW Fraser SymmansHope S RugoRebecca A ShatskyClaudine IsaacsSonali RudraPaula PohlmannCheryl EwingJasmine WongMichael AlvaradoNora JaskowiakNicolas PrionasMehra GolshanMara A PiltinOlufunmilayo I OlopadeRita A Mukhtar
Find these authors on Roon
CA
Dr. Cletus Arciero
Breast Surgical Onc · Emory University
CR
Dr. Chantal Reyna
Breast Surgical Onc · Foster G. McGaw Hospital
KA
Dr. Kamran Ahmed
Breast Rad Onc · Moffitt Cancer Center
LG
Dr. Lily Gutnik
Breast Surgical Onc · UAB
RS
Dr. Rebecca Shatsky
Breast Oncology · UCSD
Methods
I-SPY 2 is a prospective, adaptive trial for patients with molecularly high-risk clinical stage II-III breast cancer who are randomized to novel neoadjuvant chemotherapy (NAC) agents. The trial’s primary endpoint was rate of pathologic complete response (pCR). We retrospectively analyzed data from I-SPY 2 patients who completed NAC and surgical treatment during 2010–2021 across 19 participating centers. Type of axillary surgery was not mandated by the trial and was categorized as sentinel lymph node surgery (SLN)-only or axillary dissection (ALND) (± SLN), based on review of operative reports.Self-identified race was collected and categorized as White, Black, or Asian/other, and self-identified ethnicity was categorized as Hispanic/Latino or non-Hispanic/Latino. Patient age and menopausal status at diagnosis was obtained from case report forms. Region of treatment was grouped as Midwest, Northeast, South, and West, based on the Centers for Disease Control and Prevention’s geographic region guidelines for the United States. Clinicopathologic characteristics collected include clinical stage (defined by AJCC guidelines and grouped as I/II and III), tumor histology (ductal or lobular), and tumor grade. Clinical and pathological nodal status and T category were collected as cN, ypN, cT, and ypT, respectively. Nodal status was grouped as positive or negative with cN1, cN2, and cN3 categories considered cN+ and cN0 category considered cN-; cN+ status required pre-treatment needle biopsy demonstrating nodal disease. Tumor receptor subtypes were classified by hormone receptor (HR) and human epidermal growth factor receptor 2 (HER) status, and grouped as HR+HER2−, triple-negative (HR−/HER2−), or HER2+. Cases missing data on type of axillary surgery were excluded.The objective of this study was to compare ALND rates by self-identified race/ethnicity, stratified by clinical and pathological nodal status. We used chi-squared tests to evaluate relationships between race/ethnicity and factors, such as cN and ypN status, tumor receptor subtype, and region of treatment. To adjust for confounders, we used a multivariable logistic regression model including race, age, region of treatment, tumor receptor subtype, cN (+/−), cT (grouped T1-T2 and T3-T4), ypN (+/−), and ypT (grouped as T0/Tis, T1-T2, and T3-T4) to identify odds ratios (ORs) of undergoing ALND with 95% confidence intervals (CIs). Statistical analyses were performed using R version 4.3.1. Two-tailed p values < 0.05 were considered statistically significant.
Results
We identified 1394 patients in our final cohort. The average age was 48.3 ± 11.2 years, and 36.3% were postmenopausal. The most common region of treatment was the West (48.3%) (Table 1). Self-identified race was White in 1107 (79.4%), Black in 156 (11.2%), and Asian/other in 131 (9.4%), and 11.3% of all patients identified as Hispanic/Latino. Notably, we found a significant association between self-identified race and region of treatment. We found that Asian/other patients were significantly more often treated in the West, whereas Black patients were significantly more often treated in the South (78.6% of Asian/other patients treated in the West versus 49.2% White and 16% Black; 46.2% of Black patients treated in the South vs. 22.6% White and 8.4% Asian/other, p < 0.001). Table 1Study cohort characteristics shown overall and by self-identified raceVariableTotal (N= 1394)White (n= 1107)Black (n= 156)Asian/other (n= 131)Age (mean ± SD)48.3 ± 11.248.5 ± 11.148.6 ± 11.846.1 ± 10.9Hispanic ethnicity157/1394 (11.3%)149/1107 (13.5%)3/156 (1.9%)5/131 (3.8%)Postmenopausal status418/1150 (36.3%)338/916 (36.9%)37/110 (33.6%)43/124 (34.7%)RegionWest673/1394 (48.3%)545/1107 (49.2%)25/ 56 (16%)103/131 (78.6%)South333/1394 (23.9%)250/1107 (22.6%)72/ 56 (46.2%)11/131 (8.4%)Midwest310/1394 (22.2%)263/1107 (23.8%)32/156 (20.5%)15/131 (11.5%)Northeast78/1394 (5.6%)49/1107 (4.4%)27/156 (17.3%)2/131 (1.5%)Pathologic Complete Response502/1394 (36%)409/1107 (36.9%)51/156 (32.7%)42/131 (32.1%)Clinical stageI/II1023/1394 (73.4%)807/1107 (72.9%)111/156 (71.2%)105/131 (80.2%)III371/1394 (26.6%)300/1107 (27.1%)45/156 (28.8%)26/131 (19.8%)Tumor histologyDuctal1205/1329 (90.7%)950/1054 (90.1%)136/148 (91.9%)119/127 (93.7%)Lobular124/1329 (9.3%)104/1054 (9.9%)12/148 (8.1%)8/127 (6.3%)Tumor grade114/966 (1.4%)11/793 (1.4%)1/66 (1.5%)2/107 (1.9%)2282/966 (29.2%)240/793 (30.3%)13/66 (19.7%)29/107 (27.1%)3670/966 (69.4%)542/793 (68.3%)52/66 (78.8%)76/107 (71%)cT category136/1394 (2.6%)27/1107 (2.4%)3/156 (1.9%)6/131 (4.6%)2938/1394 (67.3%)747/1107 (67.5%)102/156 (65.4%)89/131 (67.9%)3368/1394 (26.4%)291/1107 (26.3%)43/156 (27.6%)34/131 (26%)452/1394 (3.7%)42/1107 (3.8%)8/156 (5.1%)2/131 (1.5%)cN positive732/1394 (52.5%)580/1107 (52.4%)91/156 (58.3%)61/131 (46.6%)cN category0662/1394 (47.5%)527/1107 (47.6%)65/156 (41.7%)70/131 (53.4%)1589/1394 (42.3%)465/1107 (42%)75/156 (48.1%)49/131 (37.4%)262/1394 (4.4%)47/1107 (4.2%)8/156 (5.1%)7/131 (5.3%)381/1394 (5.8%)68/1107 (6.1%)8/156 (5.1%)5/131 (3.8%)ypT categoryT0/Tis546/1391 (39.3%)447/1104 (40.5%)54/156 (34.6%)45/131 (34.4%)T1435/1391 (31.3%)333/1104 (30.2%)57/156 (36.5%)45/131 (34.4%)T2266/1391 (19.1%)206/1104 (18.7%)28/156 (17.9%)32/131 (24.4%)T3125/1391 (9%)105/1104 (9.5%)12/156 (7.7%)8/131 (6.1%)T419/1391 (1.4%)13/1104 (1.2%)5/156 (3.2%)1/131 (0.8%)ypN positive462/1394 (33.1%)364/1107 (32.9%)51/156 (32.7%)47/131 (35.9%)Receptor subtypeHR+/HER2−595/1394 (42.7%)490/1107 (44.3%)60/156 (38.5%)45/131 (34.4%)HR−/HER2−471/1394 (33.8%)352/1107 (31.8%)70/156 (44.9%)49/131 (37.4%)HER2+328/1394 (23.5%)265/1107 (23.9%)26/156 (16.7%)37/131 (28.2%)Data are numbers with percentages in parenthesesSD standard deviation; c clinical; yp pathologic; Tis in situ; HR hormone receptor; HER2 human epidermal growth factor receptor 2; +, receptor status positive; −, receptor status negative
Study cohort characteristics shown overall and by self-identified race
Data are numbers with percentages in parentheses
SD standard deviation; c clinical; yp pathologic; Tis in situ; HR hormone receptor; HER2 human epidermal growth factor receptor 2; +, receptor status positive; −, receptor status negative
Patients were most commonly clinically stage I/II (73.4%) with grade 3 tumors (69.4%). Most tumors were ductal (90.7%), and the distribution of tumor receptor subtype was 42.7% HR+/HER2−, 33.8% HR−/HER2−, and 23.5% HER2+. For surgical treatment of the primary tumor, most patients underwent mastectomy (55.9% compared with 44.1% lumpectomy; Table 2). Table 2Type of breast surgery, axillary surgery, and nodes removed shown overall and by self-identified raceVariableTotal (N= 1394)White (n= 1107)Black (n= 156)Asian/other (n= 131)Type of breast surgeryLumpectomy615/1394 (44.1%)476/1107 (43%)69/156 (44.2%)70/131 (53.4%)Mastectomy779/1394 (55.9%)631/1107 (57%)87/156 (55.8%)61/131 (46.6%)Type of axillary surgerySLN-only849/1394 (60.9%)692/1107 (62.5%)77/156 (49.4%)80/131 (61.1%)ALND545/1394 (39.1%)415/1107 (37.5%)79/156 (50.6%)51/131 (38.9%)Total nodes removed at surgery8.5 ± 8.48.5 ± 8.59.2 ± 8.18.5 ± 8.4Total positive nodes removed at surgery1.3 ± 3.31.3 ± 3.31.3 ± 2.61.5 ± 3.7Data are numbers with percentages in parenthesesSLN sentinel lymph node surgery; ALND axillary lymph node dissection with or without SLN
Type of breast surgery, axillary surgery, and nodes removed shown overall and by self-identified race
Data are numbers with percentages in parentheses
SLN sentinel lymph node surgery; ALND axillary lymph node dissection with or without SLN
Overall, 52.5% of the study population was cN+ and 33.1% was ypN+. There were no differences in cN or ypN category by race or by Hispanic/Latino ethnicity. Among cN+ patients, the distribution of cN1 compared with cN2/3 disease did not differ by race, with the proportion of cN2/3 cases being 10.4%, 10.3%, and 9.2% in White, Black, and Asian/other patients respectively (p= 0.4). Most cN− patients remained ypN−, with 15% being found to have ypN+ disease at surgery. Conversely, among cN+ patients, 49.6% remained ypN+, whereas 50.4% had nodal response and were ypN− at surgery. The relationship between cN status and ypN status did not differ by race (Table 3). Particularly, the rate of conversion from cN+ to ypN− was 26.6%, 32.1%, and 19.1% for White, Black, and Asian/other, respectively (p= 0.3). Table 3Changes in clinical to pathologic node status by self-identified raceVariableWhite (n= 1107)Black (n= 156)Asian/other (n= 131)pcN+ to ypN−294/1107 (26.6%)50/156 (32.1%)25/131 (19.1%)0.3cN+ to ypN+286/1107 (25.8%)41/156 (26.3%)36/131 (27.5%)cN− to ypN−449/1107 (40.6%)55/156 (35.3%)59/131 (45%)cN− to ypN+78/1107 (7%)10/156 (6.4%)11/131 (8.4%)Data are numbers with percentages in parentheses unless otherwise indicatedcN clinical node status; ypN pathologic node status; +, positive; −, negativeN= 1394
Changes in clinical to pathologic node status by self-identified race
Data are numbers with percentages in parentheses unless otherwise indicated
In the entire study cohort, 39.1% of patients underwent ALND and the remaining 60.9% underwent SLN-only). The rate of ALND did not significantly differ by ethnic identity (37.6% ALND in those who identified as Hispanic/Latino vs. 39.3%, p= 0.7). However, when evaluating ALND rates by self-identified race, Black patients had significantly higher rates of ALND compared to those identifying as White or Asian/Other (50.6%, 37.5%, and 38.9%, respectively, p= 0.007).
When stratified by nodal status, Black patients remained more likely to undergo ALND, specifically among the cN+ and ypN− subgroups (Table 4). Among cN+ patients, ALND was performed in 59.1%, 74.7%, and 60.7% of White, Black, and Asian/other patients respectively (p= 0.018). Among ypN− patients, ALND was performed in 19.2%, 33.3%, and 20.2% of White, Black, and Asian/other patients respectively (p= 0.004). For cN− and ypN+ subgroups, there was no difference in the rate of ALND by race. Despite no difference in nodal response rates by race, among those that converted from cN+ to ypN−, Black patients still had significantly higher rates of ALND than Asian/other or White patients (62% vs. 40% and 41.2%, respectively, p= 0.021). Table 4Associations between axillary surgery and self-identified race overall and by nodal statusVariableWhite (n= 1107)Black (n= 156)Asian/other (n= 131)pOverall10.007SLN-only692/1107 (62.5%)77/156 (49.4%)80/131 (61.1%)ALND415/1107 (37.5%)79/156 (50.6%)51/131 (38.9%)cN−20.3SLN-only455/527 (86.3%)54/65 (83.1%)56/70 (80%)ALND72/527 (13.7%)11/65 (16.9%)14/70 (20%)cN+30.018SLN-only237/580 (40.9%)23/91 (25.3%)24/61 (39.3%)ALND343/580 (59.1%)68/91 (74.7%)37/61 (60.7%)ypN−40.004SLN-only600/743 (80.8%)70/105 (66.7%)67/84 (79.8%)ALND143/743 (19.2%)35/105 (33.3%)17/84 (20.2%)ypN+50.2SLN-only92/364 (25.3%)7/51 (13.7%)13/47 (27.7%)ALND272/364 (74.7%)44/51 (86.3%)34/47 (72.3%)cN+ to ypN-6SLN-only173/294 (58.8%)19/50 (38%)15/25(60%)0.021ALND121/294 (41.2%)31/50 (62%)10/25 (40%)Data are numbers with percentages in parentheses unless otherwise indicatedSLN sentinel lymph node surgery; ALND axillary lymph node dissection with or without SLN; cN clinical node status; ypN pathologic node status; +, positive; −, negative
Associations between axillary surgery and self-identified race overall and by nodal status
Data are numbers with percentages in parentheses unless otherwise indicated
SLN sentinel lymph node surgery; ALND axillary lymph node dissection with or without SLN; cN clinical node status; ypN pathologic node status; +, positive; −, negative
Type of axillary surgery also differed significantly by region, with the higher rates of ALND seen in the South (49.8% ALND in the South vs. 37.4%, 30.3%, and 42.3% in the West, Midwest, and Northeast, respectively, p < 0.001; Table S1).
In a multivariable logistic regression model, we accounted for race, age, region of treatment, tumor receptor subtype, cN status, ypN status, cT, and ypT (Table 5). We found that patients with cN+ disease had significantly higher odds of receiving ALND compared with those with cN- disease (OR 6.58, 95% confidence interval (CI) 4.86–8.98, p < 0.001). Similarly, ypN+ status was also associated with increased odds of ALND compared to ypN− status (OR 7.7, 95% CI 5.45–11.0, p < 0.001). Patients with ypT3-T4 tumors also had greater odds of undergoing ALND compared with those with ypTis/T0 disease (OR 2.42, 95% CI 1.37–4.32, p= 0.002). Treatment region was also a significant factor, with patients treated in the South having significantly higher odds of undergoing ALND compared with those treated in the West (OR 2.18, 95% CI 1.52–3.14, p < 0.001). When accounting for all of these factors, we found that Black patients remained more likely to undergo ALND compared to White patients (OR 1.7, 95% CI 1.09–2.66, p= 0.02). Table 5Multivariable logistic regression model for odds of sentinel lymph node surgery adjusting for race, age at screening, region, tumor subtype, clinical and pathological nodal status, clinical T, and pathological T stageVariablesOdds ratio95% Confidence intervalp*RaceWhite(reference)Black1.71.09–2.660.02Asian/other1.260.77–2.070.4Age1.000.99–1.010.9RegionWest(reference)Midwest0.770.53–1.120.2Northeast1.010.53–1.92> 0.9South2.181.52–3.14< 0.001Tumor subtypeHR+/HER2−(reference)HR−/HER2−1.180.84–1.670.3HER2+1.360.94–1.970.1cN statusNegative(reference)Positive6.584.86–8.98< 0.001ypN statusNegative(reference)Positive7.75.45–11< 0.001cT at diagnosisT1-T2(reference)T3-T41.35.45–11< 0.001ypTT0/Tis(reference)T1-T21.030.73–1.450.9T3-T42.421.37–4.320.002HR hormone receptor; HER2 human epidermal growth factor receptor 2; +, receptor status positive; −, receptor status negative; c clinical; yp pathologic; Tis* in situ
Multivariable logistic regression model for odds of sentinel lymph node surgery adjusting for race, age at screening, region, tumor subtype, clinical and pathological nodal status, clinical T, and pathological T stage
HR hormone receptor; HER2 human epidermal growth factor receptor 2; +, receptor status positive; −, receptor status negative; c clinical; yp pathologic; Tis in situ
Discussion
In this analysis of the multicenter NAC trial I-SPY 2, we identified persistent race-based disparities in the surgical management of the axilla. Black-identifying patients had significantly higher rates of ALND compared with White and Asian/other patients. This disparity remained amongst those who achieved nodal downstaging from cN+ to ypN− following NAC, a subgroup in whom de-escalation of axillary surgery would typically be considered appropriate.3,16,17
Clinical trials are designed to standardize treatment and reduce variability in the delivery of care. Some investigators have found that Black women achieve lower rates of pCR compared with their counterparts.18–20 Others find no differences in pCR by race.21,22 A previous investigation of the I-SPY 2 trial found that there were no significant differences in rates of pCR by race.14 Also well-established in the literature, Black women have been shown to undergo increased rates of ALND compared to their counterparts.11,12 Relation et al. found that this disparity persists in the National Cancer Database, even in the context of NAC, where nodal downstaging would allow for more conservative axillary management.15 Our findings build upon this and show that even in the context of a controlled environment such as a clinical trial where no differences in pCR have been previously established, there exist differences in axillary management by race. This suggests that the lack of mandated surgery intervention in clinical trials can allow factors beyond clinicopathological characteristics to influence surgical decision-making.
Several factors could contribute to the observed disparity. Although our multivariable analysis adjusted for region, tumor subtype, clinical and pathologic nodal status, and tumor size, unmeasured institutional or provider-level differences may still play a role. Variations in regional practices may also contribute to the observed disparities, given higher rates of ALND in the South. However, even when adjusting for this, Black patients were more likely to undergo ALND, suggesting that regional variation in practice patterns alone do not account for the differential use of ALND by race. This suggests geographic differences in adherence to de-escalation practices may influence surgical decision-making independent of patient biology. Implicit bias in clinical decision-making is another possible explanation, potentially influencing whether providers feel comfortable omitting ALND. Black women have previously been shown to have more triple-negative tumors, advanced nodal involvement, higher rates of recurrence, and experience higher mortality, which could influence patient and provider treatment choices.23–28 Additionally, factors such as patient preferences, socioeconomic status, and comorbidities may all influence surgical decision-making. Reeder-Hayes et al. found that in early-stage breast cancer, educational attainment was not associated with type of axillary surgery, but patients enrolled in Medicaid had significantly lower odds of receiving SLN surgery (OR 0.61, 95% CI 0.47–0.78).29 Nayyar et al. found higher ALND use amongst those who were uninsured, and those who resided in areas with low income and low education levels compared to those with private health insurance and those with higher average income and education levels.12 Differences in insurance access, education status, and health literacy can contribute to less engagement during decision-making, leading to misunderstandings about the choice between ALND and SLN.30
The results of this study should be viewed in the context of its limitations. While we adjusted for numerous clinicopathological variables, our study did not capture factors that influence patient and surgeon choices. Although I-SPY 2 is a geographically diverse, multicenter trial, some of our racial subgroups were relatively small, which limited the statistical power to detect differences in these populations (Asian/other). Additionally, data on radiotherapy usage and BMI were unavailable. Access to postoperative radiotherapy could potentially influence ALND use, particularly for patients facing barriers to care. Body habitus may influence sentinel lymph node mapping rates, with prior studies suggesting mixed success rates in those with higher BMI.31–33 Post-NAC axillary staging information such as repeat ultrasound if performed or physical examination findings were not available for this analysis, but could also have contributed to surgical decision making. Finally, surgical management of the axilla is a rapidly evolving landscape, and our observed differences could be confounded by potential shifts in practice patterns over time.17
Conclusions
Our findings revealed persistent race-based disparities in axillary management in I-SPY 2, highlighting the need for clearer, standardized guidelines to guide axillary management decisions after NAC. Future studies should consider the collection of more granular demographic data to better understand and address disparities in breast surgical oncology. Initiatives aimed at improving provider education, reducing bias, and ensuring equitable treatment plans are essential to advancing de-escalation strategies in breast cancer care.
Supplementary Information
Below is the link to the electronic supplementary material.Supplementary file1 (DOCX 18 KB)
Supplementary file1 (DOCX 18 KB)
CR
Dr. Chantal Reyna
Surgery · Breast Surgical OncologyLoyola University Medical Centerchantal