Novel Postneoadjuvant Prognostic Breast Cancer Staging System

David J. Winchester, MD' (D; Lavisha Singh, { \mathsf { M P H } } ^ { 2 } : Stephen B. Edge, M D ^ { 3 } \oplus ; Kimberly H. Allison, { \mathsf { M D } } ^ { 4 } William E. Barlow, \mathsf { P h D } ^ { 5 } ( \mathbb { D } ) ... Veerle Bossuyt, { M D ^ { 6 } } Mariana Chavez-MacGregor, \mathsf { M D } ^ { \intercal } \boldsymbol { \mathbb { Q } } ; Emily F. Conant, M D ^ { 8 } \oplus ; James L. Connolly, { \mathsf { M D } } ^ { 9 } ; Jennifer F. De Los Santos, { M D ^ { 1 0 } } Daniel F. Hayes, M D ^ { \prime \prime } \oplus ; Nola M. Hylton, M D ^ { 1 2 } \oplus ; Elizabeth A. Mittendorf, MD, PhD, M H C M ^ { 1 3 } { 1 0 } ; Jennifer K. Plichta, M D ^ { 1 4 } \oplus .. Elena Provenzano, M D ^ { 1 5 } ; Kilian E. Salerno, M D ^ { 1 6 } : Priyanka Sharma, M D ^ { 1 7 } { \circ } ; W. Fraser Symmans, M D ^ { 7 } \oplus ; Donald Weaver, M D ^ { 1 8 } ; and Gabriel N. Hortobagyi, \mathsf { M D } ^ { 7 } \mathbb { Q } ; on behalf of Member of the Ninth Edition AJCC Neoadjuvant Breast Expert Panel

DOI https://doi.org/10.1200/JC0-24-01739

ABSTRACT

ACCOMPANYING CONTENT

PURPose Prognostic staging after neoadjuvant chemotherapy (NACT) is not included in American Joint Commission on Cancer (AJCC) staging. This study addressed this deficiency by including responses to therapy with standardized staging variables in a validated prognostic staging system for patients treated with NACT.

METHops The National Cancer Database was queried to identify 140,605 patients treated with NACT between 2010 and 2018. Three response categories (no response, partial response, and complete response [pCR]) were created on the basis of comparison of clinical and post-NAcT pathologic staging. Univariate and multivariate analyses of clinical stage, estrogen receptor, progesterone receptor, human epidermal growth factor receptor 2 (HER2), and grade were analyzed for each category. Predictive models for each response category were validated using the bootstrap technique. Calibration plots compared predicted and observed 3-year survival probabilities in the training and validation data sets.

REsULts Each validated model demonstrated statistically significant survival differences in the postneoadjuvant prognostic stage assignment. Of all patients with a pCR, 9 4 . 2 % were assigned to postneoadjuvant ypStage I compared with 3 5 . 5 % of patients with no response. Advancing clinical stage had a progressive but small impact on overall survival (OS) with pCR (high-grade, triple-negative breast cancer [TNBC]: cStage I, 9 7 % v cStage IIIB/IIIC, 9 1 % ; grade 2 luminal A: 9 7 % \nu 9 1 % ) but was associated with a profound decrease in OS with no response for TNBC or \scriptstyle { { H E R } } 2 + disease (high-grade TNBC 8 9 % \nu 5 0 % and less profound for grade 2 luminal A disease with no response ( 9 7 % \nu 8 1 %

Appendix Data Supplement

Accepted February 18, 2025
Published April 11, 2025

J Clin Oncol 00:1-13 ⊚ 2025 by American Society of Clinical Oncology

View Online Article

CONCLUSION

ISION We present a novel, validated prognostic staging system that predicts OS according to the response to NACT. These data will provide AJCC stage assignments for a growing proportion of patients treated with NACT.

INTRODUCTION

The Eighth Edition of the American Joint Commission on Cancer (AJCC) introduced a novel prognostic staging system for breast cancer that incorporated estrogen receptor (ER), progesterone receptor (PR), human epidermal growth factor receptor 2 (HER2), and histologic grade in conjunction with the anatomic extent of the disease.' This allowed more accurate staging of patients with breast cancer. Traditionally, stage assignments have only included information on the anatomic extent of the primary tumor (T), regional lymph node involvement (N), and the presence of metastases (M) combined to provide a TNM stage group. Clinical stage was assigned on the basis of clinical information only, and pathologic stage was defined by clinical criteria supplemented by the pathologic extent of cancer identified by upfront surgical resection with the exclusion of patients treated with neoadjuvant chemotherapy (NAcT). Until the eighth edition, pathologic staging was applicable for all patients, regardless of the use of chemotherapy in the adjuvant or neoadjuvant setting.2-8

In recent years, NAcT has become the initial treatment of choice for many patients with operable breast cancer, especially those with HER2-amplified tumors and triplenegative disease, as well as an increasing number of

CONTEXT

Key Objective

Postneoadjuvant pathologic prognostic staging has not been a component of American Joint Commission on Cancer breast cancer staging because of insufficient data assessing response to therapy.

Knowledge Generated

Three different validated models derived from the National Cancer Database, created for each response category (no response, partial response, and complete response) after neoadjuvant chemotherapy, predict survival and postneoadjuvant prognostic stage assignment according to clinical stage, grade, estrogen receptor, progesterone receptor, and human epidermal growth factor receptor 2 status. Response to therapy and clinical stage at presentation are important variables to predict outcome.

Relevance (G.F. Fleming)

This standardized postneoadjuvant staging should help clinicians prognosticate for individual patients and willinform future neoadjuvant or postneoadjuvant breast cancer studies.\*

\*Relevance section written by JCO Associate Editor Gini F. Fleming, MD.

patients with luminal B-type breast cancer. One of the benefits of neoadjuvant therapy is that the response provides significant prognostic information. This variable has not been previously incorporated into staging.

At the time of the creation of the AJCC eighth edition, sufficient data were not available to create a staging system for women treated with NAcT; therefore, postneoadjuvant therapy pathologic staging could not be assigned. This analysis provides data to address this deficiency in the current AJCC staging system with a postneoadjuvant prognostic staging system.

METHODS

Data Source

The 2o21 Participant User File (PUF) of the National Cancer Database (NCDB) was used. The NCDB is a facility-based nationwide data set containing information on nearly 7 0 % of newly diagnosed breast cancer cases in the United States, operated by the American College of Surgeons Commission on Cancer in collaboration with the American Cancer Society.9

Study Population

This study included female patients with breast cancer, age 18-89 years, diagnosed with clinical stage I, II, or III invasive breast cancer diagnosed from 2010 to 2018. Those who received NACT, had baseline (or pre-NACT) clinical and postneoadjuvant pathologic TNM variables, and vital status were included in the analysis. Pre-NACT ER, PR, HER2, and grade were analyzed according to the treating hospital pathology report, without a central review. Patients were excluded if their last contact or death was ^ { < 6 } months after the start of NACT or if they were diagnosed with stage o or stage IV disease at presentation. Overall survival (OS) was calculated using landmark time to allow for response evaluation, set to 6 months after diagnosis to the time of last follow-up or death, as reported by the facility registry to the NCDB. Patients were censored at last follow-up, allowing the inclusion of all available survival data in the analysis.

Response Category

The pathologic response to therapy was determined by comparing the clinical (cT and cN) categories at diagnosis and pathologic (ypT and ypN) categories after NACT and definitive surgery. Three categories (no response, partial response, and complete response [pCR]) were defined using these comparisons. No response was defined as having the same or higher T and/or N categories after the NACT (Data Supplement, Table S1, online only). Patients with disease progression during NACT were included in the no-response category. Partial response was defined as downstaging (lower category) of one or both T and N categories with no upstaging (higher category) of either but with residual invasive cancer in one or both locations. pCR was defined as any presenting clinical stage (I-IIIC) with no invasive cancer after NACT (ypToypNocMo or ypTisypNocMo). The data were divided into these three separate groups for analysis.

Stage Assignment

Survival was computed using each response model according to the clinicopathologic variables defined above. To maintain consistency of survival ranges with previous AJCC staging editions, postneoadjuvant pathologic prognostic stage assignments were defined by the 3-year OS ranges used in the

FIG 1. Patient selection map [National Cancer Data Base (NCDB) 2021 participant user file (PUF)].

AJCC Eighth Edition Staging Manual (Data Supplement, Table S2; stage 1 A > 9 4 . 0 % , stage IB 92.0 to < 9 4 . 0 % , stage IIA 88.0 to < 9 2 . 0 % , stage IIB 85.0 to { < } 8 8 . 0 % , IIIA 80.0 to { < } 8 5 . 0 % , IIIB 7 0 % to < 8 0 % , IIIC < 7 0 % ).' See the Data Supplement for detailed statistical methods.

RESULTS

A total of 3,956,621 women diagnosed with breast cancer between 2010 and 2018 were identified in the 2021 NCDB PUF. Complete data were identified for 140,605 patients who underwent NACT (Fig 1). The median follow-up period for patients who did not die was 73.7 months (range, 6.4-158.2). Among the 140,605 women, 34,572 had a pCR . 9 6 . 0 % survival), 59,764 had a partial response ( 8 7 . 6 % survival), and 46,269 had no response 8 5 . 5 % survival; log-rank P < . 0 0 0 1 ^ { * } o

Patients With No Response

Advanced clinical stage, higher grade, ER-negative status, PR-negative status, and HER2-negative status were all predictive of increased hazard of death in the univariable analysis of the training data set ( P < . 0 0 0 1 ; Data Supplement, Table S3). There was no difference between the training ( \mathbf { n } = 3 7 , 0 1 6 ) and testing \left( \mathbf { n } = \mathbf { q } , 2 5 3 \right) data sets in the frequency of variable subsets or survival (log-rank { \cal P } ~ = ~ . 4 9 ^ { \prime } . Multivariable Cox regression analysis showed that advanced clinical stage, higher grade, ER-negative, PR-negative, and HER2-negative remained predictive of worsened Os. Within this response category, patients with clinical stage IIIB/IIIC breast cancer had an OS hazard ratio (HR) of 5.89 relative to those with clinical stage I (Data Supplement, Table S4). The Data Supplement (Fig S1) displays the model calibration curves and receiver operating characteristic (Roc) association statistics for the training and testing sets (AUc of 0.79 and o.79, respectively). Table 1 lists the predicted survival and assigned postneoadjuvant prognostic stage according to the clinical stage, receptor and grade category, and response. Not all the clinical stage/receptor/grade combinations are included in this table. Statistically significant differences were noted between the postneoadjuvant prognostic stages IA, IB, IIA, IIB, IIIA, IIIB, and IIIC for patients with no response (log-rank P < . 0 0 0 1 ; Fig 2).

Patients With a Partial Pathologic Response

Advanced clinical stage, higher grade, ER-negative, PRnegative, and HER2-negative status were all predictive of increased hazard of death in the univariable analysis of the training data set ( P < . 0 0 0 1 ; Data Supplement, Table S5). There was no difference between the training ( \mathbf { n } = 4 7 , 8 1 2 ) and testing [ \mathbf { n } = 1 1 , 9 5 2 ] data sets in terms of the frequency of variable subsets or survival (log-rank P = . 6 5 { * } . Multivariable Cox regression analysis showed that advanced clinical stage (except for stage IIA [HR, 0.67]), higher grade, ER-negative, PR-negative, and HER2-negative status remained predictive of worsened OS. The Data Supplement (Fig S2) displays the model calibration curves and ROC association statistics for the training and testing sets (AUC of o.75 and 0 . 7 4 ,respectively). Table 1 lists the predicted survival and postneoadjuvant prognostic stage according to the clinical stage, receptor and grade category, and response. Not all the clinical stage/receptor/grade combinations are included in this table. Statistically significant differences were noted between the postneoadjuvant prognostic stages IA, IB, IIA, IIB, IIIA, and IIIB (log-rank P < .0001; Fig 3).

Patients With a Complete Pathologic Response

Advanced clinical stage, lower grade, ER-negative, PRnegative, and HER2-negative status were all predictive of an increased hazard of death in the univariable analysis of the training data set ( P < . 0 0 0 1 ; Data Supplement, Table S7). There was no difference between the training ( \boldsymbol { { n } } = 2 7 , 6 5 8 ) and testing { \bf \tilde { n } } = 6 , 9 1 4 , data sets in terms of the frequency of variable subsets or survival (log-rank P = . 0 7 ). Multivariable Cox regression analysis showed that advanced clinical stage, lower grade, ER-negative, and HER2-negative were predictive of decreased OS (Data Supplement, Table S8). Unlike the other two response categories, PR was not significant (HR, 1.03 for PR-negative), and grade 3 predicted improved survival (HR, 0.55) relative to grade 1 and grade 2 (HR, 0.70).

FIG 2. Survival according to assigned postneoadjuvant prognostic stage for patients with no response.

The loss of statistical significance for PR in the multivariable model suggests that its prognostic impact is not independent but mediated through related factors. ER was most likely to account for this (Cramer's V = 0 . 6 0 , strong association), followed by HER2 (Cramer's V = 0 . 2 1 , weak association), PR (Cramer's V = 0 . 1 9 , weak association), and TNM stage (Cramer's V = 0 . 0 6 ,very weak association). Clinical stages IIIB/IIIC had a hazard ratio of 3.o9 relative to clinical stage I with pCR. The Data Supplement (Fig S3) displays the model calibration curves and ROC association statistics for the training and testing sets (AUC of o.68 and 0.71, respectively). Table 1 lists the number of patients, predicted survival, and assigned postneoadjuvant prognostic stage according to the clinical stage, receptor, grade category, and response. Not all the clinical stage/receptor/ grade combinations are included in this table. Statistically significant differences were noted between the postneoadjuvant prognostic stages IA, IB, IIA, and IIB (log-rank P < .0001, Fig 4).

FIG 3. Survival according to assigned postneoadjuvant prognostic stage for patients with partial response.

FIG 4. Survival according to assigned postneoadjuvant prognostic stage for patients with complete response.

Summary of Findings

Response to therapy is one of the most important variables for predicting OS after NACT. Over 9 4 % of patients achieving pCR in this study were assigned to postneoadjuvant prognostic stage ypI, regardless of breast cancer subtype. As reflected by the hazard ratios in each of the three response group models, clinical stage has an important impact on OS, showing a progressive decrease with advancing clinical stage (Table 1). The increased hazard of death is greatest for patients with no response and most pronounced for those with high-grade triple-negative breast cancer (TNBC) or highgrade \scriptstyle { { H E R } } 2 + disease. The reduction in mortality is less pronounced in patients with intermediate-grade luminal A-like disease, likely reflective of adjuvant endocrine therapy. In patients with a pCR, a progressive increase in the hazard of death was noted with advancing clinical stage for each breast cancer subtype, but with less than a 6 % difference between clinical stage I and clinical stage IIIB/IIIC for each of the examples listed above, in contrast to a 3 9 % difference for high-grade TNBC with no response (Table 1).

Patients with a pCR had two unique findings. Grade had a paradoxical effect on patients with pCR, with low-grade tumors having a greater risk of death. Second, PR was not predictive of an increased risk of death in patients with a pCR.

DISCUSSION

AJCC breast staging systems have historically been defined solely on the basis of anatomic criteria (TNM) to predict survival at diagnosis (clinical stage) and with more refined metrics after surgery (pathologic stage).? In the eighth edition, additional prognostic factors (grade, ER, PR, and HER2 status) were used to supplement TNM to define stage groups. This has improved the precision of outcome prediction and the relevance of staging for clinicians and patients. With the eighth edition, the postsurgical staging of patients undergoing NACT was specifically excluded. A top priority for the next version of the AJCC breast cancer staging system is to comprehensively address this deficiency.

Subsequently, many studies have validated the eighth edition clinical and pathologic prognostic staging. In addition, some studies have also applied the pathologic prognostic stage to patients treated with NACT with relative success, despite the lack of response analysis.10-21

The current study provides statistically valid models for staging patients treated with NACT on the basis of anatomic and biological factors within response categories. The data elements necessary to assign the response category and postneoadjuvant prognostic stage are currently collected in cancer registries in North America and are in place within the NCDB. Thus, this postneoadjuvant staging system can be implemented without any additional variables.

Breast cancer is a heterogeneous disease that is treated according to different criteria, including stage, primary molecular/clinical subtype (luminal A, luminal B, \scriptstyle { { H E R } } 2 + 7 and basal), and identifiable mutations or receptor targets.

In the analysis that led to the changes introduced in the eighth edition, as well as this study to expand prognostic staging to postneoadjuvant therapy staging, an attempt was made to recognize specific subtypes for analysis. This strategy was also explored by evaluating eight different models (data not shown), but ultimately relied upon one construct consisting of T, N, M, grade, ER, PR, HER2, and response to provide a system that prioritized efficiency and simplicity for clinicians and patients.

The most striking finding in the models presented here for those treated with neoadjuvant therapy was the commonality of an excellent prognosis for those with a complete pathologic response, spanning across a very wide range of clinical stages, and each breast cancer subtype. These data confirm that pCR is a very important prognosticator, regardless of stage, histology, grade, and subtype, as reported in many previous studies.22-27 In addition, these models showed that clinical staging remains an important prognostic factor in all three response groups.

In a meta-analysis of 27,895 patients undergoing NACT between 1999 and 2016, pCR, event-free status, and OS were uniformly improved regardless of the receptor profile.28 Survival advantages with pCR were more pronounced in { H E R } 2 + and TNBC than in hormone receptor-positive breast cancer. Patients with TNBC and pCR had an 8 4 % 5-year OS compared with 4 7 % for those without a pCR; patients with \scriptstyle { { H E R } } 2 + breast cancer with a pCR had a 9 5 % 5-year OS compared with 7 6 % for those without a pCR; patients with hormone receptor-positive breast cancer with a pCR had a 9 8 % 5-year OS compared with 8 2 % without a pCR. Although significant, the survival of patients with TNBC and pCR was not as favorable in the metaanalysis as in the current study ( 9 4 % ) , likely reflecting improved therapy in a more recent period of analysis and 3 - year follow-up versus 5-year follow-up. It should be emphasized that a pCR has a much greater OS impact for specific subtypes of breast cancer, particularly in patients diagnosed with TNBC.

The finding of an increase in breast cancer mortality with low-grade tumors and pCR was unexpected. One explanation is the expected low response rates to NACT.29,30 Mortality from breast cancer after a pCR implies that unrecognized residual disease leads to eventual progression. It may be possible that incorrect categorization of a pCR is less common for higher-grade disease.

Future refinements of AJCC staging may include more quantified assessments of treatment response, such as the residual cancer burden (RCB) index.31 RCB has been validated in a multi-institutional pooled analysis that, like our study, confirmed the importance of receptor subtype in predicting the response to NAcT.32 However, there are insufficient data to allow the inclusion of RCB in staging at this time and it is not yet routinely reported in the United States.

This study has several limitations. In this retrospective study, it was not possible to determine why a specific patient was administered NAcT, leading to a selection bias created by institutional, physician, and/or patient preferences. However, the large number of patients included in the analysis should reduce the possibility of major bias. For the same reason, only a limited number of patients with clinical stage I cancer treated with NAcT are in the database, reflecting infrequent NAcT for that patient population. It is not possible to determine from the NCDB specific drug treatments as NAcT or as postsurgery adjuvant therapy or the duration of treatment for individual patients. Given the 9-year study period, it is highly likely that survival has improved as therapy continues to evolve through better adjuvant regimens and advances in treating recurrent cancer. The survival of patients with triplenegative and HER2-positive breast cancer is very likely underestimated in this patient population diagnosed from 2010 to 2018, which straddles the introduction of adjuvant trastuzumab, pertuzumab, emtansine, and pembrolizumab. Although not used during this study period, cDK4/6 inhibitors and neoadjuvant endocrine therapy will likely increase the survival and pCR compliments for patients treated with neoadjuvant therapy.

The absence of recurrence data and breast cancer-specific survival are recognized limitations of NCDB. These data are very important end points, particularly in clinical trials, for understanding the effectiveness of new therapies. The purpose of this study was to address an obvious deficiency in staging of patients undergoing NAcT. In line with all previous AJCC staging editions, OS has been the sole end point used in staging and is arguably the most accurate and measurable. The median follow-up in the NCDB for the patients included in this analysis was relatively short, especially for patients with hormone receptor-positive breast cancer, where late recurrences are more common. In addition to the lack of specific NAcT regimens, including trastuzumab and pertuzumab, NcDB does not record specific adjuvant endocrine therapy, cytotoxic chemotherapy, or HER2-targeted therapy for patients with residual disease after surgery. Key studies demonstrating the value of additional adjuvant therapy were reported in 2017 and 2018 and may have led to improved outcomes in the last 2 years of this study and to overestimate the hazard of death of patients with a partial response and no response.33,34 The same holds true for patients who progress to develop metastatic disease. Treatment for this circumstance has also improved, possibly leading to increased OS but not reported in the NCDB and not included in the models.

Because of patient, biomarker, and treatment heterogeneity, some combinations of stages and biomarkers are underrepresented. Such heterogeneity also reflects the current patterns of practice that routinely use NAcT for patients with HER2-amplified breast cancer and TNBC but more selectively in patients with luminal breast cancer. Improved therapy is likely to increase the

AFFILIATIONS

'City of Hope Chicago, Zion, IL
2Northshore University Healthcare, Evanston, IL
3Roswell Park Comprehensive Cancer Center, Buffalo, NY
4Stanford University School of Medicine, Stanford, CA
5swoG Statistical and Data Management Center, Seattle, WA
Harvard Medical School, Boston, MA
7The University of Texas MD Anderson Cancer Center, Houston, TX
8Hospital of the University of Pennsylvania, Philadelphia, PA
9Beth Israel Deaconess Medical Center, Boston, MA
10Washington University, St Louis, MO
1University of Michigan Rogel Cancer Center, Ann Arbor, MI
12University of California San Francisco, San Francisco, CA
13Dana Farber Cancer Institute, Boston, MA
14Department of Surgery, Duke University Medical Center, Durham, NI
15NIHR Cambridge Biomedical Research Centre, Cambridge,
United Kingdom
16National Cancer Institute, Bethesda, MD
17University of Kansas Medical Center, Kansas City, KS
18University of Vermont, Burlington, VT

CORRESPONDING AUTHOR

David J. Winchester, MD; Twitter: @DJWinchesterMD; e-mail: David.
winchester@coh.org.

percentage of patients who achieve a complete pathologic response.

In conclusion, breast cancer is a heterogeneous disease, and the prognosis of patients treated with NACT varies significantly according to the treatment response and clinical stage. These validated models, created for each of the three pathologic response categories, provided OS predictions after NACT. These data set forth a new postneoadjuvant prognostic staging system that we recommend for inclusion in upcoming versions of AJCC breast cancer staging.

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Disclosures provided by the authors are available with this article at DOI https://doi.org/10.1200/JC0-24-01739.

AUTHOR CONTRIBUTIONS

Conception and design: David J. Winchester, Lavisha Singh
Administrative support: David J. Winchester, Lavisha Singh
Provision of study materials or patients: David J. Winchester
Collection and assembly of data: David J. Winchester, Lavisha Singh
Data analysis and interpretation: David J. Winchester, Lavisha Singh,
Stephen B.Edge, William E. Barlow, Veerle Bossuyt Mariana Chavez
MacGregor, Emily F. Conant, James L. Connolly, Jennifer F. De Los
Santos, Nola M. Hylton, Elizabeth A. Mittendorf, Jennifer K. Plichta,
Elena Provenzano, Kilian E. Salerno, W. Fraser Symmans, Donald
Weaver, Gabriel N. Hortobagyi, Kimberly H. Allison, Daniel F. Hayes,
Priyanka Sharma
Manuscript writing: All authors
Final approval of manuscript: All authors
Accountable for all aspects of the work: All authors

ACKNOWLEDGMENT

The list of Ninth Edition AJCC Neoadjuvant Breast Expert Panel Members can be found in Appendix (online only).

REFERENCES

1. Amin MB, Edge SB, Greene FL, et al (eds): AJCC Cancer Staging Manual (ed 8). New York, NY, Springer, 2017
2. American Jont Ctte for Cancer Staging and End Rults Rporting. Manual for taging of ancer. Chago L, American Joint ommtte, 1977
3. Beahrs OH, Meyers MH (eds): Manual for Staging of Cancer (ed 2). Philadelphia, PA, Lippincott, 1983
4. Beahrs OH, Henson DE, Hutter RVP, et al (eds): Manual for Staging of Cancer (ed 3). Philadelphia, PA, Lippincott, 1988
5. Beahrs OH, Henson DE, Hutter RVP, et al (eds): Manual for Staging of Cancer (ed 4). Philadelphia, PA, Lippincott, 1992
6. Fleming D, Cooper JS, Henson DE, et al (eds): AJCC Cancer Staging Manual (ed 5). Philadelphia, PA, Lipincott-Raven, 1997
7. Greene FL, Page DL, Fleming ID, et al (eds): AJCC Cancer Staging Manual (ed 6). New York, NY, Springer-Verlag, 2002
8. Edge SB, Byrd DR, Compton CC, et al (eds): AJCC Cancer Staging Manual (ed 7). New York, NY, Springer, 2010
9. B lrtic18008
1i analysis. Anticancer Res 37:4615-4621, 2017
1. database.Breast 37:56-63,2018
12. b t b r 875, 2018
13. cancer. JAMA Oncol 4:203-209, 2018
14. Won n Lim J, e al. f the J h prstc t r bst ccer n si ht sti B 484, 2018
15.
16. 737-745, 2018
17. e1347-e1352, 2018
18. Results (SEER) analysis. Cancer Manag Res 11:1433-1442, 2019
19. e .12
20. Dici g B t a l f the J ptic sta f tive bt r n t til. 17:207 2019
21. negative breast cancer. Cancer Manag Res 14:719-728, 2022
2 Cancer 103:1335-1342, 2010
23. 3342-3354, 2012
24. aa i 82204
2. 2014
2. chemotherapy: Long-term follow-up from AC0s0G Z1071 (Alliance). Ann Surg 266:667-676, 2017
2 Cancer 18:e219-e229, 2018
2. Res 26:2838-2848, 2020
29. 2
3
31.0 2017
32 patients. Lancet Oncol 23:149-160, 2022
33. Masuda N, Lee SJ, Ohtan , et al Adjuant cacitaine for breast cacer afer preoperative chemotherapy. N Engl J Med 376:2147-2159 2017
34. von Minkwitz , Huang CS, Man M et al Trastuzumab etansine fr ridal inasive HER2-positive rest cancer. N Engl J Med 38.617-628, 2019

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Novel Postneoadjuvant Prognostic Breast Cancer Staging System

The following repreents disclosure information provided by author of this manuscript ll rlationships are considere compensated unless otherwise noted. Relationships are self-held unless noted. \mid \mid = Immediate Family Member, Inst \ l = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about Asco's conflic of interest policy, please refer to ww.ascoorg/wc or ascopubs.org/jco/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).

Stephen B. Edge Honoraria: North American Center for Continuing Medical Education

William E. Barlow Research Funding: Merck (Inst), AstraZeneca (Inst)

Mariana Chavez-MacGregor
Employment: MD Anderson Physician's Network
Consulting or Advisory Role: Abbott Laboratories, Exact Sciences,
Pfizer, Lilly, AstraZeneca/Daiichi Sankyo, Exact Sciences, Roche/
Genentech, Adium Pharma, Merck, AstraZeneca, Novartis
Research Funding: Novartis (Inst), Genentech/Roche (Inst), Pfizer (Inst),
Lilly
Expert Testimony: Lilly
Travel, Accommodations, Expenses: AstraZeneca, Exact Sciences,
Zodiac Pharma
Uncompensated Relationships: Legacy Healthcare Services, The Hope
Foundation
Emily F. Conant
Leadership: iCAD, Inc, Hologic
Honoraria: Medality
Consulting or Advisory Role: Hologic, iCAD
Speakers' Bureau: Medscape, iiCME, Inc
Research Funding: Hologic (Inst), iCAD, Inc (Inst) Daniel F. Hayes
Stock and Other Ownership Interests: InBiomotion, Cellworks, Xilis Consulting or Advisory Role: Cepheid, Epic Sciences, Cellworks, BioVica, Xilis, Exact Sciences, Centrix, Arvinis, bioTheranostics, Strata Oncology, Stratipath, Artera, Microbiologics, Delphi Diagnostics, Manta Cares, Spotlight Diagnostics, Mllagen
Research Funding: AstraZeneca (Inst), Menarini Silicon Biosystems (Inst), Cepheid/Danaher (Inst), Angle (Inst)
Patents, Royalties, Other Intellectual Property: Royalties from licensed technology, Diagnosis and Treatment of Breast Cancer.Patent No.US 8,790,878 B2. Date of Patent: July 29, 2014. Applicant Proprietor: University of Michigan. D.F.H. is designated as inventor/coinventor, Circulating Tumor Cell Capturing Techniques and Devices. Patent no.: US 8,951,484 B2. Date of Patent: February 10, 2015. Applicant Proprietor: University of Michigan. D.F.H. is designated as inventor/ coinventor, Title: A method for predicting progression free and overall survival at each follow-up timepoint during therapy of metastatic breast cancer patients using circulating tumor cells. Patent no. 05725638.0- 1223-US2005008602
Other Relationship: UpToDate, Cancer Expert Now

Nola M. Hylton Research Funding: Siemens Healthineers (Inst)

Elizabeth A. Mittendorf
Honoraria: Physicians" Education Resource
Consulting or Advisory Role: BioNTech SE, Merck, AstraZeneca,
Moderna Therapeutics, Alight Sciences, AstraZeneca, Merck
Research Funding: Roche/Genentech (Inst), Gilead Sciences (Inst)
Travel, Accommodations, Expenses: Merck Sharpe and Dohme
Uncompensated Relationships: Bristol Myers Squibb, Roche/
Genentech
Open Payments Link: https://openpaymentsdata.cms.gov/physician/
899522/summary

Jennifer K. Plichta Research Funding: SURGE Therapeutics (Inst)

Elena Provenzano
Honoraria: AstraZeneca, Roche
Consulting or Advisory Role: Genomic Health, AstraZeneca
Research Funding: IBEX Medical Analytics (Inst)
Travel, Accommodations, Expenses: IBEX Medical Analytics
Priyanka Sharma
Stock and Other Ownership Interests: Amgen, Johnson & Johnson/
Janssen, Roche/Genentech (l), Gilead Sciences (l), Sanofi, Pfizer (I)
Consulting or Advisory Role: Novartis, Merck, AstraZeneca, Pfizer,
Gilead Sciences, GlaxoSmithKline, Sanofi, Boston Scientific (l), Cipla (l),
Salient Pharmaceuticals (l), Lilly, Stemline Therapeutics
Research Funding: Novartis (Inst), Bristol Myers Squibb (Inst), Merck
(Inst), Novartis (Inst), Gilead Sciences (Inst)
Patents, Royalties, Other Intellectual Property: UpToDate
W. Fraser Symmans
Stock and Other Ownership Interests: ISIS Pharmaceuticals, Delphi
Diagnostics, Eiger BioPharmaceuticals
Consulting or Advisory Role: AstraZeneca, SAGA Diagnostics
Patents, Royalties, Other Intellectual Property: Intellectual Property,
Intellectual Property (expired), Intellectual Property (pending)
Uncompensated Relationships: Delphi Diagnostics
Open Payments Link: https://openpaymentsdata.cms.gov/physician/
256534 Gabriel N. Hortobagyi
Consulting or Advisory Role:Novartis, Seagan, Blueprint Medicines, AstraZeneca
No other potential conflicts of interest were reported.

APPENDIX. NINTH EDITION AJCC NEOADJUVANT BREAST EXPERT PANEL MEMBERS

David J.WinchesterStephen B.Edge, Kimberly H.Allison William E.BarlowVeerle Bossuyt, Mariana Chavez-MacGregor, Emily F. Conant, James L. Connolly, Jennifer F.

De Los Santos, Daniel F. Hayes, Nola M. Hylton, Elizabeth A. Mittendorf, Elena Provenzano, Kilian E.Salerno, Priyanka Sharma, W.Fraser Symmans, Donald Weaver, and Gabriel N. Hortobagyi.