Long-Term Prognostic Risk After Neoadjuvant Chemotherapy Associated With Residual Cancer Burden and Breast Cancer Subtype

Abstract

Purpose To determine the long-term prognosis in each phenotypic subset of breast cancer related to residual cancer burden (RCB) after neoadjuvant chemotherapy alone, or with concurrent human epidermal growth factor receptor 2 (HER2)-targeted treatment. Methods We conducted a pathologic review to measure the continuous RCB index (wherein pathologic complete response has RCB = 0; residual disease is categorized into three predefined classes of RCB index [RCB-I, RCB-II, and RCB-III]), and yp-stage of residual disease. Patients were prospectively observed for survival. Three patient cohorts received paclitaxel (T) followed by fluorouracil, doxorubicin, and cyclophosphamide (T/FAC): original development cohort (T/FAC-1), validation cohort (T/FAC-2), and independent validation cohort (T/FAC-3). Another validation cohort received FAC chemotherapy only, and a fifth cohort received concurrent trastuzumab (H) with sequential paclitaxel and fluorouracil, epirubicin, and cyclophosphamide (FEC; H+T/FEC). Phenotypic subsets were defined by hormone receptor (HR) and HER2 status at diagnosis, classified as HR-positive/HER2-negative, HER2-positive (HR-negative/HER2-positive or HR-positive/HER2-positive), or triple receptor-negative. Relapse-free survival estimates were determined from Kaplan-Meier analysis and compared using the log-rank test. Results Five cohorts (T/FAC-1 [n = 219], T/FAC-2 [n = 262], T/FAC-3 [n = 342], FAC [n = 132], and H+T/FEC [n = 203]) had median event-free follow-up of 13.5, 9.1, 6.8, 16.4, and 7.1 years, respectively. Continuous RCB index was prognostic within each phenotypic subset, independent of other clinical-pathologic variables. RCB classes stratified prognostic risk overall, within each phenotypic subset, and within yp-stage categories. Estimates of 10-year relapse-free survival rates in the four RCB classes (pathologic complete response, RCB-I, RCB-II, and RCB-III) were 86%, 81%, 55%, and 23% for triple receptor-negative; 83%, 97%, 74%, and 52% for HR-positive/HER2-negative in the combined T/FAC cohorts; and 95%, 77%, 47%, and 21% in the H+T/FEC cohort. Conclusion RCB was prognostic for long-term survival after neoadjuvant chemotherapy in all three phenotypic subsets of breast cancer. Our institutional findings should be externally validated.

Fig 1.

Kaplan-Meier plots of relapse-free survival for residual cancer burden (RCB) categories within the yp-stage categories from pooled analysis of the patients with residual disease from the five cohorts (N = 895): (A) yp-stage I (n = 218); (B) yp-stage II (n = 418); and (C) yp-stage III (n = 259). This analysis excluded 12 patients who did not undergo surgery immediately after neoadjuvant treatment (as a result of progression). Note that the original development cohort (cohort 1, Table 1) was included in these analyses. The length of each x-axis is proportional to the duration of follow-up of survivors, truncated when < 10% of the smallest group remained at risk. Response categories are shown as RCB-I (gold), RCB-II (gray), and RCB-III (red).

Fig 2.

Kaplan-Meier plots of relapse-free survival according to residual cancer burden (RCB) categories for phenotypic subsets of breast cancer: (A) triple-negative (n = 219); (B) HR-positive/HER2-negative (n = 501); (C) HER2-positive (n = 103) after paclitaxel (T) followed by fluorouracil, doxorubicin, and cyclophosphamide (T/FAC) chemotherapy; (D) all HER2-positive (n = 203); (E) HR-negative/HER2-positive subset (n = 95); and (F) HR-positive/HER2-positive subset (n = 108) after HER2-targeted therapy with trastuzumab (H) and sequential paclitaxel followed by fluorouracil, epirubicin, and cyclophosphamide chemotherapy (H+T/FEC). Patients in the HER2-positive T/FAC cohort (C) did not receive any HER2-targeted therapy (trastuzumab) as neoadjuvant or adjuvant treatment. On the other hand, all patients in the H+T/FEC cohort (D to F) received neoadjuvant and adjuvant HER2-targeted therapy (trastuzumab). Note that the original development cohort (cohort 1, Table 1) was included in the analysis for plots A to C. The length of each x-axis is proportional to the duration of follow-up of survivors, truncated when < 10% of the smallest group remained at risk. Below each survival plot is a corresponding histogram to show the relative frequency distribution of RCB index. Response categories are shown as pathologic complete response (pCR; blue), RCB-I (gold), RCB-II (gray), and RCB-III (red). HR, hormone receptors; HER2, human epidermal growth factor receptor 2.

Fig 2.

Kaplan-Meier plots of relapse-free survival according to residual cancer burden (RCB) categories for phenotypic subsets of breast cancer: (A) triple-negative (n = 219); (B) HR-positive/HER2-negative (n = 501); (C) HER2-positive (n = 103) after paclitaxel (T) followed by fluorouracil, doxorubicin, and cyclophosphamide (T/FAC) chemotherapy; (D) all HER2-positive (n = 203); (E) HR-negative/HER2-positive subset (n = 95); and (F) HR-positive/HER2-positive subset (n = 108) after HER2-targeted therapy with trastuzumab (H) and sequential paclitaxel followed by fluorouracil, epirubicin, and cyclophosphamide chemotherapy (H+T/FEC). Patients in the HER2-positive T/FAC cohort (C) did not receive any HER2-targeted therapy (trastuzumab) as neoadjuvant or adjuvant treatment. On the other hand, all patients in the H+T/FEC cohort (D to F) received neoadjuvant and adjuvant HER2-targeted therapy (trastuzumab). Note that the original development cohort (cohort 1, Table 1) was included in the analysis for plots A to C. The length of each x-axis is proportional to the duration of follow-up of survivors, truncated when < 10% of the smallest group remained at risk. Below each survival plot is a corresponding histogram to show the relative frequency distribution of RCB index. Response categories are shown as pathologic complete response (pCR; blue), RCB-I (gold), RCB-II (gray), and RCB-III (red). HR, hormone receptors; HER2, human epidermal growth factor receptor 2.

Fig 3.

Hazard function plots illustrating the estimated rate of relapse-free survival events at each time of follow-up (rate per patient per year). (A) Triple-negative, (B) HR-positive/HER2-negative, (C) HER2-positive breast cancer treated with paclitaxel followed by fluorouracil, doxorubicin, and cyclophosphamide (T/FAC) chemotherapy, and (D) HER2-positive breast cancer treated with HER2-targeted therapy (trastuzumab) and sequential paclitaxel followed by fluorouracil, epirubicin, and cyclophosphamide (H+T/FEC) chemotherapy. Response categories are shown as pathologic complete response (pCR) or residual cancer burden (RCB)-I (blue), RCB-II (gray), and RCB-III (red). Y-axes use different scales; therefore, a black bar at hazard rate 0.10 was placed for reference. HR, hormone receptors; HER2, human epidermal growth factor receptor 2.

Fig A1.

Kaplan-Meier plots of relapse-free survival for residual cancer burden (RCB) categories in four cohorts who received neoadjuvant chemotherapy: (A) paclitaxel (T) followed by fluorouracil, doxorubicin, and cyclophosphamide (T/FAC) treatment in the original development cohort for RCB (T/FAC cohort 1, n = 219); (B) T/FAC cohort 2 (validation cohort; n = 262); (C) T/FAC cohort 3 (independent validation cohort [other pathologists]; n = 342); and (D) fluorouracil, doxorubicin, and cyclophosphamide (FAC) treatment in the originally published validation cohort (n = 124). Note that the original development cohort (cohort 1, Table 1) was included in these analyses. Response categories are shown as pathologic complete response (pCR; blue), RCB-I (gold), RCB-II (gray), and RCB-III (red). The length of each x-axis is proportional to the duration of follow-up of survivors, truncated when < 10% of the smallest group remained at risk.

Fig A2.

Distant relapse-free survival according to residual cancer burden (RCB) categories for phenotypic subsets of breast cancer after paclitaxel (T) followed by fluorouracil, doxorubicin, and cyclophosphamide (T/FAC) chemotherapy: (A) triple-negative breast cancer; (B) HR-positive/HER2-negative; (C) HER2-positive breast cancer treated with T/FAC chemotherapy; and (D) HER2-positive breast cancer treated with HER2-targeted therapy (trastuzumab) plus sequential paclitaxel and fluorouracil, epirubicin, and cyclophosphamide (T/FEC) chemotherapy. Note that the original development cohort (cohort 1, Table 1) was included in these analyses. Response categories are shown as pathologic complete response (pCR; blue), RCB-I (gold), RCB-II (gray), and RCB-III (red). The length of each x-axis is proportional to the duration of follow-up of survivors, truncated when < 10% of the smallest group remained at risk. HR, hormone receptors; HER2, human epidermal growth factor receptor 2.

Fig A3.

Overall survival according to residual cancer burden (RCB) categories for phenotypic subsets of breast cancer after paclitaxel (T) followed by fluorouracil, doxorubicin, and cyclophosphamide (T/FAC) chemotherapy: (A) triple-negative breast cancer; (B) HR-positive/HER2-negative; (C) HER2-positive breast cancer treated with T/FAC chemotherapy; and (D) HER2-positive breast cancer treated with HER2-targeted therapy (trastuzumab) plus sequential paclitaxel and fluorouracil, epirubicin, and cyclophosphamide (T/FEC) chemotherapy. Note that the original development cohort (cohort 1, Table 1) was included in these analyses. Response categories are shown as pathologic complete response (pCR; blue), RCB-I (gold), RCB-II (gray), and RCB-III (red). The length of each x-axis is proportional to the duration of follow-up of survivors, truncated when < 10% of the smallest group remained at risk. HR, hormone receptors; HER2, human epidermal growth factor receptor 2.

Fig A4.

Relapse-free survival for patients who achieved pathologic complete response from any neoadjuvant treatment (across all five cohorts) to illustrate whether the first relapse was distant to the brain (blue arrow) or other site (gold arrow), local relapse (gray circle), new primary (red circle), the result of a second type of cancer (blue circle), or death unrelated to cancer (dark gold circle). Symbols above the survival curve indicate events in patients who died. Symbols below the survival curve indicate events that did not lead to death during subsequent follow-up. Four relapse events were to the brain (a chemotherapy sanctuary site that is not surveyed during staging work-up) and were from triple-negative breast cancer (TNBC; 3 events) or human epidermal growth factor receptor 2–positive (HER2+) cancer after paclitaxel (T) followed by fluorouracil, doxorubicin, and cyclophosphamide (T/FAC) chemotherapy. The table lists the subtype of cancer and the surgical and pathology methods for the patients whose subsequent relapse could potentially relate to the adequacy of pathologic evaluation. There is no clear pattern among those who relapsed and whether pathologic sampling was from a total or partial mastectomy or absence of a metal clip that was radiographed in the specimen to identify the primary tumor bed. DOD, dead of disease; HER2−, HER2-negative; HR+, hormone receptor–positive; H+T/FEC, trastuzumab plus sequential paclitaxel and fluorouracil, epirubicin, and cyclophosphamide; ipsilat, ipsilateral; LN, lymph node; M, mastectomy; MBC, metastatic breast cancer; NA, information not available; NED, no evidence of disease; TTP, time to progression; WD, with disease. (^*) Adjacent to scar from previous partial mastectomy; (†) a pretreatment computed tomography scan showed an indeterminate abnormality not conclusive for malignancy; (‡) surgery was performed in another center and details from surgical pathology were not available from the health record.

Fig A5.

Histograms of the distribution of residual cancer burden (RCB) index in the patients who had residual disease (not pathologic complete response) at surgery immediately following neoadjuvant taxane-anthracycline chemotherapy (excluding those patients whose disease progressed), according to phenotype of disease defined as triple-negative breast cancer (TNBC; A and B), HR-positive/HER2-negative (HR+/HER2-; C and D), or human epidermal growth factor receptor 2–positive (HER2+) who did not (E and F) or did (G and H) also receive neoadjuvant trastuzumab. Panels A, C, E, and G show the distributions of RCB in all patients and in the subsets with pathologic node–negative (ypN−) or pathologic node–positive (ypN+) status. Panels B, D, F, and H show the distributions of RCB in patients defined by pathologic tumor stage (ypT1, ypT2, ypT3). H+T/FEC, trastuzumab plus sequential paclitaxel and fluorouracil, epirubicin, and cyclophosphamide; T/FAC, paclitaxel followed by fluorouracil, doxorubicin, and cyclophosphamide. HR, hormone receptors; HER2, human epidermal growth factor receptor 2.

Fig A5.

Histograms of the distribution of residual cancer burden (RCB) index in the patients who had residual disease (not pathologic complete response) at surgery immediately following neoadjuvant taxane-anthracycline chemotherapy (excluding those patients whose disease progressed), according to phenotype of disease defined as triple-negative breast cancer (TNBC; A and B), HR-positive/HER2-negative (HR+/HER2-; C and D), or human epidermal growth factor receptor 2–positive (HER2+) who did not (E and F) or did (G and H) also receive neoadjuvant trastuzumab. Panels A, C, E, and G show the distributions of RCB in all patients and in the subsets with pathologic node–negative (ypN−) or pathologic node–positive (ypN+) status. Panels B, D, F, and H show the distributions of RCB in patients defined by pathologic tumor stage (ypT1, ypT2, ypT3). H+T/FEC, trastuzumab plus sequential paclitaxel and fluorouracil, epirubicin, and cyclophosphamide; T/FAC, paclitaxel followed by fluorouracil, doxorubicin, and cyclophosphamide. HR, hormone receptors; HER2, human epidermal growth factor receptor 2.

Fig A6.

Relapse-free survival for the subsets of patients with HR-positive/HER2-negative cancer according to residual cancer burden (RCB) after neoadjuvant paclitaxel (T) followed by fluorouracil, doxorubicin, and cyclophosphamide (T/FAC) chemotherapy: (A) comparison of RCB classes in patients who had at least one involved regional lymph node (ypN ≥ 1); and (B) comparison of RCB index value > 2.25 versus ≤ 2.25 (the trough in the bimodal distribution of RCB in Figure 2B) only in those patients who had RCB-II. HR, hormone receptors; HER2, human epidermal growth factor receptor 2.