Although there have been several advances in the treatment of locally advanced breast cancer (LABC), the relapse rate for patients with this disease is more than 25 percent, with most of this occurring during the first 3 years after diagnosis. Olexiy Aseyev, MD, PhD, and his colleagues at the Ottawa Hospital Cancer Center, Department of Medicine, University of Ottawa, Ottawa, Ontario, Canada, undertook a study to develop a tool to help gauge the risk of relapse in this sizable group of patients. This issue has been addressed by other groups previously; for example, in 2014, Matsuda, et al, (J Surg Oncol 2014;109(8):764-9) described a model for predicting locoregional recurrence of breast cancer in patients having undergone neoadjuvant chemotherapy followed by breast-conserving therapy. Their predictive model was generated from a study involving 520 breast cancer patients from 2001-2008, and the results were utilized to find those patients who were at higher risk for relapse, so that additional therapy could be provided.
The primary objective in the Ottawa study was to construct a predictive tool for the risk of relapse (RoR) in LABC patients. As Aseyev explained, “In our study, we also wanted to identify those patients who had a higher RoR so that appropriate treatments could be given and perhaps avoid disease recurrence. We also sought to stratify those patients who had lower RoR, as previous studies have not thoroughly evaluated the risks for that patient population.”
Studying Neoadjuvant Therapy
This was a retrospective single center study that evaluated 546 LABC patients who had received neoadjuvant therapy (NAT) at the Ottawa Hospital Cancer Center from 2005-2015. Regarding the start date of the study, Aseyev commented, “We chose this date, because it marked the date when trastuzumab was incorporated into therapy for HER2-positive breast cancer patients. The median follow-up time was 46-48 months for those patients having no progression; however, for those showing disease progression, the follow-up time was somewhat shorter, 29.6 months.”
The following patient data were recorded: demographics, cancer treatment, nodal status, tumor size, grade and stage of disease, HER2- and estrogen receptor (ER) status, and clinical outcome. The primary endpoints for this study were distant or local disease relapse (DR) rate during the first 5 years after treatment as well as the time to DR within that same 5-year period. From the data obtained, a predictive model was developed using Cox regression methodologies.
NAT was administered to 545 patients according to the following regimens: AC-Docetaxel (doxorubicin—cyclophosphamide—docetaxel) 330 patients, 60 percent; FEC-D (fluorouracil—epirubicin—cyclophosphamide—docetaxel) 91 patients, 17 percent; other regimens (e.g., AC (doxorubicin—cyclophosphamide), AC-Paclitaxel (doxorubicin—cyclophosphamide—paclitaxel), TC (Taxotere—cyclophosphamide), TCH (Taxotere—carboplatin—trastuzumab)) 124 patients, 23 percent. All HER2-positive patients (173, 34%) received trastuzumab therapy, while ER-positive patients (356, 44%) received tamoxifen and/or aromatase inhibitor therapy. Mastectomy was performed on 440 of the patients (81%), while 67 (12%) had breast-conserving surgery. Additionally, 485 patients (89%) had adjuvant radiotherapy. For the first 5 years after NAT, the overall DR rate was 17.3 percent (local DR—3.2%, distant DR—13.2%, local and distant DR—0.9%).
A Cox regression proportional hazards model was applied to the data obtained for more than 60 variables, and from this analysis only five factors were shown to have a significant influence on the RoR during the 5-year follow-up period. These risk factors in the order of their predictive value are as follows:
- residual disease (yes-4, no-0), (Hazard Ratio (HR) – 4.25, p-value = 0.000)
- lymph nodes status (positive-3, negative-0), (HR = 2.27, p-value = 0.006)
- inflammatory histology (yes-2, no-0), (HR = 1.90, p-value = 0.003)
- estrogen receptor status (ER-positive-2; ER-negative-0), (HR = 2.07, p-value = 0.001)
- adjuvant radiotherapy (yes-0, no-1), (HR = 1.76, p-value = 0.036).
When the factors for a patient are totaled, a relapse prediction (RP) score can be obtained.
A RP score of 0-5 corresponded to a low risk (7% chance) of relapse over a 5-year period, while a score of 6-7 equated to an intermediate RoR (26% chance over 5 years). RP scores of 8-12 gave the highest RoR (51% over 5 years). In this study, 153 (28%) of the patients were characterized as being in the low-risk group. Of these patients, 76 were analyzed (i.e., monitored for the full 5-year period after treatment) while 77 were censored (not monitored for 5 years after treatment). Three of the patients who were included in this risk level had local DR and three had distant DR, while none displayed local and distant DR. The largest group of patients in this study was included in the intermediate risk group (220, 40%) and, of these, 124 were analyzed and 96 were censored. Distant DR was observed in 27 of the patients in this group and five had local DR. As with the low-risk stratified patients, none of this group had local and distant DR. In the patients who had the highest RoR, 43 had distant DR, nine had local DR and, most importantly, five had local and distant DR. These data showed the RoR for the high-risk patients (RP score 8-12) was more than 7 times greater than that of the low-risk patients (RP score 0-5). This tool, which was developed using only five risk factors, was shown to have a sensitivity of 75 percent when predicting the RoR in LABC patients having NAT.
Future Treatment Needs
“The relapse of disease at sites distant from the initial tumor is clearly the most important outcome; these patients definitely need more treatment,” Aseyev noted. “This model may also have utility in suggesting future treatment strategies for high-risk patients. Currently, HER2+ patients are on a regimen of trastuzumab for 1 year; however, in the future, that may be extended to as long as 3 years in an attempt to reduce the chances of a relapse.
“Likewise, for ER+ patients, longer therapies of 5-10 years may become necessary,” he added. “This tool we have developed clearly shows that follow-up after surgery is very important, as it may suggest further treatments for the patients as well as stratify those at lower RoR.” When asked about the next steps for this model, he replied, “This tool will validated by applying it to a larger provincial or multi-center patient population.”
Richard Simoneaux is a contributing writer.