Invasive disease-free survival and brain metastasis rates in patients treated with neoadjuvant chemotherapy with trastuzumab and pertuzumab

Introduction
Approximately 15-20% of patients diagnosed with breast cancer have human epidermal growth factor receptor 2 (HER2) positive (+) disease (https://seer.cancer.gov/statfacts/html/breast-subtypes.html). These patients have a higher risk of developing brain metastases (BM), with retrospective studies and metanalyses reporting incidence rates between 30 and 40%1,2. The improved options of HER2 targeted and locally targeted therapy has helped to greatly improve outcomes in HER2(+) metastatic breast cancer (MBC) patients with BM, however the development of BM remains the limiting factor for long term survival in many patients.
In early breast cancer (EBC) HER2(+) patients who received neoadjuvant systemic therapy (NAST), the rates of BM have been reported to be approximately 2%3. However, the rate of BM development is unknown in the modern era with newer treatment options like trastuzumab emtansine (T-DM1) in the adjuvant setting in patients who have residual disease (RD). We aimed to analyze invasive disease-free survival (IDFS) and BM rates in patients with HER2(+) EBC in a modern population homogenously treated with NAST. We also report the incidence of patients with HER2 negative (−) RD and their outcomes.
Results
Baseline patient characteristics
The total cohort was 594 patients. 499 (84%) of patients had external HER2 testing done on their breast biopsy prior to receiving NAST at MSKCC. 456 (77%) and 138 (23%) received antracycline-taxane and taxane based chemotherapy, respectively, during NAST. Overall, 587 (99%) received trastuzumab and pertuzumab (HP) and 7 (1%) received trastuzumab alone. NAST was completed by 563 (95%) of patients. Baseline patient characteristics and NAST regimes received are summarized in Table 1.
Postoperative pathology results
In 594 patients, pCR (ypT0/isN0) was achieved in 325 (55%) and RD was seen in 269 (45%) patients. In 269 patients with RD, 45 (17%) did not have HER2 retesting and were excluded from the final analysis. This was mainly due to insufficient residual tissue for retesting or if the initial breast biopsy had HER2 testing done internally at our center. In the remaining 224 patients, 143 (64%) retained HER2(+) and 81 (36%) were HER2(−) (Fig. 1). The 81 patients with HER2(−) RD had pre op HER2 status of IHC 0/FISH amplified, 1 + /FISH amplified, IHC 2 + /FISH amplified and IHC 3+ in 1 (1%), 4 (5%), 38 (47%) and 38 (47%), respectively. Postoperatively, the 81 patients with HER2(−) RD were IHC 0, 1+ and 2 + /FISH non amplified in 21 (26%), 42 (52%) and 18 (22%) patients, respectively; 69 (85%) were HR(+) and 12 (15%) were HR(−).

The boxes indicate the breakdown of patient numbers according to their residual disease status. EBC early breast cancer, HER2 human epidermal growth factor receptor 2, NAST neoadjuvant systemic therapy, pCR pathological complete response, RD residual disease.
Adjuvant treatments received
For the 325 patients who achieved pCR, 295 (91%) successfully completed a year of HER2 targeted therapy with HP. Cardiotoxicity and early recurrence were the main reason for those who discontinued with 9 (2.8%) having symptomatic left ventricular ejection fraction (LVEF) decline, 5 (1.5%) having asymptomatic LVEF decline and 2 (0.6%) having early recurrence. In the 143 patients with HER2(+) RD, adjuvant T-DM1, HP, trastuzumab alone and no HER2 directed therapy were initiated in 121 (85%), 16 (11%), 1 (1%) and 5 (3%) of patients, respectively. In the 81 patients with HER2(−) RD, adjuvant T-DM1, HP, trastuzumab alone and no HER2 directed therapy were initiated in 45 (56%), 27 (33%), 3 (4%) and 6 (7%) of patients, respectively.
Out of the 224 patients with HER2 tested RD, 166 (74%) received T-DM1. In the 166 patients who received T-DM1, 124 (75%) completed the full course of therapy, 30 (18%) discontinued T-DM1 but completed a full year of HER2 targeted therapy with (27 received HP and 3 received trastuzumab alone) and 12 (7%) discontinued T-DM1 with no further HER2 targeted therapy.
Toxicity was the main reason for the patients who discontinued T-DM1 (39/42), with peripheral neuropathy, fatigue and cardiotoxicity being the most common toxicity in 23 (59%), 7 (18%) and 4 (10%) of patients respectively. Upon completion of the first year of HER2 targeted therapy, 4 (1%) patients received adjuvant neratinib. The adjuvant therapies received by patients with RD are summarized in Table 2. The update of the KATHERINE trial regarding outcomes of adjuvant T-DM1 in HER2(−) RD patients was published in Sept 20224, and thus this knowledge would not have been available during the treatment timeframe of our cohort.
BM incidence and IDFS events
With a median follow up of 24 months, 8 patients developed BM at initial recurrence, 4/325 (1.2%) with pCR and 4/143 (2.8%) with HER2(+) RD (p = 0.7). None of the patients who developed BM had HER2(−) RD. In the 8 patients who developed BM, all 4 patients with pCR presented with intracranial only disease, while in HER2(+) RD group 2 had both intracranial and extracranial metastases and 2 had intracranial disease alone. First line BM local therapy received included surgery (3 patients), stereotactic radiosurgery (4 patients), and whole brain radiotherapy (1 patient).
IDFS events occurred in 22/594 (3%) patients. In these 22 patients, distant recurrence, locoregional recurrence, contralateral invasive breast cancer and death from other causes occurred in 16 (72%), 1 (5%), 1 (5%) and 4 (18%) patients respectively. There was no difference in IDFS between 15/456 (3%) patients who received an anthracycline during NAST and 7/138 (5%) who did not receive an anthracycline (p = 0.051).
Patients with RD had a higher likelihood of having an IDFS event, 14/269 (5.2%) in those with RD and 8/325 (2.5%) in patients with pCR (p = 0.016) (Fig. 2). In the evaluable 224 patients with RD, there was no difference in IDFS between 9/143 (6%) patients with HER2(+) RD or 5/81 (6%) with HER2(−) RD (p = 0.10). Out of the 5 patients with HER2(−) RD, IDFS events were of distant recurrence (2), locoregional recurrence (1) and death from other causes (2). Overall, 4 patients received adjuvant T-DM1; the single patient who did not receive adjuvant T-DM1 had an IDFS event of death from other causes.

Kaplan-Meier estimates of IDFS compared using log-rank tests by pathological complete response. CI confidence interval, HR hazard ratio, invasive disease-free survival, pCR pathological complete response.
In the 16 patients who developed metastatic disease, 7 had pCR, 7 had HER2(+) RD and 2 had HER2(−) RD. 4/16 patients did not have metastatic biopsy HER2 status assessment due to brain only disease or insufficient tissue for analysis. Of those who had metastatic biopsy HER2 testing, 4 pCR patients all had HER2(+) metastatic disease, 6 HER2(+) RD patients had 4 HER2(+) and 2 HER2(−) metastatic disease and both of the 2 HER2(−) RD patients developed HER2(−) metastatic disease.
Discussion
With a median follow up of 24 months, we observe no statistical difference in the rate of BM development at initial recurrence, regardless of if the patient achieved pCR or had RD post NAST at 1.2% versus 2.8%, respectively. In the patients who had BM at initial recurrence, we note that all of the patients with pCR developed intracranial disease only, while those with RD had both intracranial and extracranial disease. Penetration across an intact blood brain barrier is thought to be limited for anti-HER2 agents like HP, which leads to inadequate therapeutic cerebrospinal fluid (CSF) concentrations5. Thus, a reasonable theory is that there is a selection of resistant clones that escape to the brain as a sanctuary site, which leads to the occurrence of BM relapse regardless of pCR status.
The APHINITY trial which investigated adjuvant chemotherapy/trastuzumab +/− pertuzumab in high risk HER2(+) EBC patients noted similar rates of BM incidence between the pertuzumab and placebo groups (1.9% versus 1.8%)6. The secondary analysis of the TRAIN-2 trial where patients received neoadjuvant HP with different chemotherapy regimens noted similar distant metastatic rates with the anthracycline and non-anthracycline groups (5% versus 6%, respectively). However, BM incidence rates were not reported7. In the KATHERINE trial which studied adjuvant T-DM1 versus trastuzumab in patients with RD, BM at initial recurrence rates for T-DM1 versus the control arm were 5.9% versus 4.3%, respectively. However, only 18% of the patients enrolled in the trial received NAST with HP8. As our cohort had a small number of BM events, it was not possible to further stratify our findings according to the type of adjuvant therapy received or the amount of RD.
In our cohort, none of the HER2(−) RD patients developed BM. Approximately 85% of our HER2(−) RD patients had HR(+) disease, which is known to have a lower incidence rate of BM development at initial metastatic recurrence at approximately 5%, and a longer time interval to disease recurrence9. As our median follow up is relatively short at 24 months, a possible theory is that these patients have a different metastatic pattern and longer follow up is required.
It is well established that patients with HER2(+) EBC receiving NAST have poorer outcomes if they have RD after surgery which is mirrored in our findings10. Loss of HER2 positivity in RD after NAST has been reported in literature to occur in 20–40% of patients11,12. Theories regarding the mechanism behind HER2 loss include the effect of breast cancer heterogeneity, where HER2(+) subpopulations are killed by treatment which leads to the selected survival of resistant HER2(−) cells13. Another theory postulates the effect of HER2 downregulation, where anti HER2 drugs that target the extracellular domain of HER2 like HP downregulate HER2 membrane expression by inducing HER2 internalization14. This could be a possible explanation as to why the HER2(−) RD rate reported in KATHERINE is lower at 8%, as there was less adjuvant dual-HER2 therapy given8.
There have been mixed results regarding the prognosis of patients who have HER2(−) RD, with older retrospective studies showing a higher risk of relapse compared to those who have HER2(+) RD15,16. However, more recent data with modern patient cohorts do not show inferior oncologic outcomes for patients who have HER2(−) RD, which is also reflected in our current findings12,17. A subset analysis of the KATHERINE trial suggests that patients with HER2(−) RD (8%) have better IDFS with adjuvant T-DM1 versus trastuzumab alone and suggests that adjuvant T-DM1 should be offered in this patient population4.
As events of BM recurrence are similar in patients with pCR and RD, newer HER2 targeted systemic therapies with proven brain efficacy in the metastatic setting are attractive options for use in the EBC setting which may confer greater brain protection. Tucatinib is a HER2 tyrosine kinase inhibitor studied in the HER2CLIMB trial and showed an improved OS of 21.9 months in the tucatinib/trastuzumab/capecitabine group versus 17.4 months in the placebo/ trastuzumab/capecitabine group (Hazard ratio 0.66, 95% confidence interval (CI) 0.50 – 0.88, p = 0.005) for MBC patients with BM18. Trastuzumab deruxtecan (T-DXd) is an antibody drug conjugate which has shown promising brain activity in the TUXEDO-1 study with an intracranial response rate of 73% and median PFS of 14 months (95% CI 11.0 to not reached)19 and is being further investigated in the BM space in the ongoing DESTINY-B12 (NCT04739761) and DEBBRAH trials20. Tucatinib is currently being investigated in the adjuvant setting in combination with T-DM1 (NCT04457596) and T-DXd is currently being investigated in both the neoadjuvant (NCT05113251) and adjuvant setting (NCT04622319); with results that are eagerly awaited.
Limitations of our study include it being retrospective and at a single center. The median follow up time is relatively short at 24 months with a modest number of BM events. Many of our patients had their initial breast biopsy HER2 testing done externally (84%) before seeking care at MSK. HER2 testing was not repeated on the biopsy specimen if it had already been done externally and reflects real world experience. However, we acknowledge that this introduces an element of uncertainty in being able to accurately assess whether patients with HER2(−) RD truly had different biology.
At a single center, in patients who predominantly received HP with chemotherapy as NAST, patients with RD had higher IDFS events than those with pCR. In those with RD, 36% lost HER2(+) status; IDFS events appeared similar in those with HER2(+) RD versus those with HER2(−) RD. The HER2 loss rate is higher than reported in KATHERINE (8%) possibly due to majority of patients receiving dual HP as NAST. The BM events seen in those with RD and pCR highlights the need for more effective therapy in NAST and adjuvant setting to minimize BM risk. Newer HER2 targeted agents with high CNS bioavailability like tucatinib and antibody drug conjugates are attractive options for investigation in the EBC setting to reduce the incidence of BM recurrence.
Methods
Patient cohort
Patients with HER2(+) EBC who received NAST followed by surgery at Memorial Sloan Kettering Cancer Center (MSKCC) between 1 Jan 2019 and 31 Jan 2022 were identified. Patients who were receiving novel NAST through clinical trials were excluded. External assessment of HER2 status before NAST was allowed, while HER2 status of the surgical specimens with RD were assessed internally at our center. HER2 positivity was defined according to ASCO-CAP guidelines21,22. Patients with either HER2 overexpression by immunohistochemistry (IHC) 3+ or IHC 0-2+ with gene amplification by fluorescence in situ hybridization (FISH) were included. Hormone receptor (HR) status (estrogen receptor (ER) or progesterone receptor (PR)) was assessed by IHC and defined as positive if the cancer cell nuclei had staining of ≥1%23.
Statistical methods
IDFS was defined as the time from surgery until first occurrence of invasive breast cancer locoregional recurrence, contralateral breast cancer, distant recurrence or death from any cause. IDFS was evaluated using Kaplan-Meier estimates and compared using log-rank tests by pathological complete response (pCR) and HER2 status. IDFS event and BM incidence rate at 24 months post surgery according to pCR and HER2 status were evaluated and the differences were tested using Chi-square and Fisher’s tests. Data from patients who did not have a documented event were censored at the last follow up date. Statistical analyses were conducted using R 4.1.2 software.
Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this article
Responses