Targeting mTOR and DNA repair pathways in residual triple negative breast cancer post neoadjuvant chemotherapy

Triple-negative breast cancer (TNBC) patients who do not achieve pathologic complete response post neoadjuvant chemotherapy have a poor prognosis. Alteration in PI3K/mTOR plus DNA repair

pathways are some of the major mechanisms of chemotherapy resistance. We designed an open-label phase II clinical trial to evaluate if the combination of everolimus (mTOR inhibitor) plus

cisplatin (interferes with DNA function) will improve the rate of pathologic response, as assessed by residual cancer burden (RCB). Twenty-four Stage II/III TNBC patients with residual

cancer > 1 cm post neoadjuvant anthracycline and taxane-based chemotherapy were enrolled. Patients received everolimus daily orally at 10 mg for 12 weeks and cisplatin IV at 20 mg/m2 weekly

for 4 cycles (21-day cycle), until definitive surgery. The primary endpoint was the rate of RCB-0-I at the surgery. The median age of the whole cohort was 50.1 years, with 66.7% non-Hispanic

Caucasians. Of the 24 patients enrolled, 22 were included in the efficacy analysis. Twenty-one patients underwent definitive surgery while one patient developed distant metastasis. Five

patients had RCB-I at surgery, a response rate of 23% (5/22). Patients with germline PALB2 mutation or somatic PI3KCA mutation had a pathologic response, achieving RCB-I at the surgery.

Three patients had metaplastic histology achieving RCB-I at the surgery. Estimated OS at 1 year was 100% in the RCB-I group vs. 76.5% in others, which was not statistically significant due

to the small sample size. Certain cohorts including PALB2 germline mutation carrier and somatic PI3KCA mutations warrant further investigation.

Trial registration: Clinicaltrials.gov identifier: NCT01931163. https://clinicaltrials.gov/ct2/show/NCT01931163.

Twenty-four patients with triple-negative chemorefractory breast cancer (post anthracycline plus taxane-based chemotherapy, with the biopsy-proven residual disease) were enrolled in the

study from June 1, 2013, to August 30, 2017, after obtaining informed written consent (Fig. 1). The median age of the whole cohort was 50.1 years (Table 1). The majority of patients were

non-Hispanic Caucasian (66.7%) followed by African American (25%) and Hispanic Caucasian (8.3%). Patients presented with both early-stage and locally advanced disease with 54.2% (13/24)

stage II disease while 45.8% (11/24) had stage III disease. Of these twenty-four patients, three had the rare subtype of metaplastic breast cancer (3/24, 12.5%). Patients underwent germline

mutation testing, and only 10% (2/20) of those tested had germline genetic mutation, both of which were deleterious PALB2 mutations. Notably, there were no patients with germline BRCA1 or

BRCA2 mutations in this chemorefractory TNBC patient cohort. The most common prior standard neoadjuvant regimen received by the study cohort was dose-dense doxorubicin plus cyclophosphamide

(60 mg/m2 IV doxorubicin plus 600 mg/m2 IV cyclophosphamide on Day1 every 14 days × 4 cycles) followed by weekly paclitaxel (80 mg/m2 IV every 7 days × 12 cycles).

Out of twenty-four patients enrolled, one patient progressed/metastasized prior to receiving the study treatment and a second patient withdrew after 10 days on trial without toxicity or

progression, and thus were not evaluated for treatment response. The treatment response was assessed in twenty-two patients. Of these, 10 patients did not complete all four cycles of

cisplatin/everolimus—six patients came off the study due to disease progression (one patient developed distant metastasis while five had progression of the primary lesion) and four patients

secondary to treatment toxicity. Twenty-one patients out of twenty-two patients underwent definitive surgery. The predefined primary endpoint for these chemorefractory TNBC patients was

residual cancer burden (RCB). Sample size estimations were based on historical pathologic response (RCB 0–1) of less than 5%. Here, in this trial, the overall response rate for RCB 0-I was

23% (5/22) (95% confidence interval (CI) 10.1–43.4%). While no patients achieved RCB 0, 5 patients did achieve RCB-I (Table 2). The characteristics of patients achieving RCB 0-I at the

surgery are summarized in Table 3.

At a median follow-up of 29 months (95% CI 25–36.5 months) from study enrollment, 64% (14/22) of the patient remained free of distant metastasis while 46% (6/22) developed metastatic

disease. There were 5 deaths observed among the 22 patients. The estimated overall survival (OS) at 1 year was 81% and at 4 years was 65.5%. (Fig. 2A). The 1-year OS was 100% in responders

vs. 76.5% in non-responders, which is not statistically significant given the small sample size (Fig. 2B).

(A) OS from the date of first treatment. Estimated OS at 1 year is 81% and 65.5% at 4 years. (B) OS in Responders (RCB-I at surgery) vs non-responders (RCB II-III at surgery). (p = 0.7).

The regimen had fair tolerance with only grade 1 and 2 toxicities in 95.4% (21/22) and grade 3 and 4 in 18.2% (4/22) of the patients. Three main toxicities with incidence > 20% were fatigue

(45%), nausea (41%), and mucositis (23%) (see Supplementary Table 1). One patient experienced grade 3 nausea, one patient had both grade 3 thrombocytopenia and grade 3 hyperglycemia, one

patient had both grade 3 leucocytopenia and neutropenia while one patient had grade 4 papilledema.

Somatic mutation testing by next-generation exome sequencing was performed in 45% (10/22) of the patients. Somatic mutation testing was performed on pre-treatment biopsy samples using

FoundationOne companion diagnostic. Germline mutation testing was available for all these 10 patients, germline testing was done using commercially available Myriad myRisk. Here, no germline

BRCA1/2 mutations were detected in patients with triple-negative breast cancer, contrary to expectations. Two patients had germline PALB2 mutations. The most predominant somatic mutation

was in TP53 in 60% of the patients (6/10). Of 5 patients with RCB 0–1, two had deleterious PALB2 germline mutations. All five patients underwent somatic mutation testing. Of these, two

patients had actionable somatic PI3KCA mutations. Of interest, of the three patients with metaplastic breast cancer, two patients who had somatic PIK3CA mutation responded with a residual 4 

mm (patient #1) and 8 mm (patient #8) tumor post-treatment. On comparing mutation profile between responders and non-responders (see Supplementary Figure 2), mutations affecting PI3K/mTOR

pathway and DNA repair mechanisms were enriched in the responder cohort while non-responders had enrichment of TP53 alterations (all cases). The difference in PALB2 mutation in responders

vs. non-responders (p = 0.44) and P1K3CA mutation in responders vs. non-responders (p = 0.44) were not statistically significant while the difference in TP53 alteration in responders vs.

non-responders (p = 0.04) was statistically significant using Fisher’s exact test.

In this combination study of everolimus plus cisplatin in the patient’s refractory to standard chemotherapy, where any pathologic response is notable, we report a 23% RCB-I pathologic

response. In a long term follow up of patients achieving RCB-0 (pCR) or RCB-I in TNBC post neoadjuvant chemotherapy, prognoses was superior to patients with RCB-II and RCB-III9. Relapse-free

survival and overall survival were improved in the TNBC patient subset with RCB-0 or RCB-I at surgery. In a multivariate model of relapse-free survival in TNBC, RCB was prognostic

independent of other clinical pathologic variables9. Except for one patient that developed metastasis while on the study protocol, all others were able to get definitive surgery.

On analysis of characteristics of 5 patients who had RCB-I at surgery, 2 had PALB2 germline mutation, 2 had PIK3CA somatic mutation and one had CHEK1 mutation. PALB2 is a tumor suppressor

gene that interacts with BRCA2 and is required for DNA repair10. Women with PALB2 mutations are at increased risk of developing breast cancer11. More importantly, breast cancer patients with

known PALB2 mutations are known to have a poor prognosis12. Patients with germline PALB2 mutation may be sensitive to drugs that target DNA repair mechanisms like cisplatin13 or PARP

inhibitors, which are being tested in clinical trials in various tumor types including pancreatic cancers14,15. In our trial, both patients with germline PALB2 mutation had RCB-I at the

surgery. Patient #18 with germline PALB2 mutation presented with ~ 10 cm cancer, had large 3.5 cm residual cancer following neoadjuvant anthracyclines and taxanes. Notably, after everolimus

and cisplatin, she had no residual cancer in the breast and one small 0.3 cm focus in the axillary node at the time of surgery. The other germline PALB2 mutation carrier (Patient #20) had

the chemorefractory disease, with her primary tumor progressing while on standard neoadjuvant anthracycline and taxane. After 4 cycles of everolimus and cisplatin, she responded clinically,

and her pathologic residual cancer was under 1 cm. There was no germline BRCA 1 or 2 mutations cancer in this study of chemorefractory patients, reflective of more chemosensitive disease in

germline cancers. Two patients who had somatic PI3KCA mutation also responded to cisplatin plus everolimus, both had RCB-I at the surgery. Patient #1 with PI3KCA mutation had only 0.45 cm

disease left at surgery post combination therapy, responding even when concomitant TP53 mutation was present. Patient #8 also had PI3KCA mutation and had less than 1 cm disease at surgery.

Of the three patients with metaplastic cancers which are known to be highly refractory to most chemotherapy regimens and to have one of the worst prognosis, two patients (Patient #1 and #8)

who had PIK3CA mutation were able to achieve RCB-I at the surgery. The remaining one metaplastic patient (#7) did not have PIK3CA mutations but the CHEK1-ATR pathway mutation also achieved

RCB-I at the surgery.

The overall survival of the whole cohort was 81% at 1 year and 65.5% at 4 years. Estimated OS at 1 year was higher in patients achieving RCB-I vs. others at 100% vs. 76.5% respectively,

although this was not statistically different due to a low number of patients in this trial. Patients in the responder group had enrichment of mutation affecting DNA repair (PALB2, CHEK1,

and ATR) and PI3KCA/mTOR pathway, compared to non-responders. Non-responders had enrichment of TP53 alterations compared to responders.

Everolimus has been previously studied in combination with chemotherapy in the upfront neoadjuvant setting in TNBC where its addition led to a lower pCR rate compared to chemotherapy with

increased toxicity16. In this trial, the regimen consists of everolimus in combination with cisplatin, after neoadjuvant chemotherapy where the residual tumors have been shown to have an

enrichment of both the PI3K/mTOR pathway along with DNA repair alterations as the major mechanistic pathways of chemotherapy resistance8. We report a noteworthy 23% of the patients achieved

RCB-I at the surgery. In this trial, patients who had mutations affecting the DNA repair pathway and patients with somatic PI3KCA mutations had a response to treatment with cisplatin plus

everolimus. There are limitations of this study, namely a low number of patients and a single-arm study with no randomization arm. Despite the limitations, these results add to the

understanding of targeting the PI3K/mTOR pathway in TNBC especially in patients who have residual disease post neoadjuvant chemotherapy where the current standard of care is the use of

adjuvant capecitabine17. Further studies evaluating cisplatin plus everolimus in cohorts with PI3K pathway alteration or PALB2 germline mutation is needed especially in poor prognosis

metaplastic breast cancer patients18.

The combination of everolimus plus cisplatin is active in the neoadjuvant setting in TNBC patients who have residual disease post standard neoadjuvant chemotherapy with a response rate of

23% for RCB-I at the surgery. This combination was active in a subset of patients with germline PALB2 mutation or somatic PI3KCA mutation. This is the first study using targeted therapy in a

neoadjuvant setting for TNBC patients who had residual disease post standard anthracycline-taxane neoadjuvant chemotherapy. Responders to this combination included patients with germline

PALB2 mutation and metaplastic histology who are known historically have a poor prognosis. Further studies evaluating this regimen in cohorts with PI3KCA mutation or PALB2 germline mutation

is needed.

All female patients > 18 years of age with TNBC defined as estrogen receptor-negative and progesterone receptor-negative (