Sara Tolaney, MD, MPH

At SABCS 2019, results from multiple key studies were presented that promise to further expand options for patients with breast cancer. In this commentary, I discuss the INFORM study of doxorubicin plus cyclophosphamide (AC) vs cisplatin in EBC with germline BRCA mutations; a phase III trial of oral administration with paclitaxel plus encequidar; and 2 trials assessing neoadjuvant immune checkpoint blockade in EBC.

Phase II INFORM: Neoadjuvant AC vs Single-Agent Cisplatin in Breast Cancer With BRCA Mutations
The INFORM trial is a randomized phase II study comparing neoadjuvant cisplatin to AC chemotherapy in patients with HER2-negative, germline BRCA1/2 mutation–positive EBC.²⁸ The rationale behind the INFORM trial was based on data suggesting that platinum therapy may be more effective than traditional preoperative anthracycline-based chemotherapy in patients with breast cancer and germline BRCA mutations. A retrospective analysis of patients with breast cancer and germline BRCA1 mutations after neoadjuvant chemotherapy showed a 21% pCR rate after fluorouracil, doxorubicin, and cyclophosphamide (n = 11/51) vs 83% with cisplatin (n = 10/12).²⁹ In addition, the phase III TNT study of first-line carboplatin vs docetaxel in advanced TNBC reported a higher response rate to platinum therapy (carboplatin) in BRCA mutation carriers than with taxane.³⁰ These data prompted the question of whether patients with germline BRCA mutations achieve better outcomes with platinum‑based neoadjuvant therapy than with traditional anthracycline‑based therapy, and if so, which approach should be standard treatment?

To help answer this question, Tung and colleagues²⁸ conducted the randomized phase II INFORM study (N = 118) in which patients with HER2-negative invasive breast cancer and a germline BRCA1/2 mutation were randomized to either 4 cycles of AC chemotherapy or 4 cycles of cisplatin, followed by surgery. The primary endpoint was pCR.

Of importance, results showed that rates of pCR were higher with anthracycline therapy than the platinum agent: 26% with AC vs 18% with cisplatin. This difference persisted in the subgroup of patients with TNBC (28% vs 22%, respectively) and was even more pronounced in patients with ER/PgR-positive breast cancer (21% vs 6%, respectively). As additional data have emerged from other trials, we see that the balance of the evidence is fairly consistent with these newer data from INFORM. For example, in secondary analyses from the GeparSixto trial and data from the newer BrighTNess trial, despite the fact that patients with breast cancer and germline BRCA mutations tend to be more chemosensitive, there was no significant increase in pCR rates from the addition of platinum to an anthracycline-based regimen.^31,32

The practical implication of these findings is that patients with breast cancer and BRCA mutations do not necessarily need platinum therapy added to their preoperative or adjuvant treatment. Today, my standard of care for patients with breast cancer and germline BRCA mutations is to continue to use anthracycline‑based chemotherapy, typically the dose‑dense AC plus paclitaxel (AC-T) regimen.

Phase III Trial: Oral Paclitaxel Plus Encequidar in MBC
We all have been eager to use oral chemotherapy for our patients with breast cancer, yet it has been quite challenging to develop an oral taxane. This is because administering a taxane, such as paclitaxel, orally results in excretion via efflux by the P‑glycoprotein pump and, thus, poor or no oral absorption.³³ Research has suggested that using an oral P-glycoprotein pump inhibitor with oral paclitaxel should prevent this efflux, allowing taxanes to be dosed orally, and encequidar and other agents are being investigated for this purpose.³⁴

In a randomized phase III trial presented at SABCS 2019 by Umanzor and colleagues,³⁵ the combination of encequidar with oral paclitaxel was compared with IV paclitaxel every 3 weeks in 402 patients with MBC after at least 1 year since their last taxane (for either adjuvant or metastatic treatment). Although this was a “head-to-head” design, this trial did not restrict the patient population by breast cancer subtype or by line of therapy. Approximately one half of patients enrolled on this trial were HR positive/HER2 negative (56% in the oral arm vs 49% in the IV arm); approximately 9% were HR positive/HER2 positive; 8% were triple negative in the oral arm vs 15% in the IV arm; and approximately 20% had unknown hormone receptor and HER2 status. A total of 26% of patients in the oral arm had received previous chemotherapy vs 28% in the IV arm. Most patients had received 1 line of chemotherapy (14% vs 16%), but 7% to 8% of patients in both arms had received 2 lines, and 3% to 5% had received 3 or more lines of treatment. Patients in both arms received six 3-week cycles of treatment with an extension phase continuing for 9 or more cycles. The primary endpoint was confirmed tumor response by Week 19 (defined as 2 consecutive scans showing PR/CR), which is not a traditional endpoint used, and PFS and OS were secondary endpoints.

Results showed that the combination of oral paclitaxel and encequidar was superior to IV paclitaxel, with ORRs of 40.4% vs 25.6%, respectively (P = .005). Most of these were PRs, although 1.3% of patients in the oral arm and 0.8% in the IV arm achieved a CR. Responses favored the oral formulation across most patient groups evaluated, including age (younger than 65 years vs 65 years or older), Eastern Cooperative Oncology Group performance status (0 vs 1), biomarker status (HR positive/HER2 negative vs HR positive/HER2 positive vs triple negative vs HR/HER2 unknown), and previous chemotherapy (taxane vs anthracycline). The median PFS was similar between arms (approximately 9%), but there was a trend toward a longer median OS with the oral drugs vs IV administration (27.7 vs 16.9 months, respectively; P = .114).

The oral regimen was associated with less neuropathy (7.6% vs 31.1%) and less alopecia, although approximately 30% of people did still lose their hair with the oral regimen (28.8% vs 48.1%). Also, compared with the IV formulation, the oral drugs were associated with more gastrointestinal toxicities, such as nausea and diarrhea, which were seen in approximately 25% of patients receiving oral paclitaxel vs only approximately 5% to 8% receiving IV paclitaxel. That said, grade ≥ 3 nausea and diarrhea were rare, affecting only approximately 5% of patients in the oral arm and approximately 1% in the IV arm.

The challenge in interpreting these data is that the oral regimen was compared with every-3‑week IV paclitaxel, even though weekly paclitaxel IV is superior to every-3‑week paclitaxel. It is difficult to extrapolate these data to know whether this oral regimen might have been more effective using standard weekly IV paclitaxel dosing. The other challenge in interpreting these data is the use of tumor response (ie, ORR) as the primary endpoint, particularly in a population where most patients have ER‑positive disease.

There are definite advantages to an oral regimen, including the lack of hypersensitivity reactions, less need for patients to return to the clinic for frequent infusions, and less hair loss and neuropathy. However, patients may experience more gastrointestinal toxicity, will have to take many pills for this regimen, and must be educated to be adherent to the dose scheduling. On average, patients were required to take 11 pills per dose of the oral paclitaxel and had to sequence it with fasting. Patients first took encequidar, then waited to take the oral paclitaxel, then waited 4 hours to eat; this is a lot of fasting and schedule adherence for a patient to manage; however, it was feasible and was found to be a very active regimen and still likely preferable to an IV regimen for most patients.

Another oral taxane in development is tesetaxel, which is not removed by the P‑glycoprotein pump and would not need to be accompanied with a drug like encequidar. The phase III (and registrational) CONTESSA study is comparing tesetaxel every 3 weeks plus capecitabine vs capecitabine alone in patients with HR–positive, HER2-negative locally advanced or MBC who previously received a taxane in the early disease setting (neoadjuvant or adjuvant setting).³⁶ I would also be interested to see this approach tested in EBC because it could help minimize clinic visits and decrease alopecia.

KEYNOTE‑522: Neoadjuvant Pembrolizumab Plus Chemotherapy in Early TNBC
For patients with TNBC, there are data in the metastatic setting that suggest adding a checkpoint inhibitor to chemotherapy can improve OS in patients with PD-L1–positive disease. Specifically, in the phase III IMpassion130 study (N = 902), those patients who were PD-L1 positive and randomized to first-line nab‑paclitaxel plus atezolizumab experienced a notably longer median OS than with nab‑paclitaxel alone (25 vs 18 months, respectively; HR: 0.71).^37,38 As a result, there has been interest in whether the addition of immunotherapy might be of benefit in the preoperative setting (ie, neoadjuvant therapy). Moving PD-L1–targeted agents earlier in the planned treatment course might be able to decrease or prevent recurrences in our patients with TNBC, and this was the focus of the phase III KEYNOTE-522 study.³⁹

In KEYNOTE‑522, 602 patients with newly diagnosed T1cN1-2 or T2-4N0-2 TNBC were randomized to receive neoadjuvant chemotherapy (carboplatin/paclitaxel in cycles 1-4 and doxorubicin/epirubicin/cyclophosphamide in cycles 5-8 as the backbone) with pembrolizumab or placebo. After surgery, pembrolizumab was continued as adjuvant therapy for an additional 9 cycles. The primary endpoints were pCR and event‑free survival (EFS) to determine whether recurrences are decreased in the long term. PD-L1 expression was assessed using the 22C3 pharmDx IHC assay and measured using the combined positive score (CPS) (number of PD-L1–positive cells divided by number of tumor cells; PD-L1 positive = CPS ≥ 1.)

In a previous report from KEYNOTE-522 at ESMO 2019, there was a significant 13.6% improvement in the pCR rate with the addition of pembrolizumab to chemotherapy (from 51.2% with placebo to 64.8% with pembrolizumab; P = .00055).⁴⁰ An early interim analysis of EFS at the time (15.5 months of follow up) showed a nonsignificant but definite trend toward fewer recurrences in patients who received pembrolizumab in addition to chemotherapy, with 91.3% of patients being free of recurrence compared with 85.3% with chemotherapy only.

At SABCS 2019, the investigators presented results from a KEYNOTE-522 analysis of pCR rates by patient subgroups.³⁹ Many of us were curious whether there were differences by disease stage; for example, did patients with lymph node–positive disease have a larger benefit than those with lymph node–negative disease? Results showed that patients with lymph node involvement did have a larger increase in the rate of pCR compared with patients with lymph node–negative disease. The lymph node–positive patients had an improvement in pCR from 44.1% to 64.8%, an impressive 20.7% difference. By contrast, the lymph node–negative (ie, lower-risk) patients had a 6.3% absolute improvement in pCR, from 58.6% to 64.9%.

Remarkably, improvement in pCR rates occurred regardless of PD-L1 status. Both PD-L1–negative and PD-L1–positive patients experienced improvement in the pCR rate with the addition of pembrolizumab. This is very different than previously seen with the IMpassion130 trial where the benefit with atezolizumab in metastatic TNBC was limited to the PD-L1–positive subgroup. In KEYNOTE-522, patients with a PD-L1 CPS < 1 had similar benefit from pembrolizumab (18.3% improvement in pCR rate) as those with a CPS ≥ 20 (18.5% improvement), whereas those with CPS 1-10 had improvements of approximately 15%. In other words, in the preoperative TNBC setting with an immunotherapy in combination with chemotherapy, PD-L1 does not seem to be a predictive biomarker.

One challenge is that immunotherapy is associated with a unique spectrum of toxicities. In this study, rates of grade 3‑5 toxicity were much higher with pembrolizumab plus chemotherapy compared with chemotherapy alone (12% vs 1%, respectively). Hypothyroidism was the most common immune-mediated AE, affecting 14.9% of patients in the pembrolizumab arm vs 5.7% in the placebo arm. In addition, skin reactions, hyperthyroidism, and adrenal insufficiency were all seen in > 2% of patients receiving pembrolizumab.

Particularly for patients with early-stage breast cancer, we worry about toxicities to therapy, some of which may be lifelong and permanent. For immune checkpoint inhibitors, endocrine toxicity, adrenal insufficiency, hypophysitis, and thyroid toxicity are likely to be permanent and require continued medical intervention. It will be important to weigh the risks and benefits of treatment to decide if this is a regimen that would be suitable for individual patients with triple‑negative disease. In this analysis, the benefit was greater in the lymph node–positive patients and smaller in patients with no lymph node involvement. These findings suggest this regimen should probably be restricted to higher‑risk TNBC, both because of the toxicity and because the benefit is so much greater in the high‑risk group of patients.

Another caveat is that KEYNOTE-522 used a fairly intensive chemotherapy regimen. Along with many physicians, I tend to just give AC-T in the preoperative setting and rarely include carboplatin as part of the backbone. Now, with pembrolizumab added to this intensive chemotherapy regimen, it makes me wonder if the same benefit would be seen without the carboplatin, which would spare patients some of the toxicity from this regimen. Further studies will need to tease these out.

If more mature EFS data from the second interim analysis of KEYNOTE-522 continue to show benefit, this combination may become the standard neoadjuvant regimen for patients with node‑positive TNBC. If this regimen can truly improve rates of pCR and extend EFS, it should definitely be adopted.

Regarding the role of PD-L1, it is unclear whether the lack of difference by PD-L1 expression is due more to the type of testing and the CPS assessment or possibly to PD-L1 being less relevant in this setting. In KEYNOTE-522, which evaluated pembrolizumab, the CPS (per the 22C3 assay) was used to assess PD-L1 expression. By contrast, in IMpassion130, which evaluated atezolizumab, PD-L1 scoring used the SP142 assay and was based on PD-L1–expressing immune cells (IC) as a percentage of the tumor area: IC3 (≥ 10%), IC2 (≥ 5% and < 10%), IC1 (≥ 1% and < 5%), and IC0 (< 1%). It is also difficult to tease out the role of PD-L1 in TNBC in preoperative trials because most of the patients are PD-L1 positive (eg, approximately 80% in KEYNOTE-522).

Phase III NeoTRIPaPDL1: Addition of Atezolizumab to Neoadjuvant Carboplatin/Paclitaxel in TNBC
A different preoperative approach for TNBC was developed by Gianni and colleagues⁴¹ in NeoTRIPaPDL1, a phase III study that looked at the addition of the PD-L1 inhibitor atezolizumab to a preoperative chemotherapy backbone with no anthracycline. As mentioned, in IMpassion130, the addition of atezolizumab to nab-paclitaxel significantly improved both PFS and OS in patients with PD-L1–positive metastatic TNBC.^37,38 In NeoTRIPaPDL1, 280 patients with early, high-risk, or locally advanced TNBC were randomized to receive a chemotherapy backbone of neoadjuvant carboplatin plus nab‑paclitaxel, with or without atezolizumab, for 8 cycles.⁴¹ The primary endpoint was EFS at 5 years after the last patient was randomized, with secondary endpoints including pCR (defined as an absence of invasive cells in breast and lymph nodes). After neoadjuvant therapy, patients received surgery plus 4 cycles of an anthracycline-containing regimen.

Although the EFS data are not yet mature, the pCR data from NeoTRIPaPDL1 were presented at SABCS 2019. In contrast to KEYNOTE‑522, which showed a significant increase in the pCR rate with the addition of pembrolizumab to chemotherapy, NeoTRIPaPDL1 showed no significant improvement in pCR from the addition of atezolizumab to the carboplatin and taxane backbone: The pCR rate was 40.8% in the chemotherapy arm vs 43.5% in the atezolizumab arm, a 2.63% difference. Likewise, no significant differences were seen in analyzed subgroups (PD-L1 positive; PD-L1 negative; early, high risk; and locally advanced).

Many of us are wondering why NeoTRIPaPDL1 showed no benefit from the addition of atezolizumab to chemotherapy whereas KEYNOTE‑522 did show a difference from the addition of pembrolizumab to chemotherapy. There are a couple of possibilities. One is that NeoTRIPaPDL1 did not include neoadjuvant anthracycline, which suggests that giving anthracycline concurrently with immunotherapy may be important for improving outcomes. Indeed, data from the phase II TONIC trial have suggested that, when looking at synergistic activity between chemotherapy agents and immunotherapy agents in metastatic TNBC, anthracycline therapy is the most synergistic with checkpoint inhibition.⁴² That said, platinum was a close second and should also be highly synergistic with checkpoint inhibition. So, it remains unclear what might be driving the difference in pCR rates between these similar approaches. Another reason for the differences may be that there are efficacy differences between PD-1 and PD-L1 antibodies, with PD-1 inhibition potentially being more efficacious due to its ability to inhibit interactions with both PD-L1 and PD-L2. Although it was certainly disappointing to see a lack of improvement in the pCR rate, there could still be an impact on EFS that becomes evident with longer follow up.

More SABCS 2019 Conference Coverage on the CCO Web Site!
Downloadable slideset summaries of all of the studies covered in this activity are available on our Web site.