Frontline Pembrolizumab/Chemo Represents New Standard in Metastatic TNBC With PD-L1 CPS ≥10

Hope S. Rugo, MD, discusses the final data from the phase 3 KEYNOTE-355 trial with pembrolizumab plus chemotherapy in patients with metastatic TNBC and the clinical impact on the treatment paradigm.

First-line pembrolizumab (Keytruda) in combination with chemotherapy induced a significant improvement in survival benefit over chemotherapy alone, and could represent a new standard of care for patients with locally recurrent unresectable or metastatic triple-negative breast cancer (TNBC) whose tumors have a PD-L1 combined positive score (CPS) of 10 or higher, according to Hope S. Rugo, MD.

Final results from the phase 3 KEYNOTE-355 trial (NCT02819518), which examined pembrolizumab plus chemotherapy vs placebo plus chemotherapy in patients with metastatic TNBC, were presented during the 2021 ESMO Congress and showed that the combination significantly improved progression-free survival (PFS) and overall survival (OS) over chemotherapy alone—specifically in patients with a PD-L1 CPS of 10 or more.

The median OS in this subset was 23.0 months in the investigative arm vs 16.1 months in the control arm (HR, 0.73; 95% CI, 0.55-0.95; P = .0093). Additionally, the median PFS for these patients was 9.7 months with pembrolizumab vs 5.6 months with placebo (HR, 0.66; 95% CI, 0.50-0.88).

“I do believe this to be a new standard of care. It is critical that patients, staff, and physicians be educated about how to recognize immune-related toxicity as early as possible and implement treatment that makes a difference,” Rugo said. “It is important to keep in mind that we are only benefiting 40% of our patients, so we need more treatments and to encourage enrollment on clinical trials that are looking at combination therapies and new agents to improve outcomes in patients with metastatic TNBC.”

In an interview with OncLive®, Rugo, lead study author, professor in the Department of Medicine (Hematology/Oncology), and director of Breast Oncology and Clinical Trials Education at the University of California San Francisco Helen Diller Family Comprehensive Cancer Center, discusses the final data from the phase 3 KEYNOTE-355 trial with pembrolizumab plus chemotherapy in patients with metastatic TNBC and the clinical impact on the treatment paradigm.

OncLive®: What was the rationale for the KEYNOTE-355 study?

Rugo: KEYNOTE-355 is a randomized phase 3 trial evaluating the impact and outcome of adding pembrolizumab to a menu of chemotherapy [options] for patients with metastatic TNBC in the first-line setting who have relapsed at least 6 months after their treatment in the curative setting, or who have de novo metastatic disease.

The rationale for this trial is based on 2 different aspects. One is that TNBC has a high mutational burden, a higher expression of PD-L1, and a higher presence of tumor-infiltrating lymphocytes. All these factors have been associated with a greater host immune response. They are indicators of the host immune response and have been associated with response to checkpoint inhibitors in other settings, as well as in breast cancer. The second is that limited treatment options [are available] for metastatic TNBC, and the response to treatment and overall survival [OS] is among the worst seen in any subset of [patients with] breast cancer. [As such,] there is a huge unmet need for improved treatment options.

What were the methods utilized in the study?

This is actually really important, as we are seeing the approval of checkpoint inhibitors, particularly pembrolizumab, both for the treatment of [patients with] early-stage TNBC, as well as metastatic disease, based on KEYNOTE-355. [The trial] randomized patients in the first-line, metastatic or de novo TNBC disease setting to receive either pembrolizumab or placebo with a chemotherapy backbone. The randomization was 2:1, and the chemotherapy [options] included nab-paclitaxel [Abraxane], paclitaxel, or gemcitabine and carboplatin.

The trial was designed to have 2 primary end points: PFS and OS in patients with PD-L1–positive disease and in the intent-to-treat [ITT] population. PD-L1 positivity was defined as having a combined positive score [CPS] of 10 or greater or a CPS of 1 or greater, with the alpha expended for the those with a CPS of 10 or greater, making this an enriched patient population for PD-L1. CPS uses an antibody 22C3 and includes the PD-L1 positivity on tumor cells, lymphocytes, and macrophages; it is then divided by the total number of tumor cells and multiplied by 100. [As such, CPS] is a unique test for PD-L1 positivity.

The secondary end points [of the trial] included safety in all treated patients, overall response rate [ORR], duration of response [DOR], and duration of clinical response. It is important to also keep in mind that the alpha was strictly controlled. There was a one-sided [P value of] .025, and that was split among PFS, OS, and ORR. The reason why that is so important is because sometimes we see P values that look significant, but they are not, and it has to do with this statistical design where we have this very specific allocation of alpha for different end points.

What were the key findings from the trial?

The trial randomized 847 patients 2:1 to receive either pembrolizumab or placebo with chemotherapy. The PFS was significantly improved in patients who had PD-L1–positive disease, defined as a CPS of 10 or greater. The absolute difference in PFS was 4.1 months, increasing from 5.6 months, which has been the control PFS across first-line studies evaluating checkpoint inhibitors with chemotherapy for metastatic TNBC. The median PFS in the control population receiving either nab-paclitaxel or paclitaxel was approximately 5.5 months, and this increased to 9.7 months in the PD-L1 CPS of 10 or greater population with the receipt of pembrolizumab. Those [data] have already been presented, with a [statistically significant] hazard ratio of 0.66.

At ESMO, we presented the OS data for the first time; this was significantly increased in patients who receive pembrolizumab. The difference in OS was almost 7 months, going from 16.1 months to 23.0 months, with a hazard ratio of 0.73, or a 27% relative improvement, and a P value of .0093. Something that has also become really popular has been these landmark analyses that ask, what is the percentage of patients who were alive at 18 and 24 months? The differences at 18 and 24 months were very compelling given that the median OS in the control population is less than 2 years, at 16 months. At 24 months, 34% of patients were alive in the control group, but 48.2% were alive in the pembrolizumab group [within the subset of patients with] a CPS of 10 or greater. The P value boundary had to be .01, and this was .0093 here.

Of course, we also looked at the OS in patients with a CPS of 1 or greater, which was the larger population. It is worth mentioning that 38% of patients on the KEYNOTE-355 trial has a CPS of 10 or greater. However, if we look at the patients who had a CPS of 1 or greater, we do not enrich quite as much; it is about 75% of the patient population. In that group, although there was a numerical difference of 1.6 months, overall, there was no [significant] difference in OS; that was also reflected in the ITT population where there was also no difference in OS.

If we look at the subgroup forest plots and try to figure out whether there are subgroups of patients who benefit relatively more or less [from pembrolizumab], the major difference is a CPS of 10 or greater, so PD-L1 positivity. Interestingly only 65 patients had a disease-free interval of less than 12 months, and they do not look like they benefit [from this approach]. However, this was a small group, so it is impossible to know what that impact is. This should not be a reason not to treat patients because the trial is in no way powered to look at the difference in those who have a short disease-free interval. Otherwise, all the patients seemed to benefit, and there were no compelling differences in the subgroups.

When updating the PFS, it showed the same benefit and the same hazard ratio. Also shown was the ORR, which was improved by 12% with the addition of pembrolizumab in the PD-L1 CPS of 10 or greater group, and the disease control rate was also improved. DOR is important for our patients, and the median DOR in the CPS of 10 or greater population was 7.3 months for placebo and 12.8 months for pembrolizumab combined with chemotherapy. In the landmark analysis at 12 months, 38% of patients had disease control in the control group [compared with] approximately 56% in the pembrolizumab group, which is quite remarkable.

It is interesting that in KEYNOTE-355, the duration of treatment with pembrolizumab was 2 years, [and patients] continued chemotherapy until progression; that is including stopping the treatment at 2 years based on the protocol design.

What was the safety profile of the study regimen?

Adverse effects [AEs] are important in terms of thinking about adding any new drug to standard chemotherapy. It is really encouraging that no immune-mediated AEs that led to death were reported. Immune-mediated AEs led to discontinuation of in 2.8% [of those on] the pembrolizumab group [vs 0% with] the placebo group.

Grade 3 to 5 immune-related toxicities were observed in 5.3% vs 0% [of patients], respectively. The most common immune-related toxicities were thyroid related. Pneumonitis was the next most common [toxicity to be reported], at 2.5%. This is all encouraging, and the data support pembrolizumab in combination with chemotherapy as a new standard of care in patients with TNBC in the first-line setting whose tumors express PD-L1 with a CPS of 10 or greater.

Is there a take-home message that you want to impart regarding this research?

It is great to see a survival advantage. It is interesting that the survival difference was similar to what was shown in the phase 3 IMpassion130 trial [NCT02425891 with atezolizumab (Tecentriq)]. IMpassion130 had a different statistical design, which was hierarchical, and because they did not show a survival benefit in the ITT population, they were not allowed to assess formal statistical differences in the PD-L1–positive population. Forty-one percent of patients were PD-L1 positive with SP142, and 38% with CPS. We looked at the biomarker-evaluable population in IMpassion130 and looked at CPS and SP142, and interestingly, although most overlap, there is a group that do not overlap; a little more than 10% are SP142 positive, but not positive for 22C3. Moreover, some are just positive for 22C3 CPS.

The unfortunate hierarchical design and negative results from the phase 3 IMpassion131 trial [NCT0315902] led to Roche withdrawing the approval [for atezolizumab plus chemotherapy in adult patients with unresectable locally advanced or metastatic TNBC] in the United States. This was because they did not have confirmatory data, which was required by the FDA, and they are waiting for the neoadjuvant, very large NSABP B-59/GBG 96-GeparDouze trial [NCT03281954].

This begs the question that if a patient is only positive for SP142, what would you do? Currently, I would treat them with pembrolizumab in the United States, since we cannot use atezolizumab, because I think those patients may benefit and this has not been looked at. In the rest of the world, however, atezolizumab is still approved, so that is not really a consideration. Generally, we are going to be using CPS only to evaluate patients who can be treated in the metastatic setting.


  1. Cortés J, Cescon DW, Rugo HS, et al. KEYNOTE-355: final results from a randomized, double-blind phase 3 study of first-line pembrolizumab + chemotherapy vs placebo + chemotherapy for metastatic TNBC. Ann Oncol. 2021;32(suppl 5):S1289-S1290. doi:10.1016/j.annonc.2021.08.2089