FDA Approval Summary: Atezolizumab Plus Paclitaxel Protein-bound for the Treatment of Patients with Advanced or Metastatic TNBC Whose Tumors Express PD-L1

In the United States, breast cancer is the most common cancer in women, with more than 260,000 new cases and 40,000 deaths annually (1). Triple-negative breast cancer (TNBC) is characterized immunohistologically by the lack of expression of hormonal receptors [estrogen receptor (ER) and progesterone receptor (PR)], and lack of overexpression and/or amplification of the HER2/NEU gene (2). TNBC accounts for approximately 10%–20% of newly diagnosed breast cancer cases (3). Patients with TNBC have a higher risk of both local and distant recurrence, and metastases are more likely to occur in visceral organs and the brain compared with patients with other types of breast cancers (4).

Because of lack of targets, standard of care for patients with locally advanced or metastatic TNBC has traditionally been systemic chemotherapy (e.g., paclitaxel, capecitabine, eribulin, ixabepilone; ref. 5). Despite these treatments, however, most patients have disease progression and the 5-year overall survival is estimated at only 11%. Locally advanced or metastatic TNBC remains a serious disease with unmet medical needs (1). Atezolizumab (Tecentriq, Genentech, Inc.) is an Fc-engineered, humanized, monoclonal IgG1 kappa antibody that directly binds to programmed death-ligand 1 (PD-L1) and blocks its interactions with the PD-1 and B7.1 receptors. Paclitaxel protein-bound is a microtubule inhibitor. Atezolizumab is approved for other indications including bladder cancer and lung cancer. This article summarizes the results and FDA's review leading to the accelerated approval of atezolizumab in combination with paclitaxel protein-bound for the treatment of adult patients with unresectable locally advanced or metastatic TNBC whose tumors express PD-L1 based on trial IMpassion130 (NCT 02425891).

Previously, atezolizumab had been approved for administration at 1,200 mg every 3 weeks; in IMpassion130, exposures from administration 840 mg every 2 weeks were consistent with the previously established exposure from every 3-week administration. The treatment-emergent anti-drug antibody (ADA) incidence rate was 13% (57 of 434 patients). This incidence rate is lower than seen in clinical trials of patients with non–small cell lung cancer or urothelial carcinoma who received atezolizumab 1,200 mg every 3 weeks (approximately 30%–48%). This lower ADA incidence rate seen in IMpassion130 may potentially be an underestimation due to a high percentage of patients with C_trough above the ADA assay drug tolerance threshold after the first ADA assessment at week 4. Atezolizumab clearance in patients who tested positive for treatment-emergent ADA was 22% higher compared with clearance in patients who tested negative for treatment emergent ADA. There are insufficient numbers of patients in the PD-L1–positive subgroup with ADA (21 of 178 patients) to determine whether development of ADA alters the efficacy or safety of atezolizumab.

IMpassion130 was a randomized, placebo-controlled, double-blinded, two-arm trial in patients with locally advanced or metastatic TNBC who had not received prior chemotherapy for metastatic breast cancer (6). Atezolizumab (840 mg) or placebo was administered via intravenous infusions on days 1 and 15 of every 28-day cycle. Paclitaxel protein-bound (100 mg/m²) was administered via intravenous infusion on days 1, 8, and 15 of every 28-day cycle (3 weeks on/1 week off). The study was initially designed with the coprimary endpoints of progression-free survival (PFS) in the intention-to-treat (ITT) and PD-L1–positive populations. On November 20, 2015, a protocol amendment added overall survival (OS) as a coprimary endpoint and the sample size was increased from 350 to 900 patients to achieve adequate power. The analysis of OS was a hierarchical design with OS in the ITT population needing to reach statistical significance before OS in the PD-L1–positive population could be tested. A total of three analyses of OS were planned (two interim analyses and one final analysis). There were no planned interim analyses for PFS. Secondary endpoints included investigator-assessed objective response rate (ORR), duration of response (DoR), and the patient-reported outcome (PRO) secondary endpoint of time to deterioration in global health status and health-related quality of life. ORR was included in the testing hierarchy. There was no prespecified hypothesis test with type I error allocation for DoR and the PRO endpoints and they are therefore considered exploratory.

Baseline demographics for patients in the ITT and PD-L1 population of IMpassion130 are shown in Table 1. A total of 902 patients were randomized between the two arms (451 to atezolizumab + paclitaxel protein-bound, 451 to placebo + paclitaxel protein-bound). Four (0.4%) of the patients were men. Baseline disease characteristics for patients in the ITT and PD-L1 populations are shown in Table 2. Almost all patients (99.4%) were triple negative according to local laboratory determination as required as per eligibility criteria. Two patients randomized to the placebo + paclitaxel protein-bound arm and 3 patients randomized to the atezolizumab + paclitaxel protein-bound arm did not have triple-negative disease according to the protocol-defined criteria; these patients remained part of the ITT analysis population. Overall, 41% of enrolled patients had tumors that were PD-L1–positive as tested by the Ventana PD-L1 (SP142) Assay and were balanced between the control and experimental arm. Baseline patient and disease characteristics were well balanced in the ITT and the PD-L1–positive populations and were also similar between the two populations

IMpassion130 demonstrated a statistically significant improvement in PFS in the ITT and PD-L1–positive populations. In the ITT population, a HR of 0.79 [95% confidence interval (CI), 0.68–0.92] was observed with an estimated median PFS difference of 1.5 months (7.0 months vs. 5.5 months). In the PD-L1–positive population, an HR of 0.60 (95% CI, 0.48–0.77) was observed with an estimated median PFS difference of 2.6 months (7.4 months vs. 4.8 months; Fig. 1). Other efficacy results are shown in Table 3. Adding atezolizumab to paclitaxel protein-bound in the PD-L1–negative population had no effect on PFS; the HR was 0.94 (95% CI, 0.78–1.13) and no median difference was observed (5.6 months vs. 5.6 months).

The OS results at the first interim analysis (389 events in the ITT population) did not cross the interim boundary (HR < 0.759, tested at α = 0.0065) for statistical significance, with a HR of 0.84 (95% CI, 0.69–1.02, P = 0.084; ref. 6). At this first interim analysis of OS, data were immature with 43% deaths in the ITT population. During the review of the supplemental biologic application, the results of the second interim analysis of OS became available (534 events in the ITT population). At the second interim analysis, the results also did not cross the boundary for statistical significance (HR < 0.818, tested at α = 0.021) with a HR of 0.86 (95% CI, 0.72–1.02; P = 0.077; ref. 7). Because OS in the ITT population that did not reach statistical significance, due to the hierarchical testing plan the OS results (at the second interim analysis) in the PD-L1–positive population are descriptive and are considered exploratory only (HR = 0.71; 95% CI, 0.54–0.93). Follow-up continues for OS and the final OS is planned when approximately 662 events have occurred. In the PD-L1–positive population, the ORR was 53% for the combination of atezolizumab with paclitaxel protein-bound compared with 33% in the control arm with a difference in duration of response of 3.0 months between arms.

No new safety signals were identified in IMpassion130 with the addition of atezolizumab to paclitaxel protein-bound (8). In patients who received atezolizumab + paclitaxel protein-bound, the most common treatment-emergent adverse events (TEAE ≥20%) were alopecia (56%), peripheral neuropathies (47%), fatigue (47%), nausea (46%), diarrhea (33%), anemia (28%), constipation (25%), cough (25%), headache (23%), neutropenia (21%), vomiting (20%), and decreased appetite (20%). The events which occurred at a higher incidence (≥5% difference) in the atezolizumab + paclitaxel protein-bound arm compared with the placebo + paclitaxel protein-bound arm were nausea, cough, neutropenia, pyrexia, and hypothyroidism. The most common grade 3–4 adverse reactions occurring in ≥2% were neutropenia (8%), peripheral neuropathies (9%), neutrophil count decreased (4.6%), fatigue (4%), anemia (2.9%), hypokalemia (2.2%), pneumonia (2.2%), and aspartate aminotransferase increased (2.0%). Adverse reactions leading to discontinuation of atezolizumab in the atezolizumab + paclitaxel protein-bound arm occurred in 6% of patients, with peripheral neuropathy (<1%) as the most common adverse reaction leading to discontinuation. Fatal adverse reactions occurred in 1.3% of patients in the atezolizumab + paclitaxel protein-bound arm; these included septic shock, mucosal inflammation, autoimmune hepatitis, aspiration, pneumonia, pulmonary embolism. Serious adverse reactions occurred in 23% of patients. The most frequent serious adverse reactions were pneumonia (2%), urinary tract infection (1%), dyspnea (1%), and pyrexia (1%). Immune-related adverse reactions requiring systemic corticosteroid therapy occurred in 13% of patients and are consistent with the known safety profile of atezolizumab.

Patient-reported outcomes (PRO) were assessed using the European Organization for Research and Treatment of Cancer (EORTC) Quality of Life Questionnaire Core 30 (QLQ-C30) and other measures. The scales measuring function and global health status/health-related quality of life (GHS/QoL) are transformed onto a 0–100 scoring scale, where higher scores indicate better functioning, whereas for symptom scores, higher scores indicate the symptom is worse (9, 10). In both the ITT and PD-L1–positive populations, completion rates of the EORTC QLQ-C30 at baseline was similar for both treatment arms (>90%). Completion rates for both arms were above 80% through cycle 7 when ≥50% of the ITT and PD-L1–positive populations were still on treatment. PRO endpoints were not prespecified with type I error allocation. Exploratory objectives were to evaluate PRO measures of GHS/QoL, function and disease/treatment-related symptoms associated with both treatment arms. Over the first 12 months of therapy, differences between the arms with respect to GHS/QoL and physical functioning were small. For example, at baseline for patients in the atezolizumab + paclitaxel protein-bound arm, mean physical function was 82.8, the largest decrease over the first 12 months was at cycle 5, where the mean dropped to 76.8. In the placebo arm, baseline mean was 79.4 and the largest decrease was to 75.3 at cycle 11. In both the ITT and PD-L1–positive populations, there did not appear to be large difference for treatment-related symptoms (fatigue, diarrhea, nausea, and vomiting) between the treatment arms.

This is the first FDA approval for patients with locally advanced or metastatic TNBC, and the first FDA approval for immunotherapy in breast cancer (11). IMpassion130 demonstrated a statistically significant improvement in PFS with the addition of atezolizumab to paclitaxel protein-bound in both the ITT and PD-L1–positive population. The safety profile was consistent with the known safety profile of atezolizumab and paclitaxel protein-bound and was acceptable for the intended population. The magnitude of benefit was greater in the PD-L1–positive population and patients with PD-L1–negative tumors did not appear to derive clinical benefit. Therefore, the benefit-risk profile was not deemed appropriate for treatment of patients without PD-L1 expression, and the Ventana PD-L1 (SP142) Assay was contemporaneously approved as a companion diagnostic device essential for the safe and effective use of atezolizumab to select for patients with tumors expressing PD-L1. While there are multiple biomarkers across disease types that have been used to evaluate and potentially demonstrate differential benefit of immunotherapy, PD-L1 was the studied biomarker in this trial and therefore the companion diagnostic biomarker reflected the studied population and results.

Although the IMpassion130 study was conducted in patients with previously untreated locally advanced or metastatic TNBC, given the availability of other standard-of-care options for first-line treatment, potential for benefit regardless of line of therapy, the seriousness of the condition, the acceptable safety profile, and need for additional effective therapies in patients with TNBC, the indication was not restricted to the first-line setting. This will allow for a discussion of benefit/risk between the provider and patient and will also allow those patients who were not newly diagnosed with metastatic disease at the time of approval to discuss the potential to receive this new treatment option later in their treatment course with the caveat that the exact magnitude of benefit in later lines of therapy is not known. The FDA risk–benefit analysis is shown in Table 4.

While PFS was statistically significant in IMpassion130, the median improvement in PFS in the PD-L1 subgroup was modest for an add-on design trial, indicating this improvement may not represent direct clinical benefit. However, the PFS result for the PD-L1 population appeared reasonably likely to predict clinical benefit and thus in the context life-threatening breast cancer was appropriate for accelerated approval. Because of the prespecified statistical analysis plan, OS in the PD-L1 population could not be tested, with appropriate type I error control, Because OS in the ITT population did not reach statistical significance. Results from an ongoing multicenter, randomized, double-blind, placebo-controlled study of atezolizumab or placebo with paclitaxel in patients with previously untreated, inoperable, locally advanced, or metastatic TNBC (IMpassion131, NCT03125902) may provide confirmatory evidence from PFS with supportive OS information to verify the clinical benefit in the same patient population. Alternatively, final OS results from IMpassion130 may also provide confirmatory evidence to verify the clinical benefit.

In summary, atezolizumab plus paclitaxel protein-bound for the treatment of patients with locally advanced or metastatic TNBC whose tumors express PD-L1 (PD-L1 stained tumor-infiltrating IC of any intensity covering ≥ 1% of the tumor area), demonstrates a favorable benefit-risk profile with results reasonably likely to demonstrate clinical benefit. No new safety signals were observed, and the safety profile is acceptable for patients with metastatic TNBC. Because statistical significance was not reached for OS in the ITT population, we cannot adequately determine whether any observed difference in OS in the PD-L1–positive population is due to chance. With the residual uncertainty of the benefit of the addition of atezolizumab to paclitaxel-protein bound, the results from IMpassion130 supported an accelerated approval. Continued approval for this indication may be contingent upon verification of clinical benefit from other ongoing clinical trials.

J. J. Gao is a paid consultant for Worldcare, Inc. No potential conflicts of interest were disclosed by the other authors.

The Editor handling the peer review and decision-making process for this article has no relevant employment associations to disclose.

Conception and design: P. Narayan, J.J. Gao, P.G. Kluetz, R. Pazdur, J.A. Beaver

Development of methodology: P. Narayan, J.J. Gao, R. Pazdur, J.A. Beaver

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Wahby, J.J. Gao, B.L. King-Kallimanis

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P. Narayan, S. Wahby, J.J. Gao, L. Amiri-Kordestani, S. Tang, Y. Xu, J. Liu, W. Fu, P. Song, Y. Gong, S. Ghosh, R. Philip, G.M. Blumenthal, P.G. Kluetz, R. Sridhara, R. Pazdur, J.A. Beaver

Writing, review, and/or revision of the manuscript: P. Narayan, S. Wahby, J.J. Gao, L. Amiri-Kordestani, E. Bloomquist, S. Tang, Y. Xu, J. Liu, W. Fu, P. Song, B.L. King-Kallimanis, S. Kalavar, S. Ghosh, R. Philip, K.B. Goldberg, M.R. Theoret, G.M. Blumenthal, P.G. Kluetz, R. Sridhara, R. Pazdur, J.A. Beaver

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.J. Gao, A. Ibrahim, R. Philip, K.B. Goldberg, R. Pazdur

Study supervision: J.J. Gao, R. Pazdur, J.A. Beaver

Other (regulatory project management): S. Hou