Practice Patterns and Outcomes of Novel Targeted Agents for the Treatment of ERBB2-Positive Metastatic Breast Cancer

Key Points

Question  What are the real-world survival outcomes associated with use of pertuzumab and trastuzumab emtansine (T-DM1) for ERBB2-positive metastatic breast cancer?

Findings  In this cohort study of 795 women in Ontario, Canada, median survival with pertuzumab (43 months) and T-DM1 (15 months) were substantially shorter than outcomes reported in previous pivotal trials. The reduced survival in the pertuzumab group may be explained by the older age of patients in routine practice; among the T-DM1 cohort, median survival was shorter for patients with prior receipt of pertuzumab than in the pertuzumab-naive subgroup.

Meaning  Median survival in the real-world appears inferior to results of pivotal trials, and differences in outcome likely reflect differences in patient population and previous lines of therapy in routine practice.

Importance  Clinical trials have shown that the addition of pertuzumab to trastuzumab-based chemotherapy for first-line treatment of ERBB2-positive metastatic breast cancer is associated with considerable improvement in overall survival (OS). In the second-line setting, trastuzumab emtansine (T-DM1) improves OS compared with capecitabine/lapatinib in patients previously treated with trastuzumab-based chemotherapy. However, there are few data describing long-term real-world outcomes with these agents.

Objective  To describe practice patterns and outcomes associated with pertuzumab and T-DM1 in routine clinical practice.

Design, Setting, and Participants  This population-based retrospective cohort study used the Ontario Cancer Registry linked to electronic treatment databases to identify all patients treated with pertuzumab and T-DM1 following reimbursement approval in Ontario, Canada, which has a single-payer public health system. Participants included women with stage IV ERBB2-positive metastatic breast cancer receiving treatment with pertuzumab for first-line metastatic indication from December 2013 through December 2017, and those treated with T-DM1 from May 2014 through December 2017. Pertuzumab and T-DM1 cohorts were analyzed separately. Data were analyzed December 2019 to December 2020.

Exposures  Treatment with pertuzumab or T-DM1.

Main Outcomes and Measures  The primary outcome was OS, determined using the Kaplan-Meier method. Factors associated with OS were identified using a Cox proportional hazard model.

Results  The median (interquartile range [IQR]) age of the 795 women who received pertuzumab and 506 women who received T-DM1 was 57 (49-67) and 56 (48-66) years, respectively. Among the cohort of patients who received pertuzumab, the median (IQR) OS and time on treatment was 43 (16.2-unavailable) and 14 (6.0-26.2) months, respectively. In the T-DM1 cohort, the proportion of pertuzumab-naive patients decreased over time from 68 of 91 [74.7%] in 2014 to 16 of 89 [18.0%] in 2017 (P < .001). The median (IQR) OS and time on treatment was 15 (6.7-27.7) and 4 (1.4-9.0) months, respectively. Median OS was shorter for patients with prior pertuzumab treatment than in the pertuzumab-naive subgroup (12 vs 19 months; adjusted hazard ratio, 0.70; 95% CI, 0.55-0.89; P = .004).

Conclusions and Relevance  In this population-based cohort study, the survival of patients treated with pertuzumab and T-DM1 in routine practice appeared inferior to results from pivotal clinical trials. Differences in outcome likely reflect differences in patient population and previous lines of therapy in routine practice. Further work is needed to understand the effectiveness of T-DM1 after pertuzumab exposure.

Treatment options for women with ERBB2-positive metastatic breast cancer have changed rapidly over the past 2 decades. The seminal trial reported by Slamon and colleagues¹ in 2001 established trastuzumab as standard of care in this patient population. In the past decade, there has been further interest in next-generation anti-ERBB2 treatment strategies, including pertuzumab and trastuzumab emtansine (T-DM1). In the practice-changing CLEOPATRA trial,^2-4 first-line treatment of patients with ERBB2-positive metastatic breast cancer with the combination of pertuzumab, trastuzumab, and taxane chemotherapy substantially improved progression-free survival (PFS) and overall survival (OS) compared with trastuzumab and chemotherapy alone. Median OS was 57 months in the experimental arm and 41 months in the control group (hazard ratio [HR], 0.68 [95% CI, 0.56-0.84]). In the second-line setting, the EMILIA trial^5,6 randomized patients with unresectable or metastatic ERBB2-positive breast cancer previously treated with trastuzumab and taxane chemotherapy to receive treatment with T-DM1 or a control arm of capecitabine plus lapatinib. Median OS was improved with use of T-DM1 (30 vs 26 months; HR, 0.75 [95% CI, 0.64-0.88]). Notably, patients in the EMILIA trial did not receive prior pertuzumab.

Since the initial publication of these landmark trials, they have been incorporated into clinical practice guidelines⁷ and become standard-of-care treatment options. It is not known how these novel and costly treatments are being deployed in routine practice and whether long-term outcomes are as expected based on results of pivotal clinical trials. To address these gaps, we performed a population-based study to describe use and outcomes associated with first-line pertuzumab and second-line T-DM1 in patients with ERBB2-positive metastatic breast cancer in the Canadian province of Ontario.

This was a population-based retrospective cohort study to describe practice patterns and outcomes associated with use of pertuzumab and T-DM1 in patients with ERBB2-positive metastatic breast cancer treated in the Canadian province of Ontario from 2013 to 2017. Ontario has a population of 14.5 million people (approximately 40% of Canada’s total population) and has a single-payer universal health insurance system. This study was approved by the Research Ethics Board of Queen’s University in Kingston, Ontario, Canada, and the requirement for informed consent was waived owing to use of deidentified data from administrative databases. This study was designed, analyzed, and reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.⁸

The study cohort was identified using the New Drug Funding Program (NDFP) and linked to additional databases, including the Ontario Cancer Registry (OCR). The NDFP provides provincial reimbursement for all pertuzumab, trastuzumab, and T-DM1 in the metastatic setting, as well as trastuzumab in the neoadjuvant setting. The study population included women with stage IV ERBB2-positive breast cancer receiving treatment in Ontario, Canada with pertuzumab for first-line metastatic indication from December 2013 (pertuzumab funding start date by Cancer Care Ontario) to December 2017, as well as those treated with T-DM1 from May 2014 (T-DM1 funding start date) to December 2017. Patients included in the pertuzumab cohort had not received prior trastuzumab in the metastatic setting. Patients who previously received neoadjuvant trastuzumab were included if the interval between completion of adjuvant treatment and initiation of pertuzumab was at least 12 months, in line with the CLEOPATRA study inclusion criteria.²

Databases housed at the Division of Cancer Care and Epidemiology at the Cancer Research Institute at Queen’s University in Kingston, Ontario, Canada were used and included the OCR and NDFP. Variables available in administrative databases are listed in eTable 4 in the Supplement. Full data dictionaries are available online.^9,10 The OCR is a population-based cancer registry containing diagnostic and demographic information on at least 98% of all incident cases of cancer in Ontario and also captures vital status and cause of death.¹¹ Multiple administrative health databases are linked to the OCR, including the NDFP database. Dates and duration of treatment with trastuzumab, pertuzumab, T-DM1, and other chemotherapy agents were identified through the NDFP, including prior systemic therapy in the neoadjuvant setting, defined as any previous treatment with neoadjuvant intent occurring within 1 year of previous primary surgery for breast cancer. Comorbidities were classified using the modified Charlson Comorbidity Index and based on hospital admission records for the 5-year period prior to diagnosis; these records have been validated to be complete.¹²

Descriptive statistics were used to assess the distribution and frequency of baseline variables. Survival was calculated using the Kaplan-Meier method and measured from the date of treatment initiation with pertuzumab or T-DM1. Multivariable Cox proportional hazards models were used to evaluate factors associated with OS. Proportional hazard assumptions were tested and met. Included variables were chosen a priori based on clinical relevance and availability. Results were considered statistically significant at a 2-sided P < .05. Analyses were conducted using SAS for Windows, version 9.4 (SAS Institute).

Between December 2013 and December 2017, 795 patients with metastatic breast cancer were treated with first-line pertuzumab, and 506 were treated with T-DM1. Two patients had missing values for date of death and were excluded from the survival analysis performed for the remaining 793 patients. Patient characteristics are summarized in Table 1. Median (interquartile range [IQR]) age was 57 (49-67) years in the pertuzumab cohort and 56 (48-66) years for those treated with T-DM1. Prior breast surgery was performed in 318 (40.1%) and 276 (54.5%) patients in the pertuzumab and T-DM1 cohorts, respectively, with a minority having received prior neoadjuvant chemotherapy (217 [27.3%] in the pertuzumab cohort and 202 [39.9%] in the T-DM1 cohort) and trastuzumab (179 [22.5%] in the pertuzumab cohort and 161 [31.8%] in the T-DM1 cohort).

The 795 patients treated with pertuzumab in the first-line metastatic setting received a median (IQR) of 19 (9-37) cycles, with a median (IQR) time on treatment of 14 (6-26) months. All patients received concurrent trastuzumab. Median (IQR) OS was 43 (16.2-unavailable) months. Most patients initially received concurrent chemotherapy with paclitaxel (567 [71.3%]), while 166 (20.8%) patients received docetaxel. Older age (hazard ratio [HR], 1.30; 95% CI, 1.18-1.43; P < .001) was associated with shorter OS, while time from diagnosis to pertuzumab initiation of fewer than 3 months (likely a surrogate for de novo metastatic disease) was associated with longer survival (HR, 0.55; 95% CI, 0.37-0.82; P = .001) (Table 2). To identify a population similar to that of the CLEOPATRA trial, an expected survival was derived based on Cox model by setting a median age at pertuzumab initiation of 54 years and average-level remaining covariables, which are summarized in Table 2. Expected survival based on age-adjusted model was 53 months. With adjustment for median time from diagnosis of 30 months and remainder of covariables at average level, expected survival was unchanged from outcomes seen in the overall population (43 months).

Most of the 506 patients who received T-DM1 had prior treatment with trastuzumab in the metastatic setting either alone (n = 175 [34.6%]) or in combination with pertuzumab (n = 323 [63.8%]) (Table 1). The proportion of pertuzumab-naive patients decreased over time (68 of 91 [74.7%] patients in 2014 and 16 of 89 [18.0%] patients in 2017; P < .001). For those who received prior pertuzumab, the median (IQR) time from last pertuzumab treatment to T-DM1 initiation was 24 (12-41) days. Patients received a median (IQR) of 6 (3-13) cycles of treatment with T-DM1, with more cycles observed in patients who had not received prior pertuzumab (median [range] cycles, 11 [1-73]) compared with those with prior pertuzumab treatment (median [range] cycles, 5 [1-73]) (eTable 1 in the Supplement). Median OS from start of T-DM1 treatment was 15 months and was shorter in patients who received prior pertuzumab (12 months) than for those without prior exposure (19 months) (eTable 1 in the Supplement). Median time on treatment was 4 months overall and was also shorter in patients with prior receipt of pertuzumab (3 months) compared with the pertuzumab-naive subgroup (8 months). Lack of prior pertuzumab treatment (HR, 0.70; 95% CI, 0.55-0.89; P = .004) and not having undergone primary breast surgery (mastectomy or partial mastectomy) within the previous 10 years (HR, 0.72; 95% CI, 0.54-0.97; P = .03), a potential surrogate for de novo metastatic disease, were associated with longer OS in multivariable analysis (Table 3).

In this population-based cohort study, the delivery and outcomes of novel ERBB2-directed therapies for metastatic breast cancer in the general population were assessed. We have found substantial uptake of pertuzumab and T-DM1 in routine clinical practice. Median survival of patients treated with these agents in routine practice is shorter than what was observed in pivotal randomized clinical trials^2-6; this is not unexpected given the known differences between patients in the general population and those in clinical trials.¹³

The concept of the efficacy-effectiveness gap is increasingly recognized as an important limitation in the real-world application of clinical trial results, with growing evidence that outcomes observed for patients selected to participate in trials (efficacy) differ from those observed with the same drug in real-world circumstances (effectiveness).¹³ Previous works by this group and others have identified similar gaps between efficacy and effectiveness for new treatments in pancreas cancer (FOLFIRINOX [folinic acid, fluorouracil, irinotecan, oxaliplatin]),¹⁴ gastric cancer (trastuzumab),¹⁵ prostate cancer (docetaxel),¹⁶ and renal cell carcinoma (sunitinib).¹⁷ It is likely that the efficacy-effectiveness gap applies to the majority of new cancer therapies. In a recent analysis of Medicare delivery of cancer drugs approved by the US Food and Drug Administration,¹⁸ there was striking evidence of an efficacy-effectiveness gap. Contributors to this gap may include baseline patient and disease characteristics, receipt of prior treatments, and treatment modifications.^13,18

The pertuzumab cohort population in this study was similar to that of the CLEOPATRA trial, in which patients did not receive prior trastuzumab for metastatic disease or within 12 months in the neoadjuvant setting.^2,3 In the present study, median OS with pertuzumab for first-line metastatic treatment (43 months) was shorter than outcomes seen in the experimental arm of CLEOPATRA (57 months). Moreover, median time on treatment with pertuzumab, a potential surrogate for PFS, was 14 months in the present population, which is shorter than the median PFS of 19 months seen in CLEOPATRA. These results indicate that patients with ERBB2-positive metastatic breast cancer in the general population have shorter survival than similar patients in clinical trials and may experience shorter periods of disease control on treatment. It is important to note that the number of pertuzumab cycles received in the present study was not prescribed; instead, patients continued on treatment until progression or experiencing toxic effects, as per standard-of-care practice based on CLEOPATRA (eTable 3 in the Supplement). However, given the low rate of treatment discontinuation owing to toxic effects in CLEOPATRA, time on treatment and number of treatment cycles received are likely surrogates for PFS. A comparative effectiveness study to further explore the real-world benefit of pertuzumab in this setting is currently being conducted by this group. In keeping with results from CLEOPATRA,⁴ shorter time from diagnosis to pertuzumab initiation was significantly associated with longer OS in the present study. This is likely indicative of better outcomes observed in trastuzumab-naive patients with de novo metastatic disease compared with those with recurrence, consistent with observations from CLEOPATRA where the subgroup of patients who previously received adjuvant trastuzumab had shorter OS.⁴ We also performed adjusted models to approximate a population closer to that of CLEOPATRA by adjusting median age from 57 years to 54 years, with OS (53 months) approaching trial outcomes. This exploratory analysis suggests that the observed efficacy-effectiveness gap in outcomes of treatment with pertuzumab is largely driven by differences in patient age between routine practice and the clinical trial.

The T-DM1 cohort in this study differed from that of the EMILIA study^5,6 because two-thirds of the present population received pertuzumab in the first-line metastatic setting. At the time of design and enrollment of EMILIA, pertuzumab was not approved for use in the metastatic setting. Patients received prior first-line treatment with trastuzumab and taxane chemotherapy alone, and the benefit of second-line T-DM1 in contemporary practice is unclear. One retrospective cohort study¹⁹ of 78 patients treated with T-DM1 who received prior pertuzumab showed a median duration on therapy of 4 months but did not assess long-term survival outcomes. In the present study, median OS in the overall population was 15 months, substantially shorter than observed in the experimental arm of EMILIA (30 months).^5,6 It is notable that among patients without prior treatment with pertuzumab, median OS was longer (19 months) compared with those with prior pertuzumab treatment (12 months). This suggests that real-world patients who have been pretreated with pertuzumab may be less responsive to subsequent anti-ERBB2 therapy. None of the pertuzumab-naive patients received pertuzumab subsequent to T-DM1, eliminating the possibility that these findings are attributable to posttreatment crossover. Of note, the subgroup that did not receive pertuzumab in the first line had a longer time from diagnosis to T-DM1 (4.2 years) than those with prior pertuzumab treatment (2.8 years). This is likely a reflection that these patients commenced ERBB2-targeted therapy prior to the approval and funding of pertuzumab. However, the longer median time on T-DM1 treatment observed in this pertuzumab-naive group (8 vs 3 months) points to a difference in treatment benefit rather than survival time bias and may indicate that real-world patients receiving first-line pertuzumab derive less benefit from T-DM1 than those in the EMILIA trial. Finally, it is possible that patients in the subgroup with prior pertuzumab therapy who received T-DM1 shortly after its approval represent a subset of the population with more aggressive disease. However, substantial differences in OS between the pertuzumab pretreated and naive subgroups were maintained over time in a sensitivity analysis of median OS by year of treatment initiation (eTable 2 in the Supplement). This suggests that results from EMILIA may not be fully generalizable to current practice where standard of care includes treatment with first-line pertuzumab because patients with prior receipt of pertuzumab had shorter time on treatment and survival compared with the pertuzumab-naive subgroup.

The strengths of this study include its large sample size and the fact that the cohort includes all patients treated with publicly funded treatment in the general population of Ontario, Canada. However, the results must be considered in light of methodologic limitations. The conclusions drawn from this study are limited by the lack of comparative data and rely on comparisons of real-world outcomes to those seen in randomized clinical trials; however, such comparisons are necessary to evaluate the efficacy-effectiveness gap and describe the generalizability of clinical trial results to routine practice. Accordingly, we cannot quantify the extent to which treatment with pertuzumab and T-DM1 improve outcomes in this real-world population. Furthermore, PFS cannot be accurately determined from population-based data, and time on treatment was used as a surrogate. Other reasons for treatment discontinuation, such as toxic effects or patient preference, were not captured. Finally, it is important to note that these findings do not invalidate the results of large randomized clinical trials such as CLEOPATRA and EMILIA. Rather, they reflect outcomes associated with use of novel drugs in real-word circumstances. Whether the shorter OS observed in the present real-word population is because of differences in patient and disease factors or delivery of care in routine practice can be difficult to ascertain, although the exploratory analyses conducted in the present study suggest difference in age and prior treatments may be the most important associations. The retrospective nature of this study limits quantification of the effect of potentially confounding variables. Additional data such as performance status, site of metastases, and burden of metastatic disease were also unavailable in the study data sets (eTable 4 in the Supplement).

In summary, results of this population-based cohort study show that there has been substantial uptake of pertuzumab and T-DM1 treatment in patients with advanced ERBB2-positive breast cancer since publication of the CLEOPATRA and EMILIA trials. Outcomes in the general population show inferior OS compared with results from these pivotal trials. Differences in outcome likely reflect differences in patient population and previous lines of therapy in routine practice. Further work is needed to understand the effectiveness of T-DM1 after pertuzumab exposure.

Accepted for Publication: April 26, 2021.

Published Online: July 8, 2021. doi:10.1001/jamaoncol.2021.2140

Correction: This article was corrected on September 16, 2021, to fix an error in the Results section of the Abstract.

Corresponding Author: Josee-Lyne Ethier, MD, MSc, Department of Oncology, School of Medicine, Queen’s University, 25 King St W, Kingston ON, K7K 7B2 Canada (josee-lyne.ethier@kingstonhsc.ca).

Author Contributions: Dr Ethier had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Ethier, Desautels, Robinson, Booth.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Ethier, Booth.

Critical revision of the manuscript for important intellectual content: Desautels, Robinson, Amir, Kong, Booth.

Statistical analysis: Kong.

Obtained funding: Ethier.

Conflict of Interest Disclosures: Dr Ethier reported personal fees from AstraZeneca, GlaxoSmithKline, Genomic Health, and Merck outside the submitted work. Dr Desautels reported personal fees from Novartis Canada and Amgen Canada outside the submitted work. Dr Robinson reported personal fees from Roche, AstraZeneca, Merck, and Bristol Myers Squibb outside the submitted work. Dr Amir reported personal fees from Exact Sciences, Novartis, Genentech/Roche, Sandoz, Apobiologix, and Agendia outside the submitted work. Dr Booth is supported as a Canada Research Chair in Population Cancer Care. No other disclosures were reported.

Funding/Support: This study was supported by a Canadian Institutes of Health Research Operating Grant.

Role of the Funder/Sponsor: The Canadian Institutes of Health Research had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: Parts of this material are based on data and information compiled and provided by Cancer Care Ontario. The analyses, conclusions, opinions, and statements expressed herein are solely those of the authors and do not reflect those of the funding or data sources; no endorsement is intended or should be inferred.