Clinical outcome of patients with isolated central nervous system progression on first-line pertuzumab and trastuzumab treatment for HER2-positive metastatic breast cancer in a real-life cohort

More than 10% of HER2-positive metastatic breast cancer (mBC) will develop Central Nervous System (CNS) metastases as first and isolated site of relapse on trastuzumab and pertuzumab first-line therapy. However, few clinical data are available to guide the best strategy in this setting. Patients experiencing isolated CNS progression on trastuzumab and pertuzumab first-line therapy were retrospectively identified from the French Epidemiological Strategy and Medical Economics (ESME) real-life database between 2008 and 2016. Among 995 patients treated with first-line trastuzumab and pertuzumab for HER2-positive mBC, 132 patients (13%) experienced isolated CNS progression with a median time of 12 months after mBC diagnosis. Twelves patients did not receive any treatment and were excluded from the analysis. Among the 120 patients considered, 76 (63%) received CNS-directed local therapy, 73 (60%) continued trastuzumab and pertuzumab, whereas 47 (39%) started another systemic treatment. After a median follow-up of 21 months, there was no difference in progression-free survival for patient who continued trastuzumab–pertuzumab or switched to another systemic treatment. In multivariate analysis, trastuzumab–pertuzumab continuation was associated with longer OS (HR 0,28 IC 95%: 0,14–0,54 p < 0,001). mOS was not reached (95% 37.6-NE) and was 23.2 months (95% CI 15.5–53.6) in patients who continued trastuzumab and pertuzumab therapy and in patients who switched for another systemic therapy, respectively. In this real-life cohort, trastuzumab–pertuzumab continuation after local treatment for isolated CNS progression did not negatively impact PFS and OS. Prospective trials and assessment of new strategies are warranted in this specific situation.


Background
Breast cancer (BC) is the most frequent female cancer worldwide, and remains the leading cause of death by cancer in women [1]. It is also one of the most frequently primary providing central nervous system (CNS) metastases with lung carcinoma [2]. While CNS metastases compromise survival [3,4], they are also associated with neurological symptoms and disability considerably affecting patient's quality of life [5]. Their prevention and management remain an important challenge for the oncologists.
Several risk factors of CNS dissemination or progression have already been identified in BC including younger age [6], higher grade [7], number of metastatic sites, number of previous treatment lines and relapses, endocrine receptor (ER) negativity and HER2 overexpression [3,8].

3
Indeed, more than one third of patients with HER2positive metastatic BC (mBC) will develop CNS metastases [9] which is more than other BC subtypes [10]. In the CLEOPATRA phase III trial, which demonstrated overall survival (OS) and progression-free survival (PFS) benefit of pertuzumab in addition to trastuzumab and docetaxel as first-line therapy for HER2-positive mBC, more than 10% of patients, irrespective of treatment's group, experienced CNS metastases as first site of relapse without systemic progression [11,12]. After the diagnosis of CNS metastases, current treatment strategies include CNS-directed therapy as radiotherapy [13,14] and surgery [15] for selected patients [16]. Of interest, systemic HER2-targeted therapies have also proven their efficiency on HER2-positive CNS metastases [17][18][19].
Whereas the switch of systemic therapy is the rule in case of systemic extra cranial progression, there is currently no evidence that the modification of the systemic treatment could improve outcome in case of isolated CNS progression. A pooled analysis of tucatinib study shows that maintenance of the same systemic treatment should be considered in case of isolated brain progression after local control therapy [20]. ASCO and ESMO guidelines currently recommend the use of CNS-directed therapy whenever possible and an individual case-by-case consideration of change of systemic treatment. In HER2-positive mBC, although both recommend the use of trastuzumab capecitabine and tucatinib association after progression under trastuzumab therapy [21,22], limited clinical data and guidelines are available regarding isolated brain progression, without systemic progression under trastuzumab and pertuzumab combination, [20,23].
This retrospective real-life cohort describes the characteristics, management and outcome of patients experiencing isolated CNS progression under pertuzumab and trastuzumab first-line therapy for HER2-positive mBC.

Study design and cohort population
In this retrospective study, patients were selected among the retrospective French Epidemiological Strategy and Medical Economics (ESME) mBC database. The ESME mBC database (NCT03275311), composed of 18 French Comprehensive Cancer Centers, was launched by UNI-CANCER in 2014 and built from existing information systems, treatment databases and patient's electronic medical records with homogenous on-site collected information. Among the 22,463 patients registered in the ESME mBC cohort between the first of January, 2008 and December the 31, 2016, patients experiencing isolated CNS progression (either CNS metastases occurrence or progression if already present at diagnosis) under pertuzumab and trastuzumab as first-line systemic therapy for HER2-positive mBC were identified.

Data collection
Patients' characteristics (age at diagnosis, sex, medical history of early BC and treatment), tumors' characteristics (type of metastases, number of metastatic sites, ER status, median disease-free interval from early breast cancer if concerned) and treatment's strategy (type of systemic and CNS-targeted therapy) were reported from patient's electronic medical records, patient hospitalization records and pharmacy records. Given the multicentric nature of the data collection, all coding procedures were predefined by the data manager and all data were centralized within each center using a shared anonymous format.
The present analysis was approved by an independent ethics committee (Comité De Protection Des Personnes Sud-Est II-2015-79). No formal dedicated informed consent was required but all patients had approved the re-use of their electronically recorded data. In compliance with French regulations, the ESME mBC database was authorized by the French data protection authority (Registration ID 1,704,113 and authorization N°DE-2013-117). Moreover, in compliance with the applicable European regulations, a complementary authorization was obtained on 14-Oct-2019 regarding the ESME research Data Warehouse.

Primary and secondary end-points
After description of patients' and tumors' characteristics, our primary endpoint was to estimate the OS in the entire population and in two subgroups. The first subgroup included patient who continued trastuzumab and pertuzumab after local treatment of isolated CNS progression called "continuation group" and the second subgroup included patients who switched for another systemic treatment and will be called "switched group" in the following article.
The secondary end-points were the systemic PFS and the second CNS -PFS estimations in the overall population and in the two treatment subgroups.
PFS and OS were estimated and calculated from the first CNS progression. Systemic PFS included extra-cerebral progression with or without CNS progression. First and second CNS PFS corresponded, respectively, to the time between metastatic diagnosis and the first metastatic CNS occurrence or progression (if CNS metastasis were already present at diagnosis) and the time between the first metastatic CNS occurrence or progression to the subsequent CNS progression event (Fig. 1).

Statistical analysis
The characteristics of the participants were described with median and range for quantitative data, and frequencies and percentages for qualitative data. Comparisons between groups (continuation versus switched group) were performed with Kruskall-Wallis test with Welsh's t-test if necessary for continuous variables and Pearson's χ 2 tests or Fisher tests if small class with Yates-corrected χ 2 test if necessary for binary or categorical variables.
The Kaplan-Meier method was used to calculate PFS, OS and CNS PFS. To assess the association between tumor characteristics and treatment strategy on OS and PFS, a univariate Cox model was used to calculate hazard ratio (HR) for several variables of interest. The variables considered in the univariate analysis were the following: age, grade, ER status, histological subtype, medical history of BC, adjuvant therapy, type of metastasis at diagnosis, CNS metastasis at diagnosis of metastatic relapse (if patients already have brain metastasis at diagnosis of metastatic disease or not), number of metastatic sites, endocrine therapy, disease-free interval, CNS progression-free interval, treatment after isolated CNS progression (either appearance of CNS metastasis if not present at the diagnosis of mBC or isolated progression of brain metastasis if already present at mBC diagnosis). A multivariate analysis was then performed by a backward approach checking for potential cofounding effects at each step with a subset of these variables after testing in univariate models. Treatment after first CNS progression including "CNS surgery after CNS progression", "stereotactic radiotherapy after CNS progression" and "Whole Brain Radiation Therapy (WBRT) after CNS progression" were considered as time-dependent variables for cox univariate and multivariate model. To include the variable "CNS surgery after CNS progression" in the multivariate model we performed a cox Regression with Firth's Penalized Likelihood to estimate HR as no event was observed in this subgroup of treatment. Statistical analyses were conducted with R software. Values of p < 0.05 were considered to indicate significance in all statistical tests.

Patients' and tumors' characteristics in the whole population
Between January 2008 and December 2016, 4071 patients among 22 463 patients registered in ESME cohort were diagnosed with HER2-positive mBC and 995 were treated with first-line trastuzumab and pertuzumab. Among them, 132 patients (13%) experienced isolated CNS progression (either the appearance of CNS metastasis or the progression of already present CNS metastasis, as the only progression site) under this regimen and were included in the study. Median time to first CNS progression from diagnostic of metastatic relapse in the overall population (N = 132) was 12.0 months (range 2.0; 50.0). Among them, 12 patients (9%) did not receive any systemic treatment after CNS isolated progression and were excluded from the analysis (Fig. 2). Among them, 7 (58%) patients did not receive any treatment, only 1 patient (8%) patient received only CNS surgery or stereotactic radiation therapy, and finally 3 (25%) patients wholebrain radiation therapy. Of note, these patients were older at diagnostic and have a shorter disease-free interval from early breast cancer (supplementary material Table 1).
A total of 120 patients were included in this study: 52 (43%) had previously been treated for a localized BC and 38 of them (73%) had received trastuzumab in the adjuvant setting. For these patients, median time to metastatic recurrence was 41 months (range 19; 381).

Patients' and tumors' characteristics in subgroup
Patients in the continuation subgroup (N = 73) had a median age at diagnosis of mBC of 48.4 years (range 28;79). Twenty-six patients (36%) had previously been treated for a localized BC and among them 17 (65%) received trastuzumab in the adjuvant setting. In the metastatic setting, most of them (N = 47, 64%) had visceral metastasis. Only 10 patients (14%) had CNS metastases at the time of mBC diagnosis. Forty-three patients (59%) presented ER-positive tumors and 16 of them received endocrine therapy in addition with trastuzumab and pertuzumab.
In the switched subgroup (N = 47), median age at diagnosis of mBC was 51 years old (range 27;78). More than half of them had previously been treated for a localized BC (N = 26, 55%) and among them 21 patients (81%) received trastuzumab as adjuvant therapy. In the metastatic setting, 30 patients (64%) had visceral metastasis and 10 patients (21%) already had CNS metastasis at mBC diagnosis. Seventeen patients (36%) had ER-positive tumors and among them, 8 patients received additional endocrine therapy.
Comparison between the two groups (Table 1) showed that a significantly higher number of patients had a medical history of early BC and received adjuvant trastuzumab in the switched group. Moreover, patients of the continuation group more frequently had ER-positive tumors and more often benefited from CNS surgery after CNS progression.

Therapeutic strategy description in the whole population (n = 120)
After CNS progression under trastuzumab and pertuzumab, 66 patients (55%) received systemic treatment in combination with CNS-directed therapy and 54 patients (45%) received systemic treatment alone (Table 1).
With regards to systemic PFS, no variable was statistically significant in the univariate analysis. Multivariate cox model analysis was performed including following variables: age at metastatic diagnosis (< 50 and versus > 70 years old), initial CNS metastasis (yes versus no), ER status (positive versus negative), first CNS progression-free interval (less than 9 months, between 9 and 18 months and more than 18 months), trastuzumab/pertuzumab continuation (yes versus no), CNS surgery after progression (yes versus no), stereotactic radiotherapy after CNS progression (yes versus no) and whole CNS radiotherapy after CNS progression (yes versus no). No variable was significantly associated with systemic PFS in the multivariate model. However, we observed a trend toward longer systemic PFS in favor of age < 50 years old at metastatic diagnosis (versus > 70 years old) and no initial CNS metastasis at diagnosis (versus CNS metastasis at diagnosis). However, neither trastuzumab and pertuzumab continuation nor CNS surgery after CNS progression was associated with prolonged systemic PFS (supplementary material table 3).

Discussion
In this real-life cohort of patients with HER2-positive mBC receiving first-line trastuzumab and pertuzumab, CNS isolated progression incidence was 13%, which is consistent with the CLEOPATRA trial results [11]. After CNS progression, trastuzumab and pertuzumab combination was maintained for 60% of patients according to ASCO and ESMO guidelines [22,24,25]. All patients switching for another systemic treatment received another HER2-targeted therapy. Trastuzumab and pertuzumab continuation was associated with prolonged OS in a multivariate analysis, but there were no differences in PFS and time to second CNS progression between the two groups.
Rational to keep the same systemic treatment after CNS metastases control derive mostly from the hypothesis that CNS progression may be in part the consequence of anti-HER2 antibodies having less efficacy on CNS disease [26]. As a consequence, a localized CNS progression does not imply widespread resistance to the current systemic treatment. Still, whereas intact blood brain barrier can decrease the delivery of systemic therapy in the brain, it has been shown that it can be disrupted in CNS metastasis neovascularization, leading to the concept of "Blood Tumor Barrier" with higher permeability to anti-cancer agents [27]. Indeed, randomized studies have shown that OS of patients with CNS metastases from HER2-positive mBC is improved with the best HER2 directed systemic therapy, irrespective of its theoretical brain diffusion [11,18,28]. Another hypothesis for differential efficacy of chemotherapy/targeted therapy on CNS metastasis and systemic disease may derive from clonal specificity of CNS metastasis. Indeed, some preclinical studies have shown that trastuzumab can penetrate experimental CNS metastasis as efficiently as extra-cerebral localization, but shows decreased activity on CNS metastases [29,30].
In our retrospective study, trastuzumab and pertuzumab continuation did not seem to impact negatively CNS PFS or systemic PFS, with only a small difference in favor of the continuation subgroup (7.5 months versus 6.7 months) which was not statistically significant in a multivariate cox 1 3 model analysis. These results support the hypothesis that indeed, an isolated CNS progression is not associated with widespread resistance to ongoing systemic therapy, which can therefore be kept unchanged after local CNS treatment. This strategy has also been shown as efficient in latter lines of treatment from subgroup analysis of the KAMILLA [19] and HER2CLIMB [31] studies. In the KAMILLA study, a large phase III trial of T-DM1 for patients with HER2positive mBC having progressed after at least one line of trastuzumab based treatment, 67 patients with isolated CNS progression on T-DM1 were treated with local treatment and allowed to continue T-DM1 within the study. Median PFS from CNS progression for this subgroup was 6.2 months, which compared favorably to PFS from inclusion of the whole population of this large cohort (7.7 months for the 1604 patients without CNS progression at inclusion). The HER2CLIMB study reported that tucatinib in association with trastuzumab and capecitabine improved survival in heavily pretreated HER2-positive mBC [28]. In case of isolated CNS progression while on study, patients were eligible to continue study-assigned therapy after local treatment. The benefit from such strategy was reported for a small subgroup of 21 patients. Interestingly, PFS from CNS local treatment was improved from 3.1 month to 7.6 months with the addition of tucatinib to the capecitabine/trastuzumab combination (HR, 0.33; 95% CI 0.13 to 0.85; p = 0.02), a HR that is similar to the one observed for the primary endpoint on the whole population [28,31]. These results are also consistent with the hypothesis that isolated CNS evolution does not imply systemic resistance to the ongoing systemic treatment.
The main limitation of our study is its retrospective and observational nature. Indeed, just over 50% of our study population received CNS-directed therapy in addition to systemic treatment. This could reflect the fact that some of these CNS metastases were asymptomatic at time of diagnosis and, as a consequence, the oncologist settled for active surveillance. Notably, only 9 patients already received CNS-directed therapy for CNS metastasis at diagnosis and were quite evenly distributed between our two groups, probably not influencing the observed results (data not showed). Importantly, a statistically significant higher number of patients had CNS surgery in the continuation subgroup which could be in favor of smaller number of CNS metastases in these patients. Moreover, the same proportion of patients received whole CNS radiation therapy in both subgroups (34 and 37%), suggesting that one third of patients had multiple and/or large unresectable brain metastasis which may require such local therapy [21,22]. Of note, 12 patients were not included as they did not receive any systemic therapy and do not allow to answer the main study objective concerning the strategy of systemic therapy in case of isolated CNS progression. Among them 7 did not received any treatment and 5 received only targeted therapy, mainly whole-brain radiation therapy. We cannot exclude these patients may have clinical complications and impaired general condition not allowing for further systemic therapy and therefore have worse outcome. However, mOS and mPFS did not differed from those of the studied population when we included these 12 patients in our analysis. Nevertheless, missing data from this retrospective analysis cannot be ruled out and information on the brain metastasis clinical presentation is unfortunately missing in our retrospective study. In any case, these results are consistent with a less aggressive or symptomatic CNS progression in the "continuation" cohort and could explain in part the longer survival for this subgroup even though we tried to take into account the main prognostic factors in the multivariate analysis. Moreover, as we miss clear recommendation in this specific clinical situation, therapeutic decision may differ between oncologists or cancer center practices. However, our retrospective study included patients from 18 French Comprehensive Cancer Centers and reduce the risk of bias.
Development of new and more efficient, targeted treatments may change the way patients with HER2-positive mBC and CNS metastasis are treated. Among them, small molecules and tyrosine kinase inhibitors could be of particular interest for CNS metastasis. In lung carcinoma, tyrosine kinase inhibitors have drastically changed the prognostic of patients with CNS metastasis and the use of CNS-targeted therapy can be frequently avoided, or at least, differed [32]. With regard to HER2-positive mBC, the development of second generation tyrosine kinase inhibitors, particularly neratinib [33,34] and tucatinib is of high interest. In patients with CNS metastasis prior to study entry, tucatinib added to trastuzumab and capecitabine significantly reduced the risk of CNS progression or death by 68% (HR = 0,32) and the risk of death by 42% (HR = 0,58) [31]. Furthermore, new brain progression was also significantly delayed in the whole population: median time to new brain lesion or death was 13 months for the trastuzumab-capecitabine arm, versus not reached with the addition of tucatinib (HR 0.52, 95% CI 0.33-0.82). These results suggest that early addition of new molecules, shown to be active on brain metastasis could be helpful for patients with high risk of new CNS progression, which is the case for patients having already progressed in the brain. Indeed, clinical trials are currently designed to address this question.

Conclusion
In this large real-life study, 13% of patients receiving trastuzumab and pertuzumab therapy in the first-line setting for HER2-positive mBC had isolated CNS progression. Supporting current recommendations, continuation of trastuzumab and pertuzumab was at least as efficient as modification of systemic treatment after CNS-directed therapy. Nevertheless, PFS remains low and better systemic treatments are needed to prevent further CNS progression. Prospective studies are currently being designed to evaluate new strategies in this specific situation.