Phase II study of intrathecal administration of trastuzumab in patients with HER2-positive breast cancer with leptomeningeal metastasis

Abstract

Background

Patients with HER2-positive breast cancer (HER2 + BC) develop central nervous system metastases twice as often as patients with luminal HER2-negative breast cancer. Leptomeningeal progression results in a drastically altered prognosis with limited therapeutic options. This phase II study was conducted to assess the efficacy of intrathecal (IT) trastuzumab in HER2 + BC patients with leptomeningeal metastasis (LM), based on a phase I dose-escalation study that had determined the recommended weekly dose of 150 mg for phase II.

Methods

Eligible patients received weekly administrations of 150 mg of IT trastuzumab. The primary endpoint was clinical neurologic progression-free survival (LM-related-PFS) after 8 weeks. Overall survival (OS), toxicities, and quality of life (QoL) were secondary endpoints.

Results

Among the 19 enrolled patients, 16 (84%) had concomitant brain metastases, 15 of them had received prior radiotherapy to the brain. All patients had received at least one line of systemic anti-HER2 therapy. After 8 weeks, 14 patients (74%) were free of neurological progression. The median LM-related-PFS was 5.9 months and the median OS was 7.9 months. According to the QLQ-C30 and BN20 scales, the global health-related QoL status seemed preserved and no toxicity above grade 3 was observed.

Conclusions

Conducting studies on patients with LM poses significant challenges and ethical dilemmas inherent to this population. Despite some limits, this phase II study’s findings in terms of clinical neurologic response and QoL’s control confirms weekly administration of 150 mg of IT trastuzumab as a valuable option for HER + BC patients with LM and support the interest for further investigations.

Key Points.

1. Rising incidence of meningeal carcinomatosis in HER2 + breast cancer patients

2. First phase II study for intrathecal injection of trastuzumab in this indication

3. Fourteen patients (74%) free from neurologic progression at 2 months; quality of life stabilized

Importance of the Study.

Systemic control and overall survival of HER2 + BC has greatly improved in the 2 last decades. Still, leptomeningeal progression results in a drastically altered prognosis with limited valid therapeutic options. A phase I dose-escalation study demonstrated the safety and feasibility of administering intrathecal trastuzumab to HER2 + BC patients with leptomeningeal metastasis (LM), with a recommended weekly dose of 150 mg. The aim of this subsequent phase II study was to assess the efficacy of this strategy. Consistently with phase I, no toxicity above grade 3 was observed. Clinical neurologic response and quality of life’s (QoL) control after 8 weeks of treatment confirmed weekly administration of 150 mg of intrathecal (IT) trastuzumab as a valuable option for HER + BC patients with LM. These results support the interest for larger confirmatory trials to demonstrate the survival’s and QoL’s benefits from this targeted IT therapy.

Above half of patients with metastatic HER2 + BC eventually develop central nervous system metastases,^1–5 among which 11%–20% develop leptomeningeal metastasis (LM).^6–9 This represents a critical turning point in breast cancer evolution, resulting in neurological disabilities, reduced quality of life (QoL), and a poor prognosis (median survival of 3–4 months).⁶ Despite this burden, only few studies provided evidence upon optimal treatment for LM (radiotherapy, systemic and/or intrathecal [IT] chemotherapy).^10,11 DEPOSEIN (NCT01645839) showed a benefit of IT treatment in patients with LM from breast cancer with tumor cells in the cerebrospinal fluid (CSF) or linear disease and this was also true for subgroup of HER2 patients.⁸ In a retrospective cohort of 254 patients with LM from solid cancers, the administration of either IT or systemic pharmacotherapy was associated with a better outcome in presence of tumor cells in the CSF (type I LM).⁹

The advent of trastuzumab¹² and subsequent anti-HER2 therapies greatly improved systemic disease control of HER2 + BC.^13–18 However, as large molecules poorly cross the blood leptomeningeal barrier, CSF levels of trastuzumab in patients given intravenous perfusions are largely below the cytotoxic threshold.¹² This may contribute to the increasing incidence of LM in patients with HER2 + BC, which presents a significant challenge to clinicians.

In this context, the administration of the drug directly into the CSF appears to be a valuable strategy to overcome the obstacle of the blood leptomeningeal barrier. A recent meta-analysis reported data from patients treated with IT trastuzumab either alone or in combination with systemic therapy. No severe toxicity was documented in 88% of cases and clinical improvement or stabilization was reported in 69% of cases.¹³ Consistent with these findings, a phase I dose-escalation study demonstrated the safety and feasibility of administering IT trastuzumab to HER2 + BC patients with LM,¹ with a recommended dose of 150 mg of IT trastuzumab per week. These results altogether were encouraging enough to conduct the phase II: its primary aim was to assess the clinical efficacy of IT trastuzumab for HER2 + BC patients with LM, through the clinical neurologic progression-free survival after 8 weeks (LM-related PFS) of treatment. Secondary endpoints included the evaluations of overall survival (OS), global PFS, QoL evolution, toxicity, and associations of FCGR3A and FCGR2A polymorphisms with outcome.

Material and Methods

We performed a multicenter phase I–II open-label study (NCT01373710). HER2 + BC patients with LM were enrolled between October 2013 and April 2017 in 7 French hospitals.

Patients

Eligibility criteria were like the phase I.¹ Briefly, we included patients ≥18 years with HER2 + BC and LM diagnosed by CSF cytology and/or clinical symptoms of LM and LM evidence on magnetic resonance imaging (MRI). LM could be newly diagnosed and naïve of treatment or previously known and treated. Patients with symptomatic brain metastases could be included if surgery and/or radiotherapy had been performed at least 3 weeks before inclusion.

Treatment

IT Trastuzumab (150 mg) was administered weekly by lumbar puncture, through an ventricular device or through an indwelling IT drug delivery device.¹ Systemic corticosteroids were introduced at least 3 days before IT treatment (≥20 mg/day prednisolone) and 25 mg of IT hydrocortisone hemisuccinate was administered just before IT trastuzumab. All systemic treatments except lapatinib could be initiated, maintained, or modified during the study. Patients with prior use of lapatinib (with a wash out period ≥2 weeks between last lapatinib use and the fist IT trastuzumab) were allowed. Indeed, the blood–brain barrier may be crossed more easily by small-size molecules like anti-HER2 tyrosine kinase inhibitors such as lapatinib, than larger molecules like the most chemotherapies and monoclonal antibodies. Thus, both neurological efficiency and toxicities may have been biased if lapatinib had been concomitantly used with IT trastuzumab. Tucatinib and neratinib were not yet available by the time of the study.

Evaluation

The primary endpoint was assessed after 8 weeks of treatment. Therapy was continued until patients developed clinical neurologic progression, refused treatment, or developed unacceptable toxicity. Baseline clinical and biological (CSF included) evaluations were reported up to 15 days before inclusion and then reported weekly. Cranial and spinal MRI were performed up to 28 days before inclusion, then monthly (Supplementary Appendix 1). Clinical neurologic assessment included level of consciousness, neurological symptoms (according to Common Terminology Criteria for Adverse Events version 4.0 (CTCAE v4.0), cranial nerve disorders, language or speech disorder, muscle weakness, superficial sensory disorder, ataxia, sphincter disorders, and Mini Mental State Evaluation (MMSE).

Two European Organisation for Research and Treatment of Cancer (EORTC) QoL questionnaires were weekly collected: the QLQ C30 (generic questionnaire for cancer patients), and the QLQ BN20 (specific questionnaire for brain cancer patients).^14,15,18 Higher scores in global health status and functional scales reflect better QoL. In contrast, higher scores in symptom scales indicate poorer QoL. The summary QLQ C30 score was calculated as follows¹⁴: Physical Functioning + Role Functioning + Social Functioning + Emotional Functioning + Cognitive Functioning + 100 − Fatigue + 100 − Pain + 100 − Nausea_Vomiting + 100 − Dyspnea + 100 − Sleeping Disturbances + 100 − Appetite Loss + 100 − Constipation + 100 − Diarrhea)/13.

Toxicities were also weekly estimated according to CTCAE v4.0.¹⁶

Response

Clinical neurological response was defined as (1) at least one improvement criteria persistent on 2 consecutive neurological assessments as compared to preinclusion evaluation among: a disappearance or a one grade-minimal decrease of at least one neurological impairment or symptom or a minimal increase of 3 points on MMSE, (2) without any worsening criteria among: the emergence or a one grade-minimal increase of any neurological impairment or symptom or a 4-point minimal decrease of MMSE, or any corticotherapy dosage increase.

Clinical neurologic progression was defined as any appearance or worsening of a deficit (cranial pair deficit, focal motor deficit, sensory neuropathy, and sphincter disorder) severe enough to interfere with activities of daily living (eating, washing, dressing, and moving), or higher neurological disorder with MMSE decreasing by ≥3 points. Those symptoms needed to persist for more than a week to represent a progression and had to be related to LM.^17,18 Clinical neurologic PFS was the period between the date of initiation of IT trastuzumab and the date of clinical neurological progression’s diagnosis or death.

Radiological response was defined according to classification RECIST Version 1.1 on brain and spine MRI. Biological response was defined by a disappearance of tumor cells at CSF cytology, when cytology at inclusion was positive.

Pharmacogenomics

The proteins encoded by the FCGR3A and FCGR2A genes have a role in the antibody-dependent cell-mediated cytotoxicity (ADCC) which may be one mode of action of trastuzumab. This ADCC effect is initiated when the Fc portion of trastuzumab binds to the FCGR on an immune effector cells. The resulting bridge between the trastuzumab-bound tumor cell and the effector cell triggers the release of pro-apoptotic factors, which causes trastuzumab-bound tumor cell death.^19,20

Taking into account this ADCC-mediated antitumoral effect of trastuzumab, FCGR polymorphisms were also studied. The procedures of DNA extraction and genotyping are described in Supplementary Appendix 2.

Statistics

The primary endpoint of this study was the clinical neurologic PFS after 8 weeks of treatment (LM-related PFS). The hypothesis tested (Fleming’s plan in 1 step, alpha = 5%, beta = 5%, p0 = 10%, p1 = 40%, power 95%) retains a minimum efficacy threshold of 10% in terms of survival without neurological progression at 8 weeks.

Nineteen patients were required for this phase II study, taking in account 4 patients from the phase I study who had been treated at the dose of 150 mg.¹ Beyond 4 patients free from clinical neurological progression after 8 weeks of treatment, IT trastuzumab would be considered as effective enough to warrant further investigation.

Secondary endpoints were to describe a cohort of HER2 + BC with LM including QoL scales, to assess the toxicity of IT trastuzumab, and to report on QoL evolution. Results were reported as follows: number of patients (percentage) for qualitative data, and median (min–max) for quantitative data. LM-related PFS and OS were estimated by the Kaplan–Meier method. Aiming at exploring predictive biomarkers, we performed a univariate analysis, using the log rank test to assess differences in terms of treatment efficacy according to FCGR polymorphisms. P-values < .05 were deemed to be significant.

Regulation

All patients provided written informed consent, and local ethics committee approval was obtained (CPP approval no. 100438 and ANSM approval no. 100377-77). The study was conducted in accordance with good clinical practice guidelines and the Declaration of Helsinki.

Results

Patients and Treatment Characteristics

Nineteen patients were enrolled: their baseline characteristics are presented in Table 1. The median time between diagnosis of HER2 + BC and the diagnosis of LM was 4.4 years (range 2.7–15.1 years). The median time between diagnosis of LM and inclusion in the study was 24 days (range 1–135 days). Five patients on 14 evaluated (36%) had a positive cytology in CSF at diagnosis. Among 16 patients (84%) who had concomitant brain metastases, 15 of them had received prior radiotherapy (9 in toto and 6 focal irradiations). All patients had already received a systemic anti-HER2 therapy (trastuzumab, pertuzumab, trastuzumab emtansine, and/or lapatinib) and 10 patients (53%) had received at least 2 lines of these treatments. Concomitant therapies during the study included systemic chemotherapy for 8 patients (42%), endocrine therapy for one patient (5%), systemic anti-HER2 for 3 patients (16%), and concomitant IT chemotherapy for one patient (5%) (Table 1, Supplementary Table S1). Six patients (32%) did not receive any other anticancer treatment during the study. Nine patients (47%) received the treatment through an IT drug delivery device, 8 patients (42%) through an Ommaya reservoir, and 2 patients (11%) by lumbar puncture.

Table 1.

Patient and Cancer Characteristics at Baseline

	N (%) or median (min–max) N = 19
Patient
Age (years)	51.2 (34.8–70.7)
Baseline ECOG PS
PS 0	4 (21%)
PS 1	7 (37%)
PS 2	5 (26%)
PS 3	3 (16%)
Primary tumor
Histological type of primary tumor
Invasive carcinoma of NST	18 (95%)
Invasive lobular carcinoma	1 (5%)
Hormone receptor expression
ER ≥ 10%	9 (47%)
PR ≥ 10%	7 (37%)
Both ER and PR < 10%	9 (47%)
HER2 overexpression	19 (100%)
Metastatic disease
Interval between initial diagnosis and LM (years)	4.4 (2.7–15.1)
Interval between LM diagnosis and inclusion (days)	24 (1–135)
Coexisting metastasis at enrollment (extrameningeal)
Bone metastases	7 (37%)
Visceral metastases	11 (58%)
Brain metastases	16 (84%)
Number of metastatic sites at enrollment
<3	11 (58%)
>3	8 (42%)
Previous treatment from onset of the metastatic disease
Previous systemic treatments, number of lines
Endocrine therapy	0 (0–3)
Chemotherapy	2 (1–7)
Anti-HER2 therapy	2 (1-3)
Previous brain radiation therapy	15 (79%)
Whole brain	9 (47%)
Focal	6 (32%)
Previous IT chemotherapy	6 (32%)
Liposomal cytarabine	4 (21%)
Methotrexate	2 (11%)
Thiotepa	1 (5%)
Concomitant treatments other than IT trastuzumab
Concomitant systemic treatments	13 (68%)
Endocrine therapy	1 (5%)
Chemotherapy	8 (42%)
Anti-HER2 therapy	3 (16%)
Othera	1 (5%)
Concomitant IT chemotherapyb	1 (5%)
Laboratory signs of LM in the cerebrospinal fluid at enrollment
Carcinoma cells present/suspected	5/14 (36%)
Increased level of proteins	6/14 43%)
Radiological evidence of LM at enrollment	19 (100%)

Abbreviations: ECOG PS, eastern cooperative oncology group performance status; ER, estrogen receptor; NST, no special type; PR, progesterone receptor.

^aBevacizumab.

^bThiotepa.

Thirteen patients (68%) received at least the full 8-dose course of IT trastuzumab scheduled in the trial (Figure 1). Early death (before the second week of treatment) occurred for 2 patients (11%): the first had a very high metastatic burden (multiple cerebral lesions, liver, bones, pleural, peritoneum, and cutaneous metastases), and was heavily pretreated (3 lines of endocrine therapy, 3 lines of systemic chemotherapy, 2 lines of systemic anti-HER2, and even 2 lines of IT chemotherapy). The second also had a high metastatic burden with multiple cerebral and pulmonary metastases and a deteriorated general condition prior to her enrollment (ECOG PS of 3 at inclusion). Four patients (21%) remained on weekly IT trastuzumab treatment beyond 8 weeks (total duration between 17 and 32 weeks).

Fig. 1.

Patient flow diagram. Despite completing less than the 8 scheduled IT trastuzumab administrations, the 6 patients mentioned above were included in further data and analyses, so 19 patients were actually enrolled.

Response

Considering the main endpoint, 14 patients (74%) were free of neurological progressive disease at 8 weeks including 4 patients (21%) with responsive and 10 patients (53%) with stable disease. The 5 remaining patients had neurological progression, 2 of whom died prior to 8 weeks (Table 2). The median LM-related PFS was 5.9 months and the median OS was 7.9 months (Figure 2).

Table 2.

Number of Doses Administered and Neurological Response After 8 Weeks of IT Trastuzumab

Pt. no.	No. of doses administered	Clinical neurological response	Radiological neurological response	Early termination reason
1	17	RD	SD
2	1	PD	n/a	Death
3	7	SD	n/a	8th administration omitted
4	25	SD	SD
5	32	SD	SD
6	8	PD	SD
7	8	SD	RDa
8	6	RD	n/a	Meningeal progressionb
9	8	SD	PD
10	8	PD	RDa
11	8	RD	SD
12	6	SD	PD	Meningeal progressionb
13	4	SD	n/a	Cerebral progression
14	23	SD	SD
15	8	SD	SD
16	8	SD	SD
17	8	RD	SD
18	8	PD	SD
19	1	PD	n/a	Death

Abbreviations: n/a, not applicable; PD, progressive disease; RD, responsive disease; SD, stable disease. In light gray, patients without clinical neurological progression after 8 weeks of treatment.

^aResponsive disease during the radiological assessment corresponded to partial response, according to the RECIST criteria for these 2 patients.

^bBiological and/or radiological meningeal progression without clinical progression.

Fig. 2.

Clinical neurological PFS (A) and OS (B) according to Kaplan–Meier method.

If we consider only the 13 patients with 8 or more injections, there were 10 patients (61.5%) free of clinical neurological progression after 8 weeks which also met the minimal efficacy threshold (10%).

Concerning MRI evaluation, among the 14 patients free from clinical neurological progression at 8 weeks, 1 patient had a partial radiological response (7%), 8 patients had stable imaging (57%), and 2 patients had radiological progression (14%). Among the 5 patients with a positive CSF cytology at inclusion, only 2 patients also had CSF cytology available at week 8 which was the necessary condition to evaluate the biological response. Among them, one patient had biological response (patient no. 1) and the other one remained with positive CSF cytology (patient no. 15). Considering the 9 patients with initial negative CSF cytology, one had a positive cytology at week 8, that is, biological progression.

Quality of Life

The QLQ-C30 and QLQ-BN20 scores at inclusion are depicted in Tables 3 and 4. Globally, QoL at inclusion was altered in comparison with the reference values for metastatic breast cancer patients provided by the EORTC (Supplementary Table S2).²⁰

Table 3.

Quality of Life Related to Functioning and Global QoL at Week 1, Week 5, and Week 9, Assessed by the EORTC QLQ-C30^14,15,17

	Week 1a Median (min–max)	Week 5b Median (min–max)	Week 9c Median (min–max)
Functioning
Physical functioning	40 (0–100)	40 (0–100)	7 (0–40)
Role functioning	67 (0–100)	33 (0–100)	33 (0–67)
Emotional functioning	50 (17–83)	33 (0–83)	8 (0–33)
Cognitive functioning	33 (0–100)	17 (0–100)	0 (0–33)
Social functioning	33 (0–100)	42 (0–100)	33 (0–33)
Global
Global health status and health-related QoL	51 (0–84)	58 (42–83)	63 (33–83)
Summary scored	61 (42–74)	64 (37–73)	58 (53–67)

Higher scores reflect better QoL.

^a  N = 15: 3 with progressive disease, 12 with stable or responsive disease.

^b  N = 15: 2 with progressive disease, 13 with stable or responsive disease.

^c  N = 5: 0 with progressive disease, 5 with stable or responsive disease.

^dThe summary score is calculated as follows: Physical Functioning + Role Functioning + Social Functioning + Emotional Functioning + Cognitive Functioning + 100 − Fatigue + 100 − Pain + 100 − Nausea_Vomiting + 100 − Dyspnea + 100 − Sleeping Disturbances + 100 − Appetite Loss + 100 − Constipation + 100 − Diarrhea)/13.

Table 4.

Quality of Life Related to Symptoms at Week 1, Week 5, and Week 9, Assessed by the EORTC QLQ-C30 and BN20^14,15,18

	Week 1a Median (min–max)	Week 5b Median (min–max)	Week 9c Median (min–max)
QLQ-C30 symptom scores
Fatigue	67 (0–100)	39 (0–100)	33 (0–67)
Nausea and vomiting	0 (0–83)	0 (0–83)	0 (0–0)
Pain	33 (0–100)	17 (0–83)	17 (0–17)
Dyspnea	0 (0–100)	0 (0–100)	0 (0–33)
Insomnia	33 (0–100)	33 (0–100)	33 (0–100)
Appetite loss	0 (0–100)	0 (0–100)	0 (0–0)
Constipation	33 (0–100)	0 (0–100)	0 (0–67)
Diarrhea	0 (0–67)	0 (0–67)	0 (0–33)
Financial difficulties	67 (0–100)	0 (0–100)	0 (0–100)
QLQ-BN20 symptom scores
Future uncertainty	63 (25–100)	50 (25–100)	31 (25–44)
Visual impairment	33 (25–100)	42 (25–100)	25 (25–67)
Communication deficit	25 (25–92)	33 (25–100)	25 (25–75)
Motor dysfunction	58 (25–75)	33 (25–100)	33 (25–58)
Headache	50 (25–100)	50 (25–75)	50 (25–50)
Seizures	25 (25–25)	25 (25–25)	25 (25–25)
Drowsiness	50 (25–100)	50 (25–100)	25 (25–50)
Alopecia	25 (25–100)	25 (25–100)	25 (25–75)
Pruritus	25 (25–75)	25 (25–100)	25 (25–50)
Lower limb weakness	50 (25–100)	50 (25–100)	25 (25–75)
Urinary incontinence	25 (25–75)	25 (25–75)	25 (25–25)

Higher scores reflect poorer QoL.

^a  N = 15: 3 with progressive disease, 12 with stable or responsive disease.

^b  N = 15: 2 with progressive disease, 13 with stable or responsive disease.

^c  N = 5: 0 with progressive disease, 5 with stable or responsive disease.

Main QLQ C30 symptoms present in this study were fatigue, pain, and sleeping disorder, with fatigue median score twice as much as the median score in recurrent/metastatic breast cancer or stage III/IV breast cancer patients of EORTC reference cohort.

Functioning dimensions were decreased by a third for patients of HIT study, especially for cognitive, social, and physical functioning scores. These lower scores reflect the heavy burden of leptomeningeal metastases, as compared to other metastatic localizations (Supplementary Table S2).

Among the 13 patients still in the study, only 5 patients filled questionnaires at 8 weeks (patients nos. 1, 4, 5, 7, and 14). The median score of global health status and health-related QoL improved during treatment (from 51 at week 1, n = 15, to 63 after 8 weeks, n = 5) (Table 3). All median scores related to symptoms were improved or steady after 8 weeks of treatment (Table 4, Supplementary Figures S2 and S3). But coherently with natural history of LM, functional scales median scores tended to decrease, except from social functioning median score that remained steady.

FCGR Genotypes

FCGR polymorphisms may affect ADCC-mediated antitumoral effect of trastuzumab.^19,20 Blood samples from 15 patients were available and tested. The distribution of FCGR3A and FCGR2A genotypes is summarized in Supplementary Table S3. Using the log rank test, no association was found between FCGR3A and FCGR2A genotypes and the LM-related PFS.

Toxicity

All toxicities observed during the study are listed in Supplementary Table S4. Severe toxicities possibly related to the treatment included: headache, seizure, and allergic reaction (1 grade 3 [5%] for each). No grade 4 or 5 toxicity according to CTCAE v4.0 was observed. No serious adverse events were reported.

Discussion

Owing to better systemic control of metastatic HER2 + BC and more effective LM diagnostic tools, the incidence of LM may continue to increase in this population. These patients, generally heavily pretreated, suffer from severely impaired QoL, and have a poor prognosis. Limited trial data exist to guide the choice of LM treatment. Thus, current regimens mainly rely on expert opinion (except from liposomal cytarabine).^8,21

With 14 patients (74%) free of neurological progressive disease after 8 weeks of IT trastuzumab, the primary endpoint of this phase II study, consisting in a minimal efficacy threshold of 10%, was met.

Neurological symptoms reported as potential toxicities may either be due to chemical meningitis (as a side effect of IT trastuzumab) or result from the LM itself. General and hematological disorders observed may also be related to concomitant systemic treatment and systemic disease burden.

These findings on clinical neurologic response, along with its safety and QoL maintenance, are consistent with the phase I results and support weekly administration of 150 mg of IT trastuzumab as a valid therapeutic option for HER2 + BC patients with LM.

Only one other phase I/II study has been conducted by Raizer et al. (NCT01325207) on the safety and efficacy of trastuzumab, with encouraging results (available on https://clinicaltrials.gov/ct2/show/results/NCT01325207).²² Their results have not yet been published in journal with peer review and are not strictly comparable to ours as eligibility criteria and IT trastuzumab regimen differed as well as their primary endpoint, which combined CSF cytology, radiographic, and clinical function evaluations. However, consistently with their results^22,23 and those from the recent meta-analysis,¹³ IT trastuzumab treatment appeared safe and efficient enough in our phase II study to pursue its evaluation.

Taking in account the short life expectancy of LM patients, and the lack of robustness of response determination before recent RANO²⁴ and EANO-ESMO²¹ recommendations, a clinically assessed primary endpoint seemed the most relevant to us, along with special attention paid to QoL. Despite deterioration of functioning abilities, symptoms tended to be reduced and global QoL was maintained after 8 weeks of treatment. Importantly, median OS was twice as high in our cohort as in the literature without dedicated IT treatment.^6–8 However, it was not our primary endpoint as this result could be biased by other metastatic locations independently of the efficiency of IT trastuzumab and by potential improvement of anti-HER2 therapies. Neurological-PFS was greater in our study than in the DEPOSEIN study (randomized trial testing IT liposomal cytarabine for newly diagnosed LM from breast cancer—all HR and HER2 subtypes): 5.9 months in HIT versus 3.9 months in the DEPOSEIN-IT arm and 2.2 months in the DEPOSEIN-control arm.⁸

Retrospective studies did not uniformly conclude on a significant correlation between higher affinity polymorphisms of FCGR3A or FCGR2A (FCGR2A 131 H/H and/or FCGR3A 158 V/V) and the activity of trastuzumab.^20–27 The association of FCGR genotypes with the neurological clinical response was also studied. The small size of our population set and the low prevailing FCGR3A and FCGR2A polymorphisms did not allow us to show a significant correlation with IT trastuzumab efficacy (as assessed by LM-related PFS) by univariate analysis.

Pharmacokinetics of trastuzumab after its IT administration have also been studied during this phase II study and separately published.²⁸

Limits may be noticed in this study. Exclusion criteria were few but imply that its results may not apply to patients with the most severe manifestations of LM, that is, with PS 4 and/or life expectancy < 2 months, and/or inability to express consent. Besides, concomitant systemic anticancer treatments could be a bias. Indeed, the most of patients had a change of systemic line of treatment. However, those lines were the third ones or more, which presumed of less efficiency especially on the central nervous system than previous ones and for one-third of patients, the change of systemic treatment consisted in a de-escalation. In addition, the systemic treatment was stopped for 6 patients and 4 of them were free of clinical neurological progression. Besides, CSF analyzes were not available for all patients at inclusion and at 8 weeks. CSF collection would have been preferable not only for the present study but for further exploratory analyses as well (eg, pharmacokinetics, cell-free circulating tumor DNA monitoring). Plus, at the time of study, RANO/EANO/ESMO evaluation criteria,^21–35 were not available and radiological response was assessed with RECIST criteria which could be less sensitive and more subjective.

The modality of treatment administration also requires further exploration: Ommaya reservoir has been suggested to ensure more homogeneous diffusion in CSF than IT drug delivery device or lumbar puncture (in addition to provide less invasive access to CSF).^29,30 However, it is not routinely considered, partly due to previously frequent adverse effects (infections), now better prevented and managed. Thus, larger studies may enable to stratify patients to address issues, such as the impact of the administration modality and the coadministered corticotherapy posology—among other potential confounding factors.

In continuation of this search, a controlled trial would be best to confirm our positive conclusion on IT trastuzumab efficacy, with the crucial and difficult choice of the control arm (systemic treatment only, or associated with liposomal cytarabine, with local and systemic corticosteroids). Besides, pharmacokinetics studies, which are very rare in the field of IT therapies, would also help to optimize administration schemes.

It would also be interesting to study the diffusion to subarachnoid space and neurological efficacy of the most recent systemic anti-HER2 therapies (neratinib,³¹ tucatinib,^32,33 trastuzumab deruxtecan,³⁴ and explore a potential complementary or synergic efficiency of IT trastuzumab combined with them.

Importantly, any of these potential subsequent studies definitely require to be designed with harmonized RANO/EANO/ESMO evaluation criteria to ensure reliability and to enable the best comparability of their results.^21,35 CSF cell-free circulating tumor DNA, circulating tumor cells and or NGS on the CSF monitoring may also represent promising research tools to explore molecular mechanisms of meningeal progression and to follow or early detect LM—if any clinical benefit proven—either combined with or even replacing ultimately the current clinical, biological, and radiological criteria.

Assessing QoL evolution under IT trastuzumab treatment was an important issue in our opinion but the high rate of missing data beyond the first month limits the robustness of their interpretation. There could be a selection bias in the patients answering questionnaires as 4 out of 5 patients who filled them at week 9 were those with the longest IT trastuzumab treatment durations. Thus, the implementation of facilitative tools dedicated to QoL evaluation of neurologic impaired and/or heavily asthenic patients and caregivers dedicated to clinical trial assessment could contribute to limit these missing data in future studies.

Conducting studies on patients with LM generally poses significant practical challenges and even ethical dilemmas inherent to this particularly complex population. Despite its small scale and its limits, our positive findings in terms of clinical neurologic response and QoL’s control confirms IT trastuzumab as a valuable option for selected patients and strongly support the interest for further investigations in the scope of IT trastuzumab treatment for HER-BC patients with LM.

Funding

Roche group supported this study in providing sponsorship for this trial. Funding source had no involvement in analysis and interpretation of data.

Conflict of interest statement

O.T. reports grants and personal fees from Roche (without any role neither any influence on data analyses and interpretation), grants and personal fees from MSD-Merck, grants from BMS, personal fees from Novartis-Sandoz, personal fees from Pfizer, personal fees from Lilly, personal fees from Astra-Zeneca, personal fees from Daiichi Sankyo, personal fees from Eisai, personal fees from Pierre Fabre, personal fees from Seagen, and outside the submitted work. E.L.R. has received honoraria for lectures or advisory board from Adastra, Abbvie, Daiichi Sankyo, Leo Pharma, Tocagen, and Seattle Genetics. M.G. reports grants and personal fees from Roche (without any role neither any influence on data analyses and interpretation) and Pfizer. F.O., O.T.G., I.T., A.K., C.D., S.T., S.V., and C.B. declare having no conflict of interest.

Authorship statement

All authors have significantly contributed to the manuscript. The author contributions are as follows: Experimental design: M.G., I.T., and C.B.. Implementation: E.L.R., O.T., C.D., S.T., O.T.G., and S.V. Data analysis and interpretation: F.O., C.B., I.T., M.G., O.T.G., and A.K. Manuscript writing: All authors. Final approval of manuscript: All authors.