Phase I trial of copanlisib, a selective PI3K inhibitor, in combination with cetuximab in patients with recurrent and/or metastatic head and neck squamous cell carcinoma

Background The phosphatidylinositol-3 kinase pathway is often altered in head and neck squamous cell carcinoma (HNSCC), and is involved in the resistance to EGFR inhibitors. Objective We investigated the dose-limiting toxicities (DLTs), maximum tolerated dose, pharmacokinetics, and preliminary efficacy of the combination of copanlisib, an intravenous, pan-class I PI3K inhibitor, with the anti-EGFR monoclonal antibody cetuximab in recurrent and/or metastatic HNSCC patients in a phase I dose-escalation trial. Patients and methods Copanlisib was given intravenously on days 1, 8, and 15 of 28-day cycles at the dose of 45 mg and 30 mg, in combination with standard doses of weekly cetuximab (400 mg/m2 loading dose followed by 250 mg/m2 on days 8, 15, and 22, and weekly thereafter). Results Three patients received copanlisib 45 mg, of whom two experienced grade 3 hyperglycemia during Cycle 1 that met the DLT criteria. Eight patients were then treated with copanlisib at the dose of 30 mg. Because of the occurrence of hyperglycemia, a premedication with metformine was introduced on the day of the injections. No DLTs were reported at this dose level. The trial was stopped early because of the unfavourable toxicity profile of the combination. Among eight evaluable patients for response, four patients (50%) had disease stabilization according to RECIST1.1 as best response. Conclusion Copanlisib combined with cetuximab demonstrated unfavorable toxicity and limited efficacy in heavily pretreated recurrent and/or metastatic HNSCC patients. Trial registration NCT02822482, Date of registration: June 2016.


Introduction
The phosphatidylinositol-3 kinase (PI3K) signaling pathway has been found to play an important role in the pathogenesis of HNSCC [1,2]. Both PIK3CA activating mutations and amplifications were reported in 13% and 22% of HNSCC patients, respectively [3]. The H1047R/L activating mutation has been shown to overcome the EGFR inhibition through unrestrained AKT phosphorylation in preclinical HNSCC cell lines [4]. PI3K inhibition has also been shown to restore sensitivity to cetuximab in EGFRresistant HNSCC models both in vitro and in vivo [5]. In the clinic, PI3K inhibition with buparlisib improved overall survival in combination with paclitaxel as compared to paclitaxel alone in recurrent and/or metastatic HNSCC patients [6].
Copanlisib (BAY 80-6946) is a highly selective, panclass, PI3K inhibitor with potent activity against both the α and δ catalytic subunit isoforms. A first-in-human phase I study established the maximum tolerated dose (MTD) of copanlisib single-agent as 60 mg [7]. Copanlisib demonstrated a manageable safety profile in long-term treatment in lymphoma, with hyperglycemia, diarrhea, hypertension, and neutropenia being the most common adverse events (AEs) [8]. Copanlisib was shown to overcome resistance to cetuximab in vivo [9].
We report here the results of the phase I clinical trial of copanlisib combined with cetuximab in patients with recurrent and/or metastatic HNSCC who failed platinum and cetuximab therapy.

Study design and treatment
This study was an open-label, non-randomized, dose escalation phase I clinical trial conducted between April 26, 2016 and August 17, 2019 in four centers in France. Intravenous copanlisib was administered over one hour, on days 1, 8, and 15 of 28-day cycles. Cetuximab was administered intravenously at a loading dose of 400 mg per square meter of body-surface area one hour after copanlisib, followed by subsequent weekly doses of 250 mg per square meter. Dose escalation followed a 3 + 3 design. A starting dose of 75% (45 mg) of its recommended phase II dose (RP2D) as single agent (60 mg) was selected for copanlisib, whereas, cetuximab was given at its full RP2D. Dose level two was supposed to be the RP2Ds of both copanlisib and cetuximab given as single agents. A dose level -1 with copanlisib given at the dose of 30 mg was planned if needed.
The study was compliant with the Declaration of Helsinki and Good Clinical Practice, and was approved by the appropriate ethics committees. All patients provided written informed consent. This study was registered under identifier NCT02822482 in ClinicalTrials.gov.

Patients
Patients aged ≥ 18 years with recurrent and/or metastatic histologically or cytologically confirmed HNSCC of the oral cavity, oropharynx, larynx, or hypopharynx, defined as distant metastases and/or inoperable local and/or regional recurrence, were eligible. Eligible patients were required to have disease progression on or after cetuximab and platinum therapy. Patients had to have at least one evaluable lesion according to RECIST1.1, an Eastern Cooperative Oncology Group performance status of ≤ 2, adequate bone marrow (absolute neutrophil count [ANC] ≥ 1.0 × 10 9 /L, platelets ≥ 75 × 10 9 /L), and organ functions (ALT/AST ≤ 2.5 × upper limit of normal [ULN], bilirubin ≤ 1.5xULN, creatinine clearance ≥ 60 mL/min as measured or calculated by Cockcroft and Gault formula, or serum creatinine ≤ ULN). Patients with uncontrolled diabetes mellitus were excluded.

Endpoints
The primary endpoint of the study was the occurrence of dose-limiting toxicities (DLTs) during Cycle 1 to define the MTD of the combination. Secondary endpoints included safety, the pharmacokinetics parameters, the assessment of potential predictive biomarkers, overall response rate (ORR), progression-free survival (PFS), and overall survival (OS).

Safety and tolerability
AEs were evaluated and graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events version 4.03. DLTs were defined as grade ≥ 3 hyperglycemia despite curative optimal glucose lowering therapy, grade ≥ 3 ALT/AST elevation, ANC < 0.5 × 10 9 /L lasting more than seven days, bleeding felt to be due to thrombocytopenia, febrile neutropenia (both ANC < 1.0 × 10 9 /L and fever > 38.5 °C), platelets < 25 × 10 9 /L, diarrhea ≥ grade 3, rash ≥ grade 4, hypertension ≥ grade 3 despite optimal treatment, and any clinically significant ≥ grade 3 AE related to copanlisib and/or cetuximab.

Pharmacokinetics
Blood samples were collected at Cycle 1 Day 1 and at Cycle 1 Day 15. A method was developed and validated for the determination of copanlisib in human plasma samples with heparinate at the Department of Radio-Pharmacology of the Institut Curie, using protein precipitation for sample preparation followed by Ultra Performance Liquid Chromatography with Tandem Mass Spectrometric Detection. This method was validated over the concentration range 1.00 ng/ mL to 600 ng/mL.
Copanlisib individual pharmacokinetics parameters were calculated using plasma concentration-time data by noncompartmental analysis. The maximum plasma concentration (C max ), and time to reach C max (T max ) were presented. The area under the plasma concentration-time curve (AUC) from 0 to the last measurable plasma concentration (AUC 0-last ) was calculated by the linear trapezoidal rule. AUC extrapolated to infinity (AUC 0-∞ ) was calculated as AUC 0-∞ = AUC 0-last + Ct/λz, where Ct was the last measurable plasma concentration, and λz the apparent terminal phase rate constant. The apparent plasma terminal elimination half-life t 1/2 was estimated using t 1/2 = (ln2)/λz.
A population pharmacokinetics modelling was performed in order to describe copanlisib plasma concentrations versus time curves. The pharmacokinetics of copanlisib was best described by a two-compartment open model with linear elimination. Different covariates were investigated, including both body weight and age as a continuous covariate.

Exploratory analyses
For all patients, a fresh frozen research biopsy or Formalin fixed paraffin embedded (FFPE) archived tumour sample was required for biomarker analyses. The tumour samples were analysed with a custom gene panel that included exons 10, 14, and 21 of PIK3CA, as well as the total coding sequence of PTEN allowing the detection of gene copy number alterations in addition to mutations (Supplementary Materials and Methods). For the tested oncogenes including PIK3CA, only pathogenic variants driving a gain of function and amplifications were considered, whereas for tumour suppressor genes such as PTEN, pathogenic variants driving a loss of function were considered.
Immunohistochemistry (IHC) was performed in parallel to determine the expression level of PTEN, and to validate any PTEN loss of function. A monoclonal antibody specific of the PTEN amino acids 321 to 336 was used (Zymed® laboratories).

Efficacy
Tumor assessments were performed at baseline within 28 days before enrolment, and every eight weeks after starting administration. PFS was defined as the time between inclusion and disease progression or death, whichever occurred first. OS was defined as the time between inclusion and death due to any cause. ORR was defined as the proportion of evaluable patients for response with a complete or a partial response.

Statistical analysis
The planned sample size for the dose escalation included a maximum of 12 patients. Safety analyses were performed on the full analysis set. Pharmacokinetic parameters were calculated according to the non-compartmental method. Statistical comparisons were performed using χ2 test or Fisher's exact test for categorical data, Student's test or Wilcoxon's test for continuous variables and log-rank test for censured data. OS and PFS were estimated using the Kaplan-Meier method. Median follow-up with a 95% confidence interval was calculated by reverse Kaplan-Meier method. ORR was calculated on the subgroup of evaluable patients for response according to RECIST1.1.
All analyses were performed with SAS V 9.4 and R 3.6.1 software on Windows. For clinical and biological analyses, all p-values < 0.05 (two-sided) were considered statistically significant.

Patients and treatment
Eleven patients were enrolled in the trial (three at the dose level of 45 mg, and eight at the dose level of 30 mg) (Fig. 1). Patient characteristics are reported in Table 1.
At the initial dose level of 45 mg, two patients out of three (67%) experienced a grade 3 hyperglycemia during Cycle 1 that met the DLT criteria. The next eight patients were treated at dose level -1 with copanlisib 30 mg. Because the first patient of that dose level also experienced a transient grade 3 hyperglycemia, the next two patients were premedicated with metformine 500 mg starting on the day of the copanlisib injection either at Day 15 or at Day 1 of the first cycle. The patient who was premedicated with metformine 500 mg at Day 15 still experienced a transient grade 3 hyperglycemia, and metformine premedication was therefore increased to 1000 mg in the next five patients. No DLTs were met at the dose level of copanlisib 30 mg. Transient grade 3 adverse events that did not meet the DLT criteria were reported. Among the three patients treated at the dose level of 45 mg, two patients (67%) experienced a grade 3 hypertension. Among the eight patients treated at the dose level of 30 mg, the two first patients (25%) experienced a transient grade 3 hyperglycemia. A grade 3 hypertension was reported in two patients (25%), a grade 3 anemia in one patient (12.5%), and a grade 3 lymphopenia in one patient (12.5%). Premedication with metformine 1000 mg significantly correlated with low blood glucose levels compared to patients without premedication (p = 0.026) (Fig. 2).
Across the entire cohort, two patients (18%) experienced a grade 2 acneiform rash, which was the most common AE associated with cetuximab. Eight patients (73%) experienced grade 4 AEs unrelated to copanlisib and cetuximab, including thrombopenia, oral haemorrhage, general physical health deterioration, cholestasis, device-related infection, lung abscess, septicemia, and hypokalemia in one patient each (9%). Permanent treatment discontinuation was caused by copanlisib-related grade 3 hyperglycemia in three patients (27%), disease progression in three patients (27%), grade 2 septicemia in one patient (9%), grade 4 oral haemorrhage in one patient (9%), grade 3 respiratory failure in one patient (9%), lack of clinical benefit after four cycles of copanlisib in one patient (9%), and death in one patient (9%).

Pharmacokinetics
From the 11 patients investigated, 77 time-blood samples were available for analysis (Table 4)

Efficacy
Median follow-up was 6.0 months [range: 2.4-10.8]. In the overall study population, eight out of the 11 patients were evaluable for response, three patients being not evaluable because of early death. No complete or partial response was observed. Four patients (50%) experienced a disease stabilisation, including one patient with both a PTEN loss and a PIK3CA mutation.

Discussion
Our study established that intravenous copanlisib combined with weekly intravenous cetuximab could not be safely administered to patients with recurrent and/or metastatic HNSCC without a premedication with metformin because of hyperglycemia. Although a premedication with metformin would have allowed to re-escalate, the sponsor decided to stop the study given the poor tolerance and limited efficacy. Therefore, both the MTD and the RP2D of the combination could not be established.
The most common copanlisib-related AEs were hyperglycemia and hypertension. The safety profile of cetuximab did not seem to be impacted by the addition of copanlisib, with common class-effect toxicities including rash and hypomagnesemia. The overall incidence of all-grade and grade 3 hyperglycemia were 64% and 36% respectively. In clinical trials evaluating single-agent copanlisib in lymphoma patients at the dose of 60 mg, the overall incidences of all-grade and grade ≥ 3 hyperglycemia were slightly lower [8,10]. The higher incidence of grade 3 hyperglycemia reported in our study might be related to our HNSCC patient population. Due to dysphagia, most patients were fed with means increasing the risk of hyperglycemia, including the use of a gastrostomy, parenteral nutrition, and oral nutritional supplements. In addition, most of patients were receiving corticosteroids that is known to increase glycemia. Hyperglycemia is a recognized on-target effect of p110α inhibition, given that p110α is a critical lipid kinase required for insulin signaling in two key cell types, including adipocytes and myotubes [11,12].
The incidences of all-grade (46%) and grade 3 (36%) hypertension were also higher than reported in clinical trials evaluating single-agent copanlisib at the dose of 60 mg [8,9]. These differences can be explained by the high rate of patients with hypertension at baseline in our study (27%). Furthermore, HNSCC patients often suffer from cardiovascular comorbidities due to tobacco and alcohol consumption [13]. Gastrointestinal toxicities remained low relative to orally administered PI3K inhibitors [14,15], presumably due to the intravenous route of administration, and the absence of a first-pass metabolism. A lower incidence of diarrhea and transaminases increase was observed in our study as compared with those reported in single-agent studies of copanlisib [8,9].
Pharmacokinetics parameters of copanlisib combined with cetuximab were consistent with those reported for weight-based dosing copanlisib monotherapy [7], including a short T max and a long t 1/2 . Copanlisib had a large tissue diffusion with similar C max on Days 1 and 15. Given the lack of significant accumulation, our results suggest no pharmacokinetics interaction between copanlisib and cetuximab. This was expected since the metabolism of copanlisib is mediated via CYP3A4/5 (> 90%) and to a minor extent via CYP1A1 (< 10%) [16], and cetuximab is a monoclonal antibody that is degraded by macrophages without any liver metabolism. Copanlisib pharmacokinetics was dose proportional.
Four out of eight evaluable patients achieved disease stabilization according to RECIST1.1, but no objective responses were observed. The limited efficacy observed in our study might be explained by several factors. First, none of the patients were exposed to the recommended phase II dose of copanlisib as single agent. Given that all patients had progressed on cetuximab before, it is unlikely that a suboptimal dose of copanlisib might have been able to produce antitumor activity in combination with cetuximab. Second, our patient population was heavily pretreated with a poor prognosis as exemplified by the fact that three patients died within two months following their entry in the study. Despite the suboptimal treatment received, the PFS and OS were in line with the results obtained in a phase II trial evaluating PX-866, an irreversible oral PI3K inhibitor, in combination with cetuximab in a similar patient population [17].
In summary, copanlisib combined with cetuximab demonstrated unfavorable toxicity and limited efficacy in heavily pretreated recurrent and/or metastatic HNSCC patients. The use of a premedication with metformin might have allowed to increase the dose of copanlisib to a potentially effective dose.

Supplementary information
The online version contains supplementary material available at https:// doi. org/ 10. 1007/ s10637-021-01152-z. tutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Informed consent Informed consent was obtained from all individual participants included in the study.