Design
The study was a single-arm cohort study designed as a Simon two-stage Phase II design in patients with either asbestosis or benign pleural disease with mesothelin serum level higher than 0.40 nmol/L. The study intervention was AWPC. The primary outcome measure was reduction is serum mesothelin. The secondary aim was miRNA modulation. The study was coordinated by the Ontario Clinical Oncology Group. (NCT number NCT02076672).
Patient population
Recruitment
Participants were recruited from the practices of pulmonary physicians and thoracic radiology at the Firestone Clinic at St. Joseph’s Hospital in Hamilton, Ontario under the supervision of Dr. G. Cox and Dr. M. Kolb. After confirmation of eligibility and documentation of written informed consent, patients were registered and enrolled into the study, by the clinical center, using the web-based IRIS system maintained by the OCOG Coordinating and Methods Centre located at the Juravinski Hospital in Hamilton, Ontario, Canada.
The inclusion criteria were as it follows: a) having evidence of radiographic characteristics related to asbestos exposure and consistent with a diagnosis of non-malignant asbestos-related disease, either asbestosis or benign pleural disease; b) having mesothelin serum level higher than 0.40 nmol/L.
The main exclusion criteria were any prior cancer, any biliary condition for which AWPC was a contraindication, and mental disorders.
The study was approved by the Hamilton Integrated Research Ethics Board and all participants signed a written informed consent. The study product, AWPC, was approved by Health Canada (Reg. Num. HC6-24-c170190).
The baseline assessment included the collection of patient demographics, exposure to asbestos (yes/no, estimate of years of exposure), cigarette smoking (pack-years), medical history and documentation of concurrent medications. Physical exam included measurement of height and weight, heart rate and blood pressure. Routine blood work (i.e., CBC, platelet count, creatinine, bilirubin, ALT or AST) was also collected.
Study intervention
Treatment
AWPC is derived from artichokes (Cynara Scolymus) for which the main components are caffeoylquinic acid derivatives (cynarine and chlorogenic acid), flavonoids (luteolin and apigenin) and bitters (cynaropicrin). The study agent is based on the commercially available AWPC capsules. This product contains 30% of the artichoke freeze-dried extract, used in all the preclinical experiments [4], and 70% of micronized powder from the same portion of the artichoke plant (Cynara scolymus L.) leaves. The micronized powder is used to stabilize the active freeze-dried extract and to protect it against degradation. As described in the preclinical documentation, the highest artichoke extract dose successfully used to treat mice was 100 mg/kg. Following the FDA conversion table (ref: www.fda.gov), this concentration converts to an 8.13 mg/kg human-equivalent dose. Assuming a median body weight of 90 kg in the target population, we calculated an artichoke extract dose of 732 mg per day. Each 500 mg capsule of AWPC contains 404 mg of product (121 mg of artichoke extract and 283 mg of micronized artichoke leaf powder). To achieve this dose of extract extrapolated from the preclinical data, we needed to administer 6 capsules (i.e., 3000 mg) per day of AWPC to each patient. Study subjects remained on AWPC therapy for 90 days and were monitored for toxicity on Day 45 and at the end of the study Day 90. Treatment compliance was assessed by pill count. Adherence was defined at each study visit as the use of at least 85% of pills.
Assays
Blood samples
Blood was collected at baseline and at the end of the treatment period (Day 90) for serum mesothelin levels and miRNAs assessment. After drawing, blood was centrifuged and then separated in 6 different 1.7 ml samples and stored at -80 °C until mesothelin and miRNA assessment. Mesothelin (nmol/L) was measured in serum using the MESOMARKTM assay (Cis Bio International, Gif sur Yvette, France). The estimated coefficient of variation (CV) at the levels observed in this study was 2.5% for the method detection level is 0.30 nmol/L. Based on the method detection level of 0.30 nmol/L, to detect > 25% reduction in baseline mesothelin, as absolute change between 90th day of treatment and baselines values, we included subjects with levels > 0.40 nmol/L. Serum Mesothelin was assessed in different batches and in blind condition in respect of the baseline serum concentration.
For miRNA determination, matched samples (baseline and end-of-the study) from the same individual were retrieved, thawed, and assessed together and by the same laboratory technician to control, at least in part, for effect of technical variability.
Plasma RNA extraction
Human plasma specimens were processed within 2 h from blood collection, centrifuge at 3000 rpm for 15 min and stored at − 80 °C. Total RNA from plasma specimens was extracted by the miRNeasy Serum/Plasma Kit (Qiagen, Valencia, CA, USA) according to the manufacturer’s instructions.
MiRNAs Hybridization. 200 ng of total RNA for each sample was used to generate fluorescent microRNA by using the Agilent’s microRNA Complete Labeling and Hyb Kit (Agilent) according to manufacturer’s instructions. Labeled RNA was hybridized to human microRNA Microarray V21 (Agilent), containing probes for 2570 human miRNAs. Scanning and image analysis were performed using the Agilent DNA Microarray Scanner (P/N G2565BA). Feature Extraction Software (V-10.5) was used for data extraction from raw microarray image files using the microRNA_105_Dec08 FE protocol.
Study outcomes
The primary outcome was the decrease in mesothelin serum levels after treatment with AWPC for 90 days. The secondary outcome was the assessment of circulating miRNA expression modulation between the baseline and after the 90 day-period of treatment.
Toxicity was assessed using the NCI CTCAE version 4.03. For each reaction, the highest severity grade attained since the last assessment period was reported. If a CTCAE category did not exist, the Investigator assessed the AR as Grade 1 (mild), Grade 2 (moderate), Grade 3 (severe), Grade 4 (life-threatening), Grade 5 (death) to describe maximum intensity of the adverse reaction.
Statistics
Sample size
The sample size calculation was based on the estimated proportion of subjects who had a baseline mesothelin level > 0.40 nmol/L, and Simon’s optimal two-stage Phase II design. This is a single-arm study that tests the null hypothesis of insufficient efficacy versus an alternative that the treatment has sufficient activity to merit further investigation. In our study, the treatment effect on mesothelin has been estimated using the results of Creaney et al., 2011 where a positive outcome would be a decrease of mesothelin serum levels by 25% or more after treatment [8]. If the proportion of patients with a positive outcome was expected to be less than or equal to 15%, the treatment was considered not sufficiently promising, and the trial would have been stopped. On the other hand, a response of 35% or more would be a desirable level for pursuing this product in a later phase trial. We have set α = 0.10 and β = 0.10, thus the optimal sample size for the first step of the study was 19 patients. Starting from this sample size, if only 3 or fewer patients would have responded, the treatment did not prove a substantial biological activity with the consequent interruption of the trial at this first stage. If the treatment would have created a positive response in 4 or more patients, then some biological activity was evident with a consequent continuation to the second stage with the additional recruitment of 14 patients. At that point, if only 7 or fewer patients would have responded out of the total of 33 (< 24%), the treatment would have been considered to provide too low biological activity to justify an extension to a phase III trial.
Based on mesothelin first step results, we decided to continue the trial to the second stage. However, the active patient recruitment period for the second stage brought us to the beginning of the Sars-Cov-2 pandemia with the enrollment closure at the Firestone Institute for Respiratory Health. Thus, the present report includes only all 18 patients participating in the first step of the study.
In order to validate miRNA results from the trial, we included a cohort of 41 patients affected with either mesothelial benign lesions or mesothelioma which data were previously published by our group [9]. In that study, we performed an unbiased microRNA screening of malignant pleural mesothelioma specimens (n = 29) and mesothelial benign tissues (mesothelial cysts, n = 12). The samples were collected during surgery from a series of patients admitted at the Italian National Cancer Institute (INCI) between 2009 and 2012.
Analysis
Descriptive statistics including 95% confidence intervals were used to summarize study variables. For continuous measures, we used means, standard deviations, quartiles, minimum and maximums. For variables with skew distributions, we employed a log transformation prior to summarizing. For categorical variables, we used counts and percentages as summary measures. Relationships between continuous measures (e.g., baseline and post-treatment mesothelin values) were explored with scatterplots; for categorical outcomes, we used cross-tabulations. Statistical modelling of the change scores were performed using linear models with adjustment for degree of asbestos exposure, and other baseline factors such as the severity of radiological signs of benign asbestos-related disease, cigarette smoking and age.
Microarray data analysis
Data were verified and extracted by the Agilent Extraction 10.7.3.1 software and analyzed using an inhouse built routines by Matlab R2020b. (The MathWorks Inc.). Background-subtracted signal of 2571 human miRNA assays was used in the study. All arrays were quantile normalized, assuming that all samples were measured and analyzed under the same condition, enforcing all the arrays to assume the same mean distribution.
MiRNAs expression for 18 no-treated (baseline) and 18 treated (end-of-the study) matched samples from Agilent platform were analyzed. Significantly modulated miRNAs were assessed by using a permutation test and a paired-wilcoxon test. ROC analysis was performed and miRNAs with AUC (Area Under Curve) less than 70% were excluded from further evaluations.
In silico analysis
The MiRNet web tool (https://www.mirnet.ca/miRNet/home.xhtml (accessed on 12 May 2022)), based on Tarbase v8, was used to determinate miRNA-target interaction. The graphical view of the network built by the validated targets of the 11 differently expressed miRNAs signature or miR-181a-5p and miR-193a-5p was obtained by using Cytoscape software. The enrichKEGG R function of clusterProfiler package was used to identify and graph the pathways analysis. HALLMARK pathways from MsigDB obtained using the web tool ShinyGO. The enrichment has been separately evaluated by considering predicted targets of the 11-miRNA signature and the putative targets of the first 15 miRNA modulated between malignant mesothelioma and benign cysts (IRE cohort).