Clinical Trial Details
— Status: Completed
Administrative data
| NCT number |
NCT05540925 |
| Other study ID # |
EHBH-MVIPG37 |
| Secondary ID |
|
| Status |
Completed |
| Phase |
|
| First received |
|
| Last updated |
|
| Start date |
January 1, 2022 |
| Est. completion date |
September 1, 2022 |
Study information
| Verified date |
September 2022 |
| Source |
Eastern Hepatobiliary Surgery Hospital |
| Contact |
n/a |
| Is FDA regulated |
No |
| Health authority |
|
| Study type |
Observational
|
Clinical Trial Summary
Tumor staging system based on clinicopathological charactertics has been used to guide
treatment decisions. However, therapeutic outcomes of "early-stage" hepatocellular carcinoma
(HCC) differs significantly, which strongly suggests the requirement for a re-staging of
early HCC to inform treatment selection more precisely. Microvascular invasion (MVI) reflects
malignant biological characteristics of early HCC, and has a potential role of guiding
treatment selection. As such, the objective of this study is to investigate preoperative MVI
prediction based on MVI-related genomic signatures of cell-free circulating tumor DNA (ctDNA)
to establish a re-staging of early HCC. The investigators have detected 37 mutant genes
associated with MVI in HCC tumor tissues. In this study, the investigators will design a gene
panel based on these mutant genes to perform targeted gene sequencing on preoperatively
collected ctDNA to identify genomic signatures associated with MVI. A nomogram to predict MVI
before treatment will be generated by incorporating these genomic signatures. Based on a
calculated optimal cut-off value of the nomogram, early HCC patients can be re-staged into
subpopulations based on the nomogram-predicted risks of MVI. This study will develop a
re-staging system of early HCC based on tumor biological charactertics, which is expected to
accurately and individually guide treatment decisions and improve long-term survival
outcomes.
Description:
Study Design:
The genetic profiles associated with MVI in early-stage HCC was detected to generate a gene
panel based on the WES and targeted gene NGS data of paired tumor and non-tumor tissues.
Then, genomic alternations related to MVI in preoperative cfDNA were identified using
targeted sequencing with the panel. Based on the genomic signatures in cfDNA, a nomogram
model was constructed to predict MVI risks preoperatively, and a re-staging paradigm for
early-stage HCC based on predicted high or low-risk of MVI by the nomogram was subsequently
developed. Furthermore, the clinical relevance of the re-staging system in deciding on the
optimal extent of surgical resection for HCC was examined. This study was approved by the
Institutional Ethics Committee of each center, and informed consents were obtained from all
patients for their tissues or blood samples and clinical data to be used for research
purposes.
Patients, Surgical Treatment and Follow-up:
The eligibility criteria were patients aged 18-75 years, histopathologically confirmed HCC,
tumor within the Milan criteria, Child-Pugh class A of liver function, no history of other
malignancies, no previous anti-cancer treatment including neoadjuvant therapy before surgery,
curative-intent surgical resection defined as complete removal of macroscopic nodules with
microscopic tumor-free resection margins and without distant metastasis and major vascular
invasion, and complete clinicopathological and follow-up data.
A total of 436 patients who underwent surgical resection for early-stage HCC between June
2015 and December 2017 and met the eligibility criteria were prospectively collected. Of
these patients, 150 patients who were operated between June 2015 and May 2016 at the Eastern
Hepatobiliary Surgery Hospital (EHBH) served as the panel discovery cohort. Paired tumor and
adjacent non-tumor tissues from 81 patients were used for WES, and those from another 69
patients were for targeted gene NGS by using a commercial 123-gene-panel to detect
MVI-related mutations.
Another 286 patients who underwent surgery between June 2016 and December 2017 at
multicenters were used in cfDNA testing. Peripheral blood samples of these patients were
collected 30 minutes prior to surgery to extract cfDNA. In addition to conventional
preoperative assessment, volumetric assessment for future liver remnant (FLR) was performed
using three-dimensional reconstruction of imaging studies. All resections were performed with
an intention of complete removal of tumor nodule(s) with either anatomic or non-anatomic
resections. The width of surgical resection margin was ultimately decided by the operating
surgeons based on tumor size, tumor number, intrahepatic location, local invasive features of
tumor on imaging studies, cirrhosis, estimated volume of FLR, liver function and general
condition of patients as previously reported, as well as on the surgeon's experience as a
study on real-world practice. Patients were followed-up regularly after surgery. Tumor
recurrence/metastasis was defined as appearance of new lesion(s) confirmed on at least two
radiological imaging techniques, with or without elevation of serum tumor markers.
Among these 286 patients, 125 from the EHBH comprised the training cohort for identifying
genomic features associated with MVI in cfDNA and in developing a MVI-predicting model, while
the remaining 161 patients from multicenters (the Zhongda Hospital of Southeast University,
Nanjing; Sun Yat-Sen Memory Hospital of Sun Yat-Sun University, Guangzhou; the Mengchao
Hepatobiliary Surgery Hospital of Fujian Medical University, Fuzhou; and the EHBH, Shanghai)
served as the external validation cohort to verify the model performance.
Preoperative Clinical Variables:
Both genomic signatures related to MVI in cfDNA and preoperative clinical variables which
were possibly associated with MVI were used in identifying independent risk factors of MVI to
develop the MVI-predicting model. The preoperative clinical variables included age, gender,
total bilirubin (TBIL), alanine aminotransferase (ALT), aspartate aminotransferase (AST),
albumin, platelets, prothrombin time (PT), α-fetoprotein (AFP), prothrombin induced by
vitamin K absence-II (PIVKA-II), hepatitis B and C serology, HBV-DNA levels, and imaging
features such as tumor number, tumor diameter and cirrhosis on preoperative contrast-enhanced
CT scan and/or magnetic resonance imaging (MRI).
Histopathological Diagnosis of MVI:
Surgically resected specimens were routinely examined histopathologically after surgery. To
ensure the quality of tissue samples in detecting MVI, a seven-site sampling protocol
recommended by the Pathology Branch of the Chinese Medical Society (CMS) was used. Sections
of tumor and non-tumor liver tissues were examined to observe the presence of MVI.
Histopathological diagnosis of MVI was in line with previous reports. Briefly, MVI was
identified if there was a microvascular cancer embolus or cancer cell nest in small branches
of portal vein or hepatic vein in the adjacent liver tissues, or in large capsular vessels
lined by endothelium that was visible only on microscopy. The same criteria for the
histopathological diagnosis of MVI were used in each center. The width of surgical resection
margin was defined as the nearest distance between the raw surface after hepatic resection
and the tumor capsule. All histopathological studies were independently carried out by three
pathologists who arrived at a consensus by discussion if there was any controversy.
Assessment of Suitability for Both Wide and Narrow Margin Resections Among the 286 patients,
patients who had undergone a narrow margin resection were reassessed postoperatively to
determine whether they were also suitable for a wide margin resection. Three senior hepatic
surgeons who were blinded to the prognostic information reviewed the preoperative data of
these patients. Based on imaging features including tumor size, tumor number, tumor capsule
status, intrahepatic location of tumor, estimated volume of FLR4, degree of cirrhosis, liver
function, general performance, as well as their own surgical experience. The assessment was
based on technical feasibility and surgical safety.
Tissue and Blood Samples:
Fresh tissue samples were collected and stored at -80℃ until use. Peripheral blood (10 mL per
sample) was collected in EDTA vacutainer tubes within 30 minutes prior to surgery, processed
within 1 h of collection, and separated by centrifugation at 1,600g for 10 minutes,
transferred to microcentrifuge tubes and centrifuged at 20,000g for 10 minutes to remove cell
debris. Both plasma and WBCs were collected and stored at -80℃ until use.
Genomic DNA and cfDNA extraction:
Genomic DNA from tumor tissues and WBCs were extracted by the DNeasy Tissue or Blood Kit
(Qiagen), and then fragmented to a size ranging from 200 to 400 bp using the Covaris S2
SonoLAB (Covaris). cfDNA was isolated from 3-5 mL of plasma of each patient using the QIAamp
Circulating Nucleic Acid Kit (Qiagen). DNA was extracted according to the manufacturer's
instruction, quantified by a Qubit fluorometer (Life Technologies), and kept at -80℃ until
use.
Whole-exome Sequencing:
1μg of DNA per sample was used as the input material for DNA library preparation. The
sequencing library was generated using the SureSelect XT Target Enrichment System for
Illumina Paired-End Sequencing Library (Agilent), and index codes were added to each sample.
The genomic DNA samples were fragmented by sonication to an average size of ~ 400bp. The DNA
fragments were end-polished, a-tailed and ligated with the full-length adaptor for
sequencing, followed by PCR amplification. The libraries were analyzed for size distribution
using an Agilent 2100 Bioanalyzer (Agilent). The clustering of the index-coded samples was
performed using a cBot Cluster Generation System (Illumina) according to the manufacturer's
instruction. After cluster generation, DNA libraries were sequenced on the Illumina HiSeq
2000 platform, and paired-end 2×100 nt multiplex sequencing reads were generated.
Targeted Gene Next-generation Sequencing:
In targeted sequencing of HCC tissues with the commercial 123-gene-panel, 100ng of fragmented
genomic DNA were used for NGS library construction, and probes spanning the coding sequences
of 123 genes frequently mutated in HCC were used for targeted gene capture (Baodeng Bio)
(Supplementary Table 1). NGS library was sequenced with 100 bp paired-end runs on an Illumina
HiSeq 2000 system (Illumina). The average coverage depth for all probes was at least 1000×.
In targeted sequencing of WBCs with MVI-PG37, 100ng of fragmented genomic DNA were used for
NGS library construction using a KAPA sequencing library construction kit (Kapa Biosystems).
Genomic DNA NGS library was then captured by the Accu-Act panel (AccuraGen) and was sequenced
with 100 bp paired-end runs on an Illumina HiSeq 2500 system (Illumina). The average coverage
depth for all probes was at least 500×.
In targeted sequencing of cfDNA with the MVI-PG37 panel, NGS-based assessment was performed
using the Firefly platform (AccuraGen) as previously reported13. NGS libraries were sequenced
on an Illumina Hi-Seq 2500 system (Illumina), and the unique sequencing reads were determined
using an AccuraGen proprietary algorithm. The average coverage depth for all probes was
approximately 7000×.
Bioinformatics Analysis for the WES and NGS Data:
All sequencing data were aligned to the hg19/GRCh37 human reference sequence. For WES data,
somatic mutations including SNPs and indels were identified by MuTect2 algorithm. The MOAF
for each somatic mutated gene was calculated to screen MVI-related genes using logistic
regression analysis, and genes with a P value of < 0.1 were selected as candidates to
generate the gene panel for further sequencing of cfDNA. Mutated genes of germline mutations
were identified using the MutSigCV algorithm with a false discovery rate (FDR) of <0.001 as
the cut-off in HCC with or without MVI, and then were included in enrichment analysis based
on the GO and KEGG database. The mutated genes of the most significantly enriched GO/KEGG
categories were focused, SNPs and indels from these pathways were included in logistic
regression analysis to detect SNPs and indels associated with MVI with a P value of < 0.05.
Genes containing significant SNPs and indels were selected as candidates of the gene panel.
For targeted sequencing data of tissues with the commercial 123-gene-panel, mutations were
identified by the MuTect2 algorithm. Mutated genes were identified using the MutSigCV
algorithm with an FDR of <0.05 as the cut-off. MOAF of each mutated gene was calculated to
screen MVI-related genes using the logistic regression model with a P value of <0.05 as the
cutoff.
For targeted sequencing data of cfDNA and WBCs, background noise introduced by random NGS
error was removed by the AccuraGen proprietary algorithm. The cfDNA and tumor genomic DNA
sequencing data were cross-checked with germline mutation from WBCs genomic DNA to identify
somatic mutations. MOAF was calculated for each somatic mutated gene and used as somatic
signatures. The status (yes/no) of each germline SNP/indel was used as a germline signature.
Identification of Genomic Signatures Related to MVI in cfDNA:
To detect MVI-related genomic signatures in cfDNA, the signatures including MOAF of somatic
genes and status (yes/no) of germline mutations were subjected into the univariate logistic
regression model. Signatures with a P value of <0.1 were used for the forward stepwise
multivariate selection using the maximum concordance index (C-index) criterion.
To accurately and conveniently assess the performance of the identified MVI-related
signatures in predicting MVI, the MVI-related genomic signatures were used to construct a
cfDNA-based score using the coefficients weighted by the multivariate logistic regression
analysis.
Establishment of a Nomogram to Predict MVI The nomogram model to predict MVI was developed
using the cfDNA-based score and preoperative clinical variables associated with MVI on
univariate and multivariate logistic regression analyses in the training cohort. Logistic
regression analysis and support vector machine were used to construct the model. Decision
curve analysis (DCA) was used to compare the performance between these two machine learning
algorithms. The rms package of R software was used to formulate the nomogram. The model
performance was assessed by concordance index (C-index) and calibration curve. Internal
validation for the model performance was done using leave-one-out (LOO) cross-validation in
the training cohort, and external validation was carried out using multicenter data.
Development of a Re-staging System for Early-stage HCC The total nomogram score of each
patient was calculated, and receiver operating characteristic (ROC) curve analysis was used
to calculate the optimal cutoff value of the nomogram to distinguish between high and
low-risk for MVI by maximizing the Youden index (i.e., sensitivity+specificity-1). Accuracy
of the cut-off value was confirmed by sensitivity, specificity, and positive and negative
predictive values. Using this cutoff value, patients with early-stage HCC were stratified
into two sub-stages with nonogram-predicted high or low-risks for MVI, respectively.
Statistical Analysis:
The clinical endpoints included recurrence-free survival (RFS) which was defined as the time
from surgery to the first diagnosis of recurrence or patient death without recurrence;
overall survival (OS) was defined as the time from surgery to patient death from any cause or
the last follow-up; and local recurrence was defined as any recurrence located within 2 cm of
the surgical resection margin. Survival outcomes were estimated using the Kaplan-Meier method
and log-rank test. The Cox proportional hazards model was used to identify independent
prognostic factors. A P value of < 0.05 was considered statistically significant, unless
otherwise specified. Statistical analysis was performed using the python programming language
version 2.7 (https://www.python.org/) and the R software version 3.1.1
(http://www.r-project.org/).
