David Polsky MD, PhD
Professor of Dermatologic Oncology
NYU Langone Health
New York, NY
Author: Dan Li | National Cancer Institute, Bethesda, MD
Background: Glypican-3 (GPC3) is a cell surface protein that is highly expressed in hepatocellular carcinoma (HCC).
Methods and Results: In this report, we analyzed two chimeric antigen receptors (CARs) targeting GPC3, CAR (HN3) recognizes the N-lobe of GPC3 while CAR (hYP7) targets the C-lobe that is close to the cell surface. In a real-time cell cytolytic assay, we found that CAR (hYP7) T cells showed higher cytolytic activity than CAR(HN3) T cells at an E:T ratio of 2:1 when co-cultured with HepG2 cells for up to 140 hours. The Luminex analysis showed that CAR (hYP7) T cells produced significantly more Th1/Th2 cytokines and chemokines than CAR (HN3) T cells following exposure to Hep3B and HepG2 cells. In the Hep3B peritoneal dissemination xenograft mouse model, a single intraperitoneal infusion of CAR (hYP7) cells exhibited sustainable antitumor efficacy, and all of the mice survived without recurrence 10 weeks post treatment. Strikingly, we demonstrated that single injection of CAR (hYP7) T cells derived from either healthy donor or HCC patient significantly regressed tumor growth in the orthotopic HCC mouse model. Using a highly sensitive droplet digital PCR (ddPCR), we detected high frequencies of CAR T cells in the spleen and tumor microenvironment of mice in the CAR (hYP7) group. By contrast, CD19 CAR T cells showed no sign of gene integration in either tissue. More importantly, we found that CD8+ CAR (hYP7) T cells recovered from mouse spleens exhibited much higher lytic activity against HCC cells than CD4+ CAR (hYP7) T cells. Finally, we found that the GPC3-targeted CAR T cells downregulated the Wnt signaling in HCC cells.
Conclusion: The hYP7 antibody-derived GPC3-specific CAR is a promising therapeutic that could be tested for the treatment of advanced stage liver cancer.
Author: Ramzey Abujarour | Fate Therapeutics, San Diego, CA
Abstract embargoed at this time.
Author: Matteo Benelli | Azienda USL Toscana Centro, Hospital of Prato, Prato, Italy
CDK4/6 inhibitors, such as palbociclib, target the cyclin-dependent kinases 4 and 6, essential for cell cycle regulation. CDK4/6 inhibitors are established treatment options in patients with endocrine receptor positive, HER2 negative metastatic breast cancer (BC).
We recently conducted a transcriptomic study of seven palbociclib resistant luminal BC cell lines, demonstrating a common transcriptional signature of resistance to palbociclib, including the over-expression of CCNE1 and under-expression of RB1. Genomic analysis based on ddPCR and whole exome sequencing (WES) demonstrated CCNE1 over-expression due to gene amplification in only two cell lines, and RB1 under-expression due to gene deletion in 3 out of 7 cell lines. This prompts the hypothesis that more complex molecular mechanisms of CCNE1 over-expression / RB1 under-expression may exist.
To improve the characterization of the molecular mechanisms underlying the resistance to palbociclib, we performed single-cell transcriptomic profiling (single-cell RNA-seq, scRNA-seq) of seven BC cell line resistant to palbociclib and their sensitive counterparts. Cells were isolated using the Bio-Rad ddSEQ single cell isolator and sequenced in order to obtain about 100,000 reads per cell. A total of 10,557 cells with at least 2,000 genes read per cell were selected for downstream analysis, corresponding to 5,116 palbociclib sensitive cells and 5,441 palbociclib resistant cells.
Considering the 5,000 most variable genes, dimensionality reduction analysis based on t-distributed stochastic neighbor embedding (tSNE) clearly segregated cells depending on their cell type, as expected based on previously generated bulk RNA-seq data. Using previously identified top cell-type specific differentially expressed genes between resistant and sensitive cell lines (absolute log2(fold-change) > 0.5), dimensionality reduction analysis correctly classified 98.2% of cells based on their sensitivity/resistant status, demonstrating the quality of our scRNA-seq data. Interestingly, we observed a small fraction (<1.5%) of misclassified sensitive cells showing resistant-like phenotypes. Co-expression analysis suggests modulation of key gene expression programs between sensitive and resistant cells, including E2F signaling. Further analyses to validate these data are ongoing and the results will be presented during the meeting.
This study represents, to our knowledge, the first attempt to generate a single-cell transcriptomic profiling of a large panel of luminal BC cell lines resistant to palbociclib. Results from this study might lead to the identification of new biomarkers of de-novo and acquired resistance to palbociclib, which would assist to better personalize treatment of patients with endocrine receptor positive BC.
Author: Basant Ebaid | Olivia Newton-John Cancer Research Institute, Melbourne, Australia
Introduction: Over 90% of breast cancers occur sporadically with no obvious hereditary factors at play, therefore highlighting the need for biomarkers that can identify women who are predisposed to developing breast cancer. DNA methylation of the BRCA1 promoter region is detectable at low levels in the peripheral blood of some women. This phenomenon is referred to as constitutional BRCA1 methylation. Peripheral blood methylation of the BRCA1 gene has been identified as a predisposition factor to the development of BRCA1 methylated tumors. We performed a case-control study to assess peripheral blood BRCA1 methylation from healthy women, and from women with breast cancer. This study aimed to establish the frequency of peripheral blood methylation of the BRCA1 gene in women with and without breast cancer, to further our understanding of the role of constitutional BRCA1 methylation in breast cancer predisposition.
Methods: Constitutional BRCA1 methylation was assessed in peripheral blood DNA of breast cancer patients and healthy women. Whole blood DNA was obtained from healthy women (n=327 controls) and women with breast cancer (n=300 cases) as part of the LifePool Project. Blood was collected on the day of mammography with BreastScreen Victoria. We used a probe-based droplet digital PCR (ddPCR) assay designed to quantify methylated and unmethylated alleles at the BRCA1 promoter region.
Results: Previous research using less-sensitive methodologies has reported a BRCA1 methylation frequency of 2-4% in peripheral blood of healthy women. In this study, there was no significant difference in the observed BRCA1 methylation frequency between cases and controls (6.6% and 6.4% respectively). The discrepancy between our observed methylation frequencies and previously reported data can be attributed to the higher sensitivity of the ddPCR methodology. Nevertheless, the level of BRCA1 methylation in cases was significantly higher than the level of methylation observed in controls, consistent with previous studies. In addition, the frequency of BRCA1 methylation was higher in women diagnosed with breast cancer under the age of 50 compared to women diagnosed over 50 years old (average methylation frequency = 12.5% and 5.3% respectively).
Conclusions: ddPCR methodology enabled accurate quantification of methylated BRCA1 alleles in peripheral blood DNA down to less than 0.1%. In this selected population of women undergoing mammography, there was no difference in the frequency of detectable BRCA1 methylation between cases and controls. Once age was considered, detectable BRCA1 methylation showed an association with age of onset. Tumours for the women with breast cancer will be retrieved and tested for BRCA1 methylation, to further examine the link between the methylation observed in peripheral blood DNA and development of a BRCA1-methylated tumor.
Author: Lotte Bakkerus | Amsterdam UMC, Vrije Universiteit Amsterdam, Amsterdam, Netherlands
Introduction: Presence of microRNAs (miRNAs) in blood specimens hold promise for a predictive biomarker for treatment response in clinical practice as they are relatively resistant to degradation. Previously, we identified a six miRNA expression signature from tumor tissue that predicted response to first line systemic treatment in patients with metastatic colorectal cancer (mCRC) when combined with clinicopathological factors. The six miRNA expression signature consists of miR-17-5p, miR-20a-5p, miR-30a-5p, miR-92a-3p, miR-92b-3p and miR-98-5p. The aim of this study is to determine whether circulating levels of these miRNAs will be predictive for treatment outcome as well. Therefore we first studied whether these six miRNAs can be adequately measured in blood samples of patients with mCRC undergoing first line chemotherapy. Droplet digital PCR (ddPCR) technology was used for optimal quantification of miRNAs compared to normal quantitative PCR, because of low abundance of tumor specific miRNAs in blood.
Methods: Serum samples of 120 patients with multi-organ mCRC who participated in a multi-center randomized controlled clinical trial were analyzed. RNA was isolated with miRNeasy Serum/Plasma Advanced Kit of Qiagen. Expression levels of the six miRNA expression signature was determined with ddPCR by using the miRCURY LNA miRNA PCR system combined with EvaGreen. Primer concentration and annealing temperature were first optimized per specific miRNA assay for ddPCR. RNA spike cel-miR-39-3p was used for quality control. Analysis was performed on samples with at least a minimal amount of 10.000 obtained droplets. Absolute quantification per patient is displayed as copies per µl (cop/µl) as defined by Quantasoft software.
Results: In 120 blood samples miRNAs were measured by ddPCR. The most optimal primer concentrations for the different miRNAs was 0,5 µl for cel-miR-39-3p and 1,0 µl for the other 6 miRNAs. For cel-miR-39-3p, miR-17-5p, miR-20a-5p and miR-92a-3p the most efficient annealing temperature was 53 ℃ and for miR-30a-5p, miR-92b-3p and miR-98-5p this was 58 ℃. The mean number of obtained droplets was 15.044 (SD 1922). The highest amount of detected copies were observed in miR-17-5p (644,4 cop/µl), miR-20a-5p (1037,1 cop/µl) and miR-92a-3p (1136,0 cop/µl). Lower expression levels were observed in miR-30a-5p (9,88 cop/µl), miR-92b-3p (1,6 cop/µl) and miR-98-5p (6,9 cop/µl).
Conclusion: We demonstrated that with ddPCR absolute quantification of these six miRNAs in serum samples was efficient and reliable. Further analysis will reveal whether circulating miRNA expression levels can be used as a liquid biomarker for response to first line systemic treatment in patients with multi-organ mCRC.
Author: David Polsky | New York Univ. School of Medicine, New York, NY
Purpose: The mutational spectra of melanoma has been well characterized; however, the presence of distinct subclones among multiple tumors from a given patient has been less well described. As mutational heterogeneity has been associated with decreased responses to treatments in other cancers, we sought to estimate the occurrence of distinct subclones within individual melanoma patients by analyzing commonly mutated melanoma genes using a multi-platform mutation-detection approach.
Methods: We analyzed 271 formalin-fixed, paraffin embedded tumors from 99 patients with stage III or IV melanoma enrolled in the NYU Melanoma Biorepository. All patients had two or more available tumor specimens, and complete clinical data. All samples were reviewed for adequate tumor content, and extracted DNA was assessed for mutations at BRAFV600, NRASQ61, and TERT-124C>T and TERT-146C>T using a combination of multiplex SNaPshot assays, Sanger Sequencing, Allele-specific real-time PCR, or droplet digital PCR (ddPCR). Samples undergoing ddPCR analysis for TERT mutations were treated with uracil-DNA glycolase prior to amplification to remove C>T artifacts from formalin-fixation.
Results: Sixty patients had a primary plus one or more metastatic tumors available; 39 patients had multiple metastatic tumors, but no primary tumor available. Overall, 88% of patients had tumors with at least one BRAF, NRAS and/or TERT mutation. We identified inter-tumor mutational heterogeneity in 20/99 (20%) patients, with TERT mutational heterogeneity present in 15 of these patients. Among 14 patients with heterogeneity between their primary and metastatic tumors, 6/14 (43%) had additional mutations in their metastases compared to their primaries. Most interestingly, 8/14 (57%) patients had mutations in their primary that were undetectable in at least one of their metastases; 5 of these patients had TERT mutational heterogeneity. Three patients had a BRAF mutation in their primary that was undetectable in at least 1 of their metastases. One patient had a BRAFV600E/NRASWILD-TYPE/TERTWILD-TYPE primary on the leg and 3 regional metastases lacking BRAF mutations; but carrying NRASQ61K and TERT-124C>T mutations. Another patient had 3 different metastatic tumors, with 3 different mutational spectra. We did not detect any tumors with simultaneous BRAF and NRAS mutations; however, we did detect both TERT-124C>T and TERT-146C>T mutations in 7 tumors from 7 individual patients. Four of these were primary tumors, and metastases from these patients lacked 1 of the 2 TERT mutations identified in the primary.
Conclusion: Clonal heterogeneity in melanoma is fairly common as evidenced by divergent detection of TERT, BRAF and NRAS mutations using high sensitivity multi-platform mutation detection analyses of multiple melanoma tumors from individual patients. Heterogeneity appears to occur in primary tumors.
Author: Glenn Hanna | Dana-Farber Cancer Institute, Boston, MA
Purpose: The human papillomavirus (HPV) is linked to the majority of oropharyngeal squamous cell carcinomas (OPSCC). We have previously shown that tumor-derived HPV cell-free (cf)DNA can be detected and quantified in circulation with high sensitivity and specificity using droplet digital (dd)PCR1. Further, we have shown that plasma cfDNA levels correlate with disease burden and indicate treatment response in advanced OPSCC. In this study we pair plasma with salivary HPV cfDNA to understand its clinical application in monitoring locoregional spread of disease.
Methods: We present a proof-of-concept prospective observational cohort of recurrent, metastatic (R/M) HPV+ OPSCC patients treated with standard systemic therapies or on active surveillance. We utilized ultrasensitive ddPCR to identify and quantify both plasma and salivary HPV cfDNA (subtypes 16, 18, 31, 33 and 45) at multiple time points. Salivary HPV cfDNA was normalized to total DNA concentration as measured by fluorometric quantification. We then compared matched plasma and normalized salivary HPV cfDNA concentrations at various timepoints with clinical parameters, such as tumor burden and therapeutic response.
Results: Clinicopathologic data from 15 R/M patients revealed a male cohort with a median age of 57 at diagnosis. Five (33%) were on immunotherapy and 6 (40%) on standard chemotherapy during the study period (3 months). Salivary HPV cfDNA was detected in 13/15 (87%) samples (range 0-729 copies/mL) in at least one timepoint during the study (plasma cfDNA was detected in 12/15 or 80% of the cohort, range 0-12,949 copies/mL). All patients with measurable locoregional (LR) disease had detectable salivary cfDNA. While matched plasma and salivary cfDNA levels showed no correlation among individual patients, plasma cfDNA levels correlated with site of disease (LR, pulmonary or extra-pulmonary) (p < 0.001), while median salivary cfDNA levels strongly correlated with LR tumor burden (R = 0.81, p < 0.001). Salivary cfDNA levels declined by 80% of their baseline value within a median of 6.5 days among all patients with LR disease experiencing a response to treatment. Page 1 of 1
Conclusion: Our results suggest that high sensitivity salivary HPV cfDNA levels correlate with locoregional disease burden and can be an early indicator of locoregional treatment response, while our prior work has supported the role of plasma HPV cfDNA levels in monitoring metastatic disease. Evaluating paired plasma and salivary HPV cfDNA levels in a curative HPV OPSCC population is underway to further validate their prognostic and predictive potential. 1Hanna GJ, Supplee JG, Kuang Y, Mahmood U, Lau CJ, Haddad RI, et al. Plasma HPV cell-free DNA monitoring in advanced HPV-associated oropharyngeal cancer. Ann Oncol. 2018;29:1980–6. Page 2 of 1
Author: Pashalina Kehagias | Institut Jules Bordet, Bruxelles, Belgium
Purpose: Circulating tumor DNA (ctDNA) monitoring based on an individual mutation profile during therapy is under intense investigation in modern oncology. We previously reported that the increase of ≥50% of at least one somatic mutation among multiple monitored mutations per patient is associated with a significantly worse outcome1. This study investigates whether the ctDNA monitoring of one driver gene mutation, provides enough information as compared to multiple mutations to assess response to regorafenib in advanced chemorefractory colorectal cancer (aCRC) at an early timepoint.
Experimental procedures: Archival tumor tissue and plasma samples (PL) at baseline (BL) and 14 days (D14) after treatment initiation in aCRC pts (n=141) were prospectively collected in the RegARd-C multicenter clinical trial (NCT01929616). Somatic mutations were identified based on a CRC-oriented targeted gene sequencing of tumor tissue. All available (median 2 (1-4)) driver gene mutations were monitored per patient in PL at BL and D14 via droplet digital PCR (Bio-Rad QX200 ddPCR system) to assess ctDNA dynamics.
Results: In 96 evaluable patients, the most frequently monitored mutated genes were APC (73%), TP53 (72%), KRAS (66%), and PI3KCA (23%). Among patients with ≥2 monitored mutations (73/96), one was selected at random and compared to previous methodology taking in account dynamics of all followed mutations. Optimal cutoff (CO) evaluation (Contal & O’Quigley method) separated patients (n=96) according to a ctDNA increase of ≥50% versus an increase of <50% or a decrease. The concordance of ctDNA dynamics based on one randomly selected mutation and multiple monitored mutations was 91%. Our data demonstrated that a ctDNA increase based on one single mutation taken at random is significantly associated with a worse clinical outcome in terms of progression-free survival (HR 2.42, 95% CI (1.56-3.74), P<0.001) and overall-survival (HR 2.17, 95% CI (1.41-3.34), P<0.001). In addition, when combining patients’ ctDNA dynamics to BL ctDNA levels (≥ or < 5 ng/mL optimal CO) or BL cell-free DNA (cfDNA) levels (≥ or < 50 ng/mL optimal CO), we could distinguish 4 patients’ subgroups with different prognosis. However, when performing a multivariate analysis including clinical parameters, BL ctDNA and BL cfDNA levels, BL ctDNA was not relevant in the presence of BL cfDNA.
Conclusion: The monitoring of ctDNA dynamics based on only one randomly selected driver gene mutation versus multiple is equally informative to describe adequately aCRC patients’ outcome under regorafenib after 14 days of treatment onset. Especially, combined with pre-treatment ctDNA levels, this simplifies a personalized patient monitoring. 1 P. Kehagias, et al. Circulating tumor DNA detects early response to regorafenib in advanced colorectal cancer [abstract]. AACR; Cancer Res 2018;78(13 Suppl):Abstract nr LB-221
Author: Hestia Mellert | Biodesix Inc., Boulder, CO
The adoption of third-generation tyrosine kinase inhibitors (TKIs) to treat non-small cell lung cancer (NSCLC) for EGFR positive patient cases has led to the emergence of acquired resistance pathways. Approximately 40% of EGFRT790M-positive NSCLC cases display an acquired mutation of amino acid 797 (C797S) that mediates this resistance, which can develop after a median of 10 months following treatment. The effectiveness of treatment for patients with EGFRC797S may depend on whether the mutation is present in cis or trans with EGFRT790M, the type of sensitizing mutation (EGFRdel19 or EGFRL858R), and previous treatment(s). Patients with EGFRdel19 or EGFRL858R in conjunction with EGFRT790M and EGFRC797S remain resistant to all current targeted therapies. Several fourth-generation inhibitors are in development but have yet to be validated in clinical trials. We have developed a blood-based test that can detect two of the most common C797S nucleotide mutations (T>A and G>C) in cis or trans with EGFRT790M using the droplet digital PCR (ddPCR) technology and assays. Analytic sensitivity and specificity were assessed using synthetic DNA designed to mimic the EGFRC797S variants and their locus detected in the assay. Normal healthy donors as well as reference ddPCR positive and negative NSCLC donor samples were assessed for clinical specificity and sensitivity. Finally, the precision of the assay was evaluated with both clinical and analytical samples. Specifically, we evaluated the assay at high, medium, and low mutation frequencies over three consecutive days, including repeat runs on one day, and with multiple operators. This assay is capable of accurately and precisely detecting multiple EGFRC797S variants and their locus from blood specimens in the clinical laboratory. Consistent with other ddPCR blood-based EGFR variant assays we have developed, the limit of detection was 0.02%for C797S variants detected in both cis and trans with the T790M mutation. Once validated, the EGFR C797S assay described here may be of utility in diagnosing patients who have developed resistance to osimertinib.
Author: Liana Nobre | SickKids Hospital, Toronto, ON, Canada
The use of liquid biopsies to diagnose and monitor tumors has been widely explored especially in adult cancers. In pediatric cancers, the feasibility of this method as a clinical tool is still to be established. Pediatric low grade gliomas (PLGG) are characterized by mutations in the MAPK pathway, and 20% harbour BRAFV600E mutation, which can also be found in paediatric patients with Langerhans cell histiocytoma (LCH), thyroid carcinoma and melanomas. Specific histone mutations have been described in pediatric high grade gliomas(HGG) and diffuse midline pontine gliomas(DIPG). Midline gliomas are not amenable to a gross total resection; however, a biopsy is needed for pathological and molecular diagnosis. In this scenario, liquid biopsy is of utmost importance and has the potential to spare the risk of morbidity with surgical procedures, determine diagnosis and prognosis, as well as serve as a tool to guide and monitor response to therapy with targeted agents. The objective of this study is to evaluate the use of droplet digital PCR (ddPCR) for identification of point mutations in cerebro-spinal fluid (CSF) or plasma of BRAF V600E or H3K27M positive pediatric patients. CSF was collected from a total of 51 patients, and plasma from 55 patients. ctDNA was extracted from 3 ml of plasma or 2 ml of CSF, and pre-amplified prior to ddPCR, which was conducted on the RainDance system. For BRAFV600E: Forty-five patients had CSF samples available for analysis, 8 had known positive brain tumors, 32 negative brain tumors, 5 normal controls; 5/8 positive cases had BRAFV600E mutation in ctDNA from CSF, one of the patients with negative CSF and positive tumor was on treatment with BRAF inhibitors. Plasma was analysed in 17 patients with known positive brain tumors and additionally 10 patients negative for BRAFV600E, all of the plasma samples were negative for the mutation. For H3K27M: Thirty patients had CSF samples for analysis 1 was a known positive tumor, 27 negative and 2 unknown status (radiological diagnosis of DIPG). Of the 3 DIPG samples 2 were positive, one with unknown status of the primary tumor. Plasma was also analysed for H3K27M mutation in 7 patients with known positive brain tumors, one DIPG with unknown status and 7 negative samples. All plasma samples tested negative for the mutation. Sensitivity and Specificity for this assay was respectively 62% and 97% for BRAF and 66% and 98 % for H3K27M in CSF. The sensitivity in plasma for brain tumors is poor, however seems better for non-CNS lesions. In summary, we show that liquid biopsy with analysis of CSF ctDNA is feasible with high specificity, important in the context of a diagnoses tool. Moving forward larger cohorts need to be validated towards the goal of implementing ddPCR as a clinical tool for diagnosis of specific point mutations in pediatric patients with CNS tumors.
Author: David Polsky | New York Univ. School of Medicine, New York, NY
Purpose: Challenges to accurate quantitative measurements of circulating tumor DNA (ctDNA) include variable contamination with large DNA fragments released during mononuclear cell lysis, and potential variability in extraction efficiency. We investigated optimal procedures for plasma processing and the potential utility of a DNA spike-in as a process control for extraction efficiency.
Methods: Initial experiments analyzed the potential benefits of different centrifugation protocols (1600 x g vs. 1600 x g followed by 16000 x g) for plasma separation by analyzing the size distribution of extracted DNA using a BioAnalyzer. In subsequent experiments we used droplet digital PCR (ddPCR) to compare differences in extraction efficiency between nucleosomal DNA (nDNA) and genomic DNA (gDNA) spiked into healthy donor plasma and extracted using the QIAamp DSP Circulating NA Kit (Qiagen). To evaluate the properties of the purification columns with respect to the yield of different DNA sizes, we compared the plasma results to experiments extracting purified nDNA and gDNA spiked into PBS instead of plasma. We investigated the potential competitive effect of large DNA on the extraction efficiency of small DNA by mixing nDNA with commercially available human gDNA in different concentrations. To assess if nDNA, when spiked into patient plasma, would track with endogenous ctDNA, we spiked a pool of plasma from patients with BRAFV600E mutations with known quantities of nDNA purified from NRASQ61K mutant cells (CHP212). The spiked plasma pool was divided and extracted using multiple separate columns. Copies/ml and mutant fraction were determined by ddPCR assays for BRAFV600E and NRASQ61K. Coefficients of variation were also calculated.
Results: Comparison of different centrifugation speeds for plasma processing before and after freeze-thaw indicated that a 2nd high-speed centrifugation (16000 x g), either before or after a freeze-thaw cycle, minimizes large DNA in plasma samples prior to ctDNA extraction. In initial spike-in experiments extraction efficiency for nDNA was 87.7% compared to 42% for gDNA. We obtained similar results when purifying the nDNA and gDNA from PBS instead of plasma. We observed greater variability in nDNA extraction efficiency when increasing amounts of gDNA were added as a contaminant. Finally, we found that the slight, paralell variations in extraction yields of externally spiked NRASQ61K mutant nDNA and endogenous BRAFV600E ctDNA in 7/9 replicate extractions. The overall coefficients of variation (total copies/mL) for 9 extractions for BRAFV600E and NRASQ61K were 3.9 and 4.2 respectively.
Conclusion: A second centrifugation step helps reduce potential contamination of ctDNA with large DNA, and may improve the recovery of ctDNA from plasma samples. Nucleosomal DNA can potentially be used as a process control for extracting ctDNA from plasma samples.
Author: Zhigang Kang | National Cancer Inst., Bethesda, MD
Purpose: Circulating cell-free (cf) human papillomavirus (HPV) DNA is a unique tumor marker for metastatic cervical cancers. We developed a method to genotype and quantify blood circulating HPV DNA in patients with metastatic cervical cancer for patient selection, treatment monitoring, as well as informing data on an experimental T cell therapy.
Patients and Methods: we developed a digital droplet (dd) PCR method for HPV genotyping and quantification with cfDNA. In a retrospective study, HPV cfDNA was measured in serum samples from nine metastatic cervical cancer patients received tumor-infiltrating lymphocyte (TIL) immunotherapy. cfDNA data were aligned with the tumor HPV data, drug treatment, and clinical outcome. In a HPV screening, the genotyping and DNA copy number of HPV cfDNA from 47 cervical cancer patients were measured and the allele frequencies were determined in relative to total genomic cfDNA of each patient.
Results: The ddPCR assay is highly sensitive, specific, and capable of accurate quantification of both HPV16 and HPV18 cfDNA. In the clinical validation, we detected HPV cfDNA from 19/19 (100%) patients with HPV-positive metastatic and recurrent cervical cancer but not in any of the 45 healthy blood donors. The HPV genotype harbored in the patients’ tumors was correctly identified in 87/87 (100%) sequential patient serum samples. Our data showed the use of HPV cfDNA for pharmacokinetic studies, effectiveness on an investigational therapy, and patient monitoring. In three patients who experienced objective responses, a transient HPV cfDNA peak was detected 2-3 days after TIL immunotherapy and a persistent clearance of HPV cfDNA was observed in only two patients with a complete response (CR). Among the HPV positive cases, the median HPV cfDNA allele frequency is above 30% of genomic single copy genes in these cervical cancer patients.
Conclusions: We developed and validated a highly sensitive and specific HPV cfDNA genotyping and quantification method. As a promising non-invasive tumor marker, HPV cfDNA may have value in detecting the efficacy of therapeutic agents and monitoring cervical cancer patients in remission. The cfDNA based HPV genotyping is currently being evaluated for patient selection in an experimental T-cell therapy against a specific HPV antigen. HPV cfDNA may also be a potential biomarker for early diagnosis of cervical cancer.