Quantification of human papillomavirus cell-free DNA from low-volume blood plasma samples by digital PCR

Abstract

The incidence rate of human papillomavirus-driven oropharyngeal cancer (HPV-OPC) is increasing in countries with high human development index. HPV cell-free DNA (cfDNA) isolated from 3 to 4 mL blood plasma has been successfully used for therapy surveillance. A highly discussed application of HPV-cfDNA is early detection of HPV-OPC. This requires sensitive and specific cfDNA detection as cfDNA levels can be very low. To study the predictive power of pre-diagnostic HPV-cfDNA, archived samples from epidemiological cohorts with limited plasma volume are an important source. To establish a cfDNA detection workflow for low plasma volumes, we compared cfDNA purification methods [MagNA Pure 96 (MP96) and QIAamp ccfDNA/RNA] and digital PCR systems (Biorad QX200 and QIAGEN QIAcuity One). Final assay validation included 65 low-volume plasma samples from oropharyngeal cancer (OPC) patients with defined HPV status stored for 2-9 years. MP96 yielded a 28% higher cfDNA isolation efficiency in comparison to QIAamp. Both digital PCR systems showed comparable analytical sensitivity (6-17 copies for HPV16 and HPV33), but QIAcuity detected both types in the same assay. In the validation set, the assay had 80% sensitivity (n = 28/35) for HPV16 and HPV33 and a specificity of 97% (n = 29/30). In samples with ≥750 µL plasma, the sensitivity was 85% (n = 17/20), while in samples with <750 µL plasma, it was 73% (n = 11/15). Despite the expected drop in sensitivity with decreased plasma volume, the assay is sensitive and highly specific even in low-volume samples and thus suited for studies exploring HPV-cfDNA as an early HPV-OPC detection marker in low-volume archival material.IMPORTANCEHPV-OPC has a favorable prognosis compared to HPV-negative OPC. However, the majority of tumors is diagnosed after regional spread, thus making intensive treatment necessary. This can cause lasting morbidity with a large impact on quality of life. One potential method to decrease treatment-related morbidity is early detection of the cancer. HPV cfDNA has been successfully used for therapy surveillance and has also been detected in pre-diagnostic samples of HPV-OPC patients. These pre-diagnostic samples are only commonly available from biobanks, which usually only have small volumes of blood plasma available. Hence, we have developed a workflow optimized for small-volume archival samples. With this method, a high sensitivity can be achieved despite sample limitations, making it suitable to conduct further large-scale biobank studies of HPV-cfDNA as a prognostic biomarker for HPV-OPC.

Keywords: HPV; OPC; cfDNA; digital PCR; early detection; liquid biopsy.

Fig 1

(A) Quantification of a dilution series of HPV16 E6 and beta-globin (BG) template DNA on a QX200 ddPCR system (x-axis) compared to QIAcuity dPCR (y-axis). Values shown are the minimum value out of two replicates of each sample. (B) Quantification of a dilution series of HPV16 and HPV33 DNA on QIAcuity dPCR by primers targeting HPV16 E7 and HPV33 E6. Values shown are the minimum of duplicates; the shaded area indicates the standard deviation. (C) Copy numbers of HPV16 E6 detected by QX200 ddPCR (y-axis) and QIAcuity dPCR (x-axis) in cervical swab samples with known HPV status. The red lines indicate the cutoff of three positive partitions to distinguish positive and negative samples. The black line is a regression line fitted to the data (y = 0.11 + 0.93x, r = 0.97). Not all colors are visible due to data overlay at zero copies. (D) Copy numbers of HPV16 E7 and HPV33 E6 detected by QIAcuity dPCR in cervix samples with known HPV status. The red line indicates the cutoff of three copies to distinguish positive and negative samples.

Fig 2

dPCR results of 24 patients from the training set (10 HPV−, 12 HPV16+, and 2 HPV33+). Samples categorized as positive or negative, depending on the HPV type targeted by each experiment. (A) Copy number of HPV16 E6 cfDNA by QX200 ddPCR, showing the mean value of the replicates. (B) Copy number of HPV16 E6 cfDNA by QX200 ddPCR, showing the minimum value of the replicates. (C) Copy number of HPV16 E6, HPV16 E7, and HPV33 E6 cfDNA by QIAcuity dPCR, showing the minimum value of the replicates.

Fig 3

Number of copies of different target DNA detected in plasma samples from 24 OPC patients by QIAcuity dPCR. Each point represents one of two replicates from a sample. (A) Comparison of HPV16 E6 and HPV16 E7 copies detected in each sample. (B) Comparison of copy number of HPV16 E7 and BG copies in each sample. (C) Comparison of number of copies of BG detected and the input blood plasma volume.

Fig 4

Performance of nucleic acid isolation from blood plasma. (A) Recovery rate of a 104-bp HPV16 fragment from EDTA plasma by MagNA Pure 96 (MP96) cfNA ss 2000 protocol or QIAamp ccfDNA/RNA kit (QIAamp). The red triangle indicates the position of the group mean. (B) Number of copies of native BG cfDNA isolated from aliquots of one EDTA plasma pool by MP96 and QIAamp. (C) Comparison of MagNA Pure 96 cfNA ss 2000 protocol and the Viral NA Universal 500 protocol for sodium citrate plasma.

Fig 5

QIAcuity dPCR result of 30 OPC patient plasma samples (13 negative, 2 HPV58 positive, and 15 HPV16 positive), comparing the concentration of each target (in copies/mL plasma) between cfDNA isolated from 500 µL (y-axis) and 1,000 µL (x-axis) of plasma. (A) HPV16 E7, (B) HPV16 E6, and (C) BG.