Detection of Multiple HPV Types in Liquid Biopsies of Cervical Neoplasia

Abstract

Background

More than 95% of cervical cancers and their precancerous lesions are caused by human papillomavirus (HPV). Cell-free (cf) HPV DNA detection in blood samples may serve as a monitoring tool for cervical cancer.

Methods

In our methodological study, an HPV panel for simultaneous detection of 24 types using mass spectrometry-based analysis was developed for liquid biopsy approaches and tested on HPV positive cell lines, plasmid controls, and cervical high-grade squamous intraepithelial lesions (HSIL) in positive smear samples (n = 52). It was validated in cfDNA blood samples (n = 40) of cervical cancer patients.

Results

The HPV panel showed proficient results in cell lines and viral plasmids with a limit of detection of 1 IU (international units)/µL for HPV16/18 and 10GE/µL for HPV11/31/33/39/45/51/52/58/59 and a specificity of 100% for the tested HPV types. In cervical smear samples, HPV DNA was detected with a sensitivity of 98.14%. The overall agreement between the new HPV panel and clinical records was 97.2% (κ = 0.84). In cervical cancer cfDNA, 26/40 (65.0%) tested positive for any HPV type, with most infections due to hrHPV (24/26). HPV positive samples were found in all FIGO stages, with the highest positivity ratio in FIGO III and IV. Even the lowest stage, FIGO I, had 12/23 (52.2%) patients with a positive HPV plasma status.

Conclusions

This proof-of-concept paper shows that the described assay produces reliable results for detecting HPV types in a multiplex mass spectrometry-based assay in cervical smear and cfDNA with high specificity and sensitivity in both cohorts. The assay shows potential for liquid biopsy-based applications in monitoring cervical cancer progression.

Introduction

Cervical cancer is the fourth leading cause of cancer-related deaths for women with 604 100 new cases worldwide in 2020 (1). Cervical cancer is caused by human papillomavirus (HPV) infections in >95% of the cases (2, 3). The large group of HPV types is divided into low- and high-risk with 12 different HPV types classified as high-risk HPV (hrHPV) (4) with HPV16 and HPV18 being responsible for 74% of all cervical cancers (5). The introduction of vaccination against 9 HPV types is one of the milestones in cancer research (6, 7) showing effective primary prevention against cervical cancer (8, 9). Still a large part of the world’s population, especially in developing countries, has not received the vaccination and therefore has a higher risk of developing cervical cancer (10). Whereas the different vaccines cover different HPV types, here detection of multiple HPV types also could be of importance in case of a shift regarding the most common HPV types in the future.

Cervical cancer develops through different stages of cervical intraepithelial lesions (CIN) from low-grade (LSIL/CIN1) to high-grade squamous intraepithelial lesions (HSIL/CIN2/CIN3). About 30% of HSIL transform into invasive cervical cancers (11, 12).

Cancer prevention guidelines applied to routine diagnostics of suspicious lesions are based on either liquid-based cytology (Pap smear) or HPV DNA testing (13). In many countries, HPV DNA-based testing is recommended today as the sole standard screening approach. Here, a variety of clinical molecular HPV DNA tests exist that cover different HPV types.

Liquid biopsy (LB) analysis using blood is a new approach to complement traditional diagnostic methods in cancer. Being minimally invasive LB has shown great promise for early detection and disease monitoring (14, 15). Plasma contains cell-free DNA (cfDNA), short DNA fragments (roughly 140 bp) originating from different organ sites that circulate in the blood. In cancer patients, this can also include circulating tumor DNA or viral DNA in viral-caused cancers (16). Viral HPV cfDNA offers a chance to be specifically detected and analyzed in patients with cervical neoplasia, but currently uses wide variability regarding methodology and breadth of HPV type detection, which leads to variability in overall detection (16–18). However, this approach is technically challenging due to the often very low cfDNA concentration in patient blood. Therefore, multiplexing approaches can help to increase the sensitivity (18, 19).

Our current research aims to detect the most common HPV types in one test, with high specificity and sensitivity suitable for both liquid-based cytology and LB approaches. The performance of this test is analyzed in 2 patient cohorts: an HSIL and a cervical cancer cohort.

Materials and Methods

Patient Samples

Patient samples were collected at the University Medical Center Hamburg-Eppendorf (UKE) from women attending the Gynecologic clinic between June 2019 and March 2022. Cervical smear samples were taken from 52 women before cervical conization prior to treatment. The indication for conization was an HSIL (41 CIN3, 11 CIN2). Data for clinical HPV status was received from clinical records using Anyplex™ II HPV28 Detection (Seegene) (Supplemental Table 1 in the online Data Supplement). The average age of this cohort was 39.8 years, and HPV vaccination coverage was 7.7%.

Blood samples were collected from 40 women with cervical cancer attending either the Department of Gynecology or Radiology at UKE. Here, the average age was 47.6 years, and HPV vaccination coverage was 0%. All tumor stages were normalized to the FIGO classification 2019 (20) (Supplemental Table 2).

Healthy donor (HD) blood samples were collected in EDTA-tubes from 52 women attending the Blood Donation Center at UKE. The average age was 38.2 years. Data about the HPV vaccination status was not available.

All participants gave their written informed consent, which was approved by the local ethical committee (no. PV-5392, 06/12/2016, Ärztekammer Hamburg). The study was conducted in accordance with the Declaration of Helsinki.

DNA Extraction

DNA was extracted from cervical smear samples, stored in BD SurePath solution (BD) at ambient temperature, and processed within 4 h after collection using the QIAamp MinElute Media Kit (Qiagen). The protocol was scaled up to double the volumes for higher DNA yield, allowing a total input of 500 µL of SurePath smear emulsion by leaving the elution volume at 50 µL.

Plasma from cervical cancer patients and HD was extracted from 15 mL EDTA blood using a 2-stage centrifugation protocol (10 min at 300g; followed by 10 min at 1800g) and stored at −80°C until further processing. Total plasma was used for DNA extraction using the QIAamp Circulating Nucleic Acid Kit (Qiagen) as described in the manufacturer's protocol.

DNA concentrations were determined using Qubit 4 and the Invitrogen Qubit dsDNA HS-Assay (Thermo Fisher Scientific) and DNA was stored at −20°C until further use.

Mass Spectrometry-Based Assay

HPV measurement was performed using the Agena Bioscience HPV Genotyping Panel v.2.0 on the MassARRAY® System, a matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) mass-spectrometer (Agena Bioscience) (21). This full genotyping assay can detect the status of 24 HPV types in one multiplex assay, including all 12 hrHPV, 1 probably carcinogenic, 7 possibly high-risk, and 4 low-risk HPV plus an internal control glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (4) (Supplemental Table 3).

The multiplex PCR-based approach consists of 3 steps: PCR amplification, shrimp-alkaline-phosphatase (SAP) reaction, and iPLEX Pro Extension Reaction (Agena Bioscience). In the published standard protocol, the maximum DNA input volume is 2 µL (21). To process higher DNA amounts of the low-concentrated cfDNA samples, PCR amplification was performed as an upscaled 50 µL of PCR reaction first. Because the sensitivity of the standard protocol was not achieved (data not shown), a pooling protocol was established: We performed 4-fold parallel initial multiplex PCRs, with 4 reactions per sample containing 2.5 µL of DNA and 3 µL of PCR mix each. For SAP reaction, the 4 wells were pooled (total volume of 22 µL + 8 µL of SAP mix). Following this, 7 µL of the mix were combined with 2 µL of extension mix and processed according to the standard protocol.

PCR cocktails were mixed according to the manufacturer’s standard protocol with the iPLEX Pro Reagent Set. Total DNA input was 5–40 ng DNA for cervical smear samples and 10 ng for cfDNA of cervical cancer patients. Due to low contractions of cfDNA in HD, only samples with a total DNA input of a minimum of 3 ng cfDNA (mean 6.9 ng, range 3–10 ng) and passed quality approval by Liquid IQ (Agena Bioscience) (22) were used (31 of 56 HD samples passed). For the assay panel, premixed HPV PCR primers were used. Primers are based on type-specific sequences in the genomic E6/E7 region for 24 types. All HPV types (except HPV82: 151 bp) had PCR amplicons below 140 bp.

The multiplex PCR-based assay was performed according to the manufacturer’s standard protocol as published before (21). The different masses of the HPV types in the samples were determined and measured by the mass spectrometry-based analyzer (MassARRAY Analyzer 4, Agena Bioscience). An automated report is launched from within the MassARRAY Typer software and provides all positive and negative HPV types. The panel contains an internal process control targeting GAPDH in each reaction. In each run, positive controls with known HPV types (see next) were included as well as negative controls.

All cervical smear DNA samples were tested in triplicates, all cervical cancer cfDNA samples were tested using the pooling protocol in single runs [in case of enough cfDNA amount, duplicate runs (n = 14/40)]. A smear sample result was considered positive if >1 of 3 replicates were positive. A cervical cancer sample result was considered to be positive if at least 1 (in replicates) was positive.

Control Samples

DNA from cervical cancer and head-and-neck squamous cell carcinoma cell lines were used as positive controls: UPCI-SCC-154 (HPV16 positive), UM-SCC-47 (HPV16 positive), HeLa (HPV18 positive), and UT-SCC-45 (HPV33 positive). DNA from UPCI-SCC-154, UM-SCC-47, and UT-SCC-45 was kindly provided by Thorsten Rieckmann, Department of Radiotherapy and Department of Otorhinolaryngology, UKE. HeLa was obtained from ATCC. Cell line DNA was diluted in human placental wild-type (WT) DNA. To determine further type specificity, samples of the Global HPV LabNet DNA Genotyping Proficiency Panel 2021 were used containing purified plasmids of 15 different HPV types, as singles as well as combined (Table 1), diluted in a background of 50 ng/5 µL WT DNA. HPV types that were detected at 50 international units (IU) per 5 µL of HPV16/HPV18 and 500 genome equivalents (GE) per 5 µL for the other 14 HPV types in both single and multiple infections were considered proficient (23). In our study, we also diluted the samples to 10-fold lower concentrations, respectively. All cell lines and purified plasmids were tested in triplicates for the standard protocol. In the pooling protocol cell line samples were tested in triplicates and purified plasmids in duplicates. The acceptability criterion for limit of detection (LoD) was at 100% of the specimens detected.

Statistics

Statistical analyses were performed using IBM SPSS Statistics v.29. The chi-squared test and Fisher exact test with a two-sided significance level of P < 0.05 were used to determine significant differences between patient groups and HPV status. The Spearman Rho test was used to determine correlations in ordinal (CIN grade, Pap status, FIGO stage) and metric variables (DNA input, age). The concordance between MassARRAY and Anyplex HPV testing was determined using Cohen kappa statistics using the following definitions: poor 0.00–0.20; fair 0.21–0.40; moderate 0.41–0.60; good 0.61–0.80; and very good 0.81–1.00 (24).

Results

HPV Panel Sensitivity and Specificity Testing Using Cell Lines and Global HPV LabNet DNA Samples

To establish the HPV panel suitable for LB approaches, cell lines and plasmid samples with known HPV types were tested in different dilutions (Table 1). The measurement by mass spectrometry distinguishes between the different HPV types based on their DNA mass. Figure 1A shows the spectra with positivity for HPV18 (peak at approximately 5500 Dalton).

The results of the Global HPV LabNet DNA Genotyping Proficiency Panel 2021 samples were already published as part of a ring trial (23). In this ring trial, the HPV types HPV6, 11, 16, 18, 31, 33, 35, 39, 45, 51, 52, 58, and 59 were considered proficient with the required detection level of 50 IU/5 µL for HPV16 and HPV18, and 500 HPV GE/5 µL for the other HPV types as described by Mühr et al. (23). An even better LoD of 1 IU/µL for HPV16 and HPV18 and 10 GE/µL for the other tested HPV types was achieved in our setting (Table 1). LoD was maintained when variable mixtures of HPV plasmids were tested. Only HPV56 was not detectable and therefore not proficient. HPV68 was unable to perform because the plasmid provided in the study did not contain the target region for which the PCR was designed but it was tested previously (21). There were no false positive results at all, which provides a 100% specificity.

In cell lines, the HPV Genotyping Panel detected HPV16 down to 0.1% of the viral DNA in a background of human WT DNA, both in the low copy number cell line UPCI-SCC-154 (one virus copy per β-globin copy) as well as in the higher copy numbers cell line UM-SCC-47 (15 virus copies per β-globin copy) (25). HeLa (HPV18 positive) and UT-SCC-45 (HPV33 positive) showed comparable sensitivities of 0.1% in the standard protocol. Specificity in all cell lines was 100%.

Establishment of a Liquid Biopsy Compatible HPV Assay

The pooling approach (4 × 2.5 µL of DNA input) was established with 21 different types of samples from cell lines, smear DNA, plasmids, and human WT DNA (Table 1). Smear DNA samples were tested using both protocols and showed a 100% concordance for cervical smear DNA samples of HSIL patients with the HPV types HPV16, 18, 31, 45, 52, 66, and 73. Specificity was 100% in all tested samples (data not shown). For the cell lines and purified plasmids comparable accuracy and limits of detection were reached for both protocols with an exception: for a plasmid sample containing both HPV6 and HPV35 the pooling protocol showed diminished sensitivity of 50% with 10 GE/µL. As HPV56 could not be reliably detected using either protocol, as well as HPV6 and HPV35 using the pooling protocol, these types were excluded from the correspondent analyses (Table 1).

HPV Detection in Smear Samples of Women with Cervical Neoplasia

Here, 52 smear samples from HSIL (CIN2 and CIN3) were tested first using the standard protocol with 2 µL of DNA. The initial DNA amount showed no significant correlation to the detection status (P > 0.05, Spearman Rho).

HPV DNA was detected in 94% (49/52) smear samples, with the highest prevalence for one of the hrHPV (86.5%) types in the study cohort (Fig. 1B). In 6% (3/52) the sample was negative for all tested HPV types; 2 out of 3 cases were true HPV negative according to clinical data. The third case was according to clinical records positive for HPV16 and 18. A sensitivity of 98.14% and a positive predictive value of 0.98 (49/50) were thus obtained.

For patients who tested HSIL positive before surgery, the MassARRAY HPV test detected 39/41 HPV positive patients in the CIN3 cohort (sensitivity 95.1%) and 10/11 in CIN2 (sensitivity 90.9%) (Fig. 1C). The overall distribution of the different HPV types is shown in Fig. 1D. HPV16 is detected in about half of the samples (25/52). In 15 samples, multiple HPV types were positive, with positivity in up to 4 different HPV types (Supplemental Table 4). In the study cohort, no significant correlation to clinical factors was found (Pap status, CIN grade before and after surgery, smoker, contraception, vaccination status, parity, age; data not shown).

The overall agreement between the MassARRAY and clinical detection with Anyplex HPV tests was 97.2%, with a strong level of agreement (κ = 0.84) (Table 2). The agreement for the individual types is shown in Table 2. No clinical reference data was available for HPV 34, 67, and 81 because these types were not tested in Anyplex. The MassARRAY showed for HPV16 a positive agreement of 82.6% and a negative agreement of 87.0%, HPV18 71.4% and 97.4%, HPV31 100% and 100%, and HPV33 100% and 100%, respectively.

HPV Detection in LB Samples of Cervical Cancer Patients

Out of 40, 26 (65.0%) of cfDNA samples of cervical cancer patients tested positive for any HPV type (Table 3). The distribution of HPV types classified by IARC groups showed most infections are due to hrHPV (24/26 plasma HPV positive) (Fig. 2A). The HPV MassARRAY test showed a sensitivity of 65.0% (95% CI = 50.0–80.0%). The sensitivity was defined as the proportion of patients with HPV cfDNA presence among all patients confirmed as having cervical cancer. Divided by 3 different therapy statuses, there were higher HPV rates in patients before treatment (baseline) (63.6%) and under therapy (chemotherapy/radiotherapy) (68.8%) than in those without any treatment after surgery (50%) (Fig. 2B). Regarding the FIGO stage, HPV positive samples were found in all FIGO stages, with the highest positivity ratio in FIGO III and IV (Fig. 2C). Still, 12/23 (52.2%) of FIGO stage I patients had a positive HPV status in their blood analyses. The mean cfDNA concentration increased with higher FIGO status (FIGO I 2.45 ng/µL; FIGO IV 4.02 ng/µL). However, this difference was not significant (P > 0.05). MassARRAY detection showed positivity for HPV16, 18, 33, 39, 45, 66, 68, and 73; with highest prevalence for HPV16 (Fig. 2D). Details regarding multiple HPV types are shown in Supplemental Table 4. There was no significant correlation in the tested clinical parameters (P > 0.05, n = 40).

Our results in comparison to clinical data are shown in Fig. 3. Clinical data for HPV status was available in only n = 26/40 patients for all HPV types, for HPV16 (n = 29), and HPV18 (n = 27) (Supplemental Table 2). Overall agreement between MassARRAY and recorded clinical data regarding the HPV types was 93.6%, for detailed analysis of each HPV type confer Supplemental Table 5. Agreement for HPV16 was 69.0%, HPV18 74.1%, HPV33 76.9%, and HPV45 96.2%.

The MassARRAY Panel did not detect any HPV in 20 HSIL cfDNA samples (data not shown). Twenty-eight of 31 tested cfDNA of HD were negative for all HPV types (90.3%). The 3 HPV positive HD were HPV16 (n = 2) and HPV18 (n = 1) positive. It was not possible to analyze the HSIL and HD samples in duplicates due to low DNA amounts in the samples. A significant association between cervical cancer patients and positive HPV LB status was found (P < 0.001).

Known cofactors for HPV infections, such as oral contraceptives, a higher number of births, smoking, or immunosuppression (1), showed no significant correlation with the HPV status in both the HSIL and cervical cancer cohorts (data not shown).

Discussion

Different virus-based LB assays have shown promising results for disease monitoring but also for early cancer screening approaches such as Epstein–Barr virus DNA detection in head-and-neck cancer (16, 18, 26). Also, HPV promises analogous results. Several digital-droplet-PCR (ddPCR)-based approaches have been published mainly detecting HPV16 and HPV18 [reviewed in (18, 19)]. These 2 hrHPVs are responsible for only 74% of all cervical carcinomas and are covered by the current HPV vaccine (5). Thus, a multimarker approach would be advisable to cover more HPVs, including noncovered-by-vaccine hrHPVs (27, 28). Such an assay needs to provide high sensitivity due to mostly low cfDNA amounts and will optimally also be time and cost-effective. Therefore, the main aim of this methodological work was to develop such an LB assay for simultaneous detection of multiple HPV types in cervical neoplasia.

We successfully established a panel for 24 types of HPV using smear samples of HSIL patients and ensured the high sensitivity and specificity further with HPV cell lines and plasmid controls. Our new pooling protocol combined the possibility of a higher DNA input with the sensitivity and specificity of the standard protocol. This allows using it with samples containing low DNA concentrations such as cfDNA. Variations of the standard protocol used in our study have been successfully used in tissue (21, 29–31) before. Pedersen et al. tested a MassARRAY HPV panel with 19 HPV types recently in n = 1294 cervical smear samples, where it showed high sensitivity of 95.2% for detection of HSIL lesions (21).

In our study, this method performed well for HPV detection in control samples as well as in HSIL smear samples. The LoD for HPV16 and HPV18 was down to 0.1% in cell lines and 1 IU/µL in viral purified plasmids. The LoD for other HPV plasmids was 10 GE/µL. Our panel showed a high sensitivity of 98.14% in HSIL smear samples, validating previous results (21). The specificity was at 100%. HPV16 was detected in 47.8% of the samples, reflecting general population-wide HPV distribution (32). The overall agreement between MassARRAY and clinical Anyplex data was 97.2%, with a strong level (κ = 0.84) for type-specific agreement. The assay thus offers a specific and sensitive measurement tool for HPV detection, comparable with other diagnostic tools. Furthermore, it can be processed in a single well reaction with an easy and fast detection system, short hands-on-time (8 h), and easy scalability.

In cfDNA, our LB assay achieved a detection rate of 65%. A recent meta-analysis combined data from 6 different studies with only HPV16/HPV18 cfDNA in blood of 631 cervical cancer patients and showed a pooled sensitivity of 43.8% and specificity of 96.6% for HPV16 and 25.7% and 99.0% for HPV18, respectively (19). Other studies using either ddPCR or next-generation sequencing (NGS) HPV genotyping for HPV16 and/or HPV18 have reported partially higher detection rates compared to our study (61%–88%) (18, 33–37). In contrast to ddPCR, the MassARRAY-based method, like NGS (37), allows multiplexing for simultaneous multiple HPV type detection. NGS-based assays, however, are often rather expensive and require elaborate equipment, good bioinformatics, and data storage infrastructure. Importantly, our study showed that at an early tumor stage (FIGO I), a high HPV positivity (52.2%) detection rate could be seen. This implies high sensitivity enabling future early detection and especially minimal residual disease monitoring.

Obviously, larger studies need to be performed to validate the clinical utility of this and other assays for early detection and disease monitoring. In particular, this study did not include any follow-up samples for assessing treatment response and disease relapse. Furthermore, the low cfDNA levels in HSIL patients did not permit analysis regarding the screening of HSIL blood samples. Therefore, future sensitivity studies are needed to determine whether HSIL patients shed detectable levels of HPV cfDNA into the circulation (38). Finally, and surprisingly, of the 31 HD blood samples, 3 showed a positive HPV status indicating that perhaps repeated sampling is needed to ensure correct status. Also, other studies have similarly detected HPV positive cfDNA in HD (0%–16%) [reviewed in (39)]. HD positivity could also reflect a true HPV positivity, due to high prevalence of HPV infections, with 34.4% of any hrHPV in unvaccinated women in Germany (40); or it could reflect other HPV-related diseases such as oropharyngeal carcinoma or genital warts.

A high overall agreement of 97% in smear and 94% in cfDNA was found between the clinically recorded HPV status and the new assay. Based on the Global HPV LabNet DNA Genotyping Proficiency Panel 2021 results, Anyplex-based detection was slightly more sensitive but MassARRAY-based detection was more specific (23). A discordant HPV status could derive from detection of HPV infections at very low viral loads without any cutoff values in the MassARRAY testing system.

In conclusion, our mass spectrometry-based detection of HPV is a specific and sensitive test using cervical smear DNA from patients with HSIL lesions. In cfDNA blood samples of patients with cervical cancer, it could offer the opportunity for disease monitoring once the cancer occurs. Further studies including follow-up samples will help to define the stability of this assay. Detection of cfDNA of HSIL patients still lacks sensitivity. In the future, HPV vaccines will eliminate more of the typical oncogenic HPV types while other types still lead to cervical lesions (27). Through its potential to detect a wider HPV spectrum including nonvaccine HPV types, fast and simple multimarker approaches will probably gain more importance.

Supplemental Material

Supplemental material is available at Clinical Chemistry online.

Nonstandard Abbreviations

HPV, human papillomavirus; HSIL, high-grade squamous intraepithelial lesion; IU, international units; FIGO, Fédération Internationale de Gynécologie et d'Obstétrique; hrHPV, high-risk human papillomavirus; CIN, cervical intraepithelial neoplasia; LB, liquid biopsy; cfDNA, cell-free DNA; UKE, University Medical Center Hamburg-Eppendorf; HD, healthy donors; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; SAP, shrimp-alkaline-phosphatase; LoD, limit of detection; GE, genome equivalents; WT, wild-type; ddPCR, digital-droplet PCR; NGS, next-generation sequencing.

Author Contributions

The corresponding author takes full responsibility that all authors on this publication have met the following required criteria of eligibility for authorship: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; (c) final approval of the published article; and (d) agreement to be accountable for all aspects of the article thus ensuring that questions related to the accuracy or integrity of any part of the article are appropriately investigated and resolved. Nobody who qualifies for authorship has been omitted from the list.

Johanna Herbst (Conceptualization-Equal, Data curation-Equal, Formal analysis-Equal, Funding acquisition-Equal, Investigation-Lead, Methodology-Equal, Validation-Equal, Visualization-Lead, Writing—original draft-Lead, Writing—review & editing-Equal), Vanessa Vohl (Conceptualization-Equal, Data curation-Equal, Formal analysis-Equal, Investigation-Equal, Methodology-Equal, Validation-Equal, Visualization-Equal, Writing—original draft-Equal, Writing—review & editing-Equal), Maroje Krajina (Methodology-Supporting, Software-Equal, Writing—review & editing-Equal), Markus Leffers (Methodology-Equal, Writing—review & editing-Equal), Jolanthe Kropidlowski (Methodology-Equal, Writing—review & editing-Equal), Katharina Prieske (Resources-Equal, Writing—review & editing-Equal), Anna Jaeger (Investigation-Supporting, Resources-Equal, Writing—review & editing-Equal), Leticia Ferrer (Resources-Equal, Writing—review & editing-Equal), Barbara Schmalfeldt (Supervision-Equal, Writing—review & editing-Equal), Yvonne Goy (Resources-Equal, Writing—review & editing-Equal), Eike Burandt (Resources-Equal, Writing—review & editing-Equal), Klaus Pantel (Supervision-Equal, Writing—review & editing-Equal), Caren Vollmert (Software-Equal, Writing—review & editing-Equal), Alexander Sartori (Software-Equal, Writing—review & editing-Equal), Linn Woelber (Conceptualization-Supporting, Data curation-Equal, Investigation-Equal, Resources-Lead, Writing—review & editing-Equal), Katharina Effenberger (Conceptualization-Equal, Data curation-Equal, Formal analysis-Equal, Funding acquisition-Equal, Project administration-Equal, Supervision-Lead, Validation-Equal, Writing—original draft-Equal, Writing—review & editing-Equal), and Harriet Wikman (Conceptualization-Equal, Data curation-Equal, Formal analysis-Equal, Funding acquisition-Equal, Project administration-Equal, Supervision-Lead, Validation-Equal, Writing—original draft-Equal, Writing—review & editing-Equal)

Authors’ Disclosures or Potential Conflicts of Interest

Upon manuscript submission, all authors completed the author disclosure form.

Research Funding

This research was funded by AiF GmbH, a hypothecated project-executing organization of the Bundesministerium für Bildung und Forschung, Germany, #ZF4665601SB8 to K. Effenberger and H. Wikman and Werner Otto Stiftung, Promotionsstipendium to J. Herbst. J. Kropidlowski and K. Effenberger were partially funded by AiF GmbH research funding for this study.

Disclosures

M. Krajina, employee of Agena Bioscience GMBH, holds restricted stock units of Mesa Labs Inc., the owner of Agena Bioscience. A. Jaeger, personal fees from AstraZeneca, Molecular Health, GlaxoSmithKline (GSK), Roche, Clovis Oncology, and MSD outside the submitted work. E. Burandt, payment for lectures from Eisai; payment for lectures and advisory board from AstraZeneca, Daiichi Sankyo, and Novartis. K. Pantel, Associate Editor for Clinical Chemistry, Association for Diagnostics & Laboratory Medicine (formerly AACC). C. Vollmert, employee of Agena Bioscience GMBH, holds restricted stock units of Mesa Labs Inc., the owner of Agena Bioscience. A. Sartori, employee of Agena Bioscience GMBH, holds restricted stock units of Mesa Labs Inc., the owner of Agena Bioscience. L. Woelber has received honoraria, travel expenses, or personal fees from Roche, Eisai, Novartis, MSD, Seagen, GSK, AstraZeneca, Pfizer, Lumenis, TEVA, pomedicis, and Omniamed.

Role of Sponsor

The funding organizations played no role in the design of study, choice of enrolled patients, review and interpretation of data, preparation of manuscript, or final approval of manuscript.

References

Author notes