Trastuzumab and first-line taxane chemotherapy in metastatic breast cancer patients with a HER2-negative tumor and HER2-positive circulating tumor cells: a phase II trial

PURPOSE

HER2 overexpressing circulating tumor cells (CTCs) are observed in up to 25% of HER2-negative metastatic breast cancer patients. Since targeted anti-HER2 therapy has drastically improved clinical outcomes of patients with HER2-positive breast cancer, we hypothesized that patients with HER2 overexpressing CTCs might benefit from the addition of trastuzumab to chemotherapy.

METHODS

In this single-arm, phase II trial, patients with HER2-positive CTCs received trastuzumab as addition to first-line treatment with taxane chemotherapy. Patients with detectable CTCs but without HER2 overexpression that received taxane chemotherapy only, were used as control group. The primary outcome measure was progression-free rate at 6 months (PFR6), with a target of 80%. In November 2022, the study was terminated early due to slow patient accrual.

RESULTS

63 patients were screened, of which eight patients had HER2-positive CTCs and were treated with trastuzumab. The median number of CTCs was 15 per 7.5 ml of blood (range 1-131) in patients with HER2-positive CTCs, compared to median 5 (range 1-1047) in the control group. PFR6 was 50% in the trastuzumab group and 54% in the taxane monotherapy group, with no significant difference in median PFS (8 versus 9 months, p = 0.51).

CONCLUSION

No clinical benefit of trastuzumab was observed, although this study was performed in a limited number of patients. Additionally, we observed a strong correlation between the number of evaluable CTCs and the presence of HER2-positive CTCs. We argue that randomized studies investigating agents that are proven to be solely effective in the HER2-positive patient group in patients with HER2-positive CTCs and HER2-negative tissue are currently infeasible. Several factors contribute to this impracticality, including the need for more stringent thresholds, and the rapidly evolving landscape of cancer treatments.

Can we use a simple blood test to reduce unnecessary adverse effects from radiotherapy by timely identification of radiotherapy-resistant rectal cancers? MeD-Seq rectal study protocol

BACKGROUND

Chemoradiation therapy (CRT) followed by surgery is currently the standard of care to treat patients with locally advanced rectal cancer (LARC). CRT reduces local recurrences, but is associated with significant damage to the surrounding healthy tissue that can severely impact patients quality of life. Additionally, a proportion of patients (hardly) benefit from CRT. We aim to develop a diagnostic innovation, using DNA-methylation, which can enable a more selective and thereby more effective use of the available therapies for rectal cancer patients.

METHODS

MeD-Seq Rectal is a prospective single centre, observational study. 75 patients diagnosed with rectal cancer and will receive CRT as neoadjuvant treatment are will be included. DNA-methylation profiling will be performed on liquid biopsies to predict pathological response to CRT.

DISCUSSION

To data no clinical or image-based features were found that predict response to CRT. we hypothesize that DNA methylation patterns in liquid biopsies may provide a promising and patient-friendly strategy to predict CRT resistance upfront.

TRIAL REGISTRATION

This trial is registered at ClinicalTrials.gov (NCT06035471).

Integrative whole-genome and transcriptome analysis of HER2-amplified metastatic breast cancer

BACKGROUND

In breast cancer, the advent of anti-HER2 therapies has made HER2+ tumors a highly relevant subgroup. However, the exact characteristics which prohibit clinical response to anti-HER2 therapies and drive disease progression are not yet fully known. Integrative whole-genome and transcriptomic sequencing data from both primary and metastatic HER2-positive breast cancer will enhance our understanding of underlying biological processes.

METHODS

Here, we used WGS and RNA sequencing data of 700 metastatic breast tumors, of which 68 being HER2+, to search for specific genomic features of HER2+ disease and therapy resistance. Furthermore, we integrated results with transcriptomic data to associate tumors exhibiting a HER2+-specific gene expression profile with ERBB2 mutation status, prior therapy and relevant gene expression signatures.

RESULTS

Overall genomic profiles of primary and metastatic HER2+ breast cancers were similar, and no specific acquired genomics traits connected to prior anti-HER2 treatment were observed. However, specific genomic features were predictive of progression-free survival on post-biopsy anti-HER2 treatment. Furthermore, a HER2-driven expression profile grouped HER2-amplified tumors with ERBB2-mutated cases and cases without HER2 alterations. The latter were reported as ER positive in primary disease, but the metastatic biopsy showed low ESR1 expression and upregulation of the MAPK pathway, suggesting transformation to ER independence.

CONCLUSIONS

In summary, although the quantity of variants increased throughout HER2-positive breast cancer progression, the genomic composition remained largely consistent, thus yielding no new major processes beside those already operational in primary disease. Our results suggest that integrated genomic and transcriptomic analyses may be key in establishing therapeutic options.

Circulating Tumour DNA as Biomarker for Colorectal Liver Metastases: A Systematic Review and Meta-Analysis

Circulating tumour DNA (ctDNA) is a potential biomarker that could contribute to more judicious patient selection for personalised treatment. This review and meta-analysis gives an overview of the current knowledge in the literature investigating the value of ctDNA in patients with colorectal liver metastases (CRLM). A systematic search was conducted in electronic databases for studies published prior to the 26th of May 2023. Studies investigating the association between ctDNA and oncological outcomes in patients undergoing curative-intent local therapy for CRLM were included. Meta-analyses were performed to pool hazard ratios (HR) for the recurrence-free survival (RFS) and overall survival (OS). A total of eleven studies were included and nine were eligible for meta-analyses. Patients with detectable ctDNA after surgery experienced a significantly higher chance of recurrence (HR 3.12, 95% CI 2.27-4.28, p < 0.000010) and shorter OS (HR 5.04, 95% CI 2.53-10.04, p < 0.00001) compared to patients without detectable ctDNA. A similar association for recurrence was found in patients with detectable ctDNA after the completion of adjuvant therapy (HR 6.39, 95% CI 2.13-19.17, p < 0.0009). The meta-analyses revealed no association between detectable ctDNA before surgery and the RFS and OS. These meta-analyses demonstrate the strong association between detectable ctDNA after treatment and oncological outcomes in CRLM patients.

Hypermethylation of DNA Methylation Markers in Non-Cirrhotic Hepatocellular Carcinoma

Aberrant DNA methylation changes have been reported to be associated with carcinogenesis in cirrhotic HCC, but DNA methylation patterns for these non-cirrhotic HCC cases were not examined. Therefore, we sought to investigate DNA methylation changes on non-cirrhotic HCC using reported promising DNA methylation markers (DMMs), including HOXA1, CLEC11A, AK055957, and TSPYL5, on 146 liver tissues using quantitative methylation-specific PCR and methylated DNA sequencing. We observed a high frequency of aberrant methylation changes in the four DMMs through both techniques in non-cirrhotic HCC compared to cirrhosis, hepatitis, and benign lesions (p < 0.05), suggesting that hypermethylation of these DMMs is specific to non-cirrhotic HCC development. Also, the combination of the four DMMs exhibited 78% sensitivity at 80% specificity with an AUC of 0.85 in discriminating non-cirrhotic HCC from hepatitis and benign lesions. In addition, HOXA1 showed a higher aberrant methylation percentage in non-cirrhotic HCC compared to cirrhotic HCC (43.3% versus 13.3%, p = 0.039), which was confirmed using multivariate linear regression (p < 0.05). In summary, we identified aberrant hypermethylation changes in HOXA1, CLEC11A, AK055957, and TSPYL5 in non-cirrhotic HCC tissues compared to cirrhosis, hepatitis, and benign lesions, providing information that could be used as potentially detectable biomarkers for these unusual HCC cases in clinical practice.

Liquid Biopsies for Colorectal Cancer and Advanced Adenoma Screening and Surveillance: What to Measure?

Colorectal cancer (CRC) colonoscopic surveillance is effective but burdensome. Circulating tumor DNA (ctDNA) analysis has emerged as a promising, minimally invasive tool for disease detection and management. Here, we assessed which ctDNA assay might be most suitable for a ctDNA-based CRC screening/surveillance blood test. In this prospective, proof-of-concept study, patients with colonoscopies for Lynch surveillance or the National Colorectal Cancer screening program were included between 7 July 2019 and 3 June 2022. Blood was drawn, and if advanced neoplasia (adenoma with villous component, high-grade dysplasia, ≥10 mm, or CRC) was detected, it was analyzed for chromosomal copy number variations, single nucleotide variants, and genome-wide methylation (MeD-seq). Outcomes were compared with corresponding patients' tissues and the MeD-seq results of healthy blood donors. Two Lynch carriers and eight screening program patients were included: five with CRC and five with advanced adenomas. cfDNA showed copy number variations and single nucleotide variants in one patient with CRC and liver metastases. Eight patients analyzed with MeD-seq showed clustering of Lynch-associated and sporadic microsatellite instable lesions separate from microsatellite stable lesions, as did healthy blood donors. In conclusion, whereas copy number changes and single nucleotide variants were only detected in one patient, cfDNA methylation profiles could discriminate all microsatellite instable advanced neoplasia, rendering this tool particularly promising for LS surveillance. Larger studies are warranted to validate these findings.

Genomic Alterations Associated with Estrogen Receptor Pathway Activity in Metastatic Breast Cancer Have a Differential Impact on Downstream ER Signaling

Mutations in the estrogen receptor gene (ESR1), its transcriptional regulators, and the mitogen-activated protein kinase (MAPK) pathway are enriched in patients with endocrine-resistant metastatic breast cancer (MBC). Here, we integrated whole genome sequencing with RNA sequencing data from the same samples of 101 ER-positive/HER2-negative MBC patients who underwent a tumor biopsy prior to the start of a new line of treatment for MBC (CPCT-02 study, NCT01855477) to analyze the downstream effects of DNA alterations previously linked to endocrine resistance, thereby gaining a better understanding of the associated mechanisms. Hierarchical clustering was performed using expression of ESR1 target genes. Genomic alterations at the DNA level, gene expression levels, and last administered therapy were compared between the identified clusters. Hierarchical clustering revealed two distinct clusters, one of which was characterized by increased expression of ESR1 and its target genes. Samples in this cluster were significantly enriched for mutations in ESR1 and amplifications in FGFR1 and TSPYL. Patients in the other cluster showed relatively lower expression levels of ESR1 and its target genes, comparable to ER-negative samples, and more often received endocrine therapy as their last treatment before biopsy. Genes in the MAPK-pathway, including NF1, and ESR1 transcriptional regulators were evenly distributed. In conclusion, RNA sequencing identified a subgroup of patients with clear expression of ESR1 and its downstream targets, probably still benefiting from ER-targeting agents. The lower ER expression in the other subgroup might be partially explained by ER activity still being blocked by recently administered endocrine treatment, indicating that biopsy timing relative to endocrine treatment needs to be considered when interpreting transcriptomic data.

mFast-SeqS-based aneuploidy score in circulating cell-free DNA is a prognostic biomarker in prostate cancer

Multiple prognostic biomarkers, including circulating tumour cell (CTC) counts, exist in metastatic castration-resistant prostate cancer (mCRPC) patients, but none of them have been implemented into daily clinical care. The modified fast aneuploidy screening test-sequencing system (mFast-SeqS), which yields a genome-wide aneuploidy score, is able to reflect the fraction of cell-free tumour DNA (ctDNA) within cell-free DNA (cfDNA) and may be a promising biomarker in mCRPC. In this study, we investigated the prognostic value of dichotomized aneuploidy scores (< 5 vs. ≥ 5) as well as CTC counts (< 5 vs. ≥ 5) in 131 mCRPC patients prior to treatment with cabazitaxel. We validated our findings in an independent cohort of 50 similarly treated mCRPC patients. We observed that, similar to the dichotomized CTC count [HR: 2.92; 95% confidence interval (CI);1.84-4.62], dichotomized aneuploidy scores (HR: 3.24; CI: 2.12-4.94) significantly correlated with overall survival in mCRPC patients. We conclude that a dichotomized aneuploidy score from cfDNA is a prognostic marker for survival in mCRPC patients within our discovery cohort and in an independent mCRPC validation cohort. Therefore, this easy and robust minimally-invasive assay can be readily implemented as a prognostic marker in mCRPC. A dichotomized aneuploidy score might also be used as a stratification factor in clinical studies to account for tumour load.

Liquid Biopsies for Early Response Evaluation of Radium-223 in Metastatic Prostate Cancer

PURPOSE

Reliable biomarkers for response monitoring during radium-223 treatment in patients with metastatic castration-resistant prostate cancer (mCRPC) are lacking. Circulating tumor cells (CTCs) and circulating tumor DNA (ctDNA), obtained from liquid biopsies, are shown to have prognostic value in mCRPC. The aim of this study was to determine the value of CTCs and ctDNA for response evaluation of radium-223.

METHODS

In this prospective multicenter study, longitudinal blood draws and imaging were performed in 28 patients with mCRPC and predominantly bone disease, who were treated with radium-223. CTCs were counted (CELLSEARCH CTC test), while fraction of ctDNA was estimated by measuring aneuploidy of cell-free DNA (cfDNA; modified Fast Aneuploidy Screening Test-Sequencing System). CTC counts and aneuploidy score (AS) were categorized as low (<5) and high (≥5). Primary and secondary clinical end points were failure-free survival (FFS), and overall survival (OS) and development of extraosseous metastases, respectively. Additionally, CTC count and AS were related to alkaline phosphatase (ALP) and total tumor volume in bone (TTVbone) on positron emission tomography-computed tomography with 68gallium prostate-specific membrane antigen.

RESULTS

FFS was longer in patients with a low CTC count or AS either at baseline or after 12 weeks, whereas for OS, only a significant association with CTC count was observed. Liquid biopsy results correlated well with ALP and TTVbone at baseline, but not with change in both parameters after three cycles of radium-223. AS and CTC count were significantly correlated.

CONCLUSION

CTC count and AS of cfDNA at baseline and during treatment predict clinical response to radium-223 in patients with mCRPC, warranting future evaluation of their value in treatment guidance.

Tumor-agnostic ctDNA levels by mFAST-SeqS in first-line HR-positive, HER2 negative metastatic breast cancer patients as a biomarker for survival

This prospective cohort study reports aneuploidy score by mFast-SeqS as a strong prognostic marker in MBC patients. mFAST-SeqS is an affordable and easily implementable method for the assessment of total ctDNA levels and, as such, provides an alternative prognostic tool. One mixed cohort (cohort A, n = 45) starting any type of treatment in any line of therapy and one larger cohort (cohort B, n = 129) consisting of patients starting aromatase inhibitors (AI) as first-line therapy were used. mFAST-SeqS was performed using plasma of blood in which CTCs (CellSearch) were enumerated. The resulting aneuploidy score was correlated with categorized CTC count and associated with outcome. The aneuploidy score was significantly correlated with CTC count, but discordance was observed in 31.6% when applying cut-offs of 5. In both cohorts, aneuploidy score was a significant prognostic marker for both PFS and OS. In the Cox regression models, the HR for aneuploidy score for PFS was 2.52 (95% CI: 1.56-4.07), and the HR for OS was 2.37 (95% CI: 1.36-4.14). Results presented here warrant further investigations into the clinical utility of this marker in MBC patients.

Machine learning-based somatic variant calling in cell-free DNA of metastatic breast cancer patients using large NGS panels

Next generation sequencing of cell-free DNA (cfDNA) is a promising method for treatment monitoring and therapy selection in metastatic breast cancer (MBC). However, distinguishing tumor-specific variants from sequencing artefacts and germline variation with low false discovery rate is challenging when using large targeted sequencing panels covering many tumor suppressor genes. To address this, we built a machine learning model to remove false positive variant calls and augmented it with additional filters to ensure selection of tumor-derived variants. We used cfDNA of 70 MBC patients profiled with both the small targeted Oncomine breast panel (Thermofisher) and the much larger Qiaseq Human Breast Cancer Panel (Qiagen). The model was trained on the panels' common regions using Oncomine hotspot mutations as ground truth. Applied to Qiaseq data, it achieved 35% sensitivity and 36% precision, outperforming basic filtering. For 20 patients we used germline DNA to filter for somatic variants and obtained 245 variants in total, while our model found seven variants, of which six were also detected using the germline strategy. In ten tumor-free individuals, our method detected in total one (potentially germline) variant, in contrast to 521 variants detected without our model. These results indicate that our model largely detects somatic variants.

Abstract P2-03-14: Tumor-agnostic measurement of ctDNA levels in first line ER+, HER2- metastatic breast cancer patients as a marker for survival

INTRODUCTION: Circulating tumor cell (CTC) count and levels of circulating tumor DNA (ctDNA) have proven their individual prognostic value in metastatic breast cancer (MBC). Whereas CTCs can be quantified by enumeration, levels of ctDNA are often expressed as the variant allele frequency of the dominant mutation. The modified fast aneuploidy screening test-sequencing system (mFAST-SeqS) estimates ctDNA-load by measuring chromosomal aneuploidy instead of mutations. The mFAST-SeqS method is cheap and provides a comprehensive overview of the tumor genome without prior knowledge on genomic tumor aberrations. Previously, we showed in a small MBC cohort (n=50), heterogeneous with regard to subtype and line of treatment, that the mFAST-SeqS-derived genome-wide aneuploidy score (GWA-score) was an independent prognostic biomarker (1). Our current aim was to provide evidence for the prognostic value of GWA-score in a homogeneous cohort of estrogen receptor positive (ER+), HER2 negative (HER2-) MBC patients, starting first line treatment with an aromatase inhibitor.

MATERIALS & METHODS: CTCs were enumerated using the CellSearch-system (Menarini) and screened for ctDNA fraction using mFAST-SeqS in ER+, HER2- MBC patients, starting with first line aromatase inhibitors. CTC-count was divided into high (CTChigh) and low (CTClow) using the validated cut-off of 5 CTCs/7.5 ml blood; similarly GWA-score was divided in high (GWAhigh)and low (GWAlow) based on the previously described cut-off of 5 (2). PFS and OS were estimated using Kaplan-Meier. Cox regression was used for uni- and multivariable analysis.

RESULTS: One hundred and thirty-four patients were included in this analysis. Mean age was 68 (range 40-90). Median follow up was 26 months. Sixty-five samples had both scores low and 30 both high, 25 samples were CTChigh/GWAlow and 14 were CTClow/GWAhigh. PFS and OS were not significantly different between CTChigh versus CTClow groups (p=0.07 and p=0.26, median PFS 15 vs 24 months, median OS 45 vs 49 months). However, PFS and OS were significantly different between GWAhigh versus GWAlow groups (p &lt; 0.01 for both, median PFS 11 vs 24 months, median OS 31 vs 51 months). In multivariate analysis, GWA-score was a significant prognostic factor for both PFS and OS while presence of visceral disease was also prognostic for OS (table 1).

CONCLUSIONS: In patients treated with aromatase inhibitors in the first line of treatment, GWA-score was an independent prognostic marker for a worse PFS and OS and may have added value beyond CTC-count.

1. Verschoor, N., et al., Abstract P2-01-17: Circulating tumor cell count and levels of circulating tumor DNA are complementary prognostic biomarkers in metastatic breast cancer - A pilot study. Cancer Research, 2022. 82(4_Supplement): p. P2-01-17-P2-01-17.

2. Belic, J., et al., mFast-SeqS as a Monitoring and Pre-screening Tool for Tumor-Specific Aneuploidy in Plasma DNA. Adv Exp Med Biol, 2016. 924: p. 147-155.

Table 1: Uni- and multivariate Cox regression for PFS and OS

Citation Format: Noortje Verschoor, Jaco Kraan, Vanja de Weerd, Ngoc M Van, Agnes Jager, Stefan Sleijfer, John WM Martens, Saskia M Wilting. Tumor-agnostic measurement of ctDNA levels in first line ER+, HER2- metastatic breast cancer patients as a marker for survival [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr P2-03-14.

Methylated Cell-Free DNA Sequencing (MeD-seq) of LpnPI Digested Fragments to Identify Early Progression in Metastatic Renal Cell Carcinoma Patients on Watchful Waiting

According to the current guidelines, watchful waiting (WW) is a feasible option for patients with good or intermediate prognosis renal-cell carcinoma (RCC). However, some patients rapidly progress during WW, requiring the initiation of treatment. Here, we explore whether we can identify those patients using circulating cell-free DNA (cfDNA) methylation. We first defined a panel of RCC-specific circulating methylation markers by intersecting differentially methylated regions from a publicly available dataset with known RCC methylation markers from the literature. The resulting RCC-specific methylation marker panel of 22 markers was subsequently evaluated for an association with rapid progression by methylated DNA sequencing (MeD-seq) in serum from 10 HBDs and 34 RCC patients with a good or intermediate prognosis starting WW in the IMPACT-RCC study. Patients with an elevated RCC-specific methylation score compared to HBDs had a shorter progression-free survival (PFS, p = 0.018), but not a shorter WW-time (p = 0.15). Cox proportional hazards regression showed that only the International Metastatic Renal Cell Carcinoma Database Consortium (IMDC) criteria were significantly associated with WW time (HR 2.01, p = 0.01), whereas only our RCC-specific methylation score (HR 4.45, p = 0.02) was significantly associated with PFS. The results of this study suggest that cfDNA methylation is predictive of PFS but not WW.

Circulating tumour DNA as biomarker for rectal cancer: A systematic review and meta-analyses

Background

Circulating tumour DNA (ctDNA) has been established as a promising (prognostic) biomarker with the potential to personalise treatment in cancer patients. The objective of this systematic review is to provide an overview of the current literature and the future perspectives of ctDNA in non-metastatic rectal cancer.

Methods

A comprehensive search for studies published prior to the 4^th of October 2022 was conducted in Embase, Medline, Cochrane, Google scholar, and Web of Science. Only peer-reviewed original articles and ongoing clinical trials investigating the association between ctDNA and oncological outcomes in non-metastatic rectal cancer patients were included. Meta-analyses were performed to pool hazard ratios (HR) for recurrence-free survival (RFS).

Results

A total of 291 unique records were screened, of which 261 were original publications and 30 ongoing trials. Nineteen original publications were reviewed and discussed, of which seven provided sufficient data for meta-analyses on the association between the presence of post-treatment ctDNA and RFS. Results of the meta-analyses demonstrated that ctDNA analysis can be used to stratify patients into very high and low risk groups for recurrence, especially when detected after neoadjuvant treatment (HR for RFS: 9.3 [4.6 - 18.8]) and after surgery (HR for RFS: 15.5 [8.2 - 29.3]). Studies investigated different types of assays and used various techniques for the detection and quantification of ctDNA.

Conclusions

This literature overview and meta-analyses provide evidence for the strong association between ctDNA and recurrent disease. Future research should focus on the feasibility of ctDNA-guided treatment and follow-up strategies in rectal cancer. A blueprint for agreed-upon timing, preprocessing, and assay techniques is needed to empower adaptation of ctDNA into daily practice.

Cost-Effectiveness of Abemaciclib in Early Breast Cancer Patients: One Size Fits All or Tailoring to Patients' Needs?

Purpose

The addition of two years of abemaciclib treatment to standard adjuvant endocrine therapy in all patients with high risk ER+, HER2- early breast cancer (EBC) has been approved by the US Food and Drug Administration (FDA). Pre-selection of patients with an immediate risk of recurrence within the group of clinically high risk patients using detection of minimal residual disease (MRD) using patient-informed circulating tumor DNA assays during follow-up could enhance efficacy. Here, we investigate the cost-effectiveness of the addition of two years abemaciclib in all high risk HR+, HER2- patients and in MRD-guided high risk patients only.

Methods

Two semi-Markov models were developed to evaluate the cost-effectiveness of adding two years of abemaciclib compared to "standard treatment": 1) "abemaciclib all" and 2) "MRD-guided abemaciclib" using MRD-guidance. Data of the MonarchE trial were used to model the invasive disease-free survival (iDFS). Since iDFS and overall survival (OS) data of abemaciclib were currently limited, abemaciclib effects were extrapolated using a favorable, intermediate and unfavorable effect scenario.

Results

The addition of abemaciclib in all high-risk EBC patients prolonged iDFS slightly (0.04 additional quality adjusted life years (QALYs)) and led to higher costs compared to standard ET, leading to a high incremental cost effectiveness ratio (ICER) of €1,551,876/QALY. Neither the favorable effect scenario (additional 1.09 QALYs) was cost-effective (ICER €62,935/QALY), using a willingness-to-pay threshold of €50,000/QALY. The "MRD-guided abemaciclib" strategy resulted in lower costs and an increase in QALYs (1.27) compared to "standard treatment" in the unfavorable effect scenario.

Conclusion

The addition of abemaciclib to adjuvant endocrine therapy in all high-risk ER+, HER2- EBC patients is not cost-effective. However, using MRD detection to justify the addition of abemaciclib treatment dominates standard treatment in this cost-effectiveness analysis. Further evaluation of MRD detection in EBC by means of prospective clinical trials assessing clinical utility is recommended and promising in terms of cost-effectiveness.

Circulating tumour cells are associated with histopathological growth patterns of colorectal cancer liver metastases

Histopathological Growth Patterns (HGPs) have prognostic and predictive value in patients with Colorectal Liver Metastases (CRLM). This study examined whether preoperative measurement of Circulating Tumour Cells (CTCs) is associated with HGP. CTCs were prospectively enumerated in 7.5 ml of blood using the FDA-approved CellSearch system in patients who underwent local treatment of CRLM with curative intent between 2008 and 2021. All CTC samples were collected on the day of local treatment. Patients treated with neoadjuvant chemotherapy for CRLM or with extrahepatic disease at the time of CTC sampling were excluded. HGP was scored retrospectively following the current consensus guidelines. The association between CTCs and HGP was investigated through multivariable logistic regression. Data were available for 177 patients, desmoplastic HGP (dHGP) was observed in 34 patients (19%). There were no statistically significant differences in patient and tumour characteristics between dHGP and non-dHGP at baseline. Patients with dHGP had longer overall - and disease-free survival (logrank p = 0.003 and 0.003, respectively) compared to patients with non-dHGP. CTCs were not detected in 25(74%) of dHGP patients and in 68(48%) of non-dHGP patients (chi-squared p = 0.006). Preoperative absence of CTCs was the only significant predictor for dHGP in multivariable logistic regression (Odds Ratio 2.7, 95%CI 1.1-6.8, p = 0.028), Table 3. Preoperative absence of CTCs is associated with dHGP in chemo naive CRLM patients without extrahepatic disease. Based on our results, CTC count alone is not sufficient to preoperatively identify HGPs, but integration of CTC count in multivariable prediction models may aid the preoperative identification of HGPs of CRLM.

ESR1 Methylation Measured in Cell-Free DNA to Evaluate Endocrine Resistance in Metastatic Breast Cancer Patients

ESR1 methylation was proposed as mechanism for endocrine resistance in metastatic breast cancer patients. To evaluate its potential as a minimally invasive biomarker, we investigated the feasibility of measuring ESR1 methylation in cell-free DNA (cfDNA) and its association with endocrine resistance. First, we provided evidence that demethylation in vitro restores ER expression. Subsequently, we found that ESR1 methylation in cfDNA was not enriched in endocrine-resistant versus endocrine-sensitive patients. Interestingly, we found a correlation between ESR1 methylation and age. Publicly available data confirm an age-related increase in ESR1 methylation in leukocytes, confounding the determination of the ESR1 methylation status of tumors using cfDNA.

Genome-wide aneuploidy detected by mFast-SeqS in circulating cell-free DNA is associated with poor response to pembrolizumab in patients with advanced urothelial cancer

Second-line treatment with immune checkpoint inhibition in patients with metastatic urothelial cancer (mUC) has a low success rate (~ 20%). Circulating tumour-derived DNA (ctDNA) levels may guide patient stratification, provided that an affordable and robust assay is available. Here, we investigate whether the modified fast aneuploidy screening test-sequencing system (mFast-SeqS) may provide such an assay. To this end, mFast-SeqS was performed on cell-free DNA (cfDNA) from 74 patients with mUC prior to treatment with pembrolizumab. Results were associated with corresponding tissue-based profiles, plasma-based variant allele frequencies (VAFs) and clinical response. We found that plasma-derived mFast-SeqS-based aneuploidy scores significantly correlated with those observed in the corresponding tumour tissue as well as with the ctDNA level in the plasma. In multivariate logistic regression analysis, a high aneuploidy score was independently associated with lack of clinical benefit from treatment with pembrolizumab. In conclusion, mFast-SeqS provides a patient-friendly, high-throughput and affordable method to estimate ctDNA level. Following independent validation, this test could be used to stratify mUC patients for response prior to the initiation of treatment with pembrolizumab.

Abstract P2-01-17: Circulating tumor cell count and levels of circulating tumor DNA are complementary prognostic biomarkers in metastatic breast cancer - A pilot study

Background and aim Minimally invasive liquid biopsies, usually referring to the analysis of tumor-derived material such as circulating tumor cells (CTC) and circulating tumor DNA (ctDNA), have already proven their individual prognostic value in metastatic breast cancer. Whereas CTCs can be quantified by enumeration, levels of ctDNA are often expressed as the variant allele frequency of the dominant mutation, determined by rather costly targeted next generation sequencing. The modified fast aneuploidy screening test-sequencing system (mFAST-SeqS) represents a fast and cost-effective assay to estimate the fraction of ctDNA by measuring chromosomal aneuploidy instead of mutations. Next to the low costs (EUR 50, USD 60) additional advantages of the mFAST-SeqS method include the independency of mutational status of the tumor and the comprehensive overview of the tumor genome. Our aim was to investigate the potential concordance and complementarity between CTC-count and the mFAST-SeqS-based genome-wide aneuploidy score (GWA score). Material and methods Consecutive blood samples obtained from ER positive, Her2 negative metastatic breast cancer patients, starting a new line of therapy, (n=23) were enumerated for CTCs using the CellSearch-system (Menarini Silicon Biosystems, Bologna, Italy) and screened for the percentage of ctDNA using mFAST-SeqS. This method provides a GWA-score by amplification of long interspaced nuclear elements (LINE-1 elements) throughout the genome and calculating the overall deviation from the average profile in normal controls. Spearman correlation between CTC-count and GWA-score was calculated. For survival analysis, CTC-count was divided into high (CTChigh) and low (CTClow) using 5 CTCs/7.5 ml blood as the cut-off as before; similarly GWA-score was divided in high (GWAhigh)and low (GWAlow) based on the previously described cut-off of 5. Kaplan-Meier analysis was performed to assess overall survival (OS) in subgroups. Results Mean patient age was 58 (range 42-78). Median follow up was 15 months (range 0-37). Median number of previous treatment lines of patients was 2 (range 0-11). Median CTC-count was 22 (range 0-4438) and median GWA-score was 3.94 (range 0.87-114.31). The Spearman correlation coefficient between CTC count and GWA-score was 0.44 (p=0.02). 3 out of 20 patients were alive at end of follow-up. OS was significantly different in the CTChigh group versus the CTClow group (p=0.01, log-rank test, median OS 31 and 7 months, respectively), as well as in the GWAhigh versus the GWAlow group (p =0.01, log-rank test, median OS 27 and 7 months). There was concordance in 12 out of 23 samples (52.2%). 4 samples had both scores low and 8 both high, 9 samples were CTChigh/GWAlow and 2 were CTClow/GWAhigh. OS was significantly different in these groups (p &lt; 0.01, log-rank test), with a median OS of 6 months in the group with both scores high, a median OS of 18 months for the cases with a discordant score and median OS not reached in the group with both scores low.Conclusions Our pilot study confirmed that both CTC-count and mFAST-SeqS GWA-score are prognostic and indicate that a combination of these measurements may yield complementary prognostic value. It should be noted however that the sample size was limited thus far. Our current preliminary results will be extended to 98 cases, results of which will be presented at the meeting.

Citation Format: Noraly Verschoor, Elisabeth M Jongbloed, Jaco Kraan, Vanja de Weerd, Ngoc M Van, Agnes Jager, Stefan Sleijfer, Saskia M Wilting, John WM Martens. Circulating tumor cell count and levels of circulating tumor DNA are complementary prognostic biomarkers in metastatic breast cancer - A pilot study [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P2-01-17.

Pharmacological CDK4/6 inhibition reveals a p53-dependent senescent state with restricted toxicity

Cellular senescence is a state of stable growth arrest and a desired outcome of tumor suppressive interventions. Treatment with many anti‐cancer drugs can cause premature senescence of non‐malignant cells. These therapy‐induced senescent cells can have pro‐tumorigenic and pro‐disease functions via activation of an inflammatory secretory phenotype (SASP). Inhibitors of cyclin‐dependent kinases 4/6 (CDK4/6i) have recently proven to restrain tumor growth by activating a senescence‐like program in cancer cells. However, the physiological consequence of exposing the whole organism to pharmacological CDK4/6i remains poorly characterized. Here, we show that exposure to CDK4/6i induces non‐malignant cells to enter a premature state of senescence dependent on p53. We observe in mice and breast cancer patients that the CDK4/6i‐induced senescent program activates only a partial SASP enriched in p53 targets but lacking pro‐inflammatory and NF‐κB‐driven components. We find that CDK4/6i‐induced senescent cells do not acquire pro‐tumorigenic and detrimental properties but retain the ability to promote paracrine senescence and undergo clearance. Our results demonstrate that SASP composition is exquisitely stress‐dependent and a predictor for the biological functions of different senescence subsets.

A pipeline for copy number profiling of single circulating tumor cells to assess intra-patient tumor heterogeneity

Intrapatient tumour heterogeneity is likely a major determinant of clinical outcome in cancer patients. To assess heterogeneity in a minimally invasive manner, methods to perform single circulating tumour cell (CTC) genomics at high resolution are necessary. However, due to the rarity of CTCs, development of such methods is challenging. Here, we developed a modular single CTC analysis pipeline to assess intrapatient heterogeneity by copy number (CN) profiling. To optimize this pipeline, spike‐in experiments using MCF‐7 breast cancer cells were performed. The VyCAP puncher system was used to isolate single cells. The quality of whole genome amplification (WGA) products generated by REPLI‐g and Ampli1™ methods, as well as the results from the Illumina Truseq and the Ampli1™ LowPass library preparation techniques, was compared. Moreover, a bioinformatic pipeline was designed to generate CN profiles from single CTCs. The optimal combination of Ampli1™ WGA and Illumina Truseq library preparation was successfully validated on patient‐derived CTCs. In conclusion, we developed a novel modular pipeline to isolate single CTCs and subsequently generate detailed patient‐derived CN profiles that allow assessment of intrapatient heterogeneity in future studies.

High-throughput and affordable genome-wide methylation profiling of circulating cell-free DNA by methylated DNA sequencing (MeD-seq) of LpnPI digested fragments

Background

DNA methylation detection in liquid biopsies provides a highly promising and much needed means for real-time monitoring of disease load in advanced cancer patient care. Compared to the often-used somatic mutations, tissue- and cancer-type specific epigenetic marks affect a larger part of the cancer genome and generally have a high penetrance throughout the tumour. Here, we describe the successful application of the recently described MeD-seq assay for genome-wide DNA methylation profiling on cell-free DNA (cfDNA). The compatibility of the MeD-seq assay with different types of blood collection tubes, cfDNA input amounts, cfDNA isolation methods, and vacuum concentration of samples was evaluated using plasma from both metastatic cancer patients and healthy blood donors (HBDs). To investigate the potential value of cfDNA methylation profiling for tumour load monitoring, we profiled paired samples from 8 patients with resectable colorectal liver metastases (CRLM) before and after surgery.

Results

The MeD-seq assay worked on plasma-derived cfDNA from both EDTA and CellSave blood collection tubes when at least 10 ng of cfDNA was used. From the 3 evaluated cfDNA isolation methods, both the manual QIAamp Circulating Nucleic Acid Kit (Qiagen) and the semi-automated Maxwell® RSC ccfDNA Plasma Kit (Promega) were compatible with MeD-seq analysis, whereas the QiaSymphony DSP Circulating DNA Kit (Qiagen) yielded significantly fewer reads when compared to the QIAamp kit (p < 0.001). Vacuum concentration of samples before MeD-seq analysis was possible with samples in AVE buffer (QIAamp) or water, but yielded inconsistent results for samples in EDTA-containing Maxwell buffer. Principal component analysis showed that pre-surgical samples from CRLM patients were very distinct from HBDs, whereas post-surgical samples were more similar. Several described methylation markers for colorectal cancer monitoring in liquid biopsies showed differential methylation between pre-surgical CRLM samples and HBDs in our data, supporting the validity of our approach. Results for MSC, ITGA4, GRIA4, and EYA4 were validated by quantitative methylation specific PCR.

Conclusions

The MeD-seq assay provides a promising new method for cfDNA methylation profiling. Potential future applications of the assay include marker discovery specifically for liquid biopsy analysis as well as direct use as a disease load monitoring tool in advanced cancer patients.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13148-021-01177-4.

A panel of DNA methylation markers for the classification of consensus molecular subtypes 2 and 3 in patients with colorectal cancer

Consensus molecular subtypes (CMSs) can guide precision treatment of colorectal cancer (CRC). We aim to identify methylation markers to distinguish between CMS2 and CMS3 in patients with CRC, for which an easy test is currently lacking. To this aim, fresh‐frozen tumor tissue of 239 patients with stage I‐III CRC was analyzed. Methylation profiles were obtained using the Infinium HumanMethylation450 BeadChip. We performed adaptive group‐regularized logistic ridge regression with post hoc group‐weighted elastic net marker selection to build prediction models for classification of CMS2 and CMS3. The Cancer Genome Atlas (TCGA) data were used for validation. Group regularization of the probes was done based on their location either relative to a CpG island or relative to a gene present in the CMS classifier, resulting in two different prediction models and subsequently different marker panels. For both panels, even when using only five markers, accuracies were > 90% in our cohort and in the TCGA validation set. Our methylation marker panel accurately distinguishes between CMS2 and CMS3. This enables development of a targeted assay to provide a robust and clinically relevant classification tool for CRC patients.

A panel of DNA methylation markers for the classification of consensus molecular subtypes 2 and 3 in patients with colorectal cancer

3545

Background: Consensus molecular subtypes (CMSs) can guide precision treatment of colorectal cancer (CRC). Currently available assays can identify CMS1 and CMS4 cases well, while a dedicated test to distinguish CMS2 and 3 is lacking. This study aimed to identify a panel of methylation markers to distinguish between CMS2 and 3 in patients with CRC. Methods: Fresh-frozen tumor tissue of 239 patients with stage I-III CRC was included. CMS classification was performed on RNA-seq data using the single-sample-prediction parameter from the “CMSclassifier” package. Methylation profiles were obtained using the Infinium HumanMethylation450 BeadChip. We performed adaptive group-regularised logistic ridge-regression with post-hoc group-weighted elastic net marker selection to build prediction models for classification of CMS2 and CMS3 based on 15, 10 or 5 markers. Data from TCGAwas used for validation. Results: Overall methylation profiles differed between CMS2 and CMS3. Group-regularisation of the probes was done based on their location either relative to a CpG island or relative to a gene present in the CMS classifier resulting in two different prediction models and subsequently different marker panels. For both panels, even when using only 5 markers, sensitivity, specificity, and accuracy were > 90%. Validation showed comparable performances. Conclusions: Our highly sensitive and specific methylation marker panel can be used to distinguish CMS2 and 3. This enables development of a qPCR DNA methylation assay in patients with CRC to provide a specific and non-invasive classification tool.

Circular RNA in Chemonaive Lymph Node Negative Colon Cancer Patients

Circular RNAs (circRNAs) appear important in tumor progression of colon cancer (CC). We identified an extensive catalog of circRNAs in 181 chemonaive stage I/II colon tumors, who underwent curative surgery between 2007 and 2014. We identified circRNAs from RNAseq data, investigated common biology related to circRNA expression, and studied the association between circRNAs and relapse status, tumor stage, consensus molecular subtypes (CMS), tumor localization and microsatellite instability (MSI). We identified 2606 unique circRNAs. 277 circRNAs (derived from 260 genes) were repeatedly occurring in at least 20 patients of which 153 showed a poor or even negative (R < 0.3) correlation with the expression level of their linear gene. The circular junctions for circSATB2, circFGD6, circKMT2C and circPLEKHM3 were validated by Sanger sequencing. Multiple correspondence analysis showed that circRNAs were often co-expressed and that high diversity in circRNAs was associated with favorable disease-free survival (DFS), which was confirmed by Cox regression analysis (Hazard Ratio (HR) 0.60, 95% CI 0.38-0.97, p = 0.036). Considering individual circRNAs, absence of circMGA was significantly associated with relapse, whereas circSATB2, circNAB1, and circCEP192 were associated with both MSI and CMS. This study represents a showcase of the potential clinical utility of circRNAs for prognostic stratification in patients with stage I-II colon cancer and demonstrated that high diversity in circRNAs is associated with favorable DFS.

Detection of Aneuploidy in Cerebrospinal Fluid from Patients with Breast Cancer Can Improve Diagnosis of Leptomeningeal Metastases

PURPOSE

Detection of leptomeningeal metastasis is hampered by limited sensitivities of currently used techniques: MRI and cytology of cerebrospinal fluid (CSF). Detection of cell-free tumor DNA in CSF has been proposed as a tumor-specific candidate to detect leptomeningeal metastasis at an earlier stage. The aim of this study was to investigate mutation and aneuploidy status in CSF-derived cell-free DNA (cfDNA) of patients with breast cancer with a clinical suspicion of leptomeningeal metastasis.

EXPERIMENTAL DESIGN

cfDNA was isolated from stored remnant CSF and analyzed by targeted next-generation sequencing (NGS; n = 30) and the modified fast aneuploidy screening test-sequencing system (mFAST-SeqS; n = 121). The latter method employs selective amplification of long interspaced nuclear elements sequences that are present throughout the genome and allow for fast and cheap detection of aneuploidy. We compared these results with the gold standard to diagnose leptomeningeal metastasis: cytology.

RESULTS

Leptomeningeal metastasis was cytology proven in 13 of 121 patients. Low DNA yields resulted in insufficient molecular coverage of NGS for the majority of samples (success rate, 8/30). The mFAST-SeqS method, successful in 112 of 121 (93%) samples, detected genome-wide aneuploidy in 24 patients. Ten of these patients had cytology-proven leptomeningeal metastasis; 8 additional patients were either concurrently diagnosed with central nervous system metastases by radiological means or developed these soon after the lumbar puncture. The remaining six cases were suspected of leptomeningeal metastasis, but could not be confirmed by cytology or imaging. Aneuploidy was associated with development of leptomeningeal metastasis and significantly worse overall survival.

CONCLUSIONS

Aneuploidy in CSF-derived cfDNA may provide a promising biomarker to improve timely detection of leptomeningeal metastasis.

Inhibiting CDK4/6 in Breast Cancer with Palbociclib, Ribociclib, and Abemaciclib: Similarities and Differences

The cyclin-dependent kinase (CDK) 4/6 inhibitors belong to a new class of drugs that interrupt proliferation of malignant cells by inhibiting progression through the cell cycle. Three such inhibitors, palbociclib, ribociclib, and abemaciclib were recently approved for breast cancer treatment in various settings and combination regimens. On the basis of their impressive efficacy, all three CDK4/6 inhibitors now play an important role in the treatment of patients with HR+, HER2- breast cancer; however, their optimal use still needs to be established. The three drugs have many similarities in both pharmacokinetics and pharmacodynamics. However, there are some differences on the basis of which the choice for a particular CDK4/6 inhibitor for an individual patient can be important. In this article, the clinical pharmacokinetic and pharmacodynamic profiles of the three CDK4/6 inhibitors are reviewed and important future directions of the clinical applicability of CDK4/6 inhibitors will be discussed.

Interconnectivity between molecular subtypes and tumor stage in colorectal cancer

BACKGROUND

There are profound individual differences in clinical outcomes between colorectal cancers (CRCs) presenting with identical stage of disease. Molecular stratification, in conjunction with the traditional TNM staging, is a promising way to predict patient outcomes. We investigated the interconnectivity between tumor stage and tumor biology reflected by the Consensus Molecular Subtypes (CMSs) in CRC, and explored the possible value of these insights in patients with stage II colon cancer.

METHODS

We performed a retrospective analysis using clinical records and gene expression profiling in a meta-cohort of 1040 CRC patients. The interconnectivity of tumor biology and disease stage was assessed by investigating the association between CMSs and TNM classification. In order to validate the clinical applicability of our findings we employed a meta-cohort of 197 stage II colon cancers.

RESULTS

CMS4 was significantly more prevalent in advanced stages of disease (stage I 9.8% versus stage IV 38.5%, p < 0.001). The observed differential gene expression between cancer stages is at least partly explained by the biological differences as reflected by CMS subtypes. Gene signatures for stage III-IV and CMS4 were highly correlated (r = 0.77, p < 0.001). CMS4 cancers showed an increased progression rate to more advanced stages (CMS4 compared to CMS2: 1.25, 95% CI: 1.08-1.46). Patients with a CMS4 cancer had worse survival in the high-risk stage II tumors compared to the total stage II cohort (5-year DFS 41.7% versus 100.0%, p = 0.008).

CONCLUSIONS

Considerable interconnectivity between tumor biology and tumor stage in CRC exists. This implies that the TNM stage, in addition to the stage of progression, might also reflect distinct biological disease entities. These insights can potentially be utilized to optimize identification of high-risk stage II colon cancers.

Identification of Deregulated Pathways, Key Regulators, and Novel miRNA-mRNA Interactions in HPV-Mediated Transformation

Next to a persistent infection with high-risk human papillomavirus (HPV), molecular changes are required for the development of cervical cancer. To identify which molecular alterations drive carcinogenesis, we performed a comprehensive and longitudinal molecular characterization of HPV-transformed keratinocyte cell lines. Comparative genomic hybridization, mRNA, and miRNA expression analysis of four HPV-containing keratinocyte cell lines at eight different time points was performed. Data was analyzed using unsupervised hierarchical clustering, integrated longitudinal expression analysis, and pathway enrichment analysis. Biological relevance of identified key regulatory genes was evaluated in vitro and dual-luciferase assays were used to confirm predicted miRNA-mRNA interactions. We show that the acquisition of anchorage independence of HPV-containing keratinocyte cell lines is particularly associated with copy number alterations. Approximately one third of differentially expressed mRNAs and miRNAs was directly attributable to copy number alterations. Focal adhesion, TGF-beta signaling, and mTOR signaling pathways were enriched among these genes. PITX2 was identified as key regulator of TGF-beta signaling and inhibited cell growth in vitro, most likely by inducing cell cycle arrest and apoptosis. Predicted miRNA-mRNA interactions miR-221-3p_BRWD3, miR-221-3p_FOS, and miR-138-5p_PLXNB2 were confirmed in vitro. Integrated longitudinal analysis of our HPV-induced carcinogenesis model pinpointed relevant interconnected molecular changes and crucial signaling pathways in HPV-mediated transformation.

miR-9-5p Exerts a Dual Role in Cervical Cancer and Targets Transcription Factor TWIST1

Squamous cell carcinoma (SCC) and adenocarcinoma (AC) represent the major cervical cancer histotypes. Both histotypes are caused by infection with high-risk HPV (hrHPV) and are associated with deregulated microRNA expression. Histotype-dependent expression has been observed for miR-9-5p, showing increased expression in SCC and low expression in AC. Here, we studied the regulation and functionality of miR-9-5p in cervical SCCs and ACs using cervical tissue samples and hrHPV-containing cell lines. Expression and methylation analysis of cervical tissues revealed that low levels of miR-9-5p in ACs are linked to methylation of its precursor genes, particularly miR-9-1. Stratification of tissue samples and hrHPV-containing cell lines suggested that miR-9-5p depends on both histotype and hrHPV type, with higher expression in SCCs and HPV16-positive cells. MiR-9-5p promoted cell viability and anchorage independence in cervical cancer cell lines SiHa (SCC, HPV16) and CaSki (metastasized SCC, HPV16), while it played a tumor suppressive role in HeLa (AC, HPV18). TWIST1, a transcription factor involved in epithelial-to-mesenchymal transition (EMT), was established as a novel miR-9-5p target. Our results show that miR-9-5p plays a dual role in cervical cancer in a histotype- and hrHPV type-dependent manner. MiR-9-5p mediated silencing of TWIST1 suggests two distinct mechanisms towards EMT in cervical cancer.

The genomic landscape of metastatic breast cancer highlights changes in mutation and signature frequencies

The whole-genome sequencing of prospectively collected tissue biopsies from 442 patients with metastatic breast cancer reveals that, compared to primary breast cancer, tumor mutational burden doubles, the relative contributions of mutational signatures shift and the mutation frequency of six known driver genes increases in metastatic breast cancer. Significant associations with pretreatment are also observed. The contribution of mutational signature 17 is significantly enriched in patients pretreated with fluorouracil, taxanes, platinum and/or eribulin, whereas the de novo mutational signature I identified in this study is significantly associated with pretreatment containing platinum-based chemotherapy. Clinically relevant subgroups of tumors are identified, exhibiting either homologous recombination deficiency (13%), high tumor mutational burden (11%) or specific alterations (24%) linked to sensitivity to FDA-approved drugs. This study provides insights into the biology of metastatic breast cancer and identifies clinically useful genomic features for the future improvement of patient management.

Complementarity between miRNA expression analysis and DNA methylation analysis in hrHPV-positive cervical scrapes for the detection of cervical disease

Cervical screening by high-risk HPV (hrHPV) testing requires additional risk stratification (triage), as most infections are transient and only a subset of hrHPV-positive women harbours clinically relevant disease. Molecular triage markers such as microRNAs (miRNAs) and DNA methylation markers are particularly promising, as they can be objectively tested directly on hrHPV-positive scrapes and cervicovaginal self-samples. Here, we evaluated the marker potential of 10 candidate miRNAs in 209 hrHPV-positive scrapes of women with underlying precancer (cervical intraepithelial neoplasia, grade 2–3 (CIN2-3)), cancer, or without disease (CIN0/1). A predictive miRNA classifier for CIN3 detection was built using logistic regression, which was compared to and combined with DNA methylation marker FAM19A4. Markers were correlated to histology parameters and hrHPV genotype. A miRNA classifier consisting of miR-149, miR-20a, and miR-93 achieved an area under the curve (AUC) of 0.834 for CIN3 detection, which was not significantly different to that of FAM19A4 methylation (AUC: 0.862, p = 0.591). Combining miRNA and methylation analysis demonstrated complementarity between both marker types (AUC: 0.939). While the miRNA classifier seemed more predictive for CIN2, FAM19A4 methylation was particularly high in HPV16-positive and histologically advanced CIN3, i.e. CIN3 with high lesion volume. The miRNA classifier, FAM19A4 methylation, and the miRNA/methylation combination were highest in cancer-associated scrapes. In conclusion, a panel of three miRNAs is discriminatory for CIN3 in hrHPV-positive scrapes and can complement DNA methylation analysis for the efficient detection of cervical disease. Combined analysis of the two marker types warrants further evaluation as triage strategy in hrHPV-based screening.

The circular RNome of primary breast cancer

Circular RNAs (circRNAs) are a class of RNAs that is under increasing scrutiny, although their functional roles are debated. We analyzed RNA-seq data of 348 primary breast cancers and developed a method to identify circRNAs that does not rely on unmapped reads or known splice junctions. We identified 95,843 circRNAs, of which 20,441 were found recurrently. Of the circRNAs that match exon boundaries of the same gene, 668 showed a poor or even negative (R < 0.2) correlation with the expression level of the linear gene. In silico analysis showed only a minority (8.5%) of circRNAs could be explained by known splicing events. Both these observations suggest that specific regulatory processes for circRNAs exist. We confirmed the presence of circRNAs of CNOT2, CREBBP, and RERE in an independent pool of primary breast cancers. We identified circRNA profiles associated with subgroups of breast cancers and with biological and clinical features, such as amount of tumor lymphocytic infiltrate and proliferation index. siRNA-mediated knockdown of circCNOT2 was shown to significantly reduce viability of the breast cancer cell lines MCF-7 and BT-474, further underlining the biological relevance of circRNAs. Furthermore, we found that circular, and not linear, CNOT2 levels are predictive for progression-free survival time to aromatase inhibitor (AI) therapy in advanced breast cancer patients, and found that circCNOT2 is detectable in cell-free RNA from plasma. We showed that circRNAs are abundantly present, show characteristics of being specifically regulated, are associated with clinical and biological properties, and thus are relevant in breast cancer.

A two-gene methylation signature for the diagnosis of bladder cancer in urine

AIM

To analyze the potential of 14 cancer-associated genes, including six miRNAs, for bladder cancer (BC) diagnosis in urine.

PATIENTS & METHODS

DNA methylation levels of 14 genes were analyzed in urine of 72 BC patients and 75 healthy controls using quantitative methylation-specific PCR. Multivariate logistic regression analysis was used to determine an optimal marker panel.

RESULTS

Ten genes were significantly hypermethylated in BC patients. The GHSR/MAL combination showed the best diagnostic performance, reaching a sensitivity of 92% (95% CI: 86-99) and a specificity of 85% (95% CI: 76-94).

CONCLUSION

We identified a novel two-gene panel with a high diagnostic accuracy for BC that can be applied in a noninvasive, urine-based test.

High-throughput isolation of circulating tumor DNA: a comparison of automated platforms

The emerging interest in circulating tumor DNA (ctDNA) analyses for clinical trials has necessitated the development of a high-throughput method for fast, reproducible, and efficient isolation of ctDNA. Currently, the majority of ctDNA studies use the manual QIAamp (QA) platform to isolate DNA from blood. The purpose of this study was to compare two competing automated DNA isolation platforms [Maxwell (MX) and QIAsymphony (QS)] to the current 'gold standard' QA to facilitate high-throughput processing of samples in prospective trials. We obtained blood samples from healthy blood donors and metastatic cancer patients for plasma isolation. Total cell-free DNA (cfDNA) quantity was assessed by TERT quantitative PCR. Recovery efficiency was investigated by quantitative PCR analysis of spiked-in synthetic plant DNA. In addition, a β-actin fragmentation assay was performed to determine the amount of contamination by genomic DNA from lysed leukocytes. ctDNA quality was assessed by digital PCR for somatic variant detection. cfDNA quantity and recovery efficiency were lowest using the MX platform, whereas QA and QS showed a comparable performance. All platforms preferentially isolated small (136 bp) DNA fragments over large (420 and 2000 bp) DNA fragments. Detection of the number variant and wild-type molecules was most comparable between QA and QS. However, there was no significant difference in variant allele frequency comparing QS and MX to QA. In summary, we show that the QS platform has comparable performance to QA, the 'gold standard', and outperformed the MX platform depending on the readout used. We conclude that the QS can replace the more laborious QA platform, especially when high-throughput cfDNA isolation is needed.

Genome-wide microRNA analysis of HPV-positive self-samples yields novel triage markers for early detection of cervical cancer

Offering self‐sampling for HPV testing improves the effectiveness of current cervical screening programs by increasing population coverage. Molecular markers directly applicable on self‐samples are needed to stratify HPV‐positive women at risk of cervical cancer (so‐called triage) and to avoid over‐referral and overtreatment. Deregulated microRNAs (miRNAs) have been implicated in the development of cervical cancer, and represent potential triage markers. However, it is unknown whether deregulated miRNA expression is reflected in self‐samples. Our study is the first to establish genome‐wide miRNA profiles in HPV‐positive self‐samples to identify miRNAs that can predict the presence of CIN3 and cervical cancer in self‐samples. Small RNA sequencing (sRNA‐Seq) was conducted to determine genome‐wide miRNA expression profiles in 74 HPV‐positive self‐samples of women with and without cervical precancer (CIN3). The optimal miRNA marker panel for CIN3 detection was determined by GRridge, a penalized method on logistic regression. Six miRNAs were validated by qPCR in 191 independent HPV‐positive self‐samples. Classification of sRNA‐Seq data yielded a 9‐miRNA marker panel with a combined area under the curve (AUC) of 0.89 for CIN3 detection. Validation by qPCR resulted in a combined AUC of 0.78 for CIN3+ detection. Our study shows that deregulated miRNA expression associated with CIN3 and cervical cancer development can be detected by sRNA‐Seq in HPV‐positive self‐samples. Validation by qPCR indicates that miRNA expression analysis offers a promising novel molecular triage strategy for CIN3 and cervical cancer detection applicable to self‐samples.

What's new?

MicroRNAs (miRNAs) are suspected of playing a role in cervical cancer development. They are also potential markers for the identification of human papillomavirus (HPV)‐infected women who are at risk of cervical cancer. Here, using small RNA sequencing of HPV‐positive self‐samples from women with and without cervical precancer (CIN3), the authors identify a miRNA signature consisting of multiple miRNAs that is strongly predictive of CIN3. Validation of this signature by qPCR revealed a good clinical performance for CIN3+ detection. The findings suggest that miRNA analysis is an effective means of CIN3+ prediction in HPV‐positive self‐samples obtained for cervical cancer screening.

Molecular heterogeneity in human papillomavirus-dependent and -independent vulvar carcinogenesis

Vulvar squamous cell carcinoma (VSCC) and precancerous vulvar intraepithelial neoplasia (VIN) can develop through human papillomavirus (HPV)-dependent and -independent pathways, indicating a heterogeneous disease. Only a minority of VIN progress, but current clinicopathological classifications are insufficient to predict the cancer risk. Here we analyzed copy number alterations (CNA) to assess the molecular heterogeneity of vulvar lesions in relation to HPV and cancer risk. HPV-status and CNA by means of whole-genome next-generation shallow-sequencing were assessed in VSCC and VIN. The latter included VIN of women with associated VSCC (VINVSCC ) and women who did not develop VSCC during follow-up (VINnoVSCC ). HPV-testing resulted in 41 HPV-positive (16 VINVSCC , 14 VINnoVSCC , and 11 VSCC) and 24 HPV-negative (11 VINVSCC and 13 VSCC) lesions. HPV-positive and -negative VSCC showed a partially overlapping pattern of recurrent CNA, including frequent gains of 3q and 8q. In contrast, amplification of 11q13/cyclinD1 was exclusively found in HPV-negative lesions. HPV-negative VINVSCC had less CNA than HPV-negative VSCC (P = .009), but shared chromosome 8 alterations. HPV-positive VINnoVSCC had less CNA than VINVSCC (P = .022). Interestingly, 1pq gain was detected in 81% of HPV-positive VINVSCC and only in 21% of VINnoVSCC (P = .001). In conclusion, HPV-dependent and -independent vulvar carcinogenesis is characterized by distinct CNA patterns at the VIN stage, while more comparable patterns are present at the cancer stage. Cancer risk in VIN seems to be reflected by the extent of CNA, in particular chromosome 1 gain in HPV-positive cases.

Ridge estimation of network models from time-course omics data

Time-course omics experiments enable the reconstruction of the dynamics of the cellular regulatory network. Here, we describe the means for this reconstruction and the downstream exploitation of the inferred network. It is assumed that one of the various vector-autoregressive models (VAR) models presented here serves as a reasonably accurate description of the time-course omics data. The models are estimated through ridge penalized likelihood maximization, accompanied by functionality for the determination of optimal penalty paramaters. Prior knowledge on the network topology is accommodated by the estimation procedures. Various routes that translate the fitted models into more tangible implications for the medical researcher are described. The network is inferred from the-nonsparse-ridge estimates through empirical Bayes probabilistic thresholding. The influence of a (trait of a) molecular entity at the current time on those at future time points is assessed by mutual information, impulse response analysis, and path decomposition of the covariance. The presented methodology is applied to the omics data from the p53 signaling pathway during HPV-induced cellular transformation. All methodology is implemented in the ragt2ridges package, freely available from the Comprehensive R Archive Network.

Abstract 5050: Molecular heterogeneity in HPV-dependent and -independent vulvar carcinogenesis: Chromosome 1 gain reflects the cancer risk in vulvar intraepithelial neoplasia

Vulvar squamous cell carcinoma (VSCC) and precancerous high-grade vulvar intraepithelial neoplasia (VIN) can develop through human papillomavirus (HPV)-dependent and -independent pathways, indicating a heterogeneous disease. Only a minority of VIN progress, but current clinical and histological classifications are insufficient to predict the cancer risk. Here we analyzed copy number alterations (CNA) to assess the molecular heterogeneity of vulvar lesions in relation to HPV and cancer risk. HPV-status and CNA by means of whole-genome next-generation shallow-sequencing were assessed in 24 VSCC and 41 VIN lesions. The latter included VIN of women with associated VSCC (VINVSCC) and women who did not develop VSCC during &gt;10 year follow-up (VINnoVSCC). HPV-testing resulted in 41 HPV-positive lesions (16 VINVSCC, 14 VINnoVSCC and 11 VSCC) and 24 HPV-negative lesions (11 VINVSCC and 13 VSCC). HPV-positive and -negative VSCC showed a partially overlapping pattern of recurrent CNA, including frequent gains of 3q and 8q. In contrast, amplification of 11q13/cyclinD1 was exclusively found in HPV-negative lesions. HPV-negative VINVSCC had less CNA than VSCC (P = 0.009), mainly 8q gains and 8p losses. In HPV-positive lesions, CNA frequencies were lower in VINnoVSCC than in VINVSCC (P = 0.022) and VSCC. Interestingly, a 1pq gain was detected in 81% of VINVSCC and only 29% of VINnoVSCC. In conclusion, HPV-dependent and -independent vulvar carcinogenesis is characterized by distinct CNA patterns at the VIN stage, while more comparable patterns are present at the cancer stage. Progression risk in VIN is reflected by the extent of CNA, in particularly chromosome 1 gain.

Citation Format: Maaike CG Bleeker, Dorian RA Swarts, Quirinus JM Voorham, Saskia M. Wilting, Marc van Beurden, Renske DM Steenbergen. Molecular heterogeneity in HPV-dependent and -independent vulvar carcinogenesis: Chromosome 1 gain reflects the cancer risk in vulvar intraepithelial neoplasia [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 5050.

Abstract 5059: Comprehensive molecular profiling of HPV-induced transformation over time

Although persistent infection with high-risk human papillomavirus (HPV) is generally acknowledged as necessary cause for cervical cancer, additional molecular changes are required for the development of cancer. Those changes include chromosomal aberrations that result in deregulated expression of coding and non-coding RNAs. In this study we performed a comprehensive and longitudinal molecular characterization of HPV-transformed keratinocyte cell lines, to identify the sequential order and relevance of these molecular alterations for HPV-induced transformation. Genome-wide chromosomal, mRNA and miRNA expression profiles were generated from 4 HPV-transformed keratinocyte cell lines at 8 different passages representing progressive stages of transformation. Unsupervised hierarchical clustering of DNA copy number, mRNA and miRNA profiles revealed that in all 4 cell lines most pronounced changes were associated with the acquisition of anchorage independence, a hallmark of transformation. Approximately one third of differentially expressed mRNAs and miRNAs was directly attributable to DNA copy number alterations. Focal adhesion, TGF-β signalling and mTOR signalling pathways were enriched among these genes. In addition, longitudinal analysis identified more than 600 potential miRNA-mRNA interactions that were negatively correlated during passaging. Four interactions were also predicted by 3 independent prediction algorithms, and have in part already been functionally validated, thereby demonstrating the validity of our approach. In conclusion, integrated longitudinal analysis of our HPV-induced transformation model enabled us to pinpoint relevant interconnected molecular changes and affected signalling pathways. Increased understanding of the interplay between different molecular alterations and affected pathways will be of importance for the clinical management of patients diagnosed with HPV-induced (pre)cancerous lesions.

Citation Format: Iris Babion, Viktorian Miok, Annelieke Jaspers, Wessel N. van Wieringen, Renske D. Steenbergen, Saskia M. Wilting. Comprehensive molecular profiling of HPV-induced transformation over time [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 5059.

Gene length corrected trimmed mean of M-values (GeTMM) processing of RNA-seq data performs similarly in intersample analyses while improving intrasample comparisons

BACKGROUND

Current normalization methods for RNA-sequencing data allow either for intersample comparison to identify differentially expressed (DE) genes or for intrasample comparison for the discovery and validation of gene signatures. Most studies on optimization of normalization methods typically use simulated data to validate methodologies. We describe a new method, GeTMM, which allows for both inter- and intrasample analyses with the same normalized data set. We used actual (i.e. not simulated) RNA-seq data from 263 colon cancers (no biological replicates) and used the same read count data to compare GeTMM with the most commonly used normalization methods (i.e. TMM (used by edgeR), RLE (used by DESeq2) and TPM) with respect to distributions, effect of RNA quality, subtype-classification, recurrence score, recall of DE genes and correlation to RT-qPCR data.

RESULTS

We observed a clear benefit for GeTMM and TPM with regard to intrasample comparison while GeTMM performed similar to TMM and RLE normalized data in intersample comparisons. Regarding DE genes, recall was found comparable among the normalization methods, while GeTMM showed the lowest number of false-positive DE genes. Remarkably, we observed limited detrimental effects in samples with low RNA quality.

CONCLUSIONS

We show that GeTMM outperforms established methods with regard to intrasample comparison while performing equivalent with regard to intersample normalization using the same normalized data. These combined properties enhance the general usefulness of RNA-seq but also the comparability to the many array-based gene expression data in the public domain.

Triage of high-risk HPV-positive women in population-based screening by miRNA expression analysis in cervical scrapes; a feasibility study

Background

Primary testing for high-risk HPV (hrHPV) is increasingly implemented in cervical cancer screening programs. Many hrHPV-positive women, however, harbor clinically irrelevant infections, demanding additional disease markers to prevent over-referral and over-treatment. Most promising biomarkers reflect molecular events relevant to the disease process that can be measured objectively in small amounts of clinical material, such as miRNAs. We previously identified eight miRNAs with altered expression in cervical precancer and cancer due to either methylation-mediated silencing or chromosomal alterations. In this study, we evaluated the clinical value of these eight miRNAs on cervical scrapes to triage hrHPV-positive women in cervical screening.

Results

Expression levels of the eight candidate miRNAs in cervical tissue samples (n = 58) and hrHPV-positive cervical scrapes from a screening population (n = 187) and cancer patients (n = 38) were verified by quantitative RT-PCR. In tissue samples, all miRNAs were significantly differentially expressed (p < 0.05) between normal, high-grade precancerous lesions (CIN3), and/or cancer. Expression patterns detected in cervical tissue samples were reflected in cervical scrapes, with five miRNAs showing significantly differential expression between controls and women with CIN3 and cancer. Using logistic regression analysis, a miRNA classifier was built for optimal detection of CIN3 in hrHPV-positive cervical scrapes from the screening population and its performance was evaluated using leave-one-out cross-validation. This miRNA classifier consisted of miR-15b-5p and miR-375 and detected a major subset of CIN3 as well as all carcinomas at a specificity of 70%. The CIN3 detection rate was further improved by combining the two miRNAs with HPV16/18 genotyping. Interestingly, both miRNAs affected the viability of cervical cancer cells in vitro.

Conclusions

This study shows that miRNA expression analysis in cervical scrapes is feasible and enables the early detection of cervical cancer, thus underlining the potential of miRNA expression analysis for triage of hrHPV-positive women in cervical cancer screening.

Electronic supplementary material

The online version of this article (10.1186/s13148-018-0509-9) contains supplementary material, which is available to authorized users.

A Strategy to Find Suitable Reference Genes for miRNA Quantitative PCR Analysis and Its Application to Cervical Specimens

miRNAs represent an emerging class of promising biomarkers for cancer diagnostics. To perform reliable miRNA expression analysis using quantitative PCR, adequate data normalization is essential to remove nonbiological, technical variations. Ideal reference genes should be biologically stable and reduce technical variability of miRNA expression analysis. Herein is a new strategy for the identification and evaluation of reference genes that can be applied for miRNA-based diagnostic tests without entailing excessive additional experiments. We analyzed the expression of 11 carefully selected candidate reference genes in different types of cervical specimens [ie, tissues, scrapes, and self-collected cervicovaginal specimens (self-samples)]. To identify the biologically most stable reference genes, three commonly used algorithms (GeNorm, NormFinder, and BestKeeper) were combined. Signal-to-noise ratios and P values between control and disease groups were calculated to validate the reduction in technical variability on expression analysis of two marker miRNAs. miR-423 was identified as a suitable reference gene for all sample types, to be used in combination with RNU24 in cervical tissues, RNU43 in scrapes, and miR-30b in self-samples. These findings demonstrate that the choice of reference genes may differ between different types of specimens, even when originating from the same anatomical source. More important, it is shown that adequate normalization increases the signal-to-noise ratio, which is not observed when normalizing to commonly used reference genes.

Identification and Validation of a 3-Gene Methylation Classifier for HPV-Based Cervical Screening on Self-Samples

Purpose: Offering self-sampling of cervico-vaginal material for high-risk human papillomavirus (hrHPV) testing is an effective method to increase the coverage in cervical screening programs. Molecular triage directly on hrHPV-positive self-samples for colposcopy referral opens the way to full molecular cervical screening. Here, we set out to identify a DNA methylation classifier for detection of cervical precancer (CIN3) and cancer, applicable to lavage and brush self-samples.Experimental Design: We determined genome-wide DNA methylation profiles of 72 hrHPV-positive self-samples, using the Infinium Methylation 450K Array. The selected DNA methylation markers were evaluated by multiplex quantitative methylation-specific PCR (qMSP) in both hrHPV-positive lavage (n = 245) and brush (n = 246) self-samples from screening cohorts. Subsequently, logistic regression analysis was performed to build a DNA methylation classifier for CIN3 detection applicable to self-samples of both devices. For validation, an independent set of hrHPV-positive lavage (n = 199) and brush (n = 287) self-samples was analyzed.Results: Genome-wide DNA methylation profiling revealed 12 DNA methylation markers for CIN3 detection. Multiplex qMSP analysis of these markers in large series of lavage and brush self-samples yielded a 3-gene methylation classifier (ASCL1, LHX8, and ST6GALNAC5). This classifier showed a very good clinical performance for CIN3 detection in both lavage (AUC = 0.88; sensitivity = 74%; specificity = 79%) and brush (AUC = 0.90; sensitivity = 88%; specificity = 81%) self-samples in the validation set. Importantly, all self-samples from women with cervical cancer scored DNA methylation-positive.Conclusions: By genome-wide DNA methylation profiling on self-samples, we identified a highly effective 3-gene methylation classifier for direct triage on hrHPV-positive self-samples, which is superior to currently available methods. Clin Cancer Res; 24(14); 3456-64. ©2018 AACR.

Confirmation of a metastasis-specific microRNA signature in primary colon cancer

The identification of patients with high-risk stage II colon cancer who may benefit from adjuvant therapy may allow the clinical approach to be tailored for these patients based on an understanding of tumour biology. MicroRNAs have been proposed as markers of the prognosis or treatment response in colorectal cancer. Recently, a 2-microRNA signature (let-7i and miR-10b) was proposed to identify colorectal cancer patients at risk of developing distant metastasis. We assessed the prognostic value of this signature and additional candidate microRNAs in an independent, clinically well-defined, prospectively collected cohort of primary colon cancer patients including stage I-II colon cancer without and stage III colon cancer with adjuvant treatment. The 2-microRNA signature specifically predicted hepatic recurrence in the stage I-II group, but not the overall ability to develop distant metastasis. The addition of miR-30b to the 2-microRNA signature allowed the prediction of both distant metastasis and hepatic recurrence in patients with stage I-II colon cancer who did not receive adjuvant chemotherapy. Available gene expression data allowed us to associate miR-30b expression with axon guidance and let-7i expression with cell adhesion, migration, and motility.

Aberrant methylation-mediated silencing of microRNAs contributes to HPV-induced anchorage independence

Cervical cancer and a subset of anogenital and head-and-neck carcinomas are caused by high-risk types of the human papillomavirus (hrHPV). During hrHPV-induced malignant transformation keratinocytes become able to grow anchorage independently, a tumorigenic trait at least partly associated with inactivation of tumor suppressor genes. We used hrHPV-containing keratinocytes to investigate the role of DNA methylation-mediated silencing of microRNAs (miRNAs) in the acquisition of anchorage independence.

Anchorage dependent (n=11) and independent passages (n=19) of 4 hrHPV-immortalized keratinocyte cell lines were treated with 2′-deoxy-5-azacytidine (DAC). Genome-wide miRNA expression profiles before and after treatment were compared to identify miRNAs silenced by methylation. Bisulfite sequencing and methylation-specific PCR showed increased methylation of hsa-mir-129-2/-137/-935/-3663/-3665 and -4281 in anchorage independent HPV-transformed keratinocytes and cervical cancer cell lines. Mature miRNAs derived from hsa-mir-129-2/-137/-3663 and -3665 showed functional relevance as they decreased anchorage independence in cervical cancer cell lines. Cervical (pre)cancerous lesions demonstrated increased methylation of hsa-mir-129-2/-935/-3663/-3665 and -4281, underlining the clinical relevance of our findings.

In conclusion, methylation-mediated silencing of tumor suppressive miRNAs contributes to acquisition of an anchorage independent phenotype. This study further substantiates the importance of miRNAs during early stages of carcinogenesis and underlines their potential as both disease markers and therapeutic targets.

Improved high-dimensional prediction with Random Forests by the use of co-data

Background

Prediction in high dimensional settings is difficult due to the large number of variables relative to the sample size. We demonstrate how auxiliary ‘co-data’ can be used to improve the performance of a Random Forest in such a setting.

Results

Co-data are incorporated in the Random Forest by replacing the uniform sampling probabilities that are used to draw candidate variables by co-data moderated sampling probabilities. Co-data here are defined as any type information that is available on the variables of the primary data, but does not use its response labels. These moderated sampling probabilities are, inspired by empirical Bayes, learned from the data at hand. We demonstrate the co-data moderated Random Forest (CoRF) with two examples. In the first example we aim to predict the presence of a lymph node metastasis with gene expression data. We demonstrate how a set of external p-values, a gene signature, and the correlation between gene expression and DNA copy number can improve the predictive performance. In the second example we demonstrate how the prediction of cervical (pre-)cancer with methylation data can be improved by including the location of the probe relative to the known CpG islands, the number of CpG sites targeted by a probe, and a set of p-values from a related study.

Conclusion

The proposed method is able to utilize auxiliary co-data to improve the performance of a Random Forest.

Electronic supplementary material

The online version of this article (doi:10.1186/s12859-017-1993-1) contains supplementary material, which is available to authorized users.