Background
Despite the increasing armamentarium of treatment options for metastatic clear cell renal cell carcinoma (mccRCC), its management still represents a medical challenge owing to the lack of predictive biomarkers for selection of treatments [
1,
2]. Compelling evidence demonstrated that mccRCC is an angiogenesis-dependent disease where overproduction of VEGF is the results of molecular abnormalities, including mutations of the von Hippel Lindau (VHL) gene [
3]. VHL loss-of-function is the most common genetic finding and is associated with the accumulation of hypoxia-inducible factor (HIF), which results in the excessive production of proangiogenic factors such as vascular endothelial (VEGF), fibroblast (FGF), and platelet-derived (PDGF) growth factors. The availability of VEGFR-tyrosine kinase inhibitors (TKI) alongside with mTOR inhibitors (mTORI) has increased the medical options for the management of mccRCC patients. On the other side, the introduction of anti-PD-(L)1 and anti-CTLA-4 antibodies, and the combination of immunotherapy and VEGFR-TKI positively revolutionized the clinical scenario, posing a new challenge for biomarker discovery. At the moment, apart from guidelines established on the basis of clinical trials, there is a substantial lack of predictive/prognostic biomarkers to support the clinician in her/his therapeutic decision [
4]. Candidate genes for a pharmacogenomic stratification of mccRCC include the oncosuppressors TSC1/2 and PTEN, the tumor-drivers PI3K, AKT and mTOR as well as the growth factors released by tumor cells and acting on both the stroma and cancer cells in a paracrine/autocrine fashion, including FGF, PDGF and VEGF. mTOR signaling pathway plays a crucial role in cell growth, survival, proliferation and angiogenesis. PI3K is activated by receptor tyrosine kinases such as VEGFR and insulin-like growth factor receptor (IGFR); the activation of them results in a kinase cascade through AKT and mTOR. This pathway is negatively regulated by the tumor suppressor gene PTEN through the dephosphorylation of phosphatidylinositol (3, 4, 5) trisphosphate (PIP3). Genetic alterations of mTOR pathway-related genes, including PI3K, AKT and PTEN, facilitate tumorigenesis and are common in human cancers. However, recent published results failed to identify and validate predictive biomarkers of response in mccRCC patients [
5]. The identification of biomarkers is often based on data obtained from tissue samples including biopsies, an invasive approach compared to liquid biopsy. The analysis of circulating free DNA (cfDNA) offers an alternative, non-invasive tool to interrogate the genetic profile of tumors [
6]. Previous studies already suggested cfDNA as an accurate prognostic and predictive biomarker for metastatic cancers [
6]; however it has not been validated in mccRCC [
7]. In the present study, we aimed to enhance our understanding of the genomic context of mccRCC and the predictive/prognostic value of cfDNA in this disease.
Blood sampling and cfDNA isolation
Blood specimens of 12 ml were collected at treatment baseline in EDTA tubes and centrifuged for 10 min at 1900 g within 2 h from sampling. cfDNA isolation was performed on 4 ml plasma from each patient using the MagMAX™ Cell-free Total Nucleic Acid Isolation Kit (ThermoFisher Scientific, Carlsbad, CA) according to the manufacturer’s instructions. Yields of isolated cfDNA were assessed using a Qubit fluorometer and Qubit dsDNA High Sensitivity Assay Kit (Invitrogen, Carlsbad, CA, USA).
Next-generation sequencing
The Oncomine Pan-Cancer Cell-Free Assay (Thermo Fisher Scientific, Carlsbad, CA) was used to generate libraries from the cfDNA, following the manufacturer’s instructions. The panel covered 52 genes, including hot spots (single nucleotide variants [SNVs] and short indels), copy number variations (CNVs) and gene fusions. Unique index tags were added to each fragment during library preparation. The concentration of the libraries was assessed by quantitative PCR (qPCR) with the Ion Library TaqMan Quantification Kit (Thermo Fisher Scientific, Carlsbad, CA). The quantified stock libraries were then diluted to 100 pM for downstream template preparation. The Ion Chef System and the Ion 540 Kit-Chef were used for template preparation, followed by sequencing on the Ion S5 system using Ion 540 chips. A four-plex library pool was applied to the Ion 540 chip. Raw data were processed automatically on the Torrent Server™ and aligned to the reference hg19 genome. The sequencing data were then analysed by the Ion Reporter™ Analysis Server v. 5.16.0.3 (Thermo Fisher Scientific, Carlsbad, CA) for variant calling and annotation using the workflow Oncomine TagSeq Pan-Cancer Liquid Biopsy w2.4 (Thermo Fisher Scientific, Carlsbad, CA).
Statistical analysis
Categorical variables were described by absolute and relative frequencies while quantitative factors by median and range. To compare quantitative with categorical variables, the Mann–Whitney test-two tailed was performed. Patients were grouped in two categories based on their best clinical response: patients who experienced CR/PR/SD were defined as “responders”, while patients with PD were defined as “non-responders”. The median cut-off value for analysis of cfDNA was calculated by the Receiver Operating Characteristic (ROC) curve and by the Youden’s Index analysis. PFS and OS were defined as the time from treatment start to progression disease or death, respectively. The Kaplan–Meier method was used to create survival curves and log-rank test was used to evaluate the differences between curves. The Cox hazard regression method was used to identify independent risk factors for OS and PFS. Spearman correlation coefficient was used to assess the correlation between ctDNA amount and the number of metastatic sites or number of mutations detected. Logistic regression model test (Cox-Snell’s R
2) and ROC curve analysis was computed to estimate diagnostic performance of such signatures, alone and combined with the other biomarkers analysed. χ
2-test was used to determine the association between cfDNA and PFS and differences were considered significant at p < 0.05. All statistical analyses were performed with MedCalc Statistical Software version 14.8.1 (MedCalc Software, Ostend, Belgium) and the free and open statistical software program JAMOVI
® (Version 1.1.9; downloaded from
https://www.jamovi.org).
Discussion
In this hypothesis-generating exploratory analysis, cfDNA amount and TP53 status were found to be associated with first line treatment outcome in mccRCC. Although the available therapeutic options of mccRCC are expanding [
8], a personalized treatment is still limited due to a lack of validated predictive/prognostic biomarkers. Recent evidence demonstrated the feasibility of cfDNA analysis and its high concordance with tissue genomic analysis in mccRCC patients [
9]. cfDNA assessment has numerous advantages over tissue biopsies, including minimal invasiveness, the ability to capture intratumoral heterogeneity, and the feasibility of repeated assessments of genomic profile to track tumor dynamics [
10‐
12]. Published studies already demonstrated the role of cfDNA to perform initial diagnosis or detecting disease recurrence [
13,
14]. Moreover, early variations in cfDNA levels are known to be associated with response to treatment [
15,
16]. The present study was aimed at evaluating the association of cfDNA and genomic alterations with response in 48 mRCC patients undergoing treatment with VEGFR-TKIs or immunotherapy. The study involved 12 patients treated with nivolumab plus ipilimumab, and 36 patients treated with TKIs, including sunitinib, pazopanib and cabozantinib. The PFS univariate models showed no significant differences concerning treatment types while ctDNA amount was significantly associated with PFS, regardless of the therapeutic management. This result is consistent with the one from Yamoto et al. [
16], which found that cfDNA levels were significantly associated with PFS and cancer-specific survival, suggesting its potential role as a biomarker in ccRCC patients with or without metastases. Moreover, other published data suggested an association between cfDNA detection and high radiographic tumor burden [
17]. A correlation was also found between the number of metastatic sites and cfDNA levels, thus suggesting a potential role for cfDNA in monitoring response to treatment with detectable changes occurring before radiographic evidence of disease progression [
17].
Importantly, the predictive value of cfDNA as an independent predictive biomarker was maintained at the multivariate analysis. Cut-off points enabled us to identify three groups of patients at high, intermediate or low risk to progress to treatment, potentially allowing clinicians to avoid expensive treatments in patients who will not respond to therapy. The clinical utility of cfDNA for the management of patients affected by different tumor types and treated with immunotherapy, targeted therapies or chemotherapy has been demonstrated in several studies [
18], including melanoma [
19], colorectal cancer [
20], lung [
21] and breast cancer [
22]. However, despite increasing evidence in favor of the predictive and prognostic role of cfDNA, there is still a lack of consensus for its assessment in clinical practice. Seventeen genes were found mutated alone or concurrently in the present population, including TP53, mTOR, PI3K, BRAF, EGFR, RET, GNAS, SF3B1, SMO, KRAS, PDGFRA, FGFR3, FBXW7, ALK, CTNNB1, PTEN, and APC. Being a pan-cancer assay, VHL somatic mutations were not included in the NGS panel used in this study. However, despite its clear role in ccRCC, VHL mutations are not considered a biomarker, neither tested during the diagnostic work-up [
23,
24]. Unfortunately, no sensitizing mutations predictive of response to targeted treatment (i.e. BRAF p.V600E; MET amplification) were detected in this study. TP53 was the most common mutant gene in the overall population, and was significantly associated with PFS, alone or in combination with cfDNA. It is worth to note that ccRCC harboring TP53 alterations showed deregulation of cell cycle, FAS/pentose phosphate pathway and stromal gene expression [
25]. In addition to this, it was reported that TP53 downregulates HIF-1α in RCC [
26], and tumors with loss-of-function mutations of TP53 show enhanced vascular growth. Therefore, the findings of the present study demonstrating an association of TP53 mutations with shorter PFS are justified by the relevant biologic role of TP53 loss-of-function on cell cycle and angiogenesis, including immunosuppression induced by overproduction of VEGF [
27].
The present study adds new evidence in favor of the potential usefulness of cfDNA in monitoring mccRCC response to treatment. It has been demonstrated that cfDNA and tissue-based genomic profiling are complementary to identify actionable alterations in mRCC [
28]. Furthermore, cfDNA can detect changes in genomic profiles over time and may potentially help in guiding treatment. Analysis of large cohorts of patients affectd by ccRCC demonstrated significant changes in cfDNA during the clinical course of the disease, which may have therapeutic implications [
29]. Since collection of cfDNA is feasible in patients with metastatic tumors, this test can be easily translated also into the clinical practice of mccRCC, as it has been already done in different tumors, such as lung cancer [
30].
The concordance of cfDNA with radiological imaging is also already known in different tumor types, including lung, breast, and bladder cancer [
31,
32]. In particular, it is known that cfDNA variations are able to anticipate tumor radiological and clinical progression by weeks. However, how to practically translate this information in clinical practice awaits prospective validation. Since new drug combinations may improve survival in patients with mccRCC, cfDNA levels may integrate the stratification-risk algorithm of patients, allowing for an easier choice of treatments, based on the hypothesis that high cfDNA amount may suggest a more biologically aggressive disease that may benefit from a drug combination or a more intensive therapeutic approach.
Declarations
Competing interests
Marzia Del Re—consultant/speaker: Astellas, AstraZeneca, Celgene, Novartis, Pfizer, Bio-Rad, Janssen, Sanofi-Aventis, Ipsen, Roche Elena Verzoni—consultant/advisor: Ipsen, Merck, Novartis, Pfizer. Giuseppe Procopio—consultant advisor: Astra Zeneca, Bayer, Bristol Myers Squibb, Ipsen, Merck, Novartis, Pfizer; research grants: Ipsen, Novartis. Camillo Porta—consultant/speaker: Angelini, AstraZeneca, Bristol Myers Squibb, Eisai, EUSAPharma, General Electric, Ipsen, Janssen, Merck, MSD, Novartis, Pfizer; expert testimony: EUSAPharma, Pfizer; Protocol Steering Committee Member: Bristol Myers Squibb, Eisai, EUSAPharma. Romano Danesi—consultant/speaker: Ipsen, Novartis, Pfizer, Sanofi Genzyme, AstraZeneca, Janssen, Gilead, Lilly, Gilead and EUSAPharma.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.