Multiomics Approach in Metastatic Clear Renal Cell Carcnoma

IMPACT In the last ten years the systemic treatment of metastatic renal cell carcinoma has been revolutioned with the introduction of at least ten active drugs. With the advent of novel immuno-based and tyrosine kinase inhibitors (TKIs)-based combinations, the choice of the best strategy in treatment-naive metastatic clear-cell renal cell carcinoma (mccRCC) patients (pts) is becoming an issue, since no biomarkers are available to guide the treatment allocation strategy . In recent clinical trials, combination therapies including nivolumab plus ipilimumab, pembrolizumab plus axitinib, atezolizumab plus bevacizumab, avelumab plus axitinib, pembrolizumab plus lenvatinib and nivolumab plus cabozantinib exhibited significant benefits in terms of overall survival (OS) and/or progression-free survival (PFS) for mRCC compared with sunitinib as a standard first-line treatment for mRCC . However, there is a clear need for clinical predictive biomarkers to guide optimal treatment decisions. Through the above research the investigators are confident to provide proof of concept that combine the informations from genomics and radiomics using computational approaches such as machine learning, will provide an opportunity for a molecularly driven patient's stratification.

RATIONALE AND FEASIBILITY

BIOMARKERS Risk stratification models based on gene expression pattern (both messenger and long non-coding RNA) in ccRCC have proven to have strong prognostic values. Hence, there is an interest in the identification and development of treatment predictive biomarkers to enable precision oncology increasing drug response. Multiple candidates for predictive biomarkers from plasma, tumor, and host tissues have been explored in patients with metastatic renal-cell carcinoma who are receiving systemic therapies, but, as yet, none have entered clinical practice and all require prospective validation in clinical trials.

In the era of VEGF inhibitors, the investigators counted on IMDC (International Metastatic RCC Database Consortium) model, considering Karnofsky performance status <80, time to initiation of therapy <1 year, hemoglobin < lower level of normal, serum calcium, neutrophil count, and platelet count > upper limit of normal. CheckMate 214 study showed that OS and ORR were significantly higher with nivolumab plus ipilimumab than with sunitinib among intermediate- and poor-risk pts. Extended study follow-up of KEYNOTE-426 study demonstrated that the benefit in OS and PFS is consistent in this class of patients's risk also with the IO/TKI combination therapy. The IMDC score is confirmed to be prognostic in every combos study.

PD-L1 has also been demonstrated to be a prognostic marker for poor prognosis in RCC, regardless of the type of treatment used. More recently PD-L1 expression has also been evaluated for its predictive role that is only partially confirmed in the CheckMate-214 population that received IO-IO combo and considered a poor marker for targeted therapies. Gene signatures from ImMotion 150 and 151 evidenced that two different signatures (angiogenesis versus immune signature) in RCC patients could predict the response to combo treatment.

Metabolomics have also been assessed as potential biomarkers for RCC giving new insights into the understanding of RCC clinical behavior and for the development of new therapeutic strategies. Both tumor tissue and blood are interesting source for the study of potential biomarkers. Plasma samples in particular have been analyzed to better understand the role of components of the proangiogenic and cellular proliferation pathways. Another topic is the study of cytokines, circulating endothelial cells, and gene expression controlling mechanisms. Moreover, germline genetic variations in important genes related to drug mechanism-of-action and metabolism have been under investigation as well as factors implicated in gene expression regulation by epigenetic mechanisms or by post-transcriptional regulation.

RADIOMICS Computed tomography (CT) is widely available, routinely used in the care of patients with metastatic tumors treated with antiangiogenic therapy and yields quantitative digital data. Many studies have used CT noninvasive imaging-based methods to assess the pathologic grade of renal tumor before surgery. Various radiological features such as tumor size and pattern enhancement have been shown to correlate with tumor grade. However, it is difficult to predict the pathologic grade of renal tumor with only information obtained from traditional radiologic features. Conversely, radiomics analysis involves the automatic extraction of data not recognizable to the human eye resulting in highly detailed imaging features regarding tumor structure, shape and image intensity. Radiomics may provide a novel approach to develop predictive tools by correlating imaging features to tumor characteristics including histology, tumor grade, genetic patterns and molecular phenotypes, as well as clinical outcomes. Extracting data from imaging the aim is to provide information beyond what can be achieved from human imaging interpretation alone.

In clinical practice, the prediction of RCC aggressiveness through imaging findings remains a challenge. A retrospective study by Shu et al. demonstrated that radiomics features could be used as biomarkers for the preoperative evaluation of the ccRCC Fuhrman grades. In the post-treatment setting, radiomics may assist in predicting a response to systemic therapy, including to antiangiogenic treatment, which may not be adequately assessed with traditional size-based criteria.

Smith and colleagues used a custom post-processing software and algorithm to develop a novel system to quantify changes in the amount of vascularized tumor within specific attenuation thresholds, termed the vascular tumor burden. This semi automated biomarker, in addition to other tumor metrics, such as length, area, and mean attenuation, were used to predict response to antiangiogenic therapy with sunitinib. Changes in the vascular tumor burden metric on initial post-therapy imaging after the initiation of sunitinib showed a better separation of progression free survival between non-responders and responders compared with other commonly used response criteria changes in tumor metrics, including length, area, mean attenuation, RECIST, CHOI, modified CHOI, MASS, and 10% sum long diameter. Extension of radiomic analysis through radiogenomics, radiometabolomics, and correlation with other epidemiologic, clinical, and tissue-based datasets have the potential to improve patient management in the era of personalized medicine. Understanding what these technologies can offer will allow radiologists to play a larger role in the care of patients with RCC.

CT dynamic contrast-enhanced is a capable tool to quantify tumor enhancement and its response to anti-angiogenetic therapies. Han et al found a correlation between tumor renal enhancement at baseline and response and PFS after treatment with sunitinib or sorafenib. In contrast, other studies have demonstrated that although the perfusion parameters at baseline were higher in patients with longer survival times, they were not significantly predictive of outcome, except when a cut-off analysis was established. Other CT-based methods for assessing tumor response to anti-angiogenic therapy and predicting clinical outcome are undergoing further evaluation. Among them, radiomics CT features such as heterogeneity, entropy, and texture uniformity are additional parameters that show promise for assessing the anti-angiogenic response of metastatic renal cell carcinoma. The correlation of these imaging data with genomics (ie, radiogenomics), metabolomics (ie, radiometabolomics) and beyond, offers an opportunity to generate objective, quantitative biomarkers of tumor biology that may be used to predict patient's prognosis and likelihood of response to therapy, overcoming some of the challenges associated with disease heterogeneity.

ARTIFICIAL INTELLIGENCE Artificial intelligence systems, in particular those based on Machine Learning and Deep Learning, are able to autonomously identify salient patterns and complex relationships among data by just looking at sample populations. Their ability to process heterogeneous data, for both classification and prediction purposes, may provide a valid contribution to better stratify RCC patients. Recently, some works have attempted to use Machine and Deep Learning for the differentiation between benign and malignant small renal masses, based on textural analysis of CT scans, for the prediction of the Fuhrman nuclear grade, and gene expression-based molecular signatures.

Overall, the stratification of RCC patients still remains a challenging task, especially when considering the molecular heterogeneity of kidney tumors.

The ability to combine the information from genomics and radiomics using computational approaches based on Machine Learning, provides an opportunity to re-classify patients into subgroups that could better guide treatment strategies. Both supervised and unsupervised techniques can be used to identify a biomarker signature-score as well as to predict response/resistance to therapies.

PRELIMINARY DATA The investigators have preliminary biological data on blood samples of 32 mRCC patients obtained at treatment baseline. The majority of patients received as first line the combination of nivolumab and ipilimumab, 9 patients received sunitinib, 5 were administered with pazopanib, and 6 patients received cabozantinib. The molecular analysis was conducted using the NGS OncoMine Solid Tumor Panel (Thermo Fisher). Twentyfive patients (78,1%) were carriers of a molecular alteration in one or more genes including: TP53, mTOR, PIK3CA, BRAF, EGFR, RET, GNAS, SF3B1, PDGFRA. The reported allele frequency ranged between 3% and 12%. The preliminary analysis shows that patients with multiple mutations have a better response (PR+SD vs PD) from immunotherapy.

EXPERIMENTAL DESIGN This is a multicentric prospective translational study evaluating genomics and radiomics in treatment- naïve advanced ccRCC pts candidate to receive first-line systemic therapy. Nine centers in Italy, including: Istituto Nazionale dei Tumori (INT) of Milan, European Institute of Oncology (IEO) of Milan, Istituto Oncologico Veneto (IOV), Policlinico San Martino of Genova, Istituto Nazionale dei Tumori of Napoli, University Hospital of Parma, Humanitas Research Hospital of Milan, Oncological Center of Aviano and Fondazione Policlinico Universitario Agostino Gemelli of Rome will be involved in the accrual and the treatment of patients.