THE ESCAT SCALE

Ranking genomic alterations

By prioritising genomic alterations by their evidence-based clinical actionability, the ESMO Scale for Clinical Actionability of molecular Targets (ESCAT) aims to support oncologists in taking clinical decisions. While previous scales were tailored to specific national or institutional programmes, ESMO's tool is developed to go beyond regulatory approval specific to geographical location.


In August 2018, leading cancer specialists in Europe and North America agreed on a new scale for tumour DNA mutations to simplify and standardise choices for targeted cancer treatment. The ESMO Scale for Clinical Actionability of molecular Targets (ESCAT) classes alterations in tumour DNA according to their relevance for selecting patients for targeted treatment, based on the strength of clinical evidence supporting them. Targets are ranked from Tier I (targets ready for implementation in routine clinical decisions) to Tier V (evidence supporting co-targeting approaches), with an additional Tier X category for targets lacking evidence for actionability. The ESCAT Scale represents the first classification relevant to all potential targeted cancer medicines, not just those with national regulatory approval. This dynamic classification also allows upgrading/downgrading of mutations in response to newly available data. Here, some of the experts involved in developing the ground-breaking scale speak about the need for this type of classification, how it was developed and how it will help doctors to make treatment decisions.

Fabrice André

Institut Gustave Roussy, Villejuif, France; Chair ESMO Translational Research and Precision Medicine Working Group

Can you explain why the ESCAT Scale was developed?

Many doctors are already using mutation sequencing in the context of access programmes/clinical trials or daily practice and we have observed that there is no standardised way in which genomic alterations are reported. Also, reporting generally comprises simply a list of genomic alterations, without any prioritisation for the doctor or patient and no indication as to whether targeting a particular alteration could be potentially therapeutic.

"The aim of the ESCAT is to prioritise, standardise and better inform the oncologist and patient about the likelihood of benefit from targeting an alteration."

The likelihood of benefit will drive the effort required to access appropriate drugs. For example, if all genomic alterations are ranked Tier IV, the likelihood of benefit is unknown, probably minimal. But if there is one Tier II genomic alteration, access to appropriate targeted drugs could provide real benefit.

To what extent have previous classifications proposed in recent years been implemented?

Previous attempts to propose classifications have been mainly specific to institutions. The idea of the ESCAT was to provide one global classification that would update and merge all the local ranking systems. In addition to providing commonality of language and focusing efforts on validating only one ranking system, this would enable comparisons to be made between all the different precision medicine trials. Furthermore, the ESCAT built on this foundation by including additional concepts, such as how to rank a target validated in another disease and when a combination targeted approach is appropriate.

What challenges are you expecting to the adoption of the scale and how can they be overcome?

Luckily, there are no regulatory challenges, as the ESCAT is not a regulatory tool. Instead, the biggest challenges are showing doctors that the scale is reproducible and that it has clinical utility. The recommendation by future clinical guidelines of the ESCAT for the classification of genomic alterations is crucial. In addressing these challenges, ways to validate the scale, determine its reproducibility across doctors and ensure that it is implemented and disseminated as widely as possible are being investigated. What is missing now are data that if we stratify a patient based on the ESCAT, then the value of sequencing is increased for that patient. In this respect, retrospective analyses of precision medicine trials, stratifying patients according to the ESCAT ranking, should reveal if the ESCAT correctly predicted the benefit of matched therapy.

The ESCAT

Joaquin Mateo

Vall d'Hebron Institute of Oncology, Barcelona, Spain

Can you describe how the expert group went about constructing the tier system and deciding which mutations should appear at which stage?

Initially, the group of experts had several rounds of discussion to propose classification criteria. Once these criteria were drafted, we discussed examples to ensure that they were clear and easy to use. The goal was not to derive a comprehensive list of mutations and classify them, but to provide clear wording defining tiers that have different impacts in clinical practice. Classification of individual mutations is ongoing and this is something that will change over time as new data emerge. It is important to remember that this is a rapidly evolving field and the classification system aims to facilitate up-or down-grading of mutations within the tiers as new studies demonstrate the clinical value of genomic findings.

How does this work reflect the need to combine different areas of expertise in precision medicine?

This classification system is a reflection of the understanding that sequencing data are only useful if we have the tools to interpret them. Without such tools, the information cannot be used for medical decision making. We believe that in addition to the usual oncology team, including surgeons, oncologists and nurses, multidisciplinary tumour boards will need to incorporate new specialities, such as genetic counsellors and bioinformaticians. With the ESCAT, we aim to give these individuals a common language so that they can understand each other when discussing genomics data.

"The ESCAT will provide a common language on genomic sequencing for all healthcare professionals involved in cancer care."

To what extent has the ESCAT implemented classifications proposed in recent years?

In generating ESCAT, we engaged key investigators from Europe and the US who were involved in previous classification systems and we discussed with them why these prior scales have or have not been widely implemented. We observed that, in general, previous scales were tailored to specific national or even institutional programmes. With the ESCAT, we wanted to develop a classification system that would be useful beyond national regulatory particularities.

What are the current challenges that may prevent the ESCAT from being adopted globally?

We need to prove reproducibility of the scale. Thus, the criteria should be sufficiently clear to ensure that a particular event is classified in the same tier by different people. To investigate this, we will soon be engaging a number of oncologists across Europe to participate in a test survey. Then, we need to talk and listen to oncologists, laboratory physicians and regulatory bodies to see how we can together implement the system in a way that helps everyone. In fact, this is the ultimate aim, to facilitate the understanding of genomics data to allow the use of these tests in clinical practice. We have seen already a paper reporting a comprehensive classification of breast cancer genomic events using the ESCAT system and we envision similar efforts to be pursued in other tumour types soon. In parallel, we will work with academic laboratories and private partners to discuss how to implement the ESCAT into reports of real-world genomic testing.

Debyani Chakravarty

Memorial Sloan Kettering Cancer Center, New York, USA

To what extent has the ESCAT implemented classifications proposed in recent years?

The ESCAT built upon the principles from previously proposed classifications and created a framework that prioritises genomic alterations by its evidence-based clinical actionability. This framework ranks alterations based on 1) whether data have been generated via randomised and/or prospective clinical trials; 2) clinical implications in the context of specific tumour types that distinguish routine-use versus investigational targets; 3) extrapolation from preclinical data that address mutation impact and sensitivity to drugs in experimental systems and 4) statistical recurrence of genomic alterations in patient tumour samples.

In terms of tools and data interpretation, how has multiplex sequencing changed precision medicine in recent years and what challenges have emerged with the advent of Big Data in cancer genomics?

Precision medicine in cancer care in the last decade has been driven by the shift from single analyte tests and small hotspot panels to larger gene panels and whole exome/whole genome sequencing. Currently, in the US, there are over 40 unique indications for which standard care includes routine tumour sequencing to detect specific genomic biomarkers to guide patient care. However, with this shift has come the considerable challenge of interpreting datasets that have increased by orders of magnitude, so-called “Big Data”. The challenge is to separate clinically actionable driver mutations from drivers of tumour growth with no clinical implications and from passenger mutations which have no influence on patient prognosis or treatment. Furthermore, the number and type of end-users of these data are diverse, with varying levels of knowledge and expertise in cancer genomics ranging from community oncologists to drug developers to oncology research scientists. The last several years have therefore witnessed the development of bioinformatic tools that address this interpretation challenge of genomics data within the context of clinical oncology. Knowledgebases link different cancer-relevant mutations or structural variants to their specific clinical actionability, if any exists. They do this by storing information from clinical oncology guidelines, Food and Drug Administration drug labels, the scientific literature, and proceedings from the major clinical oncology conferences in structured databases.

"The use of data repositories is limited due to difficulties in translating genomic cancer data into clinical decisions in an easy and practical way. The ESCAT provides a unified framework that is easily interpretable and accessible."

How would the use of the ESCAT simplify interpretation of genomic variations, in particular unknown and uncommon variants?

The greatest advantage of the ESCAT is that it provides a unified framework that is easily interpretable and accessible to all stakeholders in the clinical oncology space. Previous classification systems have been largely based on the regulatory approval specific to the geographic location where the scale has been developed. Instead, the ESCAT ranks clinical actionability of the target based on clinical trial outcomes and degree of benefit. The ESCAT also incorporates the different structures of clinical trials, including basket trials, and gives clear guidance regarding the utility of targets, particularly those in lower tiers of actionability, using a shared terminology system.

What are the current limitations to the use of data repositories? How would these affect any query of clinical evidence?

A key limitation to the use of data repositories is their ability to communicate, in a facile and practical way, the necessary information that translates dense cancer genomic data into clinical decision. For example, for a clinical oncologist wanting to use a data repository, those that provide every detail about the biological role of a cancer genomic biomarker and its scientific nuances will detract from the oncologist’s immediate goal of discerning which genomic alteration out of the patient’s molecular profile is considered a bona-fide evidence-based biomarker predictive of response to a targeted therapy. A possible negative outcome of this scenario is that the oncologist misses the opportunity to enrol a patient in a clinical trial testing a targeted therapy with proven clinical value. The lack of a standardised way of communicating the value of genomic biomarkers further contributes to the low rate of patient enrolment into precision medicine clinical trials. Indeed, in the US, only approximately 3% of patients enrol in a clinical trial overall. The ESCAT thus provides just such a standardised clinical utility-based ranking system that may be applied by any data repository.

How does the ESCAT collaborative paper reflect the increasing relevance of including a molecular expert in multidisciplinary oncology teams?

Precision medicine bridges clinical oncology and cancer genomics. Oncologists’ primary focus is the clinical and regulatory landscape in which they treat patients and the efficacy and toxicity considerations of specific treatments for their patients based on historical clinical data. On the other hand, molecular experts focus on why a specific cancer genome marker may be predictive of response to a specific targeted agent in experimental model systems. A multidisciplinary oncology team has the potential to successfully bridge both worlds and provide easy-to-understand interpretations of molecular data for optimal treatment decisions. Each of the ESCAT tiers gives clear examples that include a molecular rationale behind why the biomarker is predictive of therapeutic response, the supporting clinical data that include potential on-target toxicity concerns and the outcome data for the targeted agents in biomarker-defined patient subpopulations.

Lajos Pusztai

Yale Cancer Center, New Haven, USA

Can you describe how the ESCAT scale works and how it will simplify treatment decisions?

Currently, genomic anomalies revealed by molecular profiling results from academic and commercial laboratories represent a mixed bag of findings, with US Food and Drug Administration/ European Medicines Agency-approved molecular targets with clear therapeutic implication (such as HER2 amplification or BRAFV600E mutation) sitting alongside highly hypothetical targets (for example, CCND1 amplification or NF1 gene frame shift mutation, like F1472fs*11). What has been missing is an annotation system that clearly identifies the clinical utility of a given genomic variant. The ESCAT provides both the framework for this annotation and guidelines on how to perform the annotation. It is important to remember that the clinical utility of particular markers will change over time, but the guiding principles for the annotation will not.

To what extent has the ESCAT implemented classifications proposed in recent years?

Because many investigators and clinicians have recognised the need for a better annotation of tumour molecular target profiling results, many different groups have proposed clinical utility annotation systems that are very similar in philosophy. However, none of them has been widely adopted. In order to create a uniform annotation guideline that would be broadly adopted, ESMO invited many of the investigators who proposed clinical utility schemes to be involved in the development of the ESCAT.

Will the tiering system provide a more consistent approach to the treatment of patients with multiple mutations?

We hope that the tiering system will help physicians to prioritise mutations. The danger with the current situation is that a truly actionable mutation may not be acted on while another mutation that is highly hypothetical in its target value (i.e. FGR3 amplification) is.

How would the ESCAT impact on shared decision making between the oncologist and the patient? What would be the benefits for patients?

The benefits are the same for the patient and the physician. The ESCAT-based ranking of detected molecular abnormalities gives information about the clinical utility of the findings and should help with shared decision-making regarding what therapy to select.