IJC

International Journal of Cancer

Variant classification in precision oncology

Jonas Leichsenring1*, Peter Horak2,3*, Simon Kreutzfeldt2,3, Christoph Heining4,5, Petros Christopoulos6,7,
Anna-Lena Volckmar1, Olaf Neumann1, Martina Kirchner1, Carolin Ploeger1, Jan Budczies 1, Christoph E. Heilig2,
Barbara Hutter8, Martina Fröhlich8, Sebastian Uhrig8,9, Daniel Kazdal 1, Michael Allgäuer1, Alexander Harms1,
Eugen Rempel1, Ulrich Lehmann10, Michael Thomas 6,7, Nicole Pfarr11, Ninel Azoitei12, Irina Bonzheim13, Ralf Marienfeld14,
Peter Möller14, Martin Werner15, Falko Fend13, Melanie Boerries3,8,16,17, Nikolas von Bubnoff 3,8,18,19, Silke Lassmann15,
Thomas Longerich1,20, Michael Bitzer21, Thomas Seufferlein12, Nisar Malek21, Wilko Weichert11, Peter Schirmacher1,3,
Roland Penzel1, Volker Endris 1, Benedikt Brors2,3,8, Frederick Klauschen22, Hanno Glimm4, Stefan Fröhling 2,3,23† and
Albrecht Stenzinger 1,23†
1
Institute of Pathology, University Hospital Heidelberg, Heidelberg, Germany
2
National Center for Tumor Diseases (NCT) Heidelberg, Heidelberg, Germany
3
German Cancer Consortium (DKTK), Heidelberg, Germany
4
National Center for Tumor Diseases (NCT), Dresden, Germany
5
Cancer Genetics and Epigenetics

University Hospital Carl Gustav Carus, Dresden, Germany

6
Thoraxklinik, Heidelberg, Germany
7
Translational Lung Cancer Research Heidelberg (TLCR-H), German Center for Lung Research (DZL), Giessen, Germany
8
German Cancer Research Center (DKFZ), Heidelberg, Germany
9
Faculty of Biosciences, Heidelberg University, Heidelberg, Germany
10
Institute of Pathology, Hannover Medical School, Hanover, Germany
11
Institute of Pathology, Technische Universität München, Munich, Germany
12
Clinic of Internal Medicine I, University Hospital Ulm, Ulm, Germany

Key words: molecular pathology, variant classification, molecular tumor board, next-generation sequencing
Abbreviations: ACMG: American College of Medical Genetics and Genomics; AMP: Association for Molecular Pathology; ASCO: American
Society of Clinical Oncology; CAP: College of American Pathologists; CNV: copy number variation; CRC: colorectal cancer; DKFZ: German
Cancer Research Center; DKTK: German Cancer Consortium; ESCAT: ESMO Scale for Clinical Actionability of molecular Targets; ESMO:
European Society for Medical Oncology; IASLC: International Association for the Study of Lung Cancer; Indel: insertion/deletion; JCR: joint
consensus recommendation; MTB: molecular tumor board; NCT: National Center for Tumor Diseases; NGS: next-generation sequencing;
OSoverall survival; PARP: poly (ADP-ribose) polymerase; PFS: progression-free survival; RR: response rate; SNV: single nucleotide variants;
TKI: tyrosine kinase inhibitor; TMB: tumor mutational burden; TRK: tropomyosin receptor kinase; VUS: variant of unknown significance;
WES: whole-exome sequencing; WGS: whole-genome sequencing
Additional Supporting Information may be found in the online version of this article.
Grant sponsor: DKFZ-HIPO; Grant number: H021
Conflict of interests: AS: advisory board honoraria from Bayer, Novartis, BMS, AstraZeneca, ThermoFisher, Illumina; speaker’s honoraria
from Takeda, Roche, BMS, Illumina, AstraZeneca, Novartis, ThermoFisher, Bayer, MSD and research funding from Chugai and BMS. VE:
advisory board honoraria from Thermo Fischer. Speaker’s honoraria from AstraZeneca. ALV: speaker’s honoraria from AstraZeneca. PS:
advisory board honoraria, research funding and speaker’s honoraria from Novartis and AstraZeneca. WW: Advisory board and speaker’s
bureau from Roche, MSD, BMS, AstraZeneca, Pfizer, Merck, Lilly, Boehringer, Novartis, Takeda. Research funding from Roche, MSD, BMS,
Bruker. PC: Speakers honoraria from Roche, Novartis, Chugai. NP: Speaker’s honoraria and travel costs from Novartis, Roche, BMS. SL:
Research funding from Agilent, Illumina, Qiagen, BMS. Advisory boards: AstraZeneca, Novartis, Roche. Travel grants from Illumina, Agilent.
MW: Advisory boards: Roche, Novartis. Research funding from Agilent, Novartis, Roche. Honoraria and travel grants from Diakovere. MT:
advisory board honoraria from Novartis, Lilly, BMS, MSD, Roche, Celgene, Takeda, AbbVie, Boehringer, speaker’s honoraria from Lilly, MSD,
Takeda, research funding from AstraZeneca, BMS, Celgene, Roche, and travel grants from BMS, MSD, Novartis, Boehringer. NvB: Honoraria
from AstraZeneca, Amgen, Novartis, BMS. Research funding from Novartis. All other authors declared no conflicts of interest.
*J.L. and P.H. shared first authorship
†
S.F. and A.S. shared last authorship
[Correction added on September 21, 2019 after first online publication: affiliation updated.]
DOI: 10.1002/ijc.32358
History: Received 31 Jan 2019; Accepted 8 Apr 2019; Online 22 Apr 2019
Correspondence to: Stefan Fröhling, National Center for Tumor Diseases (NCT) Heidelberg, Im Neuenheimer Feld 460, 69120 Germany,
Tel.: +49-6221-566990, Fax: +49-6221-566967, E-mail: stefan.froehling@nct-heidelberg.de; or Albrecht Stenzinger, Institute of Pathology,
University Hospital Heidelberg, Im Neuenheimer Feld 224, 69120 Heidelberg, Germany, Tel.: +49-6221-5634380, Fax: +49-6221-565251,
E-mail: albrecht.stenzinger@med.uni-heidelberg.de

Int. J. Cancer: 145, 2996–3010 (2019) © 2019 UICC

 10970215, 2019, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32358, Wiley Online Library on [26/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Leichsenring et al. 2997

13
Institute of Pathology, University Hospital Tübingen, Tübingen, Germany
14
Institute of Pathology, University Hospital Ulm, Ulm, Germany
15
Institute of Pathology, Medical Center, University of Freiburg, Breisgau, Germany
16
Institute of Molecular Medicine and Cell Research, University of Freiburg, Freiburg, Germany
17
MIRACUM Consortium of the Medical Informatics Initiative, Freiburg, Germany
18
Department of Hematology, Oncology and Stem Cell Transplantation, Medical Center, Faculty of Medicine, University of Freiburg, Freiburg, Germany
19
Department of Hematology and Oncology, Medical Center, University of Schleswig-Holstein, Lübeck, Germany
20
Heidelberg-Göttingen-Hannover Medizininformatik (HiGHmed) Konsortium, Heidelberg, Germany
21
Department of Internal Medicine I, University Hospital Tübingen, Tübingen, Germany
22
Institute of Pathology, Charité Universitätsmedizin Berlin, Berlin, Germany
23
DKFZ-Heidelberg Center for Personalized Oncology (HIPO), Heidelberg, Germany

Next-generation sequencing has become a cornerstone of therapy guidance in cancer precision medicine and an indispensable
research tool in translational oncology. Its rapidly increasing use during the last decade has expanded the options for targeted
tumor therapies, and molecular tumor boards have grown accordingly. However, with increasing detection of genetic alterations,

Cancer Genetics and Epigenetics

their interpretation has become more complex and error-prone, potentially introducing biases and reducing benefits in clinical
practice. To facilitate interdisciplinary discussions of genetic alterations for treatment stratification between pathologists,
oncologists, bioinformaticians, genetic counselors and medical scientists in specialized molecular tumor boards, several systems
for the classification of variants detected by large-scale sequencing have been proposed. We review three recent and commonly
applied classifications and discuss their individual strengths and weaknesses. Comparison of the classifications underlines the
need for a clinically useful and universally applicable variant reporting system, which will be instrumental for efficient decision
making based on sequencing analysis in oncology. Integrating these data, we propose a generalizable classification concept
featuring a conservative and a more progressive scheme, which can be readily applied in a clinical setting.

What’s new?
With increasingly comprehensive molecular profiling of cancers, the clinical interpretation of genetic alterations is becoming
more and more challenging. Here the authors review several classifications for gene variant interpretation that were recently
introduced to guide clinical management. They highlight shared features and differences and point out major influencing
factors and unresolved issues that still need to be addressed. Based on this analysis, they propose a unified classification
concept that may become broadly implemented when remaining issues are solved.

Introduction alterations beyond drug-approved entity-specific targets and

During the last few years, the use of diagnostic genetic analyses
of tumor samples has markedly expanded, facilitating the
growth of precision oncology through the identification of a
continuously increasing number of therapeutic targets and
molecularly defined stratification of patients.1–7 Panel-based
next-generation sequencing (NGS) for the detection of tumor-
specific somatic mutations is already widely available in routine
clinical use8 and a number of bioinformatics tools have been
developed for their assessment, including data quality, which is
paramount for all subsequent steps (for summaries of these
tools see Refs. 9,10). While most current targeted NGS
approaches interrogate a few kilobases of the tumor exome to
identify specific targets for which approved drugs exist, more
comprehensive sequencing efforts ranging from larger gene
panels detecting aberrations on both the DNA and RNA level
to whole-exome and-genome sequencing (WES/WGS) as well
as transcriptome sequencing are entering routine diagnostics.4
As a direct consequence, an increasing number of genetic
combinations thereof are identified in a broad range of tumor
entities, offering potential access to off-label treatments and
molecularly stratified clinical trials. With growing complexity of
genetic data available for individual patients, pathological alter-
ations have to be interpreted according to predefined standards
and eventually reviewed in interdisciplinary as well as disease-
specific molecular tumor boards (MTB) engaging pathologists,
oncologists, genetic counselors, bioinformaticians and medical
scientists. To facilitate and standardize communication and dis-
cussion of molecular findings and their clinical relevance, sev-
eral classification systems have been proposed recently,4,9–17 but
no definitive universal standard has been agreed upon.
Supporting Information Appendix S1 provides an overview of
all major classifications that have been published recently.
A recent study18 compared the assessment of four fictional
patients by five MTBs from four countries. MTBs were pro-
vided with curated mutational profiles without information on
primary sequencing data. For all patients, different treatment

Int. J. Cancer: 145, 2996–3010 (2019) © 2019 UICC

 10970215, 2019, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32358, Wiley Online Library on [26/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
2998 Variant classification in precision oncology

recommendations were made by each MTB and only for one
case three MTBs agreed on the same treatment recommenda-
tions. Importantly, heterogeneity of such recommendations
increased with the complexity of genomic information avail-
able. Differences were attributed to insufficient evidence for
specific alterations as well as to uncertainty regarding the
impact of cooccurring genetic alterations. It is noteworthy that
in a recent survey by the Association for Molecular Pathology
(AMP), only 12% and 5% of participating members reported
experience with WES and WGS, respectively.10 This highlights
the importance of careful interpretation and standardized
reporting in light of ever more complex diagnostic data.
We analyze three recent and commonly applied approaches
to variant classifications with emphasis on their clinical implica-
tions. Moreover, we devise an integrated classification approach
Cancer Genetics and Epigenetics
National Center for Tumor Diseases (NCT) in Heidelberg4
and implemented by other institutions nationwide.6
The proposed classifications are supposed to facilitate
interdisciplinary communication and comparability of results,
based on a wide spectrum of tumor genetic alterations ranging
from established molecular markers with direct clinical impli-
cation (e.g., EGFR L858R in nonsmall cell lung cancer
[NSCLC]) to alterations with currently unclear/unknown or
no clinical significance.
Joint consensus recommendation of AMP, ACMG, ASCO
and CAP
The classification jointly proposed by major clinical and patho-
logical associations in the U.S.10 employs a four-tiered system
(Table 1), which considers single nucleotide variants (SNV),

based on the comparative data, which can be readily applied in
a clinical scenarios and will support harmonization of genetic
variant annotation, which is paramount for full exploitation
of the precision oncology concept.
Overview on Current Classifications
We analyzed the three most prominent classifications that
have been developed over the last years: (i) A joint consensus
recommendation (JCR) for the interpretation and reporting of
sequence variants in cancer by the Association for Molecular
Pathology (AMP), American College of Medical Genetics and
Genomics (ACMG), American Society of Clinical Oncology
(ASCO) and College of American Pathologists (CAP)10; (ii) A
framework for ranking molecular targets for cancer precision
medicine by the European Society for Medical Oncology
(ESMO)9; (iii) A classification developed for the MTBs within
the German Cancer Consortium (DKTK) applied at the
insertions/deletions (indels), gene fusions and copy number vari-
ations (CNV). Genetic alterations are classified according to their
clinical impact based on the level of clinical and experimental
evidence. Joint consensus recommendation (JCR) Tier 1 (strong
clinical significance) categorizes alterations with direct clinical
implication in a specific tumor type in terms of FDA-approved
therapies, inclusion in professional guidelines with expert con-
sensus as well as results from well-powered clinical studies. JCR
Tier 2 (potential clinical significance) includes alterations with
FDA-approved therapies in a tumor type other than the tumor
type analyzed, investigational therapies, meta-analyses with some
consensus and preclinical studies as well as case reports. JCR
Tiers 3 and 4 (unknown clinical significance and benign variants,
respectively) include alterations without convincing evidence or
lacking evidence of cancer association (so far). The JCR also
acknowledges the importance of alterations associated with resis-
tance to targeted therapies. Uniquely, this classification also
includes diagnostic and prognostic biomarkers.

Table 1. Joint consensus recommendation of AMP, ACMG, ASCO and CAP for reporting genetic variants in cancer (From Li et al., J Mol Diagn,
2017, 19, 4–23 Elsevier, adapted by permission10)
Tier LoE Explanation
Tier 1 A.1 Biomarkers that predict response or resistance to FDA-approved therapies for a specific Variants of strong
type of tumor clinical
significance
A.2 Biomarkers included in professional guidelines that predict response or resistance to
therapies for a specific type of tumor
B Biomarkers that predict response or resistance to therapies for a specific type of tumor
based on well-powered studies with consensus from experts in the field
Tier 2 C.1 Biomarkers that predict response or resistance to therapies approved by the FDA or Variants of
professional societies for a different type of tumor potential clinical
significance
C.2 Biomarkers that serve as inclusion criteria for clinical trials
D Biomarkers that show plausible therapeutic significance based on preclinical studies
Tier 3 Not observed at a significant allele frequency in the general or specific subpopulation Variants of
databases, or pan-cancer or tumor-specific variant databases. No convincing unknown clinical
published evidence of cancer association significance
Tier 4 Observed at significant allele frequency in the general or specific subpopulation Benign or likely
databases. No existing published evidence of cancer association benign variants

Int. J. Cancer: 145, 2996–3010 (2019) © 2019 UICC

 10970215, 2019, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32358, Wiley Online Library on [26/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Leichsenring et al. 2999

ESMO scale for clinical actionability of molecular targets
(ESCAT)
The classification developed by the ESMO9 focuses on clinical
relevance (Table 2). No specific molecular platform or assay
was considered during the development. Similar to Tier 1 of
the JCR, ESCAT I includes genetic alterations with matching
“standard of care” therapies, including alterations with a
defined clinical benefit based on ESMO Magnitude of Clinical
Benefit Scale (MCBS) 1.1,19 requiring the evaluation of single-
arm clinical studies by the MTB for ESCAT I-B. Within the
MCBS, progression-free survival (PFS), overall survival (OS),
quality of life as well as grade 3 and 4 toxicities are taken into
consideration. Additionally, within ESCAT I, alteration-drug
matches in basket trials with clinical benefit across tumor
types are included. ESCAT II includes alterations with
unknown magnitude of benefit. This category also considers
includes alterations for which objective responses have been
shown in the absence of improved outcome, which might
show potential benefit in combination with other therapies.
Finally, ESCAT X categorizes alterations lacking actionability.
German National Center for Tumor Diseases classification
At the National Center for Tumor Diseases (NCT) in Heidelberg,
a classification to assess molecular biomarkers beyond approved
“standard of care” targets was established in 2015,4 which
has been modified and consented and is used within DKTK.
The MTB at NCT was originally launched to determine
therapeutic choices for individual patients enrolled in the Molec-
ularly Aided Stratification for Tumor Eradication Research
(MASTER) program, a multicenter registry trial for prospec-
tive, biology-driven stratification of younger adults with

Cancer Genetics and Epigenetics

results from retrospective analyses showing clinically mean-
ingful benefit (i.e., investigational therapeutic options). ESCAT
III includes alteration-drug matches within different tumor
types as well as (downstream) alterations within the same sig-
naling pathway as ESCAT I and II biomarkers. ESCAT IV
groups alterations with preclinical evidence of utility as well as
those with predicted actionability in silico, while ESCAT V
advanced-stage cancer across all histologies and patients with
rare tumors. Individual molecular alterations detected by
WES/WGS and transcriptome sequencing are validated at the
Center for Molecular Pathology, Institute of Pathology, Univer-
sity Hospital Heidelberg,4,7 including SNVs, CNVs, gene fusions
and gene expression changes. Actionable alterations are catego-
rized according to their level of clinical (NCT evidence levels

Table 2. ESMO scale for clinical actionability of molecular targets (ESCAT; From Mateo et al., Ann Oncol, 2018, 29, 1895–1902 Oxford
Academic, adapted by permission9)
ESCAT LoE Explanation
I I-A Prospective, randomized clinical trial(s) show the alteration-drug match in a specific Drug match is associated
tumor type results in a clinically meaningful improvement of a survival endpoint with improved outcome
I-B Prospective, nonrandomized clinical trial(s) show that the alteration-drug match in a in clinical trials
specific tumor type results in clinically meaningful benefit as defined by ESMO
MCBS 1.1
I-C Clinical trials across tumor types or basket clinical trials show clinical benefit
associated with the alteration-drug match, with similar benefit observed across
tumor types
II II-A Retrospective studies show that patients with the specific alteration in a specific tumor Drug match is associated
type experience clinically meaningful benefit with matched drug compared to with antitumor activity,
alteration-negative patients but magnitude of
benefit is unknown
II-B Prospective clinical trial(s) show that the alteration-drug match in a specific tumor type
results in increased responsiveness when treated with a matched drug, however, no
data currently available on survival endpoints
III III-A Clinical benefit demonstrated in patients with the specific alteration (as ESCAT I and II Drug match suspected to
above) but in a different tumor type. Limited/ absence of clinical evidence available improve outcome based
for the patient-specific cancer type or broadly across cancer types on clinical trial data in
III-B An alteration that has a similar predicted functional impact as an already studied other tumor type(s) or
ESCAT I abnormality in the same gene or pathway, but does not have associated with similar molecular
supportive clinical data alteration
IV IV-A Evidence that the alteration or a functionally similar alteration influences drug Preclinical evidence of
sensitivity in preclinical in vitro or in vivo models actionability
IV-B Actionability predicted in silico
V Prospective studies show that targeted therapy is associated with objective responses, Drug match is associated
but this does not lead to improved outcome with objective response,
but without clinically
meaningful benefit
X No evidence that the genomic alteration is therapeutically actionable Lack of evidence for
actionability

Int. J. Cancer: 145, 2996–3010 (2019) © 2019 UICC

 10970215, 2019, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32358, Wiley Online Library on [26/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3000 Variant classification in precision oncology

m1 and m2) as well as preclinical (NCT evidence levels m3 and
m4) evidence. As responses to individual drugs based on predic-
tive biomarkers can be highly dependent on tissue context,20
further stratification is made based on tumor type. NCT evi-
dence level m1 encompasses all recommendations made for the
same tumor type, while evidence level m2 includes evidence for
the respective biomarker obtained in other tumor entities.
Within the two clinical evidence levels, three subgroups are
defined: biomarkers for which clinical efficacy of the
corresponding drug was demonstrated in a molecularly stratified
cohort of a prospective study or a meta-analysis (m1A/m2A),
biomarkers for which clinical effectiveness was shown in a
molecularly stratified retrospective cohort or a case–control
study (m1B/m2B), and case studies or single unusual responders
that suggest clinical effectiveness of the drug (m1C/m2C). NCT
Cancer Genetics and Epigenetics
Comparison of the Classification Systems
Overview on similarities and differences of the
classifications
Despite the differences between these classifications, significant
similarities exist (Fig. 1). The first and most obvious feature that
groups a set of genetic alterations, is evidence from clinical trials
showing improved outcome for a specific drug in a defined
tumor entity (i.e., JCR Tier 1/ESCAT I/NCT level m1).4,9,10 The
second group comprises genetic variants with potential clinical
benefit, derived from clinical trials and case studies in other
tumor entities (JCR Tier 2 C.1/ESCAT III-A/NCT level m2).
However, emerging genetic alterations and their corresponding
treatments are ranked differently within each classification based
on the defined level of evidence including preclinical analyses,
case reports and retrospective analyses. A third group includes

level m3 encompasses preclinical data (e.g., cell culture and ani-
mal models). NCT level m4 incorporates alterations for which a
theoretical biological rationale (including in silico prediction)
exists but clinical or preclinical data are lacking. Additionally,
the NCT classification includes suffixes for resistance and
approval by the FDA and EMA, respectively (Table 3). While
originally developed for comprehensive sequencing approaches,
the use of this classification is currently being expanded to the
interpretation of panel sequencing results.
genetic alterations, which might have a clinical benefit as
informed by preclinical models and functional rationale (JCR Tier
2 D/ESCAT III-B, -IV/NCT level m3, m4). The last group com-
prises genetic alterations with lacking or unknown association
with cancer including variants of unknown (functional/clinical)
significance (VUS) as well as likely benign variants (JCR Tier
3, 4/ESCAT X). This category does not exist within the NCT clas-
sification, as such alterations are routinely found in broad geno-
mic analyses and are not actionable. The NCT classification does

Table 3. Classification scheme developed for the MTBs of the National Center for Tumor Diseases (NCT) and within the German Cancer
Consortium (DKTK)
LoE Explanation
m1A Same entity Predictive value of the biomarker or clinical effectiveness of the corresponding drug in
a molecularly stratified cohort was demonstrated in a prospective study or a
meta-analysis in the same tumor type.
m1B Predictive value of the biomarker or clinical effectiveness of the drug in a molecularly
stratified cohort was demonstrated in a retrospective cohort or case–control study
in the same tumor type.
m1C Case study or single unusual responder indicates the biomarker is associated with
response to the drug in the same tumor type.
m2A Different entity Predictive value of the biomarker or clinical effectiveness of the corresponding drug in
a molecularly stratified cohort was demonstrated in a prospective study or a
meta-analysis in a different tumor type.
m2B Predictive value of the biomarker or clinical effectiveness of the drug in a molecularly
stratified cohort was demonstrated in a retrospective cohort or case–control study
in a different tumor type.
m2C Case study or single unusual responder indicates the biomarker is associated with
response to the drug in a different tumor type.
m3 Preclinical Preclinical data (in vitro, in vivo models or functional genomics studies) demonstrate
that the biomarker predicts response to a specific drug treatment, supported by
scientific rationale.
m4 Biological rationale Biological rationale that associates the biomarker with altered signaling pathway
activity or drug sensitivity without direct clinical or preclinical evidence for response
to the drug.
Additional modifiers is—several in situ data and studies on patient material (e.g., IHC, FISH) support the biomarker and the level of evidence.
iv—in vitro data
Z—Drug is approved for use with the specific biomarker [Z = EMA approval, Z(FDA) = FDA approval]
R—Biomarker predicts resistance to the drug.

Int. J. Cancer: 145, 2996–3010 (2019) © 2019 UICC


Leichsenring et al.
Figure 1. Comparison of the JCR, ESCAT and NCT variant classifications systems. The individual evidence levels have been color-coded from
those with direct clinical implication (dark blue) to experimental and hypothetical biomarkers (light blue) and those with no clinical
implication (gray). [Color figure can be viewed at wileyonlinelibrary.com]
not explicitly include germline variant evaluation, which is
reported separately and based on the ACMG/AMP consensus
guidelines21 as discussed below, but reports actionable targets
(e.g., pathogenic BRCA1/2 mutations) like ESCAT. Within the
ESCAT and NCT systems, in silico predictions are classified as
ESCAT IV-B and NCT level m4. No category for alterations lac-
king clinical or preclinical data has been considered within the
JCR classification. Tables 4 and 5 exemplify the clinical applica-
tion for data generated by panel-based NGS as well as whole-
genome/exome sequencing.
Comparison of core themes across classification systems
The following section addresses specific categorical topics
across classifications and highlights their impact on clinical
variant interpretation.
Impact of study design on the classification of genetic
alterations with approved targeted therapies. The basis of
oncological management is the use of approved (targeted)
therapies that have shown clinical utility in prospective trials
most often in conjunction with associated and defined bio-
markers (companion diagnostics). Hence, genetic variants
proven to be biomarkers for approved targeted therapies are
found within the highest ranked category, albeit with differ-
ences between the JCR, ESCAT and NCT systems.
Int. J. Cancer: 145, 2996–3010 (2019) © 2019 UICC
10970215, 2019, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32358, Wiley Online Library on [26/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3001
Cancer Genetics and Epigenetics
JCR Tier 1 A.1 comprises drugs approved by the FDA,
while ESCAT I-A includes alteration-drug matches in specific
tumor types with clinically meaningful improvement of a sur-
vival endpoint in a prospective, randomized clinical trial and
NCT level m1 includes predictive biomarkers with clinical
efficacy in biomarker-stratified prospective cohorts or meta-
analyses and is additionally labeled when the drug has been
approved [NCT level m1 (Z)]. However, ESCAT differentiates
between randomized (ESCAT I-A) and nonrandomized
(ESCAT I-B) clinical trials. Taking NSCLC as an example,
tyrosine-kinase inhibitor (TKI)-responsive alterations in
EGFR+ and ALK+ tumors would be classified as ESCAT I-A,
while TKI-responsive alterations in ROS1+ tumors (high
response rate [RR] in single-arm study)41 would be classified
as ESCAT I-B but JCR Tier 1 A.1 and NCT level m1A. Ana-
lyses of EGFR, ALK and ROS1 are mandatory according to
the updated molecular testing guidelines for NSCLC by the
CAP, International Association for the Study of Lung Cancer
(IASLC) and the AMP42 as well as by the ESMO,43 reflecting
the high-level classification.
As another example, an activating BRAF V600E muta-
tion in NSCLC would be classified the same way, due to
FDA/EMA approval of treatment with dabrafenib/trametinib
(Tier 1 A.1/NCT level m1A(Z)) but ESCAT I-B due to non-
randomization during clinical trials.44 Notably, the use of clin-
ical trial randomization for the evaluation of predictive
 10970215, 2019, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32358, Wiley Online Library on [26/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3002 Variant classification in precision oncology

Table 4. Clinical application of somatic variant annotation based on data generated from targeted NGS of a cancer of unknown primary in a
74-year-old male patient
SNVs Allele frequency Function Treatment JCR ESCAT NCT
22,23
ERBB2 62.7% Likely activating Afatinib, two cases Tier 2 D – Level m2C
p.Gly660Asp
ERBB2 61.8% Likely activating Trastuzumab, in vitro24,25 Tier 2 D ESCAT IV-A Level m3
p.Ser310Phe
ATM 21.3% Deleterious Olaparib, one case,26 in vitro27 Tier 2 D – Level m2C
p.Ser2259fs*13
TP53 28% Deleterious – – – –
p.Arg248Trp
FBXW7 22.7% Likely deleterious Temsirolimus, one case28 Tier 2 D ESCAT IV-A Level m2C
p.Arg385His Everolimus, one case29 Tier 2 D – Level m2C-R
Anti-EGFR antibody, – – Level m2B-R
retrospective analysis30

CNVs Chromosome
Cancer Genetics and Epigenetics

CDKN2B Whole-gene deletion 9p21.3

CDKN2A Whole-gene deletion 9p21.3
FGFR1 Low-level amplification 8p11.23
ERBB2 Low-level amplification 17q12
CDK12 Low-level amplification 17q12

biomarkers is a cause of discrepancies between the classifica-
tions and does not necessarily reflect quality of evidence,
because in some cases with existing strong evidence for the
superiority of a compound, randomization against an alterna-
tive in a prospective clinical trial may not be considered
ethical. Moreover, future regulatory approvals of targeted
therapies may not be driven primarily by randomized studies,
leading to lower evidence level assessment of biomarker–drug
combinations not always adequately reflecting their clinical
value or benefit.
Besides prospective randomized/nonrandomized single-
entity trials, basket trials recently led to FDA approvals, as in
the case of larotrectinib, a tropomyosin receptor kinase (TRK)
inhibitor approved for solid tumors based on a prospective,
nonrandomized basket trial including 55 patients with NTRK
fusions and a high RR as well as improved PFS in a diverse
range of tumors types including salivary-gland tumors, soft
tissue sarcomas, thyroid tumors, colon tumors and others,
some of which (colon cancer, lung cancer, melanoma, GIST,
cholangiocarcinoma, appendix carcinoma, breast cancer and
pancreas cancer) were represented by single cases only.45
Based on the classifications described above, the underlying
evidence would be classified as high as JCR Tier 1 A.1, ESCAT
I-B/I-C or NCT level m1A(Z) or as low as JCR Tier 2 C.1,
ESCAT I-C or NCT level m1C/m2C depending on the
tumor type.
Overall, it has to be considered that evidence levels used in
evidence-based medical practice were primarily designed to
rank the strength of a clinical trial or a research study in
molecularly unstratified cohorts in order to guide rational
clinical decisions. Such an evaluation might not be suited
to assign clinical evidence to molecular alteration-drug combi-
nations. Many molecular alterations with potential clinical
benefit might not reach higher evidence levels due to their low
incidence in a specific tumor type and would consequently
always be inferior to any clinical evidence with low clinical
benefit in an unselected cohort.
Targetable alterations lacking approval. BRAF V600-mutant
NSCLC treated with a BRAF inhibitor, such as vemurafenib or
dabrafenib, or with dabrafenib in combination with trametinib,
has also shown increased PFS in prospective nonrandomized
trials,44,46,47 therefore these mutations would be classified as
JCR Tier 1 B/ESCAT I-B/NCT level m1A. Furthermore, retro-
spective studies (molecular single entity basket trial)48 suggest
that some patients with non-V600E BRAF mutations may ben-
efit from BRAF inhibition. Despite the retrospective nature of
the study, non-V600E BRAF mutations would be classified as
JCR Tier 2/ESCAT II-A/NCT level m1B. In melanoma, all
BRAF V600-mutations are approved for vemurafenib therapy
in Europe (despite lacking trial evidence for non-V600E BRAF
mutations), while in the U.S., only BRAF V600E mutations
are approved for treatment with vemurafenib by the FDA.
However, for BRAF V600 (including K/E) mutant melanoma,
combinatorial BRAF/MEK inhibitor regimens (encorafenib/
binimetinib,49 vemurafenib/cobimetinib50 and dabrafenib/
trametinib51) have been approved by both regulatory agencies,
FDA and EMA. These examples highlight that country-specific
difference in the approval statuses of different biomarker–drug
combinations can lead to the assignment of different classifica-
tion tiers and should be considered when reading and inter-
preting diagnostic reports and MTB decisions.

Int. J. Cancer: 145, 2996–3010 (2019) © 2019 UICC


Leichsenring et al.
Table 5. Clinical application of somatic variant annotation based on data generated from whole-genome/exome sequencing of a cancer of
unknown primary in a male 59-year-old patient
SNVs Allele frequency Function
ATR 34% Likely deleterious
p.Asp2027Tyr
BRCA2 54% Likely deleterious
p.Asp1807Phe
STK11 61% Likely deleterious
p.Glu14Gln
CNVs Chromosome
IGF1R Low-level amplification 15q26.3
Complex biomarkers
Tumor mutational burden High
Mutational signatures AC4, AC24, AC2, AC13
Homologous repair deficiency High
Targetable alterations with clinical evidence in other tumor
entities. In all three classifications, cross-entity basket trials
and translation of experiences from different entities are con-
sidered (JCR Tier 2 C.1/ESCAT I-C, III-A/NCT level m2).
ESCAT specifically lists a cross-entity evidence class ESCAT
I-C for clinical basket trials that have demonstrated clinically
meaningful benefit for the target–drug pair across different
tumor entities. TRK fusions as targets of specific TKI and mis-
match repair deficiency (dMMR) for PD-1 blockade are
referred to as examples for this category. It remains unclear,
which individual biomarkers fall into this category, as particu-
larly for targets with very low prevalence, full evaluation of
entity-associated responses can be limited and hard to inter-
pret. In scenarios with more prevalent genetic aberrations,
recent data collectively show that clinical benefit across differ-
ent entities can vary.20 As exemplified by BRAF V600E-
mutated NSCLC (JCR Tier 1 A.1/ESCAT I-B/NCT level m1A)
and melanoma (JCR Tier 1 A.1/ESCAT I-A/NCT level m1A),
BRAF inhibition would be indicated with response rates of
4220 and 48%,52 respectively. At the same time, however,
response rates in colorectal cancer (CRC) are less than
5%.20,53 Therefore, histopathological evaluation of the tumor
tissue and contextual interpretation of molecular results is
paramount and genetic data can only be viewed in the histo-
pathological and clinical context of the disease. In order to
evaluate specific entity–biomarker–drug combinations in full,
we should consider clinical experiences in other entities for
treatment in structured clinical trials or well-documented
observational studies until sufficient data of individual
responses in the entity of question is available.
Alterations associated with resistance and concurrent
alterations. Alterations associated with resistance to specific
therapies are classified in the JCR and by the NCT
Int. J. Cancer: 145, 2996–3010 (2019) © 2019 UICC
10970215, 2019, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32358, Wiley Online Library on [26/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3003
Treatment JCR ESCAT NCT
– – – –
Olaparib31,32 Tier 3 ESCAT IV A Level m2A
Immune checkpoint-inhibitor, Tier 3 ESCAT IV A Level m4R
retrospective analysis33,34 Tier 3 ESCAT IV A level m4
Metformin, in vitro35
Immune checkpoint-inhibitor, Tier 2 C.1 ESCAT III-A Level m2A
prospective clinical trial36–39
Cancer Genetics and Epigenetics
– – – –
40
Niraparib, prospective clinical trial – – Level m4
classifications but are not explicitly mentioned in ESCAT. A
well-known example includes oncogenic RAS mutations in
colorectal cancer, which indicate resistance to anti-EGFR anti-
body therapy.54 Another example is RAS mutations or amplifi-
cations of BRAF in melanoma, which lead to vemurafenib
therapy resistance.55 Besides the JCR and NCT levels, several
other classifications include biomarkers of resistance in their
categorization (OncoKB, CIViC).14,56 An increasing number of
mechanisms of primary and secondary resistance, which may
also be tumor-type specific, will lead to an increasing complex-
ity of variant annotation, which might in turn complicate test-
ing for targeted therapies in molecularly distinct subsets of
tumors. Additionally, detection of specific sensitizing concur-
rent mutations can also influence interpretation of initial vari-
ant classification. In case of EGFR L858R-mutated NSCLC,
simultaneous occurrence of an EGFR T790M mutation indi-
cates resistance to first- and second-generation EGFR-TKI, but
not to osimertinib,57,58 where this constellation would indicate
drug sensitivity and is already regularly assessed in routine
molecular diagnostics. Additionally, secondary mutations
within the RAS signaling pathway result in resistance to EGFR-
TKI55 and have to be considered in the overall interpretation of
results. These fundamental examples of a complex decision-
making process also serve as a cautionary note in light of the
rising number of automated decision-support tools for MTBs.
Targetable alterations based on preclinical data or biological
rationale. Categories JCR Tier 2 C.2, D/ESCAT III-B,
-IV/NCT level m3–4 group alterations suggest benefit from
targeted therapies based on functional or in silico data. In
addition, ESCAT IV differentiates between IV-A (evidence
based on in vitro or in vivo models) and IV-B (in silico evi-
dence). The JCR recommends against the use of prediction
algorithms as sole evidence for variant classification or clinical
 Cancer Genetics and Epigenetics

3004

Figure 2. Unifying classification concept. This concept may serve as a blueprint for harmonization of variant classifications and as a (pocket) guide for clinical decision making. Based
on implementation in patient treatment, evidence levels are color-coded: established predictive biomarkers/drug targets are highlighted in green. Biomarkers with potential clinical
implication are highlighted in different shades of orange based on the strengths of evidence, while variants without any major current clinical implication are highlighted in red. [Color
figure can be viewed at wileyonlinelibrary.com]

Int. J. Cancer: 145, 2996–3010 (2019) © 2019 UICC

Variant classification in precision oncology

10970215, 2019, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32358, Wiley Online Library on [26/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

Leichsenring et al.
decision making.10 In our opinion, information obtained by
preclinical evidence is relevant for patients with advanced,
treatment-refractory disease eligible for clinical trials based on
molecular alterations or in well-documented observational
studies. These alterations may not have an established benefit
in clinical studies, though they may lead to valid therapeutic
reasoning considering the individual landscape of possible
driver mutations in the molecular context of the disease.
Assessment of case reports. Molecularly characterized case
reports and collections thereof (e.g., found in GENIE and
CIViC)56 have become increasingly valuable for the evaluation
of clinical benefit in a precision oncology setting. Such case
reports have been an integral part of the sequencing efforts
within the MASTER program at NCT Heidelberg and are
taken into consideration through NCT levels m1C and m2C.
Within the JCR, case reports are represented in Tier 2 if mul-
tiple cases support the same conclusion but no consensus was
found, while ESCAT levels do not specifically address the pos-
sibility to obtain valid evidence from individual cases.
Although this evidence may initiate further use of targeted
therapies in specific entity–biomarker–drug combinations and
may aid the development of prospective basket trials,59 obser-
vation of clinical impact in single patients represent an excep-
tional event associated with additional uncontrolled factors
that can potentially interfere with response and outcome
parameters. Hence, evidence from case studies (i.e. n = 1),
requires careful individual assessment and should not be gen-
eralized prematurely.
Alterations without clinical actionability. Tiers 3 and 4 of the
JCR represent a collection of mutations without current clini-
cal implication (mutations of unknown clinical significance
and mutations without cancer association), while ESCAT X
groups mutations lacking clinical actionability, independent of
their association with cancer. Listing benign or currently clini-
cally irrelevant alterations is beyond the scope of a routine
reporting of genetic profiles and certainly neither feasible nor
meaningful in a scenario where comprehensive genomic
assessment by WES/WGS is applied.
A Unifying Concept for Clinical Variant Classification
Based on the analysis shown above, we conceptualized our
findings to devise an integrated, unifying classification
approach that can serve as a blueprint for harmonization of
the classification schemes that were proposed until now
(Fig. 2). This conceptual approach also integrates the previous
classification systems summarized in Supporting Information
Appendix S1 and can be used as a (pocket) guide to somatic
variant classification for daily clinical decision making. This
concept acknowledges that each classification system has a
subjective component that is either supporting a more conser-
vative or a more progressive therapeutic approach. Of course,
variants and biomarkers associated with “standard of care”
Int. J. Cancer: 145, 2996–3010 (2019) © 2019 UICC
10970215, 2019, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32358, Wiley Online Library on [26/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3005
therapies are ranked first within both approaches, while data
from current clinical trials and convincing evidence from dif-
ferent entities as well as case reports might be more applicable
in a progressive setting at experienced centers. Finally, exper-
imental approaches based on clinical evidence within the
same or a different tumor entity as well as preclinical
(in vitro/in silico) data should only be considered in the con-
text of clinical programs at specialized centers independent
of the therapeutic approach. Together with potential off-label
therapies, these options have to be discussed in depth in
interdisciplinary MTBs.
Evolving Topics in Clinical Variant Classification
With the continuous and accelerated development of new
drugs and increasingly comprehensive molecular profiles,
MTBs will be facing a complex task of analyzing and inter-
Cancer Genetics and Epigenetics
preting not only a single molecular alteration but whole sets
of biomarkers, including continuous variables such as tumor
mutational burden (TMB), in the context of individual
patients’ tumor biology, histology and medical history. There-
fore, a number of additional factors, which are briefly
highlighted below, will play an increasingly important role in
the near future.
Biological themes recurring across tumor types
Some targetable genetic alterations have been identified in a
diverse number of tumor types. However, many of these alter-
ations are too rare for prospective, randomized trials and thus
histology-agnostic basket trials were initiated to assess their
clinical impact. Larotrectinib, for example, demonstrated a RR
of 75% in solid tumors with TRK fusions, resulting in FDA
approval. For cases with n = 1, the evidence level could be
considered equivalent to case reports, and the clinical rele-
vance of different fusion partners with regard to clinical effi-
cacy60 cannot be evaluated based on individual patients.
Furthermore, within these molecular basket trials, control
groups are absent, which influences outcome interpretation.
Fortunately, additional data is generated in ongoing trials61
and the resulting data will have to be reevaluated.
Another example is the response to immunotherapy with
nivolumab in different mismatch repair deficient tumor
entities.62,63 Trials including germline as well as sporadic
MSI-H colorectal cancer with nivolumab resulted in objec-
tive RR of 31.1% (n = 74; 36% of patients had Lynch-
Syndrome),62 40% (n = 10; 80% had Lynch-Syndrome)64
and 52% (n = 40; 50% had Lynch-Syndrome),65 suggesting
differences between germline and somatic mismatch repair
deficiency.
Multifactorial biomarkers: complexity is increasing
So far, most variant classifications are primarily designed to
interpret and classify SNVs, indels, gene fusions and CNVs.
However, currently, a multitude of other relevant parameters,
such as PD-1/PD-L1 expression, homologous recombination

3006
deficiency (HRD) and TMB66,67 have to be evaluated in
MTBs. Not to mention global genetic (e.g., BRCAness),68,69
gene expression and methylome profiles,70 which are currently
not captured by most precision oncology approaches but will
play an increasing role in the future.
ESCAT refers to “molecular targets” instead of genetic
alterations and therefore the PD-1/PD-L1 expression would
be classified as ESCAT I-A (at least in NSCLC).71 The JCR
focuses on somatic variants only, including SNVs, indels, gene
fusions and CNVs. NCT levels focus in a more general
approach on biomarker-based recommendations, thus includ-
ing all measurable molecular properties, including immuno-
histochemistry, in situ hybridization, but also other possible
technologies, such as methylation studies or proteomic assays.
TMB, a continuous variable of probabilistic nature, would
Cancer Genetics and Epigenetics
currently classify at best as JCR Tier 1 B/ESCAT I-C/NCT
level m1B.72 In addition, response to immune checkpoint
inhibitors has been reported to be influenced by a plethora of
molecular alterations, including STK11 or BRCA1/2 muta-
tions in association with other markers in a number of tumor
entities.73,74
Inclusion of the above alterations and biomarkers drasti-
cally increases the complexity of an all-encompassing classi-
fication and requires ranking of each therapeutic option,
taking into consideration the patient’s health status and
preference, previous lines of therapy as well as therapy
availability.
Biological equivalence of alterations and exchangeability of
therapeutic regimens
Assessing therapies such as osimertinib in NSCLC patients
with somatic EGFR T790M mutations is relatively straight
forward,57,58 but with the forthcoming implementation of
large sequencing panels, WGS/WES and transcriptome
sequencing, MTBs will frequently be challenged to assess rare
alterations, which are not strictly identical to the published
variants, or to consider therapeutic regimens which differ
from the ones used in a clinical trial. For example, the sensi-
tizing effect of BRAF V600 hotspot SNVs to selective BRAF
inhibitors is observed in several entities, and similar effects
have been assigned preclinically and clinically to certain alter-
ations close by.75,76 However, it would be incorrect to apply
these findings to other unknown V600 or K601 SNVs, as clin-
ical evidence is lacking. Furthermore, care has to be taken
before generalizing the effect of RAF/MEK inhibition in indi-
vidual responders with BRAF fusions.77
Until now, many (potential) biomarkers remain incom-
pletely studied, hindering the assignment of evidence levels,
such as the interpretation of unknown mutations in kinase
domains, mRNA expression corresponding to predictive pro-
tein expressions, assessment of gene amplifications and their
cutoffs that define “amplified” = druggable, transferability of
predictive markers for PD1 inhibitors to PD-L1 inhibitors or
CTLA4 inhibitors as well as the generalization of the
10970215, 2019, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32358, Wiley Online Library on [26/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Variant classification in precision oncology
predictive value of FGFR1 amplifications seen in BGJ398 tri-
als78 to other pan-FGFR inhibitors or even to other multi-
kinase inhibitors with FGFR1-affinity.
ESCAT groups alterations with similar predicted functional
impact in class ESCAT III-B but do not specify them further,
leaving space for interpretation. The NCT classification does
not make assumptions about biological equivalency and
assigns these unknown variants to NCT level m4, leaving the
discussions of the biological implications to the MTB in
appropriate cases. For the JCR, such alterations might be
summed up under JCR Tier 3. As of today, the issue on how
to handle diagnostic and therapeutic discrepancies with regard
to their biological equivalency in assigning evidence levels is
not ultimately solved. However, it stresses the necessity to dis-
cuss patients with extensive genomic characterization in inter-
disciplinary MTBs to appropriately address these individual
constellations.
Druggable targets associated with defects in the DNA repair
machinery
Identification and classification of BRCA1/2 variants and
others are essential for risk assessment with respect to inher-
itable disease. Such classification is performed according to
the 2015 classification consensus by the ACMG-AMP21,79
taking into consideration population data, computational
evidence, functional data, segregation data, de novo data,
allelic data and “other data”. This classification, which is
based on a multiparametric probability model and originally
meant to classify a relative risk of developing cancer for
healthy carriers with a certain genetic make-up in the
germline, has been repurposed for the evaluation of drug
response associated with deleterious mutations in DNA
repair genes. For example, likely pathogenic and pathogenic
variants (Plon 4/5 variants)80 in BRCA1/2 in breast81 and
ovarian cancer82,83 lead to homologous repair deficiency
(HRD) sensitizing the tumor to Poly (ADP-ribose) polymer-
ase (PARP) inhibition, a concept termed synthetic lethality.84
While the majority of these variants are germline events, the
classification system is also used to predict therapy response
for somatically acquired mutations.85 Such loss-of-function
aberrations are classified as JCR Tier 1 A/ESCAT I-A/NCT
level m1A in ovarian and breast cancer based on FDA
approval and a prospective, molecularly stratified, random-
ized trial.81 It should be noted, however, that the clinical rel-
evance and predictive values of deleterious variants in the
majority of the more than 100 core genes encoding DNA
repair pathways are currently only poorly understood.86,87 As
a proof-of-concept, emerging data show that for example in
prostate cancer loss-of-function alterations beyond BRCA2
can be clinically relevant.26,88 As DNA repair operates in a
tumor agnostic manner, it is reasonable to assume that
genetic lesions beyond BRCA1/2—whether germline or
somatic—can in principle be exploited for synthetic lethality
concepts across many cancer types.89
Int. J. Cancer: 145, 2996–3010 (2019) © 2019 UICC

Leichsenring et al.
Diagnostic biomarkers, pharmacogenomics and risk
assessment
Beyond predictive genetic lesions, only the JCR10 explicitly
acknowledges diagnostic (e.g., the PML-RARA fusion in acute
promyelocytic leukemia) and prognostic (e.g., NPM1 muta-
tions in acute myeloid leukemia) biomarkers and includes
them into the classification approach. However, the use of sin-
gle genetic events as clear-cut diagnostic biomarkers should be
viewed with caution. Even gene fusions involving, for exam-
ple, ALK or EWSR1 can occur in tumor types other than clas-
sic index cancer types NSCLC and Ewing sarcoma. Recently
published examples are EWSR1 translocations in pancreatic
neuroendocrine tumors90 or ALK fusions in colorectal can-
cer.91 While the overall prevalence might be lower compared
to the classic tumor types where the initial discoveries were
made,92,93 one should be aware that the concept of a single
pathognomonic genetic event has limitations, particularly
when it comes to the interpretation of individual molecular
data. Importantly, the role of pharmacogenomics, that is,
determining drug response based on sequencing results, is
becoming more important with accumulating evidence
(e.g., in colorectal cancer)94 and may be more broadly incor-
porated in oncological decision making in the near future. A
third emerging topic is specific germline aberrations that
require genetic counseling. With more comprehensive molec-
ular profiling entering routine diagnostics, this field will
strongly expand in the next couple of years: our current
knowledge of cancer risk-associated variants is confined to a
relatively small set of genes compared to, for instance, the
number of genes that are implicated in the various DNA-
repair pathways.95,96
Predictive molecular biomarkers for conventional therapies
As exemplified for BRCA1/2 variants in triple-negative breast
cancer, genetic aberrations can predict efficacy of conventional
therapeutic approaches such as platinum-based chemother-
apy.97 The authors assume that the rapid progress in our
understanding of tumor biology will lead to the discovery of
many molecular profiles that can predict response or resis-
tance to specific chemotherapy and radiotherapy regimens,
which will need to be considered in future adaptations of the
variant classification systems.
Summary and Outlook
The growing amount of molecular data correlates with
increasing diagnostic and therapeutic complexity.98 All variant
classifications analyzed herein provide a framework for the
clinical assessment of the detected genetic alterations and
facilitate prioritization of the corresponding targeted therapies,
while not absolutely ranking them in terms of overall patient
management. All three classifications are relatively easily
applicable for most well-characterized alterations, which are
already considered standard of care.
Int. J. Cancer: 145, 2996–3010 (2019) © 2019 UICC
10970215, 2019, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32358, Wiley Online Library on [26/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3007
For the remaining, newly discovered, rare or individual
genetic alterations, each classification has its advantages and
drawbacks and not all alterations are classified equally
throughout the different schemes. Differences depend on the
specific evidence levels defining the categories within each
scheme, which are influenced by clinical and biological ratio-
nale, but also by different socioeconomic circumstances
including various regulatory approvals of targeted compounds
in different health care systems.
While the JCR defines a range of evidence levels including
markers of resistance, variants of unknown significance and
benign variants, it is tailored to the U.S. health care system,
focusing on FDA approval status, which is not necessarily
applicable outside of the U.S.
ESCAT, like the JCR, covers the range from standard of
care to alterations with no evidence of therapeutic applicabil-
Cancer Genetics and Epigenetics
ity but does not take into consideration markers of resistance.
Furthermore, it relies on its own standards for determination
of a clinical benefit, which appears to be less subjective than a
broad definition of “well-powered studies” or “expert opin-
ion.” However, this adds another layer of complexity and
might sometimes be the source of difficulties or discrepancies
in diagnostic assessment. Consideration of the randomization
status of clinical trials can also be a cause of inconsistency
with the other systems and the in silico predictors mentioned
in this proposal are rarely useful.
The NCT MTB classification addresses the complexity of
evaluating predictive molecular biomarkers by focusing on two
main criteria, tumor entity and clinical relevance. By avoiding
parameters beside primary scientific data (such as the design of
prospective clinical trials, the secondary evaluation of clinical
benefit and the drug approval status) it is less complex than the
other classification schemes, thus easy to apply and communi-
cate. In our experience, the classification provides graded and
evaluable therapeutic recommendations for advanced cancer
patients and is well suited for cross-institutional use. Further-
more, similar to the JCR, it specifically allows incorporation of
case studies, which, in rare entities, might be and remain the
only level of evidence available.
A unifying integrative classification concept, as presented
by us, can support daily clinical decision making and con-
siders a more conservative and more progressive treatment
approach.
Conclusion
This head-to-head comparison of three different classifica-
tions is meant to support the clinical interpretation of com-
mon and rare druggable cancer biomarkers. As we have
shown, MTB reports must be supported by rational,
evidence-based decisions, which are reproducible and readily
transferrable between institutions. This is a prerequisite for
the implementation of interinstitutional precision oncology
programs and any future standardization in the field. Such a
common standard, for which we are providing a conceptual
 10970215, 2019, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32358, Wiley Online Library on [26/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3008 Variant classification in precision oncology

approach, needs to address the complexity and uniqueness of
molecular alterations within individual tumors while being
flexible enough to allow application in a routine clinical
setting—even for rare tumors and rare genetic events. By
focusing on implementation of molecularly targeted thera-
pies in a comprehensively characterized patient cohort in
well documented clinical trials or observational studies, we
might be able to “bootstrap” our knowledge about the
predictive value of biomarkers and the clinical efficacy of
associated drugs.
Acknowledgements
All members of the Center for Molecular Pathology at the University Hos-
pital Heidelberg, DKFZ-HIPO and the members of the Molecular Tumor
Board at NCT Heidelberg for generating the genetic data and continuous
discussions and interpretation. This work was supported by grant H021
from DKFZ-HIPO to C.H., H.G., A.S. and S.F.

References
1. Roychowdhury S, Iyer MK, Robinson DR, et al.
Personalized oncology through integrative high-
throughput sequencing: a pilot study. Sci Transl
Med 2011;3:111ra21.
2. Burkard ME, Deming DA, Parsons BM, et al.
Implementation and clinical utility of an inte-
Cancer Genetics and Epigenetics
13. Meric-Bernstam F, Johnson A, Holla V,
et al. A decision support framework
for Genomically informed investigational
cancer therapy. J Natl Cancer Inst 2015;107:
djv098.
14. Chakravarty D, Gao J, Phillips S, et al. OncoKB: a
PARP inhibitor olaparib. Transl Oncol 2017;10:
190–6.
28. Villaruz LC, Socinski MA. Temsirolimus
therapy in a patient with lung adenocarcinoma
harboring an FBXW7 mutation. Lung Cancer
2014;83:300–1.

grated academic-community regional molecular
tumor board. JCO Precis Oncol 2017;1:1–10.
3. Dalton WB, Forde PM, Kang H, et al. Personalized
medicine in the oncology clinic: implementation
and outcomes of the Johns Hopkins molecular
tumor board. JCO Precis Oncol 2017;1:1–19.
4. Horak P, Klink B, Heining C, et al. Precision
oncology based on omics data: the NCT Heidel-
berg experience. Int J Cancer 2017;141:877–86.
5. Zehir A, Benayed R, Shah RH, et al. Mutational
landscape of metastatic cancer revealed from pro-
spective clinical sequencing of 10,000 patients.
Nat Med 2017;23:703–13.
6. Hoefflin R, Geißler A-L, Fritsch R, et al. Personal-
ized clinical decision making through implemen-
tation of a molecular tumor board: a German
single-center experience. JCO Precis Oncol 2018;
2:1–16.
7. Lier A, Penzel R, Heining C, et al. Validating
comprehensive next-generation sequencing results
for precision oncology: the NCT/DKTK molecu-
larly aided stratification for tumor eradication
research experience. JCO Precis Oncol
2018;2:1–13.
8. Volckmar AL, Leichsenring J, Kirchner M, et al.
Combined targeted DNA and RNA sequencing of
advanced NSCLC in routine molecular diagnos-
tics: analysis of the first 3,000 Heidelberg cases.
Int J Cancer 2019; https://doi.org/10.1002/
ijc.32133. [Epub ahead of print]
9. Mateo J, Chakravarty D, Dienstmann R, et al. A
framework to rank genomic alterations as
targets for cancer precision medicine: the
ESMO scale for clinical actionability of
molecular targets (ESCAT). Ann Oncol 2018;29:
1895–902.
10. Li MM, Datto M, Duncavage EJ, et al. Standards
and guidelines for the interpretation and reporting
of sequence variants in cancer: a joint consensus
recommendation of the Association for Molecular
Pathology, American Society of Clinical Oncology,
and College of American Pathologists. J Mol
Diagn 2017;19:4–23.
11. Andre F, Mardis E, Salm M, et al. Prioritizing tar-
gets for precision cancer medicine. Ann Oncol
2014;25:2295–303.
12. Van Allen EM, Wagle N, Stojanov P, et al.
Whole-exome sequencing and clinical interpreta-
tion of formalin-fixed, paraffin-embedded tumor
samples to guide precision cancer medicine. Nat
Med 2014;20:682.
precision oncology Knowledge Base. JCO Precis
Oncol 2017;1:1–16.
15. Starlinger J, Pallarz S, Seva J, et al. Variant infor-
mation systems for precision oncology. BMC Med
Inform Decis Mak 2018;18:107.
16. Damodaran S, Miya J, Kautto E, et al. Cancer
driver log (CanDL): catalog of potentially action-
able cancer mutations. J Mol Diagn 2015;17:
554–9.
17. Ritter DI, Roychowdhury S, Roy A, et al. Somatic
cancer variant curation and harmonization
through consensus minimum variant level data.
Genome Med 2016;8:117.
18. Rieke DT, Lamping M, Schuh M, et al. Compari-
son of treatment recommendations by molecular
tumor boards worldwide. JCO Precis Oncol
2018;1–14.
19. Cherny NI, Dafni U, Bogaerts J, et al. ESMO-
magnitude of clinical benefit scale version 1.1.
Ann Oncol 2017;28:2340–66.
20. Hyman DM, Puzanov I, Subbiah V, et al.
Vemurafenib in multiple nonmelanoma cancers
with BRAF V600 mutations. N Engl J Med 2015;
373:726–36.
21. Richards S, Aziz N, Bale S, et al. Standards and
guidelines for the interpretation of sequence vari-
ants: a joint consensus recommendation of the
American College of Medical Genetics and Geno-
mics and the Association for Molecular Pathology.
Genet Med 2015;17:405–24.
22. Yamamoto H, Toyooka S, Ninomiya T, et al.
Therapeutic potential of Afatinib for cancers with
ERBB2 (HER2) transmembrane domain mutations
G660D and V659E. Oncologist 2018;23:150–4.
23. Ou S-HI, Schrock AB, Bocharov EV, et al. HER2
transmembrane domain (TMD) mutations
(V659/G660) that stabilize homo- and
heterodimerization are rare oncogenic drivers in
lung adenocarcinoma that respond to afatinib.
J Thorac Oncol 2017;12:446–57.
24. Herter-Sprie GS, Greulich H, Wong K-K. Activat-
ing mutations in ERBB2 and their impact on
diagnostics and treatment. Front Oncol 2013;3:86.
25. Kavuri SM, Jain N, Galimi F, et al. HER2 activat-
ing mutations are targets for colorectal cancer
treatment. Cancer Discov 2015;5:832–41.
26. Mateo J, Carreira S, Sandhu S, et al. DNA-repair
defects and Olaparib in metastatic prostate cancer.
N Engl J Med 2015;373:1697–708.
27. Wang C, Jette N, Moussienko D, et al. ATM-
deficient colorectal cancer cells are sensitive to the
29. Jardim DL, Wheler JJ, Hess K, et al. FBXW7
mutations in patients with advanced cancers: clin-
ical and molecular characteristics and outcomes
with mTOR inhibitors. PLoS One 2014;9:e89388.
30. Lupini L, Bassi C, Mlcochova J, et al. Prediction
of response to anti-EGFR antibody-based thera-
pies by multigene sequencing in colorectal cancer
patients. BMC Cancer 2015;15:808.
31. Ledermann J, Harter P, Gourley C, et al. Olaparib
maintenance therapy in patients with platinum-
sensitive relapsed serous ovarian cancer: a
preplanned retrospective analysis of outcomes by
BRCA status in a randomised phase 2 trial. Lancet
Oncol 2014;15:852–61.
32. Pujade-Lauraine E, Ledermann JA, Selle F, et al.
Olaparib tablets as maintenance therapy in
patients with platinum-sensitive, relapsed ovarian
cancer and a BRCA1/2 mutation
(SOLO2/ENGOT-Ov21): a double-blind,
randomised, placebo-controlled, phase 3 trial.
Lancet Oncol 2017;18:1274–84.
33. Rizvi H, Sanchez-Vega F, La K, et al. Molecular
determinants of response to anti-programmed cell
death (PD)-1 and anti-programmed death-ligand
1 (PD-L1) blockade in patients with non-small-cell
lung cancer profiled with targeted next-generation
sequencing. J Clin Oncol 2018;36:633–41.
34. Skoulidis F, Goldberg ME, Greenawalt DM, et al.
STK11/LKB1 mutations and PD-1 inhibitor resis-
tance in KRAS-mutant lung adenocarcinoma.
Cancer Discov 2018;8:822–35.
35. Shackelford DB, Abt E, Gerken L, et al. LKB1
inactivation dictates therapeutic response of non-
small cell lung cancer to the metabolism drug
phenformin. Cancer Cell 2013;23:143–58.
36. Hellmann MD, Callahan MK, Awad MM, et al.
Tumor mutational burden and efficacy of
Nivolumab monotherapy and in combination
with Ipilimumab in small-cell lung cancer. Cancer
Cell 2018;33:853–61 e4.
37. Hellmann MD, Ciuleanu TE, Pluzanski A, et al.
Nivolumab plus ipilimumab in lung cancer with a
high tumor mutational burden. N Engl J Med
2018;378:2093–104.
38. Rizvi NA, Hellmann MD, Snyder A, et al. Cancer
immunology. Mutational landscape determines
sensitivity to PD-1 blockade in non-small cell
lung cancer. Science 2015;348:124–8.
39. Van Allen EM, Miao D, Schilling B, et al. Geno-
mic correlates of response to CTLA-4 blockade in
metastatic melanoma. Science 2015;350:207–11.

Int. J. Cancer: 145, 2996–3010 (2019) © 2019 UICC


Leichsenring et al.
40. Mirza MR, Monk BJ, Herrstedt J, et al. Niraparib
maintenance therapy in platinum-sensitive, recur-
rent ovarian cancer. N Engl J Med 2016;375:
2154–64.
41. Shaw AT, S-HI O, Bang Y-J, et al. Crizotinib in
ROS1-rearranged non–small-cell lung cancer. N
Engl J Med 2014;371:1963–71.
42. Lindeman NI, Cagle PT, Aisner DL, et al.
Updated molecular testing guideline for the selec-
tion of lung cancer patients for treatment with
targeted tyrosine kinase inhibitors. J Mol Diagn
2018;20:129–59.
43. Planchard D, Popat S, Kerr K, et al. Metastatic
non-small cell lung cancer: ESMO clinical practice
guidelines for diagnosis, treatment and follow-up.
Ann Oncol 2018;29:iv192–237.
44. Planchard D, Smit EF, Groen HJM, et al.
Dabrafenib plus trametinib in patients with previ-
ously untreated BRAFV600E-mutant metastatic
non-small-cell lung cancer: an open-label, phase
2 trial. Lancet Oncol 2017;18:1307–16.
45. Drilon A, Laetsch TW, Kummar S, et al. Efficacy
of Larotrectinib in TRK fusion-positive cancers in
adults and children. N Engl J Med 2018;378:
731–9.
46. Planchard D, Kim TM, Mazieres J, et al.
Dabrafenib in patients with BRAFV600E-positive
advanced non-small-cell lung cancer: a single-
arm, multicentre, open-label, phase 2 trial. Lancet
Oncol 2016;17:642–50.
47. Subbiah V, Gervais R, Riely GJ, et al. Efficacy of
vemurafenib in patients (pts) with non-small cell
lung cancer (NSCLC) with BRAFV600 mutation.
J Clin Oncol 2017;35:9074.
48. Gautschi O, Milia J, Cabarrou B, et al. Targeted
therapy for patients with BRAF-mutant lung can-
cer results from the European EURAF cohort.
J Thorac Oncol 2015;10:1451–7.
49. Dummer R, Ascierto PA, Gogas HJ, et al. Overall
survival in patients with BRAF-mutant melanoma
receiving encorafenib plus binimetinib versus
vemurafenib or encorafenib (COLUMBUS): a
multicentre, open-label, randomised, phase 3 trial.
Lancet Oncol 2018;19:1315–27.
50. Ascierto PA, McArthur GA, Dréno B, et al.
Cobimetinib combined with vemurafenib in
advanced BRAFV600-mutant melanoma (coB-
RIM): updated efficacy results from a randomised,
double-blind, phase 3 trial. Lancet Oncol 2016;17:
1248–60.
51. Long GV, Eroglu Z, Infante J, et al. Long-term
outcomes in patients with BRAF V600-mutant
metastatic melanoma who received Dabrafenib
combined with Trametinib. J Clin Oncol 2018;36:
667–73.
52. Chapman PB, Hauschild A, Robert C, et al.
Improved survival with Vemurafenib in mela-
noma with BRAF V600E mutation. N Engl J Med
2011;364:2507–16.
53. Kopetz S, Desai J, Chan E, et al. Phase II pilot
study of Vemurafenib in patients with metastatic
BRAF-mutated colorectal cancer. J Clin Oncol
2015;33:4032–8.
54. Porru M, Pompili L, Caruso C, et al. Targeting
KRAS in metastatic colorectal cancer: current
strategies and emerging opportunities. J Exp Clin
Cancer Res 2018;37:57.
55. Ohashi K, Sequist LV, Arcila ME, et al.
Lung cancers with acquired resistance to EGFR
inhibitors occasionally harbor BRAF gene muta-
tions but lack mutations in KRAS, NRAS, or
Int. J. Cancer: 145, 2996–3010 (2019) © 2019 UICC
MEK1. Proc Natl Acad Sci U S A 2012;109:
E2127–33.
56. Griffith M, Spies NC, Krysiak K, et al. CIViC is a
community knowledgebase for expert
crowdsourcing the clinical interpretation of vari-
ants in cancer. Nat Genet 2017;49:170–4.
57. Mok TS, Wu Y-L, Ahn M-J, et al. Osimertinib or
platinum–pemetrexed in EGFR T790M–positive
lung cancer. N Engl J Med 2017;376:629–40.
58. Yang JC-H, Ahn M-J, Kim D-W, et al.
Osimertinib in pretreated T790M-positive
advanced non–small-cell lung cancer: AURA
study phase II extension component. J Clin Oncol
2017;35:1288–96.
59. Heining C, Horak P, Uhrig S, et al. NRG1 fusions
in KRAS wild-type pancreatic cancer. Cancer
Discov 2018;8:1087–95.
60. Childress MA, Himmelberg SM, Chen H, et al.
ALK fusion partners impact response to ALK
inhibition: differential effects on sensitivity, cellu-
lar phenotypes, and biochemical properties. Mol
Cancer Res 2018;16:1724–36.
61. Lassen UN, Albert CM, Kummar S, et al.
409OLarotrectinib efficacy and safety in TRK
fusion cancer: an expanded clinical dataset show-
ing consistency in an age and tumor agnostic
approach. Ann Oncol 2018;29:mdy279.397.
62. Overman MJ, McDermott R, Leach JL, et al.
Nivolumab in patients with metastatic DNA
mismatch repair-deficient or microsatellite
instability-high colorectal cancer (CheckMate
142): an open-label, multicentre, phase 2 study.
Lancet Oncol 2017;18:1182–91.
63. Mittica G, Ghisoni E, Giannone G, et al. Checkpoint
inhibitors in endometrial cancer: preclinical rationale
and clinical activity. Oncotarget 2017;8:90532–44.
64. Le DT, Uram JN, Wang H, et al. PD-1 blockade
in tumors with mismatch-repair deficiency. N
Engl J Med 2015;372:2509–20.
65. Le DT, Durham JN, Smith KN, et al. Mismatch
repair deficiency predicts response of solid tumors
to PD-1 blockade. Science 2017;357:409–13.
66. Allgäuer M, Budczies J, Christopoulos P, et al.
Implementing tumor mutational burden (TMB)
analysis in routine diagnostics—a primer for
molecular pathologists and clinicians. Transl Lung
Cancer Res 2018;7:703–15.
67. Chan TA, Yarchoan M, Jaffee E, et al. Develop-
ment of tumor mutation burden as an immuno-
therapy biomarker: utility for the oncology clinic.
Ann Oncol 2018;30:44–56.
68. Nik-Zainal S, Davies H, Staaf J, et al. Landscape of
somatic mutations in 560 breast cancer whole-genome
sequences. Nature 2016;534:47–54.
69. Gröschel S, Hübschmann D, Raimondi F, et al.
Defective homologous recombination DNA repair
as therapeutic target in advanced chordoma. Nat
Commun 2019;10:1635.
70. Capper D, Jones DTW, Sill M, et al. DNA
methylation-based classification of central nervous
system tumours. Nature 2018;555:469–74.
71. Reck M, Rodríguez-Abreu D, Robinson AG, et al.
Pembrolizumab versus chemotherapy for PD-
L1-positive non-small-cell lung cancer. N Engl J
Med 2016;375:1823–33.
72. Goodman AM, Kato S, Bazhenova L, et al. Tumor
mutational burden as an independent predictor of
response to immunotherapy in diverse cancers.
Mol Cancer Ther 2017;16:2598–608.
73. Marmarelis M, Bange E, Bagley S, et al. P1.01-64
impact of STK11 co-mutation on outcomes
10970215, 2019, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32358, Wiley Online Library on [26/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3009
following immunotherapy among patients with
TP53 and KRAS mutated stage IV NSCLC.
J Thorac Oncol 2018;13:S486.
74. Strickland KC, Howitt BE, Shukla SA, et al. Asso-
ciation and prognostic significance of BRCA1/2--
mutation status with neoantigen load, number of
tumor-infiltrating lymphocytes and expression of
PD-1/PD-L1 in high grade serous ovarian cancer.
Oncotarget 2016;7:13587–98.
75. Richtig G, Hoeller C, Kashofer K, et al. Beyond the
BRAFV600E hotspot: biology and clinical implica-
tions of rare BRAF gene mutations in melanoma
patients. Br J Dermatol 2017;177:936–44.
76. Ponti G, Pellacani G, Tomasi A, et al. The somatic
affairs of BRAF: tailored therapies for advanced
malignant melanoma and orphan non-V600E
(V600R-M) mutations. J Clin Pathol 2013;66:
441–5.
77. Ross JS, Wang K, Chmielecki J, et al. The distribu-
tion of BRAF gene fusions in solid tumors and
response to targeted therapy. Int J Cancer 2016;
Cancer Genetics and Epigenetics
138:881–90.
78. Nogova L, Sequist LV, Perez Garcia JM, et al.
Evaluation of BGJ398, a fibroblast growth factor
receptor 1–3 kinase inhibitor, in patients with
advanced solid tumors harboring genetic alter-
ations in fibroblast growth factor receptors: results
of a global phase I, dose-escalation and dose-
expansion study. J Clin Oncol 2017;35:157–65.
79. Nykamp K, Anderson M, Powers M, et al. She-
rloc: a comprehensive refinement of the ACMG-
AMP variant classification criteria. Genet Med
2017;19:1105–17.
80. Plon SE, Eccles DM, Easton D, et al. Sequence
variant classification and reporting: recommenda-
tions for improving the interpretation of cancer
susceptibility genetic test results. Hum Mutat
2008;29:1282–91.
81. Robson M, Im SA, Senkus E, et al. Olaparib for
metastatic breast cancer in patients with a germline
BRCA mutation. N Engl J Med 2017;377:523–33.
82. Ledermann J, Harter P, Gourley C, et al. Olaparib
maintenance therapy in platinum-sensitive
relapsed ovarian cancer. N Engl J Med 2012;366:
1382–92.
83. Moore K, Colombo N, Scambia G, et al. Mainte-
nance Olaparib in patients with newly diagnosed
advanced ovarian cancer. N Engl J Med 2018;379:
2495–505.
84. Lord CJ, Ashworth A. PARP inhibitors: synthetic
lethality in the clinic. Science 2017;355:1152–8.
85. Dougherty BA, Lai Z, Hodgson DR, et al. Biologi-
cal and clinical evidence for somatic mutations in
BRCA1 and BRCA2 as predictive markers for
olaparib response in high-grade serous ovarian
cancers in the maintenance setting. Oncotarget
2017;8:43653–61.
86. Wood RD, Mitchell M, Sgouros J, et al. Human
DNA Repair Genes. Science 2001;291:1284–9.
87. Heeke AL, Pishvaian MJ, Lynce F, et al. Preva-
lence of homologous recombination-related gene
mutations across multiple cancer types. JCO Precis
Oncol 2018;2:1–13.
88. Hussain M, Daignault-Newton S,
Twardowski PW, et al. Targeting androgen recep-
tor and DNA repair in metastatic castration-
resistant prostate cancer: results from NCI 9012.
J Clin Oncol 2018;36:991–9.
89. Pearl LH, Schierz AC, Ward SE, et al. Therapeutic
opportunities within the DNA damage response.
Nat Rev Cancer 2015;15:166–80.

3010
90. Scarpa A, Chang DK, Nones K, et al. Whole-
genome landscape of pancreatic neuroendocrine
tumours. Nature 2017;543:65–71.
91. Aisner DL, Nguyen TT, Paskulin DD, et al.
ROS1 and ALK fusions in colorectal cancer,
with evidence of intratumoral heterogeneity for
molecular drivers. Mol Cancer Res 2014;12:111–8.
92. Soda M, Choi YL, Enomoto M, et al. Identification
of the transforming EML4-ALK fusion gene in non-
small-cell lung cancer. Nature 2007;448:561–6.
Cancer Genetics and Epigenetics
93. Cantile M, Marra L, Franco R, et al. Molecular
detection and targeting of EWSR1 fusion
transcripts in soft tissue tumors. Med Oncol 2013;
30:412.
94. Patel JN, Fong MK, Jagosky M. Colorectal cancer
biomarkers in the era of personalized medicine.
J Pers Med 2019;9:E3.
95. Ciccia A, Elledge SJ. The DNA damage response:
making it safe to play with knives. Mol Cell 2010;
40:179–204.
Int. J. Cancer: 145, 2996–3010 (2019) © 2019 UICC
10970215, 2019, 11, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32358, Wiley Online Library on [26/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Variant classification in precision oncology
96. Hakem R. DNA-damage repair; the good, the bad,
and the ugly. EMBO J 2008;27:589–605.
97. Tutt A, Tovey H, Cheang MCU, et al.
Carboplatin in BRCA1/2-mutated and
triple-negative breast cancer BRCAness sub-
groups: the TNT trial. Nat Med 2018;24:
628–37.
98. Abernethy AP, Etheredge LM, Ganz PA, et al.
Rapid-learning system for cancer care. J Clin
Oncol 2010;28:4268–74.