Kinase fusions are rare and poorly characterized in colorectal carcinoma, yet they present unique opportunities for targeted therapy. In this study, we characterized kinase fusions from patients with advanced colorectal carcinoma who had MSK-IMPACT testing of their tumors between January 2014 and June 2018. Patients were analyzed for the presence of fusions, microsatellite instability (MSI), and RAS/BRAF mutations. Mismatch repair (MMR), IHC, and promoter hypermethylation status of MLH1 (MLH1ph) in microsatellite instability-high (MSI-H) colorectal carcinoma with fusions were investigated. Fusion transcripts were confirmed using a targeted RNA-seq panel assay. Of 2,314 colorectal carcinomas with MSK-IMPACT testing, 21 harbored kinase fusions. Overall 57% (12/21) of colorectal carcinoma fusions were MSI-H/MMR-D. Loss of MLH1 and MLH1ph was confirmed in all 12 and all 10 cases with available material, respectively. Fusions were present in 5% of MSI-H/MMR-D colorectal carcinoma compared with 0.4% of MSS/MMR-P colorectal carcinoma (P < 0.001) and 15% of MSI-H/MMR-D colorectal carcinoma with wild-type RAS/BRAF. Of 24 total MLH1-deficient colorectal carcinomas with MLH1ph and wild-type RAS/BRAF, 10 (42%) harbored kinase fusions. Kinase fusions in MSI-H colorectal carcinoma were associated with sporadic MLH1ph rather than with Lynch syndrome, and these patients may be eligible for kinase inhibitors, particularly following resistance or toxicity in response to immunotherapy. These findings identify a molecular subset of colorectal carcinoma with kinase fusions that may be responsive to kinase inhibitors.

Significance: A high frequency of targetable kinase fusions in BRAF/RAS wild-type, MSI-H colorectal carcinoma offers a rationale for routine screening to identify patients with colorectal carcinoma with kinase fusions that may be responsive to kinase inhibitors.

See related commentary by Valeri, p. 1041

Approximately 15% of colorectal carcinomas demonstrate mismatch repair deficiency (MMR-D)/microsatellite instability-high (MSI-H) status. The majority of these are MLH1/PMS2 deficient due to MLH1 promoter hypermethylation (MLH1ph). BRAF V600E mutations occur in approximately 50% of colorectal carcinomas with MLH1ph and have been shown to induce MLH1ph via upregulation of the transcriptional regulator MAFG (1). KRAS mutations occur in approximately 30% of MSI-H colorectal carcinoma MLH1ph (2), leaving 20% of colorectal carcinomas with MLH1ph without a known driver activating the MAPK signaling pathway. Isolated cases of MSI-H colorectal carcinoma with fusions have recently been reported (3–5), and we noted a similar trend in our clinical next-generation sequencing (NGS) data. We provide a detailed delineation of this association, defining a previously unappreciated subset of colorectal carcinomas with important therapeutic implications.

Written informed consent was obtained from patients, approval was obtained from our institutional review board, and this retrospective study was conducted in accordance with U.S. Common Rule. Colorectal carcinomas accessioned for MSK-IMPACT (6) and/or Archer NGS testing were assessed for kinase fusions. Reagents and primers for Archer NGS testing were obtained from ArcherDx. Patients with MSK-IMPACT testing had MSI status routinely assessed as a component of the assay (7). Archer fusion testing was clinically performed when sufficient remaining material was present for cases with WT KRAS, NRAS, and BRAF by MSK-IMPACT or a 95 gene Ampliseq-based assay, the latter performed when material was insufficient for MSK-IMPACT. Archer was also performed to confirm fusion transcripts in cases with novel DNA-level structural variants predicted to form kinase fusions. The custom Archer panel used covers fusions involving the kinase domains of the following genes: ALK, BRAF, EGFR, ERBB2, ERBB4, FGFR1, FGFR2, FGFR3, KIT, MET, NTRK1, NTRK2, NTRK3, RET, and ROS1. When tissue is available, MMR IHC is routinely clinically performed, and these data were recorded for patients with Ampliseq testing (which does not generate MSI status results).

Clinicopathologic characteristics of all colorectal carcinomas with kinase fusions were assessed. Primary site was classified as either proximal (cecum to transverse colon) or distal (splenic flexure to rectum). Differentiation and mucinous histology were scored on the basis of World Health Organization criteria (8). Well-differentiated colorectal carcinomas had >95% gland formation, moderately differentiated colorectal carcinomas had 50%–95% gland formation, poorly differentiated colorectal carcinomas had 0%–49% gland formation. Mucinous adenocarcinoma had an extracellular mucin component of >50%, whereas colorectal carcinoma with a mucinous component had extracellular mucin pools comprising <50% of the lesion.

MLH1ph was detected via bisulfite conversion followed by either pyrosequencing or methylation array depending on specimen availability. Colorectal carcinomas with MLH1ph and wild-type (WT) for KRAS or NRAS p. G12, G13, Q61, K117, A146, and BRAF p. V600 alleles were retrospectively screened with a custom Archer–targeted RNA-seq-based NGS assay used for fusion and alternative isoform testing (9). Confirmatory pan-Trk IHC was performed on colorectal carcinomas with NTRK fusions (10).

A subset of colorectal carcinomas with either BRAF V600E, kinase fusions, or KRAS mutations had genome-wide methylation profiling performed using the Illumina methylationEPIC (850k) platform (11). After excluding CpG sites from the MLH1 gene and X/Y chromosomes from the datasets, unsupervised hierarchical clustering was performed on the 10,000 most variable CpG sites (by standard deviation) using Euclidean distance and Ward method with R (version 3.4).

All of the above assays were clinically validated assays that were performed in CLIA-accredited laboratories.

Prevalence and spectrum of kinase fusions in colorectal carcinoma

We identified 2,314 colorectal carcinomas accessioned for MSK-IMPACT and/or Archer between January 2014 and June 2018. This dataset included 2,309 patients with colorectal carcinomas with MSK-IMPACT results, of which 189 also underwent Archer-targeted RNA-seq testing, and 5 additional patients with insufficient material for MSK-IMPACT whose tumors underwent RAS/BRAF testing by Ampliseq, followed by Archer testing. Seventeen colorectal carcinomas were positive for kinase fusions via MSK-IMPACT. Four additional colorectal carcinomas with fusions were detected using Archer-targeted RNA-seq assay: 3 cases were negative by MSK-IMPACT due to lack of coverage of breakpoints (EML4-NTRK3, FGFR3-STAB1, and FGFR2-MYH15), whereas the fourth case (TPM3-NTRK1) identified by Archer testing alone had insufficient DNA for MSK-IMPACT and had WT KRAS/NRAS/BRAF by outside NGS testing, yielding a total of 21 colorectal carcinomas positive for kinase fusions.

The detected fusions included 8 NTRK fusions (6 NTRK1 and 2 NTRK3), 5 BRAF fusions, 4 RET fusions, 2 FGFR fusions (1 each of FGFR2 and FGFR3), 1 ROS1 fusion, and 1 ALK fusion (Table 1 and Fig. 1). All detected kinase fusions were predicted to be in frame, included the kinase domain of the 3′ gene, and occurred in colorectal carcinomas that were BRAF/RAS WT. All 6 NTRK1 fusions and 1 of the 2 NTRK3 fusions were positive for pan-Trk IHC, with results as described previously (10).

Clinicopathologic characteristics of colorectal carcinoma with kinase fusions

The age at diagnosis of these 21 patients with colorectal carcinoma harboring kinase fusions ranged from 33–85 years with a median of 64 years. The majority (71%) of this cohort had colorectal carcinoma arising in the proximal colon. Poor differentiation (including medullary, n = 2) was present in 57% of the fusion cases, while 16% of cases had a mucinous component. Looking further into the fusion cohort, 83% of MSI-H colorectal carcinoma had poor differentiation or were mucinous in histologic subtype, whereas only 33% of MSS colorectal carcinoma with fusions had poor differentiation or a mucinous component. This data suggest that poor or mucinous differentiation may be associated with the MSI-H status rather than the presence of fusion. American Joint Committee on Cancer 8th edition stage at diagnosis included 6 stage II patients, 6 stage III patients, and 8 stage IV patients. Median follow-up time since diagnosis was 18 months. Sixty-eight percent of patients had distant metastasis at end of follow-up, and 76% of patients were alive at end of follow-up. These findings are summarized in Table 2.

Relationship of MSI to the presence of kinase fusions

Of the 2,314 total colorectal carcinomas, 230 were MSI-H/MMR-D and 2,084 were MSS/MMR-P. The presence of kinase fusions was mutually exclusive with BRAF V600 and RAS hotspot mutations. The MSI-H/MMR-D and MSS/MMR-P cohorts, respectively, harbored 74 (32%) versus 106 (5%) BRAF V600E mutations (P < 0.001), 83 (36%) versus 912 (44%) KRAS hotspot mutations (P = 0.322), 2 (1%) versus 86 (4%) NRAS hotspot mutations (P = 0.096), and 12 (5%) versus 9 (0.4%) kinase fusions (P < 0.001; Fig. 1). Fifteen percent of MSI-H/MMR-D and 0.9% of MSS/MMR-P colorectal carcinoma that were RAS/BRAF WT harbored kinase fusions.

MMR deficiency and relationship of MLH1 hypermethylation status to the presence of kinase fusions

Twelve (57%) of 21 colorectal carcinoma with kinase fusions were MMR-D/MSI-H. All MSI-H/MMR-D colorectal carcinoma with available material had MLH1/PMS2 loss (n = 12) and MLH1ph (n = 10). Looking further into the 71 MSI-H colorectal carcinomas that were RAS/BRAF WT, 47 were MLH1/PMS2 deficient by IHC. Twenty four of 37 of these MLH1/PMS2–deficient colorectal carcinomas with WT RAS/BRAF had MLH1ph data available were positive for MLH1ph. Of these 24 cases with MLH1 promoter hypermethylation, 10 harbored kinase fusions. Therefore, the incidence of fusions in MLH1-deficient colorectal carcinoma with MLH1ph and WT RAS/BRAF was 42% (Fig. 1).

Methylation array results

Because of the similarity of our findings relating fusions and MLH1ph to those of BRAF V600E and MLH1ph (1), we performed unsupervised hierarchical clustering of Illumina 850k methylation array data on both MSS and MSI-H colorectal carcinoma samples with fusions, BRAF V600E, and KRAS mutations after exclusion of MLH1 loci. Clear separation of hypermethylated and hypomethylated groups was evident. The hypermethylated group was composed of two predominant subclusters, suggesting CIMP-H and CIMP-L subgroupings. Eight of 11 (73%) fusion driven and 14 of 20 (70%) of BRAF V600E colorectal carcinomas localized to the hypermethylated group. All 19 (100%) KRAS mutants segregated to the hypomethylated group. Interestingly, 2 MSS colorectal carcinomas (1 fusion and 1 BRAF V600E) harbored MLH1ph.

In recent years, cancers bearing kinase fusions have shown some of the most dramatic and durable responses to kinase inhibitors (12, 13). For instance, larotrectinib has shown a response rate of 75% in adult patients with NTRK fusions, with 71% of responses ongoing and 55% of patients being progression-free at 1 year of treatment (12). Although such targetable fusions are rare in colorectal carcinomas overall, this study shows that approximately 15% of advanced MSI-H/MMR-D colorectal carcinomas, which are WT for BRAF/KRAS/NRAS, harbor kinase fusions and that all of the detected kinase fusions in MSI-H colorectal carcinoma occurred specifically in non-Lynch syndrome cases with MLH1 deficiency associated with MLH1ph. Furthermore, fusions were present in almost half of MLH1-deficient colorectal carcinomas with WT KRAS/NRAS/BRAF with MLH1ph.

A mechanistic basis for the relationship between BRAF V600E, genome-wide hypermethylation, and MSI has been proposed by Fang and colleagues, who showed that BRAF V600E mutations in colorectal carcinoma induce CpG island hypermethylation including MLH1ph via upregulation of ERK and MAFG, resulting in deficient MMR (1). The strong relationship between kinase fusions and MLH1ph suggests fusions may induce a similar phenomenon. Results from our methylation array studies show that BRAF p. V600E–mutant and kinase fusion–positive colorectal carcinomas have similar genomic CpG methylation patterns even after exclusion of data from the MLH1 promoter CpG loci. Functional studies elucidating the mechanistic relationship between kinase fusions and MLH1ph are warranted.

Our study does have several limitations. These include the rarity of kinase fusions in colorectal carcinoma and resulting relatively small cohort, the fact that none of the MSI-H/MMR-D colorectal carcinoma with fusions received a tyrosine kinase inhibitor and had available response data, and limited material on several of these cases, precluding MMR IHC, MSI testing, or MLH1ph.

To our knowledge, this study is the first to establish the relationship between kinase fusions and MSI-H colorectal carcinoma, specifically with MLH1ph. Given the rarity of fusions and the fact that fusion testing is not routinely performed on colorectal carcinoma, it is important to identify subtypes that are more likely to carry these fusions. Thus, testing for kinase fusions is warranted in advanced colorectal carcinoma with MLH1ph and WT BRAF/RAS and the current findings inform an updated proposed molecular testing workflow for colorectal carcinoma (Fig. 2). This proposed updated workflow begins with universal MSI or MMR IHC as recommended by the NCCN (14). Patients with colorectal carcinoma with MSS/MMR-P tumors should undergo NGS testing if available or KRAS/NRAS mutation analysis for eligibility for anti-EGFR therapy. Patients with MSI-H/MLH1–deficient colorectal carcinoma should undergo MLH1ph testing as part of the work-up for Lynch syndrome. If MLH1ph is not detected, MLH1 germline testing to rule out Lynch syndrome may be performed. For patients with colorectal carcinoma with deficiency of MSH2, MSH6, and/or PMS2 but not MLH1, germline testing of the deficient MMR gene is recommended because of the potential presence of Lynch syndrome. If MLH1ph is present, the patient has distant metastases, and the tumor is negative for BRAF p. V600E mutation, fusion testing may be performed because of the high likelihood of finding a kinase fusion with potential therapeutic implications.

Immune checkpoint inhibition produces response rates of 20% to 50% of MSI-H colorectal carcinoma (15, 16), and the presence of a kinase fusion would create a window of opportunity for treatment with kinase inhibitors when resistance or toxicity occurs after immune checkpoint inhibition therapy.

To conclude, while kinase fusions are rare in colorectal carcinomas overall (0.9%), 57% of kinase fusions in colorectal carcinomas occur in MMR-D/MSI-H colorectal carcinoma. These cases have MLH1ph and WT BRAF/RAS. Almost half of colorectal carcinomas with MLH1ph and WT RAS/BRAF harbor kinase fusions. This subset of advanced colorectal carcinomas may benefit from screening for oncogenic kinase fusions.

R. Yaeger reports receiving commercial research grant from Array BioPharma, Novartis Pharmaceuticals, and GlaxoSmithKline. M. Scaltriti reports receiving commercial research grant from Puma Biotechnology, Menarini Ricerche, Immunomedics, Daiichi Sankio, Targimmune and has ownership interest (including stock, patents, etc.) in Medendi Medical Travel. He is consultant/advisory board member for Menarini Ricerche, ADC Pharma, and Biocience Institute. A. Drilon reports receiving commercial research grant from Pfizer, GlaxoSmithKline, Teva, and Taiho; is a consultant/advisory board member for Ignyta, Loxo Oncology, BergenBio, Hengrui Therapeutics, Exelixis, Bayer, Tyra Biosciences, TP Therapeutics, AstraZeneca, Pfizer, Blueprint Medicines, Genentech/Roche, Takeda/Ariad/Millenium, Helsinn, and Beigene; and has provided expert testimony for Foundation Medicine, Wolters Kluwer, Merck, Medscape, OncLive, PeerVoice, PER, Targeted Oncology, and RTP. Z.K. Stadler is a consultant/advisory board member for Allergan, Genentech/Roche, Regenxbio, Regeneron, Optos, Adverum, Biomarin, Alimera Sciences, Novartis, Spark, and Fortress. D.M. Hyman reports receiving commercial research grant from Loxo Oncology, Puma Biotechnology, and AstraZeneca and is a consultant/advisory board member for Atara Biotherapeutics, Chugai Pharma, CytomX Therapeutics, Boehringer Ingelheim, AstraZeneca, Pfizer, Bayer, and Genentech. M. Ladanyi reports receiving commercial research grant from Loxo Oncology and Helsinn Therapeutics and is a consultant/advisory board member for Bayer. J.F. Hechtman reports receiving commercial research grant from Bayer, has received Speakers Bureau Honoraria from Medscape and Cor2Ed, and is a consultant/advisory board member for Axiom Biotechnologies. No other potential conflicts of interest were disclosed by the other authors.

Conception and design: E. Cocco, S. Middha, A.M. Schram, M. Ladanyi, J.F. Hechtman

Development of methodology: J. Benhamida, S. Middha, A. Zehir, L. Zhang, K. Nafa, J.F. Hechtman

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E. Cocco, J. Benhamida, A. Zehir, K. Mullaney, J. Shia, R. Yaeger, L. Zhang, D. Wong, A. Drilon, L. Saltz, A.M. Schram, Z.K. Stadler, D.M. Hyman, R. Benayed

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Benhamida, S. Middha, A. Zehir, K. Mullaney, R. Yaeger, L. Zhang, A. Drilon, L. Saltz, A.M. Schram, Z.K. Stadler, D.M. Hyman, R. Benayed, J.F. Hechtman

Writing, review, and/or revision of the manuscript: E. Cocco, J. Benhamida, S. Middha, A. Zehir, J. Shia, R. Yaeger, L. Zhang, K. Nafa, M. Scaltriti, A. Drilon, L. Saltz, A.M. Schram, Z.K. Stadler, D.M. Hyman, R. Benayed, M. Ladanyi, J.F. Hechtman

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): D.M. Hyman, J.F. Hechtman

Study supervision: K. Nafa, M. Scaltriti, A. Drilon, M. Ladanyi

Others (anything pertaining to iScan Methylation sample processing): L. Villafania

A. Drilon acknowledges Cycle for Survival award. E. Cocco acknowledges MSK society for a scholar prize. This study was funded by the NCI under the MSK Cancer Center Support Grant/Core Grant (P30 CA008748) and the R01CA226864 (to M.Scaltriti and A. Drilon). Archer testing was supported in part by a grant from LOXO Oncology (to M. Ladanyi.). A. Schram acknowledges NIH T32-CA009207 and ASCO Young Investigator Award.

1.
Fang
M
,
Ou
J
,
Hutchinson
L
,
Green
MR
. 
The BRAF oncoprotein functions through the transcriptional repressor MAFG to mediate the CpG island methylator phenotype
.
Mol Cell
2014
;
55
:
904
15
.
2.
Farchoukh
L
,
Kuan
SF
,
Dudley
B
,
Brand
R
,
Nikiforova
M
,
Pai
RK
. 
MLH1-deficient colorectal carcinoma with wild-type BRAF and MLH1 promoter hypermethylation harbor KRAS mutations and arise from conventional adenomas
.
Am J Surg Pathol
2016
;
40
:
1390
9
.
3.
Pietrantonio
F
,
Di Nicolantonio
F
,
Schrock
AB
,
Lee
J
,
Morano
F
,
Fuca
G
, et al
RET fusions in a small subset of advanced colorectal cancers at risk of being neglected
.
Ann Oncol
2018
;
29
:
1394
1401
.
4.
Yakirevich
E
,
Resnick
MB
,
Mangray
S
,
Wheeler
M
,
Jackson
CL
,
Lombardo
KA
, et al
Oncogenic ALK fusion in rare and aggressive subtype of colorectal adenocarcinoma as a potential therapeutic target
.
Clin Cancer Res
2016
;
22
:
3831
40
.
5.
Hechtman
JF
,
Zehir
A
,
Yaeger
R
,
Wang
L
,
Middha
S
,
Zheng
T
, et al
Identification of targetable kinase alterations in patients with colorectal carcinoma that are preferentially associated with wild-type RAS/RAF
.
Mol Cancer Res
2016
;
14
:
296
301
.
6.
Cheng
DT
,
Mitchell
TN
,
Zehir
A
,
Shah
RH
,
Benayed
R
,
Syed
A
, et al
Memorial Sloan Kettering-Integrated Mutation Profiling of Actionable Cancer Targets (MSK-IMPACT): a hybridization capture-based next-generation sequencing clinical assay for solid tumor molecular oncology
.
J Mol Diagn
2015
;
17
:
251
64
.
7.
Middha
S
,
Zhang
L
,
Nafa
K
,
Jayakumaran
G
,
Wong
D
,
Kim
HR
, et al
Reliable Pan-cancer microsatellite instability assessment by using targeted next-generation sequencing data
.
JCO Precis Oncol
2017
;2017.
doi: 10.1200/PO.17.00084
.
8.
Bosman
FT
,
Carneiro
F
,
Hruban R
H
,
Theise
N
.
WHO classification of tumours of the digestive system
. Fourth edition.
Lyon, France
:
IARC
; 
2010
.
9.
Zhu
G
,
Benayed
R
,
Ho
C
,
Mullaney
K
,
Sukhadia
P
,
Rios
K
, et al
Diagnosis of known sarcoma fusions and novel fusion partners by targeted RNA sequencing with identification of a recurrent ACTB-FOSB fusion in pseudomyogenic hemangioendothelioma
.
Mod Pathol
2018
.
doi: 10.1038/s41379-018-0175-7
.
10.
Hechtman
JF
,
Benayed
R
,
Hyman
DM
,
Drilon
A
,
Zehir
A
,
Frosina
D
, et al
Pan-Trk immunohistochemistry is an efficient and reliable screen for the detection of NTRK fusions
.
Am J Surg Pathol
2017
;
41
:
1547
51
.
11.
Moran
S
,
Arribas
C
,
Esteller
M
. 
Validation of a DNA methylation microarray for 850,000 CpG sites of the human genome enriched in enhancer sequences
.
Epigenomics
2016
;
8
:
389
99
.
12.
Drilon
A
,
Laetsch
TW
,
Kummar
S
,
DuBois
SG
,
Lassen
UN
,
Demetri
GD
, et al
Efficacy of larotrectinib in TRK fusion-positive cancers in adults and children
.
N Engl J Med
2018
;
378
:
731
9
.
13.
Christensen
JG
,
Zou
HY
,
Arango
ME
,
Li
Q
,
Lee
JH
,
McDonnell
SR
, et al
Cytoreductive antitumor activity of PF-2341066, a novel inhibitor of anaplastic lymphoma kinase and c-Met, in experimental models of anaplastic large-cell lymphoma
.
Mol Cancer Ther
2007
;
6
:
3314
22
.
14.
Benson
AB
 III
,
Venook
AP
,
Al-Hawary
MM
,
Cederquist
L
,
Chen
YJ
,
Ciombor
KK
, et al
Colon cancer, version 1.2017, NCCN clinical practice guidelines in oncology
.
J Natl Compr Canc Netw
2017
;
15
:
370
98
.
15.
Le
DT
,
Durham
JN
,
Smith
KN
,
Wang
H
,
Bartlett
BR
,
Aulakh
LK
, et al
Mismatch repair deficiency predicts response of solid tumors to PD-1 blockade
.
Science
2017
;
357
:
409
13
.
16.
Overman
MJ
,
McDermott
R
,
Leach
JL
,
Lonardi
S
,
Lenz
HJ
,
Morse
MA
, 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
.