Next‐Generation Sequencing in the Diagnosis of Metastatic Lesions: Reclassification of a Glioblastoma as an Endometrial Cancer Metastasis to the Brain

Abstract

Endometrial cancer is the most common gynecologic cancer in the U.S., but metastasis to the brain is rare, and diagnosis can be challenging. Traditional tools for determining if a tumor is a primary or metastatic lesion include pan‐imaging, histopathologic studies, and immunohistochemistry. Molecular testing with next‐generation sequencing has been increasingly used to augment these tests. We present a case of a patient who initially presented with a brain lesion diagnosed as glioblastoma on histology and immunohistochemistry, but whose diagnosis was later changed to metastasis from an endometrial primary based on molecular findings. The two tumors shared a common microsatellite instability signature and 51 DNA variants, including oncogenic driver mutations KRAS p.G13D, PIK3CA p.E545A, and PTEN p.I135V and p.K267Rfs*9. This highlights the power of molecular analysis in making the diagnosis in cases of rare metastases.

Key Points

• Brain metastasis from endometrial primary is rare, and histopathological features may be augmented with molecular analysis to aid in diagnosis.

• Comparison of the molecular makeup of the primary endometrial lesion with the metastatic lesion may reveal high‐risk molecular features that may be indicative of metastatic potential.

Short abstract

Metastasis of endometrial cancer to the brain is rare, and diagnosis can be challenging. This case report highlights the power of molecular analysis in making the diagnosis in cases of rare metastases.

Patient Story

A healthy 64‐year‐old gravida 0 woman presented with worsening short‐term memory, visual changes, and word finding difficulty over several weeks and a 60‐lb unintentional weight loss over the preceding year. Pan–computed tomography (CT) revealed an 8‐by‐5 cm left temporal lobe mass with surrounding edema, midline shift, and subfalcine herniation (Fig. 1A), as well as a heterogenous mass‐like expansion of the uterus. The patient underwent an urgent left frontal parietotemporal craniotomy and microsurgical gross total resection of her brain tumor. Pathology demonstrated a high‐grade malignant neoplasm, composed of areas with sheet‐like growth of poorly differentiated epithelioid cells with nuclear pleomorphism, multiple mitoses and apoptosis, and geographic coagulative necrosis. A few foci showed sarcomatous features, with malignant cells placed singly in myxoid stroma (Fig. 2A–C). Vascular proliferation was present. These morphologic features are suggestive but not specific for gliomas, although what initially prompted the diagnosis of glioma was the presence of focal glial differentiation. The tumor cells were focally positive for glial fibrillary acidic protein (GFAP) and SOX2, which are indicative of glial and neuronal cells, and negative for IDH1 R132H mutation‐specific immunomarker, which is more common in low‐grade, less aggressive, gliomas. There was loss of MLH1 and PMS2 expression and retention of MSH2 and MSH6. As discussed below, MLH1 promoter methylation was performed in the uterine tumor and not the brain lesion; however, confirmation of microsatellite instability was observed in both tumors by next‐generation sequencing. A reticulin stain highlighted the sarcomatous elements. On molecular testing, MGMT methylation was present, which is observed in approximately 50% of grade 4 glioblastomas. Based on the immunohistochemical staining and molecular testing, the pathologic diagnosis was grade 4 glioblastoma, gliosarcoma variant.

Figure 1.

(A): Brain magnetic resonance imaging (MRI) at time of initial presentation demonstrating midline shift, ring‐enhancement, and surrounding edema. Metastatic brain lesions on MRI often appear as small, well‐defined, ring‐enhancing lesions surrounded by edema; they may also show central necrosis or hemorrhage, or both. However, nonspecific gliomas may also demonstrate central necrosis surrounded by a ring of contrast enhancement along with edema causing mass effect. (B): Pelvic MRI performed as follow‐up for pelvic mass.

Figure 2.

Histopathologic features of brain and endometrial tumors. The left parietal lesion showed a poorly differentiated malignant neoplasm in brain parenchyma (40×) (A), composed of sheets of poorly differentiated epithelioid cells (200×) (B) and rarer foci with sarcomatous features in myxoid stroma (200×) (C). The endometrial adenocarcinoma showed few well‐differentiated gland‐forming areas (200×) (D) but predominantly comprised more poorly differentiated sheets of poorly differentiated carcinoma with only occasional glandular lumina (200×) (E). Metastatic poorly differentiated carcinoma was present in the ovary (200×) (F). All photomicrographs are hematoxylin and eosin stains.

The patient was referred to Neuro‐Oncology with a plan to initiate radiation and chemotherapy. Prior to beginning adjuvant treatment, she had pelvic magnetic resonance imaging (MRI) to characterize the uterus that demonstrated a diffuse heterogeneous soft tissue mass replacing the myometrium with extension to the peritoneal reflection of the pelvis, left adnexa, and sigmoid colon with mildly enlarged pelvic lymph nodes (Fig. 1B). Endometrial biopsy demonstrated grade 1 endometrial adenocarcinoma.

She underwent a radical hysterectomy, bilateral salpingo‐oophorectomy, pelvic lymph node dissection, omentectomy, and resection of tumor from the cul‐de‐sac, left ureter, and rectosigmoid to no evidence of residual disease. The uterus was densely adherent to the left pelvic sidewall with tumor extending into the rectosigmoid colon and encasing the left ureter (pT3bN1M1). The final pathology demonstrated grade 3 endometrioid endometrial adenocarcinoma with focal well‐differentiated areas corresponding to the tumor seen on preoperative biopsy (Fig. 2D–F). The tumor invaded through the myometrium to the uterine serosa as well as the cervical stroma involving the paracervical soft tissues at the radial resection margin; lymphovascular invasion was present. Metastatic endometrial adenocarcinoma was identified in the left ovary, pelvic peritoneum, perirectal fat, and a single obturator node. Immunohistochemistry demonstrated loss of MLH1/PMS2 with MLH1 promoter methylation, consistent with a sporadic microsatellite instability.

Molecular Tumor Board

Given the poorly differentiated morphology of both the endometrial adenocarcinoma and the brain tumor and the shared MLH1 and PMS2 deficiency, the molecular profiles of the two tumors were compared by a previously validated targeted next‐generation sequencing panel designed for the detection of single‐nucleotide variants, insertions and deletions, copy number alterations, and structural variants across 447 genes [1]. Mutation analysis for single‐nucleotide variants was performed using MuTect [2] and annotated by Oncotator [3], and insertions and deletions were called using Indelocator (https://software.broadinstitute.org/cancer/cga/indelocator). Integrative Genomics Viewer (version 2.0.16) and an internally developed application were used for visualization and interpretation. Variants were filtered to exclude those that occur at a populational frequency of greater than 0.1% in the Exome Sequencing Project database (http://evs.gs.washington.edu/EVS/). During the process of test validation, the filtering criteria were determined as follows: Each sequencing run, non‐neoplastic, formalin‐fixed, paraffin‐embedded liver and blood samples were included as controls. Variants identified in these control samples due to sequencing artifacts were filtered. Any filtered variants that were reported in the Catalogue of Somatic Mutations in Cancer (COSMIC v94, United Kingdom) more than twice were rescued and presented for manual review [1]. The two tumors shared a shared microsatellite instability signature and 46 DNA variants, including oncogenic driver mutations, such as KRAS p.G13D, PIK3CA p.E545A, and PTEN p.I135V and p.K267Rfs*9. Upon manual review, there was a smaller number of variants unique to each tumor, mostly present at low allelic fraction (Table 1; 17 variants in the brain lesion and 26 variants in the endometrial lesion). PTEN and KRAS are well‐established mutational drivers in endometrial cancer. However, whereas PTEN mutation is also commonly implicated in brain tumors, KRAS is rarely identified in gliomas [4]. ARID1A alterations are also frequently seen in endometrial carcinomas but are rarely part of the pathogenesis of gliomas [5].

Table 1.

Pathogenic elements of the molecular profile of patient's endometrial tumor compared with left parietal tumor

Endometrial adenocarcinoma				Left parietal tumor
Tumor mutational burden/megabase: 43.343				Tumor mutational burden/megabase: 35.739
Mismatch repair status: Deficient (MMR‐D / MSI‐H)				Mismatch repair status: Deficient (MMR‐D / MSI‐H)
Tumor %: 70%				Tumor %: 70%
Gene	Coverage	V.A.F.	AA change	Gene	Coverage	V.A.F.	AA change
APC	263	0.395	p.E1317Q	APC	326	0.475	p.E1317Q
AR	128	0.328	Intron	AR	130	0.354	Intron
ARID1A	280	0.357	p.P224Rfs*8	ARID1A	260	0.408	p.P224Rfs*8
ARID1A	310	0.248	p.M1564*	ARID1A	310	0.38	p.M1564*
ATM	281	0.32	p.S941Ifs*28	ATM	241	0.427	p.S941Ifs*28
ATR	169	0.373	p.S1965C	ATR	129	0.302	p.S1965C
ATR	245	0.322	p.P315H
				ATRX	82	0.378	Splice region
ASXL1	304	0.309	p.G645Vfs*58	ASXL1	280	0.371	p.G645Vfs*58
AURKB	270	0.322	p.P226H	AURKB	249	0.422	p.P226H
B2M	263	0.034	p.L7*
B2M	186	0.392	splice aceptor
B2M	434	0.295	p.V69Wfs*34
BCL11B	82	0.329	p.P314Q	BCL11B	68	0.515	p.P314Q
BCOR	402	0.281	p.V432I	BCOR	484	0.434	p.V432I
BCORL1	246	0.228	p.R1292Q	BCORL1	424	0.186	p.P1007H
C1orf86	79	0.342	p.A107T
				CREBBP	358	0.081	p.A298T
CSF3R	194	0.325	p.A543T	CSF3R	160	0.419	p.A543T
CTCF	279	0.326	p.T204Qfs*18	CTCF	274	0.438	p.T204Qfs*18
DICER1	225	0.307	p.L1408P	DICER1	247	0.389	p.L1408P
DIS3L2	379	0.061	p.E523K
DKC1	154	0.052	p.K498del
ERBB3	297	0.481	Splice region	ERBB3	300	0.473	Splice region
ERCC6	282	0.436	p.R557H	ERCC6	265	0.479	p.R557H
FAT1	439	0.314	p.E160G	FAT1	439	0.246	p.D2270N
FLCN	235	0.272	p.H429Tfs*39	FLCN	172	0.39	p.H429Tfs*39
HELQ	266	0.44	p.M1036I	HELQ	252	0.528	p.M1036I
IKZF1	325	0.308	p.A138V
				IKZF1	413	0.414	Intron
				JAZF1	454	0.198	p.P232Q
KAT6A	407	0.236	p.P1905H
KDM5A	499	0.299	p.G1200Dfs*9
KRAS	260	0.35	p.G13D	KRAS	259	0.425	p.G13D
				KMT2D	523	0.415	p.G1235Vfs*95
MDM2	158	0.063	Splice region
				MED12	263	0.179	p.R1214H
MET	343	0.504	Intron	MET	363	0.54	Intron
MET	151	0.298	Intron	MET	200	0.425	Intron
MITF	294	0.333	p.R74T	MITF	364	0.426	p.R74T
MYC	316	0.345	p.H374R	MYC	350	0.454	p.H374R
NTRK1	331	0.263	p.G248W	NTRK1	404	0.292	p.G248W
NSD1	362	0.318	p.G1800Wfs*3	NSD1	335	0.549	p.G1800Wfs*3
NOTCH1	257	0.233	p.D1502V
NFKBIA	228	0.057	p.E55*
NF1	305	0.357	p.G1190C
				NF1	346	0.37	p.I679Dfs*21
				NF1	368	0.457	p.Q1841Nfs*22
NT5C2	208	0.298	p.S251N
PAX5	310	0.319	p.V26Gfs*49	PAX5	324	0.358	p.V26Gfs*49
				PAX5	148	0.25	p.P330S
				PDGFRA	241	0.058	p.V129A
PIK3CA	134	0.276	p.E545A	PIK3CA	154	0.331	p.E545A
PIK3R1	220	0.341	p.K567E
POLD1	206	0.063	p.Y894C
POLE	317	0.483	p.Y623C	POLE	338	0.453	p.Y623C
POLQ	148	0.358	Splice region	POLQ	106	0.406	Splice region
PRKDC	233	0.047	p.R1136L	PRKDC	298	0.064	Splice donor
PRSS1	468	0.064	p.A173G	PRSS1	445	0.043	p.A173G
PRF1	373	0.067	p.L478*
				PTCH1	124	0.137	p.R1308Efs*64
PTEN	276	0.322	p.I135V	PTEN	274	0.394	p.I135V
PTEN	230	0.383	p.K267Rfs*9	PTEN	228	0.447	p.K267Rfs*9
RARA	91	0.066	p.P440Rfs*203
RASA1	237	0.35	p.M718T	RASA1	285	0.435	p.M718T
				RB1	174	0.115	p.W78*
				RBBP8	243	0.412	p.K357Nfs*3
SETD2	421	0.314	p.S2382Lfs*47	SETD2	366	0.331	p.S2382Lfs*47
SH2B3	332	0.476	p.E292Q	SH2B3	455	0.479	p.E292Q
SLX4	278	0.299	p.R204H	SLX4	219	0.484	p.R204H
SMARCE1	236	0.22	p.R322M
SMARCA4	250	0.296	p.T910M	SMARCA4	311	0.354	p.T910M
STAG2	182	0.253	p.A550T	STAG2	184	0.386	p.A550T
TDG	408	0.304	p.S352N	TDG	464	0.332	p.S352N
TERT	142	0.493	p.V251I	TERT	140	0.529	p.V251I
TOPBP1	341	0.34	p.Q1233H	TOPBP1	330	0.47	p.Q1233H
TSC1	372	0.341	p.A616T
TSC1	202	0.045	p.Y195C
				TSC2	61	0.508	p.C32Vfs*24
TRIM37	241	0.261	p.R152W	TRIM37	217	0.438	p.R152W
XPO1	113	0.097	Splice region
XRCC4	166	0.373	p.P119T	XRCC4	173	0.509	p.P119T
ZNF217	463	0.313	p.D528N	ZNF217	539	0.44	p.D528N
Intergenic variant	421	0.29	p.E48del
Intergenic variant	197	0.508	p.T193S
Intergenic variant	411	0.365	p.W343Gfs*7
				Intergenic variant	404	0.396	Intergenic variant
				Intergenic variant	150	0.513	Intergenic variant
				Intergenic variant	469	0.452	Intergenic variant

Gray shading indicates the unique variants present in each tumor sample. The minimal coverage for each region was at least 50 reads.

Abbreviations: AA, amino acid; MMR‐D, mismatch repair deficient; MSI‐H, microsatellite instability high; VAF, variant allele fraction.

Copy number alterations are estimated using an equation to calculate copy numbers based on the normalized log₂ ratio and tumor percentage: number of copies = (2^{avg. log2ratio}) × (100 ÷ [tumor percentage]) × 2. The avg.log2ratio is calculated across all of the segments represented a single gene. In assessing copy number alterations, including gains and losses, the results highlighted the clonal relationship between the primary endometrial and metastasis to the brain (Fig. 3). There was an isolated gain of chromosome arm 1q in the primary endometrial carcinoma. Endometrioid carcinomas are generally slightly hyperdiploid with chromosome gains involving mainly the long arm of chromosome 1 (70% of the cases), although isochromosomes or unbalanced translocations can be also seen [6]. The metastatic brain tumor shows additional loss of short arm of chromosome 5 that could also represent a sign of tumor evolution. This was interpreted as strong molecular evidence for a clonal relationship between the endometrial and brain tumors.

Figure 3.

Copy number plots of primary endometrial and brain tumors. (A): Endometrial carcinoma showing isolated gain of chromosome arm 1q. (B): Metastatic brain tumor with gain along 1q and a broad loss of 5q denoting some degree of tumor evolution with the additional loss on chromosome 5. The similarity of copy number alterations is remarkable and reinforces that the tumors are related.

Additional immunohistochemical stains were then performed on the brain tumor, which showed rare cells positive for pan‐cytokeratin markers and for OLIG2 (a marker of glial differentiation). The tumor cells were negative for PAX8. On integration of the molecular findings, despite multiple unusual morphologic features and a nonspecific immunophenotype, the brain tumor was classified as a poorly differentiated malignant neoplasm with glial differentiation, most compatible with metastasis from the endometrial primary. This ultimately changed the diagnosis from two separate primaries to metastatic poorly differentiated endometrial cancer.

KRAS is a proto‐oncogene located at chromosome 12p12.1 that encodes a signaling protein that moderates response to extracellular signals via downregulation of the MAPK or PI3K/AKT pathways [7]. KRAS plays a role in both an early checkpoint of transition from endometrial hyperplasia to carcinoma (present in 6%–16% of endometrial hyperplasia) and is a marker of invasive potential in the case of grade 1 endometrioid carcinoma [7]. KRAS mutation has also been associated with increased activation of estrogen signaling that is mediated through the Ras/MAPK pathway and provides a route to circumvent MEK inhibition to ultimately upregulate MAPK signaling. From a targetable therapy standpoint, this suggests that combination treatment of MEK inhibitors (e.g., trametinib) with antiestrogen therapy is required for efficacy in patients with KRAS mutant tumors; other combinations of MEK inhibitors, such as AKT inhibitors, are currently under investigation [8].

PIK3CA mutation is present in approximately 50% of endometrial cancers, and PTEN acts as a negative regulator of the PI3K‐AKT pathway and has been identified in up to 55% of endometrial cancers [9]. PTEN protein loss has also been found to be correlated with improved overall survival [10]. Several PI3K pathway inhibitors have been developed and are currently under investigation in preclinical studies and early clinical trials. PTEN mutations also trigger cellular senescence through a TP53‐dependent pathway associated with mTOR activation; modest effects have been shown with mTOR inhibitors in patients with these mutations [11].

Lastly, microsatellite instability (MSI) was also one of the key molecular alterations in this tumor. MSI is caused by aberrant mismatch repair gene and is identified in 20%–30% of patients with endometrial cancer, with Lynch syndrome accounting for approximately 25% of these cases [12]. A recent study showed that MSI analysis is effective as a predictive biomarker for the effect of immune checkpoint inhibitors, which are new anticancer drugs, including anti–programmed cell death protein 1 (PD‐1)/programmed cell death ligand 1 (PD‐L1) antibody [13].

The mutational profile in this case, including mutations in KRAS, PTEN, and PIK3CA as well as microsatellite instability, is characteristic of endometrioid neoplasia and therefore in keeping with the histologic findings in the endometrial tumor. The biological basis for dedifferentiation and distant metastasis in this tumor is not entirely clear, although next‐generation sequencing did reveal a hotspot (pT910M) mutation in the SWI/SNF protein SMARCA4, which is reported in a subset of dedifferentiated endometrial carcinomas and is associated with aggressive behavior [14]. SMARCA4‐deficient tumors, such as small cell carcinoma of the ovary and non‐small cell lung cancer, have been reported to respond to PD‐L1 inhibition and other novel immunotherapies [15].

Patient Update

The patient recovered well from her abdominal surgery and shortly afterward began adjuvant treatment for her presumed glioblastoma with temozolomide and brain radiation. Unfortunately, repeat imaging after 6 weeks demonstrated recurrent/residual tumor at the vaginal cuff and new peritoneal carcinomatosis. Temozolomide was discontinued based upon the change in diagnosis from primary glioblastoma to metastatic endometrial cancer. She completed pelvic radiation and two cycles of carboplatin/paclitaxel but tolerated chemotherapy poorly with weight loss, malnutrition, and cytopenia. She received one cycle of single agent paclitaxel with a subsequent rapid decline. Given her poor performance status, targeted treatments were not initiated. She was admitted with failure to thrive, transitioned to comfort measures only, and passed away 6 months after her initial presentation.

Discussion and Conclusion

The diagnostic dilemma of identifying a primary versus metastatic brain lesion has been well summarized in the literature [16]. Gliomas are the most common type of primary brain tumor (5 in 100,000), but overall, metastasis is more common than primary tumor (12 in 100,000). Presenting symptoms in brain metastases are similar to those of primary brain tumors (e.g., headache, aphasia, localized neurological deficits, and seizures). CT and MRI are commonly used for detecting brain lesions. Perfusion MRI has been used to differentiate primary gliomas and brain metastases, where measurements of relative cerebral blood volume of the edema are lower in metastases compared with glioma. However, radiologic appearance alone will not always lead to a specific diagnosis, as was the challenge in the present case. In general, multiple brain lesions are more suspicious for metastatic process than are single lesions.

The brain is a relatively uncommon site of metastasis for endometrial cancer (0.3%–1.16%), making this diagnosis difficult [17, 18, 19]. Although deep myometrial invasion has been suggested to be the strongest predictor of hematogenous dissemination and thus risk of spread to lung, liver, bone, and brain, the majority of patients with brain metastases do not have high‐risk factors (e.g., lymphovascular space invasion; microcystic, elongated, and fragmented (MELF) pattern; positive cytology), which suggests that despite early stage and low‐grade disease, there may be other inherent high‐risk features [19].

The use of molecular genetic testing has been increasingly used to assist in diagnosis, prognosis, and tailored treatment. In the case presented, we highlight the power of molecular testing as a diagnostic tool, especially in identifying tissue of origin for metastatic cancer to a rare site. Traditional tools for identifying tissue of origin in the setting of metastases with unknown or uncertain primary site include pan‐imaging, histopathologic studies, and immunohistochemistry. Initial accurate histopathologic diagnosis in this case was limited by the poorly differentiated nature of the patient's brain lesion and the lack of clinical and radiological correlation to the occult uterine primary at the time of brain biopsy. Molecular genetic testing has been increasingly used to augment these traditional tools. In one prospective study, molecular testing predicted tissue of origin in 247 of 252 patients with metastatic cancer of unknown primary site (98%) [20]. In this case, although immunohistochemistry gave the initial clue to the relationship between the tumors (as both had MLH1 and PMS2 deficiencies), the similarities in molecular signature led to the ultimate diagnosis.

Integrated genomic and proteomic analysis from The Cancer Genome Atlas (TCGA) has expanded our understanding of disease biology and diagnostic classification of endometrial cancer [21]. TCGA established four distinct endometrial cancer subclasses based on the extent of mutational load and somatic copy number alterations, each with its own prognostic significance: POLE‐mutant/ultramutated, MSI high/hypermutated, copy number low, and copy number high [21]. Since the development of TCGA, various next‐generation sequencing methods have been used to explore the genomic makeup of endometrial cancer. TCGA focused primarily on primary disease; 75% of samples were collected from patients who did not develop recurrent disease. Thus, less is known about the genomic landscape of tumors that metastasize or recur at extrapelvic sites. Furthermore, as highlighted in the present case, confounding elements are present in distinguishing primary from metastatic brain lesion even in the molecular mutational profile. For example, glioblastomas have been reported to harbor PTEN and PIK3CA alterations, among others, in whole‐exome sequencing of 543 glioblastoma samples [22]. In contrast, the MGMT promoter methylation seen in the patient's brain lesion is infrequent in endometrial cancer [23]. The variation in molecular signatures would be clarified as the genomic landscape of tumors that metastasize versus those that do not is further elucidated.

A study from Memorial Sloan Kettering assessed the clinical utility of prospective molecular characterization in 189 patients with advanced endometrial cancer in which 51% had tumor samples from metastatic sites [24]. They identified 63% with TP53 mutation, 56% with PTEN alterations, 70% with PI3K/AKT/mTOR pathway alteration, and 65% with RTK/RAS/β‐catenin pathway alterations. Phase II trials have investigated PI3K/AKT/mTOR inhibitors with encouraging preliminary results. Furthermore, high MSI was identified in 16% of the cohort, which has significant clinical implications, as pembrolizumab, a PD‐1 inhibitor, and lenvatinib, a multikinase inhibitor, have promising antitumor activity in MSI‐high/mismatch repair–deficient endometrial cancer [25]. Thus, an in‐depth understanding of the molecular makeup of endometrial carcinomas could have significant treatment and outcome implications among patients with metastases.

Additional high‐risk features may be revealed by molecular analysis. In the Memorial Sloan Kettering cohort mentioned above, the researchers noted four patients with matched primary and metastatic tumors with some genomic heterogeneity that involved potentially actionable hotspot mutations, including mTOR L2427Q and PIK3R1 E468* [24]. This suggests that metastatic lesions may have a higher rate of unique targetable mutations. However, lack of paired samples limited their ability to comment on genetic alterations in the tumor in response to therapy, and metastases were not categorized based on site.

In conclusion, given the importance of correct identification of tissue of origin in developing an appropriate oncologic treatment plan, use of molecular testing in patients with metastatic disease represents a significant opportunity. In cases of unclear tissue of origin, rare metastatic presentation, and/or other diagnostic uncertainty, we advocate for the use of molecular testing to both arrive at a diagnosis and to develop a targeted treatment plan based on specific mutations.

Glossary of Genomic Terms and Nomenclature

AKT: v‐akt murine thymoma viral oncogene.

ARID1: AT‐rich interactive domain‐containing protein 1A.

KRAS: Kirsten rat sarcoma viral oncogene.

MAPK: mitogen‐activated protein kinase.

mTOR: mammalian target of rapamycin.

PTEN: phosphatase and tensin homolog.

PI3K phosphoinositide‐3‐kinase.

PIK3CA: PI3K catalytic subunit α.

Author Contributions

Conception/design: Shuk On Annie Leung

Provision of study material or patients: Marisa Nucci, Michael G. Muto, Susana Campos

Collection and/or assembly of data: David Chapel, Annacarolina DaSilva

Manuscript writing: Shuk On Annie Leung, Olivia Foley, David Chapel, Annacarolina Da Silva

Final approval of manuscript: Shuk On Annie Leung, Olivia Foley, David Chapel, Annacarolina Da Silva, Marisa Nucci, Michael G. Muto, Susana Campos

Disclosures

The authors indicated no financial relationships.