Keywords

3.1 Introduction

Next-generation sequencing (NGS)-based tumor molecular profiling has become a fundamental component of precision medicine for cancer patients, which enables us to identify genetic alterations in genes and pathways for molecular-targeted therapies [1,2,3]. Several types of cancer genomic profiling (CGP) have proven their utility in cancer precision medicine, and two types of CGP (The OncoGuide™ NCC Oncopanel System and FoundationOne CDx Cancer Genomic Profile) were approved in Japan [4, 5]. However, most CGPs are based on analysis of genomic DNA of target genes, isolated either by polymerase chain reaction (PCR) or probe hybridization, enabling the detection of SNV, small insertions/deletions (indels), and CNV.

At The University of Tokyo, an original CGP assay, named Todai OncoPanel (TOP), was developed, which consists of a DNA panel and an RNA panel [6]. For the analysis, tumor DNA and RNA were prepared from FFPE tissues, and normal-paired DNA was extracted from peripheral blood collected from the same patients as a control to distinguish somatic and germline variants [6]. We analyzed >600 clinical samples since February 2017 under the approval of the institutional ethics committee. This clinical sequence assay consists of DNA and RNA hybridization capture-based next-generation sequencing panels that enable the comprehensive characterization of cancer-related genes. The TOP DNA panel can detect single-nucleotide variant (SNV), indels, and copy number variations (CNV) in 464 genes in version 3 (478 genes in version 4). Tumor mutational burden (TMB) and allele-specific copy number variations can be evaluated, and over 1000 microsatellite probes are included in the TOP DNA panel. The TOP RNA panel covers 463 genes in version 4 (678 genes in version 5). The current version can detect gene fusions in 504 genes, as well as exon skipping (such as MET and CTNNB1) and provide gene expression profiling [6]. Fusion genes in the TOP RNA panel include BCR-ABL1, EML4-ALK, RET, ROS1, NTRK1/2/3, FGFR1/2/3, and NRG1, all of which can be (or are promising to be) candidates for specific molecular-targeted agents [7, 8]. Here, we describe the efficacy and the utility of TOP for gynecological malignancies, using our comprehensive analysis of 54 gynecological malignancy cases.

3.2 Patient Characteristics

Between February 2017 and April 2018, 54 gynecological cancer patients (with 79 FFPE tumor specimens) were analyzed by the TOP panel at The University of Tokyo Hospital. Tumor specimens were mainly obtained by surgically resected tumors. Under written informed consent, we returned the results to the patients. The number of patients in each tumor type is 6 in cervical cancer, 15 in endometrial cancer, 21 in ovarian cancer (as well as 2 synchronous endometrial and ovarian carcinomas), 9 uterine sarcomas, and 1 choriocarcinoma (Fig. 3.1a). Comprehensive analysis of all tumor types was reported previously [6].

Fig. 3.1
figure 1

Cancer types and evidence level classification. (a) Distribution of tumor types among the 54 gynecological malignancy patients. (b) TOP evidence level classification. The evidence level classification was used to annotated gene alterations in TOP between 2017 and 2018. (c) Clinical actionability of gene alterations detected in gynecological cohort by the TOP panel. The maximum evidence level for each case is listed

Full size image

3.3 Clinical Annotations and Recommendation of Clinical Trials

We annotated somatic variants by using various types of public databases, including OncoKB, a curated knowledge database of oncogenic effects and treatment implications of gene alterations (http://oncokb.org); CIViC (Clnical interpretation of Variants in Cancer), a community-based curation database (http://civicdb.org), ClinVar (https://www.ncbi.nlm.nih.gov/clinvar/), and COSMIC (Catalogue of Somatic Mutations in Cancer, http://cancer.sanger.ac.uk/cosmic) to assess variants frequency in cancer. We additionally used specific databases for annotation of gene variants, including germline variants by BRCA Exchange (http://brcaexchange.org) for BRCA1/2, IARC TP53 (http://p53.iarc.fr) for TP53, and InSiGHT (https://www.insight-database.org/genes) for mismatch repair genes. We constructed our knowledge database using a website of the Pharmaceuticals and Medical Devices Agency, US Food and Drug Administration (FDA), and National Cancer Institute. We also included clinical trial databases, such as ClinicalTrials.gov and the Japanese clinical trial databases UMIN, JAPIC, and JMACCT.

Annotated variants were classified according to the level of evidence and drug availability [6] (Fig. 3.1b). In summary, Tier 1 was annotated to (pathogenic/likely pathogenic) variants with biomarkers to PMDA-approved drugs in the matched tumor type; Tier 2 was annotated to (pathogenic/likely pathogenic) variants with biomarkers applicable to clinical trials/FDA-approved drugs/PMDA-approved drugs in other tumor types. Tier 3 was for (pathogenic/likely pathogenic) variants, which were supported by knowledge databases for prediction of response to drugs or oncogenic alterations. Tier 4 corresponded to biomarkers, which were to be oncogenic. Tier 5 corresponded to biomarkers, which were recurrently reported in knowledge databases (Fig. 3.1b).

Recently, clinical and/or Experimental Evidence Levels have been standardized in Japan, which were defined by the Center for Cancer Genomics and Advanced Therapeutics (C-CAT) and Equivalent Evidence Levels in Other Guidelines [5].

3.4 Genetic Alterations in Gynecologic Malignancies by the TOP Panel

We evaluated the clinical utility of genetic alterations in each tumor in the TOP panel. All the tumors were subjected to both DNA and RNA panels. Each alteration in each case was discussed and annotated by the molecular tumor board at The University of Tokyo Hospital, which are currently applicable for various types of PMDA-approved CGPs. Therefore, information on clinical trials was confirmed between 2017 and 2018. Overall, 91% (49 out of 54) harbored one or more clinically annotated alterations (Tier 1 to Tier 5), and 46% (25 out of 54) harbored one or more actionable variants (Tier 1 or Tier 2) in gynecologic cancers (Fig. 3.1c). The proportion of actionable variants in cervical cancer, endometrial cancer, and ovarian cancer was 50%, 73%, and 57%, respectively. In these 54 cancer patients, the most frequently mutated gene with clinical annotation was TP53, followed by PIK3CA, PTEN, PIK3R1, KRAS, and ARID1A.

3.5 Cervical Cancer

Among the six cervical cancer patients, actionable mutations (Tier 2) were identified in three cases (50%) (Fig. 3.2a). Two were somatic oncogenic variants of PIK3CA, which matched to the clinical trial for an AKT inhibitor in Japan. Another case harbored the GOPC-ROS1 fusion gene (Table 3.1). ROS1 is a proto-oncogene located on chromosome 6q and encodes a receptor tyrosine kinase involved in the regulation of cancer cell growth and differentiation. ROS1 is often involved in genomic rearrangements resulting in constitutively active kinases that stimulate multiple pathways such as JAK-STAT, PI3K-AKT-mTOR, and RAS-RAF-MEK-ERK [9,10,11]. Fusion products of ROS1 have been observed in a variety of types of cancer, including lung, gastrointestinal tract and hepatobiliary tract, and central nervous system [9]. The ROS1 fusions are now considered as therapeutic biomarkers of crizotinib and entrectinib [12, 13].

Fig. 3.2
figure 2figure 2

Genetic alterations and clinical actionability by the TOP panel in gynecological cancers. (a–d) Detected gene alterations were annotated using the TOP classification. All the cases are assigned to the level with the most actionable alterations: (a) cervical cancer, (b) endometrial cancer, (c) uterine sarcoma, and (d) ovarian cancer. (e) Histological types of ovarian cancers. (f) Genetic alterations identified in 23 ovarian cancers

Full size image

3.6 Endometrial Cancer

We analyzed 15 endometrial cancer cases and identified 18 actionable gene alterations in 11 patients (73%) (Fig. 3.2b). Somatic variants of the PI3K-AKT pathway are most frequent, with 6 pathogenic mutations of PIK3CA, 7 of PTEN, and 2 of PIK3R1. These were considered for clinical trials with AKT inhibitors. Pathogenic somatic variants of BRCA1 were identified in one patient, which may be associated with homologous recombination deficiency (HRD). Thus, poly (ADP-ribose) polymerase (PARP) inhibitors were recommended in this patient, although no PARP inhibitors have been clinically approved in endometrial cancer. Pathogenic germline variants of MMR genes were identified in two patients (MSH6 (p.F858Sfs*12) and MSH2 (p.Q170*)) (Table 3.1). In addition to pathogenic variants in MMR genes, the primary tumors of these two patients were TMB-High, which also supports the recommendation of immune checkpoint inhibitors. Pembrolizumab was approved for microsatellite instability-high (MSI-High) solid tumors by FDA in 2017 and by PMDA in 2018 [14, 15].

3.7 Uterine Sarcoma

We analyzed nine uterine sarcomas and found that found no actionable mutations, except for one patient (Fig. 3.2c). No pathogenic alterations were detected in three patients (33%); however, two novel gene fusions were identified by the TOP RNA panel, which may be associated with tumorigenesis of uterine sarcomas. As NTRK gene fusions were identified at 4% in uterine leiomyosarcomas [16], the TOP RNA panel would be useful to identify actionable and/or novel gene fusions.

3.8 Ovarian Cancer

We enrolled 23 ovarian cancer cases. Totally, 15 actionable gene alterations were detected in 13 patients (Fig. 3.2d). Histological distribution was as follows: 14 (61%) with high-grade serous carcinomas, 2 with endometrioid carcinomas, 1 with clear cell carcinoma, 2 with mucinous borderline tumors, 1 with low-grade serous carcinoma, and 3 with malignant transformation of mature cystic teratomas (MCTMT) (Fig. 3.2e). The genome profile of ovarian cancers highly depends on histological types. For example, TP53 is mutated in >90% of high-grade serous carcinomas. Pathogenic somatic variants are listed in Fig. 3.2f. TP53 is the most frequently mutated gene, followed by PIK3CA, PTEN, KRAS, and ARID1A. Fourteen (12 high-grade serous carcinomas and 2 MCTMT) harbored pathogenic somatic variants of TP53. Pathogenic germline variants were identified in four patients (BRCA1 in two and BRCA2 in two patients, Table 3.1), and pathogenic somatic variant of BRCA1 was identified in one patient, all of which could be targeted by PARP inhibitors. Pathogenic somatic variants of the PI3K-AKT pathway were identified in five patients, which were the candidate targets of AKT inhibitors. Actionable alterations in FGF and FGFR were identified in four patients, which led to the enrollment of a clinical trial with an FGFR inhibitor. Pathogenic BRAF somatic variant was identified in one case with low-grade serous carcinoma (Table 3.1).

3.9 Choriocarcinoma

We analyzed one choriocarcinoma and detected three clinical annotated genes (Tier 5). However, no actionable gene alterations were identified.

3.10 Germline (Secondary) Findings

In gynecological malignancy, the ratio of germline variants is expected to be high, due to the prevalence of hereditary breast and ovarian cancer syndrome (HBOC) and Lynch syndrome [17, 18]. While the primary purpose of CGP tests has been considered to identify pathogenic somatic variants, germline findings may also guide personalized therapy and may be clinically significant, especially in gynecological malignancies. Indeed, we found pathogenic germline variants of mismatch repair genes and BRCA1/2 genes in endometrial and ovarian carcinomas, respectively (Table 3.1). The prevalence of pathogenic germline variants can be assessed by the TOP panel, as it analyzes paired-normal DNA from peripheral blood samples. The TOP panel basically includes cancer-related genes, which are recommended to report as germline (secondary) findings in clinical CGP [19, 20], and it is expected to contribute to find germline findings, which would be useful for both patients themselves and their relatives. As described above, TOP analysis of the 54 individuals in this study revealed 6 pathogenic germline variants (2 in BRCA1, 2 in BRCA2, 1 in MSH2, and 1 in MSH6) (11%), all of which were disclosed to each patient by the physicians and certified genetic counselors. Two patients wished confirmatory single-site test and were found to be pathogenic.

Table 3.1 Actionable mutations and candidate molecular-targeted drugs in gynecologic cancers
Full size table

3.11 Tumor Mutational Burden

The TMB representing the number of somatic variants per Mb is a biomarker to estimate the efficacy of immune checkpoint inhibitors. In 2019, Pembrolizumab was approved for adult and pediatric patients with TMB ≥10 [21]. In this study, we defined a threshold of 10 mutations/Mb as TMB-High and found that 7 cases (1 cervical cancer, 3 endometrial cancer, and 3 ovarian cancer) were TMB-high (Fig. 3.3). One cervical cancer patient exhibited TMB of 14.3/Mb. This patient received concurrent chemoradiation as primary treatment and the sample was obtained from the recurrent site in the heart. Careful caution is required to address TMB-high, as treatment by either chemotherapy or irradiation may increase the number of mutations. In endometrial cancer cases, three patients exhibited TMB-high (206.2/Mb, 21.4/Mb, and 20.2/Mb). One of the three patients harbored pathogenic germline variant of MSH2, one with a variant of uncertain significance (VUS) of MSH2, and the remaining one with somatic variants of POLE (206.2/Mb). In ovarian cancer, three cases were TMB-high. Two of them harbored somatic variants of POLE, which might induce hypermutation genotype, although one of the POLE variants was uncertain for pathogenicity. The remaining one patient with TMB-high (20.4/Mb) was initially diagnosed as double primary cancer (endometrial cancer and ovarian cancer) and harbored a germline VUS of MSH2. We confirmed that her sample from endometrial cancer also exhibited TMB-high with overlapping genotype. Therefore, we amended the diagnosis as endometrial cancer with metastasis to the ovary.

Fig. 3.3
figure 3

Tumor mutational burdens in gynecological cancers. (a) Distribution of the somatic tumor mutational burden (TMB), defined as a number of coding mutations per megabase. The threshold for hypermutation is 10 mutations/Mb. A plot of 206.2/Mb in endometrial cancer is an ultra-mutated case with a variant of POLE

Full size image

3.12 Discussion

Our study confirmed that NGS-based CGP, using the TOP panel, is useful to identify “actionable” mutations in gynecological malignancies. It has been uncertain which types of gynecological cancers are suitable for CGP, especially for the RNA panel. Here, we revealed by our cohort that the TOP panel, a twin-panel system of DNA and RNA, can efficiently detect molecular profiling of gynecological malignancies, including identification of gene fusions. More than 90% of all patients harbored at least one clinically annotated gene alteration and 46% harbored actionable alterations (Tier1 or Tier2). The ratio of actionability is higher than that from the pan-cancer analysis (actionability rate at 32.2%) [6].

Allele-specific analysis of CNV is useful to identify homozygous deletion (and uniparental disomy) of tumor suppressor genes [6, 22]. Another merit of TOP is the RNA panel. The junction-capture method enabled us to accurately and cost-effectively detect hundreds of fusion genes, as well as aberrantly spliced transcripts [6]. The RNA panel can detect novel gene fusions if one of the constituent genes is targeted by the capture panel. Although the clinical actionability of the fusion gene may be limited in gynecological carcinomas [23, 24], detection of ROS1 gene fusion in one cervical cancer suggested that exploring gene fusions may lead to the best-personalized therapy. In addition, the identification of fusion genes using junction-capture RNA sequencing may be useful for molecular diagnosis of sarcomas, which are characterized by various fusion genes [16, 25, 26]. We identified two novel gene fusions in uterine sarcomas in this study. Although these two alterations were not clinically actionable, the existence of a gene fusion was useful to differentially diagnose high-grade endometrial stromal sarcoma from undifferentiated uterine sarcoma.

By using the TOP panel, we constructed the infrastructure of a clinical sequencing laboratory (with ISO15189 certification) within The University of Tokyo Hospital and established an expert annotation team to properly assess the results of sequencing data and provide final reports to each patient. Building this in-house clinical sequencing system should be advantageous to further propel personalized medicine by the cutting-edge technology.

The low rate of clinical trial enrollment after CGP is still a major problem [1, 4, 6, 27]. In this study, only one patient with an oncogenic variant of PIK3CA was enrolled in the clinical of AKT inhibitor, although we included cancer patients regardless of the patients’ status (both nonrecurrent patients and those who receive standardized treatment can be enrolled). Larger numbers of basket-type clinical trials are anticipated to provide more options for personalized therapy. We conducted the prospective TOP analysis as Advanced Medical Care Category B (Japanese medical system: Senshin-iryou B) in 200 patients toward approval from the Ministry of Health, Labor and Welfare from August 2018 (Patient accrual was completed in December 2019) (UMIN000033647). The data of the 200 patients will disclose the ratio of the actionability in patients who (almost) finished all the standardized treatments. We believe that the TOP panel system will accelerate personalized medicine and broaden cancer treatment options for cancer patients in the near future.