Integrating comprehensive genomic profiling in the management of oncology patients: Applications and challenges in Taiwan

Abstract

Comprehensive genomic profiling (CGP) refers to the detailed genomic analysis of cancers for oncology patients. With the rapid development of next-generation sequencing (NGS) technologies, CGP has been widely applied to clinical practice and managing oncology patients. CGP can be performed on the tumor DNA and RNA, as well as non-tumor tissues (e.g., blood, pleural effusion, and ascites). In this article, we review the current evidence supporting the use of CGP in the management of oncology patients, both in real-world practice and the bridging to clinical trials. We also discuss the role of the molecular tumor board on the application of CGP in oncology patients. We provide an overview of the current scheme of CGP reimbursement in Taiwan and the precision oncology branch of the National Biobank Consortium of Taiwan. Finally, we discuss about the potential barriers and challenges of applying CGP in managing oncology patients and the future perspectives of CGP in precision oncology.

1. Background and introduction

Ever since the development of next-generation sequencing (NGS) and its growing application in scientific research and in clinical practice, the current scheme of managing oncology patients is experiencing a paradigm shift. NGS can provide in-depth analysis of various genomic alterations, including single nucleotide variations (SNVs), small insertions and deletions (indels), somatic copy number alterations, structural variations, and other genomic features [1]. In oncology patients, NGS is a powerful tool to inform genomic alterations, which may help the clinical management and potentially improve the treatment outcome. Comprehensive genomic profiling (CGP) refers to the detailed analysis of a patient's tumor at the genomic level. Compared to traditional molecular testing, NGS-based approaches profile a larger number of genes in a single test, providing a thorough overview of the genomic alterations of the cancer.

In precision oncology, substantial evidence supports the integration of CGP into routine care practice. CGP could identify predictive, prognostic, and diagnostic biomarkers in cancers, which may eventually tailor the management of oncology patients [2]. In this review article, we aim to review and summarize the growing evidence of integrating CGP in the management of oncology patients. We also discuss the role of the molecular tumor board (MTB) and summarize the current scheme of CGP reimbursement in Taiwan. Finally, we discuss some of the challenges and future direction in the application of CGP in precision oncology.

2. Overview of the NGS technologies

NGS is a massively sequencing technique that allows billions of sequencing processes to happen simultaneously in parallel on a solid surface (glass or beads) [3]. Short-read sequencing is the mainstream technology in NGS that generates sequences of less than 1000 bp (typically 150–300 bp) per read. Prior to the sequencing, library preparation is required to obtain templates corresponding to the target of interest for sequencing. Amplicon-based and hybridization capture-based methodologies are the two major approaches employed for target enrichment. Both methods are available in commercialized forms, but the performance may vary. For example, hybridization capture-based enrichment is more sensitive in detecting insertion and deletions (indels) of larger size [4]. On the other hand, amplicon-based enrichment is more sensitive in detecting variants with lower allele fractions [5]. Overall, both methods reach high concordance and accuracy in detecting small genetic variants [4].

Based on the region of genome enriched for sequencing, the NGS application can be categorized into three types: whole-genome sequencing (WGS), whole-exome sequencing (WES), and targeted-panel sequencing (TPS). WGS covers a complete genomic region, and WES captures more than 95 % of the coding regions. In contrast to WGS or WES, TPS often focuses on a panel of genes representing 2–3 % of coding regions [6]. These genes are often highly associated with clinical outcomes or responses to certain therapies. In addition, TPS reduces costs and data burden, but still offers greater sequencing depth and diagnostic power [6]. Currently, TPS is the most common method for CGP in clinical practice. Compared to targeted-panel sequencing, CGP that covers the complete coding mutations and gene fusions (i.e., WES and WGS) is a preferred method for predicting personalized tumor-associated neoantigen [7].

The number of genes included in a single CGP test ranges from 11 to 596 [8]. However, the most optimal size of gene panel covered in one CGP test remains unknown. Comparing to patients with advanced non-small cell lung cancer (NSCLC) who received CGP focusing a smaller gene panel (<53 genes), patients who received CGP with a larger gene panel had improved detection of actionable targets and greater use of matched therapies, both of which were associated with significant increases in survival [9]. However, the additional benefits of CGP with a larger size of gene panel for patients with other types of cancers remain unclear.

Despite the striking achievements of short-read sequencing in genomic profiling, this technology is still underpowered to resolve some genomic regions, including repetitive regions (e.g., centromere and telomere) and complex structural variations [10]. The emergence of long-read sequencing technologies, such as PacBio and Oxford Nanopore Technology, has potentiated the ability to tackle this problem by generating sequences up to tens of thousands or million bp per read [10]. Long-read NGS is also useful in phasing the somatic mutations and genotyping HLA molecules [11,12]. However, long-read NGS suffers from lower base-calling accuracy and much higher sequencing costs, which limits their application in precision oncology [13]. The evidence that supports the use of long-read NGS in clinical practice is scarce. Currently, the use of long-read NGS in precision oncology is primarily in the context of clinical research.

3. Genomic profiling using RNA as sequencing material

Gene fusions, which can be caused by intrachromosomal or interchromosomal rearrangement, may introduce fusion transcripts that translates to chimeric proteins, resulting in signaling activation and cancer promotion [14]. Oncogenic fusions can be found in 16.5 % of cancer cases and function as the sole driver in 1 % of them [14]. Currently, fusions involving ALK, ROS1, NTRK, RET, NRG1, MET, or FGFR genes are tier I clinical actionability across various cancer types [15]. Although these fusion variants can be detected with commercially available DNA-based assays [16], variants involving large introns, repetitive sequences, or unknown fusion partners may be overlooked in some circumstances. For detecting of these fusion variants, RNA-based assays are generally more sensitive than DNA-based assays [17,18].

RNA-based assays are the best method of choice to study variants with alternative splicing, some of which are oncogenic and susceptible to targeted therapies. For example, MET exon-14-skipping mutation is an uncommon genomic alteration in NSCLC that leads to skipping loss of exon 14 and converts to sensitivity to MET-targeted therapies [19]. In NSCLCs harboring MET exon-14-skipping variants, a study found that RNA-based assays are more sensitive than DNA-based assays [20].

However, RNA-based CGP requires high quality of RNA contained in the tissue, which can be suboptimal in some clinical samples. Currently, RNA-based genomic profiling assays are available for clinical use [21]. Overall, RNA-based genomic profiling is more sensitive in detecting gene fusions and alternative splicing variants. This approach may be critical for managing patients with malignancies that harboring these genomic alterations [22]. Table 1 summarizes the features and applications of different NGS assays using DNA or RNA as sequencing material.

Table 1.

Clinical applications of different NGS assays using DNA or RNA as sequencing material.

Assay type	DNA-based	DNA-based	DNA-based	RNA-based
NGS method	Whole-genome sequencing	Whole-exome sequencing	Targeted-panel sequencing	RNA sequencing
Region of human genome covered	>99 %	∼2 %	∼0.05 %	0.01–2 %
Number of coding genes covered	∼20,000	∼20,000	50–1000	20–20,000 (Depend on the assay)
Average read depth per target region	30–60 ×	50–200 ×	500–1000 ×	(Depend on the transcript abundance)
Accuracy in detecting small variants (SNVs and indels)	Good	Good	Excellent	Poor
Accuracy in detecting copy number variation	Good	Intermediate	Poor	Poor
Accuracy in detecting noncoding variants	Excellent	Poor	Poor	Poor
Accuracy in detecting gene translocation	Good	Poor	(Probe-dependent)	Intermediate
Accuracy in detecting large structural variation	Good	Intermediate	Poor	Poor
Accuracy in detecting alternative splicing variants or splice site variants	Good	Intermediate	Poor	Excellent
Sequencing cost	$$$$	$$	$	$$

4. CGP using non-tissue samples

Although tissue-based CGP remains the standard approach to identify actionable genomic aberrations, recent advancement has enabled CGP to be performed on non-tissue samples, such as peripheral blood, pleural effusion, or ascites [23]. Plasma circulating tumor DNA (ctDNA) has been found to represent the genomic alterations of primary tumors, highlighting the feasibility of CGP using ctDNA in clinical practice [24,25]. Published studies showed that the concordance of CGP between ctDNA and tissue samples was highly variable, ranging from 8.3 % to 93 % across different cancer types [23]. Despite the inconsistencies in recapitulating the genomic landscapes, the detection of oncogenic driver variants using cfDNA-based CGP demonstrated great sensitivity, ranging between 75 % and 93 % [23]. Clinical activity of targeted therapies or immunotherapy in patients with advanced NSCLC based on the CGP using ctDNA is currently under investigation [26,27].

Currently, CGP using ctDNA still faces many technical and biological challenges that limit its application in the management of oncology patients. The ctDNA in blood is often fragmented (with an average size of 150–300 bp) and low in concentration (5–10 pg/μL) [28]. This makes the detection of target variants extremely noisy and unstable. Overall, ctDNA detection rate was higher for clonal mutations (75–90 %) than subclonal mutations (20–30 %) [29]. In addition, some cancers are less prone to releasing ctDNA into the circulation, also known as the non-shedding tumors [30]. These include brain, kidney, thyroid, breast, and liver cancers [31]. Studies also reported a lower detection rate of ctDNA in colorectal cancer patients without liver metastasis compared to those with liver metastases [32]. These biological features may as well impact the sensitivity and the application of ctDNA testing in clinical practice.

Malignant pleural effusions and ascites are common complications in patients with advanced cancer. When tumor tissues are not available, these samples can be alternative options for molecular testing or CGP. When applied in malignant pleural effusions, CGP have comparable (65–87 %) concordance of molecular alteration with corresponding tissue biopsies [33,34]. The concordance was highly variant for malignant ascites (25–90 %) [35,36]. Although NGS can be applied to pleural effusion and ascites, more studies are needed to validate the accuracy of detecting clinical actionabilities in cancer patients. Nevertheless, CGP using non-tissue samples is still a valuable diagnostic tool in managing oncology patients.

5. The development and application of CGP in precision oncology

Underpinning precision oncology is the core concept of somatic mutations as the foundation of cancer development [37]. Many tumor-specific and tumor-agnostic mutations are well-proven to be predictive biomarkers of response or resistance to targeted therapies, with novel biomarkers continuously emerging [38]. Variants of non-actionable genes may inform prognosis and predict response to treatment. Conventional tests, such as immunohistochemistry staining, fluorescence in situ hybridization, Sanger sequencing, and quantitative polymerase chain reaction (qPCR), are reliable tools to identify these alterations in daily practice [39]. However, as the number of clinically meaningful biomarkers expands, CGP has greater efficiency and a shorter turn-around time compared to those of traditional testing [40]. Using NSCLC as an example, there are currently 11 genes have been approved as tier I actionable targets [41]. This includes oncogenic mutations in EGFR, ALK, ROS1, KRAS, MET, ERBB2, BRAF, NTRK, and RET genes. To date, there are 53 genes approved as level 1 actionable targets across all cancer types [41]. Some of the tumor-agnostic genomic alterations are rare but yet highly responsive to targeted therapies, such as BRAF^V600E mutations, RET fusions, and NTRK fusions. CGP assay is also frequently used for the diagnosis of complex biomarkers, such as microsatellite instability (MSI), tumor mutational burden (TMB), and homologous recombination deficiency (HRD) [[42], [43], [44]]. Considering the number actionable genes are still expanding, CGP is an ideal diagnostic tool in clinical practice [45].

To evaluate whether CGP should be integrated in managing patients with advanced cancer, several large cohorts were conducted to address this issue. The SHIVA study that included 741 refractory cancer patients found that 40 % of patients had at least 1 druggable molecular alteration [46]. The MOSCATO-01 study that included 1035 patients with advanced cancer demonstrated that 49 % (411 of 843) of adult patients and 61 % (42 of 69) of pediatric patients had actionable molecular alterations [47,48]. The ProfiLER study included 1980 patients with refractory cancers who successfully underwent CGP and showed that 52 % of patients had at least 1 actionable mutation [49]. The Long March Pathway trial that included 520 Chinese cancer patients showed that 22 % of patients had actionable targets [50]. Of note, the frequencies of tumors harboring actionable targets vary across cancer types, with the highest in gastrointestinal stromal tumors, non-small cell lung cancer, and thyroid cancer [51]. In addition, the definition of clinical actionability has expanded substantially in the past few years. An overall increase from 8.9 % to 31.6 % in the fraction of tumors harboring a standard care (level 1 or 2) was observed in a large retrospective study [51]. Although progress has been made in expanding the use of CGP and recognizing clinical actionability, a continued unmet need for innovative therapeutic strategies remains a major issue in precision oncology, particularly for cancers with currently undruggable oncogenic drivers.

6. Evidence supporting the use of CGP in the management of oncology patients

The introduction of CGP allowed the cancer genome to be systematically studied, and guide the therapy in managing oncology patients. In a meta-analysis including 29 publications, more than half of the studies reported favorable outcomes in oncology patients who received NGS-based testing [8]. Among patients who were matched to targeted treatment, progression-free survival (PFS) was significantly longer in 10 of 14 publications across various tumor types (range of HRs 0.24–0.67, with median of 0.47). Overall survival (OS) was significantly longer in 16 of 26 publications (range of HRs 0.34–0.84, with median of 0.47) [8]. In 16 publications reporting outcomes of patients who received NGS testing, only a median of 25 % (range, 2–66 %) were matched to targeted treatment [8]. The reasons for not receiving matched treatment included therapies were not available, physician chose an alternate treatment, disease has progressed, patient declined treatment, and among others. However, these publications reported outcomes from different types of solid tumors, often mixed in a single cohort. The authors concluded that this makes it difficult to aggregate and compare the impact of CGP across various cancer types [8].

To date, clinical guidelines of the National Comprehensive Cancer Network [52], the European Society for Medical Oncology [53,54], and the American Society of Clinical Oncology recommend genomic testing for some cancers [55]. Similarly, the Asia-Pacific Oncology Drug Development Consortium (APODDC) recommends CGP for patients with NSCLC, epithelial ovarian cancer, or prostate cancer [56].

7. Bridging the standard-of-care to clinical trials

With the ability to profile multiple genes in a single test, CGP helps clinicians evaluate investigational therapies that target rare genetic variants [57]. The "umbrella trial" is designed to study more than one therapy for a particular cancer type defined by both pathological and molecular criteria, which is often guided by CGP nowadays [58]. For example, Lung-MAP is a phase I/II clinical trial conducted in patients with squamous cell lung cancer that aimed to individualize treatment based on the CGP results [59]. This is a multi-institutional, multi-cooperative group trial that aimed to rapidly identify new active drugs and bring them to the patients timely.

Conversely, the "basket trial" is designed to study a therapy targeting particular genes for more than one cancer types [58]. Several large basket trials include the NCI-MATCH, TAPUR, MyPathway, and the ProTarget trials [25,[60], [61], [62]]. Molecular targets and pathways that are frequently targeted in the basket trials include AKT, HER2, HER3, BRAF, Hedgehog, EGFR, BRCA, KIT, PDGFRA, and PDGFRB [63]. Although these genomic alterations can be detected with traditional testing, CGP has greatly facilitated the identification of these variants in a single test, regardless of the prevalence in cancer [64]. Based on CGP, the effectiveness and toxicity of 13 off-labeled targeted therapies in advanced cancers are currently investigated in a phase 2 trial [25].

8. Role of molecular tumor board in the integration of CGP

To facilitate the implementation of CGP into clinical practice, the Molecular Tumor Boards (MTBs) have been established to aid frontline physicians and solve the challenges. MTB usually consists of physicians, pathologists, bioinformaticians, pharmacologists, and many other experts [65]. During the MTB meeting, the physician and/or other members of the care team provide clinical information on the patient, and various scientists provide interpretation of genomic and other molecular data derived from the patient [66]. Subsequently, the participants of the board work together to provide matched treatment options in the form of consensus recommendations [66]. The primary aim of MTB is to provide a multidisciplinary approach to potential therapeutic strategies based on the molecular characteristics of a patient's tumor [67].

In the real-world experience, MTB is a valuable tool for matching CGP data with treatment (in-label use or off-label use) or allocating patients to clinical trials [[68], [69], [70]]. In a meta-analysis study, the percentage of patients receiving MTB-recommended targeted therapies ranged from 11 % to 43 % [71]. Although data quality is limited by a lack of prospective randomized controlled trials, the authors concluded that MTBs appear to improve clinical outcomes for patients with cancer [71]. However, the operation of MTBs usually requires substantial resources and time, indicating that MTBs are not easily scalable [66]. For non-academic institutions or community hospitals, centralization of MTBs might decrease the barrier to promoting treatment decision-making [66]. Most urgently, guidelines for MTB implementation and standardized quality requirements are needed to improve the quality of CGP interpretation for oncology patients.

9. Current scheme of CGP reimbursement policy in Taiwan

Starting from May 2024, Taiwan's National Health Insurance (NHI) has granted the reimbursement of NGS-based CGP for patients with solid cancer and leukemia as part of precision medicine for cancer treatment [72]. The fixed-amount reimbursement encompasses tests for either the BRCA gene, a small panel (≤100 genes), or a large panel (>100 genes), and covers nine solid tumor types. The reimbursement policy also mandates the health-care providers to set up MTBs within or across the institutes. This measure is estimated to benefit over 20,000 cancer patients per year, with a budget of approximately 300 million NTD (about 9 million USD).

Prior to the reimbursement of CGP, the National Biobank Consortium of Taiwan (NBCT) was first inaugurated in 2022 under the auspices of the Ministry of Health and Welfare in Taiwan [73]. The precision oncology branch of the NBCT project was designed to expand the repository of CGP data of Taiwanese oncology patients, fortifying the clinical evidence for precision medicine in this population. As of May 2024, the NBCT project has enrolled 2000 oncology patients and granted tissue-based CGP using the FoundationOne CDx assay [74]. As part of the NBCT project, our experience showed that clinical actionability was detected in 21.1 % (96/456) of the patients enrolled in the study [75]. The clinical outcome of patients who received matched therapy was superior to that of patients who did not receive matched therapy (median survival, 26.1 months and 10.6 months, respectively, hazard ratio of 0.28, 95 % confidence interval of 0.14–0.55, p < 0.001) [75]. The NBCT project also aimed to share the genomic and clinical data of all patients through an integrated platform [76]. Ultimately, the goal of the NBCT project is to facilitate the use of CGP in clinical practice, furnish matched therapies, and eventually improve the therapeutic outcomes and quality of life for cancer patients in Taiwan [76].

10. Challenges of implementing CGP in clinical practice

Despite the advancements of NGS technology and rapid reduction of sequencing cost for the human genome in the past two decades, the cost of CGP remains to be the major barrier for most of the patients to access CGP testing in clinical practice. A national policy is an important enabler for oncology patients to access the CGP. Countries including France, Germany, the United Kingdom, and the United States have introduced national policies designed to increase investments in genomics and increase diagnostic capacity [77]. For patients with metastatic NSCLCs, upfront or sequential (after EGFR and ALK tests) CGP testing was associated with substantial cost savings [78,79]. For patients with advanced gastrointestinal stromal tumor, an uncommon solid cancer harboring somatic KIT or PDGFRA mutations, CGP was associated with an increase of 0.10 quality-adjusted life years (QALYs) at a cost of USD$9513 compared with the empirical imatinib approach [80]. Overall, the economic impact of CGP in other types of malignancies is mostly understudied. To improve the accessibility of CGP for oncology patients, a comprehensive evaluation of the economic impact of CGP is critical to facilitating reimbursement across regions. Other key factors that positively affect NGS implementation in clinical practice include the equipment, infrastructure, routine utilization, expert consultation, and educational programs [81].

Variants of uncertain significance (VUS) are poorly characterized variants in the scientific literature that make their biological and clinical significance unclear [82]. VUSs represent the majority of the variants identified by CGP, account for 60–80 % reported variants in driver genes [82]. These variants are often hard to interpret and leave both patients and physicians perplexed. Recently, machine-learning approaches became an appealing method to predict the pathogenicity of VUS. Tools like AlphaMissense, REVEL, and ESM1b demonstrated high accuracy in predicting the pathogenicity of VUS [[83], [84], [85]]. However, the integration of these variant effect prediction tools into CGP is still at a slow pace.

11. Future direction of CGP in precision oncology

To further profile the cancer cells and provide useful information for oncology patients, a multi-omic approach and data integration from genomics, transcriptomics, epigenomics, proteomics, and microbiomics is a rising trend for precision oncology. More recent developments in tumor profiling enable dissection of tumor molecular architecture and the functional phenotype at cellular and subcellular resolution [86]. In particular, spatial omics are the most promising technologies that combine molecular characterization with spatial resolution, providing unique insights into tumor molecular architecture, intratumor heterogeneity, and tumor microenvironment [86]. Spatial transcriptomic studies were able to dissect intra-tumoral heterogeneity, to understand the tumor-stroma interaction, and to identify biomarkers in lung cancer, colorectal cancer, liver cancer, breast cancer, pancreatic cancer, melanoma, and other cancers [87]. However, clinical translation of high-resolution tumor-stroma profiling and integration of multi-omics data into precision oncology still pose significant challenges. Prospective studies and validations are needed to implement the multi-omic tumor profiling into precision oncology.

12. Conclusions

In the management of oncology patients, the clinical utility of CGP will continue to grow, with broader indication and lower costs over time. Currently, CGP can be applied to DNA and RNA as sequencing material, and performed on tumor tissue and non-tumor tissue. This approach is efficient in identifying actionable targets in cancers and may improve outcomes for oncology patients. CGP also helps to bridge the standard of care with clinical trials, providing opportunities for more treatment options for oncology patients. Consequently, the need for MTB will also increase to support and guide clinicians in interpreting and acting on CGP results. Although it still faces challenges in managing oncology patients, CGP with integrated and multi-omic tumor profiling will probably impact precision oncology in the near future.