Implementing somatic mutation testing in clinical setting: recommendations from a panel of experts.

new tumor classifications as well as to select better tailored therapies for the patients. Some of the genomic markers identified are also prognostic and predictive factors. Additionally, many technologies have been used to investigate these alterations, each with different benefits and caveats. The Genomics Committee from the Sociedade Brasileira de Oncologia Clínica (SBOC) put together a group of specialists, from different regions of Brazil that work both in the private and public scenario, to gather and organize the information regarding the utility of somatic mutation testing in solid tumors. This special article summarizes their recommendations on how to better incorporate this information into clinical practice.


INTRODUCTION
Genomic Medicine is an emerging area of medicine that is characterized by the use of data derived from an individual's DNA and that is part of both his medical follow-up (disease prevention, diagnosis, or therapeutic decisions) and health policies. This concept was adapted from the Genomic Medicine Working Group, a working group developed by the North American NIH (National Institutes of Health) to increase the speed of translating data obtained in the laboratory to clinical practice.
Genomic medicine is part of a broader concept of individualized medicine, called Precision Medicine. The Precision Medicine includes new imaging tests, with or without radionuclides, nanotechnology, and the assessment of non-nucleic biomarkers in body fluids and tissues (https://www.genome.gov/health/ Genomics-and-Medicine).
This consensus refers to somatic mutational panels, investigated in nucleic acids from tumor tissues. Platforms that evaluate gene expression and methylation tests are not discussed here. Analyzes of somatic mutations in circulating tumor DNA can be discussed in an exploratory way, but they are not part of the scope of these recommendations.

Genomic Medicine and Oncology
Nowadays, the oncologist is faced with an enormous amount of data from discoveries about the variants in the individual structure of DNA and RNA molecules. These acquired changes can be associated with different phenotypes in the tumor cell, resulting in changes in the behavior of them. We know that the integration of this knowledge in clinical practice has already started and it is increasing. The fact itself is not surprising, since several steps in the process of fighting cancer can be decoded with greater precision from the data of the tumor genetics.
Cancer epidemiology, prevention, risk, prognosis, and therapeutic decision are already affected by the There has been a rapid increase in the volume of genomic data gathered from different cancers, this has helped to develop new tumor classifications as well as to select better tailored therapies for the patients. Some of the genomic markers identified are also prognostic and predictive factors. Additionally, many technologies have been used to investigate these alterations, each with different benefits and caveats. The Genomics Committee from the Sociedade Brasileira de Oncologia Clínica (SBOC) put together a group of specialists, from different regions of Brazil that work both in the private and public scenario, to gather and organize the information regarding the utility of somatic mutation testing in solid tumors. This special article summarizes their recommendations on how to better incorporate this information into clinical practice.
Despite the rapid progress in the area, there are many challenges to be faced in the coming years: the lack of familiarity with this technology amongst patients and doctors, which can result in resistance to use it; the small number of professionals trained in genomics; the scarcity of specialists and bioinformatics laboratories; the difficulty of access, and the cost of exams. The faster rate that the potentially relevant genetic data are obtained without the previous discoveries having yet been assimilated is another difficult.
The disproportion of data about cataloged genetic variations in populations in some regions of the world in comparison to developing countries, with very diverse ethnicities, is another obstacle that needs to be overcome in the coming years. Finally, a practical aspect still delays the implementation of genomic medicine on large scale in oncology: the evidence is still limited on the efficacy and cost-effectiveness of genomic studies on large scale, except for some mutations clinically validated in a small group of genes. Therefore, SBOC decided to create the genomics in oncology chapter in its committees, as other oncology societies have done in other countries, in an attempt to improve the understanding of the technology available to the clinician today, how it can be used and how to interpret the data obtained.
The following recommendations are based on the best current evidence, divided by subspecialty. Groups of specialists in each of the major areas of oncology included oncologists from different treatment centers for cancer patients, both public and private, from all geographic regions of Brazil. Each group was asked to discuss only methods available in the country and that have a real clinical practical use, according to treatments approved by ANVISA or available in open clinical trials in Brazil. The participants of each group met to answer the following The answers to these questions were organized in the form of topics, with a subsequent discussion session, to put the data into context and discuss briefly their clinical application.
It is important to comment that when the evidence of a test is evaluated and the recommendation is graded considering the quality of the studies, there is a risk of bias inherent to the molecular testing, including the consistency of the results, its precision, and applicability. For a given test to have a strong recommendation, it is strictly necessary that the test has published data and test validation.
Next-generation sequencing panels (NGS) can identify a specific actionable mutation, but with fair outcomes. Thus, although the test may be sensitive, its applicability has less than ideal results. The degree of recommendation depended on the perception of a panel of specialists and the consensus between them. Thus, a moderate or weak recommendation does not indicate that this is a bad test, but that there was no consensus as to the benefit of its use; this certainly varies depending on the environment in which it is performed. A panel of NGS may be strongly recommended in a center with dozens of phase I studies, but if there are scarce treatments and the absence of clinical trials with new therapies, the degree of recommendation will be weak.

EVOLUTION OF COMMERCIAL PANELS
In the last few decades, the identification of targeted therapies has paved the way for precision oncology. In parallel, the rapid development of modern sequencing technology has enabled the expansion of the molecular characterization of cancer and sped up the discovery of actionable mutational processes, in addition to the development of targeted therapies. First introduced in 1977 1 , the Sanger sequencing represented a major advance in determining the sequence of nucleotides in the DNA molecule. The Sanger sequencing allowed important discoveries, such as the oncogenic mutations of RAS in 1982 2 and EGFR in 2004 3 . Likewise, the Human Genome Project 4 was entirely carried out using this platform. However, large-scale projects, such as TCGA (The Cancer Genome Atlas), have used new methods of massive parallel sequencing, which have become known as NGS 5 .
NGS methods offer a wide range of possibilities to characterize the cancer genome. First, the NGS allows the evaluation of hundreds or thousands of genes simultaneously, covering the entire genome or transcriptome within a few days 6 . In addition, the NGS is more sensitive and comprehensive than the Sanger method, as it assesses not only the changes in single nucleotides but also the variation in the number of copies and the multiple structural changes, such as insertions, deletions, and rearrangements, which commonly occur in the cancer genome 6 .
NGS methods have been progressively customized to reduce costs, to prioritize genes associated with cancer and actionable mutations, to be applicable in formalin-fixed paraffin-embedded tissue samples; and to provide results on time to become clinically useful 7 . Targeted gene sequencing panels were first introduced in research centers to accelerate precision oncology and patient inclusion directed by biomarkers in phase I and II trials [7][8][9] . However, commercial panels have emerged as an opportunity for patients and doctors to individualize the use of targeted therapies in clinical practice 10 . Meanwhile, international consensus has been developed to set the recommendations for the use of gene sequencing panels in different scenarios 11 , which further increased the enthusiasm for their use.
Commercially available NGS panels can be classified into two major strategies defined by the target enrichment method: hybridization capture or amplification 12,13 . Target enrichment is the core step in the NGS, as the genes of interest are isolated from the rest of the genome and amplified. This step generates a DNA library, which contains the target regions that will be sequenced and analyzed.
The amplification-based NGS was quickly implemented in clinical practice because it is a simpler process and it has lower cost, shorter delivery time, and its analysis is objective 14 . In addition, local laboratories can adapt in-house panels to cover the gene regions of interest. As a limitation, amplification-based NGS is best suited to cover only DNA hotspot regions and often only to a few dozen genes. On the other hand, hybridization-based NGS (also known as comprehensive genomic profiling) is more easily scalable, it has a high throughput, and represents a more comprehensive strategy, as it allows access to hundreds of genes simultaneously [15][16][17] . This approach may be more accurate to look for all forms of genomic changes and it includes analyzes such as tumor mutational load and determine microsatellite instability. Other differences between the two NGS techniques include the sequencing process and the bioinformatics algorithm. Such particularities have been reviewed in other publications 12 and are beyond the scope of this article.
A crucial step in the implementation of somatic mutation panels is the expertise in molecular pathology and bioinformatics techniques, which must be accompanied by adequate validation for each step of the process 12 . There are several guidelines on this Given the complexity of the NGS panels, the complete process must be validated, from the extraction of the nucleic acid from the biological material to its final analysis. It is important that the validation process includes the evaluation of the quality of the clinical samples in the paraffin blocks; demonstrate the ability to detect different types of genetic changes; define metrics for routine testing and for supplementary testing in genomic regions that are not reliably sequenced. Common actionable variants must be specifically assessed for accuracy and reproducibility. Orthogonal methods must be applied to confirm the results of the validation process. It is noteworthy that changes in the design of the panels require revalidation before the implementation of a new panel in the clinical practice.
The laboratories must inform whether the NGS panels developed in-house have been properly validated, as well as the possible limitations detected during the process. In addition, the proficiency of the test and the laboratory must be certified to ensure that the NGS processes are followed. Examples of internationally recognized proficiency testing programs are the College of American Pathologists (CAP), Clinical Laboratory Improvements Amendments (CLIA), and United Kingdom National External Quality Assessment (UK NEQAS) 19 . Such efforts to confirm standard testing and analysis is also useful to educate laboratories in the proper reporting of clinically relevant variants 20 .
The USA Food and Drug Administration also provides specific mechanisms for the regulatory approval of NGS for the evaluation of genomic profiles in tumors and it has defined levels of evidence that support the actionability and clinical utility of NGS tests. These programs are crucial to ensure high-quality testing and to support national coverage policies or coverage by health insurance companies.

CLINICAL APPLICATION FOR THE DIFFERENTS TYPES OF CANCER TYPE Thoracic Cancer
When should a somatic panel test be requested and for which patients? When is the best time to request?

DISCUSSION
The use of targeted palliative therapy in patients selected according to the molecular profile increases progression-free survival and overall survival when compared to cytotoxic chemotherapy 21 . Controlled studies with targeted therapy included only patients with adenocarcinoma, a predominance of adenocarcinoma, or mixed histology with an adenocarcinoma component. Patients without a definitive diagnosis of adenocarcinoma (for example, carcinoma not otherwise specified, NOS) may have driver mutations (particularly in inadequate/unrepresentative biopsy samples) and should also be treated with targeted therapy in this situation. Driver mutations are also found in patients with squamous histology, with little or no exposure to tobacco (1 to 10 pack-years) 22 .
The decision to test patients with disease at an earlier stage should consider the cost of obtaining such tests in individuals who can be cured (and for which the results will be of little or no use) versus the advantage of knowing the molecular profile earlier if there is cancer recurrence 11,22 . Driver mutations of clinical relevance, in general, are present since the initial diagnosis in patients with less advanced stages and they remain stable throughout the natural history of the disease 23 . The targeted therapy as adjuvant or non-palliative therapy is being studied 24 . The use of adjuvant osimertinib is beginning to be considered as it was associated with increased disease-free survival versus placebo in patients with resected stage II-III NSCLC 25 .
The minimum molecular panel should include assessment of EGFR, ALK, ROS, BRAF, and NTRK1-3. This panel reflects the approval of targeted therapies in Brazil and should be expanded as new drugs demonstrate clinical benefits (Table 1). EGFR inhibitors are recommended as first-line treatment of patients with EGFR sensitivity mutations due to increased progression-free survival, response rate, and quality of life when compared to chemotherapy 26 . Patients with EGFR mutations treated with first or second-generation inhibitors, often develop the secondary T790M resistance mutation at the time of progression. The use of osimertinib in these patients is considered the gold standard of treatment, according to the results of a phase 3 study, which makes the assessment of T790M mutation mandatory in this scenario 27 . The presence of ALK gene rearrangements is associated with high sensitivity to ALK-TKIs (crizotinib, ceritinib, alectinib, brigatinib and lorlatinib) 28 .
Due to the high degree of homology in the tyrosine kinase domains of ALK and ROS, crizotinib showed a high response rate and disease control in patients with ROS1 fusion, being the drug of choice in the firstline treatment of these patients 29 . The presence of a BRAF V600E mutation does not occur with other molecular changes, except for the KRAS mutation, which can coexist with other types of BRAF mutations 30 . The combination of dabrafenib and trametinib for the treatment of patients with the BRAF V600E mu-tation is approved in Brazil and is based on phase 2 studies 31 . Fusion in one of the three TRK receptors confers sensitivity to larotrectinib in lung cancer and other tumors TRK fusion-positive. This is the first approved agnostic therapy in Brazil 32 .
The expanded panel testing can detect genetic changes that have drugs approved for other types of tumors but with activity in lung cancer (for example, HER2 blockade with trastuzumab, afatinib, or T-DM1 in the presence of ERBB2 mutations or HER2 amplification) 33 . It also allows repositioning of approved drugs to other scenarios (for example, use of crizotinib for tumors with MET amplification or exon 14 skipping mutations 34 and the use of drugs already approved in other countries (for example, capmatinib for lung cancer exon 14 skipping mutation of MET 35 and selpercatinib for lung cancer with RET translocation) 36 and inclusion of patients in studies with new drugs 37 .
Often, a hotspot panel is used to test patients for the most common genomic alterations 38 . Sequences of single-gene testing can also be used without a hotspot panel, testing the most frequent alterations initially, and after that the rarest ones. In both cases, running a sequence of single-gene testing is time-consuming and may require a relatively large amount of tissue sample 39 .
Next-generation sequencing (NGS) has emerged as a reliable method to test several alterations simultaneously using a single tissue sample 40 . Computerized models demonstrated that NGS was associated with the same duration to test result (when compared to the hotspot panel) or less (when compared to sequential or exclusion tests) at reduced costs 41 . The use of NGS resulted in the identification of almost 40% more patients with genomic alterations, some with targeted therapies not yet approved by regulatory agencies. These patients could be candidates for clinical trials 42 .
Liquid biopsy in plasma may overcome some limitations of solid tissue biopsy 43 . The circulating tumor DNA may reflect the genetic profile of the tumor, therefore, the possibility to characterize it may have prognostic and therapeutic value 44,45 . Nevertheless, liquid biopsy can fail to detect low levels of circulating tumor DNA (either due to low tumor load or low DNA released by the tumor), which can lead to false-negative results. The agreement of the liquid biopsy with biopsy of the tumor tissue depends on the percentage of tissue changes found in the blood 46,47 , the size of the tumor tissue sample, the timing of the sample, in addition to other factors, such as tumor heterogeneity, treatment interval, and method used 45,48,49 .
In addition to gene sequencing, the molecular profile of lung cancer can be determined by alternative methods, including IHC and FISH for specific genes. The D5F3 and 5A4 antibodies to ALK have a sensitivity of 96% and specificity of 100% for the diagnosis of NSCLC with ALK translocation 50 , and they are considered as standard testing for patients' selection ac-

Implementing somatic mutation testing in clinical setting
Brazilian Journal of Oncology cording to ASCO, ESMO, and NCCN guidelines 11,24,46 . FISH for ALK is also an acceptable method for detecting ALK translocation, but it has disadvantages in comparison with NGS, such as the impossibility of defining the translocation partner and less sensitivity 50 . IHC for ROS may be used as a screening test for tumors with ROS1 translocation. The D4D6 antibody to ROS can label non-neoplastic cells and, therefore, positive results must be confirmed by FISH or NGS 51 . The FISH for ROS1 has a sensitivity of 100% and a specificity of 92% 51 .
Several alternative methodologies can be used to detect NTRK fusions, including FISH and IHQ. The FISH may need up to three probes for a complete analysis 11 . The anti-BRAF V600E monoclonal antibody is commercially available 11  Squamous cell carcinoma of the upper aerodigestive tract 1. Routine somatic panel is not indicated. IHC to assess overexpression of p16 (determine whether HPV positive or negative) in the oropharynx is mandatory for staging and prognostic assessment (type of recommendation: evidence-based; strength of evidence: strong). In metastatic disease, PD-L1 assessment using IHC (CPS) is necessary to define treatment (type of recommendation: based on evidence; strength of evidence: strong).
Nasopharyngeal carcinoma 2. Routine somatic panel is not indicated. IHC or in situ hybridization (preferred) may be performed in the tumor sample to assess etiological correlation with the Epstein-Barr virus (EBV) (type of recommendation: evidence-based; strength of evidence: strong). Salivary gland carcinomas 3. In metastatic disease: 1) IHC or in situ hybridization to assess HER2 overexpression (type of recommendation: evidence-based; strength of evidence: moderate); 2) IHC to assess androgen receptor expression (type of recommendation: evidence-based; strength of evidence: moderate); 3) Assess NTRK1-3 fusion, initially by IHC-pan-TRK and, if positive, confirmation with NGS (type of recommendation: evidence-based; strength of evidence: moderate).

Thyroid cancer
Differentiated thyroid carcinoma 4. Routine somatic panel is not recommended. In metastatic disease, it is recommended: 1) investigation of NTRK1-3 fusion, either by initial IHC pan-TRK screening (EPR 17341 Abcam or Roche / Ventana) and, if positive, confirmation with NGS, or direct testing by NGS (type of recommendation: evidence-based; strength of evidence: moderate); 2) NGS for RET fusion (type of recommendation: evidence-based; strength of evidence: moderate).
Medullary thyroid carcinoma 5. Routine somatic panel is not recommended. It is recommended in advanced or metastatic disease: NGS to assess RET mutations in tumor tissue and to assess germline mutation in suspected cases of multiple endocrine neoplasia type 2 (recommendation type: evidence-based; strength of evidence: strong).
Anaplastic thyroid carcinoma 6. Routine somatic panel is not recommended. It is recommended in locally advanced or metastatic disease: 1) RT-PCR (real-time polymerase chain reaction) or NGS to assess BRAF V600E mutation (type of recommendation: evidence-based; strength of evidence: moderate); 2) IHC for pan-TRK as a screening test and, if positive, confirmation with NGS, preferably RNA (type of recommendation: evidence-based; strength of evidence: moderate). Routine assessment of PD-L1 or TMB (tumor mutational burden) is not recommended.

DISCUSSION
Head and neck carcinoma is a heterogeneous group of epithelial tumors that initiate in the oral cavity, larynx, pharynx (upper aerodigestive pathways), paranasal sinuses, salivary glands, and thyroid.
Head and neck squamous cell carcinoma (HNSCC) HNSCC represents 90% of head and neck tumors. In addition to clinical, phenotypic, and etiologic heterogeneity, these tumors have high molecular heterogeneity 52 . There are currently no predictive genetic markers of response useful to determine the treatment. Therefore, routine panels are not recommended to assess somatic mutations for HNSCC.
P16 protein overexpression by IHC in the sample of oropharyngeal tumors (positivity higher than 70% of the sample) is a marker of the presence of the human papillomavirus (HPV) and it is mandatory for staging and prognostic evaluation 53 . For metastatic HNSCC in cervical lymph nodes of unknown primary site, the presence of p16 (by IHC) should be assessed in the tumor tissue and, if negative, assess the presence of the EBV virus by IHC or in situ hybridization, preferably 53 . In metastatic disease, it is required IHC assessment of PD-L1 expression (22C3 pharmDx), characterized by CPS (combined proportional score).
A phase III study (KEYNOTE 048) compared pembrolizumab alone or in combination with cisplatin and 5-fluorouracil versus the EXTREME regimen (cisplatin, 5-fluorouracil, and cetuximab) 54 . This study showed higher overall survival with isolated pembrolizumab in overall survival in patients with HNSCC with CPS≥1% (12.3 versus 10.3 months, HR = 0.74) and with pembrolizumab associated with cisplatin and 5-fluorouracil, regardless of CPS (13.0 versus 10.7 months, HR = 0.72), compared to the EXTREME arm. Patients with CPS<1% did not show any benefit in comparison to the use of immunotherapy in first line therapy 55 .

Salivary glands cancer
In salivary gland cancer, the assessment of therapeutic targets is recommended in some patients. In the head and neck neoplasms, this would be the scenario in which somatic mutation panels would be closer to use in clinical practice.

Implementing somatic mutation testing in clinical setting
Brazilian Journal of Oncology treatment and the presence of NTRK fusion, firstline treatment with larotrectinib 32 is recommended. Entrectinib is another option in this scenario 67 , although it is not available in Brazil.
HER2 overexpression is present in up to 30% of mucoepidermoid carcinomas or adenocarcinomas not otherwise specified (NOS) 68,69 , and in up to 40% of salivary ductal carcinomas 68,[70][71][72][73][74][75] . Although there is no consensus, most studies evaluate the HER2 expression in breast cancer 76,77 , being considered positive IHC 3+ or 2+ and FISH with ratio HER2/CEP17 ≥2 68,78 . In metastatic disease, treatment with a combination of chemotherapy (taxanes with or without platinum) and trastuzumab is supported by a case series 79,80 and a phase II study 78 . There are also data for double blockade with trastuzumab and pertuzumab 81,82 . Trastuzumab-emtansin (T-DM1) can be used in the second line 83,84 and there are some data on the efficacy of trastuzumab-deruxtecan 85 . The use of anti-HER2 therapy as adjuvant treatment was evaluated only in retrospective studies 79,86 .
IHC androgen receptor (RA) expression, characterized as nuclear, strong, and diffuse (>70%), is observed in most salivary ductal carcinomas [87][88][89][90][91] and adenocarcinomas NOS. The use of combined hormonal blockade, with a GnRH agonist and bicalutamide, has shown efficacy in metastatic disease with RA expression in retrospective studies [92][93][94] and a prospective phase II study 95 . A randomized phase 2 study comparing hormonal blockade and chemotherapy in this scenario is ongoing 96 . The use in the adjuvant setting was evaluated only retrospectively 97 .

Thyroid cancer
In metastatic radiodine-refractory differentiated thyroid carcinoma (DTC), the presence of NTRK fusions is estimated in up to 12% of patients 64 . It is recommended to screen with IHC composed of pan-TRK antibody (EPR 17341 Abcam or Roche / Ventana) 65 , followed by NGS confirmation, with NTRK1, NTRK2, and NTRK3 fusion, preferably based on RNA 66 . For metastatic patients with indication for systemic treatment and presence of NTRK1-3 fusion, it is recommended first-line treatment with larotrectinib -already approved in Brazil. This approval was based on a pooled analysis of three studies with seven individuals with advanced thyroid cancer. There was a 100% response to larotrectinib in this population 32 . RET rearrangement was found in approximately 20% of these patients 98 .
The FDA recently approved selpercatinib for first-line treatment of metastatic radiodine-refractory DTC with RET fusions. This drug showed very encouraging results in a phase 1/2 study. Particularly in DTC, the response rate was 100% when used in the first-line and 79% in other treatment lines 36 . It is recommended to assess RET fusion by NGS 98 .
In medullary thyroid carcinoma (MTC), RET activation is proven to be one of the main mechanisms of oncogenesis. In patients with sporadic MTC, RET somatic mutations are found in approximately 40-60% of patientse 99 . The results of the LIBRETTO-001 study with selpercatinib demonstrated a response rate of 73% in the first-line, particularly in MTC, and 69% in previously treated patients 36 . NGS is indicated to assess RET mutations in tumor tissue 98 .
In anaplastic thyroid carcinoma (ATC), the BRAF V600E mutation is found in 20% to 50% of patients 100 . The combination of dabrafenib and trametinib is already approved for the treatment of patients with locally advanced or metastatic ATC with BRAF V600E mutation without effective locoregional treatment options. A 69% response rate was demonstrated with the combination, being the preferred strategy in this condition 101 . For these patients, it is recommended the assessment of this mutation by RT-PCR or NGS 102 . NTRK fusions are also relevant in ATC.
As previously mentioned, larotrectinib shows encouraging results in this population and the screening of these patients should be performed with IHC composed of pan-TRK antibody (EPR 17341 Abcam or Roche / Ventana) 65, followed by NGS for confirmation 66 .

Gastrointestinal Cancer
When should a somatic panel test be requested and for which patients? When is the best time to request?
1. In the diagnosis of unresectable metastatic or locally advanced disease in esophageal, gastric, pancreatic, biliary, intestinal (small intestine and colorectal) adenocarcinomas (type of recommendation: evidence-based; strength of recommendation: strong); 2. In the diagnosis of the localized esophageal and gastric adenocarcinoma stage II and III (type of recommendation: evidence-based; strength of the recommendation: strong); 3. In the diagnosis of the localized colorectal adenocarcinoma stage II (type of recommendation: evidence-based; strength of the recommendation: strong).
What should be assessed in a somatic mutation panel (which genes should be included and what alterations are expected from each gene)? 4. In metastatic or unresectable adenocarcinomas or poorly differentiated carcinomas: a) Microsatellite instability (MSI) assessment (type of recommendation: evidence-based; strength of recommendation: strong). b) NTRK1-3 fusion assessment (neurotrophic tyrosine receptor kinase) fusion research (recommendation type: evidence-based; recommendation strength: strong).

Esophageal and gastric adenocarcinoma:
a) unresectable metastatic/locally advanced disease: assessment of HER2 overexpression or ERBB2 amplification (Human epidermal growth factor receptor; overexpression or amplification (type of recommendation: evidence-based; strength of recommendation: strong); b) localized disease: MSI (microsatellite instability) assessment (type of recommendation: evidence-based; strength of recommendation: strong).
6. Biliary tract cancer: a) unresectable metastatic / locally advanced disease: FGFR2 (fibroblast growth factor receptor 2) fusion or rearrangement assessment (type of recommendation: formal consensus; strength of recommendation: strong); IDH1 mutation (isocitrate-dehydrogenase) (type of recommendation: evidence-based; strength of recommendation: moderate); BRAF V600E mutation (type of recommendation: formal consensus; strength of the recommendation: strong); ERBB2 amplification (type of recommendation: formal consensus; strength of the recommendation: weak).  11. NTRK: NGS. IHC maybe be used for screening, since it is rapid and less expensive. However, positive cases must be evaluated with sequencing to confirm the fusion;

KRAS/NRAS:
The ideal test is the sequencing of the gene by real-time PCR, which may or may not be part of a panel with other genes. In the absence of the method, it is common and reliable to assess hotspots by simple sequencing, or even by using conventional PCR with restriction enzymes; 13. BRAF V600E: The ideal test is the sequencing of the gene by real-time PCR, whether or not it may be part of a panel with other genes. In the absence of the method, it is common and reliable to assess the most common alterations in hotspots by simple sequencing, or even by using conventional PCR with restriction enzymes; Obs. Due to the cost of each analysis and the time spent, panels by NGS tend to quickly replace the above technologies, addressing all of these genes with deep sequencing simultaneously.

Agnostic alterations
Microsatellite instability (MSI) -Microsatellites are simple sequences (repeats) of nucleotides that occur throughout the genome. Its instability is a marker of mismatch repair (MMR) deficiency, a system composed of four enzymes encoded by the MLH1, MSH2, MSH6, and PMS2 genes, whose dysfunction can be germinative (Lynch syndrome) or somatic, more often associated with epigenetics changes (methylation of the promoter region of the gene). The detection of this alteration in somatic panels by different methodologies (section 1), in addition to screening individuals and families with Lynch syndrome, is associated with a potential response to immunotherapy with immune checkpoint inhibitors (ICI) in several histologies [103][104][105] . This approach was the first agnostic cancer therapy approved in many countries after failure in at least one treatment line.
At the ASCO 2020 plenary sessions, the KEYNOTE-177 study showed benefit in overall survival (OS), progression-free survival (PFS), and response rate in favor of pembrolizumab when compared to chemotherapy chosen by the investigator and monoclonal antibody in the first line of metastatic colorectal cancer 106 . It is also important to cite that the presence of POLE or POLD1 mutations are also associated with a better prognosis in the initial disease and benefit with immunotherapy 107 . In stage II colorectal cancer, the presence of MSI-H was associated with a better prognosis, so that adjuvant chemotherapy is considered ineffective and is not recommended in this scenario. 108 In esophageal and gastric adenocarcinoma, the presence of MSI-H was associated with lack of efficacy of chemotherapy for localized disease in post-hoc analysis of phase III clinical trials 109,110 , and its use is considered controversial, particularly in the perioperative scenario.
NTRK fusions and rearrangements -The NTRK genes encode the tropomyosin receptor kinase (TRK). Fusion of these genes leads to overexpression of fusion proteins with TRK, which results in persistent signaling in different tumors 111 . The activity of NTRK tyrosine kinase inhibitors (TKI) occurs in several histologies 32 .
The use of TRK inhibitor agents should be considered in therapies after the first-line in these patients. Larotrectinib is the drug currently available in Brazil.
Note: In a cohort of 2,314 patients with metastatic colorectal cancer, NTRK alterations were found in patients without mutations, that is, KRAS, NRAS, and BRAF wild-type. Seven of the eight NTRK fusions in this analysis occurred in MSI-H patients 112 . If confirmed, this finding may restrict the metastatic colorectal cancer population for whom the test would be recommended.

Esophageal and gastric cancer
HER2 -The HER2 protein is a transmembrane receptor tyrosine kinase, a member of the epidermal growth factor (EGFR) receptor family and responsi-ble for the regulation of cell proliferation, differentiation, and survival 113 . About 7-20% of esophageal and gastric adenocarcinomas are HER2 positive, that is, they have IHC expression for HER2 3+ or 2+ with positive FISH (fluorescence in situ hybridization), or CISH (chromogenic in situ hybridization) 114 . The positivity rates are similar between European and Asian patients (23.6% vs. 23.9%), but they are higher in the intestinal type than in the diffuse type (31.8% vs. 6.1%), and in esophagogastric junction adenocarcinoma than in gastric tumors (32.2% vs. 21.4%) 115 .
The ToGA study evaluated a humanized monoclonal antibody against HER2 -the trastuzumab -demonstrating better PFS and OS with the trastuzumab and chemotherapy in patients with locally advanced, unresectable, or metastatic gastric or esophageal cancer, HER2 positive, compared to chemotherapy alone 115 . The DESTINY-Gastric01 study was an open-label, phase 2, randomized study that demonstrated that trastuzumab deruxtecan (DS-8201), compared to the treatment chosen by the investigator, increased the response rate and OS in refractory patients in at least two lines of treatment, including trastuzumab 116 .

Biliary Tract Cancer
FGFR -The fibroblast growth factor receptor is a transmembrane receptor with a tyrosine kinase domain, divided into four subtypes (FGFR1-4). Between 6 and 15% of cholangiocarcinomas have some alteration in FGFR gene. Fusion is the most frequent alteration 117 . They are more common in intrahepatic cholangiocarcinomas and, in general, are associated with better prognosis 118 . Pemigatinib is a FGFR1, -2 and, -3 inhibitor drug approved by the FDA in 2020 for use in patients with FGFR2 fusion. The approval is based on the phase 2 study FIGHT-202 that included 107 patients with FGFR2 fusion who had failed at least one previous chemotherapy and were treated with pemigatinib. The response rate and the disease control of the disease were 35% and 88%, respectively, with a median PFS of 6.9 months 119 . Mutated patients showed no benefit from using the drug.
IDH1/2 -IDH1/2 mutations are found in up to 7% of extrahepatic cholangiocarcinomas, and up to 15% of intrahepatic cholangiocarcinomas. Generally, these mutations are mutually exclusive to FGFR alterations. Ivosidenib is a IDH1 inhibitor used for the treatment of leukemia with IDH alteration. The phase III study ClarIDHy randomized 185 patients with chemotherapy failure to receive ivosidenib or placebo. The ivosidenib group had the highest number of patients with stable disease and the highest PFS 121 .
BRAF -BRAF mutations occur in 1-7% of biliary tract cancer, most in intrahepatic cholangiocarcinomas. The most common alteration is V600E mutation 118 . Particularly in patients with BRAF V600E mutation, the dabrafenib-trametinib combination was evaluated in the ROAR study that included in which 35 patients were treated with a response rate of 36% and a PFS of 9.2-month 122 .
HER2 -ERBB2 amplifications occur in 1-3% of intrahepatic cholangiocarcinomas and 10-16% of extrahepatic and gallbladder tumors 118 . There are reports of HER2 overexpression as a mechanism of acquired resistance to FGFR inhibitors. Data on anti-HER2 drugs are limited to a retrospective case series, with response rates of up to 50% for gallbladder tumors treated with trastuzumab, with or without pertuzumab 123 . The largest prospective study on anti-HER2 therapies for biliary tract cancer is MyPathway, in which 11 patients with ERBB2 amplification or mutation were treated with trastuzumab-pertuzumab, resulting in a response rate of 36% and a duration of response of 4.2 months 124 .

Colorectal carcinoma
KRAS, NRAS, and HRAS -The three human RAS genes (KRAS, NRAS, and HRAS) are frequently altered by somatic mutations in several tumors, including colorectal 125 . RAS is a downstream component of the EGFR pathway. Monoclonal anti-EGFR antibodies act by blocking the signal chain and preventing cell proliferation. However, when there are mutations in the genes that encode RAS, this pathway is constantly activated. KRAS is mutated in approximately 40% of colorectal tumors 126,127 , representing an early event in carcinogenesis and it was the first mutation identified as a negative predictive marker concerning to the use of anti-EGFR monoclonal antibodies. NRAS mutation, although less frequent (5-10%), is also a negative predictor of response to anti-EGFR.
Currently, it is being tested together with KRAS and named all-RAS. Cetuximab and panitumumab are monoclonal anti-EGFR antibodies that have shown benefit in patients with wild RAS metastatic colorectal cancer, particularly when the primary tumor is located in the left colon [128][129][130][131][132] .
BRAF -the BRAF gene codes the protein kinase serine-threonine and acts as a downstream effector of RAS signaling, and is a component of the RAS-RAF-MEK-MAPK pathway 133 . The V600E mutation occurs in 8-10% of metastatic colorectal adenocarcinomas. It is more common in tumors on the right side and it is associated with a worse prognosis 134 . Mutations in the RAS and BRAF V600E family genes are mutually exclusive 135 .
Studies have shown that treatment with the triple chemotherapy combination FOLFOXIRI, associated or not with bevacizumab, is superior to double combinations 136 . In patients with BRAF V600E mutation and previous treatment, a randomized phase II study showed benefits with an association of a BRAF inhibitor (vemurafenib), chemotherapy (irinotecan) and anti-EGFR antibody (cetuximab) 137 . More recently, the BEACON phase III study showed benefit increasing OS, PFS, and response rate with the combination of BRAF inhibitor (encorafenib) plus anti-EGFR antibody (cetuximab), with or without MEK inhibitor (binimetinib), in previously treated patients. Subsequently, it was approved by the FDA 138,139 .
The BRAF V600E mutation is also a negative predictor of benefits with anti-EGFR therapy, as shown in a Implementing somatic mutation testing in clinical setting  [148][149][150][151] .
Patients in the POLE and MSI groups are considered to have hot tumors with high neoantigen formation, high TMB and, therefore, are excellent candidates for immunotherapy 146 . Several studies have shown activity of anti-PD1 (e.g. pembrolizumab and dortaslimab) and anti-PD-L1 (e.g. atezolizumab and avelumab) agents with global response rates ranging from 25 to 50% for patients with MSI [151][152][153][154] . In a recent analysis of PORTEC 3, patients with TP53 mutation had benefits with combined treatment with chemoradiotherapy, particularly when compared to radiotherapy alone. In the PORTEC 2 study, patients with a TP53 mutation had better survival when treated with pelvic radiotherapy compared to brachytherapy 155 .
The HCN subgroup presents ERBB2 amplification in approximately 25% of patients 146 . In a randomized phase II study, patients with stage III/IV or relapsed serous carcinoma and HER2 expression in IHC (based on the ASCO/American College of Pathology 2007 guidelines) had better progression-free survival (PFS) and global survival (OS) when trastuzumab was added to carboplatin and paclitaxel 156,157 .
Patients with positive hormone receptor EC appear to be more likely to respond to endocrine therapy. In a randomized study, the response rate observed in patients with positive RE and PR was 25% and 37%, respectively, but the response rate was only 7% to 8% in patients with negative RE/PR disease 158,159 . Hormone therapy is the preferred systemic treatment for patients with grades 1 or 2 RH positive tumors and the absence of rapidly progressive disease 160 .
L1CAM is an adhesion protein that has been recognized as an adverse prognostic marker in EC. In a multicenter study with 1,021 patients with endometrial cancer, L1CAM positive tumors had worse PFS and OS 161 .

Ovarian Cancer
What should be assessed in a somatic mutation panel (which genes should be included and what alterations are expected from each gene)?

DISCUSSION
Pathogenic variants in one of the BRCA genes are found in about 21% of patients with EOC: 14% are germline mutation and 7% are somatic mutation. Defects in DNA repair secondary to deficiency of homologous recombination (HR) pathways are detected in about 50% of patients. HR, particularly by BRCA mutation, is associated with response to PARP inhibitor (PARPi) therapy 162 .
Four randomized studies support the use of this drug class in the first line of patients with EOC. The SOLO-1 study demonstrated the benefit of olaparib maintenance therapy following first line in patients with BRCA1/2 mutation and partial or complete response to platinum. It was observed reduced risk of progression or death by 70% (95% CI: 0.23 -0.41; p <0.0001) 163 . These results supported the approval of olaparib in this scenario in Brazil.

Implementing somatic mutation testing in clinical setting
Brazilian Journal of Oncology The PRIMA, VELIA, and PAOLA-1 studies evaluated, respectively, maintenance with niraparib 164 , veliparibe concomitant with chemotherapy followed by maintenance for three years 165 , and olaparib maintenance for two years in association with bevacizumab 166  In an unselected manner, 10-12% of epithelial ovarian cancer may have dMMR 172 and, although pembrolizumab has not received agnostic approval in Brazil, these patients are potential candidates for immunotherapy.

Breast Cancer (BC)
When should a somatic panel test be requested and for which patients? When is the best time to request?
1. The first-line treatment used in metastatic breast cancer (MBC) is supported by extensive literature. The use of panels may only be considered in patients with MBC who need additional cancer treatment -particularly when standard/registered treatment options are limited. It is crucial that patients and their families understand that somatic panels provide useful results in only a minority of patients.
Often, access to the recommended treatment is very restricted -since most treatments based on these panels are not approved by Brazilian regulatory agencies and, therefore, are not available, either in public health or supplementary health. There is also a very limited number of clinical studies that are based on the results of somatic panels in our country. Families should also understand that, in general, they may need to pay for the panels themselves and any treatment. Therefore, this document should not be used as a justification for requesting health plans or the public health system, whether by judicial or other means.
Another potential use of the panels is the assessment of multiple biomarkers required (for registered treatments) in a single test than the use of individual and sequential testing of a rapidly growing number of biomarkers.
What should be assessed in a somatic mutation panel (which genes should be included and what alterations are expected from each gene)? How? Which platforms or tests are the most appropriate?
2. It is suggested to assess somatic genomic alterations with a strong level of evidence for an intervention, including ERBB2 amplification, activating PIK3CA mutation, microsatellite instability, NRTK1, NRTK2, and NRTK3 fusion, and high mutational tumor burden -TMB). All of these alterations are predictive of benefits in different therapies and are approved by the FDA. In addition, it is suggested to assess BRCA1 or BRCA2 somatic mutation and ERBB2 mutation when they have a weak/moderate level of evidence for modification of clinical management.
There are still few situations in which the identification of specific mutations based on somatic panels leads to the availability of other therapies with documented clinical benefit -in addition to case reports or small case series. In Brazil, where the availability of phase I and II clinical studies is limited, the performance of somatic panels in patients with MBC must be very careful, since it will only rarely expand treatment options outside a research environment.
The MBC treatment continues to be mostly decided based on the evaluation of hormone receptor expression (by IHC), HER2 overexpression (by IHC) and/ or amplification (by ISH -in situ hybridization), identification of PIK3CA activating mutation (by PCR, in 3 hotspots), PD-L1 expression (by IHC), BRCA1, BRCA2, and possibly PALB2 germline mutation (by genomic sequencing of the DNA host in blood or saliva), microsatellites instability (by IHC and/or PCR) and NTRK fusion (PCR or NGS). By the way, many of these tests are provided free of charge by drug manufacturers.
It is important to mention that, for the assessment of the aforementioned genomic alterations, it is possible to perform individual tests, with reliability similar to the use of somatic panels with wide coverage and at a lower cost. However, with the increasing number of targets to be assessed, issues regarding the depletion/circulation of biological material must be considered, since the panels represent a practical al-ternative to testing multiple biomarkers individually. Even for this purpose, it is emphasized that there is still no health insurance coverage in the country.

DISCUSSION
Advanced cancer has gone through changes during the disease and, thus, the genomic profile becomes more complex than that of early BC 173 . Although sequencing is traditionally performed on tumor tissue, which is limited by the availability of the sample and the biopsy risk, the use of tumor DNA sequencing in plasma is an alternative with increasing use 174 .
It is important to assess whether the alteration found corresponds to targeted therapy and whether it results in a clinically relevant antitumor effect.
The identification of genomic alterations related to sensitivity and resistance can help in the selection of treatments for MBC. Although advanced sequencing methods have enabled to detect important genomic alterations, before considering the test, it is essential to determine whether sequencing is clinically recommended and how the results would affect treatment decisions. In addition, the evidence associated with treatment decisions based on genomic alterations discovered in the sequencing needs to be

Type of cancer Gene / Alteration -Comments
All (agnostic approval)

Microsatellite instability (MSI-high)
Microsatellite instability assessment by IHC, RT Table 4. Main somatic alterations in endometrial and ovarian cancer.

Implementing somatic mutation testing in clinical setting
Brazilian Journal of Oncology versus 5.7 months in the arm that received placebo and fulvestrant (HR 0.65; p = 0.00065). There was no difference between groups in the cohort without PIK-3CA mutation 177 .
Microsatellite instability: the incidence in BC is estimated to be around 1% 178 . Tumors with a deficiency in the repair system by unpaired bases recombination are more responsive to PD-1 blockade by pembrolizumab 103 . This drug has received agnostic approval in the USA (regardless of histology), based on the analysis of 149 patients (2 with BC) included in five prospective, single-arm cohort studies. The objective response rate was 39.6%. The two patients with BC had a partial response. Although there are methods of detecting this alteration using NGS, it is important to recognize that the current gold standard for detection is PCR or IHC 179 .
NTRK fusion: Tropomyosin receptor kinase (TRK) family is composed of three transmembrane proteins (TrkA, TrkB, and TrkC), which are coded by the NTRK1, NTRK2 and NTRK3 genes, respectively. Chromosomal alterations that lead to fusions of different genes with NTRK genes determine the transcription of chimeric TRK proteins with kinase function, activated or overexpressed, giving oncogenic potential to these cells 180 . Currently, two TRK inhibitors are approved: Larotrectinib, and entrectinib (not approved in Brazil). Larotrectinib demonstrated effectiveness in the LOXO-101 study. In this study, 55 patients were included and treated with larotrectinib, including one patient with BC (2%). The objective response rate was 75%. After one year, 71% of the patients who had objective response remained with the response and 55% of the patients remained without progression. The median duration of the response and the median PFS have not been reached 32 , however, the frequency of NTRK fusion in the BC is very low; a study that evaluated 12,214 consecutive patients with MBC found that 0.13% of the tumors harbored NTRK fusion 181 . Among the BC subtypes, it is important to note that NTRK fusions are most commonly found in mammary analog secretory carcinoma (carcinoma of the salivary glands) and in secretory breast carcinoma 182 . It is important to note that the NTRK fusion has multiple partners and not all are oncogenic. In addition, NTRK1 G595R, and NTRK3 G623R hotspot mutations are probably associated with resistance to larotrectinib 32 .
High tumor mutational burden: the FDA recently approved the use of pembrolizumab for solid tumors with a high tumor mutational burden 183 . This approval is also considered agnostic and is based on a mutational burden > 10 mutations per megabase (mut / Mb), determined by the FoundationOne CDx somatic panel (Foundation Medicine, Inc.). The approval of this treatment was based on a retrospective analysis of the KEYNOTE-158 study, which included ten cohorts of tumors treated with pembrolizumab 200 mg every three weeks. Tumors with high mutational burden were present in 102 patients (13%). The objective response rate was 29%, the complete response rate was 4%, and the partial response rate was 25%. The median duration of response was not achieved, as 57% of the patients had a duration of response ≥12 months and 50% of the patients had a duration of response ≥24 months. It is important to note that no patient with MBC was included in this analysis. In the MBC scenario, different groups have pointed out a prevalence of a high tumor mutational burden around 10% [184][185][186] . Some preliminary studies suggest a benefit in PFS and OS with the use of checkpoint inhibitors in these patients 187,188 .
Below are mentioned alterations with a weaker level of evidence, but with potential utility, depending on more scientific data: BRCA1/2 somatic mutations: as long as there is robust evidence about the fact that BRCA1 or BRCA2 germline mutations predict benefits in using PARP inhibitors 189,190 or platinum agents 191 , the data about somatic mutations are still preliminary. Recently, Tung and colleagues presented the results of the TBCRC 048 study, a phase II, single-arm study that showed a response rate of 50% with the use of olaparib as monotherapy for patients with somatic mutations in one of these two genes 192 . It should be noted that somatic panels may not capture all BRCA1/2 and PALB2 germline mutations (possibly in 10-20% of patients) 193 , and these patients may respond very well to PARP inhibitors 192 .
ERBB2 mutation: ERBB2 alteration, in addition to amplification, represent up to 20% of the total ERBB2 alterations in these panels (and 2-3% of BC patients), and it is not detected by conventional IHC or FISH 194 . Preliminary data suggest possible response to anti-HER2 therapies 195 .
Despite these potential and uncommon benefits, prospective clinical studies that have attempted to assess the impact of these methods have failed [196][197][198] .
The main justifications are the intratumoral heterogeneity, the lack of effective drugs for most of the molecular targets until now, the heterogeneous patient populations, and the previous and intense treatment of the vast majority of recruited patients. In addition, the studies may have selected patients with tumors that developed several resistance mechanisms.
Considered these data and the aforementioned exceptions, it is concluded that the use of somatic panels in MBC remains largely restricted to clinical research. The ASCO positioning and the ESMO Advanced Breast Cancer (ABC) guidelines 4 affirm that multigene panels should not be used in clinical practice for MBC 199 . However, the somatic panel may be used in prospective molecular screening programs that include patient selection for clinical trials, or as a practical substitute for testing multiple individual markers 199 .

Prostate Cancer
When should a somatic panel test be requested and for which patients? When is the best time to request?

DISCUSSION
Patients with localized prostate cancer can benefit from molecular tests, both for prognostic stratification and treatment selection 200,201 .
Although therapeutic decisions in patients with localized prostate cancer are based on clinical (PSA, clinical stage) and pathological (Gleason score) factors, some molecular tests may help in the management of patients with low-risk disease and in some patients with favorable intermediate-risk 200 The phase III PROFound study evaluated olaparib in patients with CRPC 213 . Patients with deleterious alterations in genes related to the HR pathway and who had disease progression receiving antiandrogenic therapy with abiraterone or enzalutamide were included. Olaparib demonstrated benefit in progression-free survival by imaging (primary outcome), both in cohort A (BRCA1, BRCA2, and ATM alteration) and in the general study population (including other alterations related to the HR pathway). Cohort A patients who received olaparib showed benefit in overall survival, demonstrating that this therapy can increase survival in selected patients 215 . Despite being analyzed as one, each type of genetic alteration in the HR pathway probably is associated with a different benefit from PARP inhibitor olaparib therapy. Each patient must be individualized, weighing the risks and benefits.
Several retrospective series suggest that defects in the DNA repair by HR pathway have also been associated with better responses with the use of radium-223 216,217 and platinum-based chemotherapy 218 .
However, these findings must be interpreted with caution until they are validated in prospective studies.
The benefit of using pembrolizumab in patients with CRPC is derived from this drug as agnostic therapy in patients with defects in the mismatch repair pathway 151,219 , which have alteration in up to 8% of patients with CPRC 207 . Despite preliminary data showing a benefit with the use of PD1 inhibitors in patients with CDK12 mutation 220 , new studies with a greater number of patients did not confirm that the mutation in this gene is a biomarker of response to immunotherapy 218,221 .
Some genes that are included in most commercially available NGS panels may provide prognostic information and information related to resistance to some therapies. Genes such as PI3K, AKT, PTEN, TP53, RB1, CTNNB1, APC, and RNF43 are associated with worse prognosis and resistance to antiandrogenic therapies. Despite being clinically relevant, these data should be interpreted with caution until validated in prospective studies [222][223][224][225] .
From the data exposed above, patients with prostate cancer, at different times of the disease, may benefit from somatic molecular tests for both prognostic information and treatment selection 201 . Thus, currently, the molecular test to be considered in advanced urothelial carcinoma is the evaluation of FGFR-2 and -3, which defines the recommendation for the use of FGFR inhibitor. Other molecular panels can be considered for the inclusion of patients in clinical studies. 3. Tests and platforms available: not applicable.

Solid Hematologic Malignancies
Only the most frequent lymphomas are addressed here with the name solid hematological tumors. There are dozens of lymphoma subtypes. The two main categories of lymphomas are B-cell lymphomas and T-cell lymphomas (not covered in this text). In addition, lymphomas can also be divided between Hodgkin's Lymphomas (HL) and Non-Hodgkin's Lymphomas (NHL). About 90% of lymphomas are NHL and, among them, diffuse large B-cell lymphoma and follicular lymphoma are the most frequent.

Diffuse large B-cell lymphoma (DLBCL)
When should a somatic panel test be requested and for which patients? When is the best time to request?

DISCUSSION
The determination of the cell-of-origin (COO) is considered mandatory by the most recent version of the World Health Organization (WHO). COO can be determined using IHC. Hans's algorithm uses CD10, BCL6, and MUM1 expression, and can distinguish the types of germinal center (GC) and activated/unclassifiable B cell (not CG or ABC) 235 . When compared to gene expression profiling (GEP), Hans' algorithm has an accuracy of about 85-90% 236 . Due to its easy application, low cost, and good correlation, it is indicated for use in clinical practice. The distinction between CG and CBA is important since the last subgroup is associated with the worse prognosis 237 .
More recently, the role of MYC and BCL2 protein expression by IHC has been associated with poor prognosis 238 , regardless of COO 239 . A positive result should be considered when it is greater than 40% for MYC and greater than 50% for BCL2. The scenario in which both are positive, but without genetic translocation, is called "double expressor". Based on the results of COO and MYC and BCL2 expression by IHC, new drugs are being tested combined with standard chemotherapy protocol, R-CHOP, to try to improve the outcomes in this population 240 .
Studies using NGS have demonstrated the difference between the CG and ABC subtypes, in addition to discovering new mutations with prognostic and therapeutic potential. The most frequent alterations found in patients with CG subtype include the BCL2 gene (34%, translocation and mutation), while those with ABC subtype include TNFAIP3 (30%, mutation and deletion), and MYD88 (30%, mutation) 241 . BCL6 translocation (35%) and KMT2D mutation (35%) have similar frequencies in the two COO subtypes. After evaluating almost 600 biopsies of DLBCL 242 , four genetic subgroups were proposed: MCD (MYD88 and CD79B mutation), BN2 (BCL6 fusion and NOTCH2 mutation), N1 (NOTCH1 mutation), and EZB (EZH2 mutation and BCL2 translocation). The BN2 and EZB groups seem to have a better clinical outcome than the MCD and N1 groups.
Alterations in MYC, BCL2, and BCL6 genes may result from different mechanisms 243 . Molecular abnormalities of these genes tend to produce more aggressive phenotypes of the disease, in the case of translocations, than by point or indel mutations. The most traditional method for the assessment these translocations is using in situ hybridization (FISH). Lymphomas that simultaneously host the aforementioned translocations are called Double Hit (MYC + BCL-2) or Triple Hit (MYC + BCL-2 + BCL-6). High-grade B lymphomas with MYC and BCL2 or BCL6 translocation are recognized as a new entity by the recent WHO classification 236 . This group with a worse prognosis has a very poor response to conventional chemotherapy and there is still no consensus on how these patients should be approached 244,245 .
In situations of a scarcity of resources for genetic translocation assessment, it is possible to perform FISH initially only for MYC, reserving BCL2 and BCL6 assessment for situations in which the first one is positive 246 . Another marker with a possible prognostic role is IRF4/MUM1 by IHC. It has already been demonstrated that its expression may be associated with a higher response rate in subtype CBA 247 .

FOLLICULAR LYMPHOMA (FL)
When should a somatic panel test be requested and for which patients? When is the best time to request?
1. Currently, a specific genetic panel is not recommended for patients with FL that helps in the assessment of evolution, therapeutic response, or risk of transformation to aggressive lymphoma (type of recommendation: formal consensus; strength of evidence: strong).
What should be assessed in a somatic mutation panel (which genes should be included and what alterations are expected from each gene)? How? Which platforms or tests are the most appropriate?
1. All patients with FL and uncertain diagnosis should be referred for direct assessment of t(14;18) (q32.3,q21.3), using a specific method (FISH) (type of recommendation: evidence-based; strength of evidence: strong); 2. The routine use of FLIPI-m7 is not recommended as a prognostic tool, since it is not validated in different cohorts of patients with FL (type of recommendation: evidence-based; strength of evidence: moderate); 3. The loss of BCL2 translocation and activation-induced deaminase (AID) overexpression may be a

Implementing somatic mutation testing in clinical setting
Brazilian Journal of Oncology clue for the diagnosis of follicular lymphoma transformed to DLBCL (type of recommendation: informal consensus; strength of evidence: weak).

DISCUSSION
The most characteristic genetic alteration of FL is the BCL2 proto-oncogene translocation with the immunoglobulin heavy chain (IgH) gene locus 248 . The result is a reciprocal translocation t(14;18)(q32.3,q21.3), which results in BCL2 constitutive overexpression and gives an anti-apoptotic effect to the tumor cell 248 . This translocation alone is not able to promote lymphomagenesis, requiring other changes to be added for this process to occur. Healthy individuals can have t(14;18) (q32.3,q21.3) in circulating B lymphocytes without developing the disease 249 . Among the molecular changes, the most important and frequent is the KMT2D (or MLL2) mutation, which occurs in 70-80% of cases 250 . In general, epigenetic changes are often seen in FL.
To gather data related to molecular changes with clinical data in a combined prognostic index, FLI-PI-m7 was developed 251 . This prognostic score integrated the risk factors of the FLIPI (score that uses age, number of nodal sites, LDH value, hemoglobin, and Ann Arbor staging) to the performance status and added seven genes frequently mutated in the FL (EZH2, ARID1A, MEF2B, EP300, FOXO1, CREBBP, and CARD11). FLIPI-m7 was validated in patients treated with R-CHOP or R-CVP, classic first-line regimens in the treatment of the disease, but not in patients exposed to bendamustine or rituximab as monotherapy 252 . Patients with FL and disease progression within the first two years after first-line treatment with R-CHOP (POD24) have a worse prognosis, particularly when compared to patients with progression after two years 253 . In a study that prospectively evaluated FLIPI-m7 in patients with POD24, almost half of the population was classified as low risk, showing that this is not a sensitive tool to identify a group with worse outcomes.
The transformation of the FL into an aggressive lymphoma (histological grade 3B) is an event that occurs in about 10-15% of the cases 254 and represents one of the main causes of mortality related to the disease 255 .
When histological transformation occurs, the morphology resembles a new DLBCL in most cases. This phenomenon seems to be related to activation-induced deaminase (AID) overexpression 256 and loss of BCL2 translocation 236,257 . However, so far there is no combination of mutations that can be used for the diagnosis of this process.

Hodgkin's lymphoma (HL)
When should a somatic panel test be requested and for which patients? When is the best time to request?
1. Currently, no specific genetic panel is recommended for patients with HL to help in the evaluation of the therapeutic response or outcomes (type of recommendation: formal consensus; strength of evidence: strong).

DISCUSSION
There is a close relationship between HL and EBV virus, that is present in tissue samples of lymph node biopsy 258 , as well as an inversely proportional association with common childhood infections, particularly measles, rubella, and mumps 259 . HL can be divided into classic HL and nodular lymphocytic-predominance 236 . The first is the most common type, representing about 90% of cases. The characteristic neoplastic cell is the Reed-Stenberg (RS) cell, which is derived from B cell and it is giant, multinucleated and with an inflammatory infiltrate around it 260 . Considering all tumor tissue, the RS cell represents between 0.1-10% and the other cells are lymphocytes, histiocytes, and peripheral reactional eosinophils 236 . The RS cell has increased expression of ligands 1 and 2 of PD-1 (programmed death-1), PD-L1, and PD-L2, which protects against the mechanisms of death induced by T lymphocytes (immune evasion phenomenon) 261,262 . Variation in the number of chromosomal copies is frequent in HL, particularly of the chromosome 9p24.1 (location of JAK2, PD-L1, and PD-L2 genes), a frequent finding in patients with advanced disease and associated with reduced progression-free survival with chemotherapy [261][262][263] .
The most common genetic alterations in HL lead to changes in the three main signaling pathways: NF-kB (TNFAIP3 mutation in about 40% of patients and more frequently in EBV+ patients), JAK / STAT (SOCS1 and STAT6 mutation in 30-40% of patients) and MHC1 (B2M mutation in up to 70% of patients, particularly in the nodular lymphocytic predominant subtype) 264 . As the number of neoplastic cells in comparison to the tumor tissue is very small, studies of genetic alterations in this lymphoma have always been quite challenging. The use of circulating tumor DNA (ctDNA) is increasing in HL and Italian authors who used this method were able to demonstrate STAT6 mutation in 40% of patientes 265 , in concordance with other studies.

Sarcomas
Sarcomas are rare and heterogeneous malignancies. Soft tissue extremities and retroperitoneal sarcomas, bone tumors, and GIST will be included in this recommendation.
When should a somatic panel test be requested and for which patients? When is the best time to request? What should be assessed in a somatic mutation panel (which genes should be included and what alterations are expected from each gene)? How? Which platforms or tests are the most appropriate?
1. Consider molecular tests for assessment of somatic alterations in patients whose morphology and IHC are not sufficient to establish an accurate diagnosis, or it is necessary prognostic/predictive information. The method to be considered can be guided according to the suspected diagnosis, and availability. Alternatively, broad sequencing of multiple genes using NGS may be considered (type of recommendation: evidence-based evidence; strength of recommendation: strong); 2. Consider NGS sequencing test for the diagnosis, prognosis and therapeutic approach for non-GIST sarcomas (type of recommendation: consensus-based; strength of recommendation: weak); 3. Consider mutational assessment in GIST (genotyping) when planning adjuvant treatment. There is no preferred method (type of recommendation: consensus-based; strength of recommendation: moderate); 4. Consider mutational assessment in GIST (KIT, PDGFRA, NF1, RAS, and SDH) when planning treatment for the metastatic or inoperable disease. There is no preferred method (type of recommendation: consensus-based; strength of recommendation: moderate); 5. If the initial treatment of GIST has not been guided by molecular assessment, PDGFRA D842V assessment must be done in patients with progression of disease or evidence of primary resistance to imatinib (type of recommendation: evidence-based; strength recommendation: strong); 6. ALK translocation in patients diagnosed with inflammatory myofibroblastic tumors may be performed to confirm the diagnosis and guide potential therapy with an ALK inhibitor. It occurs in approximately 50% of IMFT 266

DISCUSSION
The somatic mutations identified in most sarcomas are not driver mutations and, consequently, their assessment will result in limited immediate clinical benefit to the patient. On the other hand, the somatic mutations found at the time of diagnosis are important to detail the histological diagnosis and, subsequently, direct the best therapy. IHC, combined with the histology analyzed by a pathologist specialized in sarcomas, is an accessible method for the classification of the sarcomas. For a diagnostic definition, molecular tests are strongly recommended as auxiliary and complementary methods to IHC 268 . More than 30% of sarcomas have a known translocation 269 . Thus, the use of in situ hybridization (FISH), or even the real-time PCR method to detect these alterations, should be performed in patients whose histological diagnosis cannot be made with precision only by morphology or IHC 270 . An example of the importance of fusion assessment is the identification of several translocations in round cell tumors. EWS translocation is numerous and is increasingly recognized as a prognostic marker in Ewing's sarcomas and Ewing-like sarcomas [270][271][272] .
NGS with a large panel can identify numerous somatic alterations in sarcomas. However, the chance of finding any molecular alteration that is a target for treatment is low. This strategy may be used when there is the possibility of including the patient in clinical studies 273 .
In a study conducted at the MD Anderson Cancer Center with 102 patients with recurrent and metastatic sarcomas, the main alterations found were mutations in TP53 (31%), CDK4 (23%), MDM2 (21%), RB1 (18%), and CDKN2A (13%). Only 14/102 patients had a mutation that was the target of two approved drugs: pazopanib and imatinib. However, these drugs have an off-target effect in PDGFRA, FGFR, and KIT 274 . Therefore, broad sequencing may be used to aid in the histological classification and for the identification of patients for inclusion in clinical studies 275 .
Phase II studies evaluated the efficacy of CDK4 inhibitors (palbociclib) in patients with well-differentiated and dedifferentiated liposarcoma with MDM2/CDK4 amplification. The results show that this strategy results in disease control with promising progression-free survival, but with a low objective response rate 276 .
Crizotinib and ceritinib are ALK inhibitors that have shown activity in patients with inflammatory myofibroblastic tumors (IMFT), with ALK translocation 277,278 .
Patients diagnosed with PEComa and lymphangi-

Implementing somatic mutation testing in clinical setting
Brazilian Journal of Oncology oleiomyomatosis have been treated with mTOR inhibitors with promising results 279 .
On the other hand, patients diagnosed with GIST may have tumor genotyping performed at the time of diagnosis of the localized disease or at the time of the treatment of recurrent or metastatic disease 280 . The most frequent alterations in GIST are KIT and PDG-FRA mutations. In approximately 15% of patients, no type of mutation was found in these two genes (KIT and wild PDGFRA). However, the wild type has been characterized by NF1, BRAF, SDH1, RAS, and NTRK mutations [281][282][283] . The presence of a mutation in exon 11 of the KIT gene is the most frequent and it is related to the increased sensitivity to imatinib in the setting of metastatic disease. Other mutations confer partial or total resistance to imatinib 284 . Prior knowledge of these mutations may better guide the therapeutic approach with alternative drugs, such as sunitinib, in the first mutation line in KIT exon 9 285 .
Recently, avapritinib, which potently inhibits the PDGFRA D842V286 mutation, was approved in the USA. Patients who develop secondary resistance to imatinib acquire new KIT or PDGFRA mutation, and the identification of the mutation may facilitate inclusion in clinical studies.
Molecular changes in NTRK genes are uncommon in adult sarcomas (0.76%) 63 . However, they can occur in more than 70% of patients in childhood fibrosarcoma, a rare disease that affects children, usually under one year of age 287 .

Melanoma
When should a somatic panel test be requested and for which patients? When is the best time to request? 7. Several tests are available to assess the BRAF V600 mutation in melanoma, employing both DNA and antibody analysis. These tests are based on techniques such as real-time polymerase chain reaction (RT-PCR), mutation-specific or single gene, Sanger-type or NGS, pyrosequencing, high-resolution melting, and IHC, the latter using the antibody monoclonal

DISCUSSION
BRAF mutation that constitutively activates the MAPK pathway is present in approximately 40 to 60% of melanomas. In 80 to 90% of patients, this activating mutation consists of the substitution of valine for glutamic acid at codon 600 (mutation V600E), and most of the others consist of an alternative substitution (valine for lysine) (V600K) 288 .
In the adjuvant setting, the phase III study (2) evaluated the use of the combination of dabrafenib plus trametinib versus placebo in 870 patients with a recent diagnosis of completely resected stage III cutaneous melanoma considered to be at high risk of recurrence (lymph node metastases > 1 mm), IIIB or IIIC, BRAF mutation (V600E or V600K). It was observed a statistically significant reduction of 51% in the risk of disease recurrence. After a median follow-up of five years, relapse-free survival was higher with dab-  289 .
In the metastatic scenario, BRAF and MEK inhibitors have also been to delay the development of resistance to treatment and to reduce some toxicities directly associated with BRAF inhibition 290 .
In the phase III COMBI-D study, 423 metastatic patients with BRAF V600E or V600K mutations, treatment-naïve, were randomized to receive dabrafenib plus trametinib, or dabrafenib plus placebo 290 . Progression-free survival (PFS) was significantly higher with the combination than with the use of dabrafenib alone (median 11.0 vs. 8  It is important to note that the three different combinations were not compared to each other in a phase III study. In addition to BRAF mutation, the MAPK pathway can also lead to the development of tumors due to NRAS mutation. Binimetinib is a MEK inhibitor that has been studied particularly in patients with NRAS mutation. A phase III study randomized 402 patients with advanced melanoma and NRAS mutation to receive binimetinib or dacarbazine 296 . PFS was prolonged with binimetinib in comparison to dacarbazine (2.8 vs. For the diagnosis of oligodendroglioma, the presence of the IDH1/2 mutation associated with codeletion 1p/19q is necessary. This is caused by an unbalanced translocation between chromosomes 19 and 1, with total loss of a hybrid chromosome (1p;19q) and loss of heterozygosis. One of the most practical tests for detecting the 1p/19q codeletion is FISH (fluorescence in situ hybridization), although it may result false positive in partial or incomplete deletions. The 1p/19q codeletion is widely recognized as a prognostic and predictive marker, associated with prolonged patient survival, in addition to a better response to chemotherapy 302

NGS for the other alterations
Assessment of KIT mutations in acral or mucosal melanoma Table 9. Main somatic alterations in melanoma.

Implementing somatic mutation testing in clinical setting
Brazilian Journal of Oncology mutation, have activating mutations in the promoter region of the telomerase reverse transcriptase (TERT) gene, however, it is worth remembering that these mutations are also frequent in wild IDH glioblastomas, conferring a worse prognosis in this scenario 303 .
The TP53 mutation is identified in 36-60% of adult gliomas. Loss of ATRX expression is strongly associated with IDH1/2 mutations and was identified in 65-97% of astrocytomas with IDH1/2 mutation. Its agreement with TP53 mutations occurs in 70-94% of cases. It is important to note that the ATRX and TP53 mutations are almost mutually exclusive of the presence of 1p/19q codeletion, therefore, their identification can be used as a screening method 304 .
The diffuse midline glioma with H3K27M mutation is identified as a subgroup with K27M mutation in the histone H3 family gene 3A (H3F3A), or in the H3 histone family gene in cluster 1/B (HIST1H3B/C). It presents a glial phenotype and a diffuse growth pattern, in addition to being located in the midline. Morphology and molecular changes are important for its definition since H3K27M mutations are not exclusive to midline gliomas. This tumor occurs predominantly in young patients, located in the midline (spinal cord, thalamus, brainstem, and cerebellum), and has a worse prognosis, with a two-year survival rate below 10% and a median survival of nine months. The presence of the H3K27M mutation can be demonstrated reliably using IHC 305 .
The understanding of how molecular changes affect the typing and classification of CNS tumors continues to evolve. In addition to the previously mentioned mandatory markers for diagnosis, updates proposed by the Consortium to Inform Molecular and Practical Approaches to CNS Tumor Taxonomy (cIMPACT-NOW consortium) suggest other changes with a potential prognostic impact. According to the proposal, a diffuse wild IDH astrocytoma, with TERT promoter mutation and/or amplification of the epidermal growth factor receptor (EGFR) gene and/or combined gain of chromosome 7, associated with the loss of chromosome 10, would be classified as a grade IV tumor by the WHO. The survival of these patients is similar to that of patients with glioblastoma (with classic histological findings), wild IDH, WHO grade IV 306 .
Likewise, anaplastic astrocytomas (WHO grade III) with IDH mutation should be tested for homozygous CD-KN2A/B deletion and, if present, the tumor should be designated as grade IV. Mitotic activity remains a grading criterion. Microvascular proliferation and necrosis remain criteria for diagnosing grade IV tumors, although these tumors may behave less aggressively, particularly if they do not have a homozygous CDNK2A/B deletion. In neuropathological practice, FISH or high-resolution cytogenetic methods (e.g., array-CGH, SNP arrays, and methylation arrays) may be used to detect homozygous CDKN2A/B deletions 303 .
BRAF alterations may characterize subtypes of gliomas. In adults, BRAF V600E point mutations are present in up to two-thirds of pleomorphic xanthoastrocytomas. Occasionally, diffuse gliomas may also present this alteration, with predictive potential for targeted therapy. A specific antibody is available for detection of the V600E mutation by IHC, but it may also be detected by sequencing 307 .
The TRK family of transmembrane receptors is composed of three proteins, TRKA, TRKB, and TRKC, which are coded, respectively, by the genes NTRK1, NTRK 2, and NTRK 3. These receptors play a crucial role in the development of the nervous system during embryogenesis and remain expressed in neuronal tissue after birth. Pathogenic fusions involving the NTRK genes result in oncoproteins. It is a rare agnostic alteration, but potentially present in any tumor type. In primary CNS tumors, the reported frequency is <1%, but it may be higher in pediatric gliomas 308 . The development of targeted therapy with TRK-inhibitor agents has made it mandatory to assess these alterations in all tumors. Specifically in tissue from CNS tumors, IHC, although available, is inadequate due to the high rate of false-positive results 309,310 .
Current neuro-oncology practice is increasingly dependent on the molecular diagnosis of tumor tissue. The classification of tumors according to histological findings, integrated with molecular findings, has a diagnostic, prognostic, and potentially therapeutic role.

FINAL CONSIDERATIONS IN THE INTERPRETA-TION OF SOMATIC MUTATIONS AND CLINICAL AC-TIONABILITY
In the last decade, somatic panels were quickly adopted for the identification of genomic alterations that could assist in the decision making regarding the selection of targeted therapies and patient management 10,311-315 . Currently, hundreds of laboratories around the world provide results of genomic tests based on somatic panels, generating tens of thousands of reports each year. However, there is still little uniformity as to the mechanisms of analysis of variants and standardization of sequencing reports [316][317][318] .
Despite the development of recommendations for validating NGS 319 tests, many challenges remain in the detection of somatic mutations. Among these challenges is the detection of subclonal or variant in low-purity tumor samples, as well as the distinction of changes in the germline or artifacts related to amplification or sequencing during the polymerase chain reaction (PCR). Direct comparisons of NGS tests developed in laboratories that use different sample processing and sequencing techniques demonstrate disagreement in results, which raises concerns regarding the accuracy of such tests [320][321][322] .
A recent analysis of the in silico algorithms, most commonly used to call mutation, revealed that the existing methods for the detection of somatic mutation can be influenced by factors that generate false-positive and false-negative results 323 . Computational approaches that implement machine learning for direct analysis of raw data from massive parallel sequencing may be useful in minimizing the amount of false-positive calls, optimizing the sensitivity for detecting real alterations in the tumor 323 .
The variants final annotation clinically relevant requires validated computational support for the accurate interpretation of actionable mutations of a given neoplasia 324,325 . On the other hand, it is known that manual curation is an essential part of the process of generating reports and of the public databases of somatic mutations 326,327 . The standardization of the curation process can improve quality control and interoperability between the available databases, facilitating the regulatory approval of these efforts for the clinical interpretation of variants 20,328 . In recent years, two important databases have stood out as public tools for curating somatic variants in cancer: OncoKB and CIViC, developed, respectively, by the Memorial Sloan Kettering Cancer Center 329 , and the Washington University School of Medicine 330 .
The OncoKB (Precision Oncology Knowledge Base) database includes biological, clinical and therapeutic information, curated by resources from unstructured information, including recommendations and guidelines from the FDA (Food and Drug Administration), NCCN (National Comprehensive Cancer Network), and expert groups 329 . Since the clinical implications vary substantially based on the specific alteration in a given gene and the context of the tumor, the in-formation in OncoKB is hierarchically organized by gene, alteration, type of cancer and potential clinical implications. OncoKB information is publicly available through an interactive website (http://oncokb. org/) and incorporated into cBioPortal for Cancer Genomics (http://cbioportal.org/) 331,332 , facilitating the interpretation of complex genomic data for oncologists and cancer investigators. To date, OncoKB has noted more than 5,300 changes in 682 cancer-associated genes in 55 types of cancer.
The CIViC database (Clinical Interpretation of Variants in Cancer) currently contains 7,532 cured interpretations of clinical relevance for 2,622 variants that affect 431 genes (https://civicdb.org/home). These interpretations were selected from 2,737 studies published by 256 CIViC curators 330 . CIViC evidence records are supported by a wide range of levels of evidence, currently focused on somatic alterations and positive associations with response to treatment. At least one evidence record has been created for 309 cancer subtypes and 454 drugs, with most data available for gene actionability in the lung, breast, colorectal cancer, and hematological tumors.
Since the public release of CIViC in June 2015, external curators (not affiliated with the Washington University) have contributed with almost half of all evidence records in this database, indicating the importance of external longitudinal collaborations in curating somatic variants. Like OncoKB, the CIV-iC database can be accessed free of charge without the need for registration or login. Both academic and commercial adoption of these databases should be widely encouraged by oncologists, oncology surgeons, radio-oncologists, pathologists, and investigators in general.