Molecular profiling of colorectal cancer in a genetically admixed Hispanic population

Abstract Backgorund Colorectal cancer (CRC) is among the leading causes of cancer‐related deaths among Hispanics living in the United States (USH). Understanding the most common carcinogenic molecular pathways that affect Hispanics with CRC is crucial to guide research efforts in developing new therapeutic modalities incorporating genomically diverse populations. Tumor profiling techniques help identify actionable alternatives to recommend treatment and improve survival in cancer patients. Methods We conducted a secondary data analysis to evaluate the mutational profile of 218 CRC tumors in Hispanics living in Puerto Rico (PRH) who underwent next‐generation sequencing (NGS) testing from 2015 to 2020. We compared the prevalence of CRC tumor somatic mutations in PRHs with the mutational profiles reported for CRC from The Cancer Genome Atlas (TCGA) Pan‐Cancer Clinical Data, the AACR Project Genomics Evidence Neoplasia Information Exchange (GENIE)‐Non‐Hispanic, and GENIE‐Hispanic datasets. Results Among the top mutated genes in CRC tumors in PRHs were APC, TP53, and KRAS, which had significantly higher mutational frequencies in PRH compared to the examined datasets, including GENIE‐Hispanics. The most frequent gene amplifications for PRH were CDX2, CDKN1B, and HNRNPA2B1. Targetable biomarkers for CRC, such as microsatellite instability‐high (MSI), wild‐type KRAS, wild‐type NRAS, V600E BRAF, and ERBB2 gene amplifications were found in 2.0%, 43.8%, 97.8%, 3.9%, and 2.3%, respectively, of PRH patients. Conclusion This is the first study to report the mutational profile of CRC tumors in PRHs and make comparisons to other non‐Hispanic and USH populations.


| INTRODUCTION
Colorectal cancer (CRC) is the third most common cancer diagnosed in men and women in the United States (US) and the third leading cause of cancer-related deaths. 1 CRC is a highly heterogeneous disease with different clinicopathological manifestations and clinical outcomes among racial/ ethnic populations. 2 Furthermore, CRC incidence and mortality rates have differed consistently among other racial/ ethnic groups, including Hispanic subpopulations. Among Hispanics living in the mainland United States (USH), CRC is the second and fourth most commonly diagnosed cancer for men and women, respectively, and the third-leading cause of cancer-related deaths for both sexes. 3 However, the term Hispanic aggregates various subpopulations, masking the significant variability within subgroups regarding CRC incidence and mortality. 4 For example, Cubans and Puerto Ricans living in the United States have disproportionately higher CRC incidence and mortality rates than other USH subgroups. 5,6 For Hispanics living in Puerto Rico (PRH), CRC is the second most diagnosed cancer for both sexes and is the leading cause of cancer-related deaths. 7 Overall, PRHs have a higher incidence/mortality rate due to CRC than USHs. 8 The genetic makeup of Hispanics exhibits a complex population structure, arising from more than 500 years of genetic admixture with varied proportions of African, European, and Native American ancestry. 9 Moreover, mutational differences in genes that promote tumor progression among different ethnic groups may be important in CRC heterogeneity observed among Hispanic populations. 2 This suggests that genetic factors may play a significant role in CRC disparities, underscoring the need to characterize the molecular features of admixed populations to develop effective targeted therapies to impact clinical outcomes. This study describes the mutational profile of CRC tumors from PRH, the second-largest Hispanic group in the United States. 10 Additionally, we compared the mutational frequency of actionable and driver genes in CRC tumorigenesis for PRH with other non-Hispanic and USH populations. Together, these datasets highlight the molecular mechanisms underpinning these biological differences.
Most CRC cases arise from somatic mutations, but approximately 5%-10% are due to germline mutations. 11 Sporadic CRC tumors arise heterogeneously from the accumulation of genetic and epigenetic alterations leading to the transformation of adenomas into adenocarcinomas. A multi-step sequence for CRC tumorigenesis has been proposed where alterations in genes involved in cellular pathways, such as EGFR (ERBB1/HER1), MAPK, PI3K, TGF-β, and WNT/β-catenin signaling pathways, occur at different stages of tumor progression. 12,13 Descriptive information regarding the genetic mutational frequency of colorectal tumors among Hispanics using next-generation sequencing (NGS) technologies is limited. We present for the first time a comprehensive somatic molecular characterization of CRC from PRH and compared it to three independent cohorts of patients with CRC from The Cancer Genome Atlas (TCGA) dataset 14 and AACR Project Genomics Evidence Neoplasia Information Exchange Non-Hispanics (GENIE-NH) and Hispanics (GENIE-H) datasets. 15 By evaluating colorectal tumors from molecularly characterized ethnically diverse cohorts, we hope to gain a better understanding of the common carcinogenic molecular pathways that affect our Hispanic population. The characterization of the oncogenic drivers in Hispanics will inform clinicians, scientists, and health policy stakeholders about actionable mutations to guide research, treatment, and health policy efforts.

| METHODOLOGY
Using a cross-sectional design, we evaluated CRC tumors' mutational profiles from 218 PRH that underwent NGS testing from 2015 to 2020. The data were provided by the Precision Oncology Alliance (POA) (https://www.caris lifes cienc es.com/ colla borat ion/) using the CARIS Life Sciences NGS platform, which uses two-gene panel versions (one containing 592 genes and the other containing 54 genes) to detect mutations, indels, and copy number amplification (CNA). In some samples, genes were ruled indeterminate due to the low coverage of some or all exons. The CNA is calculated using the average sequencing depth of the sample and the sequencing depth of each exon and comparing this result to a pre-calibrated value. The resulting categories were: (1) Amplification: all exons within the gene of interest have an average of ≥3 copies, and the average copy number of the entire gene is ≥6 copies; (2) Intermediate: an average of ≥4 but <6 copies of a gene were detected, or if the average copy number of the gene is ≥6 copies, but contains exons with an average of <3 copies; (3) No amplification detected: an average of <4 copies of a gene are detected. No clinicopathological data were available for the mutational data provided by the POA. The age and sex data provided cannot be correlated with the mutational data.
To determine the actionable mutations, the OncoKB (https://www.oncokb.org/levels) therapeutic levels of evidence 1, 2, and R1 were considered: (1) FDA-recognized biomarker predictive response to an FDA-approved drug; (2) Standard care biomarker recommended by the NCCN or other professional guidelines predictive of response to an FDA-approved drug; and (R1) Standard care biomarker predictive of resistance to an FDA-approved drug. Additionally, to classify the gene alterations found for PRH as oncogenic, likely oncogenic, or unknown, we used the OncoKB prior knowledge about specific variants, which contains information about the oncogenic effects and treatment implications or variants. 16 My Cancer Genome website was used to identify the clinical impact of the observed biomarkers for CRC tumors in PRH and to identify clinical trials that used the observed biomarkers as inclusion criteria for CRC or solid tumors. The information was obtained from FDA labels, the NCCN, and clinical trials, among other scientific platforms.
We estimated the prevalence of somatic mutations in PRH CRC tumors and compared them with the mutational profiles reported for CRC tumors from the TCGA Pan-Cancer Atlas Clinical Data and CRC tumors from the AACR Project GENIE (public release version 9.0), both available in the cBioPortal for Cancer Genomics. The mutational data reported for the TCGA cohort was based on exome-sequencing and Affymetrix 6.0 microarrays, 14 while for the GENIE cohort, the data reported was based on a hybridization-based or amplicon-based sequencing approach according to each participating center. 15 The GENIE database had 724 cases of Hispanics; these cases were taken as an independent subset to compare PRH with the Hispanics in GENIE (GENIE-H) and the 9427 cases available for GENIE-Non-Hispanics (GENIE-NH).

| Statistical analysis
We used descriptive statistics to characterize the datasets. For descriptive purposes, the age variable was dichotomized into <50 and ≥50. However, the mutational data for CRC in PRH were not stratified by age. This variable was reported as age at diagnosis for PRH and TCGA; meanwhile, for GENIE, it represented the age at which sequencing was performed. The equality of the proportions between our study population and the TCGA, GENIE-NH, and GENIE-H were evaluated using the two-sample proportion test (prtesti); a p-value less than 0.05 was considered statistically significant or marginally significant (p-value = 0.05). All statistical analyses were conducted using Stata version 17 for Microsoft Windows (StataCorp LLC). Lollipop graphs using the cBioPortal Mutation Mapper (https://www.cbiop ortal.org/mutat ion_mapper) and the GraphPad software were used to visualize t genes and frequency of mutations impacting our Hispanic study population. Table 1 compares sex, age at which sequencing was reported, and ethnicity for the PRH, TCGA, and GENIE cohorts. Of the 218 PRH cases, 24.0% had a CRC diagnosis before 50 years of age (early-onset CRC) since these patients were <50 years old when sequencing was performed. For the PRH population, males accounted for more than half of the sampled population (55.5%), similar to the distribution among the TCGA (52.5%), GENIE-NH (54.6%), and GENIE-H (56.1%) datasets. For TCGA CRC tumors, 13.3% were diagnosed before 50 years of age. Like PRH, the GENIE-NH (27.0%) and GENIE-H (28.5%) datasets have a higher percentage of sequenced individuals younger than 50.

| Comparison of the population characteristics of PRH, TCGA, GENIE-NH, and GENIE-H datasets
The following sections describe only genes and mutations with a statistically significant difference (p < 0.05) when comparing PRH versus TCGA, GENIE-NH, and GENIE-H datasets.

| Mutational profile of CRC tumors from PRH: Top mutated and actionable genes and comparison with TCGA and GENIE datasets
The most commonly mutated genes among CRC tumors for PRH were APC, TP53, KRAS, PIK3CA, SMAD4, AMER1, FBXW7, BRAF, and ARID1A. Table 2 includes the top mutated genes in CRC tumors from PRH, along with their most frequent gene-specific alterations alongside those reported in the TCGA, GENIE-NH, and GENIE-H datasets. Figure 1 shows the frequency of the most common genetic alterations found in CRC tumors for PRH leading to deregulation of the WNT/β-catenin, MAPK, PI3K, TGF-β, and p53 signaling pathways. Figure 2 shows the mutational frequencies of these genes for PRH compared to other datasets. Only 2.0% of the 150 PRH-profiled tumors were microsatellite instability-high (MSI-H). A comparison between the frequency of MSI tumors from PRH and the TCGA and both GENIE datasets was not performed since this information was not available for these datasets. In this section, we will discuss only those genes that had significant differences when comparing the mutational frequencies of PR with other datasets.

| APC
The APC mutational frequency for PRH was 9.4% higher than TCGA, 18.1% higher than GENIE-NH, and 16.5% higher than GENIE-H. In terms of specific gene alterations, 134 unique gene alterations were reported for the APC gene: 132 (98.5%) were truncating mutations (including 68 frameshift and 64 nonsense mutations) and two splice variants (1.5%). Figure 3 shows the mutational distribution of APC for PRH. The most frequent APC gene alteration for PRH was E1309*-8.2%, 8.0%, and 8.1% higher than that reported for TCGA, GENIE-NH, and GENIE-H, respectively. The second most frequent gene alteration for PRH

| TP53
PRH had the highest TP53 mutational frequency among the compared datasets: 16.5% higher than TCGA, 7.7% higher than GENIE-NH, and 7.4% higher than GENIE-H. A total of 88 unique gene alterations were reported for the TP53, of which 51 (58.0%) were missense mutations, 28 (31.8%) truncating mutations (including 15 frameshift and 13 nonsense mutations), 7 (8.0%) splice variant mutations, and 2 (2.3%) insertion/deletion mutations (see Figure 3). In TP53, R175H and R282W were the most frequent variants for PRH, which match the frequencies in the TCGA, GENIE-NH, and GENIE-H datasets. G245S was the third most frequent alteration among PRH, with a frequency 2.8% higher than that reported for TCGA. PRH had a 1.6% higher frequency of R342* than that reported for GENIE-H and a higher frequency of P278S compared to other datasets.

| KRAS (actionable gene for CRC)
The mutational frequency of KRAS for PRH was 15.4% higher than TCGA, 13.0% higher than GENIE-NH, and 9.5% higher than GENIE-H. For PRH, 15 different missense mutations were reported for the KRAS gene. Figure 3 shows the frequencies of KRAS alterations. In detail, KRAS G12D, G13D, G12C, and G12V were the most frequent gene alterations for PRH. Compared to other datasets, PRH had a 6.1% and 5.7% higher frequency of G13D than that reported for TCGA and GENIE-NH, respectively, and a 4.2%, 4.0%, and 4.2% higher frequency of G12C than that reported for TCGA, GENIE-NH, and GENIE-H, respectively. PRH also had a 2.0%, 1.6%, and 1.9% higher frequency of KRAS Q61H when compared to TCGA, GENIE-NH, and GENIE-H, respectively.

| PIK3CA
The mutational frequency for PIK3CA was 9.6% lower for PRH when compared to TCGA. For PRH, a total of 13 unique gene alterations were reported for the PIK3CA gene, of which 11 (84.6%) were missense mutations and 2 (15.4%) were insertion/deletion mutations (see Figure 3). For PRH, E545K was the most frequent alteration for PIK3CA, having a similar frequency across all the compared datasets.

| FBXW7
The mutational frequency for FBXW7 was 7.8% lower for PRH when compared to that reported for TCGA, and it was similar to the GENIE and GENIE-H datasets. A total of 12 different gene alterations in FBXW7 were reported for PRH, of which 6 (50%) were truncating mutations (including one frameshift and five nonsense mutations), 3 (25.0%) missense mutations, and 3 (25.0%) intronic variants. The most frequent alterations reported for PRH were R456C, R505C, and R4965H, which had similar mutational frequencies compared to other datasets.

| BRAF (actionable gene for CRC)
For PRH, BRAF mutational frequency was 6.6% and 5.7% lower than TCGA and GENIE-NH, respectively, which is comparable to GENIE-H. This gene has three different gene alterations for PRH: V600E, D594G, and F595L. The frequencies of these alterations among the other datasets were similar, except for F595L, which was significantly lower for PRH when compared to that reported for GENIE-NH.

| NRAS (actionable gene for CRC)
PRH had a 4.0% lower mutational frequency for NRAS when compared to that reported for TCGA. For PRH, only three different missense mutations were reported: G13D, Q61K, and Q61L. Additionally, G13D frequency was 1% higher than that reported for GENIE-NH.

| ERBB2
For PRH, ERBB2 mutational frequency was 3.2%, 2.9%, and 4.1% lower than those reported for TCGA, GENIE-NH, No mutations were found for these genes among CRC tumors from PRH. On the contrary, mutations for these genes were reported for TCGA, GENIE-NH, and GENIE-H.

| Copy number variants: Gene amplification
The most common gene amplifications found for PRH CRC tumors using NGS are presented in Table 3. For PRH, the most frequent gene amplifications were: CDX2, which was 13.7% higher than TCGA; CDKN1B, which was 6.3% and 6.2% higher than GENIE-NH and GENIE-H, respectively; and HNRNPA2B1, which was 5.1% higher than TCGA.
Other gene amplifications significantly different between PRH and other datasets were: ZNF703, MNX1, ARFRP1, FGF3, and RARA. The amplification of the ERBB2 gene (an actionable mutation for CRC) was similar among all the compared datasets.

| DISCUSSION
Molecular characterization of CRC tumors has resulted in the identification of genetic alterations in cancer driver genes, such as KRAS, NRAS, BRAF, and ERBB2, which are now used to guide first-line therapies. 17 The introduction of targeted therapies, such as monoclonal antibodies (mAb) and small molecules to inhibit tyrosine kinases, has improved overall survival in metastatic CRC (mCRC). 17 Despite the development of biomarker-based stratified treatment, there are still marked survival disparities among different racial and/or ethnic groups, including Hispanics, 8 underscoring the necessity to identify new predictive and prognostic biomarkers for the development of effective treatments. In this study, we analyzed NGS data from PRH CRC tumors and provided a detailed description of their molecular profile. We also compared the mutational profiles reported for CRC tumors from other national and international datasets. Here we only discuss actionable genes and driver genes with overall mutational frequencies significantly different in PRH compared to other datasets. Our goal was to identify possible alterations that may be clinically significant in Hispanics.
Microsatellite instability-high (MSI-H) tumors were less frequent in PRH compared to previously reported for African Americans (12%), Hispanics (12%), and Caucasians (14%) in the United States. 18  with what was previously reported by our group, and only 4.3% of tumors from PRH had negative MMR-protein expression. 19 According to the literature, MSI varies depending on CRC stage. Approximately 12-20% of MSI has been observed in the early stages, whereas <5% incidence of MSI has been observed in advanced stages. 20 Tumor stage information was unavailable for CRC tumors from PRH that were used in this study, and no assumptions could be made about tumor stage and MSI status. Currently, MSI-H status is a biomarker predictive of response to anti-PD-1/ PD-L1 therapy (see Table 4). 21 Significant differences in the mutational frequencies of CRC driver genes, such as APC, TP53, PIK3CA, FBXW7, CDX2, CDKN1B, and HNRNPA2B1, were reported for PRH when compared to the datasets examined. Most of these driver genes are members of the WNT/β-catenin, MAPK, PI3K, TGF-β, and p53 pathways (see Figure 1). PRH had the highest mutational frequency for APC, a tumor suppressor gene involved in the regulation of the WNT/β-catenin signaling pathway, and mutations in this gene occur as an early event in CRC tumorigenesis. 22 Previous studies showed lower APC mutational frequency for Caucasians (15.5%), Asians (25.2%), and African Americans (30.9%). 23 However, the mutational frequency of APC for PRH was similar to the Brazilian population (71.4%). 24 The marked differences in APC mutational frequency among populations may reflect on the molecular pathways that lead to tumor progression, suggesting that for PRH, the most common one is the CIN pathway. Even though APC is not an actionable gene for CRC treatment, in vitro studies have shown that small-molecule inhibitors targeting APC truncated proteins either restore the WNT/β-catenin signaling pathway or cause cytotoxicity. 25,26 Similarly, PRH had the highest TP53 mutational frequency among the compared datasets. TP53 tumor suppressor gene mutations occur as late events in CRC tumorigenesis and are mostly associated with metastasis and poor overall survival. 27 Considering the high mutational burden of APC and TP53 for PRH, this population will benefit greatly from targeted therapies against mutations in these genes. Moreover, ongoing clinical trials (see Table S1) evaluating new treatment approaches based on these biomarkers could result in new biomarker-based therapies that will benefit Hispanic populations with a high mutational burden for these genes.
In this study, the mutational frequencies of the PIK3CA and FBXW7 genes were lower for PRH. However, PRH has a higher mutational frequency than those previously reported for other minority groups, including African Americans (13%) and Asian/Pacific Islanders (10.0%). 28 PIK3CA mutations play a key role in CRC pathogenesis, with an incidence of 20% in CRCs. 17,22 For PRH, most of the mutations were found in PIK3CA mutational hot spots, located in exons 9 and 20, which have a distinct molecular impact on CRC: exon 9 mutations are associated with KRAS mutations, and exon 20 mutations are associated with KRAS, BRAF mutations, and MSI. 29 PIK3CA mutations may predict resistance to first-line chemotherapy, anti-EGFR therapy, and response to radioembolization therapy in mCRCs. 30 Even though contradictory findings assessing the prognostic value of PIK3CA mutations in CRC, targeting the PI3K/AKT1 pathway has emerged as a therapeutic alternative for CRC. 31 Mutations in FBXW7, a tumor suppressor gene that downregulates transcriptional activators involved in cell growth, are found in approximately 6%-10% of CRC. 32 Previous studies have shown that mCRC patients with FBXW7 mutations have a lower overall survival rate than those with the wild-type gene. 32 The mutational frequency of FBXW7 reported for PRH in this study is consistent with those reported for Asians (6.5%), African-Americans (8.0%), Whites (7.6%), and Hispanics (7.4%). 32 Even though multiple studies describe the predictive/prognostic value of these biomarkers in CRC-targeted therapies and overall survival, there is no significant evidence to recommend selected treatments for mutations in PIK3CA and FBXW7. However, clinical trials use these biomarkers as inclusion criteria; see Table S1.
PRH had higher mutational frequencies for gene amplification in CDX2, CDKN1B, and HNRNPA2B1. A reduced expression of CDX2, a transcriptional regulator of intestinal cell lineage observed in approximately 20% of CRCs, is associated with poor outcomes and has been identified as a prognostic biomarker in stage II/III CRC. 33 In this study, we reported that PRH had a three times higher frequency for CDX2 gene amplification when compared to TCGA. CDX2 amplification is associated with the activation of the Wnt/β-catenin signaling pathway, playing a role as a lineage-survival oncogene in CRC. 34 The role of CDX2 expression in CRC may rely on the underlined molecular pathways leading to carcinogenesis, 34 which implies that CRC tumors from PRH are molecularly distinct from other populations. PRH had six and five times more CDKN1B and HNRNPA2B1 gene amplification, respectively, compared to the other datasets. For CRC, it has been reported that the absence or reduction of CDKN1B expression is associated with a poor prognosis. 35 In contrast, CDKN1B gene amplification is associated with poor prognosis for gastric carcinomas. 36 High CDKN1B expression predicts sensitivity to hormone therapies and chemotherapy in luminal breast cancer patients, while its downregulation predicts resistance to radiotherapy and anti-ERBB2 therapies. 37 HNRNPA2B1 amplification correlates with a higher gene expression observed in several cancers and is associated with tumor progression and poor prognosis. 38 Similarly, a recent study reported that high expression of HNRNPA2B1 in late-stage CRC (stages III and IV) was -GRT-C903 and GRT-R904: are neoantigen cancer vaccines that stimulates host immunity to mount specific cytotoxic T-lymphocyte response against tumor cells.
Tyrosine kinase inhibitors-inhibition of EGFR and ERBB2/HER2 signaling pathways lead to tumor cell death, inhibition of cell growth, angiogenesis, and metastasis.
-Pertuzumab: mAb that targets subdomain II of ERBB2/HER2.  associated with tumor progression, metastasis, and poor prognosis. 39 Further studies are needed to investigate the role of these gene amplifications in CRC carcinogenesis and how they correlate with the high mortality rates observed in PRH.
We also compared the mutational frequencies of actionable genes for CRC and reported significant differences for PRH in KRAS/NRAS, BRAF, and ERBB2 amplification. The high mutational frequency of KRAS reported for PRH differs from a previous study with PRH that found KRAS mutations in 39% of CRC tumors. 40 However, a similar KRAS mutational frequency was reported for USH (59%), 41 while a higher frequency was reported for Mexicans (86%). 42 In contrast, lower mutational frequencies have been reported for Caucasians (38%), 41 Asians (37%), 41 and African Americans (44.4%) 43 in the United States, for Chileans (37%), 44 and Brazilians (52%). 24 CRC patients with tumors harboring KRAS/NRAS mutations do not respond to anti-EGFR mAb therapy and have lower overall survival when compared to those with wild-type tumors. 45 Table 4 shows the current biomarkers used to guide CRC treatment, indicating that only patients with wild-type KRAS/NRAS have the clinical benefit of anti-EGFR mAb therapies. For PRH, 43.8% had a wild-type KRAS status, whereas higher frequencies were observed for the TCGA, GENIE-NH, and GENIE-H (59.2%, 56.8%, and 53.3%, respectively). The majority of the missense mutations found in KRAS for PRH occur at exon 2 at codons 12 and 13, followed by exon 3 at codon 61 and exon 4 at codon 146. Our findings correspond with previous reports that the most common mutations for KRAS are found in exon 2 (70%-80% at codon 12 and 15%-20% at codon 13 46 ), and least frequently in exon 3 at codon 61, all resulting in constitutively active GTPase increasing the downstream signaling pathway. 47 PRH had a significantly higher frequency for the specific oncogenic mutations G12C, G13D, and Q61H when compared to the other datasets. Even though there is no actual treatment for CRC cases harboring KRAS mutations, several drugs have entered clinical trials targeting mutations at codons 12, 13, and 61 for mCRC. Some of these clinical trials are listed in Table 4. Recently, a drug called sotarasib, which targets the KRAS G12C mutant, received the FDA approval for treatment in non-small cell lung cancer (NSCLC), and several investigational clinical trials are evaluating this drug for the treatment of mCRC patients harboring KRAS G12C and other mutations at codons 12 and 13. These clinical trials show promising benefits for PRH, who have significantly higher mutational frequencies of KRAS codon 12 and 13 mutations when compared to other populations. The high mutational burden of KRAS for PRH translates to a higher proportion of patients who cannot benefit from anti-EGFR mAb therapies, could be correlated to the higher mortality rates observed for this population.
PRH had a lower BRAF mutational frequency when compared to TCGA and GENIE-NH as well as to those previously reported for Hispanics (6%), 41 Whites (12%), 41 Asians (6%), 41 and African Americans (6.4%) 48 in the United States, and for Mexicans (6%). 42 Even though BRAF and KRAS mutations usually occur in a mutually exclusive manner, 49 the biological consequence of these mutations is similar since both genes are part of the EGFR/MAPK signaling pathway. Patients harboring BRAF mutations have poorer outcomes when compared to their wild-type counterparts, especially those carrying the V600E mutation (90% of BRAF alterations). 50 Several studies have reported variable BRAF V600E mutational frequency (0%-15%) among Hispanics from Central and South America, with an average of 7.8%, 51 higher than reported in this study for PRH (3.9%). Small molecule inhibitors targeting BRAF V600E in combination with anti-EGFR targeted therapies have been demonstrated to improve overall survival in patients with mCRC carrying this mutation. 52 As a result (see Table 4), the use of encorafenib, a BRAF kinase inhibitor, in combination with cetuximab (a mAb targeting EGFR), has been approved to treat mCRCs with BRAF V600E.
The frequency of ERBB2 amplification for PRH was similar to that reported for the other studied datasets. ERBB2 gene amplification represents 3%-4% of mCRCs and is associated with poor survival. 53 The ERBB2 V777L missense mutation was detected in only 0.6% of CRC tumors from PRH, which was lower compared to other datasets. Upregulation of RAS/RAF/ERK and PI3K/PTEN/ AKT signaling pathways has been observed in CRC tumors having ERBB2 gene amplification and/or V777L activating mutation. 54,55 Currently, targeted therapies with mAb against the ERBB2 receptor are the standard of care for mCRC tumors positive for ERBB2 amplification. There are several ongoing clinical trials with immune checkpoint inhibitors, anti-EGFR/ERBB2 mAb, and/or small molecule tyrosine kinase inhibitors for mCRC tumors harboring ERBB2 activating mutations (see Table 4). Numerous studies have shown up-regulation of ERBB2 signaling pathways as a mechanism of resistance for anti-EGFR therapies in mCRC, which has led to consider ERBB2 amplification/mutations as predictive biomarkers of resistance for anti-EGFR therapies in KRAS and BRAF wild-type mCRC. 56 Although cancer outcomes are associated with ethnicity, the participation in clinical trials of individuals from diverse racial/ethnic groups remains substantially low (25.9%, 5.0%, 1.1%, 0.2%, and 0.9% for Whites, Asians, African Americans, Hispanics, and other minorities, respectively). 57 Our team has examined the association between ancestral groups and CRC risk in PRH, showing that African American ancestry was associated with an increased risk of developing rectal tumors. 58 This could partly explain the mutational frequency differences observed among PRH and other ethnic/racial groups, including Hispanic subgroups. For example, a lung cancer study performed with Latin Americans showed that Native American ancestry was strongly correlated with mutations in the EGFR and KRAS genes, underscoring the importance of providing genetic testing for Latin American patients with lung cancer with admixed ancestries. 59 Future investigations are needed to assess ancestry-specific germline variation among Hispanic CRC tumors. Efforts to increase the participation of ethnic/racial diverse populations in early-phase clinical trials are of paramount importance to help elucidate the effect of germline variation as prognostic and predictive therapeutic biomarkers.

| CONCLUSION
Some factors contributing to the observed differences in the mutational frequencies between PRH and the other datasets are secondary to differences in sample selection and methodologies used to assess mutations and ethnicity. Additional limitations of the present study are the limited clinical data on CRC tumors from PRH since the analysis was restricted to data of patients who underwent NGS in a clinical setting through commercial laboratory testing. Thus, there is limited information on socio-demographic data, additional clinicopathologic variables, treatment regimens, and outcomes data. Nonetheless, this is the first study to systematically examine the somatic mutational profile and molecular alterations of PRH and contrast them with two large national and international datasets. Our study provides data on the specific mutational landscape for Hispanics with CRC and the implications on therapeutic options and clinical outcomes. Hence, future investigations into the carcinogenic pathways predominantly affecting Hispanic populations and the development of targetable therapies that will impact clinical outcomes among diverse populations are warranted.
The development of novel, comprehensive, and more accessible tumor interrogation methods, such as NGS, has paved the way for the identification of genomic drivers of cancer. Subsequently, this has improved our understanding of cancer health disparities for diverse populations, including underrepresented populations such as Hispanics. Our study provides the first comprehensive somatic molecular evaluation of colorectal tumors among PRH, USH, and non-Hispanics demonstrating important differences in targetable biomarkers and carcinogenic pathways. The findings of this study will serve as a guide for future research and the identification of tailored cancer-targeted therapies to improve overall CRC survival and decrease disparities among diverse populations.