English
Arabic
French
Genomic Landscape of Advanced Solid Tumors in Middle East and North Africa Using Circulating Tumor DNA (ctDNA) in Routine Clinical Practice
Abstract
Introduction
Next-generation sequencing (NGS) of tumor DNA can detect actionable drivers and help guide therapy for patients with advanced-stage cancers. While tissue-based genotyping is considered standard of care, blood-based genotyping is emerging as a valid alternative. Tumor genomic profiles may vary by region, and data from the Middle East and North Africa (MENA) are not widely available. This study elucidates the genomic landscape of advanced solid cancers in patients from the MENA region by retrospectively analyzing results from NGS ctDNA testing.
Methods
In routine clinical practice, 926 plasma samples from 767 patients with advanced cancers from the MENA region were profiled using a comprehensive NGS assay (Guardant360®). We conducted a pan-cancer analysis and sub-analyses focusing on lung, breast, and colorectal cancers.
Results
In the pan-cancer group, TP53 (58.5%), EGFR (20.4%), and KRAS (18.9%) were the most frequently mutated genes. EGFR (10.2%), FGFR1 (4.9%), and PIK3CA (4.9%) showed the most amplifications, while fusions were observed in 2.7% of patients, including ALK, FGFR2, and RET. For lung adenocarcinoma, EGFR (30.5%), KRAS (19.3%), and ERBB2 (4.6%) were the most frequently identified alterations among the genes recommended for evaluation by the National Comprehensive Cancer Network (NCCN). In patients with breast cancer, PIK3CA (35.3%), ESR1 (21.7%), and BRCA1/2 (13.3%) had the most prevalent alterations among NCCN-recommended genes. In colorectal cancer, KRAS (39.0%), NRAS (8.0%), and BRAF (V600E, 4.0%) were the most observed mutations among genes recommended by the NCCN. Comparing this cohort to publicly available Western and Eastern datasets also indicated similarities (including PIK3CA in breast cancer) and variances (including EGFR in lung adenocarcinoma) in key genes of interest in the analyzed cancer types.
Conclusion
Overall, our findings provide insight into the genomic landscape of individuals with advanced solid organ malignancies from the MENA region and support the role of ctDNA in guiding therapeutic decisions.
Introduction
Malignant neoplasms are the second leading cause of death worldwide and are associated with an enormous public health burden . The disease burden of cancer is also increasing in Middle Eastern and African countries, leading to significant productivity loss and healthcare spending .
The routine implementation of precision oncology depends on the availability of technologies that can reliably identify biomarkers targeting specific medicines. Next-generation sequencing (NGS) has enabled the identification of such genomic biomarkers.
Tumor tissue-based DNA sequencing remains the mainstay for evaluating somatic genomic alterations; however, it is associated with certain limitations . Approximately 10%–20% of patients are ineligible for a biopsy. For up to 20% of eligible individuals, the available tumor tissue does not meet the quantity or quality requirements for NGS analysis . Furthermore, a single tumor biopsy may not fully represent tumor heterogeneity and may miss emerging resistance alterations. The turnaround time to complete comprehensive tumor tissue profiling may be more than a month, which can delay the initiation of treatment [6,7]. Moreover, repeated biopsies may incur incremental costs to patients or health systems because of associated complications .
NGS analysis of plasma-derived ctDNA enables a comprehensive assessment of the tumor genomic landscape with a high genotyping success rate. It has emerged as a non-invasive alternative that may help overcome several limitations of tumor tissue-based analysis. The average turnaround time for liquid biopsy is 7–10 days, shorter than that for tumor tissue-based profiling . Liquid biopsy can better capture tumor heterogeneity caused by multiple clonal populations in a tumor . It is minimally invasive and can help monitor tumor evolution and resistance over time without the risk of complications associated with tumor biopsies .
However, ctDNA analyses also have limitations. ctDNA concentration in plasma depends on tumor DNA shedding, which varies among tumor types . In one study, plasma analysis failed to detect any cancer-derived somatic alterations in 15% of patients, representing over 50 types of late-stage cancers . In addition, the likelihood of detecting ctDNA in blood depends on the location of the tumor .
Several factors may contribute to differences in the genomic profiles reported for specific populations of patients with cancer, including their geographical area of origin. For example, distinctions in the genomic profiles of patients with non-small cell lung carcinoma (NSCLC) from Western and East Asian populations have been identified using ctDNA-based NGS . While there have been numerous reports describing cancer genomic profiling studies of Western and, to a lesser extent, Asian populations, less is known about cancer genomic profiling from other geographic areas, such as the Middle East . Therefore, the current study aimed to retrospectively analyze the genomic landscape of advanced solid tumors from patients in the Middle East and North Africa (MENA) region who were tested by ctDNA NGS as a part of routine clinical practice.
Materials and Methods
Study Design
We analyzed data from Guardant360 tests ordered between September 2016 and January 2023 from Saudi Arabia, Kuwait, Bahrain, UAE, Qatar, Jordan, Lebanon, Iraq, Egypt, Tunisia, and Morocco. Whole blood samples were collected from patients with advanced solid tumor malignancies in Streck Cell-Free DNA blood collection tubes (BCTs), stabilizing cfDNA and nucleated blood cells . We excluded patients enrolled in prospective clinical trials.
Circulating Tumor DNA Next Generation Sequencing
Guardant360® (Guardant Health Inc; Redwood City, CA, USA) is a comprehensive genomic profiling assay that identifies single-nucleotide variants (SNVs), insertions and deletions (indels), fusions, and amplifications (amps) using a single blood sample [17,18]. It is the first blood-based comprehensive genomic profiling test approved by the US Food and Drug Administration (FDA) . It is indicated for use in patients with advanced solid tumors at diagnosis and at the time of disease progression. The assay covers complete exon sequencing of multiple genes, including EGFR, ERBB2, and KRAS. During the sample collection period, the assay included 70 to 74 genes (v2.9 to v2.11). Analysis of microsatellite instability was introduced with v2.11.
All samples were analyzed in a single Clinical Laboratory Improvement Amendments (CLIA)-certified and CAP-accredited laboratory in California.
Statistical Analyses
The present analysis focused primarily on clinically informative genomic alterations as described by National Comprehensive Cancer Network® (NCCN) guidelines. Synonymous mutations and variants of unknown significance were not considered clinically relevant and were excluded from this analysis.
Descriptive statistics were used to report sample characteristics, mutation types, and frequencies. In patients with multiple concurrent potential driver alterations, the following conventions were applied to categorize the primary driver. In most cases, the alteration with a higher variant allelic frequency (VAF) was considered the primary driver. Potential driver mutations in separate genes with similar or relatable VAFs were classified as compound mutations.
Database Analysis
The frequency of NCCN-recommended biomarkers in our patients with advanced lung adenocarcinoma, breast cancer, and colorectal cancer were also compared to those in publicly available databases in cBioPortal. We used the Memorial Sloan Kettering – Metastatic Events and Tropisms (MSK-MET) database for Western population representation . For the Asian population, data were retrieved from the China Pan-Cancer (CPC) database . Descriptive statistics were used to report frequencies of relevant biomarkers.
Results
Sample Characteristics
Nine hundred twenty-six plasma samples from patients with advanced solid organ malignancies were included. Samples that were canceled, had no ctDNA detected, or contained only synonymous mutations were excluded. The remaining 786 eligible samples with profiling results came from 767 patients (Fig. 1a). This patient population included 56% men. The median patient age was 63 years (range, 27–92). Most patients (95%) were profiled once, and the remaining were genotyped twice or more. The median number of alterations per sample was 3 (range, 1–91). The average turnaround time from the sample receipt in the laboratory to reporting the results was 7.8 days (range, 4–27; median, 7) (Table 1).
Pan-cancer Analysis
The analysis included patients with lung adenocarcinoma (25%), breast carcinoma (23%), colorectal adenocarcinoma (14%), pancreatic ductal adenocarcinoma (8%), cholangiocarcinoma (5%), ovarian carcinoma (4%), prostate adenocarcinoma (3%), and other cancer types (18%) (Fig. 1b).
In the overall cohort, the most prevalent SNVs were detected in TP53 (58.5%), EGFR (20.4%), KRAS (18.9%), PIK3CA (13.9%), and APC (8.7%) (Fig. 2a). Amplifications were most frequently observed for EGFR (10.2%), FGFR1 (4.9%), PIK3CA (4.9%), MYC (4.6%), and BRAF (4.4%) (Fig. 2b). Fusions were observed in 2.7% of the profiled patients. They included ALK, ROS1, FGFR2, FGFR3, and RET (Fig. 2c). Microsatellite instability-high (MSI-H) was detected in 2% of patients tested. Activating ERBB2 mutations, KRAS G12C, and BRAF V600 mutations were observed in 4.2%, 4.0%, and 1.8% of patients, respectively (Supplementary Fig. S1).
Lung Adenocarcinoma Analysis
From 227 patients diagnosed with lung cancer, after excluding patients with neuroendocrine or homogeneously squamous histology, we analyzed genomic alterations in the remaining 185 patients. The most frequently occurring SNVs were in TP53 (62.4%), EGFR (30.5%), KRAS (19.3%), STK11 (13.7%), and ATM (12.2%) (Supplementary Fig. S2a). The most commonly amplified genes were EGFR (13.2%), CCNE1 (3.1%), CDK4 (3.1%), MYC (3.1%), and FGFR1 (2.0%). The lack of distinction between focal amplification and aneuploidy for EGFR amplification in this version of the Guardant360 assay may have influenced this observation. MET and ERBB2 were amplified in 2.0% and 1.0% of patients, respectively (Supplementary Fig. S2b). Fusions were observed in 6.5% of patients and included EML4-ALK (50.0%), other ALK fusions (17.0%), ROS1 (17.0%), RET (8.0%), and FGFR2 (8.0%) (Supplementary Fig. S2c). MSI-H was detected in 1.3% of the patients tested for this.
Next, we focused on the frequency of NCCN guideline-recommended genes . The most common mutations involved EGFR (30.5%), KRAS (others) (10.2%), KRAS (G12C) (9.1%), ERBB2 (4.6%), and BRAF V600E (2.5%); ALK fusions were detected in 3.1% of patients (Fig. 3).
The landscape of EGFR mutations was analyzed in further detail. The most prevalent alterations were exon 19 deletions (40.0%), EGFR L858R (26.0%), exon 20 insertions (9.0%), T790M (9.0%), and C797S (4.0%) (Supplementary Fig. S3). Among patients with exon 19 deletions and L858R, analyses revealed the presence of multiple co-alterations. For exon 19 deletions, common co-alterations included MET amplifications (10.4%), EGFR amplifications (6.9%), T790M and/or C797S (6.9%), ERBB2 SNVs (6.9%), PIK3CA E542K (6.9%), and BRAF V600E (6.9%). For those with EGFR L858R, MET amplifications were detected in 11.8% (Supplementary Fig. S3).
Uncommon EGFR SNVs accounted for (13.6%) of all EGFR variants (Supplementary Table S1). ERBB2 modifications were observed in 4.6% of the patients. Among them, 65.0% were ERBB2 exon 20 insertions, 26.5% were ERBB2 KD mutations, and 8.5% were ERBB2 amplifications (Supplementary Table S2). BRAF mutations were detected in 2.7% of the patients, and among them 66.5% were BRAF V600E, whereas 33.5% were BRAF non-V600E (Supplementary Table S2).
Breast Cancer Analysis
In 166 patients with advanced breast cancer of any histology (i.e., unselected), the most frequently mutated genes included TP53 (53.0%), PIK3CA (35.3%), ESR1 (21.7%), BRCA1/2 (13.3%), and ATM (13.2%) (Supplementary Fig. S4a). The most commonly amplified genes were FGFR1 (15.1%), CCND2 (12.1%), CCND1 (9.6%), PIK3CA (9.6%), and MYC (8.4%) (Supplementary Fig. S4b). Fusions were found in 3% of patients and involved ALK (40%), FGFR2 (40%), and NTRK1 (20%). MSI-H was found in 1.8% of the patients. Clinically informative alterations according to the NCCN included mutations in PIK3CA (35.3%), ESR1 (21.7%), BRCA1/2 (13.3%), ERBB2 (10.2%), and amplification of ERBB2 (4.8%) (Fig. 4).
An in-depth, gene-specific analysis was also conducted on ERBB2 (Supplementary Fig. S5a), PIK3CA (Supplementary Fig. S5b; Supplementary Fig. S6), ESR1 (Supplementary Fig. S7), and BRCA genes (BRCA1 and BRCA2; Supplementary Fig. S8). In PIK3CA, the most prevalent mutations were observed in the PI3_PI4 domain (46%), followed by the helical domain (27%) (Supplementary Fig. S5b). Nearly one-third (32%) of all PIK3CA mutations occurred outside of typical hotspots. Additionally, insertions or deletions predicted to be activating alterations were identified in 5% of samples with any PIK3CA mutation (Supplementary Fig. S6).
All alterations in ESR1 occurred in the ligand-binding domain. Common mutations included Y537S/N/D (41%), D538G (37%), and E380Q (13%) (Supplementary Fig. S7). Overall, 34 genetic alterations were observed in BRCA1/2 (12 SNVs and 22 indels) in 22 (13.3%) of patients with breast cancer. Among them, multiple BRCA alterations were detected in 18.2% of patients, BRCA + ESR1 in 13.6%, BRCA + PIK3CA + ESR1 in 13.6%, BRCA + PIK3CA in 13.6%, and BRCA + PIK3CA + PTEN in 9.1%. Most BRCA alterations appeared to be suspected germline (Supplementary Fig. S8). Among patients with ERBB2 mutations, most (76%) were in the kinase domain (KD), 18% occurred in the extracellular domain (ECD), and 6% were exon 20 insertions.
Colorectal Cancer Analysis
Data from 100 patients with colorectal cancer were analyzed. The most frequently mutated genes included TP53 (83.0%), APC (47.0%), KRAS (39.0%), PIK3CA (23.0%), and SMAD4 (18.0%) (Supplementary Fig. S9a). Frequently amplified genes included EGFR (27.0%), BRAF (9.0%), FGFR1 (7.0%), CCNE1 (7.0%), and KRAS (4.0%) (Supplementary Fig. S9b). No fusions were found.
Considering the NCCN recommendations for genomic testing in colon cancer , we observed mutations in KRAS (39.0%) and NRAS (8.0%), BRAF V600E (4.0%), ERBB2 amplification (3%), and MSI-high (2.0%) (Fig. 5).
In the KRAS gene, the most frequent variants were KRAS G12A/D/L/V (59.2%), followed by KRAS G12C (14.3%), KRAS Q61H (11.9%), KRAS G13D (4.8%), and KRAS A146T (4.8%) (Supplementary Fig. S10a). Most KRAS variants were located in exon 2 (78.0%), followed by extended RAS variants (20.0%) and atypical variants (2.0%) (Supplementary Fig. S10b). BRAF V600E and nonV600E mutations were each seen in 4.0% of patients. ERBB2 KD mutations and ERBB2 amplifications were observed in 2.0% and 1.0% of patients, respectively (Supplementary Table S3). MET amplification was observed in 3% of patients.
Database Analysis
In lung adenocarcinoma, for the most prevalent NCCN biomarkers (EGFR, KRAS, and ERBB2 mutations), the nominal values for frequencies of genetic alterations in our sample fell between the values recorded in the MSK-MET and CPC databases. A higher prevalence of ALK fusions was observed in the CPC database. The frequency of remaining biomarkers was similar across the various databases (Fig. 6a).
For breast cancer, the most frequently altered NCCN biomarker, PIK3CA, had a similar prevalence rate in our cohort to that in the MSK-MET and CPC databases; however, ESR1, BRCA1/2, and ERBB2 mutations were more frequently observed in our dataset (Fig. 6b).
For colorectal cancer, the most frequently affected NCCN gene, KRAS, was less commonly mutated in our patient cohort than in the MSK-MET and CPC databases. In contrast, NRAS was more commonly mutated in the current study (Fig. 6c), which may reflect a bias in patient selection for liquid comprehensive genomic profiling.
Studies have described geographical variance in the genomic landscape of advanced cancers but with less information regarding African, Central Asian, and South American regions . Regarding ctDNA analyses, only a handful of studies have described genomic findings in MENA, usually with small cohorts (≤100 patients) and focusing on single cancer types [26-29]. Our data therefore expand this information with a larger sample from various countries in the MENA region diagnosed with different cancer types.
Implications of the Lung Cancer Analysis
We focused on lung adenocarcinoma because it was the most common histologic sub-type in our database and is most likely to be associated with actionable alterations. We found that 65% of samples had clinically informative alterations in genes defined by the NCCN (EGFR, KRAS, ERBB2, ALK, BRAF, MET, ROS1, and RET).
The frequency of most biomarkers was similar to those in Western and East Asian cohorts, but with notable exceptions. The rate of EGFR mutations in patients with lung adenocarcinoma from the MENA region (30.5%) was between those reported for East Asian (40%–50%) and Western populations (10%–15%) [30-32]. Our findings are in line with those from tissue analysis in MENA, with EGFR mutation frequency in NSCLC of 17.2% . Similarly, the prevalence of KRAS mutations among patients with NSCLC is lower in Asian countries (5%–11%) than in Western countries (20%–26%) . Our findings show that, among MENA patients with lung adenocarcinoma, KRAS mutation frequency (19.3%) was between the MSK (33.1%) and CPC cohorts (11.8%), but with the same relative rates of specific KRAS mutations. For example, KRAS G12C accounted for 47.3% in our cohort, 41.4% in MSK-MET, and 41% in the CPC database.
Comprehensive genomic profiling can improve the detection of uncommon and complex EGFR variants, including exon 20 insertions . In our cohort, 9% of the EGFR alterations were exon 20 insertions, similar to the rate reported for Western populations ; however, given that the overall number of patients harboring EGFR mutations is higher than in the West, there were proportionately more patients with lung adenocarcinoma in MENA with such alterations. Additionally, 13.6% of the actionable EGFR variants detected by our analysis occurred outside of exon 19 deletions, L858R, and T790M (Supplementary Table S1). These could have been missed by hotspot-based testing methods. EML4-ALK was the most common gene rearrangement; however, we also detected a rare FGFR2-NRAP fusion. FGFR2/3 fusions are rare resistance alterations in lung cancer that are acquired after treatment with EGFR TKIs . Our cohort had 1.3% of patients with MSI-H, which is consistent with previous data .
Implications of the Breast Cancer Analysis
In addition to germline testing for BRCA1/2 mutations and PALB2 mutations, the NCCN guidelines for advanced breast cancer recommend testing for somatic alterations, such as ERBB2 amplifications, ESR1 mutations, activating AKT1 and PIK3CA mutations or PTEN alterations, NTRK fusions, MSI-H, and RET fusions; assessments can be considered for emerging biomarkers, such as ERBB2 activating mutations and somatic BRCA1/2 mutations. The NCCN guidelines propose limiting the use of liquid biopsy to recurrent/stage IV (M1) disease, where it can serve as an alternative to tissue biopsy to identify candidates for targeted therapies.
A study comparing liquid and tissue biopsy in patients with metastatic breast cancer found that plasma-based genotyping could detect high rates of actionable mutations. Our findings are consistent with limited single-institution studies from MENA, such as those reporting 25.9%–31% prevalence for PIK3CA mutations , .
The frequencies of most alterations in our cohort are consistent with those from the MSK-MET and CPC databases. These databases were suitable comparators, because, like our real-world observational cohort, they included patients with unselected advanced breast cancer. However, there are notable differences in our findings, such as a lower incidence of ERBB2 amplification (4.8% vs 10.9% in MSK-MET and 23.8% in CPC). Other studies from MENA have described ERBB2 amplification or protein overexpression in 20%–29% of patients with advanced breast cancer.
The ability of tissue- or plasma-based NGS platforms to detect ERBB2 amplification in tumors known to demonstrate HER2 (protein) overexpression by immunohistochemistry or ERBB2 amplification by fluorescence in situ hybridization (FISH) has been challenged. However, one recent study in patients with ERBB2-positive breast cancer demonstrated that, in routine clinical practice, the concordance between ERBB2 amplification by NGS and ERBB2 positivity by immunohistochemistry and/or FISH was approximately 70% .
Although our database does not include results of tissue-based testing, we hypothesize that our cohort is over-represented by patients with HER2-negative advanced breast cancer (hormone receptor-positive or triple-negative). This is supported by a higher incidence of ESR1 (21.7% vs. 7.2% and 3.7%), BRCA1/2 (13.3% vs. 4.4% and 5.6%), and activating ERBB2 mutations (10.2% vs. 3.1% and 3.7%) than in the MSK-MET and CPC databases. The higher incidence of ESR1 mutations in our cohort suggests a prior exposure to aromatase inhibitors for hormone receptor-positive disease. Furthermore, a higher incidence of BRCA1/2 mutations is expected in ER/PR-positive or triple-negative disease relative to the HER2-positive phenotype, and ERBB2 mutations are expected after exposure to ER-directed or anti-HER2 therapy in metastatic breast cancer .
Implications of the Colorectal Cancer Analysis
The prevalence of BRAF V600E was similar in the MENA and CPC cohorts, while it was lower in MSK-MET, suggesting regional variance. Compared to MSK-MET, the rate of MSI-H in our MENA cohort was lower, mainly due to the ordering bias. Typically, dMMR testing is performed in earlier lines, while ctDNA testing is used later to detect resistance mechanisms, usually in patients who have experienced progression on anti-EGFR monoclonal antibodies. The frequency of KRAS mutations (39.0%) was lower compared to MSK-MET (44.1%) and CPC (48.2%). Our findings are consistent with a study of CRC tissue testing from MENA (32.2%–48%).
Among 100 patients with advanced CRC, eight had NRAS mutations. This is numerically higher than reported in previous studies from the region (NRAS: 2.2%–3.2%) and in the MSK-MET (4.4%) and CPC databases (3.3%). However, given the relatively small cohort size, our findings could be due to chance. Similarly, while no fusions were observed, it is expected that only 1.0% of all tumors would have clinically actionable rearrangements (FGFR2, FGFR3, RET, ALK, NTRK1, and ROS1 combined). The cohort may not have been large enough to include such rare events.
Study Strengths and Limitations
Our study addresses an information gap, as there are no large-scale analyses of ctDNA genomic profiling for advanced cancer patients in the MENA region. Moreover, our findings provide comprehensive real-world data regarding the mutational landscape of such patients.
This study was retrospective and reflected patterns of real-world testing, where ordering biases may have occurred. Detailed pathological features, such as ER, PR, and HER2 status in breast cancer, were not available for all patients. Treatment history, results of previous or concurrent biomarker testing, and each patient’s line of testing were unknown. For some, ctDNA NGS testing may have been the first and only analysis, while for others it could have been used as a reflex test following a negative result from another test. Therefore, the frequency of alterations observed in our study may not reflect those for treatment-naive advanced cancers in MENA.
Nevertheless, ctDNA-based NGS has certain advantages. For example, it detects a wide range of genomic alterations from a single blood sample within a week, with performance similar to that of tissue-based genotyping. However, the plasma concentration of ctDNA depends on tumor DNA shedding, which may be less with well-differentiated cancers, lower tumor burden, or when tumors are limited to the central nervous system. In cases of specific complex genomic alterations, such as fusions, large insertions, or deletions, sensitivity may be decreased compared to that seen with SNVs. Furthermore, while the presence of non-tumor mutations derived from clonal hematopoiesis of indeterminate potential (CHIP) may be a confounding factor, most actionable driver mutations are not among the common alterations associated with CHIP.
Conclusion
Our findings provide insight into the genomic landscape of individuals with advanced cancer from the MENA region and further support the role of ctDNA NGS to identify patients for treatment with appropriate targeted therapy.