Impact of a Molecular Sequencing Systematic at Diagnosis in Digestive Oncology: Experience of a French Center

Materials Methods: We analysed the and the data of with who received molecular high-throughput sequencing analyses at diagnosis. The primary objective of this single-center study was to compare the overall survival of patients who were treated with molecularly matched therapy with patients who received standard therapy. Median overall survival was calculated from initial disease diagnosis to death. Results: 528 patients were referred to the Digestive Oncology Department of the Timone Hospital in Marseille between January 2018, and November 2020 for management of digestive cancer and received high-throughput molecular sequencing. Among them, 461 patients had a digestive carcinoma (75 of them were excluded because of the presence of a GIST or a neuroendocrine tumor, a digestive localization of extra digestive cancer or the absence of follow-up in our center) and 275 had metastatic disease (synchronous or metachronous). For metastatic patients, actionable molecular alterations were identified in95 patients (43.5%) and for 13 patients (4.7%) a molecularly matched therapy was administered. There was no significant difference in median overall survival between patients who received matched therapy than patients who did not receive molecularly matched therapy (2.89 [95%CI 1.84 - 3.93] vs. 2.86 [95%CI 1.52 - 4.19], p=0.671). Conclusion: This study suggests that high-throughput genomics can improve management of patients. Although these results did not show a benefit in overall survival for tumors who harboured such actionable molecular alterations and who received molecularly matched therapy, than patients who did not receive molecularly matched therapy, they are promising. Randomized trials are needed to confirm that there is a benefit to treating patients with matched therapy based on NGS.


Introduction
There were an estimated 3.9 million new cancer cases and 1.9 million deaths from cancer in Europe in 2018 excluding non-melanoma skin cancer). Digestive cancers represent a quarter of the overall incidence (Esophagus 1.4%, Stomach 3.4%, Colorectal 12.8%, Liver 2.1%, Pancreas 3.4%, i.e., 23.1%) and a third of cancer mortality (Esophagus 2.3%, Stomach 5.3%, Colorectal 12.6%, Liver 4%, Pancreas 6.6%, i.e., 30.8%) [1]. Among these, colorectal cancer is the second leading cause of cancer death in males and the third in females. Pancreatic cancer, whose incidence is increasing, is the fourth cause. The observed trends in digestive cancer mortality rates in Europe are stable or slightly increasing. Therefore, the number of deaths is expected to increase with the continuous aging of the European population [1]. The molecular pathogenesis of cancer is characterized by the successive acquisition of genetic alterations that lead to aberrant activation of proto-oncogenes and inactivation of tumor suppressor genes. The molecular landscape of cancers of the digestive system are becoming well known, but we need to improve the implications of this new knowledge for the therapeutic management of patients [2,3]. Now, several testing methods to identify genomic alterations are used in routine clinical practice: detection of a mutant or loss of protein expression immunohistochemistry (IHC), search for mutations in the DNA sequence by Polymerase Chain Reaction (PCR), as well as molecular profiling based on Next Generation Sequencing (NGS). The PCR method (the so-called Sanger method) focuses on a region of interest, such as a single gene or a group of genes selected according to the clinical signs collected and the diagnostic hypothesis made by the prescriber. Thanks to NGS sequencing, it is possible to sequence millions of fragments simultaneously. This high-throughput process results in the sequencing of hundreds to thousands of genes at a time, allowing the detection of novel or rare variants [4]. This method also requires less cellular material, and is faster, less expensive and more accurate. This NGS sequencing includes DNA sequencing looking for genomic mutations, but some genes (notably FGFR2 or NTRK) once transcribed, may form fusions with multiple partners for which additional RNA sequencing is required.
Thanks to this technology, new genes associated with cancer and the clinical significance of these genomic alterations in diagnostic, prognostic and therapeutic management were discovered [5]. Today, based on molecular anomalies, some patients benefit from targeted therapy diagnosed by these genomic methods, either routinely or within the framework of clinical trials or ATU (Temporary Authorization for Use). Promising targeted therapy and personalized medicine are making molecular profiling of tumors a priority. The main actionable mutations and their targeted therapies in digestive oncology are summarized in (Figure 1). The main objective of our study was to determine the benefit of overall survival of these targeted therapies administered on the basis of NGS data systematically performed at the time of digestive cancer management.

I Study Design and Patients
In order to identify the impact of systematic targeted sequencing at diagnosis, we retrospectively collected data from all patients who underwent NGS in the digestive oncology department of Timone Hospital, APHM, Marseille, France, between January 2018 and November 2020 for DNA (mutation NGS) and November 2018 to November 2020 for RNA (fusion NGS). Patients eligible for inclusion in our study were those with i) histologically proven digestive cancer in our center with a sequencing performed at diagnosis and for whom ii) a Multidisciplinary Concertation Meeting has been performed at least for the first line of treatment. The patients excluded from our analysis were those who had no follow-up in our center (death before treatment, absence of data in the medical file), had nondigestive cancers (lymphoma, lung, bladder) and had some particular digestive cancers such as GIST, neuroendocrine tumors, in situ cancers.
The follow-up of patients started at diagnosis with systematic molecular profiling and decision during an RCP (Reunion de concertation pluridisciplinaire) of a first line of treatment. Tumor samples were collected by endoscopic biopsy or surgical sampling and could be from primary or secondary lesions. These samples were analysed by the Ion Torrent S5 XL ThermoFisher technology, which is an amplicon-based technique, after signing a written and informed authorization. This technique is based on semiconductor chips, filled with wells, which release a proton when a nucleotide is incorporated by the DNA polymerase. This reaction causes a local pH change that is detected by sensors on the chip and converted into raw data in the form of an ionogram, before being transformed into sequence data. Patient data were collected from the computerized medical record of our institution, in particular from the minutes of the multidisciplinary consultation meeting for clinical data and the molecular biology report for mutations.  [6]. Specific treatments were proposed by the oncologist and discussed in a multidisciplinary consultation meeting, then explained to the patients (inclusion in a trial, ATU or exceptional treatment) who had the choice of accepting or not, after clear information during a dedicated consultation.

II Molecular Sequencing
Apart from RAS mutations in colorectal cancer and HER2 amplification in gastric cancer, which are already investigated and necessary for therapeutic decision, the actionable mutations retained, according to the literature microsatellite instabilities, mutations affecting the DNA polymerase POLE proofreading domain in colorectal cancer, BRAF mutations, in particular V600E in colorectal cancer, cholangiocarcinoma or pancreatic cancer, ERBB2/HER2 mutations in colorectal cancer or cholangiocarcinoma, MET amplification in colorectal cancer, ALK, ROS1 rearrangements and EGFR mutations in colorectal and pancreatic cancer, NTRK1/3 fusions, BRCA1/2 or even PALB2, ATM, RAD51 mutations, IDH1 and FGFR1/2 fusions or rearrangements in cholangiocarcinoma, KRAS G12C mutations, and more broadly and nonexhaustively thanks to the 'basket' clinical trials: PTEN loss or mutations in PIK3CA, AKT1, VEGFR, mTOR pathway (PDGFRA, KIT, DDR1/2), RET fusions or mutations .

III Outcomes
The primary objective of our single-center retrospective study was to evaluate the overall survival of patients who received decided and targeted therapy on initial NGS data, regardless of the treatment line. We performed exploratory indirect comparisons against patients who received standard lines of therapy and against literature data. Patients were followed throughout their treatment at our center. We constituted two separate analyses of the overall survival of patients with metastatic cancer and those with cancer that remained localized throughout the management. In the metastatic group, we evaluated the overall survival of patients with a 'hypermutated' mutation status (with at least 4 mutations found) and those without a hypermutated or mutation status, as well as survival by organ and by mutation.

IV Statistical Analyses
To compare proportions (e.g., MSI status) a χ 2 test was used for categorical variables and a Mann-Whitney test for ordinal variables. We performed survival analyses using the Kaplan-Meier method and compared the curves with a log-Rank test; estimates are reported for the medians with 95% confidence intervals. A significance level for p values was set at 0.05 for all statistical tests. All the analyses were performed with IBM SPSS Statistics 26.0 (IBM Inc., New York, USA).

Results
Of the 535 patients who had NGS sequencing requested by a Timone APHM digestive functional unit between January 1, 2018, and November 30, 2020, 461 were included in our study ( Figure 2). Among the patients excluded from the analysis, 21 patients (3.9%) had GIST or NET, and were therefore excluded from our analysis due to significant differences in management and tumor oncogenesis of these particular tumor types. Fifteen patients (2.8%) had a digestive location of a nondigestive primary (melanoma, bladder cancer, lymphoma, pleural mesothelioma, lung cancer). Two patients (0.4%) died before any medical management. Eight patients (1.5%) had been sequenced on surgical specimens or biopsies, but the histological analysis did not reveal any carcinomatous proliferation. Four patients (0.7%) had no clinical data available and twenty-four (4.5%) were followed up in another center.

Discussion
In our study, no overall survival benefit was found for patients treated with a therapy adapted to a molecular target detected on an NGS systematically performed at the time of medical management. However, there is a large mutational panel of tumors in digestive oncology as shown in this study. Indeed, 82.9% of patients have at least one mutation in the metastatic situation, and many targets potentially actionable outside the current marketing authorization were found (43.2% in the overall population and 34.5% in the metastatic situation). Our study is large, with a large number of patients, collected in real life and with a follow-up of nearly 3 years, and descriptive data on the mutational landscape of digestive cancers provided on a large database for future studies. Moreover, the tumor locations were diversified with almost 50% of non-colorectal cancers, accounting for differences in oncogenesis between the different tumor subtypes and may allow the selection of patients for future trials targeting a specific molecular alteration. We also included patients without limiting ourselves to metastatic patients, giving us an idea of the temporal dynamics of genomic alterations. Some patients showed a 'hypermutated' status, which seems to predict a response to immunotherapy, in the KEYNOTE 028 trial as well as in the preliminary results of the TAPUR trial and thus could allow the indication for immunotherapy to be extended, especially in those without microsatellite instability [30,31].
This study, although based on overall survival which is a strong and relevant assessment, is limited by the fact that it is retrospective. Indeed, some patients included at the beginning of the study showed mutations that had not yet been identified as potentially actionable and could have been the subject of targeted treatment if they had been discovered at the end of the study. Thus, despite the duration of nearly 3 years, which may seem relatively short, the targeted therapies proposed in 2020 were more numerous than those existing in 2018. Regarding the main objective of the benefit of targeted therapy on overall survival, the results obtained in this study cannot be generalized to each tumor subgroup because of the heterogeneity of our cohort. Indeed, each tumor subgroup is not impacted in the same way by the implementation of a targeted treatment on a mutation found by NGS. This tumor heterogeneity would lead to a lack of power if we were to analyse the benefit of a targeted therapy based on each location.
There has long been controversy about whether the use of genomics could improve survival outcomes in patients with difficult-to-treat cancers. Despite an early negative study, recent studies now show a clear survival benefit from searching for and treating activatable tumor abnormalities, with prolonged survival in subgroups and sometimes clinical and objective responses [32]. Schwaederle et al. conducted meta-analyses of phase 1 and phase 2 trials of a personalized oncology treatment strategy targeting any biomarker, and in both cases demonstrated that the use of molecularly targeted therapy resulted in a higher response rate and improved progression-free survival and overall survival compared with standard therapy [33,34]. In the IMPACT/COMPACT trial, conducted by Stockley et al. 1893 patients were enrolled and 1640 were tested. 84 patients were treated with molecularly targeted therapy (5%) with a significantly higher overall response rate [35]. In the WINTHER trial, 107 patients received molecularly targeted therapy based on DNA sequencing (n=69) or RNA sequencing (n=38), the rate of stabilized disease > 6 months, partial response or complete response was 26.2% [36]. These results were also found to a greater extent in the MOSCATO-01 trial conducted by Massard et al. where tumor sequencing improved survival for 33% of patients with advanced cancers, particularly in advanced biliary cancers [37,38]. This benefit was also found in the Pishvaian et  in the group treated with molecularly targeted therapy, and no other therapeutic modality offers a benefit of this magnitude in this patient population [39].
The realization of a systematic NGS at diagnosis is an important question today. At the economic level in France, this analysis is not yet part of the social security acts. It is registered in the list of acts outside the nomenclature (RIHN), with a cost of 882.90 euros per sequencing, that is to say 1765.80 euros for a DNA and RNA research. The hospital is reimbursed at about 50%. In addition, logistical issues and the aggressiveness of the disease are also factors to be taken into account. In our study, we performed this sequencing at diagnosis for all patients regardless of their disease status and this may have caused an important selection bias, 7 patients died less than one month after their sequencing, 186 patients remained localized. Limiting the prescription of NGS to metastatic/treatment-refractory situations, or to young patients still able to receive treatment is a key issue [40].
Given the limited number of genes whose analysis is currently of proven medical interest, NGS is currently used in the form of a panel (Targeted sequencing) to rapidly perform an analysis of a selected number of genes of theranostic interest on a large number of patients in order to reduce delays and costs. But an important contribution of the new sequencing technologies in cancer is to have access to the globality of the molecular mechanisms of oncogenesis, by a complete sequencing of the exome (Whole exome sequencing) which enables the identification of molecular profiles and recurrent mutations of nosological, prognostic or theranostic interest, to assess the tumor and dynamic genetic heterogeneity, which is at the origin of the secondary resistance to treatments. In a trial presented at ASCO in 2014 by Lim et al., the results of full tumor sequencing were compared with targeted sequencing (AmpliSeq panel) on 56 patients with advanced breast, lung, and colorectal cancers. Whole genome sequencing was more informative than targeted sequencing (70% of cases compared with 30% of cases) [41]. However, in the MOSCATO-01 trial, whole exome sequencing detected mutations in 8 patients who had no alterations on targeted sequencing and in situ hybridization tests, which represents less than 10% of the patients included and suggests that targeted sequencing can ensure detection of the most common alterations (Table 4) [37,42]. High-throughput sequencing technologies are not part of international recommendations, although they are performed on a clinical routine basis in many centers, and similar observational trials are appearing, such as the study conducted in Paris by the team of Bayle et al. [43]. Prospective multi-center studies on larger series are considered and will allow a more significant analysis of the impact of precision medicine on patient prognosis. The current challenge, supported by learned societies and the France Genomics 2025 Plan, is to improve the effectiveness of precision medicine, its indications, its accessibility and its use, in order to extend the number of patients treated and have an impact on survival.

Conclusion
Although our study did not find an overall survival impact of using a targeted therapy on an actionable molecular abnormality, the mutational panel of digestive cancers is large, and these results pave the way for the future of prospective clinical trials guided by molecular profiling. Only prospective, therapeutic interventional trials by molecular or tumor alteration subgroups will definitively demonstrate the benefits of these therapies and are the next logical step in clinical trials for patients with digestive cancer.

Ethical Approval
The study was conducted in accordance with the good clinical practices of the International Conference on Harmonization and the recommendations of the Declaration of Helsinki.

Informed Consent
Written informed consent was obtained from all for the DNA/RNA sequencing analysis.