Contribution of the IdyllaTM System to Improving the Therapeutic Care of Patients with NSCLC through Early Screening of EGFR Mutations

Epidermal growth factor receptor (EGFR) genotyping, a critical examen for the treatment decisions of patients with non-small cell lung cancer (NSCLC), is commonly assayed by next-generation sequencing (NGS), but this global approach takes time. To determine whether rapid EGFR genotyping tests by the IdyllaTM system guides earlier therapy decisions, EGFR mutations were assayed by both the IdyllaTM system and NGS in 223 patients with NSCLC in a bicentric prospective study. IdyllaTM demonstrated agreement with the NGS method in 187/194 cases (96.4%) and recovered 20 of the 26 (77%) EGFR mutations detected using NGS. Regarding the seven missed EGFR mutations, five were not detected by the IdyllaTM system, one was assayed in a sample with insufficient tumoral cells, and the last was in a sample not validated by the IdyllaTM system (a bone metastasis). IdyllaTM did not detect any false positives. The average time between EGFR genotyping results from IdyllaTM and the NGS method was 9.2 ± 2.2 working days (wd) (12.6 ± 4.0 calendar days (cd)). Subsequently, based on the IdyllaTM method, the timeframe from tumor sampling to the initiation of EGFR-TKI was 7.7 ± 1.2 wd (11.4 ± 3.1 cd), while it was 20.3 ± 6.7 wd (27.2 ± 8.3 cd) with the NGS method (p < 0.001). We thus demonstrated here that the IdyllaTM system contributes to improving the therapeutic care of patients with NSCLC by the early screening of EGFR mutations.


Introduction
Lung carcinoma remains the most common cause of cancer death worldwide. Nearly 85% of lung cancers are non-small cell lung cancer (NSCLC), and 15% are small cell lung cancer (SCLC). Accounting for approximately 50% of cases of NSCLC, adenocarcinoma (ADC) is the most prevalent histological subtype [1].

Patients from the ID-MUT Study and Paraffin-Embedded Specimens
From January 2019 to August 2020, 223 patients with NSCLC diagnosed by pathologists from Caen University Hospital (CHU; n = 79) and the François Baclesse Center (CFB; n = 144) were routinely tested to evaluate EGFR mutations with both reference methods, i.e., NGS and the Idylla TM system (Table 1). All tumor samples were fixed in formalin for 6 to 48 h: (i) "biopsy" type samples (including bone samples) were addressed to the pathology department from the Caen University Hospital or from the CFB in formalin buffered at 4%, (ii) the "fine needle aspiration" type samples were addressed to the pathology departments in a fixing liquid (CytoRich TM , BD, Le Pont de Claix, France); the cells were pelleted by centrifugation then incubated in 4% buffered formalin. Following the fixation step, the bone samples undergwent an additional decalcification step by being incubated in a buffered EDTA solution (Osteosoft TM , Merck, Germany). After fixation (and decalcification for bone samples), the tumor samples were embedded with paraffin. Sections of paraffin-embedded specimens for EGFR mutation testing by NGS method were performed before those for EGFR mutation testing by the Idylla TM system. At the end of the two series of sections, a slide for hematoxyllin, eosin, saffron (HES) staining and morphological verification of the residual tumor material was systematically carried out.
The clinical data were recovered from electronic medical records, including the date of (1) tumor sampling, (2) multidisciplinary consultation meetings, (3) diagnosis announcement, and (4) initiation of anti-EGFR therapy.
Specific informed consent was obtained for the biological study (ID-MUT NGS was performed using S5 Prime (Thermo Fisher Scientific). The average depth was >500×; on target >90%. Bioinformatic analyses (alignment, call of variants, and annotations) was run on LifeTechnologies: Torrent suite 5.10, Variant caller 5.10, Ion reporter 5.10-Nextgene (Softgenetics, State College, Pennsylvania, USA) 2.4.1.2. The copy number variant (CNV) analysis was expressed as the ratio of mean depths by amplicons ±2 standard deviations. The detection limit was set to 3% for punctual mutations and 5% for insertions/deletions for a minimum depth of 100× per amplicon. Variations of the sequences recognized as non-pathogenic (classes 1 and 2) were not mentioned. The allelic frequency of variants (VAF) of an alteration was evaluated, including panel CLv3 sequence exons 18-21 of the EGFR gene.

EGFR Mutation Assay by the Idylla TM System
EGFR mutation testing by the Idylla TM system was centralized and carried out every working day in the Department of Pathology from the CHU de Caen from unextracted paraffin-embedded specimens according to an adaptation of the manufacturer's recommendation. Briefly, three sections of 20 µm thickness from the paraffin-embedded specimens were loaded into the Idylla TM EGFR Mutation cartridge for the following fully automated test, previously described by others [32]. The tumor sample had to contain at least 10% of tumor cells; a macrodissection was carried out to enrich the sample with tumor cells when necessary. The EGFR mutations detected in the Idylla TM EGFR Mutation cartridge are listed in Supplementary Table S1. PCR curves were visualized through the web-based interphase Idylla TM explore to evaluate the quantification cycle (CQ) from the mutation signal if present, the CQ for the internal control (total EGFR) signal, the difference between the two CQs (∆CQ), the sigmoid aspect of the amplification curve of the mutation when found, and the maximal fluorescence, similarly to Momeni-Boroujeni et al. [32]. The total EGFR CQ for all samples ranged from 16 to 26.

Statistical Analysis
Sensitivity and specificity were the proportion of concordant results against the sum of concordant and discordant results (true positives/(true positives + false negatives) and true negatives/(true negatives + false positives), as detailed by [24]. The male/female distribution, smoking status, and age of patients between NSCLC patients with or without EGFR mutation were evaluated using a chi-square test and a non-parametric test for unpaired data from Mann-Whitney. The influence of the used method (Idylla TM or NGS) on the time required for delivering EGFR genotyping results was evaluated by a two-way (techniques and time) analysis of variance (ANOVA), followed by a post-hoc Bonferroni test (GraphPad Prism version 8.0.0 Software, San Diego, California, USA).
Statistical differences of the timeframe according to the method used for EGFR genotyping (Idylla TM or NGS) were determined using a t-test (GraphPad Prism version 8.0.0 Software, San Diego, California, USA). Statistical significance was set at p ≤ 0.05.

Patients' Characteristics
In total, 223 patients (126 men and 97 women; medium age at diagnosis: 65.4 ± 9.8 years old) newly diagnosed with NSCLC, mostly adenocarcinomas or probable adenocarcinomas (83.1%), were enrolled in the ID-MUT study from January 2019 to August 2020 (Table 1, Figure 1).
internal control (total EGFR) signal, the difference between the two CQs (ΔCQ), the sigmoid aspect of the amplification curve of the mutation when found, and the maximal fluorescence, similarly to Momeni-Boroujeni et al. [32]. The total EGFR CQ for all samples ranged from 16 to 26.

Statistical Analysis
Sensitivity and specificity were the proportion of concordant results against the sum of concordant and discordant results (true positives/(true positives + false negatives) and true negatives/(true negatives + false positives), as detailed by [24]. The male/female distribution, smoking status, and age of patients between NSCLC patients with or without EGFR mutation were evaluated using a chi-square test and a non-parametric test for unpaired data from Mann-Whitney. The influence of the used method (Idylla TM or NGS) on the time required for delivering EGFR genotyping results was evaluated by a two-way (techniques and time) analysis of variance (ANOVA), followed by a post-hoc Bonferroni test (GraphPad Prism version 8.0.0 Software, San Diego, California, USA).
Statistical differences of the timeframe according to the method used for EGFR genotyping (Idylla TM or NGS) were determined using a t-test (GraphPad Prism version 8.0.0 Software, San Diego, California, USA). Statistical significance was set at p ≤ 0.05.

Patients' Characteristics
In total, 223 patients (126 men and 97 women; medium age at diagnosis: 65.4 ± 9.8 years old) newly diagnosed with NSCLC, mostly adenocarcinomas or probable adenocarcinomas (83.1%), were enrolled in the ID-MUT study from January 2019 to August 2020 (Table 1, Figure 1).

Figure 1.
Flow chart of the ID-MUT study and concordance between the Idylla TM system and NGS genotyping for EGFR. Figure 1. Flow chart of the ID-MUT study and concordance between the Idylla TM system and NGS genotyping for EGFR.
We excluded from this study: (i) patients not then treated at the CHU or CFB (n = 14), (ii) samples with insufficient material (tumor sample of less than 50 mm 2 and/or without sufficient tumor cells (less than 10% of tumoral cells); n = 6), (iii) a malpighia metaplasia sample (n = 1), and (iv) samples without results for both the Idylla TM and NGS tests (n = 8) (Figure 1).
However, seven EGFR mutations were detected by NGS but not by Idylla TM ; five of these missed mutations were not assayed by the Idylla TM system and were therefore not true mismatches between Idylla TM and NGS. The two other missed mutations were mutations evaluated by Idylla TM . However, one of them was missed probably because, following the scraping of the FFPE block for Idylla TM to analyze, only 5% of the tumor cells remained in the sample (the sensitivity threshold of the Idylla TM technique is 10% of tumor cells), and the deletion of the exon 19 of the EGFR reported by NGS had a low allelic frequency of 2.80%. The second missed mutation was from a bone metastasis sample of an undifferentiated carcinoma presumed to be of pulmonary origin, because it was TTF1-positive. It should be noted that the Idylla TM system has not been certified for such type of sample.

Consideration of the EGFR Genotyping by the Idylla TM Method in the Treatment Decision
Among the 194 patients with EGFR mutations diagnosed by the combo "Idylla TM /NGS", 158 patients received first-line systemic treatment. For these 158 patients, EGFR genotyping results by Idylla ® were all reported before those by NGS. For the majority of these patients, i.e., 118/158 of them (75%), the multidisciplinary consultation meeting leading to the therapeutic decision took place with the knowledge of EGFR genotyping by the Idylla TM method and without the knowledge of the result of the analysis by NGS. For 23/158 patients, EGFR genotyping results were known (10/23 EGFR genotyping results were based only on the Idylla TM method, 13/23 EGFR genotyping results were known from both NGS and Idylla TM methods). Finally, 17/158 patients (11%) were discussed in the multidisciplinary consultation meeting before the results of the mutation status of EGFR.

Turnaround Time (TAT)
To appreciate the TAT of EGFR genotyping from the tumor sample to the initiation of EGFR-TKI treatment according to the NGS or Idylla TM method, we then measured the timeframe between (1) tumor sampling and EGFR genotyping request, (2) EGFR genotyping request and result, (3) tumoral sampling and EGFR genotyping result, (4) results from both techniques, and (5) tumoral sampling and initiation of treatment ( Figure 2).  We expressed these time periods in both working (wd) and calendar (cd) days, calendar days being more representative for patients.
The TAT from tumor sampling to EGFR genotyping request was comparable whether genotyping was ordered from Idylla TM or performed using the NGS method, with a timeframe of 5.3 ± 3.6 wd (7.4 ± 5.1 cd) and 5.6 ± 3.7 wd (7.8 ± 5.1 cd), respectively (Figure 2A). By contrast, the TAT from the EGFR genotyping request to the EGFR genotyping results was almost six times faster with Idylla TM than with the NGS method, with a timeframe of 1.5 ± 1.3 wd (1.9 ± 1.8 cd) and 10.3 ± 2.0 wd (14.2 ± 3.5 cd), respectively (p < 0.001) (Figure 2A). The average time between EGFR genotyping results by Idylla TM and the NGS method was thus 9.2 ± 2.2 wd (12.6 ± 4.0 cd) (Figure 2A). All these TAT (timeframe between (1) tumor sampling and EGFR genotyping prescriptionrequest, (2) EGFR genotyping prescription request and result, (3) tumoral sampling and EGFR geno-typing result, (4) results from both techniques, and (5) tumoral sampling and initiation of treatment) were comparable between the CHU and the CFB (p > 0.05). Figure 2B,C illustrates the availability of EGFR genotyping since the tumor sampling in calendar days ( Figure 2B) and working days ( Figure 2C) from the point of view of patients (calendar days) and practitioners (working days), respectively. As shown, for half of the patients from the ID-MUT study, the EGFR genotyping result was determined to be 6-7 cd (4-5 wd) after the tumor sampling with the Idylla TM method, against 19-20 cd (13-14 wd) for the NGS method. Similarly, for 90% of the patients from the ID-MUT study, the EGFR genotyping result was determined to be 15-16 cd (10-11 wd) after tumor sampling with the Idylla TM method, against 23-24 cd (19-20 wd) for the NGS method. The influence of the "technique" used (Idylla TM or NGS) on the time required to deliver EGFR genotyping results was evaluated by a two-way (technique and time) analysis of variance (ANOVA), followed by a post-hoc Bonferroni test, which confirmed statistically the time-saving benefit of the Idylla TM technique whether evaluated in cd (p = 0.002) or wd (p = 0.0017).
During the first part of the ID-MUT study (the first nine months), in one of the centers (Caen University Hospital), the EGFR genotyping results by both the Idylla TM and NGS methods were expected to initiate EGFR-TKI treatment, the time for the confirmation of EGFR genotyping concordance between the two methods in our hand. During the second part of this study, EGFR-TKI treatment was initiated on EGFR genotyping by Idylla TM in both Caen University Hospital and the François Baclesse Center. For this reason, among the 22 NSCLC patients from the ID-MUT study with EGFR mutations treated by EGFR-TKI, 12 NSCLC patients harboring EGFR mutations received treatment following EGFR genotyping by the NGS method, while the other 10 patients with NSCLC and EGFR mutations received EGFR-TKI treatment according EGFR genotyping by the Idylla TM system (i.e., before the EGFR genotyping result by NGS was known), allowing the evaluation of the contribution of the Idylla TM system to improving the therapeutic care of patients with NSCLC by early screening of EGFR mutations (Table 2). Table 2 details the timeframes for the available EGFR genotyping according to the Idylla TM or NGS method for each of the 22 patients, as well as the time required for the initiation of EGFR-TKI treatment since the tumor sample was processed.
The time to initiation of EGFR-TKI was defined as the time between the interventional procedures leading to the histological confirmation until initiation of TKIs for patients harboring EGFR mutations. The TAT from tumor sampling to initiation of EGFR-TKI was 7.7 ± 1.2 wd (11.4 ± 3.1 cd) when the decision was based on the Idylla TM method, while it was 20.3 ± 6.7 wd (27.2 ± 8.3 cd) when the decision was based on the NGS method, i.e., reduced by more than two-fold with the Idylla TM system (p < 0.001).

Discussion
In this study, we confirmed the good sensitivity and specificity of the rapid detection of EGFR mutations using the Idylla TM system and mainly reported that EGFR mutation detection with this assay is associated with a significantly reduced turnaround time compared to the use of NGS testing. In turn, patients were observed to begin systemic EGFR-TKIs therapy an average of two weeks earlier than if waiting for the NGS result.
The good sensitivity and specificity of the rapid detection of EGFR mutations using the Idylla TM system were previously reported in other publications on lung cancer, as well as on other cancers such as melanoma and colorectal cancer [17]. Similarly, as detailed in the introduction, the concordance of EGFR genotyping has already been reported between the Idylla TM system and NGS [12][13][14][15][16][17][18][19][20][21][22] or other techniques [23][24][25][26][27][28][29][30]. In our center, we chose to introduce the Idylla TM system rather than another rapid assay, because our main objective was to reduce, at maximum, the result of the EGFR genotyping for clinicians and to reduce the need of DNA extraction being performed daily, not allowing this objective to be achieved. In fact, the Idylla TM system is one of the rare solutions avoiding DNA extraction and allowing reliable EGFR genotyping directly from FFPE sample slides [20,25,[33][34][35] in all patients with advanced NSCLC, mainly adenocarcinomas and squamous cell carcinomas in never-smokers, as in patients with stage IB to IIIA resected EGFR mutation-positive NSCLC [8]. However, the Idylla TM system has several limitations. First, biological materials obtained from biopsies are often limited and can be an issue when multiple analyses are required. Indeed, the Idylla TM assay requires additional sections of FFPE sample and therefore risks exhausting the tumor sample, especially because, contrary to what is recommended by Biocartis, which markets the Idylla TM EGFR cartridge, we did not perform EGFR genotyping on a single section of 5 µm but on three sections of 20 µm thickness from the paraffin-embedded specimens, because, in our hands, during preliminary tests, we observed that, by following the recommendations of Biocartis, there was a risk of missing out on an EGFR mutation with low allelic frequency. This did not put the performance of the tests at risk because no analysis failure due to a saturation of a cartridge was reported. As we have an excellent agreement (96.4%) of results between the methods by the Idylla TM system and NGS, we concluded that our procedure for EGFR mutation testing by the Idylla TM system allowed us not to miss a mutation due to lack of sensitivity for the Idylla TM method. The off-label use of CE-IVD methods was thoroughly validated before being used in routine testing during a retrospective study not reported here. The risk to exhaust the tumor sample could, however, be lifted by the reuse of H&E, immunohistochemistry (IHC), and fluorescence in situ hybridization (FISH) diagnostic slides [36], by the use of plasma from patients with lung cancer [37,38] or of the DNA extracted from FFPE sections for NGS analysis [14,19,26,28,39]. Indeed, using the same DNA for Idylla TM and NGS assays could discard divergent results linked to tumor heterogeneity and to the fact that the two analyses are not carried out on strictly the same part of the tumor sample. Moreover, only a little DNA is needed; Bocciarelli et al. reported that >25 ng of DNA and >10% of tumor cells are sufficient to detect EGFR mutations with the Idylla TM method [19]. DNA use seems a good alternative, however, and this is also the second limitation of the Idylla TM system-Idylla TM EGFR cartridges are not certified for samples other than primary tumor biopsies included in FFPE. This lack of certification is regrettable, but like others [13,27,30,40], we also analyzed samples of other kinds (fine needle aspiration and metastasis of bronchial origin) on Idylla TM EGFR cartridges and, except for one of them, the results of EGFR genotyping were consistent with the analysis by NGS. As long as the result of the EGFR genotyping is confirmed in a second step by another technique, it seems to us that this second limitation can therefore be avoided.
Studies comparing the Idylla TM system and NGS performance [12][13][14][15][16][17][18][19][20][21][22] reported a concordance between the two techniques ranging from 94% to 100%, which is consistent with the 96.7% of concordant results between the Idylla TM system and NGS that we evaluated here, especially because the seven apparent discordant results that we reported were not trues discordant: Five missed mutations were not detected by the Idylla TM system, one missed mutation was assayed in a sample with insufficient tumoral cells, and the last missed mutation was sought in a sample not validated on the Idylla TM system (a bone metastasis).
That some mutations detected in the NGS panel were absent in the Idylla TM test panel is unfortunate, but most of the time there are few consequences for the patient considering that approved therapies are missing for most of those missed mutations (insertion in exon 20 of the EGFR gene), and the others were finally found with the complementary analysis of the sample by NGS, such as the C797S mutation, a second acquired resistance mutation arising in tumors that have progressed after treatment for T790M+ disease and not detected in the Idylla TM system [41]. For us, the Idylla TM system or another rapid system of EGFR mutation detection is essential for reducing the timeframe of EGFR genotyping and initiating therapy in patients with lung cancer. However, it cannot be the only analysis realized to process this genotyping because of the risk of missing some EGFR mutations, either in samples with few tumor cells or rare EGFR mutations not detected by rapid genotyping, but for which we will soon know whether they do or do not predict the response to EGFR-TKIs, and because NGS allows the analysis of a large panel of genes whose alterations (mutations and copy gain) can also guide the treatment decisions of patients with lung cancer. Considering the simultaneous evaluation of numerous genomic alterations across several genes with NGS, and even if the system is presented as being available in any laboratory, because it does not require a molecular biologist, we believe that the links between these laboratories and platforms equipped with NGS technology must be preserved. NGS panels remain essential in molecular sub-type diagnosis of lung cancer and cannot be replaced due to the rapid emergence of new targeted therapies for different genomic alterations. Therefore, it can screen mutations that allow some patients to be included into clinical trials. Molecular testing is also essential in the treatment strategy because studies have demonstrated that immunotherapy before targeted therapies increases the occurrence of serious side effects [42][43][44]. While comprehensive molecular screening is essential in academic centers with access to clinical trials, it is questionable in smaller centers who do not have access to NGS assays. In those centers, Idylla TM assays can be part of the solution to improving the time to initiate therapies.
Molecular testing requires a good-quality sample, enough tumor cells, and even multiple interventional procedures to be conclusive, which would lengthen delays. In 2016, a survey from the French National Cancer Institute showed that the median turnaround time (TAT) from test prescription to reception of results by the clinician for EGFR molecular test was 18 days [7]. A limitation with the NGS technique is that the TAT is usually longer than the TAT associated with a specific assay. However, a delayed turnaround time for biomarker reports can lead to delays in treatment initiation, decreased efficacy of treatment, and inappropriate treatment decisions. Indeed, EGFR mutations are an oncogenic driver occurring especially in patients with pulmonary adenocarcinoma and never-smokers. For patients with metastatic NSCLC harboring EGFR mutations, the development of EGFR tyrosine kinase inhibitors (TKIs) is an important improvement in therapeutic care, as shown by the increase in the progression-free survival (PFS) and limitation of toxicities using EGFR-TKIs for patients with EGFR-mutated NSCLC compared to chemotherapy [5], especially with the use of osimertinib versus first-line TKIs (gefitinib and erlotinib) [4,6]. However, a long delay to initiate EGFR-TKIs can result in rapid disease progression and deterioration in performance status associated with a worse prognosis [45]. Thus, it is essential to accelerate the availability of molecular sub-type results. Because pathology processing is reduced with rapid techniques, we were able obtain a result for EGFR genotyping 12.5 calendar days earlier with Idylla TM compared to NGS assays. As a consequence, the TAT from tumor sampling to initiation of EGFR-TKIs was reduced by two weeks when the decision was based on the Idylla TM method compared to when the decision was based on the NGS method. Besides faster delivery of the appropriate treatment for patients with NSCLC, therefore increasing their chance of survival, we can also assume that by improving the deadlines, we can improve patients' adherence to participating in clinical trials.

Conclusions
This study demonstrated, for the first time and to the best of our knowledge, the benefit for the patient of the introduction into routine practice of the rapid EGFR genotyping test, in addition to NGS in the initiation of its therapeutic care. EGFR mutation assays by the Idylla TM system, in addition to NGS testing, increase the costs of patient care but improve it through the timely completion of biomarker results and the facilitation of appropriate treatment decisions [46].