The present study was devised to evaluate a large series of microdeletions at exon 19 of the EGFR gene by the 454 GS Junior system. This new technical approach is based on the amplification of single DNA molecules by emulsion PCR giving rise to a sort of chemical cloning before pyrosequence analysis. Since expanded wild type and mutant alleles are examined separately, the method can allow an accurate evaluation of mutations without the interference of wild type alleles as it occurs when conventional sequencing methods are used. The results of NGS were compared with those obtained by conventional SS. In about 12% of the samples analyzed, SS failed in finding the accurate sequence of deleted bases, either using a dedicated software or manually. In particular, in 6 cases, the starting-ending point of the deletion point was not correctly detected. This can be ascribed to the fact that, by using Sanger sequencing, the wild type and mutated alleles are superimposed on the same electropherogram. In some samples, it can happen that the alignment of the two alleles is not perfect, as shown in , so that the operator can misinterpret the starting-ending point of the deletion. In other 7 cases, the starting-ending point of the deletions were clear but the exact sequence of the mutant alleles was partially obscured (not assessable).
An accurate detection of EGFR
microdeletions is highly recommendable, in that different deletions could have different effect on tumor development and progression or patient outcome after treatment with EGFR TKIs. Moreover, different deletion may have specific effects on the antigenicity of proteins carrying deletion. In our study, we have tested this hypothesis by evaluating the possibility to detect EGFR
deletions with specific monoclonal antibodies in tumors with different deletions accurately characterized by NGS. We have shown that the 6B6 anti-mutant EGFR antibody strongly reacts to EGFR proteins carrying the E746_A750del, whereas the immunoreaction was absent in tumors affected by other types of deletions. Our data are in agreement with previously published data obtained with this antibody in larger series of NSCLCs 
. However, due to the limited number of cases examined in these reports, additional studies will be required to definitely clarify this point. Our results suggest that the development of new monoclonal antibodies or cocktails of antibodies would require the exact knowledge of the deletions and that NGS could be a very accurate and reliable technique to address this point.
The alignment of the numerous sequences obtained by NGS by dedicated softwares, allowed to formulate a new interpretative hypothesis on the nature of particular EGFR
deletions. A series of frequent (20% of cases) complex deletions, most of which reported as deletions associated to insertions in previous reports 
, may also be ascribed to the presence of non-in frame double or multiple deletions producing a net in-frame loss of genetic material. This interpretation is in keeping with the hypothesis that a short region within exon 19 is particularly fragile and preferentially subjected to microdeletions. The frequency of these complex mutations was statistically higher in cases with longer (18 bp) losses: about 80% of double/multiple deletions resulted in a loss of 18 bps. Five novel mutations were observed by NGS in this study and all of them were complex, double/multiple mutations, that in 3 (60%) cases were not resolved by Sanger Sequencing. This clearly confirms the superiority of NGS in the characterization of EGFR
microdeletions at exon 19. Rare cases of double/multiple deletions have been reported in previous studies, especially when conventional sequencing was conducted on multiple samples after cloning of genomic DNA into plasmids 
. Cloning was essential to better characterize multiple deletions and rule out the possibility that they were present on different alleles. However, biological cloning and sequencing of multiple samples is time consuming and not suitable for routine clinical diagnostic purposes. The GS Junior Technology, based on PCR cloning before pyrosequencing of multiple samples, is in our opinion an ideal approach for a fine characterization of complex deletions in exon 19 of the EGFR
The NGS assay is one of the most sensitive methods available for the detection of somatic mutations when used in deep sequencing, and the sensitivity of NGS is dependent on the number of sequences obtained per sample 
. In this study we decided to perform a deep NGS analysis taking a median of more than 3.000 sequences per sample. The high sensitivity of this NGS assay allowed us to detect in about 70% of cases, subpopulations of DNA molecules carrying exon 19 deletions different from the main mutation, but in most cases structurally related to it (). In the majority of cases these subpopulations carried in frame simple or double deletions. In about one third of cases, less expanded non-in frame deletions were observed. These subpopulations were confirmed in independent PCR-NGS assays in 43% of the tumors investigated. However, only 4 tumors (4%) showed substantial subpopulations of deletions with at least one of them representing more than 2% of genomic DNA.
The presence of these subpopulations in a large proportion of cases examined is at moment unclear. It is extremely unlikely that they are due to cross-contaminations, since they were highly heterogeneous, usually related to the main deletion, and they were not seen in control samples. In addition, particular strategies were adopted to minimize cross-contaminations in our study (see Materials and Methods
). These subpopulations could represent modifications of the main deletion or ex novo deletions acquired in cell clones during tumor progression. In both cases, this finding would support the hypothesis that this region within exon 19 of EGFR
is highly instable in most patients affected by NSCLC carrying EGFR
deletions. This genetic fragility may enable the development of both complex deletions and multiple subpopulations. To the best of our knowledge, this is the first demonstration of subpopulations of EGFR
deletions in NSCLC. Our findings could have important clinical implications. Recent evidence indicate that tumor genotype may evolve dynamically under the selective pressure of targeted therapies 
. We are tempted to hypothesize that the genetic fragility of exon 19 in particular patients could take a role in the dynamic evolution of lung tumors subjected to different therapeutic strategies. It would be interesting to monitor by NGS on repeated biopsies the main EGFR
deletion as well as the deletions in minor clones during treatment.
In conclusion, our results indicate that NGS is particularly suitable for the study of EGFR deletions. This technique can accurately characterize EGFR deletions, even in cases in which conventional methods fail. Data obtained by NGS analysis allowed us to formulate a new interpretative hypothesis for complex deletions which represent about 20% of EGFR mutations in exon 19 as well as to identify 5 novel deletions. In addition, we report, for the first time to our knowledge, the presence of subpopulations of different deletions in most of the tumors investigated with pothential pathogenetic and clinical impact.