Lung cancer, specifically lung adenocarcinoma (ADC), is frequently diagnosed in an advanced stage with a global 5-years survival not exceeding 17% [1–3]. Even when it is detected at an early stage, the prognosis is poor, especially in terms of tumor recurrence [4, 5]. Pulmonary ADC has an extraordinarily high mutational burden and somatic genomic alterations can be found in more than 75% of the cases. Fortunately, a vast proportion of the oncogenic driver alterations has potential therapeutic implications [6]. The acquisition of driver mutations in still histologically non-tumoral cells appears to be very important during the ADC carcinogenesis since it will potentially lead to a clonal cell expansion [7]. Moreover, two studies have demonstrated the presence of cancer-related mutations in non-tumoral cells of subjects with endometriosis and arteriovenous malformations of the brain [8, 9]. These findings demonstrate that cancer-related mutations may not necessarily be present only in malignant cells, but also in histologically non-tumoral cells. Our group has recently demonstrated that subjects with localized lung ADC with epidermal growth factor receptor-EGFR or KRAS proto-oncogene alterations showed the same driver mutation in the non-tumoral lung tissue as far as a 21.3% of the cases [9]. These findings were associated with a significantly lower disease-free survival at 12 months and are in line with other previously published evidence related with molecular alterations in non-tumoral lung [10]. Our hypothesis is that cancer-related mutations can appear in histologically non-cancerous cells even in the absence of molecular alterations in the primary tumor during the field cancerization process. Accordingly, the aim of the present study was to demonstrate the presence of EGFR or KRAS mutations in non-tumoral lung cells even in subjects with early-stage ADC with negative mutational testing.

The main clinical, functional and tumor characteristics of the cohort are shown in Table 1. All subjects were stratified in stages based on the present TNM classification (IASLC, 8th edition) and only stage I or II subjects were included [3]. The surgical procedures were performed in accordance with the institution clinical-practice recommendations. The most common procedure was a lobectomy (67.7%) followed by segmentectomy (22.6% and bilobectomy (9.7%). Almost all the subjects (30 of 31, 96.8%) were smokers or former smokers.

We identified five mutations in EGFR or KRAS oncogenes in the normal lung parenchyma among three subjects (9.7%). The exon 21 substitution L858R in EGFR was detected in two cases while the exon 19 deletion E746-A750 in the EGFR, the codon 12 substitution G12C and G12D in the KRAS were detected once. Surprisingly, in one patient, three different mutations were identified in NLP (EGFR_L858R, KRAS_G12C and KRAS_G12D). More details of the three subjects with mutated NLP can be found in Table 2.

In these three subjects, the negative-mutation status of the tumor for the specific mutation detected in the NLP was confirmed using highly sensitive and specific TaqMan PCR (CAST-PCR). The confirmation of the positive-mutation status in the NLP was also performed for the previously mentioned five specific mutations. In all the assays, the PCR efficiency was between 95 and 105%. To improve specificity and avoid false positives we only considered as positives the assays for EGFR mutations when the amplification occurred before the cycle 35. For the two KRAS mutations detected, the amplification occurred between the cycle 35 and 38. The mutation was confirmed in both cases. The amplification plots are shown in Figure 1. Real-time PCR results for positive cases were confirmed in independent experiments. We have also performed digital PCR (dPCR) to confirm the presence of the mutations reported and, in addition, with the samples with enough material, we have repeated the qPCR assays.

This study confirms the presence of driver mutations in the histologically normal lung parenchyma cells coexisting with the absence of mutations in the primary tumor. Our findings are consistent with the hypothesis that during the carcinogenesis process multiple cells can gain somatic mutations without necessarily producing a clonal expansion. In a previous study, we confirmed that the same driver mutation detected in the lung adenocarcinoma was also present in non-tumoral samples in a fifth of the subjects. However, this study has the limitation that this detection could be secondary to contamination by tumor DNA from blood or tumor cells, which may not have been detected by the usual histopathological methods. Although this limitation was unlikely, it could not be one hundred percent ruled out. In the present this study, having excluded the presence of mutations in EGFR or KRAS by highly specific and sensitive techniques in the primary tumor, we are able to confirm that the mutations detected in other lung cells were not the result of contamination from the tumor. These findings make us change the way we understand and define a driver-mutation.

The prevalence of EGFR and KRAS mutations detected in the present study is only 9.7%, lower than the one detected in our previous study (21.3%) and in those studies on endometriosis (26%) and arteriovenous malformations of the brain (48%). The prevalence observed in the present study is likely to be higher if more extensive molecular studies that included other molecular alterations were carried out. Additionally, to stress our ambitious hypothesis, we decided to increase the specificity not including the mutations occurred in later cycles of PCR as well as to exclude the substitution of T790M in exon 20 that are usually considered as secondary mutations in the final analysis.

In the present study, we could not find differences in clinical outcomes such as recurrence or disease-free survival, between the two groups, possibly because of the sample size. One patient (33.3%) in the group of the mutated NLP presented progression at 18 months in the form of hepatic metastases; while in the non-mutated NLP group 8 (28.6%) did, not reaching however statistical significance. Subjects in the mutated NLP group were significantly less-frequent smokers and predominantly women when compared with the non-mutated NLP group, this is in line with the evidence that is consistent with the higher prevalence of driver-mutations in non-smoker women.

Interestingly, one patient presented three different mutations in the NLP sample wich included one in the EGFR and two in the KRAS. Normally the mutations in EGFR and KRAS are considered as mutually exclusive mutations in lung cancer. Although this finding is surprising, it does not seem improbable either, since, unlike a tumor, where all the cells come from the clonal expansion of a single malignant cell, the normal lung parenchyma samples may contain multiple different cell lines. Our results support our prior hypothesis that molecular changes can occur in multiple cells even without malignancy changes.

There are multiple limitations in the present study, mainly the small sample size, although this limitation does not affect the main objective of the study, which was to demonstrate the presence of driver-mutations in the absence of tumor mutations. The sample size surely affects clinical outcomes, which was not the objective of the study. Contamination or sample mix-up of the studied samples with tumor cells that contain molecular alterations has always been a topic of discussion, but we consider it highly unlikely for several reasons. First, by ruling out molecular alterations in the tumor, contamination would be much less relevant. Second, a “non-tumoral lung” sample was considered when resection margins were tumor-free assessed by an expert pathologist and the area of lung farthest from the tumor was selected for the molecular analysis.