This was a prospective observational study evaluating the prognostic significance of ctDNA, such as exon 19 deletion, L858R, T790M, and C797S EGFR mutations, in patients with previously treated NSCLC harboring T790M EGFR mutation who received osimertinib. The detection of ctDNA in plasma samples by digital PCR exactly reflected the change in exon 19 deletion, L858R, and T790M at 1 month and PD after osimertinib administration. We found that the detection of T790M at PD after osimertinib initiation was a significant independent prognostic factor for predicting worse OS, and that the detection of major EGFR mutations at pretreatment and T790M at PD also affected the outcome after osimertinib initiation. Currently, osimertinib is clarified as a first-line EGFR-TKI; thus, the frequency of its administration as a second-line or more lines is decreasing. However, the detection of blood samples using digital PCR is a noninvasive technique that is convenient for daily practice, and an improvement in the detection rate is warranted to realize the development of an optimal predictor.
The analysis of EGFR mutation in tissue and plasma from the AURA3 trial reported that the detection of plasma T790M was related to a larger baseline tumor size, and PFS after osimertinib (median, 12.5 vs. 8.3 months) was longer in patients with a Cobas plasma T790M-negative status at pretreatment compared to those with a plasma T790M-positive status [14]. PFS in patients with NSCLC harboring T790M who received osimertinib tended to be worse in patients with high T790M copy number at pretreatment, as assessed by cell-free plasma DNA using digital, rather than those with low T790M copy number. Moreover, no significant difference in ORR was recognized according to T790M copy numbers [15]. Although another study also assessed the association between the utility of plasma T790M mutant copy number at pretreatment and the prognostic role of osimertinib using digital PCR, a high plasma T790M copy number was associated with worse PFS and OS compared to a low T790M copy number [16]. In addition, a high copy number of active EGFR mutations in plasma samples has been reported to be closely associated with poor outcomes after osimertinib treatment [11]. The results of these studies indicated that a high copy number of plasma T790M or activating EGFR mutations was a useful marker for predicting worse PFS in patients with NSCLC harboring positive EGFR T790M who were treated with osimertinib [11, 14–16]. Ding et al. reported that high plasma T790M level at pretreatment was related to superior disease control in patients with NSCLC with advanced EGFR T790M treated with osimertinib [10]. In our study, the detection of major plasma EGFR mutation (exon 19 deletion or L858R) at pretreatment could predict shorter PFS after osimertinib administration, but not plasma T790M copy number. Although previous studies have focused on the association between plasma EGFR mutation copy numbers and the efficacy of osimertinib at pretreatment, Sakai et al. evaluated the clinical significance of monitoring the ctDNA of EGFR-TKI-sensitizing mutations and EGFR T790M mutation in EGFR T790M-positive patients with NSCLC at pretreatment, on day 1 of treatment cycle 4 or 9, and at the diagnosis of PD using digital PCR [17]. In their study, rebound of sensitizing EGFR mutation and T790M was observed at PD, and ctDNA monitoring for sensitizing EGFR mutation at four cycles was better for predicting the outcome after osimertinib [17]. We found that the detection rate of ctDNA by digital PCR significantly decreased 1 month after osimertinib initiation, and that the detection of plasma T790M at PD was closely associated with worse outcomes. Although ctDNA monitoring in the early phase after osimertinib may not be useful for prognostic prediction, the detection of major EGFR mutation at pretreatment is predictive of PFS after osimertinib based on previous studies.
Several studies have investigated the prognostic significance of the ratio of T790M to major EGFR mutation at baseline in patients with advanced EGFR T790M-positive NSCLC receiving osimertinib and reported that a higher ratio is closely related to tumor shrinkage and favorable survival [11, 18]. Two researchers observed that the ratio was significantly higher in patients with CR/PR/SD than in those with PD [11, 18]. In the present study, the ratio of T790M/major EGFR mutation at pretreatment could not predict the outcome or tumor shrinkage after osimertinib administration. A previous study confirmed that the amount of plasma T790M at pretreatment is not a reliable biomarker for tumor response and survival [11, 18]. This finding is similar to the results of the present study. Although the absence of major EGFR mutation in plasma at pretreatment may be a significant surrogate marker for the outcome after osimertinib treatment, we believe that baseline T790M level is not closely associated with response and prognostic prediction.
Our study has some limitations. First, the sample size was small, which may have biased the results. Second, the current study focused on patients harboring the T790M EGFR mutation after first- or second-generation EGFR-TKI administration. Currently, osimertinib is frequently administered to patients with naive EGFR-TKIs. Therefore, our digital PCR technique should be examined as an exploratory investigation for a predictive marker of first-line osimertinib. Moreover, we defined a copy number of 0 (copies/µL) as the cut-off value for further investigation. The cut-off values of copy number are different according to individual studies; therefore, it is uncertain whether an optimal cut-off value is determined by digital PCR. Finally, we were unable to investigate the mechanisms of resistance to osimertinib, except for C797S. Resistance mechanisms, such as the bypass signaling pathway, PTEN loss, MET amplification, MYC amplification, and small cell lung carcinoma transformation, have been described in previous reports [19–21]. Since resistance to osimertinib is closely associated with many tumor mutations, it is difficult to identify specific markers.
In conclusion, the detection of plasma T790M at PD after osimertinib treatment is identified as a significant predictor of worse outcomes after osimertinib administration. The detection of major EGFR mutations during pretreatment and PD also affects the outcome after treatment. Further investigation using a large-scale sample is warranted to confirm the results of our study.