LOXO-101 Number of tissue and plasma samples obtained at eac
Number of tissue and plasma samples obtained at each time point are described in Fig. 1. One tumor tissue specimen did not contain tumor LOXO-101 so we analyzed 56 samples. During the first 12 weeks of treatment, about 90% of planned plasma samples were collected. At 24 weeks, 40 plasma samples were collected.
Regarding tissue samples, results of sensitive EGFR mutation analyses using digital PCR corresponded to those based on institutional mutation analyses. In one patient (1.8%), de novo Exon 20 T790 M mutation was additionally detected by digital PCR analysis. Analyzing plasma samples, 35 of 56 patients (62.5%, 95% CI: 44.2–70.1%) were positive for sensitive EGFR mutation at baseline. Sites of EGFR mutation were completely identical between tissue and corresponding plasma samples. Characteristics of the patients who were positive or negative for EGFR mutation in plasma are shown in Table 1. Clinical stage was the only factor that correlated with detection rate. Patients with stage IV demonstrated the highest detection rate compared with patients who were stage III and post-operative relapse (81.6% versus. 0% and 31.2%, respectively). In more detail, the systemic spread of tumor seemed to correlate with higher detection rate (Fig. 2). Among stage IV disease, detection rate tended to be higher in patients with distant metastasis (M1b, 85.2%, 23 of 27 patients) than those without (M1a, 72.7%, eight of 11 patients). Similar tendency was observed among patients with post-operative relapse (57.1% in patients with metastatic relapse versus 11.1% in patients with localized relapse). Regarding efficacy, cfDNA-positive patients showed slightly shorter PFS than cfDNA-negative patients, but not significantly (median 13.0 months versus not reached, p = 0.11, Fig. 3).
At baseline, mutant allele frequency of EGFR in tissue and plasma samples did not correlated with anti-tumor response (Supplementary Fig. 1). Among 35 patients who were positive for cell-free DNA (cfDNA) at baseline, changes of cfDNA during treatment were analyzed. Negative conversion (NC) of cfDNA was observed in 60.6% at two weeks, 87.5% at four weeks, 93.8% at eight weeks, 87.1% at 12 weeks and 83.3% at 24 weeks, respectively. Patients who achieved NC at two weeks had significantly longer PFS than those without (13.6 months vs 7.5 months, p = 0.0001, Fig. 4a). Similarly, patients who achieved NC at four weeks demonstrated longer PFS than those without (13.6 months versus 5.1 months, p < 0.0001, Fig. 4b). PFS was not different between patients who achieved NC in two weeks and those who achieved NC in two to four weeks (p = 0.59).
At the time of analysis, 17 patients experienced disease progression. We managed to collect nine tissue samples and 14 plasma samples. Most common reason we were not able to obtain as many tissue samples as expected was difficulty in obtaining tissue samples due to invasiveness and patients’ poor condition. Exon 20 T790M was detected in five patients (62.5%) from plasma samples. Sequential changes of allele frequencies of EGFR mutation in each individual were shown in Supplementary Fig. 2. Of 14 patients, 11 achieved negative conversion, but finally plasma recurrence was found in 8 patients (57.1%). Five patients (37.5%) showed recurrence in plasma DNA earlier than radiological progression. Of those, median time between plasma recurrence and radiological progression was 45days (range 12–123 days). However, PFS curve based on plasma recurrence did not show significant difference compared to that based on RECIST (Fig. 5).
Discussion Emergence of highly sensitive PCR assay has enabled us to detect driver mutation from plasma samples. In fact, liquid biopsy became one of the standard method to identify the subtypes of NSCLC. However, there still remains some debatable issues. First, its detection rate has been discussed based on various types of assays and patient populations. Pivotal studies demonstrated that detection rate is about to 50–60% [6,13]. On the other hand, a large-scale observational study conducted in Japan showed that detection rate is only 20% in a real world setting . Sensitivity of each assay might influence on these results. Additionally, our analysis suggests that utility of cfDNA analysis was limited only in patients with disseminated disease. Detection rate of our patients with stage III or post-operative relapse was below 60%, while detection rate among patients with distant metastasis increased to 85%. Similarly, differences in PFS between cfDNA-positive and –negative patients might reflect the tumor volume. Because amount of cfDNA in plasma was scarce, multiplexed assay like ours or next-generation sequencing are clinically useful to analyze both sensitive and resistant mutation in one reaction.