Background
Currently, NSCLC with
HER2 mutation entered into the era of targeted therapy [
1,
2]. Based on the promising results of phase II trials that investigated the efficacy and side effects of ado-trastuzumab emtansine (T-DM1) [
3,
4] and trastuzumab deruxtecan (T-DXd) [
5], national comprehensive cancer network guideline has recommended them as options for patients with advanced
HER2-mutant NSCLC, and the later got the approval by FDA in later line setting [
6]. Meanwhile, TKIs targeting
HER2 mutation also achieved a breakthrough, pyrotinib [
7‐
9] or poziotinib [
10] showed an inspiring antitumor activity in this setting in phase II trials, and the former got the recommendation by the Chinese society of clinical oncology (CSCO) guideline. Besides, several other novel TKIs including tarloxotinib (NCT03805841) [
11], TAK-788 (NCT02716116) [
12] are under development.
However, both the ADCs and TKIs showed moderate efficacy, with an ORR of 31–55% and median PFS of 4.4–8.2 months [
3‐
5,
7]. As a result, identifying the benefit population or improving the efficacy by combination is essential in the clinical setting. Currently, several strategies, such as combining with anti-angiogenesis [
13,
14], immunotherapy [
15], were ongoing and reported inspiring preliminary results. In this study, aiming to clarify molecular features of responders in patients of
HER2 mutant advanced NSCLC treated with pyrotinib, we collected the ctDNA and performed the biomarker analysis from the pooled analysis of our two previous phase II trials [
7,
8].
Methods
Patients and sample collection
Patients were recruited from two phase II clinical trials of pyrotinib (ClinicalTrial.gov: NCT02535507, NCT02834936). Briefly, eligible patients with advanced HER2-mutant NSCLC who previously received systemic treatment were enrolled. All enrolled patients received pyrotinib 400 mg or 320 mg per day, until intolerable toxicity, disease progression, death or withdrawal of consent. Peripheral blood sample collection was performed at baseline, 40 days and 80 days after pyrotinib administration. A total of 50 patients were enrolled in the biomarker analysis. Among them, 21 pretreated tissue samples and 112 serial blood samples were collected. The study protocol was approved by ethics committees and relevant health authorities. All patients signed informed consent forms of our study.
DNA extraction and sequencing
DNA was isolated from formalin-fixed paraffin-embedded (FFPE) tumor specimens with the TIANamp genomic DNA kit (TIANGEN, China) according to the manufacturers’ instructions. Genome DNA is extracted by TGuide S32 magnetic blood genomic DNA kit (TIANGEN, China) from peripheral blood lymphocyte (PBL), and circulating cell-free DNA (cfDNA) is extracted by MagMAX cell-free DNA isolation (ThermoFisher, USA) from the plasma sample. Fragmented DNA libraries were constructed with a KAPA HTP library preparation kit (Illumina Platform) (KAPA Biosystems, Massachusetts, USA) according to the manufacturer’s instructions. All libraries were quantified using AccuGreen high sensitivity dsDNA quantitation kit (Biotium, USA), with library size assessed on agilent bioanalyzer 2100 (Agilent, USA). DNA libraries from baseline tissue samples were captured with Panel 1, which was a designed panel spanning 769 cancer-related genes (Genecast, Wuxi, China), while DNA libraries from plasma samples were captured with panel 2, which covered exon regions of 95 genes (Genecast, Wuxi, China) related to drug resistance. The captured library was sequenced on Illumina Novaseq 6000 with paired end 150 bp mode.
Variant calling
After filtering low quality reads by Trimmomatic(v0.36) [
16], clean reads were aligned to the human reference genome (hg19, NCBI Build 37.5) with the Burrows-Wheeler aligner (version 0.7.17) [
17]. Then Picard toolkit (version 2.23.0) [
18] was applied for making duplicates and genome analysis tool kit (version 3.7) [
19] was used for realignment. VarDict (version 1.5.1) [
20] was used to call single nucleotide variant (SNV) mutations while compound heterozygous mutations were merged by FreeBayes (version 1.2.0) [
21]. Sentieon software (genomics-201911) was also used to improve the detection rate of mutations in plasma samples, and the mutations were annotated through ANNOVAR [
22]. Typical QC-filtering such as variant quality and strand bias was applied to the raw variant list. Additionally, variants in low complex repeat and segmental duplication regions that matched to the lowly mappable regions were defined by ENCODE [
23], and variants in an internally developed and validated list of recurrent sequence-specific errors (SSEs) were removed.
Somatic mutation filtering of tumor tissues and plasma
After filtering germline or hematopoietic origin mutations by comparing with paired normal sample, somatic mutations met the following criterions were used for the following analysis: (i) The variant allele frequency (VAF) threshold of mutations was 5% in tumor and 1% plasma; (ii) All low-frequency mutations in samples from the same patient in different time points were retained. For plasma samples, we used a pre-defined blacklist to remove false positive variants introduced for special processing of UMI data. These quality cut-offs were predetermined during the analytical validation of the internal NGS panel to optimize the test performance and measured according to sensitivity, specificity, repeatability and reproducibility.
Statistical analysis
Chi-square test or Fisher’s exact test was used to compare the categorical variables. PFS and OS curves were estimated by Kaplan–Meier method and compared by Log-rank test. Cox proportional hazard model was performed for univariate and multivariate survival analyses to calculate the hazard ratio (HR) and 95% confidence interval (CI). Mann–Whitney U tests was introduced to analyze the significant difference of mutation frequency between defined groups. And statistical significance was defined with P-value < 0.05. All of the statistical analyses were performed using R V.3.6.1. and SPSS statistical software (version 22.0; IBM Corporation, Armonk, NY, USA).
Discussion
In this study, we firstly presented the up-to-date largest cohort of dynamic ctDNA profiling in HER2 mutant NSCLC on the basis of a prespecified biomarker analysis of two prospectively trials. We found that TP53 wild type at baseline were independently correlated with better clinical outcomes, including superior disease control rate (p = 0.010), longer PFS (p = 0.001) and OS (p < 0.001), than those with mutation. Our study also revealed that nonshedding tumor or ctDNA clearance were associated with superior efficacy. These findings shed light on the individual targeted therapy in patients with HER2 mutation.
Therapeutic landscape has been largely changed in patients of advanced NSCLC with
HER2 mutation [
25,
26]. Doublet chemotherapy used to be the standard-of-care for
HER2-mutated NSCLC, however, previous results including ours showed a discouraging result with the ORR of 10–43.5% and median PFS of 4.3–6 months [
27]. Subsequently, it was also found that patients with
HER2 mutation were unsuitable for immunotherapy, such as anti-PD-1/PD-L1 monotherapy [
15,
28,
29]. Encouragingly, the recent advance of ADCs such as T-DM1 [
4] and T-DXd [
5,
30], TKIs such as pyrotinib [
7,
31] or poziotinib [
10] showed inspiring antitumor activity in
HER2-mutated NSCLC in phase II trials, which made
HER2 mutation a druggable target. However, their moderate efficacy indicated that there’s an urgent need to establish effective predictive biomarkers in the clinical practice.
As far as we know, this is the first study to investigate the predictive role of genomic alternations through the ctDNA detection in
HER2 mutant NSCLC, and we found that
TP53 mutation was associated with the inferior efficacy of pyrotinib. Previously, several studies revealed that concurrent mutations would deteriorate the anti-cancer effect of EGFR-TKIs in patients with
EGFR mutations [
32,
33]. Moreover,
TP53 mutations was further found to promote genetic evolution and accelerate occurrence of resistance both in patients with
ALK fusion and
EGFR mutation [
34]. Similarly, this study for the first time reported that
TP53 mutation was of vital importance in the resistance of pyrotinib treatment in
HER2 mutant NSCLC. Taken together, these findings suggested that
TP53 mutations played an important role in the resistance of targeted therapy and needed to be considered as a stratification factor in study design in the future.
Moreover, we found that 10% patients had nonshedding ctDNA of lung tumors at baseline and after pyrotinib therapy. Importantly, patients with nonshedding tumors had a higher ORR of 60% and longer median progression free survival of 10.2 months. Recently, it was found that the presence of nonshedding tumors in the minimal residual disease (MRD) detection was associated with longer relapse-free survival (RFS) and higher possibility of cure [
35]. Although the nonshedding observation might also be attributed to the insufficient detection sensitivity, the possible underlying mechanisms still needed to be further explored. Our findings showed that nonshedding might also be served as a potential biomarker for the superior prognosis and efficacy of pyrotinib treatment.
Additionally, we also investigated the predictive role of ctDNA dynamics. Previously, several reports including ours consistently demonstrated that ctDNA clearance after 2 cycle of chemotherapy [
36] or immune checkpoint inhibitors (ICIs) therapy [
37,
38] was associated with a better efficacy in advanced NSCLC, indicating that ctDNA dynamics was a useful marker for systemic therapy in lung cancer. In this study, we also observed that the pyrotinib treatment efficacy was superior in patients with ctDNA clearance after 40 days of treatment. Taken together, this study highlighted the importance of ctDNA detection and found that
TP53 wild type, nonshedding tumor and ctDNA clearance could be used to identify the patients benefit from pyrotinib treatment in NSCLC.
Several limitations must be mentioned in this study. First, the number of patients finally enrolled into the analysis was still small even this was a pooled analysis of two phase II trials. Thus, selection bias might be inevitable. Second, only blood samples were used for the biomarker analysis due to the insufficient tissues and difficulty of re-biopsy at the disease progression, some genomic information might be missed in the liquid-based NGS testing. Thirdly, currently pyrotinib was only recommended by CSCO NSCLC guidelines in the later line setting, therefore, these findings might not suitable for a wide generalization.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.