Concomitant mutation status of ALK-rearranged non-small cell lung cancers and its prognostic impact on patients treated with crizotinib.

BACKGROUND: In non-small cell lung cancer (NSCLC), anaplastic lymphoma kinase (ALK) rearrangement characterizes a subgroup of patients who show sensitivity to ALK tyrosine kinase inhibitors (TKIs). However, the prognoses of these patients are heterogeneous. A better understanding of the genomic alterations occurring in these tumors could explain the prognostic heterogeneity observed in these patients.
METHODS: We retrospectively analyzed 96 patients with NSCLC with ALK detected by immunohistochemical staining (VENTANA anti-ALK(D5F3) Rabbit Monoclonal Primary Antibody). Cancer tissues were subjected to next-generation sequencing using a panel of 520 cancer-related genes. The genomic landscape, distribution of ALK fusion variants, and clinicopathological characteristics of the patients were evaluated. The correlations of genomic alterations with clinical outcomes were also assessed.
RESULTS: Among the 96 patients with immunohistochemically identified ALK fusions, 80 (83%) were confirmed by next-generation sequencing. TP53 mutation was the most commonly co-occurring mutation with ALK rearrangement. Concomitant driver mutations [2 Kirsten rat sarcoma viral oncogene homolog (KRAS) G12, 1 epidermal growth factor receptor (EGFR) 19del, and 1 MET exon 14 skipping] were also observed in 4 adenocarcinomas. Echinoderm microtubule associated protein-like 4 (EML4)-ALK fusions were identified in 95% of ALK-rearranged patients, with 16.2% of them also harboring additional non-EML4-ALK fusions. Nineteen non-EML4 translocation partners were also discovered, including 10 novel ones. Survival analyses revealed that patients concurrently harboring PIK3R2 alterations showed a trend toward shorter progression-free survival (6 vs. 13 months, P=0.064) and significantly shorter overall survival (11 vs. 32 months, P=0.004) than did PIK3R2-wild-type patients. Patients with concomitant alterations in PI3K the signaling pathway also had a shorter median overall survival than those without such alterations (23 vs. 32 months, P=0.014), whereas progression-free survival did not differ significantly.
CONCLUSIONS: The spectrum of ALK-fusion variants and the landscape of concomitant genomic alterations were delineated in 96 NSCLC patients. Our study also demonstrated the prognostic value of concomitant alterations in crizotinib-treated patients, which could facilitate improved stratification of ALK-rearranged NSCLC patients in the selection of candidates who could optimally benefit from therapy. 2021 Translational Lung Cancer Research. All rights reserved.

Introduction

Lung cancer remains the most common malignancy worldwide, the morbidity and mortality of which are at the forefront of global research (1). With the recent development in targeted therapy, treatment for patients with non-small cell lung cancer (NSCLC) has changed substantially (2). Molecular studies have shown that 64% of lung adenocarcinomas have driver gene alterations (3).

Similar to that of Kirsten rat sarcoma viral oncogene homolog (KRAS) and epidermal growth factor receptor (EGFR), the discovery of echinoderm microtubule-associated protein-like 4 fused with anaplastic lymphoma kinase (EML4-ALK) has been hugely significant to the individualized treatment of NSCLC. The ALK gene encodes a transmembrane tyrosine kinase receptor. EML4-ALK translocation can result in constitutive ALK kinase activity and represents an oncogenic addiction pathway in lung cancer. EML4-ALK fusion protein serves as a therapeutic target for an ALK-TKI, and has shown promising results when used to treat NSCLC patients carrying ALK rearrangement. Over the last few years, ALK inhibitors, including the TKI crizotinib, have shown significant benefits in the management of ALK-positive NSCLC compared to conventional chemotherapy (4). Of patients with NSCLC, approximately 3–7% express EML4-ALK and can benefit from individualized treatment (5). There is a subset of ALK rearrangement-positive NSCLCs that respond differently to ALK inhibitors according to variations in EML4-ALK fusion. EML4-ALK fusion variant V3 is a high-risk feature in ALK-positive NSCLC and confers early metastatic spread, treatment failure after tyrosine kinase inhibitor (TKI) therapy, chemotherapy, and cerebral radiotherapy, as well as inferior overall survival (OS) (6). However, it has been reported that there is no correlation between variants of EML4-ALK and patients’ clinical responses to crizotinib (3). Although several ALK inhibitors, such as crizotinib, ceritinib, alectinib, and brigatinib, have been approved for cancer therapy, a large number of NSCLC patients go on to develop disease progression after the application of ALK TKIs (7). The clinicopathological features of patients with both ALK fusion and an oncogene mutation are critical to progressing individualized treatment for NSCLC (8).

So far, crizotinib has the longest follow-up of any drug for the treatment of ALK NSCLC, as it was the first drug to be approved (9). However, most patients who respond to frontline TKIs eventually acquire resistance within 1–2 years (10). According to published research reports, the mechanisms of ALK-TKI resistance can be roughly divided into 2 main forms: ALK-dependent changes and ALK-independent changes. ALK-dependent changes include secondary ALK mutations and ALK gene amplification, whereas ALK-independent changes mainly consist of the upregulation of a bypass signaling pathway and changes in lineages. Among the genetic factors, co-alterations of ALK fusions and gene mutations are the leading cause of crizotinib resistance. Preclinical data have shown that various concomitant mutations (11,12) are associated with low sensitivity to crizotinib (13). In terms of the molecular pathology, rearrangement of ALK is independent of EGFR and KRAS mutations (14-18), although these alterations are not absolutely mutually exclusive (18,19). For instance, in a recent lung cancer mutant complex series, 8% of ALK-positive adenocarcinoma patients also had EGFR or KRAS mutations (20). Studies have illustrated that patients with concomitant EGFR mutation and EML4-ALK translocation are insensitive to EGFR TKIs but sensitive to crizotinib (21). KRAS mutations are the most common concomitant mutations with EML4-ALK fusions, accounting for approximately 2.5% of patients with EML4-ALK NSCLC, and are associated with reduced reactivity to crizotinib and a poor prognosis (22).

In the present study, we analyzed the basic clinical information of 99 patients with ALK-positive NSCLC who were treated in Tianjin Medical University Cancer Institute and Hospital from 2012 to 2016. Next-generation sequencing (NGS) was conducted on 96 paraffin-embedded tissue samples from these patients. According to the NGS results, we analyzed the landscapes of ALK fusion variants and the co-occurring genomic alterations in ALK fusion-positive NSCLC, and explored their associations with patients’ clinical outcomes.

We present the following article in accordance with the STROBE reporting checklist (available at http://dx.doi.org/10.21037/tlcr-21-160).

Methods

Patient information

All patients (n=99) with ALK fusion-positive lung cancer, as assessed by immunohistochemistry [IHC; VENTANA anti-ALK (D5F3) Rabbit Monoclonal Primary Antibody], who received treatment in Tianjin Cancer Hospital between January 2012 and August 2016 were enrolled in the study. Because the samples from the patients were placed for too long, 3 tissue samples failed the quality control of sequencing data, and only 96 samples were tested by NGS for subsequent analysis. The patients’ medical records were retrieved to collect data including clinicopathological characteristics, treatment history, and survival outcomes. Progression-free survival (PFS) after treatment and overall survival (OS) were assessed. This study was approved by the Institutional Review Board of Tianjin Cancer Hospital (No. bc2019089), and was conducted in accordance with the Declaration of Helsinki (as revised in 2013). Written informed consent was obtained from all participants.

DNA isolation and capture-based targeted DNA sequencing

DNA was isolated from tumor tissues using a QIAamp DNA FFPE Tissue Kit (Qiagen, Hilden, Germany). The NGS library was prepared with a minimum of 50 ng of DNA as described previously (23). Briefly, DNA fragmentation was performed using a Covaris M220 Focused-ultrasonicator (Covaris, MA, USA), after which end repair, phosphorylation, and adapter ligation were carried out. Selective purification of DNA fragments between 200–400 bp was conducted using magnetic beads (Agencourt AMPure XP Kit, Beckman Coulter, CA, USA), followed by hybridization with RNA probes of a panel consisting of 520 cancer-related genes spanning 1.64 megabases (Mb) of the human genome (OncoScreen Plus, Burning Rock Biotech, Guangzhou, China). The targeted library was subsequently enriched by polymerase chain reaction (PCR) amplification. The quality and size of the library were assessed using a Qubit 2.0 Fluorometer with the dsDNA High Sensitivity Assay Kit (Life Technologies, Carlsbad, CA, USA). Sequencing of indexed samples was performed on the NextSeq 500 platform (Illumina, Inc., USA) with paired-end reads and a median sequencing depth of 1,636×.

Data analysis

Sequencing data in FASTQ format were aligned to the reference human genome (hg19) using Burrows–Wheeler Aligner v.0.7.10 (24). Local alignment optimization, duplication marking, and variant calling were conducted using the Genome Analysis Tool Kit v.3.2 (25) and VarScan v.2.4.3 (26). Variants with a loci depth <100 were filtered out with the VarScan fpfilter pipeline. Variants with population frequencies over 0.1% in the Exome Aggregation Consortium (ExAC), 1,000 Genomes, Single Nucleotide Polymorphism Database (dbSNP), or ESP6500SI-V2 databases were excluded from further analysis. The remaining variants were annotated with ANNOtate VARiation (ANNOVAR) (February 1, 2016, release) (27) and SnpEff v.3.6. (28) DNA translocation and copy number variation (CNV) were analyzed using Factera v.1.4.3 (29) and an in-house algorithm based on sequencing depth, respectively. Fusions, large genomic rearrangements, and mutations occurring on the kinase domains of EGFR and ALK were excluded from the mutation count.

Statistical analysis

Statistical analyses were performed using R version 3.3.3 software. Differences among groups were calculated by Fisher’s exact test, paired two-tailed Student’s t-test, or analysis of variance, as appropriate. Survival outcomes were estimated with Kaplan-Meier curves, and the log-rank test was applied to determine the difference in the survival curves between groups. A P value <0.05 was considered to be statistically significant.

Results

Clinicopathological characteristics of the patients

Qualified sequencing data were generated from tissue samples from 96 of the 99 patients and used in subsequent analyses. details the clinicopathological characteristics of the 96 patients. The median age of this cohort was 56.5 years, and 52 (54.2%) patients were male. Twenty-nine (30.2%) patients had a history of smoking. There were 81 (84.4%) and 9 (9.4%) diagnoses of lung adenocarcinoma and squamous carcinoma, respectively. Six patients had pulmonary tumors with other histologies (1 adenosquamous carcinoma, 1 mucoepidermoid carcinoma, 1 typical carcinoid, 1 atypical carcinoid, and 2 large cell neuroendocrine carcinoma). There were 31 (32.3%), 17 (17.7%), 28 (29.2%), and 20 (20.8%) patients in stage I, II, III, and IV, respectively. Brain, bone, and visceral metastases were reported in 14 (14.6%), 9 (9.4%), and 7 (7.3%) patients, respectively. A total of 39 (40.6%) patients received crizotinib, 26 of whom were in stage III or IV. Of the 39 patients who were treated with crizotinib, there was prognostic information available for 37. A median progression-free survival (mPFS) of 12 months and a median overall survival (mOS) of 23 months were observed. Patients treated with crizotinib had a poorer prognosis than untreated patients, which was attributable to crizotinib-treated patients having more advanced tumors.

Table 1

Clinicopathological characteristics of patients

Characteristics	No. (%)
Age, years
Median [min, max]	56.5 [22, 75]
Sex
Female	44 (45.8)
Male	52 (54.2)
Smoking history
No	67 (69.8)
Yes	29 (30.2)
Histology
Adenocarcinoma	81 (84.4)
Squamous carcinoma	9 (9.4)
Others	6 (6.2)
Primary lesion site
Right lung	55 (57.3)
Left lung	39 (40.6)
Both lungs	2 (2.1)
Tumor stage
I	31 (32.3)
II	17 (17.7)
III	28 (29.2)
IV	20 (20.8)
Lymph node
N0	36 (37.5)
N1	11 (11.5)
N2	34 (35.4)
N3	15 (15.6)
Metastasis
No	58 (60.4)
Yes	38 (39.6)
Brain metastasis
No	82 (85.4)
Yes	14 (14.6)
Bone metastasis
No	87 (90.6)
Yes	9 (9.4)
Visceral metastasis
No	89 (92.7)
Yes	7 (7.3)
Crizotinib treated
No	57 (59.4)
Yes	39 (40.6)
PFS, months
Median (min, max)	24.5 [2, 64]
Crizotinib treated	12 [2, 52]
Crizotinib untreated	50 [4, 64]
OS, months
Median (min, max)	34 [3, 71]
Crizotinib treated	23 [3,7]
Crizotinib untreated	50 [4, 64]

The genomic landscape of the cohort

Qualified sequencing data were generated from 96 of the 99 patients. NGS identified ALK fusions in 83% (80/96) of patients, and 1 patient harbored concomitant ALK p.S267I mutation and amplification. Among the ALK-rearranged tumors, we identified 4 adenocarcinomas with other concurrent driver mutations: 2 KRAS G12, 1 EGFR 19del, and 1 MET exon14 skipping (). The most frequently mutated gene in this cohort was TP53 (24%), followed by lysine (K)-specific methyltransferase 2C (KMT2C; 12%), SET domain containing 2 (SETD2; 10%), and telomerase reverse transcriptase (TERT; 8%). The other concomitant alterations are illustrated in .

Figure 1

The genomic landscape of patients with anaplastic lymphoma kinase (ALK) fusion (n=80). Top represents the number of mutations detected in each sample; bottom represents the histological type; right represents the genes; left indicates the detection rate of mutation.

The distribution of ALK-fusion variants

Of the 80 ALK fusion-positive patients, detected with both NGS and IHC, 63 (78.8%) harbored a single EML4-ALK fusion, and 13 (16.2%) patients harbored more than 1 ALK fusion, with EML4-ALK being 1 of the variants. Four (5%) patients carried only non-EML4-ALK fusions ().

Figure 2

The distribution of anaplastic lymphoma kinase (ALK) rearrangements. (A) The distribution of different ALK partners (n=80). (B) The distribution of echinoderm microtubule associated protein-like 4 (EML4)-ALK variants (n=76). (C) The list of non-EML4-ALK partners identified. * a and b indicate the partners reported in the literature and in our internal database, respectively.

Among the 76 patients carrying EML4-ALK, variant 1 (36.8%) and variant 3 (31.6%) were the most common variants, followed by variants 4 (17.1%), 6 (7.89%), 2 (5.26%), and 5 (1.32%). Thirteen patients harbored EML-ALK variants other than variants 1–6 (). Overall, 19 non-EML4-ALK partners, including 10 novel partners, were identified ().

The prognostic value of ALK–fusion variants and concomitant genomic alterations in patients treated with crizotinib

To identify potential prognostic biomarkers, we investigated the correlations of ALK-fusion variants and concomitant genomic alterations with PFS and OS in patients treated with crizotinib. Patients harboring EML4-ALK showed no significant difference in PFS (P=0.51) (95% CI: 0.429–5.493) or OS (P=0.33) (95% CI: 0.447–11.413) when compared with those carrying non-EML4-ALK partners. Also, the PFS and OS of patients with EML4-ALK variant 3 were comparable to those of patients harboring EML4-ALK variant 1 [P=0.47 (95% CI: 0.389–7.857) and P=0.49 (95% CI: 0.339–9.232)] or other non-V3 fusions [P=0.15 (95% CI: 0.781–4.903) and P=0.37 (95% CI: 0.548–5.121)]. The significance of these results is limited by the small number of patients in our study.

Patients concurrently harboring PIK3R2 alterations showed a trend toward a shorter PFS [6 vs. 13 months, P=0.064 (95% CI: 0.928–14.937), ] and significantly shorter OS [11 vs. 32 months, P=0.004 (95% CI: 1.507– 27.643), ] than PIK3R2-wild-type patients. We next explored the association of alterations in the PI3K signaling pathway with patient prognosis, and a shorter mOS was observed in patients with these alterations than in patients without them [23 vs. 32 months, P=0.014 (95% CI: 0.072–0.746), ]; in contrast, we did not find a significant difference in the PFS of patients with and without alterations (). Furthermore, a shorter mOS was observed in patients harboring concomitant TP53 mutations [22.5 vs. 30 months, P=0.046 (95% CI: 1.023–10.527), ] and patients with mutations in the P53 signaling pathway [23 vs. 32 months, P=0.022 (95% CI: 0.729–3.871), ] than in patients with TP53/P53-wild-type; however, no difference was found in PFS ().

Figure 3

The correlation of concomitant alterations in PIK3R2 or in the PI3K signaling pathway with progression-free survival (PFS) and overall survival (OS) (n=37). (A,B) Alteration in PIK3R2; (C,D) alterations in PI3K signaling pathway genes (MAP2K2, JAK3, PIK3R2, KIT, CDK6, MYC, FGFR4, JAK1, RPTOR, INSR, BRCA1, IRS1, STK11, EGFR, KRAS, and MET). The P value was adjusted for age, sex, smoking history, surgical history, and brain metastasis.

Figure 4

The correlation of concomitant mutations in TP53 or in the P53 signaling pathway with progression-free survival (PFS) and overall survival (OS) (n=37). (A,B) Mutations in TP53; (C,D) mutations in P53 signaling pathway genes (TP53, MDM4, and CDK6). The P value was adjusted for age, sex, smoking history, surgical history, and brain metastasis.

Discussion

In this study, 96 patients with ALK fusions identified by IHC were retrospectively evaluated (Figure S1). Among them, 80 (83%) patients had rearrangement of ALK confirmed by NGS. Also, we discovered 10 novel ALK-fusion partners and identified concurrent mutations that could affect the clinical outcomes to crizotinib therapy. Notably, NGS positivity was correlated with a higher disease control rate and longer PFS than NGS negativity (P=0.02 and P=0.09), while fluorescence in situ hybridization (FISH) and IHC status were not able to distinguish the outcome after treatment with crizotinib. Although it is considered the gold standard, FISH presents a certain false-negative rate. Ventana-D5F3 IHC is qualified as a screening tool, while NGS positivity may more accurately predict clinical benefit of crizotinib, allowing for efficient testing for specific variants and concurrent genomic alterations (30).

The mutual exclusivity of the different driver mutations in adenocarcinoma is generally agreed upon. However, we identified concurrent KRAS G12, EGFR 19del, and MET exon14 skipping in 4 ALK-rearranged adenocarcinomas (4/80; 5.0%). Treatment for this subset of patients remains controversial with respect to the choice of single- vs. dual-targeted therapies (2), and a decision guided by evidence from large-scale studies is still awaited. We also found that 13 patients had other non-classical ALK fusions accompanying the EML4-ALK classic fusion. However, due to the insufficient number of samples, it is impossible to make an effective statistical analysis on the prognosis of these patients, so a large sample of cases is needed to verify it in the future.

Consistent with previously reported studies, we found EML4-ALK fusion to be the most common type of ALK fusion, with variants 1 and variants 3 being the most frequent EML4-ALK variants. Furthermore, we identified 19 different non-EML4-ALK partners, including 10 novel partners (ANKRD36BP2-ALK, ANO4-ALK, ASB9-ALK, LOC100507443-ALK, DNAJC10-ALK, DUSP22-ALK, MIR148A-ALK, MIR641-ALK, SH3YL1-ALK, and ZNF155-ALK) that have not been reported before (31). Although NGS analyses have identified a variety of ALK rearrangements, whether or not the diversity of fusion sites leads to different responses to ALK inhibitors remains unclear. The investigation on the prognostic impacts of ALK-fusion partner/variants in ALK inhibitor-treated patients would guide the accurate stratification of ALK-rearranged NSCLC patients for optimizing the therapeutic selection for individual patients. In line with some other reports (32), but in contrast with other retrospective studies (33,34) and results from the randomized phase 3 trial ALTA-1L (35), we found no significant difference in PFS (P=0.47) or OS (P=0.49) between patients with variant 1 and variant 3 (Figure S2). It is noteworthy that, instead of using DNA-based NGS, most studies that detected a difference in patient outcome according to the EML4-ALK fusion variant used RNA-based NGS (36), which has demonstrated a higher yield for the detection of oncogenic fusions in lung adenocarcinoma (37), or highly sensitive circulating tumor DNA assays, such as ctDx-Lung in case of the ALTA-1L analysis (38). In our study, patients with EML4–ALK variant 3 also had comparable PFS and OS to patients harboring other non-V3 fusions (P=0.15 and P=0.37) (5,33,39,40). A recent study reported that the specific ALK variant seemed to impact the ALK resistance mutations acquired. Further study of the relationship between ALK variants and ALK-TKI-resistant mutations using tissue samples obtained from patients with disease progression is worthy of attention in the future (32).

Besides, in our study, NGS identified TP53 mutations to be the most common type of concomitant mutation with ALK rearrangement, with an incidence of 24%, which was comparable to the 20–29% reported by other investigators (1,41). In addition to identifying the mutations concomitant with ALK rearrangement, we also analyzed the relationships of these combinations with prognosis. Consistent with the findings of other studies, concomitant TP53 mutations were predictive of poor survival (42). Also, we observed that patients with alterations in the PI3K signaling pathway had a shorter mOS than patients without these alterations (23 vs. 32 months, P=0.014). However, due to the small number of cases in the present study, it is necessary to increase the number of samples for further analysis in the future. In line with our findings, another study reported that PI3K/AKT signaling activation led to treatment with ALK inhibitor being ineffective, while PI3K inhibitor increased sensitivity to ALK inhibitor in EML4-ALK-positive cells (43). We hope that further research on the ALK signaling pathway will lead to the emergence of novel generations of ALK inhibitors, which will aid in improving the PFS and OS of patients with ALK rearrangements. Since the incidence of mutations concomitant with ALK rearrangement has been discovered to be high, the relationship between concomitant mutations in the ALK signaling pathway and prognosis needs to be clarified so that the targeted therapy that will maximize the benefit to the patient and substantially improve their survival can be chosen (44).

Limitations of our study which might weaken the strength its findings should also be noted; these include its retrospective nature and the small sample size of patients with ALK rearrangements, especially those receiving ALK inhibitors. Therefore, the prognostic impact of the concomitant alterations we discovered requires confirmation by further prospective studies with large cohorts. Through this, accurate stratification of patients with ALK-rearranged NSCLC for selecting those who may receive the optimal therapeutic benefit may be realized.

The article’s supplementary files as