Characteristicsof patients and ALK rearrangements
All of 6576 samples from NSCLC patients were profiled with DNA-based NGS between January 2018 and July 2020. The clinical characteristics of the patients are described in Table 1. ALK fusions were identified in 343 (5.2%) cases with higher incidences in female, age<60 or adenocarcinoma patients. Canonical EML4-ALK fusions occurred most frequently accounting for 78.4% (269/343). Most of the genomic breakpoints of the ALK gene were detected within intron19, while the EML4 potential breakpoints differ and may generate various fusion protein variants at the genomic level. As shown in Figure 1, EML4-ALK variant 3 (E6:A20, 109/269, 40.5%) was the most predominant type, followed by variant 1 (E13:A20, 77/269, 28.6%) and variant 2 (E20:A20, 35/269, 13.0%).
Identification and validation of complex ALK rearrangements
Among the 343 ALK fusion cases, complex ALK rearrangements in 14 cases were identified using targeted DNA-based NGS across 86 cancer-related genes panel with multiple probes tilling selected intronic regions of fusion partner genes (Table 2). These cases could be divided into three types by integrating various genomic features, including intergenic (n=3), intragenic (n=5) and “bridge joint” rearrangements (n=6). A subset of 13 cases retained enough specimens were validated for additional RNA-based NGS tilling all coding exons of common fusion genes. Surprisingly, we found that the fusion genes and breakpoint positions had significant discrepancies between DNA and RNA sequencing. All thirteen cases actually expressed canonical EML4-ALK fusion transcripts. Besides, positive ALK IHC was detected in 13 of 13 cases, and 9 of 11 cases were positive in FISH testing.
Case 1, a representative intergenic complex rearrangement case, harbored WDR43-ALK (3’intron1: 3’intron19) and EML4-intergenic fusions identified by DNA-based NGS (Figure 2A), with positive results detected using ALK IHC assay (Figure 2D), but RNA-based NGS detected the canonical EML4-ALK fusion transcript joining EML4 exon 13 to ALK exon 20 (Figure 2B). Sequencing data indicated that the intergenic complex rearrangement involved multiple fusion junctions, comprising EML4, LINC01913 upstream intergenic region, WDR43 and ALK (Figure 2C). Evidences of such intergenic complex rearrangements were also detected in case 2 and case 3 by DNA-based NGS, which harbored a canonical EML4-ALK variant 3 (E6:A20) transcript identified by RNA sequencing, and were positive by IHC and FISH assays (Table 2).
The more remarkable observation, shared by cases 4-8, was the rare and complicated intragenic rearrangement of ALK or EML4 gene identified at DNA level. Case 4 typically harbored multiple distinct rearrangements involving ALK locus, consisting of 5’ EML4 (intron 13) and 3’ ALK (intron 3) fusion, ALK-ALK fusion in which intron 3 of ALK was jointed to intron 19 of ALK with a 9-bp insertion, GALM-3’ EML4 fusion, and 5’ ALK-intergenic fusion (Figure 3A). Only the first two connecting fusion-oncogene-associated rearrangements appeared capable of producing a functional pathogenic fusion transcript joining EML4 exon 13 to ALK exon 20 detected by RNA-based NGS data (Figure 3B and 3C). The other two fusions without transcription product may be the reciprocal fusions. Meanwhile, clear split signals of ALK gene were detected by FISH using a break-apart probe kit (Figure 3D), and IHC test of the surgically resected sample revealed a positive result (Figure 3E). Similarly, case 6 harbored 5’ EML4 (intron 6) and 3’ ALK (intron 4) fusion with a 39-bp insertion and ALK-ALK fusion in which intron 4 of ALK was jointed to intron 19 of ALK with a 57-bp insertion, indicating to product the canonical EML4-ALK variant 3 (E6:A20) transcript without enough specimens for validation assays. In case 5, an special inversion of ALK gene from intron 18 to intron 19 was detected, in which 3’ intron 18 of ALK was jointed to 3’intron 19 of ALK and 5’ intron 19 of ALK was jointed to 5’ intron 6 of EML4 and RNA-based NGS detected the canonical EML4-ALK variant 3 (E6:A20) transcript. Similarly, inversions of EML4 intron 6 were identified in case 7 and 8, which also harbored the canonical EML4-ALK variant 3 (E6:A20) transcript and were IHC and FISH positive (Table 2).
In cases 9-13, multiple gene fusions were identified, herein defined as “bridge joint” rearrangements owing to that both EML4 and ALK jointed with an identical gene at the genomic level, respectively. Taking case 9 for example, DNA-based NGS detected that the intron 13 of EML4 fused with the downstream region of intron 1 of LCLAT1, and the upstream region of intron 1 of LCLAT1 joined to the intron 19 of ALK (Figure 4A and 4C). Due to intronic splicing, it is reasonable that RNA-based NGS identified the canonical EML4-ALK variant 1 (E13:A20) transcript without intron1 of LCLAT1 (Figure 4B and 4C). IHC assays showed clearly positive ALK protein expression, but FISH revealed negative results, perhaps due to break-apart probe design or technical aspects yielding a risk of false-negative result (Figure 4D and 4E) [19]. Similarly, in cases 10-13, DNA-based NGS revealed that the intron of EML4 fusion partner gene firstly joined to the intronic region of an novel “bridge” gene and then to the intron of ALK kinase gene, as “bridge joint” complex rearrangements. Most of the intronic regions of the novel “bridge” gene were removed by splicing, leading to canonical EML4-ALK transcripts. In particular, the exon6 of “bridge” gene (RUNX1) was involved in the complex rearrangement and the RUNX1-ALK (exon6: exon20) transcript was detected in case 12, which may be part of the EML4-RUNX1-ALK (exon5: exon6: exon20) transcript hardly to be identified. Moreover, the EML4- ALK (exon5: exon20) transcript was also detected in case 12, perhaps due to the alternative splicing. Interestingly, case 14, harboring EML4-RPIA and MAP4K3-ALK fusions, was identified as the canonical EML4-ALK variant 1 (E13:A20) transcript, suggesting that RPIA and MAP4K3 were both the “bridge” genes and their intronic regions were connected together (Table 2).
Targeted therapies and clinical outcomes of complex ALK rearrangements
Among the 14 cases with complex ALK rearrangements, only 8 patients received targeted ALK inhibitors (crizotinib or alectinib) treatment, including 2 intergenic complex rearrangements, 3 intragenic complex rearrangements and 3 “bridge joint” rearrangements. Treatment and response to therapy, as defined by RECIST v1.1, were outlined in Table 3, which showed that 6 patients (75%) achieved clinical objective response, including 5 partial responses (PR) and 1 complete response (CR).
Case 1 and case 3 with intergenic complex rearrangements both had positive response to crizotinib, and the endpoint of progression-free survival (PFS) was still not reached, lasted for at least 8 and 7 months, respectively (Table 3). Differential ALK inhibitor responses were observed among intragenic rearrangements variants in ALK-positive lung adenocarcinoma (case 5, case 6 and case 8). The identical EML4-ALK fusion cases 5 and 8, both got transcript joining EML4 exon 6 to ALK exon 20 and positive results in FISH and IHC, achieved quite discrepant clinical outcome to crizotinib, CR for case 5 and progressive disease (PD) after 5 months treatment for case 8. We speculated that the other variant occurred in case 8, TP53 p.R273C mutation, enhanced cancer cell proliferation, invasion and drug resistance [20]. As to the “bridge joint” rearrangements, one of the three cases, case 14, exhibited stable disease (SD) 4 weeks after crizotinib treatment (Table 3), different with the PR states of the other two cases, which implied that the poor clinical outcomes for ALK inhibitor in some patients could be caused by primary drug resistance to targeted therapies [21].
Patients with complex ALK fusions (n=5) received crizotinib treatment exhibited comparable median progression-free survival (mPFS) with patients harboring canonical ALK fusions (n=34), which displayed in Figure 5A with the values of 7.0 months (95%CI: 0.3-2.0) versus 9.0 months (95%CI: 0.5-3.3) and the P value of 0.7616. Similarly, no significant difference in mPFS was observed between complex and canonical ALK fusions when patients with alectinib and crizotinib treatment were analyzed together (8.0 months [95%CI: 0.4-2.1] versus 9.0 months [95%CI: 0.5-2.7], P=0.9291, Figure 5B).