IDH1 and IDH2 mutations in lung adenocarcinomas: Evidences of subclonal evolution

Abstract Background Selective IDH1 and IDH2 inhibitors have been approved for targeted therapy of acute myeloid leukemia. Clinical trials for solid tumors with IDH1 and IDH2 (IDH1/2) mutations are ongoing. Reports of IDH1/2‐mutated non–small cell lung cancers (NSCLCs), however, are limited. Methods We evaluated IDH1/2 mutations in 1,924 NSCLC specimens (92% adenocarcinoma) using a next‐generation sequencing assay. Results Retrospective quality assessments identified false detection of IDH1 c.395G>A (p.R132H) resulting from cytosine deamination (C:G→T:A) artifact in one specimen. IDH1/2 mutations were detected in 9 (0.5%) adenocarcinomas taken by fine‐needle aspiration (n = 3), thoracentesis (n = 2) or core biopsy (n = 4). All nine adenocarcinomas showed high‐grade features. Extensive clear cell change, however, was not observed. High expression (50% or greater) of PD‐L1 was observed in two of five specimens examined. IDH1/2 mutations were associated with old age, smoking history, and coexisting KRAS mutation. Lower than expected variant allele frequency of IDH1/2 mutants and coexistence of IDH1/2 mutations with known trunk drivers in the BRAF, EGFR, and KRAS genes suggest they could be branching drivers leading to subclonal evolution in lung adenocarcinomas. Multiregional analysis of an adenocarcinoma harboring two IDH2 mutations revealed parallel evolution originating from a KRAS‐mutated lineage, further supporting subclonal evolution promoted by IDH1/2 mutations. Conclusions IDH1/2 mutations in NSCLCs are uncommon. They occur in adenocarcinomas with high‐grade features and may be branching drivers leading to subclonal evolution. Accumulation of more IDH1/2‐mutated NSCLCs is needed to clarify their clinicopathological characteristics and implications for targeted therapy.


| INTRODUCTION
Mutational profiling identifies genomic alterations for targeted therapy in metastatic non-small cell lung cancers (NSCLCs). Several targeted therapeutic agents have been approved for metastatic NSCLCs with EGFR mutations, BRAF p.V600E, ALK translocations, and ROS1 translocations. 1,2 Mutational profiling of these genomic alterations is considered standard of care for patients with metastatic NSCLCs. 3 Integrated multiplatform analyses including whole-exon sequencing and whole-genome sequencing have uncovered additional genomic alterations in NSCLCs with potential implications for targeted therapy, such as ERBB2 mutations, MET mutations and translocations of the RET, NTRK1, NTRK2, and NTRK3 genes. 1,2 IDH1 mutations involving codon 132 and IDH2 mutations involving codons 140 and 172 occur in a variety of human cancers, including acute myeloid leukemia (AML), diffuse gliomas, cholangiocarcinoma, and chondrosarcoma. [4][5][6][7][8][9][10][11] and IDH2 (IDH1/2) mutations were also reported in NSCLCs with a much lower incidence (0.4%-1.1%). 12-14 IDH1/2 mutants lead to accumulation of D-2-hydroxyglutarate through neoenzymatic conversion, and subsequent oncogenic effects including epigenetic alterations. 15,16 IDH2 inhibitor (Enasidenib or AG-221) and IDH1 inhibitor (Ivosidenib or AG-120) have been approved by the Food and Drug Administration in the United States for targeted therapy of AML. 6,7 Several clinical trials of IDH1/2 inhibitors for advanced solid tumors, such as NCT02073994 (AG-120 for IDH1 mutations), NCT02746081 (BAY1436032 for IDH1 mutations), and NCT02481154 (AG-881 for IDH1/2 mutations) are ongoing. Clinical pharmacokinetics and pharmacodynamics studies have shown robust and persistent inhibition of plasma D-2-hydroxyglutarate by oral ivosidenib. 17 In this study for quality assessment, next-generation sequencing (NGS) was examined in a large cohort of NSCLC specimens to elucidate the incidence of IDH1/2 mutations and the clinicopathological and molecular characteristics of IDH1/2-mutated NSCLCs.

| Materials
NGS results from 1924 lung cancer specimens (1778 adenocarcinomas, 24 adenosquamous carcinomas, 12 adenocarcinomas in situ, and 110 NSCLCs) submitted between April 2013 and December 2018 were analyzed for mutations in IDH1 and IDH2 genes. For multiple specimens taken from the same tumor (such as biopsy and resection specimens, or primary and metastatic tumor specimens) and showing an identical mutation status, only one specimen was included.
Specimens with prior EGFR tyrosine kinase inhibitor therapy were also excluded. Accompanied hematoxylin and eosin-stained slides were reviewed by a pulmonary pathologist (PI) and/or a molecular pathologist (MTL). DNA was isolated from formalin-fixed paraffin-embedded (FFPE) tissues using Pinpoint reagents (ZymoResearch) and purified using QIAmp DNA kit (Qiagen) as described previously. 18 After April 2017, DNA was isolated from FFPE tissues using Tissue Preparation System (Siemens) according the manufacturer's protocol. Concentration of DNA was determined by Qubit 2.0 Fluorometer (Life Technologies). The Johns Hopkins Institutional Review Board granted approval to this study.

| Immunohistochemical stains
Immunochemical stains for TTF1, Napsin A, and programed death ligand 1 (PD-L1) were performed as routine clinical assays using Ventana XT (Ventana Medical Systems) and Leica Bond III (Leica Microsystems) automated immunohistochemistry platform as described previously. 22 The monoclonal antibody clone 22C3 (KEYTRUDA) (Neogenomics) and OptiView Detection System (Ventana Medical Systems) were used for PD-L1 staining. High expression is defined as 50% or greater Tumor Proportion Score.

| Statistical analysis
The Fisher exact test or χ 2 test was performed to calculate P values.

| IDH1/2 mutations in lung adenocarcinomas
NGS detected 11 IDH1/2 mutations in 10 specimens (Table 1). These included 8 specimens with an IDH1 mutation (4 with p.R132L or c.395G>T, 3 with p.R132H or c.395G>A, 1 with p.R132G or c.394C>G), 1 specimen with an IDH2 mutation (p. R172S or c.516G>T), and 1 specimen with 2 IHD2 mutations (p.R140Q or c.419G>A and p.R172M or c.515G>T). The variant allele frequency (VAF) was less than 5% in 4 specimens, including cases 3 and 10 harboring a cytosine deamination change (p.R132H or c.395G>A). Repeat of NGS showed a concordant VAF (2.9% p.R132H in case 3, 3.3% p.R132G in case 5, 3.5% p.R132L in case 6, and 4.5% p.R132H in case 10). DNA was isolated from the adjacent nonneoplastic tissues of case 3 (one subarea with no tumors) and case 10 (two separate subareas with no tumors). NGS showed c.395G>A in both nonneoplastic and neoplastic subareas of case 10 ( Figure  1A,B), but only in the neoplastic subarea of case 3 ( Figure  2A,B). Analysis of the entire NGS panel revealed a TP53 p.C242W (c.726C>G) in the neoplastic subarea, but not the two nonneoplastic subareas of case 10 ( Figure 1C,D). The results indicated cytosine deamination artifact of case 10 leading to a false detection of IDH1 c.395G>A. This specimen was excluded for further analysis of IDH1/2-mutated NSCLCs.
IDH1/2 mutations were detected in nine (0.5%) of 1924 patients with NSCLCs ( Table 2). All nine patients were diagnosed as adenocarcinoma. There were three males and six females. The age ranged from 55 to 89 years with a median of 78 years. Seven (78%) of nine patients are 70 years or older. Six were current or former smokers and two were never smokers. Seven patients presented as a newly diagnosed lung cancer (3 with stage IA, 1 with stage IIIB, and 3 with stage IV). Case 9 revealed a residual lung adenocarcinoma, 18 months after the initial diagnosis of stage IIIB.

| Histomorphology and immunophenotypes of IDH1/2-mutated lung adenocarcinomas
Specimens were taken by fine-needle aspiration (FNA) of lung or lymph node (n = 3), thoracentesis of pleural effusion (n = 2), or core biopsy of lung or pleura (n = 4). Histomorphology of the core biopsy specimens and the cell blocks of FNA or thoracentesis specimens were reviewed by a cytopathologist (ER) and a pulmonary pathologist (PI). Necrosis was seen in three biopsy specimens and two FNA

Cases
Tumor % a 7-gene specimens. All five cytology specimens were characterized by high-grade cytomorphology, such as high nuclear to cytoplasmic ratio, marked pleomorphism and/or prominent nucleoli ( Figure 3A). All four biopsy specimens also showed high-grade histopathology with marked pleomorphism (n = 4) and a predominantly solid pattern (n = 3) ( Figure  3B). Extensive clear cell change, as described by Toth et al, was not observed in all nine adenocarcinomas. Tumor cells were immunoreactive with TTF1 in eight of eight specimens examined and immunoreactive F I G U R E 1 False detection of IDH1 p.R132H (c.395G>A) change resulting from cytosine deamination artifact. In case 10, IDH1 c.395G>A was detected in the non-neoplastic subarea (A) and neoplastic subarea (B). TP53 c.726C>G was not detected in the non-neoplastic subarea (2/3792 = 0.05%, below the limit of detection) (C), but was present in the neoplastic subarea (D). Percentage in parentheses indicates variant allele frequency F I G U R E 2 Subclonal evolution of IDH1 p.R132H (c.395G>A) mutation. In case 3, NGS revealed absence of IDH1 (7/3344 = 0.2%, below the limit of detection) and EGFR mutations in the non-neoplastic tissue (A and D), 3.0% IDH1 and 31% EGFR mutations in the neoplastic tissue of biopsy specimen (B and E), and only EGFR mutation in the neoplastic tissue of resection specimen (6/5592 = 0.1% IDH1, below the limit of detection) (C and F). Percentage in parentheses indicates variant allele frequency with Napsin A in six of six specimens examined, confirming a diagnosis of lung adenocarcinoma (Table 2). Immunohistochemical stains were not performed for case 3 taken from the right upper lobe and harboring an EGFR exon 19 deletion mutation, supporting a diagnosis of lung adenocarcinoma. High expression (50% or greater) of PD-L1 was observed in two of five specimens examined.

| Subclonal IDH1/2 mutations
VAFs of IDH1/2 mutants were higher than or concordant with those of KRAS mutants in cases 1 and 2, but were lower than those of KRAS, EGFR, or PIK3CA mutants in the remaining cases (Figure 4). The findings suggest a higher incidence of mutant allele-specific imbalance of the KRAS, EGFR, or PIK3A mutations or presence of subclonal tumor populations harboring IDH1/2 mutations. The observations of less than 5% VAF of the IDH1 mutation in cases 3, 5, and 6 in a contexts of 51%-70%, 31%-50%, and 41%-60% estimated tumor cellularity and 33%, 13%, and 18% VAF of the coexisting EGFR, KRAS, or PIK3CA mutation suggest presence of IDH1 mutation in a tumor subpopulation. Subclonal evolution is also supported by examination of the resection specimen taken from case 3 at 2 months after biopsy. While NGS revealed 31% EGFR p.E746_A750del mutation and 3.0% IDH1 p.R132H mutation in the biopsy specimen, the resection specimen showed the same EGFR mutation at 25%, but no IDH1 mutation ( Figure 2). Furthermore, the biopsy specimen showed a high-grade features ( Figure  3C,D), but the resection specimen showed a well-differentiated adenocarcinoma with a lepidic growth pattern.
Tumor heterogeneity has important clinical applications in targeted therapy. 23,24 Trunk (initiating) driver mutations initiate the development of a founding cancer cell. Subsequently, branching driver mutations promote subclonal evolution. Therefore, trunk drivers are present in each cancer cell of the primary and metastatic tumors, while branching drivers may be present in a dominant subpopulation or a minor subclone. Targeting a mutation present only in a minor subclone likely results in a shorter progression-free survival, even though the initial efficacy is observed. A combinatory therapy to include trunk drivers, if present, for targeting may provide better treatment outcomes. We have previously shown that lower than expected VAF indicates intratumor heterogeneity while higher than expected VAF indicates mutant allele-specific imbalance. 25 In this study, the high incidence of coexisting KRAS and EGFR driver mutations and the lower than expected IDH1/2 VAF suggest that IDH1/2 mutations may be branching drivers promoting subclonal progression of lung adenocarcinomas.
By tracking the evolution of lung adenocarcinomas, most BRAF, EGFR, and KRAS activating mutations are trunk drivers. 26 These trunk drivers are usually mutually exclusive, although rare cases with coexisting EGFR and KRAS mutations have been reported. 27 In contrast, TP53 and PIK3CA mutations are branching drivers and can be seen in BRAF-, EFGR-, or KRAS-mutated lung adenocarcinomas. In a total of 25 IDH1/2-mutated lung adenocarcinomas reported by Toth et al, TCGA, MSK-IMPACT, and this study, known trunk F I G U R E 4 Correlation of variant allele frequency (VAF). VAF of IDH1/2 mutants vs VAF of KRAS (7 specimens), EGFR (one specimen) and PIK3CA (one specimen) mutants (mut 1 of the x-axis), including case 9 with two IDH2 mutations drivers were seen in 23 cases. 14 These included 18 cases with a KRAS mutation, four cases with an EGFR mutation, and one case with BRAF p.V600E. Coexistence of IDH1/2 mutations with trunk drivers within different genes suggests IDH1/2 mutations are branching drivers. In case 3 from our cohort, the observation of an EGFR mutation in both the biopsy and resection specimens of the same tumor, but the presence of IDH1 mutation only in the biopsy specimen also supports that the IDH1 mutation was a branching driver occurring during the subclonal evolution in lung adenocarcinoma. Immunohistochemical stains could also be helpful to identify tumor heterogeneity of the IDH1 p.R132H mutant.
Mutations within the same signature transduction pathway are redundant and, therefore, most mutually exclusive. Detection of coexisting mutations within the same pathway raises the concern for potential laboratory errors. 27 In a previous study for quality assessment, we confirmed that KRAS and NRAS mutations may be present in the same population or different subpopulations of colorectal cancers according to an operating procedure proposed for validation of unexpected coexisting mutations. 28 Concurrent IDH1 and IDH2 mutations have been reported in two AMLs, 5 four anaplastic gliomas, 8 and three chondrosarcomas. 29 Whether multiple IDH1/2 mutations occurred in the same or different tumor subpopulations was not described. In this study, multiregional analyses of case 9 support the presence of two distinct IDH2 mutations in different subpopulations harboring the same trunk driver KRAS mutation.
Parallel evolution in neoplasms refers to evolution of distinct subpopulations from a common ancestral clone as a consequence of independent mutations affecting the same gene or genes in the same pathway. 30,31 Parallel evolution has been shown in a variety of neoplasms, including lung cancers involving genetic alterations in the MUC1, CDK4, CHD8, and NKX2-1 genes. 26,30,31 Results of multiregional analyses in case 9 indicate paralleled evolution involving the IDH2 gene and support subclonal evolution promoted by two IDH2 mutations as branching drivers.
Previous studies have confirmed the role of IDH1 and IDH2 driver mutations in tumorigenesis of AML and gliomas. 16,32,33 In vitro studies showed IDH1 p.R132H mutation enhances migration and proliferation of NSCLC cells through downregulation of fibulin 5 by hypermethylation of the promoter. 34 In lung cancers, plasma IDH1 and IDH2 proteins are elevated and could be novel biomarkers for diagnosis of NSCLCs. 35,36 A germline variant (rs11540478) within the IDH2 gene is associated with an increased risk of lung cancers. 37 Cannataro et al have proposed a novel concept, the cancer effect size of genetic alterations, to estimate the intensity of selective advantage to cancer cells. 38 As expected, mutations of IDH1 codon 132 show a very high effect size (greater than 10 6 or 10 7 ) in lower-grade glioma of the brain. IDH1 p.R132C also shows a relatively higher effect size (10 4 -10 5 ) in lung adenocarcinomas. These findings suggest IDH1/2 mutations may confer selective advantage for clonal evolution in lung adenocarcinomas.
IDH1 and IDH2 mutations are detected in approximately 60%-80% and 1%-5% of patients with WHO grades II-III astrocytoma or oligodendroglioma and are predominantly IDH1 p.R132H (>90%). 8 In gliomas of the brain, IDH1/2 mutations are trunk drivers to confer cancer initiation and are favorable prognostic markers. 8,16,39,40 Combined analysis of 26 IDH1/2 mutations in NSCLCs reported in the previous and current studies revealed predominantly IDH1 p.R132C (35%) and p.R132L F I G U R E 5 Multiregional analysis of KRAS mutated lung adenocarcinoma harboring two IDH2 mutations. In case 9, subarea 2 showed a dominant population with IDH2 p.R140Q and subarea 3 showed a dominant population with IDH2 p.R172M mutation. Percentage in parentheses indicates variant allele frequency (27%). IDH1 p.R132H was seen in only two cases (7.7%). In this study, IDH1/2 mutations occurred in adenocarcinomas with high-grade features and may be branching drivers leading to subclonal evolution. Further prospective studies in a larger cohort are warranted to elucidate if IDH1/2 mutations are markers for theranostics and/or prognostication in lung cancers.
We and others have shown that cytosine deamination changes induced by formalin fixation or thermocycling of polymerase chain reaction is a major cause of background noise in both Sanger sequencing and NGS. 21,41,42 Pretreatment of DNA specimens with uracil-N-glycosylase can reduce the level of cytosine deamination artifact. 42,43 During our validation of this NGS panel, we have shown that background noise is consistent with spontaneous and formalin-induced cytosine deamination change. 21 Therefore, a cutoff of background noise at 2% was chosen for single-nucleotide variations. In our daily routine practice, background noise resulting from cytosine deamination changes is often less 2%. However, up to 2%-5% of cytosine deamination artifact occurring at multiple-nucleotide positions can be observed when the DNA input is suboptimal (usually less than 10 ng), consistent with previous observations that the chance of cytosine deamination artifact was inversely correlated with input of DNA quantity for sequencing. 41 Repeat of NGS often shows a low level of cytosine deamination changes occurring at other nucleotide positions. The observation of IDH1 c.359G>A (p.R132H) in case 10 was uncommon for cytosine deamination artifacts. In this specimen, cytosine deamination changes with frequency of 2% or more at other nucleotide positions were not observed and repeat of NGS showed persistent IDH1 c.359G>A change. However, NGS analyses of two subareas containing no tumor cells also show c.359G>A change indicating a false detection of IDH1 c.395G>A.
Although the incidence of IDH1/2 mutations is only 0.4%-1.1% in NSCLCs, it is worthwhile to investigate IDH1/2 inhibitors because of the high prevalence of lung cancers in the United States and worldwide. On the other hand, IDH1/2 mutations may be branching drivers leading to subclonal evolution, which may affect the benefit of IDH1/2 inhibitors. Accumulation of more IDH1/2-mutated NSCLCs is needed to elucidate their clinicopathological characteristics and implications for targeted therapy.