Genetic Variants in Histone Modification Regions Predict Clinical Outcomes of Pemetrexed Chemotherapy in Lung Adenocarcinoma

Abstract

Objective

This study was conducted to investigate the association between genetic variants in histone modification regions and clinical outcomes of PEM chemotherapy in patients with lung adenocarcinoma.

Methods

Potentially functional SNPs were selected using integrated analysis of ChIP-seq and RNA-seq. The associations of 279 SNPs with chemotherapy response and overall survival (OS) were analyzed in 314 lung adenocarcinoma patients who underwent PEM chemotherapy.

Results

Among the SNPs investigated, 18 were significantly associated with response to chemotherapy, while 28 with OS. Of these SNPs, rs549794A>G in an enhancer which is expected to regulate the expression of ribosomal protein S3 (RPS3) gene was significantly associated with both worse response to chemotherapy and worse OS (adjusted odds ratio = 0.59, 95% CI = 0.36–0.97, p = 0.04; adjusted hazard ratio = 1.44, 95% CI = 1.09–1.91, p = 0.01, respectively). Previous studies suggested that RPS3, a multi-functional protein with various extraribosomal activities, may play a role in chemotherapy resistance. Therefore, it is postulated that rs549794-induced change in the expression level of RPS3 may affect the response to PEM chemotherapy and consequently the survival outcomes in lung adenocarcinoma patients.

Conclusion

This study suggests that genetic variants in the histone modification regions may be useful for the prediction of clinical outcomes of PEM chemotherapy in advanced lung adenocarcinoma.

Introduction

Lung cancer is the leading cause of cancer-related death worldwide, representing approximately one in 5 cancer deaths, with an average 5-year survival rate of 22% [1]. The breakthrough targeted therapy and more recent innovative immunotherapy have shown unprecedented clinical success in the treatment of NSCLC patients [2]. However, only a subset of all lung cancer patients who have targetable driver mutations can get the therapeutic benefit of targeted therapy. In addition, only a proportion of patients obtain response to immune checkpoint inhibitors because of the diverse mechanisms of resistance [3]. Therefore, chemotherapy alone or in combination with immune checkpoint inhibitors is still an important modality of systemic treatment in advanced NSCLC patients. Lung adenocarcinoma is the most common subtype of lung cancer, comprising approximately 50% of all lung cancer cases [4]. Pemetrexed (PEM) is the most frequently used chemotherapeutic agents for lung adenocarcinoma. PEM is an anti-folate agent that inhibits the synthesis of DNA and RNA by targeting three enzymes involved in purine and pyrimidine synthesis: thymidylate synthase, dihydrofolate reductase, and glycinamide ribonucleotide formyltransferase [5]. Although the efficacy of PEM has been proven as a first-line treatment in combination with cisplatin (CIS), second-line, or as a maintenance therapy in advanced stage lung adenocarcinoma [6, 7, 8], therapeutic outcome of PEM varies significantly among individual patients with similar clinical profiles. Despite the considerable effort to discover and validate molecular biomarkers to predict efficacy of PEM, no viable biomarkers have yet been clearly established in clinical practice.

Cancer has traditionally been considered a set of diseases that are caused by the accumulation of genetic mutations [9]. It is now well accepted that epigenetic changes, including DNA methylation, microRNA regulation, and histone modifications, also play important roles in the pathogenesis of cancer [10, 11]. A variety of epigenetic modifications are known to be linked with the development and progression of lung cancer, and the possible involvement in the resistance to chemotherapy has also been suggested [12, 13]. Histones are the central component of the nucleosomes which are the fundamental building blocks of chromatin. The histone tails are subject to extensive posttranslational modifications such as methylation and acetylation, which may contribute to transcriptional processes [14]. In human cancers, dysregulation of these processes leading to aberrant gene expression is frequently observed [15]. Studies have revealed the effects of genetic variations in the regulation of epigenome [16, 17, 18] and their association with the risk or clinical outcomes of human cancers [19, 20, 21].

In this study, we hypothesized that genetic variants of histone modification regions may affect the chemotherapy response and survival outcomes by regulating the expression of genes involved in the pathogenesis of lung cancer. To test this hypothesis, we investigated the association between functional single-nucleotide polymorphisms (SNPs) in histone modification regions and chemotherapy response and survival of patients with lung adenocarcinoma who were treated with PEM chemotherapy.

Methods

Study Population

This study included a total of 314 lung adenocarcinoma patients with stages III/IV or recurrent disease after curative surgery who had available DNA samples and received PEM chemotherapy at Kyungpook National University Hospital (KNUH) in Daegu, Korea, from March 2007 to July 2015. Patients were treated with either PEM + CIS for up to four to six cycles as the first-line chemotherapy with or without PEM maintenance therapy, or PEM alone as the second- or further line chemotherapy. Patients who were treated with PEM + CIS before the approval of maintenance PEM therapy in Korea in 2009 or who had a progressive disease (PD) with the first-line therapy did not undergo PEM maintenance therapy. Assessment of tumor response was carried out by computed tomography scan every two cycles. Responses were determined using response evaluation criteria in solid tumors [22]. Patients with a complete response (CR) or a partial response (PR) were defined as responders, and patients having stable disease (SD) or PD were defined as nonresponders. Overall survival (OS) was defined as the time between the date of first chemotherapy and the date of death or last follow-up. Genomic DNA samples from the patients were provided by the National Biobank of Korea, KNUH, which is supported by the Ministry of Health, Welfare and Family Affairs. Written informed consent was obtained from all patients. This study was approved by the Institutional Review Boards of the KNUH (Approval No. KNUH 2017-07-012) and was performed in accordance with relevant guidelines and regulations.

Cell Culture and Antibodies

For chromatin immunoprecipitation (ChIP) assay, H2087 cells were obtained from the American Type Culture Collection (ATCC) and anti-Histone H3 antibodies (ab8580 and ab4729) were purchased from Abcam (Cambridge, UK).

ChIP-Sequencing

ChIP-sequencing was performed as described previously [23]. ChIP assays were performed using the Pierce Magnetic ChIP kit (Thermo Fisher Scientific, Waltham, MA, USA). H2087 cells were crosslinked with 1% formaldehyde for 10 min, and the crosslinking was inactivated by 0.125 M glycine for 5 min at room temperature. After washing, the cells were lysed, sonicated to shear DNA. To immunoprecipitate protein/chromatin complexes, the diluted supernatants were incubated with 10 μg of H3K4me3 or H3k27ac antibody overnight, to which 50 μL of agarose/protein A or G beads was added and incubated for 2 h. Several washing steps were followed by protein digestion using proteinase K. Reverse crosslinking was carried out at 65°C. DNA was subsequently purified. Library preparation was conducted using the TruSeq ChIP Library Preparation Kit (Illumina, San Diego, CA, USA), and sequencing was performed on an Illumina HiSeq4000. Sequence reads for each sample were aligned to the human genome using Bowtie [24]. The reference genome sequence of Homo sapiens (hg19) and annotation data were downloaded from the UCSC table browser (http://genome.uscs.edu). Peaks were called in the aligned sequence data using a model-based analysis of ChIP-seq (MACS2 version 2.1.1) (https://bioweb.pasteur.fr/packages/pack@macs@2.1.1) [25]. ChIPseeker (version 1.6.6) (http://www.bioconductor.org/packages/release/bioc/html/ChIPseeker.html) [26], a bioconductor package within the statistical programming environment R to facilitate batch annotation of enriched peaks identified from ChIP-seq data, was used to identify nearby genes and transcripts from the peaks obtained from MACS2.

RNA-Sequencing

Total RNAs from H2087 cells were isolated using TRIzol (Invitrogen, Carlsbad, CA, USA). Sequencing was performed on an Illumina HiSeq4000 and aligned the processed reads to the Homo sapiens genome sequence (hg19) using HISAT v2.0.5 [27]. Transcript assembly and abundance estimation were performed using StringTie v1.3.3b [28, 29], which provides the relative abundance estimates as fragments per kilobase of exon per million fragments mapped (FPKM) values of transcript and gene expressed in each sample. FPKM values have been normalized with respect to the library size.

SNP Selection and Genotyping

We conducted an integrated analysis of ChIP-seq and RNA-seq for the SNP selection. As a result of the ChIP-seq, SNPs within H3K4me3 and H3K27ac peak regions were selected. Next, using the FuncPred utility for functional SNP prediction in the SNPinfo web server (https://snpinfo.niehs.nih.gov), potentially functional variants were collected. And then, using RNA-seq we chose genes with a high expression level (FPKM ≥100), and SNPs within or closest to the genes were extracted. Genotyping was performed using iPLEX Assay and MassARRAY System (Agena Bioscience, San Diego, CA, USA).

Statistical Analysis

Hardy-Weinberg equilibrium was tested using a goodness-of-fit χ2 test with 1 degree of freedom. The genotypes for each SNP were analyzed as a three-group categorical variable and also analyzed under dominant, recessive, and codominant genetic models. The association between clinical variables or genotypes and chemotherapy response was tested by odds ratio (OR) and 95% confidence intervals (CIs) using unconditional logistic regression analysis. Estimated survival rate was calculated using the Kaplan-Meier method, and the difference in OS according to different clinical variables or genotypes was compared using log-rank tests. Cox's proportional hazard regression model was used for the multivariate survival analyses, estimating the hazard ratio (HR) and 95% CI. A cut-off p value of 0.05 was adopted for all the statistical analyses. All analyses were carried out using Statistical Analysis System for Windows, version 9.4 (SAS Institute, Cary, NC, USA).

Results

Patient Characteristics and Clinical Outcomes

The baseline clinical and pathologic characteristics of study population and their association with response to chemotherapy and OS are presented in Table 1. The overall response rate was 41.4%. The observed number of deaths in the cohort was 239 (76.1%) and median survival time (MST) was 24.4 months (95% CI = 21.7–26.2 months). Age ≥64, no benefit from tyrosine kinase inhibitor (TKI) treatment which includes wild-type EGFR/ALK or no response to EGFR/ALK-TKIs, PEM maintenance therapy, and PEM + CIS regimen were associated with better response in the univariate analyses. Regarding the OS, female sex, never-smoking status, recurred disease after surgery versus stage III/IV, Eastern Cooperative Oncology Group (ECOG) performance status (PS) 0, benefit from TKIs, PEM maintenance therapy, and PEM alone were significantly associated with the better OS. The conflicting results may be partly explained by the fact that patients with wild-type EGFR/ALK received first-line PEM + CIS while those with upfront EGFR/ALK-TKI treatment underwent second-line PEM monotherapy, because the PEM + CIS provides higher response rate than PEM monotherapy, but EGFR/ALK-TKI treatment is associated with better survival than chemotherapy.

Table 1.

Univariate analysis for response to chemotherapy and OS by clinical variables

Variables	Case, n	Response to chemotherapy				OS
		responders (CR + PR, %)a	nonresponders (SD + PD, %)a	OR (95% CI)	p value	MST (months)	95% CI	Log-rank P	HR (95% CI)	p value
Overall	314	130 (41.4)	184 (58.6)			24.4	21.7–26.2
Age (years)
<64	153	52 (34.0)	101 (66.0)	1.00	0.01	25.9	23.6–32.9	0.06	1.00	0.06
≥64	161	78 (48.5)	83 (51.6)	1.83 (1.16–2.88)		22.5	17.2–25.1		1.28 (0.99–1.65)
Sex
Male	150	63 (42.0)	87 (58.0)	1.00	0.84	16.6	14.6–20.9	0.0001	1.00	0.0001
Female	164	67 (40.9)	97 (59.2)	0.95 (0.61–1.50)		28.6	25.1–34.3		0.60 (0.47–0.78)
Smoking status
Never	166	70 (42.2)	96 (57.8)	1.00	0.77	28.6	24.7–34.2	0.0002	1.00	0.0002
Ever	148	60 (40.5)	88 (59.5)	0.94 (0.60–1.47)		16.6	14.5–23.2		1.63 (1.26–2.11)
Stage
III/IV	244	101 (41.4)	143 (58.6)	1.00	1.00	22.9	18.8–24.7	0.001	1.00	0.002
Recurred	70	29 (41.4)	41 (58.6)	1.00 (0.58–1.72)		34.9	25.9–44.6		0.60 (0.43–0.82)
ECOG PS
0	102	45 (44.1)	57 (55.9)	1.00	0.50	29.5	24.6–35.4	0.01	1.00	0.01
1–2	212	85 (40.1)	127 (59.9)	0.85 (0.53–1.37)		21.9	17.6–24.7		1.45 (1.09–1.93)
TKI benefit
No	198	94 (47.5)	104 (52.5)	1.00	0.01	15.2	14.0–17.2	<0.001	1.00	<0.001
Yes	116	36 (31.0)	80 (69.0)	0.50 (0.31–0.81)		38.7	34.9–44.1		0.32 (0.24–0.42)
Maintenanceb
No	112	51 (45.5)	61 (54.5)	1.00	0.01	14.8	10.8–17.1	1×10−5	1.00	2×10−5
Yes	67	45 (67.2)	22 (32.8)	2.45 (1.30–4.60)		27.2	16.6–		0.41 (0.27–0.62)
Regimen
PEM/CIS	179	96 (53.6)	83 (46.4)	1.00	<0.001	16.7	14.7–20.9	0.0001	1.00	0.0001
PEM alone	135	34 (25.2)	101 (74.8)	0.29 (0.18–0.47)		28.8	25.4–35.4		0.60 (0.47–0.78)

CR, complete response; PR, partial response; SD, stable disease; PD, progressive disease; OR, odds ratio; MST, median survival time; CI, confidence interval; HR, hazard ratio; ECOG, Eastern Cooperative Oncology Group; PS, performance status; TKI, tyrosine kinase inhibitor.

^aRow percentage.

^bAmong patients with PEM/CIS.

Identification of Potentially Functional SNPs Using ChIP-Seq and RNA-Seq

We performed ChIP-seq using two posttranslational modifications of histone H3 (H3K4me3 and H3K27ac). Based on the ChIP-seq data for two histone marks, 31,582 SNPs located within H3K4me3 peaks and 34,591 within H3K27ac peaks were retrieved. And then, based on the NCBI SNP database (https://snpinfo.niehs.nih.gov), 1,654 and 1,229 potentially functional SNPs within H3K4me3 and H3K27ac peaks, respectively, with minor allele frequency (MAF) ≥0.1 were collected after excluding those in linkage disequilibrium (r² ≥ 0.8). Among those, 383 SNPs within or closest to the genes with expression level of FPKM ≥100 on the RNA-seq data were extracted, of which 320 were identified as tag SNPs. Finally, the 279 SNPs were selected after subtracting 41 SNPs located within the overlap between the H3K4me3 peaks and H3K27ac peaks.

Associations between the SNPs and Chemotherapy Outcomes

Of the 279 SNPs genotyped, 213 were analyzed for the association study after excluding 10 with genotyping failure and 56 with deviation from the HWE (p < 0.05) or call rate <95% (online Suppl. Table 1; for all online suppl. material, see www.karger.com/doi/10.1159/000527492). Among these, 18 were significantly associated with response to chemotherapy (Table 2), while 28 with OS (Table 3). One SNP, rs549794A>G, was significantly associated with both worse response to chemotherapy and worse OS under dominant model (adjusted OR [aOR] = 0.59, 95% CI = 0.36–0.97, p = 0.04; adjusted HR [aHR] = 1.44, 95% CI = 1.09–1.91, p = 0.01, respectively) and under codominant model (aOR = 0.69, 95% CI = 0.47–1.00, p = 0.05; aHR = 1.24, 95% CI = 1.02–1.51, p = 0.03, respectively) (Table 4; Fig. 1). The rs549794 resides within a H3K27ac peak which marks an active enhancer but not within a H3K4me3 peak which marks active promoter, although rs549794 is located in the 5′ UTR of in Kelch Like Family Member 35 (KLHL35) gene. The enhancer containing rs549794 is within 30 kb from ribosomal protein S3 (RPS3) gene with a high expression level (FPKM ≥100) and expected to regulate its expression. Next, subgroup analysis was performed to investigate the influence of chemotherapy regimen. In patients who underwent PEM + CIS, significant associations between the rs549794A>G and both worse response to chemotherapy and worse OS remained (aOR = 0.60, 95% CI = 0.37–0.97, p = 0.04; aHR = 1.36, 95% CI = 1.02–1.83, p = 0.04, respectively, under codominant model) (Table 5). However, in patients with PEM alone, the association was not significant although a test for heterogeneity revealed that the difference between PEM + CIS and PEM subgroups was not significant (P for heterogeneity test >0.05).

Table 2.

Summary of 18 SNPs and response to chemotherapy

Polymorphism IDb	P for response to chemotherapya
	gene	alleles	CR (%)	MAF	HWE-P	dominant	recessive	codominant
rs549794	KLHL35	A>G	96.8	0.34	0.63	0.04	0.40	0.05
rs800351	TSSC4	T>C	98.4	0.48	0.79	0.03	0.06	0.79
rs1529959	RPS5	G>A	98.4	0.49	0.70	0.002	0.17	0.01
rs2168762	KRT87P	G>A	96.8	0.45	0.55	0.01	0.05	0.01
rs2302175	NDUFB10	C>T	96.2	0.34	0.46	0.002	0.62	0.05
rs2012124	MIF	C>T	97.8	0.38	0.41	0.03	0.62	0.22
rs2735784	ELF3-AS1	A>G	95.5	0.45	0.08	0.03	0.94	0.15
rs2983641	CST9	C>T	96.8	0.19	0.34	0.02	0.51	0.03
rs7897156	BUB3	C>T	96.5	0.33	0.60	0.03	0.01	0.01
rs11246331	TSPAN4	C>T	97.8	0.22	0.96	0.27	0.01	0.05
rs2070876	KRT18	T>C	98.1	0.44	0.91	0.83	0.03	0.16
rs2486256	TAGLN2	A>G	97.8	0.31	0.85	0.35	0.01	0.67
rs10902227	TSPAN4	C>T	98.7	0.37	0.28	0.94	0.03	0.23
rs2712429	−	C>A	98.1	0.16	0.26	0.59	0.03	0.71
rs9427715	ELF3	T>G	98.1	0.49	0.31	0.81	0.03	0.14
rs972892	NDUFS6	A>G	96.8	0.44	0.60	0.71	0.04	0.36
rs2748240	RPS14	G>A	98.1	0.43	0.64	0.06	0.20	0.05
rs2079786	TRMT112	T>G	98.7	0.12	0.75	0.06	0.99	0.03

MAF, minor allele frequency; HWE-P, P for Hardy-Weinberg equilibrium test.

^ap values were calculated using multivariate Cox proportional hazard models, adjusted for age, sex, smoking status, stage, ECOG PS, and PEM/CIS or PEM alone.

^bInformation about polymorphisms and IDs was obtained from NCBI database (http://www.ncbi.nlm.nih.gov/SNP).

Table 3.

Summary of 28 SNPs and OS

Polymorphism IDb	P for OSa
	gene	alleles	CR (%)	MAF	HWE-P	dominant	recessive	codominant
rs549794	KLHL35	A>G	96.8	0.34	0.63	0.01	0.58	0.03
rs296887	HNRNPK	G>A	97.5	0.27	0.76	0.03	0.31	0.03
rs1889532	ARHGEF2	G>A	98.1	0.10	0.27	0.003	0.12	0.002
rs2240275	KIAA0753	G>A	97.5	0.36	0.18	0.02	0.58	0.05
rs284573	RPL37A	C>G	98.4	0.21	0.66	0.02	0.06	0.11
rs3767199	QSOX1	G>T	98.4	0.10	0.90	0.01	0.45	0.01
rs3812445	PYCR3	A>G	96.5	0.15	0.11	0.04	0.44	0.11
rs6502051	CCDC57	A>C	98.7	0.11	0.76	0.01	0.69	0.01
rs2913861	PHYKPL	T>C	97.1	0.16	0.30	0.02	0.52	0.02
rs1242095	LGMN	A>C	98.4	0.36	0.28	0.02	0.83	0.09
rs1279738	RACK1	C>G	98.1	0.39	0.38	0.02	0.77	0.13
rs2425047	EIF6	A>C	98.7	0.17	0.52	0.02	0.74	0.03
rs2853834	CCT8	A>G	98.1	0.42	0.84	0.01	0.43	0.04
rs7648309	TKT	G>A	98.4	0.34	0.61	0.04	0.49	0.07
rs7935835	-	G>A	98.7	0.16	0.69	0.004	0.92	0.02
rs2227564	C10orf55	C>T	98.7	0.27	0.61	0.82	0.01	0.26
rs36615	RPS9	G>A	98.7	0.28	0.31	0.84	0.01	0.45
rs1461496	HSPA8	G>A	96.8	0.48	0.93	0.82	0.02	0.18
rs2073098	PQLC2	C>T	98.4	0.23	0.59	0.71	0.01	0.17
rs2958153	PTGES3	G>A	98.4	0.24	0.48	0.33	0.01	0.95
rs4669	TGFBI	T>C	97.5	0.36	0.72	0.18	0.05	0.05
rs8003631	ACIN1	T>C	98.7	0.42	0.30	0.58	0.01	0.10
rs937215	SRP14	T>C	98.1	0.21	0.92	0.15	0.03	0.06
rs2071391	HNRNPA1	A>G	97.1	0.49	0.40	0.08	0.13	0.05
rs8316	DDIT4	T>C	98.7	0.11	0.37	0.20	0.03	0.09
rs11609253	TMEM106C	A>C	98.7	0.14	0.41	0.15	0.03	0.06
rs17528989	WDR77	G>A	98.7	0.15	0.83	0.41	0.02	0.18
rs3806515	HADHA	C>T	98.1	0.12	0.20	0.06	0.01	0.03

MAF, minor allele frequency; HWE-P, P for Hardy-Weinberg equilibrium test.

^ap values were calculated using multivariate Cox proportional hazard models, adjusted for age, sex, smoking status, stage, ECOG PS, PEM/CIS or PEM alone, TKI benefit, and maintenance therapy.

^bInformation about polymorphisms and IDs was obtained from NCBI database (http://www.ncbi.nlm.nih.gov/SNP).

Table 4.

Genotypes of rs549794 and their associations with the response to chemotherapy and OS

Polymorphism/genotype	Location	Cases, n (%)a	Response to chemotherapy				OS
			responders (%)b	nonresponders (%)b	OR (95% CI)c	p valuec	L-R P	HR (95% CI)d	p valued
rs549794e
AA		129 (42.4)	62 (48.1)	67 (51.9)	1.00		0.33	1.00
AG		141 (46.4)	51 (36.2)	90 (63.8)	0.61 (0.36–1.02)	0.06		1.46 (1.09–1.95)	0.01
GG	5′UTR	34 (11.2)	10 (29.4)	24 (70.6)	0.55 (0.23–1.29)	0.17		1.38 (0.88–2.15)	0.16
Dominant					0.59 (0.36–0.97)	0.04	0.18	1.44 (1.09–1.91)	0.01
Recessive					0.70 (0.31–1.59)	0.40	0.84	1.12 (0.74–1.70)	0.58
Codominant					0.69 (0.47–1.00)	0.05		1.24 (1.02–1.51)	0.03

OR, odds ratio; CI, confidence interval; L-R P, log-rank P; HR, hazard ratio.

^aColumn percentage.

^bRow percentage.

^cORs, 95% CIs, and their corresponding P values were calculated using multivariate regression analysis, adjusted for age, sex, smoking status, stage, ECOG PS, and PEM/CIS or PEM alone.

^dHRs, 95% Cis, and their corresponding P values were calculated using multivariate Cox proportional hazard models, adjusted for age, sex, smoking status, stage, ECOG PS, PEM/CIS or PEM alone, TKI benefit, and maintenance therapy.

^eGenotype failure in 10 cases for rs549794 A<G.

Fig. 1.

Overall survival curves according to rs549794A>G. p values by multivariate Cox proportional hazard models.

Table 5.

The association between rs549794 genotypes and clinical outcomes according to chemotherapy regimens

	Polymorphism IDa	Response to chemotherapy								OS
	dominant			recessive			codominant			dominant			recessive			codominant
	OR (95% Cl)b	P valueb	PHd	OR (95% Cl)b	P valueb	PHd	OR (95% Cl)b	P valueb	PHd	HR (95% Cl)c	P valuec	PHd	HR (95% Cl)c	P valuec	PHd	HR (95% Cl)c	P valuec	PHd
rs549794
PEM/CIS	0.54 (0.29–1.01)	0.05	0.68	0.47(0.15–1.47)	0.19	0.30	0.60 (0.37–0.97)	0.04	0.39	1.42 (0.96–2.10)	0.08	0.98	1.62 (0.87–3.01)	0.13	0.17	1.36(1.02–1.83)	0.04	0.42
PEM alone	0.68 (0.28–1.65)	0.39		1.11 (0.34–3.61)	0.86		0.85 (0.45–1.61)	0.61		1.43 (0.90–2.25)	0.13		0.88 (0.49–1.61)	0.68		1.14(0.84–1.55)	0.39

OR, odds ratio; CI, confidence interval; HR, hazard ratio; PEM, pemetrexed; CIS, cisplatin.

^aInformation about polymorphisms and IDs was obtained from NCBI database (http://www.ncbi.nlm.nih.gov/SNP).

^b ORs, 95% CIs, and their corresponding p values were calculated using multivariate regression analysis, adjusted for age, sex, smoking status, stage, and ECOG PS.

^cHRs, 95% CIs, and their corresponding p values were calculated using multivariate Cox proportional hazard models, adjusted for age, sex, smoking status, stage, ECOG PS, TKI benefit, and maintenance therapy.

^dWald test for homogeneity of adjusted ORs or HRs between the two study samples.

Discussion

In this study, we aimed to investigate whether genetic variants in histone modification regions are associated with the treatment outcomes of PEM chemotherapy in lung adenocarcinoma patients. This study revealed that rs549794A>G was associated with worse response to PEM chemotherapy and worse OS in lung adenocarcinoma patients. Our results suggest that the rs549794A>G may be a potential biomarker that may contribute to predicting therapeutic outcomes in lung adenocarcinoma patients treated with PEM chemotherapy and help for an optimal personalized treatment strategy.

PEM inhibits replication and growth of tumor cells by interrupting folate metabolism crucial for DNA synthesis and repair [30]. Despite the proven efficacy and low toxicity profile, therapeutic outcome of PEM varies significantly among patients with lung adenocarcinoma and chemoresistance develops eventually for which various causative mechanisms have been suggested [31]. Upregulation of DNA repair systems, including base excision repair (BER), nucleotide excision repair (NER), as well as CHK1 and MSH2, has been suggested as the mechanisms of resistance to PEM chemotherapy [31, 32, 33, 34]. In this study, the rs549794A>G in an enhancer which is expected to regulate RPS3 expression was associated with clinical outcomes after PEM chemotherapy. RPS3 is a component of the 40S ribosomal subunit and participates in ribosomal maturation and translation initiation through cooperation with initiation factors [35]. RPS3 is also known to be a multi-functional protein with various extraribosomal functions, including DNA damage processing, apoptosis, tumorigenesis, and transcriptional regulation [36, 37, 38], which is puzzling and may be context dependent [37]. A previous study revealed that RPS3 interacts with and positively affects BER proteins, OGG1 and APE/Ref-1, suggesting a potential benefit to BER activities [39]. A recent study suggested that RPS3 augments NER efficiency by enhancing XPD helicase activity and expediting the NER process by increasing turnover rate of transcription factor IIH [40]. Therefore, RPS3 may be involved in the resistance to PEM chemotherapy by promoting repair of defective DNA. Other possible mechanisms of chemotherapy resistance by RPS3 have been suggested in some cancers. Huang et al. [41] showed that CIS induced overexpression of lipocalin 2 and that lipocalin2-RPS3 complex promoted the activation of NF-kB, leading to CIS resistance in oral squamous cell carcinoma. Sun et al. [42] demonstrated that RPS3 expression levels were significantly elevated in CIS-resistant gastric cancer cell exosomes, and exosomal RPS3 derived from CIS-resistant cells promoted the chemoresistance of CIS-sensitive cells through the activation of PI3K-Akt-cofilin-1 signaling pathway. In NSCLC, it has been proposed that RPS3 is involved in the mechanism of resistance to radiation therapy and radiation-induced metastatic conversion via activation of NF-kB [43, 44]. However, there has been no study that investigated the effect of RPS3 on clinical outcomes of chemotherapy in lung cancer patients. Based on the roles of RPS3 in previous studies, it is postulated that rs549794-induced change in the expression level of RPS3 may contribute to the resistance to PEM chemotherapy and consequently the clinical outcomes in lung adenocarcinoma patients. We also evaluated the effect of rs549794A>G on the clinical outcomes of lung adenocarcinoma patients who underwent curative surgery in a recently published study, but found no significant effect [23]. Considering these findings, it can be assumed that RPS3 is involved in molecular mechanisms of resistance to PEM chemotherapy, rather than directly affecting the disease course of lung adenocarcinoma. Further studies are required on the role of RPS3 in chemotherapy resistance and the molecular mechanism of the association between the rs549794 and clinical outcomes of PEM chemotherapy in lung adenocarcinoma.

In summary, this study demonstrates that genetic variations in histone modification regions, especially rs549794A>G, are associated with the clinical outcomes of PEM chemotherapy in patients with advanced lung adenocarcinoma. Our result suggests that RPS3 may play an important role in the resistance to PEM chemotherapy in lung adenocarcinoma. Evaluation of rs549794 may be useful for refining therapeutic decisions in the treatment of lung adenocarcinoma by helping to identify subgroups of patients who will have poor prognosis after PEM chemotherapy. Future studies with a larger population are needed to validate our findings.

Statement of Ethics

This study was conducted in accordance with the Declaration of Helsinki and Good Clinical Practice guidelines and was approved by the Institutional Review Board of the Kyungpook National University Hospital (Approval No. KNUH 2017-07-012). All patients provided written informed consent.

Conflict of Interest Statement

The authors declare no conflict of interests.

Funding Sources

This work was supported by the National R&D Program for Cancer Control, Ministry of Health and Welfare, Republic of Korea (Grant number: 1720040), and the National Research Foundation of Korea (NRF) grant funded by the Korean government (MIST) (NRF-2020R1A5A2017323).

Author Contributions

Conceived and designed the experiments: Jae Yong Park and Shin Yup Lee. Performed the experiments: Sook Kyung Do, Hyo-Gyoung Kang, Jin Eun Choi, Mi Jeong Hong, Jang Hyuck Lee, and Sunwoong Lee. Acquired clinical data: Yong Hoon Lee, Shin Yup Lee, Ji Eun Park, Sun Ha Choi, Hyewon Seo, Kyung Min Shin, Ji Yun Jeong, Seung Soo Yoo, Jaehee Lee, Seung Ick Cha, Chang Ho Kim, and Jae Yong Park. Analyzed and interpreted the data: Yong Hoon Lee, Sook Kyung Do, Shin Yup Lee, Won Kee Lee, and Jae Yong Park. Wrote the main manuscript text: Yong Hoon Lee, Sook Kyung Do, and Shin Yup Lee. Supervised the study: Jae Yong Park and Shin Yup Lee. All authors reviewed the manuscript.

Data Availability Statement

The datasets for the RNA-seq and ChIP-seq (H3K4me3, H3K27ac) have been deposited and are available at Gene Expression Omnibus (GEO accession No. GSE182385) (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc = GSE182385).

Supplementary Material

Supplementary data

Funding Statement

This work was supported by the National R&D Program for Cancer Control, Ministry of Health and Welfare, Republic of Korea (Grant number: 1720040), and the National Research Foundation of Korea (NRF) grant funded by the Korean government (MIST) (NRF-2020R1A5A2017323).