This article is mentioned in:

Introduction

Non-small-cell lung cancer (NSCLC) may establish an immunosuppressive tumor microenvironment that is conducive to tumor growth (1). Natural killer (NK) cells play a pivotal role in innate immunity and immunological surveillance (2). Activation of NK cells partly depends on the interactions between killer-cell immunoglobulin-like receptors (KIRs) and human leukocyte antigen (HLA) class I ligands (3). KIRs are a family of cell surface receptors expressed by NK cells and certain subpopulations of T lymphocytes. The KIR family consists of inhibitory as well as activating receptors characterized by both allelic (high numbers of variants) and haplotypic (different numbers of genes for inhibitory and activating receptors on individual chromosomes) polymorphisms (4). The ligands of KIRs are specific epitopes on HLA class I molecules (5). HLA-C allotypes fall into the C1 or C2 groups (asparagine and lysine, respectively, at position 80) that are recognized by KIR2DL2/KIR2DL3 and KIR2DL1/KIR2DS1, respectively. Two previous studies have implicated HLA-C1 and HLA-C2 as potential ligands for KIR2DS4 (6,7). KIR3DL1/KIR3DS1 recognizes HLA-Bw4 epitopes of HLA-A and HLA-B alleles, which are classified according to whether isoleucine or threonine is present at position 80 (4).

An increased understanding of the complexities of the biology of the KIR/HLA system has provided opportunities to leverage NK cell function as a novel avenue of immunotherapy for cancer (8). Epidemiological studies have associated particular KIR and HLA genotypes with the susceptibility to and clinical outcome of leukemia and certain types of solid tumors (9–11). There is a lack of clinical information on the role of KIR/HLA interactions in metastatic NSCLC (mNSCLC). Therefore, the present study was performed to investigate the effect of KIR/HLA ligand on the prognosis of Chinese Han mNSCLC patients treated with chemotherapy.

Patients and methods

Patients and follow-up

A total of 70 eligible Chinese Han patients who attended the Fudan University Shanghai Cancer Center (Shanghai, China) were enrolled in the present study. Eligible patients were aged ≥18 years and had pathologically confirmed metastatic or recurrent NSCLC. Patients were considered ineligible if they had received previous systemic anticancer therapy <1 year prior, had severe drug allergies, had an active infection or other serious diseases or conditions, or were unable to provide consent. Blood samples and laboratory blood test results were collected prior to chemotherapy. The performance status of the cancer patients was defined according to the Eastern Cooperative Oncology Group (ECOG). Eligible patients received platinum-based first-line chemotherapy. All the patients were followed up for 2 years after enrollment. Follow-up data for all patients were obtained from their most recent medical review, which consisted of a clinical examination and an assessment of chest X-rays or computed tomography scans. Progression-free survival (PFS) was defined as the time from the date of inclusion to the date of disease progression or death, whichever came first. Overall survival (OS) was defined as the time from the date of inclusion to the date of death. Data were censored if patients were non-progressing or alive at the time point of evaluation. If a patient missed a follow-up or had an unknown event date, the patient was censored at the last radiological assessment for PFS and the last contact date for OS. The Ethics Committee of Fudan University Shanghai Cancer Center approved this study and all the included patients provided written informed consent.

Sample preparation

Genomic DNA was extracted from peripheral blood mononuclear cells using the QIAamp DNA Mini kit (Qiagen GmbH, Hilden, Germany) according to the manufacturer's instructions. Total RNA was extracted from whole-blood samples with the PAXgeneÔ Blood RNA system (PreAnalytiX GmbH, Hombrechtikon, Switzerland). cDNA was then synthesized with the QuantiTect® reverse transcription kit (Qiagen).

KIR and HLA ligand typing

The presence or absence of 15 human KIR genes plus two pseudogenes at the level of genomic DNA and mRNA was analyzed by sequence-specific primer polymerase chain reaction (SSP-PCR) using the Miltenyi Biotec KIR typing kit (Miltenyi Biotec GmbH, Bergisch Gladbach, Germany) according to the manufacturer's protocol. Differentiation between KIR2DS4del (the 22 bp-deleted form of KIR2DS4) vs. KIR2DS4full (the full-length form of KIR2DS4) was also possible. The presence or absence of KIR-HLA ligands, including HLA-CAsn80 (HLA-C1), HLA-CLys80 (HLA-C2), HLA-BBw4+Thr80 (HLA-Bw4T80), HLA-BBw4+Ile80 (HLA-Bw4I80), HLA-ABw4+ and HLA-BBw4+Asp77,Thr80 was determined by SSP-PCR using the Olerup SSP KIR HLA Ligand kit (Olerup SSP, Stockholm, Sweden) according to the manufacturer's protocol.

Statistical analysis

Differences in categorical variables were analyzed with the Fisher's exact test. Frequencies of >20% and <80% for KIRs or HLA ligands were included in the univariate analysis. Survival curves were estimated using the Kaplan-Meier method and compared with the log-rank test. Only variables with P<0.1 in the univariate analysis (Tables II and III), apart from the epidermal growth factor receptor gene (EGFR) mutation status (17.1% of patients not tested), were entered in the multivariate model. The Cox proportional hazard regression model was applied for multivariate survival analysis, with the stepwise selection, to determine independent predictors of survival. The Benjamini-Hochberg method (12) was used to perform multitest correction. The contingency analysis was computed and the Fisher's exact test was used to investigate the independence between the significant KIR/HLA ligand and other variables. P<0.05 was considered to indicate statistically significant differences. All the analyses were conducted using SAS® 9.4 software (SAS Institute Inc., Cary, NC, USA).

Table II. Association of KIR and HLA class I ligands with survival prognosis of patients with mNSCLC by univariate analysis.

Table II.

Association of KIR and HLA class I ligands with survival prognosis of patients with mNSCLC by univariate analysis.

	PFS (negative vs. positive)		OS (negative vs. positive)
KIR and HLA typing (n=70)	HR (95% CI)	P-value	HR (95% CI)	P-value
KIR genotype
KIR2DL5all	0.58 (0.32–1.05)	0.072	1.05 (0.54–2.05)	0.888
KIR2DL5A	0.59 (0.32–1.10)	0.095	0.96 (0.49–1.90)	0.912
KIR2DL5B	0.64 (0.35–1.17)	0.148	1.06 (0.54–2.08)	0.870
KIR2DS4del	1.07 (0.62–1.85)	0.797	0.60 (0.31–1.16)	0.126
KIR2DS4full	1.29 (0.69–2.41)	0.430	1.39 (0.66–2.91)	0.387
KIR2DS4del+/KIR2DS4full−	0.97 (0.50–1.91)	0.935	0.60 (0.26–1.37)	0.227
KIR2DS4full+/KIR2DS4del−	1.11 (0.64–1.92)	0.716	1.57 (0.81–3.06)	0.182
KIR2DS4ful+/KIR2DS4del+	1.20 (0.67–2.16)	0.540	0.85 (0.40–1.82)	0.673
KIR3DS1	0.70 (0.38–1.28)	0.243	0.99 (0.50–1.95)	0.979
KIR2DS1	0.56 (0.30–1.05)	0.071	0.93 (0.47–1.85)	0.843
KIR2DS2	0.80 (0.38–1.69)	0.557	0.89 (0.38–2.11)	0.795
KIR2DS3	0.58 (0.34–2.14)	0.736	1.48 (0.55–3.97)	0.436
KIR2DS5	0.60 (0.32–1.14)	0.118	1.12 (0.56–2.26)	0.744
KIR cDNA type
KIR2DL5all	0.70 (0.39–1.27)	0.244	0.90 (0.45–1.80)	0.760
KIR2DS1	0.72 (0.39–1.34)	0.300	1.43 (0.62–3.27)	0.399
KIR2DS4del	1.18 (0.64–2.18)	0.588	0.51 (0.26–1.01)	0.049
KIR2DS4full	1.22 (0.67–2.24)	0.516	1.55 (0.77–3.11)	0.219
KIR3DL1	1.40 (0.80–2.46)	0.231	1.16 (0.58–2.31)	0.669
KIR3DS1	0.59 (0.33–1.06)	0.075	0.96 (0.48–1.91)	0.908
KIR2DS4full+/KIR2DS4del−	1.19 (0.68–2.09)	0.540	1.84 (0.93–3.66)	0.082
KIR HLA ligand
HLA-C2	1.21 (0.67–2.18)	0.523	1.23 (0.59–2.57)	0.574
HLA-Bw4T80	0.69 (0.39–1.22)	0.199	0.54 (0.28–1.05)	0.066
HLA-Bw4I80	0.70 (0.39–1.25)	0.225	0.93 (0.43–1.99)	0.851
HLA-ABw4	1.31 (0.73–2.35)	0.364	1.72 (0.81–3.67)	0.156
HLA-C genotype
C1C1	1.31 (0.69–2.47)	0.614	1.33 (0.60–2.95)	0.746
C1C2	0.78 (0.24–2.57)		0.86 (0.20–3.62)
C2C2	0.60 (0.17–2.12)		0.64 (0.14–3.04)

[i] Bold print, P-values <0.1. mNSCLC, metastatic non-smallcell lung cancer; OS, overall survival; PFS, progression-free survival; HR, hazard ratio; CI, confidence interval; KIR, killer-cell immunoglobulin-like receptor; HLA, human leukocyte antigen. Frequencies of >20% and <80% for KIRs or HLA ligands were included in the survival analysis.

Table III. Patient characteristics and their value in predicting survival by univariate analysis.

Table III.

Patient characteristics and their value in predicting survival by univariate analysis.

			OS		PFS
Variables	Subgroups	No. (%)	Median	P-value	Median	P-value
Age, years	≤60	41 (58.6)	19.3	0.583	10.5	0.178
	>60	29 (41.4)	17.5		4.4
Gender	Female	29 (41.4)	NR	<0.001	12.1	0.006
	Male	41 (58.6)	11.9		4.5
Smoking status	Never smoker	32 (45.7)	NR	0.001	10.5	0.152
	Ever smoker	38 (54.3)	11.9		4.4
ECOG PS	0	13 (18.8)	23.2	0.073	6.5	<0.001
	1	49 (71.0)	19.3		10.5
	2/3	8 (11.4)	4.5		2.2
EGFRa	Mutation	21 (36.2)	NR	<0.001	11.6	0.082
	Wildtype	37 (63.8)	15.6		6.3
Histologyb	AD	58 (86.6)	NR	<0.001	10.3	0.002
	SCC	9 (13.4)	10.0		2.8
Metastatic site	Brain/liver	13 (18.6)	14.4	0.266	7.3	0.622
	Others	57 (81.4)	21.4		8.1
CRP (mg/l)d	≤10	45 (67.2)	NR	0.001	10.3	0.140
	>10	22 (32.8)	11.6		3.5

a Data were unavailable for 12 patients.

b-d Data were unavailable for 3 patients. PFS, progression-free survival; OS, overall survival; CI, confidence interval; ECOG PS, Eastern Cooperative Oncology Group performance status; EGFR, epidermal growth factor receptor gene; AD, adenocarcinoma; SCC, squamous cell carcinoma; S-1, fluorouracil derivative; CRP, C-reactive protein; NLR, neutrophil/lymphocyte ratio; PLR, platelet/lymphocyte ratio; NR, not reached.

Results

Concordance between KIR typing using DNA and mRNA in NSCLC patients

Raw frequency data of all tested KIRs at the DNA and mRNA level, as well as HLA ligands, were calculated (data not shown). The framework genes 3DL3, 3DP1, 2DL4 and 3DL2 were positive in all samples at the DNA level, while the framework genes 3DL3 (0%) and 3DP1 (4.3%) were absent in nearly all individuals at the mRNA level. The frequencies of 2DL1, 2DL2, 2DL3, 3DS1, 2DS4full and 2DP1 at the DNA level were similar to those at the mRNA level, whereas the frequencies of other genes at the DNA level were significantly higher compared with those at the mRNA level. Furthermore, the frequencies of the full-length and deleted versions of KIR2DS4 were analyzed and the difference between the DNA and mRNA levels was compared (Table I). The majority of the 70 samples [62 at the mRNA (88.6%) and 65 (92.9%) at the DNA level] from the Chinese Han population were positive for KIR2DS4. The KIR2DS4del variant was found in 47.1% of the 70 individuals at the DNA level, similar to what was previously reported (13). However, at the mRNA level, the presence of the KIR2DS4del variant decreased to 27.1% in this population. The ratio of deleted:full-length form of KIR2DS4 was 1:2.3 at the DNA level and 1:3.9 at the mRNA level (Table I).

Table I. Frequency of KIR2DS4 alleles at the mRNA and DNA level.

Table I.

Frequency of KIR2DS4 alleles at the mRNA and DNA level.

Type	2DS4del	2DS4full	2DS4full-positive/2DS4del-positive	2DS4full-positive/2DS4del-negative	2DS4full-positive/2DS4del-positive	2DS4negative	2DS4dela/2DS4fullb
mRNA (n=70)	27.1% (19)	72.9% (51)	15.7% (11)	61.4% (43)	11.4% (8)	11.4% (8)	1:3.9
DNA (n=70)	47.1% (33)	72.9% (51)	20.0% (14)	45.7% (32)	25.7% (18)	7.1% (5)	1:2.3

a 2DS4full-negative/2DS4del-positive.

b 2DS4full-positive/2DS4del-negative. KIR, killer-cell immunoglobulin-like receptor.

Effect of individual KIR and HLA class I ligands on the prognosis of mNSCLC

The vast majority of the patients (95.7%) were positive for HLA-C1, a frequency which is similar to that of the Japanese population (14). Among the 15 types of KIRs and the 6 types of HLA ligands, with the exception that positive HLA-Bw4I80 expression was associated with a poor performance status, there were no other significant associations between KIRs/HLA ligands and patient characteristics (data not shown). Patients without KIR2DS4del gene expression (Fig. 1A) exhibited a significantly better OS [hazard ratio (HR)=0.51, P=0.049]. The median OS was not reached in patients without KIR2DS4del gene expression vs. 14.5 months in patients with KIR2DS4del gene expression. As regards the HLA genotype, only HLA-Bw4T80 was found to be associated with decreased OS at the 10% level (HR=0.51, P=0.066) (Fig. 1B). The median OS was 12.5 months in the HLA-Bw4T80-positive group vs. 21.4 months in the HLA-Bw4T80-positive group. No other KIRs or HLA ligands were significantly associated with clinical outcome according to the univariate analysis (Table II).

Figure 1. Kaplan-Meier overall survival curves for mNSCLC patients. (A) Kaplan-Meier curves of the association between KIR2DS4del and overall survival. (B) Kaplan-Meier curves of the association between HLA-Bw4T80 and overall survival. P-value was calculated by the log-rank test. mNSCLC, metastatic non-small-cell lung cancer. KIR, killer-cell immunoglobulin-like receptor; HLA, human leukocyte antigen.

Multivariate analysis considering patient characteristics

The effect of known clinicopathological confounding factors [age, gender, smoking status, performance status, EGFR mutation status, histology, metastatic site and C-reactive protein (CRP) level] on OS was evaluated by univariate analysis (Table III). The results revealed that male gender, adenocarcinoma and a performance status of 2 were significantly associated with disease progression. Male gender, adenocarcinoma, smoking, wild-type EGFR status and CRP level were factors that significantly affected OS. In the stepwise multivariate Cox model, which includes significant clinical factors and KIR/HLA ligands at the 10% level, positive KIR2DS4del gene expression (KIR2DS4del mRNA) (adjusted P=0.014, HR=3.33), positive HLA-Bw4T80 (adjusted P=0.012, HR=3.58), smoking (adjusted P=0.02, HR=2.87) and a CRP level >10 mg/l (adjusted P=0.004, HR=6.48) remained independent predictors of decreased OS; no KIR/HLA type or clinicopathological variables were selected as independent predictors of PFS following multitest correction (Table IV).

Table IV. Multivariate Cox analysis of PFS and OS.

Table IV.

Multivariate Cox analysis of PFS and OS.

		Multivariate analysis
Variables	Groups	HR (95% CI)	P-value	Adjusted P-value
OS (n=64)
CRP, mg/l	>10 vs. ≤10	6.48 (2.11–19.91)	0.001	0.004
Smoking	Yes vs. no	2.87 (1.18–6.96)	0.020	0.020
HLA-Bw4T8	Positive vs. negative	3.58 (1.51–8.47)	0.004	0.012
KIR2DS4del mRNA	Positive vs. negative	3.33 (1.39–7.94)	0.007	0.014
PFS (n=64)
CRP, mg/l	>10 vs. ≤10	2.94 (1.36–6.37)	0.006	0.054
Histology	SCC vs. AD	2.38 (0.95–5.95)	0.065	0.195
Gender	Male vs. female	1.71 (0.84–3.49)	0.138	0.204
ECOG PS	2 vs. 0	4.20 (1.10–16.09)	0.036	0.160
HLA-Bw4T8	Positive vs. negative	2.41 (1.21–4.83)	0.013	0.091
KIR2DS1 DNA	Positive vs. negative	17.08 (1.28–227.54)	0.032	0.160
KIR2DL5all DNA	Positive vs. negative	34.14 (1.52–765.94)	0.026	0.156
KIR2DL5A DNA	Positive vs. negative	0.047 (0.005–0.461)	0.047	0.072
KIR2DL5B DNA	Positive vs. negative	0.063 (0.002–1.737)	0.063	0.204

[i] Bold print, P-values indicating statistically significant differences. OS, overall survival; PFS, progression-free survival; HR, hazard ratio; CI, confidence interval-; CRP, C-reactive protein; SCC, squamous cell carcinoma; AD, adenocarcinoma; ECOG PS, Eastern Cooperative Oncology Group performance status; KIR, killer-cell immunoglobulin-like receptor; HLA, human leukocyte antigen.

Discussion

Several studies have investigated the association between KIR/HLA genotypes and autoimmune disease, infectious disease, transplantation, stem cell disease and malignancy (15–17). Regarding malignant diseases, certain KIR/HLA types have been associated with the susceptibility or worse clinical outcome in several types of cancer, such as leukemia, colorectal cancer and lung cancer (18–20).

In lung cancer, Wisniewski et al reported a possible protective effect of the HLA-C C1C2 genotype on susceptibility to NSCLC and its association with the KIR2DL2/KIR2DS2/C1C1 genotype, which was correlated with better response to therapy and longer survival in the Caucasian population (20). Due to the ethnic differences and small sample size, we did not observe these consistent findings in our samples. In the present study, both cDNA and genomic DNA were typed, in contrast to other reports where genomic DNA was used, considering that only expressed genes may play a biological role and non-expressed KIR gene(s) may exert an effect on a different, causative gene, due to linkage disequilibrium. We observed that the majority of KIR genes had a different typing profile at the DNA and mRNA levels. Furthermore, only KIR cDNA exerted a significant effect on the outcome of mNSCLC in this study population following multitest correction. We observed that KIR2DS4del expression was associated with a significantly decreased survival time, and KIR2DS4full-positive/KIR2DS4del-negative genotype was significantly associated with long-term survival according to the univariate analysis. However, following multivariate Cox regression analysis and multitest correction, only KIR2DS4del mRNA expression was considered as an independent factor associated with OS.

The activating KIR2DS4 receptor has been reported to bind to group 1 and 2 HLA-C, HLA-A11 and non-HLA class I ligands (21,22). KIR2DS4del is the mutant form of KIR2DS4 with a 22-bp deletion in exon 5, resulting in a frameshift mutation. This truncated KIR2DS4 protein would be secreted due to the loss of the transmembrane/cytoplasmic domains. Individuals who are homozygous for haplotype A and with a deletion in both alleles are likely to have a dysfunctional KIR receptor activator, considering that a proper allele of KIR2DL4 may function as an activating receptor. KIR2DS4del frequencies vary among different populations. In the Caucasian population, the ratio of full-length:deleted KIR2DS4 is ~1:2. However, in the Chinese, Japanese and Korean populations, the ratio of KIR2DS4full to KIR2DS4del is reversed (13), which is similar to what was observed in the present study.

The associations and functions of KIR2DS4 in infectious diseases have been extensively investigated. KIR2DS4full is associated with increased viral loads and transmission rates of HIV-1 (23,24). Additionally, KIR2DS4full promotes HIV-1 pathogenesis during chronic infection, probably through the maintenance of an excessive pro-inflammatory state (25). The combination of KIR2DS4 and KIR2DS4del was associated with disease progression from hepatitis to hepatocellular carcinoma (HCC) development via cirrhosis (26). In terms of malignant diseases, Giebel et al (27) reported that the absence of KIR2DS4full was associated with susceptibility to chronic myeloid leukemia, but the effect of KIR2DS4full appeared to be the opposite in endometrioid ovarian cancer (28). Although the associations of all known 15 human KIRs and their known HLA ligands with the prognosis of mNSCLC were evaluated in the present study, we only observed that the KIR2DS4full-positive/KIR2DS4del-negative type combined with its potential ligand HLA-C1 was significantly associated with improved OS in the univariate analysis (data not shown). Almost all patients expressed HLA-C1 (95.7% positive) in our samples; therefore, it was hypothesized that the improving effect of KIR2DS4full on OS may be associated with potentiating NK cell-mediated cytotoxicity and tumor surveillance through the interaction between HLA-C1 and activating KIR2DS4full.

KIR3DL1 will only inhibit NK cytotoxicity against target cells expressing a discrete subset of HLA-B and HLA-A allotypes. In consideration of the clinicopathological variables, we observed that HLA-Bw4T80 was the only HLA type significantly affecting OS in multivariate analyses following multitest correction. This observation is consistent with a previous report (29), which suggested an association of HLA-Bw4T80 with decreased OS in patients with HCC. HLA-Bw4T80 is an allotype of the HLA-Bw4 motif with a threonine at position 80, which exhibits lower affinity for KIR3DL1 compared with the HLA-Bw4 motif with an isoleucine at the same position (30). It is possible that HLA-Bw4T80 binds with KIR3DL1 to deliver inhibitory signals to NK cells, which would then enhance the ability of tumor cells to escape immune detection (31). However, a positive association of the combination of KIR3DL1/HLA-Bw4T80 with survival was not observed (data not shown).

Notably, the KIR/KLA system exhibits extensive genetic diversity. It would be interesting to perform a similar analysis of an association of KIR2DS4del mRNA and HLA-Bw4T80 with clinical outcome in other ethnic populations. To the best of our knowledge, this is the first study addressing the relevance of KIRs and HLA ligands regarding the clinical outcome of mNSCLC in a Chinese population. If confirmed in a larger cohort of patients in independent studies, KIR2DS4del mRNA and HLA-Bw4T80 may be prognostic indicators for treatment stratification and patient management. Furthermore, functional studies will be required to determine the role of KIR2DS4del mRNA and HLA-Bw4T80 in the immune response against mNSCLC.

Acknowledgements

The present study was supported by Transgene SA, France and by grants from the Wu Jieping Medical Foundation (no. 320675014278) and the Shanghai Municipal Commission of Health and Family Planning (no. 201440423).

Glossary

Abbreviations

Abbreviations:

References