Changes in serum amino acid levels in non-small cell lung cancer: a case-control study in Chinese population

Introduction

Lung cancer is one of the most common malignant tumors which threaten human health and life. According to global cancer statistics in 2020, lung cancer is the second most common cancer in terms of incidence, and the first leading cause of death globally, accounting for approximately 18% of global cancer deaths (Sung et al. , 2021). Non-small cell lung cancer (NSCLC) occupies over 85% of all the lung cancers, and the most common sub-types are adenocarcinoma and squamous cell carcinoma (SCC) among them (Goldstraw et al. , 2011). The 5-year survival rate for localized lung cancer is 57%, while for metastatic lung cancer it is only 5% (Siegel, Miller & Jemal, 2020). In China, approximately 733,300 lung cancer patients occurred in 2015 according to the latest data of the national cancer registries (Chen et al., 2016). It is no doubt that lung cancer is still a serious public health problem in both China and the global.

As basic metabolites and regulatory factors, amino acids (AAs) play extensive effects on cancer cells (Wei et al., 2021). For instance, AAs account for the majority of the carbon-based biomass production, and are the dominant nitrogen source for hexosamines, nucleotides, and other nitrogenous compounds in cancer cells (Hosios et al., 2016; Ma & Vosseller, 2013; Meng et al., 2010; Zetterberg & Engstrom, 1981). AAs can also regulate gene expression and the protein phosphorylation cascade. For example, adequate essential amino acids (EAAs) stimulated MYC oncogene mRNA translation by inhibiting the GCN2-eIF2A-ATF4 stress pathway, leading to MYC-dependent transcriptional amplification (Yue et al., 2017). Some AAs such as glutamine (Gln), arginine (Arg) and leucine (Leu) stimulated the cell-specific phosphorylation of mTOR1, thereby regulating the conversion of intracellular proteins (Wu, 2009). Meanwhile, Gln, citrulline (Cit) and methionine (Met) could also activate the mTOR1 signaling pathway through modulating the concentration of phosphatidic acids (PAs) and activity of Ras homologue enriched in brain (Rheb) in the late lysosome/endosome system (LEL) (Bourgoin-Voillard et al., 2016). On the other hand, the metabolism of AAs is strictly regulated and closely related to other metabolic networks, such as lipid and glucose metabolism.

Although the concentrations of circulating AAs remain constant in a healthy state, numerous studies have found that various cancers, such as colorectal, lung and breast cancers, can influence the levels of AAs (Miyagi et al., 2011). In recent years, the AminoIndex has aroused wide attention, which facilitates the evaluation of certain health conditions and the possibility of disease by establishing multivariate indices from blood free AA profile information (Noguchi et al., 2006). Using these indices, researchers have discriminated various cancer types from healthy controls, including lung, colorectal, breast, gastric, prostate, gynecological and pancreatic cancers (Miyagi et al., 2011; Mikami et al., 2019). For lung cancer, AminoIndex has also been utilized for monitoring cancer treatment (Higashiyama et al., 2020). These evidences suggested that AAs are suitable candidates for the study of NSCLC metabolomics, and it might be of potential value in clinical application.

At present, several studies have reported the alterations of blood free AA profile in lung cancer patients from Japan, Korea and Europe countries (Maeda et al., 2010; Shingyoji et al., 2013; Kim et al., 2015; Klupczynska et al., 2016). However, there is still little data regarding AA profile in NSCLC patients in China. Hence, a case-control study with 200 NSCLC patients and 202 healthy controls was conducted to characterize the feature of AA profile in Chinese NSCLC patients in this study. A total of 22 AAs, including eight common EAAs and 14 common non-essential amino acids (NEAAs), were determined by a liquid chromatography-tandem mass spectrometer (LC-MS/MS) method (Choi & Coloff, 2019; Sivashanmugam et al., 2017; Papadia et al., 2018). The alteration of AA profile in NSCLC was identified and the related metabolic pathways were discussed. We also evaluated the performance of the AA profile for the detection of NSCLC across different histological types and stages in Chinese patients.

Materials & Methods

Results

AA concentrations in patients with different histological types and clinical stages

The results of subgroup analysis based on histological type are shown in Table 3. After multiple testing using FDR method, the concentrations of Trp, Gly, Ser, Asp, Asn, Cys, Glu, Gln, Orn, and Cit were significantly altered in serum of adenocarcinoma patients when comparing with healthy controls (all P-FDR < 0.05). Among these, three AAs including Asp, Cys and Orn were substantially up-regulated (FC ≥ 1.2) in serum of adenocarcinoma patients. And the concentrations of Phe, Asp, Cys, and Glu were significantly altered in serum of SCC patients when comparing with healthy controls (all P-FDR < 0.05). Among these, three AAs including Asp, Cys and Glu were substantially up-regulated (FC ≥1.2) in serum of SCC patients. Besides, the concentration of Asp was significantly altered and substantially down-regulated in serum of adenocarcinoma patients when comparing with SCC patients (P-FDR < 0.05, FC = 0.69).

Table 3:

Subgroup analysis of amino acid concentrations in patients based on histological types.

Amino acid	Adenocarcinoma N = 133					SCC N = 57					Fold changee	P-valuee	P-FDRe
	Median	IQR	Fold changec	P-valuec	P-FDRc	Median	IQR	Fold changed	P-valued	P-FDRd
Phenylalaninea	99.60	34.80	1.03	0.212	0.301	114.00	59.70	1.18	<0.001	<0.001	0.87	0.008	0.088
Tryptophana	51.70	25.00	0.90	0.006	0.019	48.10	25.80	0.83	0.087	0.216	1.07	0.925	0.969
Isoleucinea	86.50	35.00	1.06	0.049	0.098	87.00	52.20	1.06	0.113	0.216	0.99	0.789	0.882
Leucinea	170.00	50.00	1.06	0.180	0.292	175.00	108.00	1.09	0.038	0.120	0.97	0.245	0.797
Methioninea	28.90	11.00	0.97	0.856	0.856	26.70	13.10	0.90	0.460	0.633	1.08	0.316	0.797
Lysinea	234.00	106.00	1.05	0.219	0.301	210.00	158.50	0.95	0.879	0.938	1.11	0.519	0.882
Valinea	281.00	95.50	1.02	0.285	0.369	295.00	149.00	1.07	0.250	0.394	0.95	0.759	0.882
Threoninea	154.00	68.00	1.04	0.065	0.119	143.00	86.50	0.97	0.553	0.715	1.08	0.103	0.567
Glycineb	378.00	162.50	1.16	0.001	0.006	349.00	224.50	1.07	0.113	0.216	1.08	0.582	0.882
Serineb	191.00	68.00	1.12	0.007	0.019	190.00	88.50	1.11	0.023	0.100	1.01	0.719	0.882
Asparagineb	63.90	24.40	1.11	0.005	0.018	58.80	35.00	1.02	0.115	0.216	1.09	0.802	0.882
Aspartic acidb	39.90	26.70	1.23	<0.001	<0.001	57.50	32.90	1.77	<0.001	<0.001	0.69	<0.001	<0.001
Cysteineb	67.40	47.80	1.25	<0.001	0.009	76.10	47.80	1.42	0.001	0.006	0.89	0.280	0.797
Glutamic acidb	122.00	77.50	1.17	<0.001	<0.001	135.00	93.90	1.29	<0.001	<0.001	0.90	0.689	0.882
Glutamineb	528.00	213.50	1.05	0.011	0.027	488.00	287.50	0.97	0.733	0.896	1.08	0.042	0.308
Histidineb	65.90	31.60	1.10	0.186	0.292	57.80	45.60	0.96	0.881	0.938	1.14	0.311	0.797
Prolineb	217.00	97.00	1.02	0.493	0.542	223.00	124.00	1.04	0.330	0.483	0.97	0.675	0.882
Arginineb	111.00	72.80	1.03	0.492	0.542	114.00	85.70	1.06	0.118	0.216	0.97	0.326	0.797
Tyrosineb	75.80	49.70	1.04	0.612	0.641	76.00	57.90	1.05	0.896	0.938	1.00	0.635	0.882
Alanineb	477.00	201.50	1.01	0.411	0.502	433.00	299.00	0.91	0.954	0.954	1.10	0.593	0.882
Ornithineb	116.00	76.80	1.25	<0.001	<0.001	110.00	90.20	1.19	0.031	0.112	1.05	0.611	0.882
Citrullineb	30.20	15.60	1.13	0.017	0.037	32.00	27.30	1.20	0.186	0.315	0.94	0.991	0.991

DOI: 10.7717/peerj.13272/table-3

Notes:

aEssential amino acids.

bNot essential amino acids.

cCompared between adenocarcinoma and control.

dCompared between squamous cell carcinoma and control.

eCompared between adenocarcinoma and squamous cell carcinoma.

Mann–Whitney U test was used for continuous variables that did not normally distributed.

Benjamini–Hochberg procedure was used for multiple testing correction.

AA levels in patients with different clinical stages are shown in Fig. 1. After multiple testing using FDR method, the concentration of Glu was significantly altered in serum of early-stage NSCLC patients (P- FDR = 0.044), whilst the concentrations of Phe, Trp, Ile, Asp, Cys, Glu, Orn, Gly, Ser, Asn and Arg were significantly altered in serum of late-stage NSCLC patients when comparing with healthy controls (all P-FDR < 0.05). Four AAs including Asp, Cys, Glu and Orn were substantially up-regulated (FC ≥ 1.2) in serum of late-stage NSCLC patients. Although there was no significant difference in AA levels between the early-stage and late-stage NSCLC, several AA levels showed consecutive changes with the disease progression. For example, the Trp concentration gradually decreased in the control group, early-stage and late-stage NSCLC patients successively, while concentrations of AAs such as Ile, Gly, Ser, Asn, Asp, Cys, Glu and Orn showed a gradually increased trend.

Metabolic pathway analysis of AAs in NSCLC

Metabolic pathway analysis was performed based on the altered 11 AAs between the NSCLC group and the control group. As illustrated in Fig. 2 and Table S8, we identified three significant pathways with important impact (impact factor > 0.4, P- FDR < 0.05), including Gly, Ser and Thr metabolism which hits three AAs including Ser, Gly, Cys (P- FDR = 0.016, impact factor = 0.463), Ala, Asp and Glu metabolism which hits three AAs including Glu, Asp, Asn (P- FDR = 0.017, impact factor=0.421), and Arg biosynthesis which hits four AAs including Glu, Asp, Cit, Orn (P- FDR ≤0.001, impact factor = 0.406).

The classification of NSCLC patients and control group

The AA panel for NSCLC classification was composed of the 11 AAs which of the levels were altered in NSCLC patients compared with controls, including Phe, Trp, Ile, Gly, Ser, Asn, Asp, Cys, Glu, Orn, and Cit, and the model was generated using logistic regression. As a result, the area under the curve (AUC) of the ROC was 0.80 (95% CI [0.74–0.85]) in the training data set (Fig. 3A). In the validation data set, the AUC of the model also reached 0.79 (95% CI [0.71–0.87]) (Fig. 3A). These results demonstrated that the AA panel may be of potential value in NSCLC classification and detection.

To further investigate the performance of the model in discriminating NSCLC patients across different histological types and clinical stages, the AUCs for each histological type and stage were also estimated. For early-stage NSCLC, the AUC was 0.75 (95% CI [0.68–0.81]), while for late-stage NSCLC, the AUC reached 0.86 (95%CI [0.82–0.91]) (Fig. 3B). Moreover, the model showed large discrepancy in the detection of adenocarcinoma and SCC, with the AUC of 0.77 (95% CI [0.71–0.82]) for adenocarcinoma, while as high as 0.90 (95%CI [0.85–0.95]) for SCC (Fig. 3B).

Discussion

Many previous reports have revealed that the metabolism was perturbed in cancer patients, including the metabolism of AAs (Li & Zhang, 2016; Vander Heiden, Cantley & Thompson, 2009). As basic building blocks of proteins, AAs play crucial roles in maintaining normal life activities and meeting the nutritional and energy requirements of tumor cells as well (Vettore, Westbrook & Tennant, 2020; Tabe, Lorenzi & Konopleva, 2019). Previous studies showed that there were substantial changes in AA profile in lung cancer, and an index derived from the altered AAs, the AminoIndex, has been utilized for lung cancer screening and cancer treatment monitoring (Miyagi et al., 2011; Mikami et al., 2019; Higashiyama et al., 2020; Maeda et al., 2010; Kim et al., 2015; Klupczynska et al., 2016; Proenza et al., 2003). At present, the common methods for determining the content of AAs in fluid samples include capillary electrophoresis (CE), ion-exchange chromatography (IEX), high performance liquid chromatography (HPLC), and LC-MS/MS, etc (Smith, 1999; Smon et al., 2019; Mo et al., 2013; Phipps et al., 2020). Compared with other methods, the LC-MS/MS method which adopted in our study has the advantages of high throughput, good selectivity, time-saving and cost-effective, thus has been widely used in clinical assessment of AAs. Based on the application of amino derivative markers, this method has greatly improved the accuracy, precision, and detection range of AAs determination (Le et al., 2014). In this study, we found that the levels of 11 out of 22 AAs were significantly altered between the NSCLC and the healthy people in Chinese population. Three important metabolic pathways were involved in the metabolism of these AAs, including glycine, serine and threonine metabolism; alanine, aspartate and glutamate metabolism; and arginine biosynthesis. In addition, this AA panel were of potential value in sample classification of NSCLC.

The current study identified 11 altered AAs in serum of NSCLC, included Phe, Trp, Ile, Gly, Ser, Asp, Asn, Cys, Glu, Orn, and Cit. In previous studies, circulating Phe, Ile, Ser, Asp and Orn were found to be elevated while Trp showed a downward trend in lung cancer, which coincided with our results (Miyagi et al., 2011; Maeda et al., 2010; Shingyoji et al., 2013; Kim et al., 2015; Klupczynska et al., 2016; Proenza et al., 2003; Zhao et al., 2014). However, some inconsistent results were also found in several studies. For instance, Zhao’s team and Kim’s group (Kim et al., 2015) reported down-regulated Gly concentration, while Maeda’s team, Miyagi’s group (Miyagi et al., 2011) and our study observed that Gly was up-regulated in NSCLC patients. Remarkably, abnormal levels of these AAs have been found in different populations as well (Maeda et al., 2010; Kim et al., 2015; Klupczynska et al., 2016; Proenza et al., 2003). The reasons for the inconsistent changes in AA levels may be related to the diversity of genetic background, regional differences, nutritional status, and lifestyle factors.

We found three important metabolic pathways which were involved in the metabolism of the altered AAs. Among these, Glu, Asp and Asn were involved in alanine, aspartate and glutamate metabolism. Glu could be converted to α-ketoglutarate (α-KG) and further entered the tricarboxylic acid cycle (TCA), which is the main way for tumor cells to meet the demands of bioenergy, biosynthesis and redox balance in tumor microenvironment (Todisco et al., 2019). Meanwhile, α-KG can produce Asp through a series of TCA catalytic reactions in which Asn could also be produced by the catalysis of Asn synthase (ASNS). Therefore, Asp and Asn are also involved in the energy metabolism of the tumor (Wei et al., 2021). Likewise, Gly and its precursor Ser participate in TCA through serving as exchange factor of Asn. Notably, ASNS is frequently upregulated in tumors and is associated with poor prognosis of several types of cancer, such as prostate cancer, glioma and neuroblastoma (Sircar et al., 2012; Zhang et al., 2014). On the other hand, L-asparaginase, an enzyme catalyzing the degradation of Asn, has been widely applied in clinical treatment against multiple types of leukemia and lymphoma (Haskell & Canellos, 1969). These evidences suggested a strong dependence of cancer cells for Glu metabolism and energy from TCA.

We noticed that Asp, as well as Orn and Cit, also participated in the Arg biosynthetic pathway. This pathway is a part of the urea cycle, in which intermediates including Orn, Cit, Asp, arginine succinate and Arg are produced by the successive catalysis of carbamyl phosphate synthase 1, arginine succinate synthase 1 and arginine succinate lyase (Wei et al., 2021). Although the complete urea cycle works primarily in the liver, some enzymes could relink the intermediates outside the liver. These substances support the survival and proliferation of tumor cells through strong interactions with other metabolic pathways (Lee et al., 2018; Keshet et al., 2018). Arg is the most consumed AA in the inner necrotic core of tumor mass, indicating its high demand for the survival of tumor cells. Preclinical studies have reported essential roles of Arg in cancer metabolism. For example, Arg starvation caused global transcriptional suppression of metabolic genes including those involved in oxidative phosphorylation, mitochondrial functions, and DNA repair in breast and prostate cancer cells (Qiu et al., 2014; Changou et al., 2014). Meanwhile, Arg can import into cells, directly activate mTOR pathway, a nutrient-sensing kinase strongly implicated in carcinogenesis. By mTOR mediation, Arg thus has profound impacts on protein synthesis, lipid synthesis and nucleotide synthesis of cancer cells (Ban et al., 2004; Saxton & Sabatini, 2017). In addition, Arg can also bind G-protein coupled receptor to activate down-stream RAS/ERK or PI3K pathway, another essential pathway in cancer development and progression (Chen et al., 2021). Therefore, as the building blocks of Arg biosynthetic pathway, Asp, Orn and Cit might be signaling metabolites and act as transcriptome reprogrammer in cancer metabolism. Besides, as mentioned above, CIT could activate the mTOR1 signaling pathway through modulating the concentration of PA and Rheb activity in the LEL, which also indicating the role of these AAs in cancer signaling transduction (Bourgoin-Voillard et al., 2016).

In addition, numerous tumors rely on the availability of extra-cellular Ser to proliferate rapidly (Jones & Schulze, 2012). In the conversion process of Ser to Gly, a single-carbon unit could be generated to assist the production of 5, 10-methylene tetrahydrofolate (CH2-THF), while the Gly can also be cleaved by the mitochondrial Gly cleavage system to produce CH2-THF. CH2-THF could enter the folate cycle for the synthesis of other substances that are involved in cancer metabolism, such as other AAs, nucleotides and lipids (Tibbetts & Appling, 2010; Locasale, 2013; Amelio et al., 2014). Moreover, Ser metabolic pathway probably contribute to NSCLC by serving as a carbon source of nucleotide synthesis and DNA methylation (Pavlova & Thompson, 2016). Interestingly, in addition to the de novo-produced Ser, the exogenous donors perhaps affect one-carbon and tumor behavior as well. For instance, a study has found that dietary Ser and Gly deficiency can inhibit the growth of colon cancer cells in mice (Maddocks et al., 2013). Besides, Cys might be involved in cancer metabolism through regulating ferroptosis (Poursaitidis et al., 2017). Thus, Ser metabolic pathway, as well as Gly and Cys metabolic pathways, may play essential roles in NSCLC metabolism.

The established AA panel in our study was effective in discriminating NSCLC patients from controls, as well as classification of NSCLC patients with different stages and histological types. In the study of Kim et al. (2015) and Shingyoji et al. (2013), six AAs (Ser, Gln, Ala, His, Orn, and Lys) were used to establish an index for distinguishing lung cancer cases from healthy controls. Besides the Ser and Orn, we included some other AAs as well, such as Phe, Trp, Ile, Asp, Asn, etc. Although we included more AAs, the AUCs were similar to the results of previous studies. This may be related to sample size, sample composition of clinical stages and pathological types, nutritional status, etc. Notably, a Polish study applied both the Korean and self-built models, and found that the latter has better distinguishing ability (Aal, His, Phe, Cit, Asp, Asn) (Klupczynska et al., 2016). Taken together, AAs may have potential value for the classification of NSCLC. However, it may be necessary to establish population-specific AA panels in different countries and regions.

To our knowledge, this is the first case-control study in Chinese population to investigate the alteration of AA levels as well as the involved AA pathways in NSCLC patients, and to evaluate the potential value of AAs in the classification of NSCLC. The enrolled cases and controls were comparable in the basic characteristics. And the approach for determination of AA levels were validated with widely approved standards. However, several limitations should be addressed. Firstly, there are many types of AAs and their derivatives, however, only 22 common ones were selected in our study. More kinds of AAs thus need to be determined and compared in the future. Secondly, AA levels in the human body are affected by a variety of factors including nutritional status, lifestyle factors such as diet, physical activity and other underlying diseases. Although we have matched the cases and controls with gender and age, our results might be still influenced by other potential confounding bias. Therefore, our results still need to be verified in future studies.

Conclusions

In conclusion, the current study found that the level of Trp, Phe, Ile, Gly, Ser, Asp, Asn, Cys, Glu, Orn, and Cit were significantly altered in serum of NSCLC patients in comparison with controls. The altered AAs were involved in three important metabolic pathways, including Gly, Ser and Thr metabolism, Ala, Asp and Glu metabolism, and Arg biosynthesis. In addition, this AA panel might pose potential in NSCLC classification.

Supplemental Information