Preoperative AICS (lung) abnormalities predict postoperative recurrence in patients undergoing curative resection for non-small cell lung cancer

[Background] AminoIndex TM Cancer Screening (AICS (lung)) was developed as a screening test for lung cancer using a multivariate analysis of plasma-free amino acid (PFAA) proles. According to the developed index composed of PFAA, the probability of lung cancer was categorized into AICS (lung) ranks A, B, and C in order of increasing risk. The aim of the present study was to investigate the relationship between the preoperative AICS (lung) rank and surgical outcomes in patients who underwent curative resection for non-small cell lung cancer (NSCLC). [Methods] Preoperative blood samples were collected from 297 patients who underwent curative resection for NSCLC between 2006 and 2015. PFAA concentrations were measured. The relationship between the preoperative AICS (lung) rank and clinicopathological factors was examined. The effects of the preoperative AICS (lung) rank on postoperative outcomes were also analyzed. [Results] The AICS (lung) rank was A in 93 patients (31.3%), B in 82 (27.6%), and C in 122 (41.1%). The AICS (lung) rank did not correlate with any clinicopathological factors, except for age. Based on follow-up data (median follow-up period of 6 years), postoperative recurrence was observed in 22 rank A patients (23.7%), 15 rank B (18.3%) and 49 rank C (40.2%). In the univariate analysis, preoperative AICS (lung) rank C was a worse factor of recurrence-free survival (p=0.0002). The multivariate analysis identied preoperative AICS (lung) rank C (HR: 2.17, p=0.0005) as a signicant predictor of postoperative recurrence, particularly in patients with early-stage disease or adenocarcinoma. in NSCLC

preoperative AICS (lung) rank C (HR: 2.17, p=0.0005) as a signi cant predictor of postoperative recurrence, particularly in patients with early-stage disease or adenocarcinoma.
[Conclusion] Preoperative AICS (lung) abnormalities are a high-risk predictor of postoperative recurrence in NSCLC patients undergoing curative resection.

Background
Metabolite pro ling remains a relatively underrepresented eld of biomarker development for identifying and characterizing various types of cancer. Metabolic changes in patients with cancer may re ect differences in the concentration of a single metabolite or alterations in the constituents of an entire metabolic pathway [1][2][3][4][5]. Metabolic changes in the blood of patients with cancer are closely associated with alterations in amino acid, protein, glucoside, and fatty acid metabolism. Therefore, metabolic pro les may be used to differentiate between patients with and without cancer [1,2,[4][5][6]. In particular, amino acids are one of the most suitable candidates for focused metabolomics because they are ingested or endogenously synthesized and play important physiological roles as essential metabolites and their regulators [1][2][3][4][5][6].
Based on these ndings, we developed a novel cancer screening test for the early detection of various types of cancers using AminoIndex™ technology, which scores health conditions and the probability of disease by a multivariate analysis with plasma-free amino acid (PFAA) concentrations included as a variable [1,2]. We demonstrated that the PFAA concentration ratio was altered in 7 types of cancers: gastric, lung, colorectal, prostatic, gynecological, breast, and pancreatic cancers [1,7]. In Japan, this technology, designated the AminoIndex™ Cancer Screening (AICS) test, is widely used in health and cancer screening [1,7]. Since the lung-speci c AICS (AICS (lung)) test, one of 7 cancer screening tests, identi es lung cancer based on a speci c PFAA pro le composed of signi cantly altered plasma concentrations of 6 representative amino acids (serine, glutamine, alanine, histidine, ornithine, and lysine) [1,8,9], it is now used to screen for lung cancer. The probability of lung cancer is noted as AICS (lung) values of 0.0-10.0, which are categorized as ranks A (0.0-4.9), B (5.0-7.9), and C (8.0-10.0). Since higher AICS (lung) values are associated with an increased risk of lung cancer, when judged as rank C in the screening, medical work-up for lung cancer is recommended as needed [1,[7][8][9].
However, the AICS (lung) test has not yet been su ciently examined in clinical practice for lung cancer. Recently, we reported that the majority of lung cancer patients with high preoperative AICS (lung) values exhibited a reduction in AICS rank after curative resection, but that the absence of postoperative reduction in AICS rank was strongly associated with tumor recurrence [10]. Therefore, the AICS (lung) rank is a marker that re ects the presence or absence of lung cancer. However, the relationship between the AICS rank and clinicopathological factors of cancer remains unclear, and, thus, its clari cation may lead to the AICS (lung) test becoming a screening test and predictor of recurrence.
To investigate the clinicopathological and prognostic signi cance of the AICS (lung) rank among patients with lung cancer on a large scale, the relationship between the preoperative AICS (lung) rank and surgical outcomes was retrospectively investigated in 297 patients who underwent curative resection for nonsmall cell lung cancer (NSCLC). To the best of our knowledge, this is the rst study to report the prognostic importance of the AICS rank in surgically resected patients with cancer.  Table 1. Their mean age was 65.3 years (range, 3086), and there were 187 males and 110 females. A total of 112 patients were non-smokers, while 185 were current or ex-smokers. Elevated preoperative serum CEA levels were observed in 72 patients. The surgical procedure selected was standard surgery (including lobectomy or more than lobectomy) for 224 patients and limited surgery (including segmentectomy or wedge resection) for 73, with all undergoing curative resection. The histological types were adenocarcinoma in 216 patients, squamous cell carcinoma in 61, and other types in 20. Regarding the pathological stage (pstage), 163, 58, 9, 20, 30, 16, and one case were classi ed as stage IA, IB, IIA, IIB, IIIA, IIIB, and IV, respectively. Stage IV disease was due to a postoperative status of preceding brain metastasis. These stage classi cations were based on the 7th Edition of the TNM classi cation [11].

Methods
Postoperative adjuvant therapy was administered to 61 patients: adjuvant chemotherapy to 58, adjuvant radiotherapy to 2, and adjuvant chemoradiation therapy to one. The regimen for adjuvant chemotherapy was as follows: oral UFT (Tegafur-Uracil) for 15 patients with stage I disease, for 2 with stage II disease, and for 1 with stage III disease. Gemcitabine (GEM) or vinorelbine (VNR) monochemotherapy was administered to one patient with stage I disease and 5 with stage II, respectively. Platinum (cisplatin or carboplatin)-based combined chemotherapy was administered to 4 patients with stage I disease, 6 with stage II disease, and 23 with stage III disease. Combined chemotherapy using GEM and VNR was administered to one patient with stage III disease. Most regimens were performed as described previously [12,13]. Adjuvant radiotherapy was administered to two patients because of the short margin in the surgical stump of the bronchus in one and of the mediastinal eld in the other. Adjuvant chemoradiation therapy was performed using the combined carboplatin-based regimen and 64 Gy on the surgical margin of the chest wall. No patients received preoperative therapy. Followup examinations were generally performed as follows: During the rst 36 months after surgery, systemic and local screening examinations were conducted using blood tests, chest computed tomography (CT) scans were routinely obtained every 6 months, and uoro2deoxyglucose positron emission tomography (FDGPET) scans were performed where necessary. Brain CT or magnetic resonance imaging scans were also performed as required. During the rst 2 years, these examinations were routinely performed and the ndings obtained were carefully assessed. Thereafter, intensive examinations were performed as needed. Therefore, the median postoperative follow-up time was 6.0 years (range, 0.1 to 11.5 years) in this cohort.
The time of initial recurrence was assessed based on the onset of clinical symptoms, the detection of blood test abnormalities (e.g., serum CEA levels), or the detection of recurrent lesions on imaging, and the recurrence-free survival (RFS) period was de ned as the period between surgery and the time of initial recurrence. When a tumor recurred, the initial recurrence site was evaluated and generally categorized into three patterns: local, distant, and multiple.
Measurement of PFAA concentrations. Patients underwent preoperative blood sampling within 1 week prior to surgery. Preoperative blood samples (5.0 mL) were collected from a forearm vein on the morning after overnight fasting using tubes containing ethylenediaminetetraacetic acid, and were immediately placed on ice. Plasma was prepared by centrifugation at 3,000 rpm at 4 °C for 15 minutes and stored at − 80 °C until analyzed. Plasma samples were deproteinized using acetonitrile at a nal concentration of 80.0%. PFAA concentrations were measured using high-performance liquid chromatography/electrospray ionization tandem mass spectrometry with pre-column derivatization, which have been described previously [8,10,14]. AICS (lung) values and ranks. AICS (lung) values for patients with NSCLC were classi ed into the following three groups: rank A, B, or C. The AICS (lung) test detects lung cancer from plasma concentrations of serine, glutamine, alanine, histidine, ornithine, and lysine. AICS (lung) values range between 0.0 and 10.0, with 5.0 producing 80.0% speci city and 8.0 producing 95.0% speci city. A high AICS (lung) value was associated with a greater probability of lung cancer. Rank A was considered to be normal (0.0-4.9), rank B was relatively high (5.0-7.9), and rank C was high (8.0-10.0). Rank C indicated a high risk of lung cancer [1,2,8,9]. Statistical analysis. RFS and overall survival (OS) curves were calculated using the Kaplan Meier method, and differences were examined by the Log-rank test. Cox's proportional hazards regression model was used to perform a multivariate analysis of factors associated with RFS and OS. P-values were calculated using a one-way ANOVA for means or Fisher's exact test for comparisons of rank. P < 0.05 was considered to indicate a signi cant difference. All statistical analyses were conducted using GraphPad Prism (software version 8; GraphPad Software Inc., San Diego, CA, USA) and R software (version 3.6.1; The R Foundation for Statistical Computing, Vienna, Austria).
Ethical considerations. This retrospective study was performed in accordance with the Declaration of Helsinki, and the study protocol was approved by the Ethics Committees of the Osaka International Cancer Institute (Osaka, Japan) (formerly the Osaka Medical Center for Cancer and Cardiovascular Diseases) (1404015008) and the Kochi University School of Medicine (Kochi, Japan) (ERB-000486). All participants gave their written informed consent for inclusion in this study. All data were analyzed anonymously.

Each AICS (lung) rank and clinicopathological factors
The relationship between each rank and clinicopathological factors was summarized in Table 1. There were more elderly patients (p = 0.003) in rank C. Regarding the surgical mode, standard surgery (lobectomy or more) for lung cancer was more frequently selected for rank C (p = 0.053). No signi cant differences were observed in gender, the smoking status, preoperative CEA level, p-stage, or treatment method among patients with each rank.

Effects of each AICS (lung) rank on tumor recurrence and OS
The median follow-up period in the present study was 6.0 years, ranging between 0.1 and 11.5 years.
There were 86 (30.0%) patients with tumor recurrence, and 42 (14.1%) died (Table 1). Tumor recurrence was observed in 22 (23.7%) patients with rank A, 15 with rank B (18.3%), and 49 with rank C (40.2%) ( Table 1). The incidence of recurrence was signi cantly higher in rank C than in rank A or B (p = 0.0014). Based on this result, the following comparison was performed in two groups: rank A + B vs. rank C.
The 5-year RFS rate in all patients was 71.7%. RFS curves in ranks A + B and C were shown in Fig. 1. Five-year-RFS rates were 79.9% in rank A + B and 62.5% in rank C, and were signi cantly different (p = 0.000113). Figure 2 shows the RFS curves of these two groups according to the p-stage and histological type. A signi cant difference was observed in p-stage I ( Fig. 1-a, p = 0.0026) and p-stage II ( Fig. 2-b, p = 0.0402), but not in p-stage III (Fig. 2-c, p = 0.141). Regarding the histological type, only adenocarcinoma showed a signi cant difference in RFS (Fig. <link rid=" g2">2</link>-d and 2-e, p = 0.0013). According to the univariate analysis of RFS in the present study, signi cant variables were the smoking status (non vs. current/ex), surgical mode (limited vs. standard), stage (I vs. II-IV), histological type (non-squamous cell carcinoma, non-Sq vs. squamous cell carcinoma, Sq), treatment method (surgery alone vs. surgery + adjuvant therapy), and preoperative AICS (lung) rank (rank A + B vs. rank C) ( Table 2). Table 3 shows a multivariate analysis of RFS. Of the six variables tested, the p-stage, AICS (lung) rank, and treatment method were identi ed as signi cantly independent factors affecting RFS.
Forty-two (14.1%) patients died. Death was not associated with the AICS rank ( Table 1). The 5-year OS rates were 89.5% in all patients, 91.5% in rank A + B, and 86.5% in rank C. A slight difference was observed in OS between rank A + B and rank C (Fig. 3, p = 0.0708).
Each AICS (lung) rank and the initial recurrence site Since a marginal difference was found in surgical mode among each rank (Table 1), the postoperative recurrence pattern was carefully examined. Table 4 shows the relationship between the AICS (lung) rank and initial recurrence site in this cohort. No signi cant differences were observed in the recurrence site between rank A + B and rank C.
The hypothesis for its development was based on the highly sensitive detection of minute changes in metabolic pro les with a focus on PFAA, even if the disease was in the early stages. The AICS™ test is currently performed to detect 7 types of cancers, and the AICS (lung) test for lung cancer is widely used in health and cancer screening in Japan. Several promising studies on similar or the same concepts have been reported abroad [5,21,22,24,25]. According to our previous ndings [1,8,9], this test may be an optional tool for the early detection of lung cancer, regardless of the tumor stage or progression, in contrast to other tumor markers, such as serum CEA and CYFRA. However, since the mechanisms responsible for AICS (lung) abnormalities, namely, increases in AICS (lung) values with the occurrence or progression of lung cancer, remain unclear, the usefulness of this test has not yet been examined in detail in clinical practice. For example, clinicopathological differences between patients with and without this AICS (lung) abnormalities remain unclear. We recently reported that AICS (lung) abnormalities in patients with lung cancer may be normalized (from rank C to A or B), when the tumor is completely removed, indicating that AICS (lung) values show some reversibility based on the disease condition [10]. Furthermore, similar ndings were reported for other types of cancers [26,27]. Therefore, changes in amino acid pro les, e.g., AICS values, before and after treatment may be used to predict prognosis and monitor recurrence in patients with several cancers [4,10,26,27].
In the present study, we examined the clinicopathological and prognostic signi cance of AICS (lung) abnormalities in 297 patients with NSCLC, who underwent curative surgery in our two institutes. The preoperative AICS (lung) status in patients with resectable NSCLC was observed on a relatively large scale: 93 with rank A (31.3%), 82 with rank B (27.7%), and 122 with rank C (41.1%). This distribution was similar to that in our previous study [10]. No relationships were observed between the preoperative AICS (lung) rank and clinicopathological factors, except for patient age and surgical mode. Consistent with previous ndings [1,2,8,9], the present results con rmed that AICS (lung) abnormalities was not related to the tumor stage at the start of surgical treatment. Moreover, preoperative o AICS (lung) abnormalities themselves had a signi cant effect on the frequency of postoperative recurrence among NSCLC patients undergoing curative surgery: patients diagnosed with AICS (lung) rank C in preoperative blood samples showed a signi cantly higher incidence of tumor recurrence even after complete resection than those diagnosed with the two other ranks. Although a signi cant impact on OS was not observed, the AICS (lung) rank was associated with the grade of tumor malignancy. This result was more apparent in patients with early-stage disease (p-stage I and II), particularly those with adenocarcinoma. This is the rst study to investigate the clinical relationship between the preoperative amino acid pro le status and tumor recurrence in and the prognosis of patients with cancer.
In the multivariate analysis, three variables (p-stage, treatment method, and AICS rank) were identi ed as signi cant high-risk predictors of tumor recurrence in the present cohort. As described previously [28], the p-stage had the greatest impact on recurrence after curative surgery. It currently remains unclear why the treatment method was regarded as a high-risk predictor; however, a bias was speculated because patients treated with surgery plus adjuvant therapy were intentionally selected as having a high-grade tumor malignancy and potentially unfavorable prognosis in each stage. In contrast, the AICS (lung) rank was rstly suspected to be a confounder associated with the surgical mode. In fact, limited surgery was more selected for rank A + B. Considering that the predictive power of the AICS (lung) rank was mainly limited to a subpopulation of patients with stage I and/or adenocarcinoma, it was also suspected that most patients with adenocarcinoma subtype indicating a predominant ground glass opacity (GGO) on preoperative CT scan image and/or a predominantly lepidic pattern in histology were included in rank A + B. However, such a prognostic good adenocarcinoma subtype was distributed regardless AICS (lung) rank (data not shown). Therefore, the importance of preoperative AICS (lung) abnormalities was clearly determined in a multivariate analysis.
We recently reported that most patients preoperatively diagnosed with rank C potentially exhibited a reduction in the AICS (lung) rank if they were cured after surgery [10]. In contrast, some NSCLC patients without AICS (lung) abnormalities in the preoperative test, namely, those diagnosed with rank A or B, also showed tumor recurrence, even after curative resection. These ndings demonstrated that the condition of AICS (lung) abnormalities themselves may not only re ect the tumor-bearing status in patients, but may also be strongly associated with some biological grade of malignant characteristics associated with tumor proliferation, invasiveness, and/or metastasis. However, no signi cant difference was observed in the clinicopathological factors tested among these ranks. Furthermore, although standard surgery for lung cancer was more frequently performed for rank C, no signi cant differences were observed in recurrence patterns in the present cohort, suggesting that AICS (lung) abnormalities are not associated with the metastatic route, such as lymphatic, venous, or disseminated spread.
In contrast to the results obtained for RFS, AICS (lung) abnormalities only slightly affected OS. We speculated that this discrepancy was mainly due to some therapeutic effects after recurrence because the predictive value of tumor recurrence was demonstrated in adenocarcinoma. However, typical biomarkers for adenocarcinoma, such as EGFR mutations and ALK abnormalities, were not adequately examined because of the old times of some enrolled patients. The potential relationship between AICS (lung) abnormalities and these therapeutic biomarkers needs to be examined in future studies.
There were some limitations that need to be addressed. Although the number of enrolled patients was relatively large, the predictive value of AICS (lung) abnormalities for tumor recurrence after curative resection was not satisfactory due to the retrospective nature of the analysis. Since its effect on tumor recurrence was limited in patients with early-stage disease and adenocarcinoma, the number of patients was too small to assess its signi cance. Moreover, the biological mechanisms and condition of AICS (lung) abnormalities in human cancer currently remain unknown. Several amino acids may exert promoting or inhibitory effects on the proliferation of cancer cells [4]. Alanine is regarded as an important key amino acid in apoptosis as well as the proliferation of cancer cells in vitro [4]. Moreover, according to recent ndings on novel tissue free amino acid (TFAA) pro le analyses [24,29], the TFAA pro le status was strongly associated with malignant characteristics as well as carcinogenesis in cancer patients, and PFAA pro les have been suggested to re ect the status of cancer tissues. Nevertheless, the present study was the rst to show that minute changes and/or alterations in PFAA pro les are strongly associated with the biological grade of malignant characteristics. To clarify the predictive signi cance of AICS (lung) abnormalities on tumor recurrence in NSCLC patients undergoing surgical treatment, not only further validation in other cancers, but also basic research of the underlying mechanisms are warranted.

Conclusions
The present study showed that preoperative AICS (lung) abnormalities were a high-risk predictor of postoperative recurrence in NSCLC patients undergoing curative resection. Although AICS was originally regarded as a screening modality for the risk of cancer, these abnormalities may have a biological role in increasing the grade of malignant characteristics in cancer patients.

Declarations
Competing interests HY and SK are employees of Ajinomoto Co., Inc. (Kanagawa, Japan). MH and TA have received research grants from Ajinomoto Co., Inc. (Kanagawa, Japan). This does not alter the authors' adherence to journal's policies. The remaining authors declare that they have no competing interests. Figure 1 Kaplan-Meier curves for recurrence-free survival (RFS) of pre-operative AICS (lung) rank A+B and rank C Figure 2 Kaplan-Meier curves for recurrence-free survival (RFS) of pre-operative AICS (lung) rank A+B and rank C according to the p-stage and histological type a) p-stage I b) p-stage II c) p-stage III d) Adenocarcinoma e) Squamous cell carcinoma Figure 3 Kaplan-Meier curves for overall survival (OS) of pre-operative AICS (lung) rank A+B and rank C