Determination and evaluation of serum monosaccharides in patients with early-stage lung adenocarcinoma

To the Editor: Lung cancer remains the leading cause of cancer-related deaths worldwide.[1] Its histological classification includes non-small cell lung cancer (NSCLC), which encompasses adenocarcinoma, squamous cell carcinoma, and large cell carcinoma. Among these, lung adenocarcinoma (LUAD) is the most common subtype. Visceral pleural invasion (VPI) is a significant factor influencing tumor grade and serves as an independent risk factor for the prognosis of patients with lung cancer. However, due to the lack of measurable markers in most cases, it can only be detected through imaging techniques and pathological examinations.[2] Currently, commonly used biomarkers in clinical screening, such as carcinoembryonic antigen (CEA), still exhibit a high false-positive rate in diagnosing early-stage lung cancer. Therefore, identifying new markers or methods to more accurately evaluate the characteristics of minimally diseased pulmonary nodules holds considerable importance, especially in the early stages of lung cancer diagnosis. The biological glycans in blood are synthesized without a template, and their content and structure are regulated by both genes and internal and external environmental influences. These factors encompass genetic and environmental information, which can serve as potential biomarkers for system malfunctions, enabling the early detection and assessment of lung cancer. The integrated structural information of glycans holds promise for development into system malfunction-based biomarkers for cancer diagnosis.[3] However, current glycomics-based biomarker development faces technical challenges for clinical application, and biomarkers based on the simple composition of circulating blood monosaccharides have been seldom reported. This study was performed in accordance with the Declaration of Helsinki. The Institutional Review Board of the Affiliated Hospital of Qingdao University approved this retrospective study (Approval No. QYFYWZLL26174). In this study, we collected the blood samples from age- and gender-matched patients with stage I LUAD (n = 268), including 53 patients with VPI, and healthy controls (n = 268). Additionally, in the stage I LUAD group, 53 patients without VPI (NVPI) were matched with 53 patients with VPI and served as subgroup controls for subsequent analyses. Furthermore, 71 patients with early-stage LUAD (VPI, n = 19; NVPI, n = 52) and 26 healthy individuals undergoing physical examination were enrolled during the same period as the retrospective validation group [Supplementary Table 1, https://links.lww.com/CM9/C237]. All serum samples were obtained from the laboratory of our hospital and stored at –80°C for future use. Our lab has developed a polymerase chain reaction (PCR) instrument-based hydrolysis method coupled with high-performance liquid chromatography (HPLC) and pre-column 1-phenyl-3-methyl-5-pyrazolone (PMP) derivatization for the quantification of monosaccharides in serum samples. Detailed procedures for this analysis can be found in Spring Natural Protocol Exchange.[4] All serum samples were subjected to hydrolyzation-independent analyses [Supplementary Figure 1A, https://links.lww.com/CM9/C237], which included PCR instrument-based hydrolysis followed by monosaccharide PMP derivatization for the quantification of fucose (Fuc), galactose (Gal), galactosamine (GalN), glucose (Glc), glucosamine (GlcN), acetylglucosamine (GlcNAc), glucuronic acid (GlcUA), mannose (Man), rhamnose (Rha), and xylose (Xyl). Monosaccharide standards were purchased from Sigma-Aldrich (Saint Louis, Missouri, USA), and PMP was also purchased from Sigma-Aldrich. All other reagents used were of analytical-grade purity. For HPLC analysis, ZORBAX XDB-C18 columns (150 mm × 4.6 mm, 5 µm) were purchased from Agilent Technologies (Santa Clara, California, USA). The ten kinds of monosaccharides standard solution were mixed evenly in the same volume, and the mixed solution was diluted twofold step by step for analysis by HPLC. The liquid chromatogram of the corresponding concentration can be obtained, which can be used to draw the standard curve. The chromatograms of the standard monosaccharides were obtained by HPLC analysis of the mixture of the standard monosaccharides at the initial concentration [Supplementary Figure 1A, https://links.lww.com/CM9/C237]. Due to the design conditions of this study, the hydrolysis yield of GlcNAc, GlcUA, and Xyl will be affected and Rha was used as an internal control to evaluate the stability of the operation, therefore, the four aforementioned monosaccharides are not contained in the subsequent studies. Taking the abscissa as the concentration of the monosaccharide standard and the ordinate as the corresponding peak area, the corresponding monosaccharide standard curve and the corresponding equation were drawn [Supplementary Figure 2A, 2B, Supplementary Table 2, https://links.lww.com/CM9/C237]. Comparing the content of six monosaccharides between the healthy control group and the early-stage LUAD group, we found a statistically significant difference in serum monosaccharide degradation between the two groups, as shown in Supplementary Figure 1B and Supplementary Table 3, https://links.lww.com/CM9/C237 (P <0.001). In the LUAD group, the levels of Man, GlcN, GalN, Glc, Gal, and Fuc were higher than those in the control group. We first performed a univariable logistic regression analysis for each monosaccharide, and the results are shown in Supplementary Table 4, https://links.lww.com/CM9/C237. Monosaccharides that showed statistical significance in the univariable regression were selected for subsequent multivariable regression analysis. The multivariable logistic regression included Man, GlcN, Glc, Gal, and Fuc for model construction. We also plotted receiver operating characteristic (ROC) curve for each monosaccharide and for the final fitted model, as shown in Supplementary Table 5 and Supplementary Figure 1C, https://links.lww.com/CM9/C237. The combined diagnostic model (Model-L = –8.035 – 0.033 × Man + 0.009 × GlcN + 0.002 × Glc – 0.027 × Gal + 0.029 × Fuc) effectively distinguished patients with early-stage LUAD from healthy individuals, with an area under the ROC curve (AUC) of 0.8652. For visualization, we created a nomogram for Model-L, shown in Supplementary Figure 1D, https://links.lww.com/CM9/C237, and the calibration curve is shown in Supplementary Figure 3, https://links.lww.com/CM9/C237. Compared with the NVPI group, Man, GlcN, Gal, and Fuc exhibited an increasing trend in the VPI group. However, the differences in GalN and Glc between the two groups were not statistically significant [Supplementary Figure 1E, Supplementary Table 6, https://links.lww.com/CM9/C237]. Subsequently, univariable logistic regression was performed for each monosaccharide, with the results presented in Supplementary Table 7, https://links.lww.com/CM9/C237. Following the same method used previously, we constructed a multivariate logistic regression model to distinguish between patients with VPI and NVPI. A multivariable logistic regression model (Model-V = –3.504 + 0.044 × Man – 0.008 × GlcN – 0.013 × Gal + 0.042 × Fuc) was developed, incorporating the monosaccharides Man, GlcN, Gal, and Fuc, which demonstrated high diagnostic value for detecting the VPI pathological state with an AUC of 0.8829 [Supplementary 1F, Supplementary Table 8, https://links.lww.com/CM9/C237]. The nomogram for visualizing Model-V is shown in Supplementary Figure 1G, https://links.lww.com/CM9/C237, and the calibration curve is presented in Supplementary Figure 4, https://links.lww.com/CM9/C237. Finally, we validated the two constructed models using the retrospective validation group. Model-L demonstrated stability with a sensitivity of 92.0%, a specificity of 46.8%, a positive predictive value (PPV) of 64.8%, a negative predictive value (NPV) of 84.6%, and an overall accuracy of 70.1%. In the same group, Model-V showed a sensitivity of 63.3%, a specificity of 100%, a PPV of 100%, an NPV of 78.8%, and an accuracy of 84.5%. Our data showed that circulating monosaccharides levels not only differed between patients and healthy controls but also showed significant differences among patients with varying pathological characteristics of nodules. Binary logistic regression analysis showed that, compared with the detection efficiency of individual monosaccharides, the model based on the regression results exhibited superior AUC and accuracy. Among the monosaccharides examined, Man played a particularly significant role in our diagnostic model, as indicated by its strong statistical performance. The structure of carbohydrate chains is complex, with different chain structures potentially sharing similarities and exhibiting various forms of expression.[5] The detection of carbohydrate chains requires advanced equipment and skilled operators. Currently, there are several methods for measuring monosaccharides, with HPLC being a particularly efficient and simple detection technology. HPLC allows for the stable and accurate detection of degraded monosaccharides in serum, providing rapid and reproducible results. The findings can reveal changes in patients’ serum from a glycomic perspective. Developing a “glycomic fingerprint” for early-stage lung cancer offers a multi-dimensional approach to describe and evaluate the disease, thereby broadening the clinical application of glycobiology. Statistically significant differences were observed in serum monosaccharide degradation between patients with early-stage LUAD and healthy individuals from the same period (P <0.05). The two groups of patients exhibited distinct “monosaccharide fingerprint” characteristics. Specifically, the distributions of Man, GlcN, Gal, and Fuc differed between patients with early-stage LUAD in the VPI and NVPI groups. Compared with the NVPI group, the serum contents of Man, GlcN, Gal, and Fuc were higher in patients in the VPI group. Although the specificity of Model-L was relatively unsatisfactory in the retrospective validation group, which may be due to bias introduced by the composition of our retrospective validation group, the diagnostic model constructed using monosaccharides has meaningful reference value when considering diagnostic indicators. The diagnostic model based on serum monosaccharide content showed good diagnostic efficacy in distinguishing the pathological features of VPI in patients with early-stage LUAD, providing certain reference significance for clinical diagnosis and evaluation. Additionally, the changes in monosaccharide components are different among patients with different pathological characteristics, highlighting the potential of monosaccharides as promising biomarkers. Our study has some limitations. Our study is a single-center study and lacks external validation. Additionally, the study population was limited to patients with early-stage LUAD. The sample size is not large enough, and a larger sample size would be more advantageous in supporting the conclusions of our research. Despite efforts to match the study population, selection bias may still be present. Furthermore, in our methodology, sialic acid, glucuronic acid, and iduronic acid were below the detection limits in certain serum samples, and as a result, these monosaccharides were excluded from data analysis in the current study. Acknowledgments We sincerely thank the patients and healthy individuals who made this study possible by providing access to their blood test results. We also thank the personnel in the data management department of our hospital for their assistance in obtaining these blood test results. Funding This research was supported by the Natural Science Foundation of China (Nos. 30981672585 and 11804151) and the Natural Science Foundation of Shandong Province of China (Nos. ZR2020QH286 and ZR2020QC059). Conflicts of interest None.