Circulating levels of high mobility group box-1 and nucleophosmin/B23 proteins and clinical significance in debut non-small cell lung cancer patients.

Introduction
Lung cancer remains one of the leading causes of cancer-related deaths each year worldwide, largely due to its late clinical presentation and the limited availability of biomarkers capable of predicting early diagnosis, disease progression, clinical response, and overall survival. However, currently, the majority of the identified biomarkers for non-small cell lung cancer (NSCLC) are actionable gene mutations, which promote personalized molecular targeted therapies. The above therapies yield a successful objective clinical response in ~ 30% of patients1.
Alarmins, which can promote, recruit, activate, and mature dendritic cells2, could potentially serve as soluble biomarkers easily measurable for the diagnosis, progression, and/or treatment of NSCLC.
More particularly, the present study primarily focused on the clinical significance of circulating levels of high mobility group box 1 (HMGB-1), which is a widely studied alarmin not only in cancer but also in autoimmunity and diabetes3. Additionally, the coexistence of circulating HMGB-1 with nucleophosmin/B23 (NPM/B23), a molecule that has not been studied as an alarmin in cancer, was also investigated.
On the other hand, a dual role, both pro- and anti-tumor, has been suggested for alarmin by other authors4. For example, a study indicated that intracellular HMGB-1, a non-histone chromatin-associated protein, could play a crucial role in tumor growth, affect response to therapy, and promote both cell death and survival5. However, it was also reported that this alarmin could be involved in different cancer progression-related pathways and could enhance the efficacy of immune checkpoint blockade in patients with cancer5. As a putative biomarker in cancer, the role of HMGB-1 in cancer has been supported by several studies. For example, a study demonstrated that the expression levels of HMGB-1 were increased in tumor tissues and serum of patients with urothelial carcinoma of the bladder6. Additionally, in advanced lung cancer, small cell lung carcinoma (SCLC) and mesothelioma, high HMGB1 levels were associated with shorter overall survival in patients with NSCLC7. In pancreatic ductal adenocarcinoma (PDAC), HMGB1 displayed a better diagnostic value compared with the traditional biomarkers, such as carbohydrate antigen 19 − 9 or carcinoembryonic antigen8. Furthermore, other studies revealed that HMGB-1 could predict gastric cancer progression9 and response to chemotherapy in breast cancer10.
Intracellular NPM/B23, also known as NPM1, B23, No38 or numatrin, is a phosphoprotein mainly localized in the nucleolus11. It is involved in several biological processes, including mRNA transport, chromatin remodeling, genome stability, and apoptosis. NPM/B23 is commonly overexpressed in solid tumors, and it is therefore mainly associated with increased mitotic activity and metastasis12. However, there is limited data on the presence of circulating NPM/B23 in the serum of patients with cancer and/or its potential role as an alarmin.
To the best of our knowledge, no prior studies have investigated the clinical significance of the coexistence of circulating HMGB-1 and NPM/B23 in the serum of patients with cancer. Therefore, the present study aimed to investigate whether these two proteins were associated with each other and/or with clinicopathological parameters in newly diagnosed patients with NSCLC.

Results

Demographic and clinicopathological characteristics
The mean age of patients in the NSCLC group was 61.2 ± 8 years, while that in the control group was 62.4 ± 11.8 years. Among the total of 232 patients with NSCLC enrolled in the present study, 98 (42.2%) and 134 (57.8%) were females and males, respectively. The control group was composed of 30 male and 30 female subjects. Among all patients with NSCLC, 162 (69.8%) were diagnosed with LUAD and 70 (30,2%) with LUSC. Clinical stages were categorized into two major cohorts: Stage I and stages II, III and IV. The mean tumor size, according to the maximum diameter, was 2.3 ± 1.3 cm. In addition, other clinical conditions considered in the current study included the presence or absence of lymphatic metastasis (23 vs. 191 patients), pleural invasion (23 vs. 191 patients), and STAS (79 vs. 135 patients). Tumor differentiation was classified as low (65 patients), mid (125 patients), mid-low (5 patients), high (23 patients), and high-mid (1 patient; Table 1). Overall, 162 NSCLC patients with paired samples (serum versus tissue) were used for analysis of correlation in this work and five patients that lacked optimal serum samples were added just for tissue sample analysis only (167 in total). Data on clinical response to standard anticancer therapy and overall survival were collected.

Levels of circulating alarmins in patients with NSCLC
As shown in Fig. 1A, the median HMGB-1 levels were significantly lower in patients with NSCLC (30.04 ng/ml) compared with healthy individuals (37.30 ng/ml; P = 0.02). Interestingly, patients with LUAD, but not those with LUSC, displayed markedly reduced HMGB-1 levels compared with the control group (P < 0.01). However, HMGB-1 levels were not significantly different between patients with LUAD and LUSC. By contrast, the median NPM/B23 levels were notably enhanced in patients with NSCLC (812 pg/ml) compared with healthy individuals (551 pg/ml; P = 0.001; Fig. 1B). When both histological subtypes were analyzed separately, circulating NPM/B23 levels were also significantly higher in both patients with LUAD and LUSC compared with the control group (P < 0.001; Fig. 1B). No significant difference in circulating NPM/B23 levels was observed between patients with LUAD and LUSC.

Association between circulating alarmin levels and clinicopathological features of patients with NSCLC
After assessing the serum levels of both alarmins in patients with NSCLC, the present study then investigated whether these levels were associated with any particular clinicopathological feature. Interestingly, the univariate analysis shown in Table 1 revealed a strong association between tumor size and HMGB-1 serum levels in patients with NSCLC. More particularly, significantly higher HMGB-1 levels were positively associated with tumors of > 2.0 cm in maximum diameter (P = 0.004), regardless of tumor histology. However, no other associations were recorded between circulating HMGB-1 levels and other clinicopathological features. When generalized linear models were applied using HMGB-1 as a dependent variable and incorporating all available clinical features, it was verified that only tumor size was associated with HMGB-1 levels (Table SI). Subsequently, potential associations between circulating NPM/B23 levels and particular clinicopathological features in patients with NSCLC were evaluated. As shown in Table 2, the presence of STAS pattern (P = 0.043) and a tumor size of > 2 cm (P = 0.0001) were associated with higher serum levels of NPM/B23 in patients with NSCLC. However, data from the multivariate analysis revealed that only a tumor size of > 2 cm was significantly associated with circulating NPM/B23 levels (Table SII).

Correlation between circulaing HMGB-1 and NPM/B23 serum levels in patients with NSCLC
To verify whether circulating HMGB-1 and NPM/B23 levels were associated with each other, the presence of both alarmins in the serum of healthy individuals and patients with NSCLC was analyzed. Therefore, as shown in Table 3, the serum levels of both HMGB-1 and NPM/B23 were associated with each other in both cohorts. However, their association was more significant in patients with NSCLC compared with healthy individuals (r = 0.679 vs. r = 0.278). As tumor size and the presence of STAS pattern were associated with the serum levels of circulating alarmins in the NSCLC group, the present study aimed to further investigate the association of the above two clinicopathological features with HMGB-1 and NPM/B23 levels. Interestingly, the analysis showed that the levels of both alarmins were significantly associated with a tumor size of > 2 cm (r = 0.70), regardless of the presence or not of STAS (Table 4). However, the strongest association between both circulating alarmins was observed in patients with NSCLC with tumors of ≤ 2 cm in diameter and the presence of STAS (r = 0.900; P = 0.0001; Table 4).

Expression of HMGB-1 and NPM/B23 in tumor and adjacent healthy lung tissues
Finally, the current study aimed to explore the intratumoral expression of both HMGB-1 and NPM/B23 alarmins in patients with NSCLC by immunohistochemistry, using tumor biopsies and adjacent healthy lung tissues located 5 cm from tumor, as internal control. Representative immunohistochemistry images of intracellular HMGB-1 expression in both tumor and adjacent healthy tissues are illustrated in Fig. S1. HMGB-1 was predominantly expressed in the nuclear compartment, not only in alveolar epithelial cells, but also in phagocytes in the alveolar space, fiber cells in the alveolar septum and fibrocytes. To semi-quantify the immunohistochemistry results, the HMGB-1 positive expression rate was calculated as the sum of patients exhibiting low (+), moderate (++), and high (+++) intracellular expression. As shown in Table 5, among 167 patients with NSCLC, the positive expression rate of HMGB-1 in tumor tissues was 78.44% compared with 86.83% in adjacent healthy lung tissues (P = 0.043). Notably, a distinct dichotomous expression pattern of HMGB-1 was observed based on NSCLC histological subtypes. Therefore, in patients with LUAD, the positive expression rate of HMGB-1 was significantly lower compared with that in adjacent healthy tissues (P = 0.0001). By contrast, in patients diagnosed with LUSC, the positive expression rate was notably higher in tumor tissues compared with adjacent healthy ones (P = 0.001). Interestingly, more than 50% of adjacent healthy tissues also displayed positive HMGB-1 expression.
Subsequently, whether a potential association existed between intracellular HMGB-1 levels in tumors and adjacent healthy tissues was assessed. Therefore, a statistically significant, but weak, positive association was found between intracellular HMGB-1 expression in tumors and their corresponding adjacent normal tissues counterpart (P = 0.0001; r = 0.348; Table SIII). However, no significant association was observed between circulating serum HMGB-1 levels and tumor or normal tissues (Table SIII). By contrast, the immunohistochemistry analysis revealed a nuclear or nuclear/nucleolar expression pattern for the NPM/B23 alarmin (Fig. S2). Unlike HMGB-1, the overall positive NPM/B23 expression rate in tumor tissues was notably higher compared with that observed in adjacent healthy tissues (92.22 vs. 85.03%; P = 0.0387; Table 6). More particularly, the positive NPM/B23 expression rate in tumor tissues from patients with LUAD was 91.54% compared with 83.08% recorded in adjacent healthy tissues. Interestingly, in patients with LUSC, the positive expression rate of NPM/B23 was not significantly different between tumor and adjacent healthy tissues (94.60 vs. 91.89%; P = 0.6433; Table 6). Nonetheless, NPM/B23 showed a significantly increased expression rate in patients with NSCLC. Similarly, with HMGB-1, a significant but weak positive association was also observed between intracellular NPM/B23 expression in tumor and adjacent healthy tissues (P = 0.0001; r = 0.394; Table SIV). Consistently, no association was obtained between circulating NPM/B23 levels in serum and tissue expression in either tumor or normal tissues (Table SIV). Given the increased positive expression rate of both HMGB-1 and NPM/B23 in adjacent healthy tissues, the expression of Ki-67, a cell proliferation biomarker that is not an alarmin, was determined to verify specificity. As expected, Ki-67 expression was very weak in adjacent healthy tissues compared with NSCLC tissues (2.2 vs. 56.7%; P < 0.001).
Finally, to assess whether HMGB-1 and NPM/B23 alarmins were associated at the tissue level, a Spearman coefficient analysis was performed in both malignant and adjacent normal tissues. Unlike the findings in serum, only a weak correlation between the two alarmins was observed in tumor tissues (P = 0.02; r = 0.223; Table SV), but not in adjacent healthy tissues (Table SVI).

Discussion
Early detection of novel disease biomarkers in NSCLC can optimize clinical decision-making for therapy, but can also complement computed tomography (CT) scan and chest radiography, thus helping to mitigate cancer-related mortality and improve overall prognosis in lung cancer. As mentioned above, alarmins have been proposed as biomarkers for the diagnosis, progression, and/or treatment of NSCLC. In particular, previous meta-analyses suggested that the HMGB-1 alarmin was upregulated in both tumors and serum of patients with NSCLC compared with healthy individuals1314, supporting its potential as a biomarker for predicting disease progression and survival. However, other studies demonstrated that HMGB1 was downregulated in NSCLC tumor tissues compared with matched normal lung tissues5. The present study investigated the relevance of pretherapeutic circulating and tissue levels of HMGB-1, as well as those of MPN/B23, which has been suggested as a secreted alarmin in sepsis and found to be elevated in patients with NSCLC1617. More particularly, herein, the clinical significance of the co-expression of both proteins in relation to key clinicopathological parameters in newly diagnosed patients with NSCLC was assessed. In addition, the association between HMGB-1 and NPM/B23 at both systemic and tissue levels was investigated in tumor samples. The results demonstrated that the median global circulating HMGB-1 levels in the serum of patients with NSCLC were lower compared with those observed in healthy individuals, and more particularly in patients diagnosed with LUAD (Fig. 1A). It is currently known that different oxidation states of HMGB-1 can lead in three mutually exclusive functions, acting as an alarmin18, and promoting chemoattraction19 or tolerance20. Therefore, it would be of great interest to determine in the future whether a specific redox status of HMGB-1 occurs in patients with LUAD and whether this status could be associated with clinical outcome, disease progression, or survival. Nonetheless, the results showed that elevated individual HMGB-1 levels were positively associated with a tumor size of > 2 cm in diameter (P = 0.004; Table 1), regardless of NSCLC histological subtype. This biological association merits further clinical validation as putative indicator of tumor progression in both LUAD and LUSC. These results were consistent with previous ones showing positive association between circulating HMGB-1 levels and tumor size, among other clinicopathological features14. Although a correlation between HMGB-1 and other clinicopathological parameters was not found in patients with NSCLC, the current study was the first to report a 2-cm tumor size threshold as clinically significant in relation with HMGB-1 expression. Notably, none of the correlations observed between alarmins and clinicopathological features were associated with the NSCLC histological subtype, which could be due to the sample size imbalance in the cohort included in the study (LUAD, n = 124 vs. LUSC, n = 38).
NPM/B23, a multifunctional protein involved in cancer, exerts chaperone activity and plays a key role in ribosomal biogenesis. Previous studies demonstrated that NPM/B23 was upregulated in patients with cancer111221. Although only a few studies have investigated the extracellular role of NPM/B23 as an alarmin, experimental evidence has suggested that beyond being a potential biomarker, it can also serve as a therapeutic target. For instance, a previous study indicated that NPM/B23 was physiologically secreted by endothelial cells under stress, thus exerting proangiogenic activity both in vitro and in vivo through the secretion of vascular endothelial growth factor-A, hepatocyte growth factor, stromal derived factor-1, fibroblast growth factor-2, platelet derived growth factor-B, and matrix metallopeptidase 9 22. Consistently, recombinant NPM/B23 could induce the release of the proinflammatory cytokines, tumor necrosis factor-, IL-6 and monocyte chemoattractant protein-1 from RAW264.7 cells and increase the expression levels of intercellular adhesion molecule 1 in human umbilical vein endothelial cells, thus supporting its additional role in systemic inflammation16. Interestingly, the toll-like receptor 4/myeloid differentiation factor 2 complex has been identified as a receptor which is required for the extracellular function of NPM/B23 23. Consistently, other studies reported elevated circulating levels of NPM/B23 in patients with NSCLC compared with healthy individuals, a positive association with the presence of nodules, and markedly higher levels of NPM/B23 in patients with LUAD compared with those with LUSC17. In agreement with the aforementioned findings, the results of the present study revealed significantly higher levels of NPN/B23 in patients with NSCLC across both histological subtypes (Fig. 1B). Furthermore, both the univariate and multivariate analyses identified a notable association between elevated NPM/B23 levels and tumor size of > 2 cm (P = 0.0001; Table 2), regardless of histological subtype. Interestingly, higher circulating levels of NPM/B23 were modestly associated with the presence of STAS pattern (P = 0.043). Since the presence of STAS pattern in NSCLC indicates early spread of tumor cells through airspaces adjacent to the primary tumor and is associated with a higher risk of recurrence2425, additional investigation could be useful to decipher whether soluble NPM/B23 elicits an instrumental role in this particular clinical condition of NSCLC or it is rather a consequence. Otherwise, the univariate and multivariate analyses did not reveal any significant associations between NPM/B23 levels and other clinicopathological features. The above findings indicated that both HMGB-1 and NPM/B23 were individually associated with clinical indicators of disease progression, such as tumor size and STAS pattern, regardless the NSCLC histological subtype. In particular, these findings reinforce the underexplored role of NPM/B23 as putative alarmin in newly diagnosed patients with NSCLC.
After evaluating the clinical relevance of HMGB-1 and NPM/B23, the present study aimed to analyze the association between both proteins in serum and tissue samples from patients with NSCLC. Importantly, the results revealed a positive association between both proteins in the serum of newly diagnosed patients with NSCLC (Table 3), reinforcing the hypothesis that NPM/B23 could also act as an alarmin in this clinical setting. Considering that the individual analysis of circulating HMGB-1 and NPM/B23 levels was associated with clinical parameters involved in disease progression, such as tumor size and the presence of STAS, whether this correlation was maintained across the abovementioned clinicopathological parameters was subsequently investigated. A significant positive correlation was observed between HMGB-1 and NPM/B23 levels when tumor diameters were < 2 cm and in the presence of STAS (P = 0.0001; r = 0.900; Table 6). For tumors > 2 cm, a weak but significant association was obtained between both alarmins, regardless of the presence of STAS pattern. To the best of our knowledge, this study was the first to report a direct positive association between circulating HGMB-1 and NPM/B23 levels in patients with cancer, as well as their association with specific tumor size thresholds and STAS. Although elevated individual levels of circulating HMGB1 and NPM/B23 could reflect changes associated with disease progression, the strong biological association observed in a fraction of patients between both proteins in smaller tumors with STAS should be further investigated as putative biomarker for the early detection of the disease and/or for predicting the risk of NSCLC progression. Such biomarkers could be valuable, especially when combined with non-invasive imaging procedures, such as CT scan and Magnetic Resonance Imaging. In addition, early detection could assist oncologists in developing more effective therapeutic strategies. Therefore, the results of the current study warrant validation in larger patient cohorts.
Considering that alarmins can be released from tumor cells during the tumor-host immune interactions and the necrotic processes inherent to solid tumors, subsequently, the expression levels of HMGB-1 and NPM/B23 were detected in tumor and normal adjacent lung tissues. The results indicated that, although the positivity rate or HMGB-1 expression in tumor tissues was > 50%, it was lower than that observed in matched adjacent healthy tissues, especially among patients with LUAD (Table 5). By contrast, in patients with LUSC, this expression pattern was rather reversed, possibly due to the function of HMGB-1 as an alarmin. Therefore, further studies are needed to elucidate the significance of the differential expression of HMGB-1 in tumor tissues compared with adjacent healthy lung tissues in patients with LUAD (Table 6). On the other hand, the positive expression rate of NPM/B23 was > 80% in both tumor and healthy tissues, which was a novel observation in this study. By contrast, Ki-67, a common biomarker of cell proliferation that does not act as an alarmin, showed a very low positive expression rate (2.2%) in healthy tumor tissues of patients with NSCLC. To evaluate whether the co-expression of HMGB-1 and NPM/B23 in tumor tissues behaved similarly to that observed in the circulating levels, a correlation analysis was carried out. Interestingly, a weak correlation (P = 0.02; r = 0.223) was obtained, thus suggesting that future studies with a greater number of patients with NSCLC are needed to strengthen the above finding. Nonetheless, the weak correlation might be influenced by the semiquantitative output of immunohistochemistry compared with ELISA, which is fully quantitative. No relevant associations between the intracellular expression levels of both proteins and individual clinicopathological parameters were identified. Therefore, circulating alarmin-based biomarkers could be more useful than such tissue biomarkers to support disease progression of NSCLC; this notion merits additional investigation.
In summary, the present study investigated the individual and correlative features of HMGB-1 and NPM/B23 alarmins in the serum and tissues of patients with NSCLC, who had not received any anticancer therapy. This study was the first to describe the correlations between these two proteins in the serum of patients with NSCLC and their clinical characteristics at the onset of the disease. However, future directions of the research must be focused on the validation of this potential alarmin-based biomarker of disease progression in larger NSCLC patient cohorts.

Methods

Subjects
A total of 232 patients with NSCLC were initially enrolled in the present study. A total of 162 NSCLC patients with high-quality paired samples (serum versus tissue) and 60 age-matched healthy individuals were enrolled in this prospective study and further used for all the correlation analysis. Five patients who lacked optimal serum samples were added just for tissue sample analysis only (167 in total). The study and procedures were approved by the Ethics Committee of the First Affiliated Hospital of the University of South China, Hunan Province, and reviewed with lot number No.2021LL0115001. Also, the study was conducted in accordance with ethical principles from the Helsinki declaration (2013), the regulations on Biomedical Ethics Review involving human (2016), the regulations on Clinical Research Ethics review in traditional Chinese medicine and the International Ethical Guidelines for Biomedical Research. In addition, patients signed a pan-informed consent form when they were admitted to the hospital, and informed the specimens to be used for research purposes. Although patient identification records are not published in this work, consent for publication of the general data in this work was also obtained.

Enrollment criteria for NSCLC patients and sample collection
Only debut patients without previous chemoradiotherapy, targeted therapy, or immunotherapy before surgery, were included. Also, patients with paraffin histology diagnosis after radical resection of lung cancer with LUAD or LUSC were eligible for the study. Those patients with poor quality of serum, other histology types, or previous lung cancer surgery were excluded from the analysis. All the patients’ samples originated pairs of wax blocks (cancer and its corresponding normal lung tissue taken 5 cm away from tumor). Five milliliters of blood from patients and healthy individuals were obtained and serum was collected and kept at −80 °C until used for measuring HMGB-1 and NPM/B23 by ELISA. Those enrolled patients with missing clinical data, recent infectious or inflammatory disease during the study or inadequate paired samples were excluded from the final analysis.

Measurement of HMGB 1 and NPM/B23 in serum
HMGB-1 in serum was quantified used a commercially available Human High-Mobility Protein 1 (HMGB 1) ELISA Kit (ml026429), Shanghai mlbio Technology Co., LTD. On the other hand, NPM/B23 was determined by using a Human Nucleolar Phosphoryl Protein (NPM) ELISA Kit (ml058225), Shanghai mlbio Technology Co., LTD.

Immunohistochemistry and scoring criteria in tissue samples from NSCLC patients
In total, 167 paraffin-embedded NSCLC and healthy tissue samples were submitted to immunohistochemistry analysis by using a Ready to Use Immunohistochemical Hypersensitivity UltraSensitiveTM SP kit (rabbit/rat) (KIT-9729), Fuzhou Maixin Biotechnology Development Co., LTD. Serial sections of tumor specimens were cut and stained with antibodies specific for both alarmins. Recombinant Anti-HMGB1 Antibody [EPR3507] (ab79823), and Recombinant Anti-Nucleophosmin Antibody [EP1848Y] (ab52644), Abcam (Shanghai) Trading Co., Ltd, were used as primary antibodies for detecting HMGB-1 and NPM/B23 respectively. Samples were scored independently by two observers. Briefly, the slides were scored based on intensity (negative, 0; weak, 1+; weak positive, 2+; positive, 3+; and strongly positive, 4+), and percentage of positively stained tumor cells < 5% (negative); 5–20% (1+); 21–50% (2+); 51–75% (3+) and > 75% (4+). Data were expressed as a positive expression rate which corresponds to the number of positive sections with respect total sections × 100.

Statistical analysis
The quantitative variables are reported as mean, standard deviation (SD), median, interquartile range, and range. The qualitative are reported as absolute and relative frequencies. The expression of NPM/B23 and HMGB-1 in patients and healthy subjects was compared using the Mann-Whitney U test. A generalized linear model was used to explore the association between NPM/B23 and HMGB-1 (each one) regarding clinicopathological parameters. Spearman’s rank correlation was used to assess relationships among NPM/B23 and HMGB-1 globally and stratified by tumoral size and Spread Through Airspaces (STAS). Statistical analyses were done using SPSS version 26. The level of significance chosen was α = 0.05.

Supplementary Information
Below is the link to the electronic supplementary material.