A combination of hepatic leukemia factor and circulating tumor cells serve as effective biomarkers for lung adenocarcinoma prognosis

High-throughput data acquisition and analysis

The RNA sequencing and copy number variation profiles of LUAD from The Cancer Genome Atlas (TCGA; https://tcga-data.nci.nih.gov/tcga/) were downloaded. LUAD samples were divided into two groups of EGFRw.t. and EGFR mutations, for which the expression value of HLF was procured (the unit was RNA-Seq by Expectation Maximization, RSEM), and the log2 value of each sample was analyzed using Excel 2019.

Patients and clinical samples

A total of 343 cases of LUAD from 2017 to 2019 were collected from Jiangmen Central Hospital (Guangdong, China). Detailed follow-up information is provided in Table S1. In the follow-up data, tumor stage, grade, overall survival and progression-free survival were the key information. The tumor-node-metastasis (TNM) system (Lababede & Meziane, 2018; Telloni, 2017), includes: tumor size and local growth (T), extent of lymph node metastases (N), and occurrence of distant metastases (M), was used to define the extent of disease. In the analysis of tumor metastasis, NX, N0 and M0 were included as the group without metastasis, while N1, 2, 3 and M1 were included as the group with metastasis. The EGFR mutations were detected using HybriBio (Guangzhou, China; https://www.hybribio.cn/) using Sanger sequencing, and the mutation sites were compared to Cosmic (http://www.sanger.ac.uk/genetics/CGP/cosmic/). Full details are provided in Table S2. Approval for this study was provided by the Ethical Review and Approval Committee of Jiangmen Central Hospital (Approval number: 2019-002). All patients provided signed informed consent.

Immunohistochemical staining and analysis

HLF expression in paraffin sections was detected by immunohistochemical staining. The experiments were performed by a professional pathological technician. Immunohistochemistry and scoring of HLF expression were performed as previously described (Chen et al., 2020; Zhang et al., 2017; Wu et al., 2020). The staining index (SI) was scored at 20X magnification using an M8 digital microscope (Precipoint, Garching, Germany). Expressions of HLF were classified into low (SI ≤ 4) or high (SI > 4) subgroups.

Identification of CTCs and analysis

For clinical samples, CTC enrichment was performed using the Human Circulating Tumor Cell (hCTC) Subtraction Enrichment Kit (SHE-011; Cytelligen, San Diego, CA, USA) or the Tumor Fisher CTC Detection Kit (Nanopep Biotechnology, Beijing, China) according to the manufacturer’s instructions with six mL of peripheral blood. The cell mixture was obtained using the same protocol as in a previous study, as well as for CTC identification (Liu et al., 2021). Immunofluorescence staining with CD31, CD45, DAPI, and FISH with CEP8 were performed to identify the CTCs. Cells with CD31⁻/CD45⁻/CEP8⁺/DAPI were recognized as single CTC, and cells with CD31⁻/CD45⁻/CEP8⁺/DAPI ≥ 2 were recognized as CTC cluster. Those cells with CD31⁻/CD45⁺/CEP8⁺/DAPI were regarded as white blood cells (WBC) and with CD31⁺/CD45⁻/CEP8⁺/DAPI were regarded as circulating epidermal cell (CEC). High and low quantities of CTCs were classified into low (CTCs ≤ 3) or high (CTCs > 3) subgroups.

Statistical analyses

The mRNA expression levels of HLF from TCGA are presented as means ± standard deviation (SD) and were analyzed with unpaired t-tests and Tukey’s multiple comparisons test. The chi-square test was used to analyze the relationship between HLF expression and EGFR variation, as well as the relationship between the quantity of CTCs and EGFR variation. Survival curves were plotted using the Kaplan–Meier method and compared using the log-rank test. P < 0.05 was considered significant. All experiments were repeated three times. Significant differences were determined using GraphPad Prism 10.

HLF is upregulated in EGFR-mutated LUAD specimens

Genes associated with EGFR mutations in lung cancer samples were analyzed using TCGA. HLF expression was associated with EGFR mutation, with HLF was significantly upregulated in EGFR-mutated samples (Fig. S1A). Among the EGRF-mutation samples, there were 15 cases of E19del., 18 cases of L858R and 33 cases of other mutation types. HLF was significantly upregulated in samples from the E19del. and L858R mutations compared to the wild type. There was a tendency for an increase in the remaining types of mutation samples, but the difference was not statistically significant (Fig. S1B). In total, 343 clinical LUAD samples were collected from our hospital and subjected to Sanger sequencing. There were 139 cases of EGFR mutation, including 70 cases of E19del., 56 cases of L858R and 13 cases of other types of mutation. Further details are presented Table S1. Table S2 provides details of the sequencing results of the 343 LUAD cases. Immunohistochemical staining of HLF was performed on surgical or puncture specimens from the collected cases. HLF was expressed to varying degrees in both surgical and puncture specimens (Fig. 1A). After scoring, the SI of the HLF was distributed from 0 to 12, as shown in Fig. 1B, with SI = 4 as the grouping criterion. After analysis, HLF was found to be upregulated in patients with EGFR mutations compared to the wild type (P = 0.061) (Fig. 1C). Compared with the HLF low group, the proportion of the HLF high group (SI > 4) was significantly higher in the EGFR mutant group (Fig. 1D). HLF also showed no differences between the different EGFR mutant site samples (Fig. 1E), and the proportion of high HLF was significantly increased in the at E19del group. However, the difference was not significant (P = 0.053, Fig. 1F).

The number of CTCs was negatively correlated with HLF expression, but not with EGFR mutation

Using a fully automated scanning fluorescence microscope, single-nucleated cells were tagged with CD31, CD45, and CEP8. The phenotypes of CEC, WBC, and CTC were respectively marked with CD31⁺CD45⁻CEP8⁺, CD31⁻CD45⁺CEP8⁺, and CD31⁻CD45⁻CEP8⁺. The results demonstrated that CTCs were effectively separated and collected from peripheral blood (Fig. 2A). The distribution of cases is shown in Fig. 2B. The number of cells was employed as a differentiation metric, with more than three cells being classified as the HLF high group and three or fewer cells being classified as the HLF low group. One to sixteen CTCs were captured from the collected cases, and the number of CTCs was not related to EGFR mutation (Figs. 2C, 2E). In comparison, the percentage of high or low CTCs was also not significantly associated with EGFR mutation, despite gene locus mutation (Figs. 2D, 2F). When HLF high or low was used as a distinction, there was a significant increase in CTCs captured in HLF low cases compared to HLF high (Fig. S2A), and the percentage of CTCs was significantly higher in patients with HLF low expression (Figs. S2B, S2C). These results confirmed that the number of CTCs was not associated with EGFR mutations but was inversely correlated with HLF expression.

The expression of HLF and the number of CTCs were significantly associated with the prognosis of LUAD patients

Follow-up of 343 LUAD cases revealed that EGFR mutations had no effect on the degree of pathological differentiation of the patients’ tumor size, metastasis, or stage, as documented in detail in Table S3. In contrast, HLF expression level was closely associated with tumor progression. By comparison, the number of cases with poorly differentiated tumors, large tumor size, tumor metastasis, and advanced stage were statistically higher in patients in the low HLF group; detailed information is presented in Table S4. After Univariate COX regression analysis, the factors associated with progression-free survival in patients with LUAD were tumor stage, tumor size, proximal metastasis, distant metastasis, HLF expression, and the number of CTCs (Fig. 3A). Detailed follow-up information and comparative analyses are presented in Tables S3 and S4. The results of the Multivariate COX regression analysis demonstrated that high HLF expression was an independent protective factor for progression-free survival in patients with lung adenocarcinoma (P < 0.001), whereas high CTCs were an independent risk factor for recurrence or death in patients with LUAD (P < 0.001) (Fig. 3B). Patients with high HLF expression in tumor tissues showed longer survival, with a median survival of 55.2 months (Fig. 3C). In contrast, more CTCs captured in the peripheral blood was associated with a poor prognosis, with a median survival of 28.7 months (Fig. 3D). When low HLF expression and high CTCs coexisted (two points), patients had the shortest median survival of 20.1 months (Fig. 3E). When low HLF or high CTC levels appeared alone (1 point), the median patient survival was 53.3 months. If high HLFs and low CTCs were present together (0 points), the median survival exceeded 60 months. Metastasis is one of the leading causes of death in cancer patients. As ‘seed cells,’ CTCs are closely related to tumor metastasis. Patients with high CTCs had a much higher rate of metastasis than those with low (Fig. 3F). Patients with a high HLF showed the possibility of reduced metastasis, with a lower rate of metastasis than those with a low HLF (Fig. 3G).

In EGFR wild-type cases, univariate COX regression analysis showed that tumor size, distant metastases, HLF expression, and the number of CTCs were correlated to progression-free survival (Fig. 4A). Multivariate COX regression analysis revealed that HLF high expression was an independent protective factor for progression-free survival in patients with LUAD (P = 0.005), whereas CTCs high was an independent risk factor (P = 0.005) (Fig. 4B). The median survival time of patients with high HLF expression was 54.4 months, which was significantly longer than 21.9 months of patients with low HLF group (Fig. 4C). Patients in the CTCs high group had a shorter median survival time of 24.0 months (Fig. 4D). When low HLF expression and high CTCs coexisted (2 points), the patients had the worst median survival, 21.7 months. When low HLF or high CTCs appeared alone (1 point), the median survival was 30.6 months, and if high HLF and low CTCs were present together (0 points), the median survival was greater than 60 months (Fig. 4E). Patients with elevated CTC levels of CTCs experienced a significantly higher incidence of metastasis compared to those with lower CTCs levels (Fig. 4F). Conversely, high HLF expression exhibited the potential for decreased metastasis, with a reduced metastasis rate compared to that of low HLF (Fig. 4G).

In cases with EGFR mutations, Univariate COX regression analysis revealed that tumor pathologic grade, proximal metastases, HLF expression, and the number of CTCs were correlated with progression-free survival (Fig. 5A). Multivariate COX regression analysis revealed that high HLF expression was an independent protective factor (P < 0.001), whereas high CTC level was an independent risk factor (P = 0.014) (Fig. 5B). Patients exhibiting high HLF expression demonstrated a median survival time of 55.2 months, which was markedly longer than the 28.7 months observed in the low HLF group (Fig. 5C). Conversely, the median survival time for patients in CTCs high group was notably reduced, to 35.4 months (Fig. 5D). When low HLF expression and high CTCs coexisted (2 points), the patients had the shortest median survival, 20.2 months. When HLF low or CTCs high appeared alone (1 point), the median survival was 55.2 months, while HLF high and CTCs low presented together (0 point), the median survival exceeded 60 months (Fig. 5E). Individuals with high CTC counts were more likely to develop metastases than those with low CTC (Fig. 5F). In contrast, patients with high HLF levels showed a lower metastasis rate than those with low HLF group (Fig. 5G).

In summary, the expression of HLF and the number of CTCs are the main factors affecting the PFS and survival time of patients with LUAD, regardless of the presence or absence of EGFR mutations, and they are the main indicators of LUAD prognosis.