Predictive value of direct bilirubin and total bile acid in lung adenocarcinoma patients treated with EGFR-TKIs

Abstract

Epidermal growth factor receptor (EGFR)-tyrosine kinase inhibitors (TKIs) have been the standard treatment for patients with sensitizing EGFR mutation. However, almost all patients eventually acquire resistance to EGFR-TKIs. Therefore, easily available parameters to estimate the outcome of lung adenocarcinoma patients treated with EGFR-TKIs are in urgent need. Lung adenocarcinoma patients harbored EGFR sensitive mutant and received EGFR-TKIs as first-line or second-line treatment were recruited in the study. X-tile software were utilized to determine the optimal cut-off value of Alkaline phosphatase (ALP), direct bilirubin (DB), total bile acid (TBA), and high-density lipoprotein-cholesterol (HDL-C). The prognostic value of ALP, DB, TBA, and HDL-C for Progression-free survival (PFS) in patients were evaluated by the Kaplan-Meier curve. We applied univariate and multivariate survival analysis to identify the independent predictor for PFS in patients with EGFR-mutant advanced lung adenocarcinoma and received EGFR-TKIs. A total of 131 lung adenocarcinoma patients with a median age of 58 years old were included in the final analysis. Patients with elevated level of DB and HDL-C showed a longer PFS, while high level of ALP and TBA indicated shorter PFS in response to EGFR-TKI treatment. The multivariate survival analyses revealed a significant association of prolonged PFS with increased DB, and decreased TBA. In conclusion, these findings suggest that DB and TBA were significant independent predictors of PFS in EGFR-TKI-treated patients with advanced lung adenocarcinoma.

Background

Lung cancer, one of the most prevalent malignant diseases worldwide, is the leading cause of death among men and the second leading cause of death among women [1]. Among all cases of lung cancer, around 85% was non-small cell lung cancer (NSCLC) [2]. Adenocarcinoma is the major histologic subtype. About 50% of East Asian patients with lung adenocarcinoma have been reported with activating epidermal growth factor receptor (EGFR) mutation [3]. Treatment with EGFR-tyrosine kinase inhibitors (TKIs) prominently prolongs progression-free survival (PFS) compared with chemotherapy [4] in patients with EGFR-mutant advanced lung adenocarcinoma. Based on these evidence, EGFR-TKIs are recommended to be the first-line treatment for lung patients who harbor TKI-sensitizing EGFR mutation [5]. Despite 80% of patients treated by EGFR-TKIs achieved remarkable disease control, there is a great distinction in the duration of response [6]. It is urgent to seek prognostic factors that could identify patients with a worse outcome, and develop new treatment strategies to improve prognosis.

Biochemical parameters have been identified as valuable predictive factors in patients with solid tumors. For example, a cohort study of 5207 Danish women demonstrated an inverse relationship between high-density lipoprotein-cholesterol (HDL-C) and breast cancer [7]. A similar result was identified in another prospective investigation, showing the correlation of high risk in colorectal cancer patients with blood lipids [8]. Lv et al.. found that high-level HDL-C was an independent biomarker for patients harboring activating EGFR mutation [9]. Another separate study demonstrated that elevated HDL-C, lower direct bilirubin (DB), and higher total cholesterol (TC) were correlated with the risk of death in NSCLC patients with EGFR mutation [10]. Intriguingly, DB has been confirmed to be a protective factor for lung cancer patients in some investigations [11, 12], due to its antioxidative role in tumorigenesis [13]. It is meaningful to re-investigate the prognostic role of DB in forecasting the efficacy to the treatment of EGFR-TKIs in lung adenocarcinoma patients with EGFR-activating mutation.

Bile acids play a key role in the modulation of epithelial cell proliferation, gene expression, and regulation of cholesterol and lipids [14]. Whereas, the relationship between bile acids and prognosis of advanced lung patients administrated EGFR-TKIs have not been explored. Alkaline phosphatase (ALP) is a homomeric enzyme that expresses in almost all tissues or organs. Elevated serum ALP level generally reflects damage of liver, kidney, or bone [15]. The increased ALP level has been also detected in several kinds of cancer patients [16,17,18]. However, the association of ALP and total bile acid (TBA) with the survival of NSCLC patients with EGFR mutation remains unclear.

Therefore, we conducted a retrospective study to evaluate the correlation between ALP, DB, TBA, and HDL-C level before treatment of EGFR TKIs with PFS in lung adenocarcinoma patients with EGFR mutation.

Table 1 Distribution of basic characteristics according to ALP, DB and TBA group in lung adenocarcinoma patients treated with EGFR-TKIs

Patients & methods

Data collection

A total of 131 patients diagnosed with non-small-cell lung cancer in Wuhan Union Hospital between 2014 and 2018 were enrolled in the present study. All these patients were identified with mutations of the EGFR gene and treated with EGFR-TKIs. All patients were eligible for inclusion criteria in the study: (1) patients > 18 years of age; (2) histological diagnosis of pulmonary adenocarcinoma; (3) Unresectable stage according to the 8th edition of the American Joint Committee of Cancer staging system or R1 resection; (4) harboring sensitizing EGFR mutation and treated with EGFR-TKI as the first- or second-line therapy. Ethics approval was obtained from the Ethics Committee of Wuhan Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China. A consent form was not applicable to our study for minimal ethical implications and issues was involved in this study since this is a retrospective study. A specific serial number for each patient was applied to protect patient identity and confidentiality.

Detailed clinical and pathological information, including patients’ age, gender, smoking status, pathological data, treatment regimens, and follow-up outcomes were collected from medical records. Laboratory test, including DB, TBA, ALP, AND HDL-C before treatment were obtained through hospital-based laboratory service.

Fig. 1

The optimal cut-off points of ALP, DB, TBA, and HDL-C. The optimal cut-off points of ALP, DB, TBA, and HDL-C were determined by the X-tile model. X-tile plots for PFS were shown in the left panels, Optimal cut-off values highlighted by the black circles in the left panels were shown in histograms in the right panels. (panel a) ALP, (panel b) DB, (panel c) TBA, and (panel d) HDL-C

Follow-up assessment

Follow-up was performed through a review of the electronic medical records by trained doctors. The endpoint of this study was PFS, which was defined as the time interval from treatment of EGFR-TKIs to progression or the last follow-up.

Table 2 Relationship of ALP, DB, TBA, and HDL-C with mPFS in lung adenocarcinoma patients treated with EGFR-TKIs

Fig. 2

The Kaplan–Meier curves for the PFS. The Kaplan–Meier curves for the PFS among lung adenocarcinoma patients treated with EGFR-TKIs according to pretreatment ALP, DB, TBA, and HDL-C. (panel a) ALP, (panel b) DB, (panel c) TBA, and (panel d) HDL-C

Statistical analysis

X-tile Software combined with patients’ PFS was used to identify the optimal cut-off point of ALP, DB, TBA, and HDL-C. SPSS statistical software (SPSS 23.0, New York, USA) was applied for the rest of analysis. The correlations between variables were performed using the Chi-squared or Fisher’s exact tests. We utilized the Kaplan-Meier (K-M) method to generate the survival curves of PFS, and compare the differences between groups. The univariate and multivariate survival analysis were employed to confirm the independent prognostic factors of PFS, using cox regression models. A two-tailed p value of < 0.05 was considered significant difference.

Fig. 3

The Kaplan–Meier survival of PFS in patients without liver metastasis. (Panel a) ALP, (panel b) DB, (panel c) TBA, (panel d) HDL-C

Fig. 4

The Kaplan–Meier survival of PFS in patients without bone metastasis. (Panel a) ALP, (panel b) DB, (panel c) TBA, (panel d) HDL-C

Results

Participants

A total of 131 eligible pulmonary adenocarcinoma patients with sensitizing EGFR mutations were recruited in the study (Table 1). Among these patients, 101 (77.1%) patients were treated with EGFR-TKIs as the first-line treatment. The median age was 58 years old, ranging from 33 to 73 years old. Females accounted for 64.1% and never smokers for 77.1% patients. 107 (81.7%) patients were with Eastern Cooperative Oncology Group performance status (ECOG PS) 0 or 1 and 24 patients with ECOG PS 2 or 3. 31 (23.7%) patients were diagnosed with bone metastasis and 7 (5.3%) patients were diagnosed with liver metastasis. EGFR Exon 19 deletion (51.1%) and 21 L858R (42.7%) comprised the most common mutation types. Most patients (78.6%) received treatment with gefitinib or icotinib. Only 9 (6.9%) patients received treatment for hyperlipidemia before EGFR-TKI treatment. The median follow-up time was 11.0 (range 1–41) months. 95 (72.5%) patients had progressed and 36 patients had still not progressed at last visit. 36 (72.5%) patients received concurrent chemotherapy combined with EGFR-TKIs, while 95 (72.5%) patients only received EGFR-TKI treatment (Table 1).

Table 3 Univariate and multivariate analysis for progression-free survival in lung adenocarcinoma patients treated with EGFR-TKIs

Correlation of ALP, DB, TBA, and HDL-C with clinical characters

The optimal cut-off threshold of ALP, DB, TBA, and HDL-C were confirmed by X-tile software, taking PFS into account. The appropriate cut-off points of ALP, DB, TBA, and HDL-C were considered 124 U/L, 3.3 µmol/L, 4.0 µmol/L, and 1.3 mmol/L, respectively (Fig. 1). Among the 131 patients, there were 31 patients with an HDL-C level > 124 U/L, 51 patients with a DB level > 3.3 µmol/L, 85 patients with a TBA level > 4.0 µmol/L, 45 patients with an HDL-C level > 1.3 mmol/L. The distribution of ALP, DB, TBA, and HDL-C before treatment of EGFR-TKI is shown in Table 1. Evidently decreased HDL-C was identified in male patients (p = 0.030). The level of DB was significantly associated with the clinical feature of smoker (p = 0.032). No significant differences were detected between ALP, TBA level and variables such as gender, smoking status, chemotherapy, and treatment lines.

K-M survival analysis of ALP, DB, TBA, and HDL-C for PFS

The K-M curve indicated that patients with low-level ALP (HR = 1.214; 95% CI: 1.076–1.370, p = 0.002), high-level DB (HR = 0.567; 95% CI: 0.360–0.895, p = 0.015), low-level TBA (HR = 1.824; 95% CI: 1.170–2.843, p = 0.008), and high-level HDL-C (HR = 0.578; 95% CI: 0.368–0.907, p = 0.017) had better PFS (Fig. 2). The estimated median PFS was 12.00 (95% CI: 10.694–13.306) months (Table 2). The median PFS was 14.00 (95% CI: 10.920–17.080) months for patients with low ALP levels compared with 12.00 (95% CI: 10.694–13.306) months for patients with high ALP level. For patients with higher DB levels, the median PFS was 19.00 (95% CI: 8.620–29.380) months and 12.00 (95% CI: 10.675–13.325) months for patients with low DB level. Patients with low TBA levels showed better survival with a median PFS 18.00 (95% CI: 14.230–21.770) months compared with median PFS 12.00 (95% CI: 10.788–13.212) months for patients with high TBA level. As for HDL-C, patients with a higher level had a longer PFS than those with a lower level (p = 0.013) (Table 2).

Subgroup analysis

We also conducted subgroup analysis to investigate the relationship of ALP, DB, TBA, and HDL-C with survival in patients without bone or liver metastasis respectively (Figs. 3 and 4). In the subgroup without liver metastasis, patients with high ALP level (HR = 1.221; 95% CI: 1.077–1.385, p = 0.002) or TBA level (HR = 1.841; 95% CI: 1.163–2.913, p = 0.009) showed remarkably worse PFS. Whereas high level DB (HR = 0.554; 95% CI: 0.343–0.895, p = 0.016) and HDL-C (HR = 0.550; 95% CI: 0.343–0.882, p = 0.009) was significantly correlated with better PFS. Similar results were observed in patients without bone metastasis.

Survival and prognostic value of ALP, DB, TBA, and HDL-C

To further estimate the prognostic value of ALP, DB, TBA, and HDL-C, we performed univariate and multivariate analyses (Table 3). Univariate analysis showed that combination with common EGFR mutation (Exon 19 deletion or L858R) (HR = 3.168 95% CI: 1.343–7.471, p = 0.008), chemotherapy (HR = 0.509; 95% CI: 0.308–0.843, p = 0.009), high level of HDL-C (HR = 0.578; 95% CI: 0.368–0.907, p = 0.017), low level of ALP (HR = 1.214; 95% CI: 1.076–1.370, p = 0.002), high level of DB (HR = 0.567; 95% CI: 0.360–0.895, p = 0.015), and low level of TBA (HR = 1.824; 95% CI: 1.170–2.843, p = 0.008) were related with longer PFS. The multivariate survival revealed that longer PFS was associated with increased DB (HR = 0.569; 95% CI: 0.334–0.968, p = 0.038), and decreased TBA (HR = 1.769; 95% CI: 1.059–2.954, p = 0.029). While LDL-C, and ALP were not significantly associated with PFS. Chemotherapy and first line therapy were also independent prognostic factors. These results suggest that DB, and TBA were significant independent predictors of PFS in EGFR-TKI-treated patients with advanced lung adenocarcinoma.

Discussion

In recent years, some blood and serum parameters have showed prognostic value in lung cancer patients. The parameters are easily detected and validated in clinical practice, making them promising prognostic factors. To our best knowledge, this is the first study focusing on the relationship between ALP, TBA and PFS in advances lung adenocarcinoma patients treated with EGFR-TKIs. We showed that high-level DB, and low-level TBA were favorable predictors for PFS in NSCLC patients.

Previous studies have shown that a pretreated higher level of HDL-C reduced the risk of death compared with a lower level of HDL-C in NSCLC with EGFR mutations [9, 10]. In accordance with the previous studies, we found HDL-C before EGFR-TKIs treatment was inversely related to PFS in lung adenocarcinoma patients by univariate analysis in this study. ALP has been implicated to participate in inflammation by the regulation of purinergic signaling [19]. Inflammation is closely linked with cancer for the inflammatory disease, increasing the risk of cancer development [20]. Emerging evidence have shown that ALP is an independent prognosis factor of survival in patients with cancers, such as breast cancer [16], prostate cancer [17], and nasopharyngeal carcinoma [21]. In our study, we showed that lung adenocarcinoma patients with an elevated ALP had shorter PFS. In consistent with our results, Pujol et al. showed that small cell lung cancer (SCLC) patients with serum ALP levels higher than 125 U/l have a worse survival [22]. Nevertheless, multivariate analysis showed HDL-C (p = 0.621) and ALP (p = 0.091) was not an independent indicator for PFS in lung adenocarcinoma patient receiving EGFR-TKIs, perhaps due to the limited number of patients included in the present study.

DB is an endogenous antioxidant compound that could scavenge free radicles and protect cells [13]. Oxygen radicals can attack and damage DNA, which is important in the etiology of many human cancers [23]. It has been suggested that an increased concentration of serum bilirubin decreased the risk of cancer mortality [24]. Interestingly, previous studies also revealed controversial results. In a prospective cohort study, the risks of lung cancer incidence increased in male smokers with a decrease of bilirubin [25]. Inconsistent with the result, a study with 10-year follow-up showed that increased concentration of bilirubin decreased the risk for cancer mortality in patients with curatively resected NSCLC [11]. Additionally, Song et al. demonstrated that the elevated level of DB is associated with favorable prognosis in NSCLC [12]. Our results were in agreement with these previous studies. The protective effect of bilirubin may be related to the antioxidant and anti-tumor effect of bilirubin in cancer patients. On the contrary, Zhang’s study suggested that high-level direct bilirubin is correlated with a higher percentage of lymph node metastasis and lymphovascular invasion in stages II and III colorectal cancer after radical resection [26]. Thus, elevated concentration of bilirubin is an unfavorable predictor for OS and DFS in patients with colorectal cancer [27]. Several studies conducted in rectal cancer patients and NSCLC patients [10] showed similar conclusions. As it has been demonstrated that pretreated DB level is inversely correlated with overall survival in NSCLC patients harbored EGFR mutation [10], we also investigated the relationship of DB and PFS in this study. Whereas we showed that the patients with an elevated level of DB had longer median PFS compared with the other patients. The discrepancy in the prognostic role of DB may due to the dual effect of antioxidants. The disparity of optimal cut-off of bilirubin levels perhaps also contribute to the contrary conclusion. The optimal cut-off point in our study was 3.3 µmol/L, while the previous study chooses 6.1 µmol/L as the cut-off point.

In our study, as far as we know, we investigated the prognostic effect of TBA in lung adenocarcinoma patients receiving EGFR-TKIs for the first time. We found that an elevated level of TBA was related to shorter PFS. Additionally, we identified serum TBA as an independent predictor of PFS in lung adenocarcinoma patients. Bile acids are detergents synthesized from cholesterol and secreted into bile by the liver and act as signaling molecules by interacting with the receptors [14]. Bile acids can activate plasma membrane G protein-coupled receptor, TGR5 by interacting with membrane receptor and regulate various cellular responses [28]. Previous studies suggested that bile acids activated TGR5 signaling pathway in many types of solid tumors, including ampullary adenocarcinoma [29], endometrial cancer [30], pancreatic cancer [31], and NSCLC [28]. Aberrantly expression of TGR5 was observed in NSCLC [28]. TRG5 was found positively correlated with an advanced clinical stage in NSCLC patients, indicating that TRG5 played a pivotal role in the course of NSCLC [28]. Besides, patients with higher level of serum deoxycholic acid (DCA), one kind of the bile acid were correlated with stronger TGR5 expression [28]. While elevated TGR5 was related to worse prognosis in NSCLC patients, indicating bile acid may influence the prognosis of cancer patients in a TGR5-dependent manner [28]. Thus, we speculated that elevated TBA triggered TGR5 signaling pathway in lung adenocarcinoma patients in our study, leading to the progression of malignancy and shorter survival.

There were some limitations in our study. First, this study was a retrospective study, which had selection bias and confounders. Second, the serum level of DB, ALP, TBA, HDL-C could be influenced by many factors. For instance, DB elevation may reflect hepatic metastasis or hepatitis. For the DB level in serum, we could not provide an upper limit value for analysis. Third, we only evaluated lung adenocarcinoma patients harbored EGFR mutation and receiving EGFR-TKIs treatment, which limits the application of the results.

Conclusions

Conclusively, our study showed that DB and TBA levels before EGFR-TKIs treatment were prognostic factors in patients with lung adenocarcinoma. Our data highlighted the value of serum biochemical markers in forecasting PFS for lung adenocarcinoma patients receiving EGFR-TKIs. Further large-scale prospective studies are needed to validate our findings.