Abstract
Background/Aim: To identify predictors of severe adverse events (≥grade 3) in patients with advanced hepatocellular carcinoma treated with lenvatinib. Patients and Methods: Of 41 patients, 25 and 16 were stratified into the severe and non-severe adverse events groups, respectively. Of these, 19 formed a lactulose–mannitol test subgroup, which was divided into severe adverse events (n=11) and non-severe adverse events (n=8) groups. Severe adverse events were assessed by liver disease etiology and modified albumin–bilirubin grade. Intestinal permeability by lactulose–mannitol test and serum soluble CD163, soluble mannose receptor, and zonulin levels. Results: Severe adverse event incidence rates were higher in patients with advanced hepatocellular carcinoma related to alcoholic liver disease and nonalcoholic fatty-liver disease than in those with advanced hepatocellular carcinoma of other etiologies (p=0.014). The rates were higher for modified albumin–bilirubin grades 2a and 2b compared to modified albumin–bilirubin grade 1 (p=0.0104). Zonulin levels were higher in the severe adverse event group (p=0.0331) and were independently associated with severe adverse events (odds ratio=140, 95% confidence interval=1.66-11800; p=0.029). Patients with high zonulin levels (≥0.518 ng/ml) experienced more severe adverse events than those with low levels (<0.518 ng/ml) (p=0.0137). In the lactulose–mannitol test subgroup, the urine lactulose:mannitol ratio was higher in the severe vs. non-severe adverse event group (p=0.0164). Moreover, it was higher in patients with alcoholic liver disease and nonalcoholic fatty-liver disease-related advanced hepatocellular carcinoma compared to those with other advanced hepatocellular carcinoma etiologies (p=0.0108). Conclusion: Serum zonulin levels predict severe adverse events in patients with advanced hepatocellular carcinoma treated with lenvatinib.
Hepatocellular carcinoma (HCC) is the most frequent fatal cancer worldwide) (1). Six systemic therapies have been approved for advanced HCC (a-HCC) based on Phase III trials: atezolizumab with bevacizumab, sorafenib, lenvatinib, regorafenib, cabozantinib, and ramucirumab (2). A combination of atezolizumab + bevacizumab is the new standard of care for the first-line treatment of a-HCC (3). If disease progression occurs on atezolizumab + bevacizumab, the optimal sequence of therapy is molecularly targeted agents (2, 4). Lenvatinib is an oral multiple tyrosine kinase inhibitor that targets vascular endothelial growth factor receptor (VEGFR)1-3, fibroblast growth factor receptor (FGFR)1-4, RET, KIT, and platelet-derived growth factor receptor (PDGFR). However, severe adverse events (SAEs; grade 3 or higher) frequently occur (5). Maintaining the dose intensity of lenvatinib can be challenging because of treatment-related adverse events (4), which results in possible reduced efficacy of treatment. Therefore, focusing on the management of AEs is crucial to obtain favorable outcomes in patients with a-HCC. Furthermore, over recent decades, cirrhosis related to alcoholic liver disease (ALD) and nonalcoholic fatty-liver disease (NAFLD) has become the chief etiology of HCC, outweighing viral hepatitis infection (6).
Compared with healthy controls, patients with ALD and NAFLD show higher intestinal permeability (IP) due to chronic alcohol abuse and increased endotoxin sensitivity, respectively (7, 8). Increasing evidence of the role of the gut–liver axis has shown that microbiome dysbiosis causes impaired intestinal barrier function and increased IP, which results in increased passage of endotoxins from the intestinal lumen to extraintestinal organs, including the liver (9). The lactulose-mannitol test (LMT) is clinically useful for assessing IP (10). An elevated lactulose-to-mannitol ratio (LMR) is evident in patients with serious toxicity and predicts possible AEs in patients with ovarian and breast cancers who receive intravenous chemotherapy (11). Potentially, LMR is a favorable indicator for assessing mucosal barrier dysfunction in patients with gastrointestinal cancers who receive chemotherapy (12). However, the LMT requires 2 h of fasting after ingesting sugars (in water) before the urine is collected over 6 h (13). Several tight junction proteins including zonulin, claudins and occludins have been suggested as biomarkers of disrupted barrier function in humans (14). The serum levels of macrophage activation markers and soluble CD163 (sCD163) as well as mannose receptor (sMR) have been proven to be correlated with LMR in patients with cirrhosis (15).
Recent evidence shows that reduced skeletal muscle mass is a favorable predictor for lenvatinib toxicity in patients with HCC (16). However, no biomarkers are predictive of AEs in patients with HCC who are treated with systemic therapies. It is becoming increasingly urgent to develop new biomarkers that predict SAEs in lenvatinib treatment and allow the maintenance of relative dose intensity in patients with a-HCC. This study aimed to examine the surrogate biomarkers of lenvatinib-related SAEs in patients with a-HCC.
Patients and Methods
Patients. This retrospective study included 41 patients with a-HCC who received lenvatinib (Lenvima®; Eisai, Tokyo, Japan) between October 2018 and September 2020. Previous research showed that gut microbiota composition affects the incidence of diarrhea in 22 patients with a-HCC who received lenvatinib (17). All participants gave informed consent to participate prior to the study. The primary outcome was a factor that would predict lenvatinib-related SAEs in patients with a-HCC. ALD diagnosis was established based on elevated levels of liver enzymes in patients with a history of excessive alcohol consumption. NAFLD was diagnosed based on clinical and laboratory features in patients with metabolic risk factors, provided that other causes of liver disease were excluded. The study protocol conformed to the ethical guidelines of the Declaration of Helsinki and was approved by the Research Ethics Committee of Nara Medical University (Nara-med, 070-0056; approved on February 13, 2018). All patients provided written informed consent before blood specimens were obtained. Patients with Child–Pugh class C or those with other types of cancer were excluded. The lenvatinib dose was administered based on body weight. The initial dose was 12 mg/day for patients with a body weight of 60 kg or higher and 8 mg/day for those with a body weight of less than 60 kg. For patients with Child–Pugh class B, the lenvatinib dose was 8 mg/day for those with a body weight of ≥60 kg and 4 mg/day for those with a body weight of <60 kg.
Safety assessment. Patients were evaluated every 2-4 weeks during lenvatinib therapy using laboratory tests and physical examinations for safety assessment, including the Eastern Cooperative Oncology Group performance status, vital signs, physical findings, laboratory tests, and urinalysis. AEs were assessed by a medically qualified individual at each visit and graded according to the Common Terminology Criteria for Adverse Events (CTCAE) version 4.0 (18). Patients who developed ≥ grade 3 or unacceptable AEs were subjected to a dose interruption or treatment discontinuation, according to the manufacturer’s instructions, until toxicities recovered to CTCAE Grade 1 or less.
Intestinal permeability markers. For noninvasive evaluation of IP, sCD163 levels were measured using a sCD163 Enzyme-Linked Immuno Sorbent Assay (ELISA) kit (Cusabio Technology, Houston, TX, USA) (19). sMR levels were also measured using a sMR ELISA kit LifeSpan BioSciences, Seattle, WA, USA) (15). Serum zonulin levels were determined using a human zonulin ELISA kit (Elabscience, Wuhan, PR China) (20). The ELISA analyses of sCD163, sMR, and zonulin levels were conducted according to the respective manufacturers’ instructions.
Endotoxin activity measurements. Whole-blood endotoxin activity was measured using the commercially available endotoxin activity assay (EAA) kit (Spectral Diagnostics, Toronto, Canada) (21). Briefly, the EAA measures the ability of lipopolysaccharide–antibody complexes, transported to neutrophils by complement receptors, to augment the production of reactive oxygen species. In the presence of β-glucan and luminol, the neutrophils undergo a respiratory burst and emit light. A chemiluminometer was used to measure the light produced; its intensity increased with increasing concentrations of endotoxin (22). EAA was expressed in relative units (a scale of 0-1) and derived from the integral of the basal level and the enhanced chemiluminescent reaction. EAA was conducted within 24 h after blood sample collection.
Lactulose-mannitol test. After an overnight fast, patients completely emptied their bladders. Subsequently, a urine sample was obtained to check for possible endogenous mannitol production. They then received 5g of mannitol and 10 g of lactulose in 200 ml drinking water followed by 300 ml of drinking water. Patients drank an additional 300 ml of water 3 h after receiving the mannitol and lactulose doses. Further, 19 urinary samples from 19 patients (one sample per patient) who agreed to the lactulose–mannitol test (LMT) were collected for the following 6 h. Urinary estimations of lactulose and mannitol are measured using the EnzyChrom IP Assay Kit (BioAssay Systems, Hayward, CA, USA). The percent absorption of lactulose and mannitol was determined by measuring the amount excreted during the first 6 h after ingestion (23). The degree of IP is reflected by the ratio of the lactulose-to-mannitol absorption percentages. An increase in this ratio indicates increased IP since lactulose is exclusively absorbed though intercellular spaces. The IP is determined by calculating the ratio between the excretions of the ingested doses of lactulose and mannitol in urine (LMR) (24).
Measurement of the skeletal muscle index. Skeletal muscle area was measured from cross-sectional computed tomography (CT) images at the third lumbar vertebra from scans that were performed before treatment initiation. Using the Synapse Vincent software (FUJIFILM Medical, Tokyo, Japan), measurements were calculated as follows: total bilateral psoas muscle area/height2. Skeletal mass index (SMI) derived from CT (SMI-CT) is expressed as cm2/m2.
SMI was also determined as the sum of skeletal muscle mass in the upper and lower extremities, divided by hight2, using data from bioelectrical impedance analysis (25). SMI derived from BIA (SMI-BIA) is expressed as cm2/m2 (26).
Statistical analysis. All statistical analyses were performed using the R software version 4.0.2 (R Foundation for Statistical Computing, Vienna, Austria). Data are expressed as medians with interquartile range between the 25th and 75th percentiles. Categorical variables are presented in a contingency Table I and Table II. The baseline characteristics between groups were compared using the unpaired t-test or the Mann–Whitney U-test. We used the unpaired t-test for continuous data with normal distribution and Mann–Whitney U-test for continuous data without normal distribution. Correlations among continuous variables were evaluated using Spearman’s correlation coefficient. The areas under the receiver operating characteristic (ROC) curves (AUCs) were used to evaluate the diagnostic values of IP markers regarding the correct identification of SAEs. Statistical significance was defined as a two-tailed p-value of <0.05.
Background characteristics of patients with advanced hepatocellular carcinoma (HCC) (n=41).
Baseline characteristics in hepatocellular carcinoma (HCC) patients with and without adverse events (n=19).
Results
Baseline patient characteristics. Table I shows a summary of demographics and baseline characteristics for the entire population. There were 33 men (80.5%); the median age was 77.0 years at the time of lenvatinib initiation (range=70.0-81.5 years). Alcohol consumption was the most common cause of a-HCC (n=14, 34.1%). The other causes were hepatitis C virus (HCV) (n=9, 22.0%), hepatitis B virus (HBV) (n=8, 19.5%), NAFLD (n=8, 19.5%), and other (n=2, 4.9%). Most patients were classified as having a Child–Pugh score (CPS) of 5 (n=29, 70.7%) or 6 (n=9, 22.0%). The remaining patients were classified as CPS 7 (n=1, 2.4%) and CPS 8 (n=2, 4.9%). The m-ALBI grades were as follows: grade 1 in 13 patients, grade 2a in 13 patients, and grade 2b in 15 patients. The median serum alpha-fetoprotein and des-γ-carboxy prothrombin levels were 46.6 ng/ml (range=0.8-26998 ng/ml and 665 mAU/ml (range=11-83134 mAU/ml), respectively. Of the 41 patients who received prior systemic therapy, sorafenib and regorafenib had been administered to 7 patients (17.1%). Patients were divided into two groups: those with AEs of grade 3 or higher (SAE group; n=25) and those with AEs of lower grades [non-SAE group (NSAE); n=16] (Figure 1). The incidence rates of SAEs were significantly higher in patients with ALD- and NAFLD-related a-HCC than in those with a-HCC of other etiologies (p=0.0140) (Table I). Furthermore, the incidence rates were significantly lower for patients with m-ALBI grade 1 than for those with m-ALBI grades 2a and 2b (p=0.0104). Table II summarizes the baseline characteristics of the LMT subgroup, including 19 patients treated with lenvatinib. NAFLD was the most common cause of a-HCC (n=6, 31.6%). The other causes were alcohol (n=3; 15.8%), HCV (n=3; 15.8%), HBV (n=3; 15.8%), and other (n=2; 10.5%). Most patients were classified as having a CPS of 5 (n=14, 73.7%). Four patients presented with a CPS of 6 (21.1%), and one patient presented with a CPS of 7 (5.3%). The m-ALBI grades were as follows: grade 1 in 4 patients, grade 2a in 9 patients, and grade 2b in 6 patients. As for prior systemic therapy, sorafenib and regorafenib were administered in three patients (15.8%). There were no significant differences in characteristics between patients with SAEs and those with NSAEs.
Flow chart of the study design. A prospective study of 41 patients with advanced hepatocellular carcinoma (a-HCC) treated with lenvatinib. Among them, a subgroup of 19 patients treated with lenvatinib underwent a lactulose-mannitol test. Two patients who were initiated at a reduced dose were excluded.
Adverse events. Lenvatinib-related AEs were summarized. The overall incidence of drug-related AEs of any grade was 97.6% (n=40), and of SAEs, 61.0% (n=25) (Table III). The most frequent lenvatinib-related SAEs were loss of appetite (n=5; 12.2%), diarrhea (n=5; 12.2%), peripheral edema (n=4; 9.8%), fatigue (n=4; 9.8%), and hepatic encephalopathy (n=4; 9.8%). Most AEs were controllable. Four patients discontinued lenvatinib due to hepatic encephalopathy. In the LMT subgroup, the incidence of drug-related AEs of any grade was 89.5% (n=1). The overall incidence of SAEs was 57.9% (11 patients) (Table IV). The most frequent lenvatinib-related SAEs were loss of appetite (n=3; 15.8%), diarrhea (n=2; 10.5%), peripheral edema (n=3; 15.8%), hand-foot syndrome (n=1; 5.3%), fatigue (n=1; 5.3%), and hypertension (n=1; 5.3 %).
Number of patients with adverse events (n=41).
Number of patients with adverse events (n=19).
Incidence rate of SAEs according to etiology and m-ALBI. The incidence rates of SAEs were 10/14 (71.4%), 4/9 (44.4%), 3/8 (37.5%), 7/8 (87.5%), and 1/2 (50.0%) in patients with a-HCC caused by ALD, chronic HCV, chronic HBV, NAFLD, and others, respectively (Figure 2A). The incidence rates of SAEs were 5/13 (38.4%), 9/13 (69.2%), and 11/15 (73.3%) in patients with m-ALBI Grades 1, 2a, and 2b, respectively (Figure 2B). No significant differences in the incidence rates of SAEs according to either the etiology of liver disease or the m-ALBI grade were observed (p=0.23 and 0.14, respectively).
Incidence of severe adverse events in 41 patients with advanced hepatocellular carcinoma. (A) Incidence rates of severe adverse events (SAEs) according to etiology. (B) Incidence rates of SAEs according to the modified albumin-bilirubin (m-ALBI) grade.
SAEs with IP markers. We evaluated correlations between SAEs and IP markers in 41 patients with a-HCC who were treated with lenvatinib. Serum zonulin levels were significantly higher in the SAE group than in the NSAE group (p=0.0331) (Figure 3A). In contrast, no significant differences in the serum levels of either sCD163 or sMR were observed between the two groups (p=0.4837 and 0.2683, respectively) (Figure 3B and C).
Comparison of severe adverse events (SAE) with intestinal permeability markers between patients with SAE and non-SAEs (A) Zonulin. (B) Soluble CD163. (C) Soluble mannose receptor.
Factors associated with the development of SAEs. We examined factors associated with the development of SAEs in 41 a-HCC patients who were treated with lenvatinib. Univariate analysis revealed that zonulin levels were independently associated with the development of SAEs (Table V). The ROC analysis showed that zonulin levels were predictive of SAEs (AUC=0.705; 95% confidence interval=0.527-0.882; cut-off value=0.518 ng/ml) (Figure 4). The rate of SAEs was significantly higher in patients whose serum zonulin levels were ≥0.518 ng/mL than in those whose levels were <0.518 ng/ml (p=0.0137).
Univariate analyses of predictive markers for SAEs in patients with hepatocellular carcinoma treated with Lenvatinib.
Correlation between serum zonulin levels and severe adverse events (SAEs) in 41 patients with advanced hepatocellular carcinoma. Receiver operating characteristic curves for predicting SAEs.
LMR and IP markers. The 19 patients upon which the LMT was performed (the LMT subgroup) were also divided into SAE (n=11) and NSAE (n=8) groups. We evaluated the correlation between LMR and IP markers in 19 patients with a-HCC who underwent the LMT. The correlation between serum zonulin levels and LMR tended to be positive (R=0.469, p=0.0578) (Figure 5A). However, no significant correlation was observed between the LMR and CD163 levels (R=0.17, p=0.514) or between LMR and sMR levels (R=0.326, p=0.202) (Figure 5B and C). The LMR and EAA levels were significantly higher in the SAE group than those in the NSAE group (p=0.0164 and 0.033, respectively) (Figure 6A and B). Similar to the incidence rates of SAEs, the LMR was significantly higher in patients with ALD- and NAFLD-related a-HCC than in those with a-HCC of other etiologies (p=0.0108) (Figure 7A). No significant correlation was observed between the LMR and m-ALBI grade (R=−0.00634, p=0.7624) (Figure 7B).
Correlation between intestinal permeability markers and lactulose-to-mannitol ratio in 19 patients with advanced hepatocellular carcinoma. (A) Zonulin, (B) soluble CD163 and (C) soluble mannose receptor.
Comparison of the lactulose-to-mannitol ratio (LMR) and endotoxin activity levels between patients with severe adverse events (SAEs) and non-SAEs. (A) LMR. (B) Endotoxin activity levels.
The association of lactulose-mannitol test (LMT) with etiology of liver disease and with the modified albumin-bilirubin grade. (A) Comparison of LMR between alcoholic liver disease-related and nonalcoholic fatty-liver disease-related hepatocellular carcinoma (HCC) versus HCC of other etiologies. (B) Correlation of LMR with the albumin-bilirubin score.
SAEs and SMI. We assessed the correlations between the incidence rates of SAEs and SMI. No significant differences were found between the occurrence of SAEs and either BIA-SMI or CT-SMI (p=0.5535 and 0.2023, respectively) (Figure 8).
Comparison of skeletal mass index between patients with severe adverse events (SAE) and non-SAEs in 41 patients with advanced hepatocellular carcinoma. (A) Biochemical imbalance analysis. (B) Computed tomography.
Discussion
Insufficient management of AEs may result in poor adherence and treatment discontinuation in patients with HCC who received lenvatinib therapy. Therefore, the management of lenvatinib-related SAEs is essential for increasing relative dose intensity and achieving full therapeutic effect (27). To identify potential biomarkers for the prediction of SAEs, we performed a comparative assessment of lenvatinib-associated AEs. We found that an increased serum zonulin level (an IP marker) can predict SAEs in patients with a-HCC treated with lenvatinib.
Metabolic and immunologic barrier functions of the intestinal mucosa may be closely related to chemotherapy in patients with cancer. Intestinal villous atrophy and increased IP are observed even before starting chemotherapy in patients with advanced breast cancer and acute myeloid leukemia (25, 28). IP is modified by multiple factors, including intestinal epithelial damage, gut microbiota and mucus alterations, and impaired barrier function in intestinal epithelial cells (29). Intestinal drug absorption might be affected by changes in IP. Several reports showed that gastrointestinal toxicity is modified also by several proinflammatory cytokines, including interleukin-11 (30) or interleukin-15 (31). Moreover, Yamamoto et al. reported that microbiome modifications were significantly correlated with hand-foot skin reaction and diarrhea in patients with a-HCC treated with sorafenib (32). Furthermore, increased IP associated with fecal microbiota dysbiosis causes lenvatinib toxicity. These findings support our hypothesis that elevated blood endotoxin levels, as measured by EAA, are a major contributor to SAEs. This warrants future investigation into the impact of endotoxin on the development of SAEs in patients with a-HCC who are treated with lenvatinib.
More recent studies showed high frequency rates of liver function–related AEs (33, 34). However, the incidence of SAEs did not correlate with the m-ALBI grade in our study population, probably because most patients demonstrated relatively well-preserved liver function.
Skeletal muscle depletion has been shown to be related to liver function reserve in patients with liver cirrhosis (35). Decreased SMI has been previously associated with the occurrence of SAEs in patients with a-HCC treated with lenvatinib (16). However, we did not find a correlation between the incidence of SAEs and SMI. A relationship between SMI and the incidence of lenvatinib-related SAEs remains unclear, which might be partially due to differences between studies in the numbers of patients and the operational methods used for evaluating skeletal muscle mass. Our findings are congruent with our hypothesis that the incidence of SAEs is associated with IP, but not with liver function reserve in patients with a-HCC who are treated with lenvatinib.
Zonulin (prehaptoglobin 2) is secreted by intestinal epithelial cells. It regulates IP via protease-activated receptor-2-mediated epidermal growth factor receptor transactivation (36). The serum zonulin level has been reported to be an IP marker in patients with malignant, autoimmune, neurodegenerative, and metabolic diseases (37, 38). Gut microbiota dysbiosis triggers inappropriate production of zonulin, causing functional loss of gut barrier integrity, followed by increased translocation of endotoxin from the gut lumen to the lamina propria (37). In our study, the development of SAEs was correlated with the zonulin level but not with the levels sCD163 or sMR. Conversely, markers of macrophage activation (sCD163 or sMR), but not zonulin, correlated well with LMR in patients with cirrhosis (39). Several studies found no significant correlation between LMR and serum zonulin levels (40, 41). The reason why different IP markers demonstrate different relationships with the incidence of SAEs remains unclear. It might be partly explained by the fact that the three IP markers we measured demonstrated different distributions and variations in the ELISA tests (41, 42). Our findings reiterate the potential role of zonulin in predicting SAEs, suggesting a potential new biomarker that can predict SAEs prior to treatment initiation.
This study has several limitations. First, the number of patients was small. In particular, the number of patients who received the LMT was extremely small because it required a six-hour hospital stay. Second, we did not determine plasma lenvatinib concentrations before and after lenvatinib treatment. Third, we did not evaluate the gut microbiome. Fourth, SAE-associated risk factors should be identified through multivariate logistic regression analysis. Nevertheless, our findings suggest that the gut–liver axis plays a significant role in the development of SAEs in patients with a-HCC who are treated with lenvatinib. Baseline serum zonulin levels can help identify potential SAEs to assist safe management. However, a larger study is required for the validation of surrogate biomarkers for estimating the occurrence and severity of AEs in systemic chemotherapy for HCC, such as with lenvatinib.
Acknowledgements
The Authors would like to thank Enago (www.enago.jp) for English editing and proofreading the article.
Footnotes
Authors’ Contributions
Conceptualization, T.N. and H.T.; methodology, Y. Fujim.; software, K.M.; validation, M.E. and Y.T.; formal analysis, T.I and K.M.; investigation, A.M.; resources, H. Takay.; data curation, Y. Fujin and Y.S.; writing – original draft preparation, S.T.; writing – review and editing, T.N.; visualization, T.A. and H.K.; supervision, K. Kaji.; project administration, K.K.; funding acquisition, N.N. All Authors have read and agreed to the published version of the manuscript.
Conflicts of Interest
The Authors declare no conflicts of interest.
- Received July 9, 2022.
- Revision received July 28, 2022.
- Accepted August 6, 2022.
- Copyright © 2022 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.
References
In this issue
Jump to section
Related Articles
Cited By...
- The Macrophage Activation Marker Soluble CD163 Predicts the Response to Atezolizumab and Bevacizumab in Advanced Hepatocellular Carcinoma
- Clinical Outcomes With Lenvatinib in Patients Previously Treated With Atezolizumab/Bevacizumab for Advanced Hepatocellular Carcinoma
- Efficacy and Safety of Lenvatinib After Progression on First-line Atezolizumab Plus Bevacizumab Treatment in Advanced Hepatocellular Carcinoma Patients