Clinical consequence of hypophosphatemia during antiviral therapy for chronic hepatitis B
Article information
Abstract
Background
Antiviral therapy is an essential treatment for chronic hepatitis B (CHB) infection. Although hypophosphatemia is an important adverse effect of antiviral agents, its clinical significance remains unclear. We investigated the incidence and clinical consequences of hypophosphatemia in a large cohort of CHB patients.
Methods
This retrospective cohort study included CHB patients who started antiviral therapy between 2005 and 2015 and continued it for at least 1 year. Patients with decompensated liver cirrhosis, diabetes mellitus, hypertension, concomitant diuretic administration, and end-stage renal disease were excluded. The primary outcome was a change in renal function. Secondary outcomes included the incidence of infection and changes in serum potassium, uric acid, and total carbon dioxide (tCO2).
Results
Among the 4,335 patients, hypophosphatemia developed in 75 (1.7%). During the median 2-year follow-up period, patients with hypophosphatemia showed a lower estimated glomerular filtration rate than those in the control group. The incidence of infection and changes in serum potassium, uric acid, and tCO2 were similar between groups.
Conclusion
Hypophosphatemia was associated with a renal function decline in patients with CHB receiving antiviral therapy.
Introduction
Chronic hepatitis B (CHB) is an important liver disease and a major public health concern. According to the World Health Organization, the estimated global prevalence of hepatitis B (HBV) carriers was 296 million in 2019, and 820,000 deaths globally per year were caused by HBV-related liver diseases [1]. Antiviral therapy that slows down and prevents the progression of CHB to cirrhosis and hepatocellular carcinoma is essential for patients with CHB.
The nephrotoxicity of antiviral agents was reported in previous studies [2–5]. The nephrotoxicity of antiviral agents is known to be mediated by various mechanisms such as transporter defects or apoptosis of the renal tubular epithelium, vascular injury, and mitochondrial injury. Fanconi syndrome, characterized by hypouricemia, hypokalemia, and metabolic acidosis, is caused by antiviral agent-induced dysfunction of the proximal tubule [6,7]. Proximal tubular damage induced by antiviral agents might be related to hypophosphatemia because 75% to 85% of the filtered phosphate is reabsorbed in the renal proximal convoluted tubules [8]. However, the precise clinical association between antiviral agent-induced hypophosphatemia and renal function decline or Fanconi feature tubulopathy has yet to be reported.
Serum phosphate is a fundamental element related to skeletal development, membrane composition, nucleotide structure, and cellular signaling [9]. Hypophosphatemia may affect the hematopoietic system by decreasing the intracellular adenosine triphosphate and suppressing phagocytosis and chemotaxis of granulocytes [10]. Nevertheless, the clinical significance of antiviral agent-induced hypophosphatemia in real-world practice remains unclear.
This study aimed to investigate the incidence and clinical consequences of hypophosphatemia in a large cohort of CHB patients without well-known risk factors for renal insufficiency.
Methods
Study population
This was a single-center retrospective cohort study. The data was collected from the medical records of 16,829 CHB patients who received antiviral therapy for at least 1 year from 2005 to 2015 at the Samsung Medical Center, a 2,157-bed tertiary hospital in Seoul, South Korea (Fig. 1). All patients were over 18 years of age. Patients with diabetes mellitus (DM), hypertension, end-stage renal disease, decompensated liver cirrhosis (defined as the presence of ascites, variceal bleeding, hepatic encephalopathy, hepatorenal syndrome, hepatopulmonary syndrome, portopulmonary hypertension, and cirrhotic cardiomyopathy), and concomitant diuretic administration were excluded to minimize the confounding effects of these well-known risk factors for renal insufficiency [11]. The antiviral agents used included lamivudine, entecavir, adefovir, tenofovir, telbivudine, and clevudine.
The study protocol was in compliance with the tenets of the Declaration of Helsinki and approved by the Institutional Review Board of Samsung Medical Center (No. 2019-11-055-002), which waived the need for informed consent owing to its retrospective nature.
Variables
After starting the antiviral agents, the patients were followed up for 1 to 5 years. Demographic variables (age, sex, and race) and laboratory variables, including serum phosphate, potassium, creatinine, uric acid, and total carbon dioxide (tCO2) levels, were collected. All serum phosphate levels from the initiation of antiviral agents (baseline) to the last follow-up were serially assessed. Hypophosphatemia was defined as a median serum phosphate level ≤2.5 mg/dL during the follow-up period. To compare the nutritional status of the patients, their body mass index (BMI) was calculated. Malnutrition was defined as a BMI less than 18.5 kg/m2.
Outcomes
The primary outcome was changes in the renal function as evaluated by the estimated glomerular filtration rate (eGFR) and calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation and the Modified Diet in Renal Disease (MDRD) equation. The secondary outcomes were changes in serum potassium, uric acid, and tCO2 levels. To determine the effect of hypophosphatemia on infection, the incidence of infections requiring antibiotics was analyzed.
Statistical analyses
Continuous measures are expressed as mean ± standard deviation, while skewed data are expressed as median with interquartile range (IQR). Categorical variables are expressed as proportional numbers, and the chi-square test was used to assess the differences between them. The Wilcoxon signed-rank and Kruskal-Wallis tests were performed to examine non-normally distributed categorical variables.
Because unbalanced, repeated-measured data was collected, the linear mixed model was used to determine the relationship between the serum phosphate levels and changes in eGFR. Serum phosphate level, age, and baseline serum creatinine level were included as fixed effects. To test for the significance of each fixed effect specified in the model, type III tests were performed. The significance was set at p < 0.05. All statistical analyses were performed using IBM SPSS version 27.0 (IBM Corp.), SAS version 9.4 (SAS Institute), and R version 3.4.4 (R Foundation for Statistical Computing).
Results
Baseline characteristics
A total of 4,335 patients were included in the analysis. The mean follow-up period was 6.7 years. The median age was 51 years (IQR, 42–59 years); 71.1% of the patients were men. The prevalence of hepatocellular carcinoma was 65.4% (2,835 of 4,335 patients). The median baseline serum creatinine level was 0.8 mg/dL (IQR, 0.7–0.9 mg/dL). The median baseline eGFR was 97 mL/min/1.73 m2 (IQR, 87.7–105.3 mL/min/1.73 m2), estimated by the CKD-EPI equation, and 90.3 mL/min/1.73 m2 (IQR, 80.5–101.2 mL/min/1.73 m2), estimated by the modified MDRD equation. The mean BMI of patients was 23.85 kg/m2.
Hypophosphatemia developed in 1.7% (75 of 4,335) of all patients. Those in the hypophosphatemia group were older and had higher baseline serum creatinine levels than those in the control group (Table 1). The percentages of patients with a BMI less than 18.5 kg/m2 were 2.7% (2 of 75 patients) in the hypophosphatemia group and 1.7% (72 of 4,260 patients) in the control group, there were no significant differences (p = 0.37).
Changes in renal function according to serum phosphate levels
There was a significant intergroup difference in renal function for up to 24 months. The mean CKD-EPI eGFR at 6 months after antiviral agent initiation was 94.8 mL/min/1.73 m2 and 86.5 mL/min/1.73 m2 in the control and hypophosphatemia groups, respectively. At 24 months after antiviral agent initiation, the median eGFR was 89.9 mL/min/1.73 m2 and 79.6 mL/min/1.73 m2 in the control and hypophosphatemia groups, respectively. Renal function was significantly decreased in the hypophosphatemia group at all time points during the follow-up period (p < 0.05) (Fig. 2). To compare the changes in eGFR over time between the groups, a linear mixed model was used. As shown in Table 2, statistical analysis using a linear mixed model revealed a significant independent effect of serum phosphate (group) on eGFR, such that the hypophosphatemia group had a lower eGFR (p < 0.05). There were no significant differences in the degree of eGFR change between the groups over time (p = 0.57), indicating no interaction between the main characteristics of the groups. Although the patients in the hypophosphatemia group patients were older and had a lower baseline eGFR, as evidenced by the significantly higher serum creatinine levels and age at the initiation of antiviral agents (p < 0.001), this disparity did not increase over the 2 years of follow-up (Table 2). After adjusting for serum baseline creatinine level, time, and age, the hypophosphatemia group had a lower eGFR than the control group in all the follow-up periods (Fig. 3).

Changes in renal function.
Serial changes in estimated glomerular filtration rate (eGFR; mean ± standard deviation) were calculated using the CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration) equation according to the following: development of hypophosphatemia after antiviral agent initiation.
*p < 0.05 compared with the control group.

Results of the linear mixed model for the eGFR, estimated by CKD-EPI according to the median phosphate levels

Changes in renal function analyzed with the linear mixed model.
Serial changes in estimated glomerular filtration rate (eGFR) were calculated by the CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration) equation according to the following: development of hypophosphatemia after antiviral agent initiation.
*p < 0.05 compared with the control group. p-value was analyzed by type III test.
Subgroup analysis according to the antiviral agent
Subgroup analyses were performed using a linear mixed model according to the antiviral agents (Table 3). There were no significant independent effects in the lamivudine, entecavir, and adefovir groups. However, the tenofovir group showed a significant independent effect on eGFR (p = 0.04).
Comparison of changes in serum potassium, uric acid, and tCO2 levels
For the 2 years of follow-up, no significant differences in serum potassium, uric acid, and tCO2 levels were noted between the control and hypophosphatemia groups (Fig. 4).

Serial changes in serum potassium, uric acid, and total carbon dioxide levels according to hypophosphatemia status.
(A) Serum potassium (mean ± standard deviation). (B) Serum uric acid (mean ± standard deviation). (C) Serum total carbon dioxide (mean ± standard deviation).
*p < 0.05 compared with the control group.
Incidence of infection
During the entire follow-up period, antibiotics were prescribed to 64.7% (2,756 of 4,260) of the patients from the control group and 62.7% (47 of 75) of the patients from the hypophosphatemia group. The incidence of infection requiring antibiotics was comparable between the control and hypophosphatemia groups (p = 0.72).
Discussion
Here, we demonstrated that hypophosphatemia was significantly associated with renal function impairment in patients with CHB receiving antiviral therapy. Furthermore, our study showed that hypophosphatemia during antiviral therapy is not related to electrolyte disturbances, including hypokalemia or hypouricemia. The incidence of infection was also comparable between groups regardless of the hypophosphatemia status.
The development of hypophosphatemia during antiviral therapy in CHB patients has been reported in several studies. Shimizu et al. [12] reported six cases of hypophosphatemia among 17 patients with CHB treated with adefovir. Persistent hypophosphatemia developed in 14 of 292 patients receiving adefovir and lamivudine for CHB [13]. The incidence of hypophosphatemia was reportedly 25.4% (18 of 71) among patients who received tenofovir [14]. The incidence of hypophosphatemia in previous studies was significantly higher than that in our cohort. These discrepancies in the incidence of hypophosphatemia seem attributable to the small sample size, study population with comorbidities related to hypophosphatemia, short-term follow-up, or cross-sectional study design of previous studies. Our study thoroughly analyzed a larger number of CHB patients without potential confounding comorbidities affecting their serum phosphate levels or renal function and demonstrated a relatively low incidence of hypophosphatemia induced by antiviral therapy.
Although the actual incidence of hypophosphatemia was low in our study, its development was associated with a decline in renal function. Several previous studies reported hypophosphatemia and renal function decline with antiviral agent treatment among patients with CHB [12,13,15,16]. A decline in eGFR of more than 30% was reported in patients with hypophosphatemia among 292 CHB patients treated with adefovir and lamivudine [13]. In our study, serum phosphate level was a useful surrogate marker for predicting changes in renal function with correction for the effects of time and group using a linear mixed model. Essig et al. [17] suggested the use of serum phosphate levels as a sensitive marker for predicting kidney injury in patients with human immunodeficiency virus. As shown in our results, serum phosphate level may also be a surrogate marker for predicting renal outcomes in CHB patients treated with antiviral agents.
This study excluded patients with comorbid diseases, such as DM and hypertension, that could affect renal function or electrolytes. DM and hypertension are well-known risk factors for the development and progression of CKD [18–21]. In CHB patients receiving antiviral agents, age, hypertension, DM, liver or kidney transplantation, underlying CKD, and diuretic use are reportedly risk factors for renal function decline [22]. In clinical practice, frequent or careful monitoring of renal function is usually performed in patients with these risk factors. Renal function monitoring can be neglected in patients with CHB without CKD risk factors. Our study findings suggest that a declining serum phosphate level, even in the absence of known risk factors, may be an early sign of renal dysfunction that requires more careful monitoring of renal function and renoprotective management.
Phosphate plays an important role in cell structure and signaling [9]. Hypophosphatemia is known to impair hematopoietic cell function and subsequently increase infection susceptibility, especially by suppressing the phagocytosis of white blood cells [10]. In our study, the incidence of significant infections requiring antibiotic treatment was comparable between groups regardless of hypophosphatemia status. These results suggest that hypophosphatemia, induced by antiviral agents in patients with CHB, may not be serious enough to significantly suppress immune cells.
Antiviral drug-related proximal tubular damage can cause Fanconi syndrome, which is characterized by hypokalemia, glycosuria, hypouricemia, and hyperchloremic metabolic acidosis [2]. In previous studies, adefovir and tenofovir reportedly induced proximal tubular defects [23–26]. However, in our study, the prevalence of hypokalemia, hypouricemia, and metabolic acidosis was comparable among groups divided by serum phosphate level. These results suggest that the main mechanism of antiviral agent-induced hypophosphatemia in CHB patients without risk factors for CKD is the isolated dysfunction of a phosphate transporter rather than generalized proximal tubular dysfunction [8].
This study has some limitations. First, some baseline characteristics differed between the groups because of the retrospective cohort design. To overcome this limitation, we used a linear mixed model. As the data used in this study were unbalanced and repeatedly measured, the variables were modeled as fixed effects. We analyzed the effect of each variable on eGFR through this model and confirmed that the group depending on serum phosphate levels showed a significant difference in renal outcome over time. Second, this was a retrospective study, and only 2.1% (90 of 4,335) of all patients were tested for the urine spot protein-to-creatinine ratio. In this study, significant renal decline occurred over time when hypophosphatemia developed. Since this study confirmed only a decrease in eGFR, additional studies should be considered to determine whether proteinuria occurs significantly in the hypophosphatemia group. Third, detailed nutritional status such as vitamin D deficiency was not included in the analyses. Phosphate homeostasis is regulated by dietary intake, gastrointestinal absorption, bone resorption, and renal excretion [9]. Although we could not include dietary factors related to phosphate balance in the analysis, we believe that the overall nutritional status of our cohort would exert little or a negligible effect on the results because we excluded patients with comorbidities such as decompensated liver cirrhosis, DM, and hypertension. In addition, the number of patients with malnutrition, defined as a BMI less than 18.5 kg/m2 was relatively small in this study [27]. Also, we used the median value of multiple serum phosphate levels during the follow-up period in the analyses to minimize the confounding effects of variability in serum phosphate concentration influenced by diet. Fourth, the interaction between hypophosphatemia and the skeletal system was not estimated because of the limited data available for this single-center retrospective cohort study. Adult chronic hypophosphatemia can lead to impaired mineralization of the bone and subsequent osteomalacia [28,29]. Further prospective studies with long-term follow-up are required to elucidate the clinical relevance of hypophosphatemia in terms of metabolic bone disease in patients with CHB.
In conclusion, hypophosphatemia induced by antiviral agents is significantly associated with a decline in renal function. Although the incidence of hypophosphatemia during antiviral therapy is relatively low, our results support the clinical significance of serum phosphate monitoring and renoprotective management in patients with CHB and hypophosphatemia or declining serum phosphate levels.
Notes
Conflicts of interest
All authors have no conflicts of interest to declare.
Funding
This study was supported by the Korean Health Technology R&D Project (grant number: HC20C0085) through the Korean Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea.
Acknowledgments
The authors thank Ji Hyun Kang of Samsung Medical Center for her dedicated efforts as a clinical research coordinator. We would like to thank Editage (www.editage.co.kr) for English language editing.
Data sharing statement
The data presented in this study are available from the corresponding author upon reasonable request.
Authors’ contributions
Conceptualization, Methodology: MYP, HRJ
Data curation, Formal analysis, Investigation: All authors
Funding acquisition: HRJ
Writing–original draft: MYP, HRJ
Writing–review & editing: All authors
All authors read and approved the final manuscript.