Abstract
Accumulating evidence has shown that serum calcium and vitamin D may be associated with or influence various cancer risks. However, no prospective studies have evaluated the independent and joint associations between prediagnostic levels of serum calcium and vitamin D and future risk of incident primary liver cancer.
We used a nested case–control design to evaluate subjects over 22 years of follow-up. Serum calcium, 25-hydroxy vitamin D [25(OH)D], and three markers of hepatitis B virus and hepatitis C virus were measured in baseline serum from 226 incident primary liver cancer cases and 1,061 matched controls. We calculated ORs and 95% confidence intervals (CI) using logistic regression to estimate the associations between calcium, 25(OH)D, and primary liver cancer risk.
Multivariable adjusted models showed that subjects with both low (ORLow/Medium = 1.48, 95% CI = 1.01–2.17) or high (ORHigh/Medium = 1.92, 95% CI = 1.34–2.76) calcium had an increased primary liver cancer risk, while those with high 25(OH)D had a decreased risk of primary liver cancer (ORHigh/Medium = 0.54, 95% CI = 0.35–0.82). In joint analyses, when compared with subjects with medium calcium and 25(OH)D, subjects with high calcium and medium 25(OH)D had elevated odds of developing primary liver cancer (OR = 1.89, 95% CI = 1.17–3.05); those with medium calcium and high 25(OH)D had reduced odds of developing primary liver cancer (OR = 0.34, 95% CI = 0.17–0.67); and subjects in other classifications of calcium and serum 25(OH)D levels had no change in the odds of developing primary liver cancer (all P > 0.05).
In a nutrient-deficient population, we found that serum calcium and serum 25(OH)D could potentially be modifiable risk or protective factors.
Our findings provide potential targets for primary liver cancer prevention and control.
This article is featured in Highlights of This Issue, p. 1841
Introduction
Primary liver cancer is the third leading cause of cancer death in China, where there are 392,868 new cases and 368,960 primary liver cancer deaths each year (1). Chronic infection with hepatitis B virus (HBV) or hepatitis C virus (HCV) is the predominant cause of primary liver cancer in China (2).
As the most abundant mineral element in the body, calcium is an important component of bone structure and is involved in processes related to various biological processes including proliferation, apoptosis, cell aging, cell signaling, and lethal cell injury by acting as a chemical transmitter or second messenger in cell signal transduction (3, 4). Serum and intracellular calcium concentrations are in a dynamic homeostasis and are necessary for cells and organs to maintain their structure and function (5, 6). The intestinal absorption of calcium is affected by vitamin D, and calcium is metabolically interrelated with vitamin D in several normal physiologic states (7). Serum 25-hydroxyvitamin D [25(OH)D] is currently considered the best marker of vitamin D status in humans (8), and 25(OH)D may work by inhibiting cellular proliferation, cell migration, and angiogenesis, and by inducing apoptosis, which could potentially prevent tumor development (9–11). A previous study by our group showed that serum 25(OH)D was a protective factor for primary liver cancer at higher concentrations, compared with lower concentrations, in some subgroups, but it did not evaluate the independent effects of serum calcium or the joint effects of 25(OH)D and calcium on the risk of primary liver cancer in detail (12).
Abnormal calcium levels and a compromised vitamin D status have been shown to be risk factors for numerous chronic diseases (13), including cancers (14–24), but there have been no prospective studies of serum calcium levels and primary liver cancer risk, and only a few studies of the association between serum 25(OH)D levels and risk of this important malignancy (12, 22, 24). Because the physiologic effects of calcium and vitamin D are so interrelated, it is important to evaluate both their individual and joint effects on primary liver cancer risk in the same study.
The aim of this study was to examine the independent and joint associations between prediagnostic serum calcium, serum 25(OH)D, and the risk of primary liver cancer.
Materials and Methods
Study population and data acquisition
This study was a prospective case–control study nested in the Linxian Nutrition Intervention Trial (NIT) cohorts in China. The populations in these cohorts were deficient in many nutrients (25), and they had extremely high rates of esophageal squamous cell carcinoma and gastric cardia adenocarcinoma (26). The purpose of these trials was to evaluate whether vitamin/mineral supplements in physiologic doses could reduce the rates of these cancers in this nutrient-deficient population (27). Subjects were selected from both the Dysplasia Trial cohort and the General Population Trial cohort. The population and detailed design of the Linxian NITs have been described previously (27–29). Briefly, the Dysplasia Trial enrolled individuals between the ages of 40 and 69 years who were cytologically diagnosed with esophageal dysplasia and lived in three communes located in northern Linxian from August to October of 1984. A total of 3,318 residents were randomized and received either a daily multiple mineral/vitamin supplement (12 minerals and 14 vitamins), including 324 mg calcium and 800 IU vitamin D, or matching placebo for 6 years, from May 1985 to April 1991.
The General Population Trial enrolled individuals between 40 and 69 years old from the general population of four communes in Linxian from March to May of 1985. A total of 29,584 healthy adults were randomized and received one of four daily mineral/vitamin supplement combinations for 5.25 years from March 1986 to May 1991. Calcium and vitamin D were not among the supplemented vitamins in this trial (28). Individuals with cancer, debilitating disease including liver disease, or those who required daily medications were excluded from both trials.
At the baseline exams conducted between August 1984 and May 1985, all subjects were interviewed using a structured questionnaire recording data on age, smoking, alcohol consumption, etc., given a physical examination, and had a 10 mL blood sample drawn before either intervention started. These samples were stored on ice for 3–6 hours during transportation to the field station lab. Then they were centrifuged and aliquoted into 1 mL vials, frozen and stored at −85°C for long-term storage until thawed for the current laboratory measurements. The study was performed according to the guidelines of the Helsinki declaration. All consent procedures, including human specimen collection, were approved by the Institutional Review Boards of the U.S. National Institutes of Health and the Chinese Academy of Medical Sciences (Beijing, China), and all participants provided written informed consent.
Follow-up and identification of outcome events
During the trial (1985–1991) and posttrial follow-up periods (after 1991), follow-up was performed and incident cancer cases were identified by several methods to ensure essentially complete ascertainment of events. Village health workers visited each participant monthly and a panel of Chinese experts confirmed new cancer diagnoses by reviewing medical records from the local hospitals. Most incident primary liver cancers were diagnosed by combined evidence from biochemical assays, clinical examination, ultrasound, and CT scan.
Nested case–control design and subject selection
A total of 255 incident primary liver cancer cases were identified from baseline through the end of 2007. We previously reported the effects of serum 25(OH)D on these same 255 primary liver cancer cases and 310 cases of chronic liver disease death, using somewhat different methods of analysis (12). For both groups of cases, incidence density-matched controls (two frequency-matched controls per case) were selected by age at baseline (±3 years), gender, and trial from among the NIT participants who were alive and free of cancer at the time of case diagnosis. After exclusions for unavailable baseline serum, we included 226 primary liver cancer cases (Dysplasia Trial = 29; General Population Trial = 197) in the current analysis. Because in the previous study, the controls for the primary liver cancer cases and the controls for the chronic liver disease deaths were not significantly different in age, sex, or trial, we used the entire set of 1,061 controls (Dysplasia Trial = 131; General Population Trial = 930) in the current analysis, to increase statistical power.
Laboratory measurements
Each specimen's tube was labeled with a previously assigned unique serial number and all laboratory technicians were blinded to case–control status and identification information. Each serum sample was tested for total serum calcium concentration, 25(OH)D concentration, HBV surface antigen (HBsAg), antibody to HBV core antigen (anti-HBc), and antibody to HCV (anti-HCV). Total serum calcium was measured using the Calcium Test Kit on the Cobas c501 automatic biochemistry system (Roche Diagnostic Corp.). Serum 25(OH)D was measured using a commercial enzyme Immunoassay Kit (IDS Inc.). HBsAg was analyzed using the Bio-Rad Genetic Systems HBsAg EIA 3.0 Kit (Bio-Rad Laboratories); anti-HBc was analyzed using the HBc (recombinant) ORTHO ELISA Test System (Ortho-Clinical Diagnostics); and anti-HCV was analyzed using the ORTHO HCV version 3.0 enzyme-linked immunosorbent assay (Ortho-Clinical Diagnostics). All tests were performed according to the instructions of the reagent manufacturers. Every 36 samples were accompanied by three pooled serum samples as internal controls. Pooled samples were made from 70 NIT serum samples that had undergone the same storage conditions as the study samples but were not included in this study. The coefficient of variation (CV) of 126 blinded quality control (QC) samples for total serum calcium measurements was 3.9%. Vitamin D measurements were performed in 42 batches. Each of the 42 batches included 39 samples (36 samples from cases or controls, and three QC samples). Each sample was individually measured in duplicate. The overall mean CV of these 126 blinded replicate QC samples for measuring vitamin D was 8.0%. In a sensitivity analysis, we excluded results from the eight batches with CVs >10%. As this did not change the results, we presented the full results from all 42 batches here.
Statistical analysis
Nonnormally distributed continuous variables are presented as median (interquartile range) and compared using Wilcoxon rank-sum test. Categorical variables are presented as number (percentage) and are compared using the χ2 test. A restricted cubic spline analysis was used in the multivariate logistic regression models to explore the nonlinear associations between continuous calcium and 25(OH)D concentrations and primary liver cancer, using R software version 3.5.3. Unconditional logistic regression models were used to estimate ORs and 95% confidence intervals (CI). Calcium and 25(OH)D concentrations were categorized using quartiles in the control population. In the independent and joint association analyses, quartile 1, quartiles 2 and 3, and quartile 4 of the indicators [serum calcium or serum 25(OH)D] were defined as low, medium, and high, respectively, to evaluate the associations between these indicators and the risk of incident primary liver cancer. Multivariable analyses were adjusted for other known primary liver cancer risk factors and potential confounders, including age at baseline (continuous), gender (female or male), body mass index (BMI; continuous), smoking (yes: lifetime smoking ≥6 months, or no: lifetime smoking <6 months or no smoking), drinking (yes: any alcohol consumption in the last 12 months, or no: no alcohol consumption in the last 12 months), trial (Dysplasia Trial or General Population Trial), HBsAg (negative or positive), anti-HBc (negative or positive), and anti-HCV (negative or positive). Subgroup analyses were performed to evaluate the possible impact of residual confounding or effect modification. To evaluate whether preclinical disease may have influenced the results, we conducted sensitivity analyses by excluding incident primary liver cancer cases that were diagnosed during the first 2, 5, or 10 years of follow-up. P values of interactive effects were calculated by including two multiplied variables into the model. For each fully adjusted logistic regression model, statistical significance and goodness-of-fit were evaluated using likelihood tests and Hosmer and Lemeshow tests, respectively. All statistical analyses were conducted using SPSS (version 17.0, IBM Corp). All tests were two-sided and P < 0.05 was considered statistically significant.
Data available in article supplementary material
The data that supports the findings of this study are available in the Supplementary Materials and Methods of this article.
Results
Baseline characteristics
We evaluated 226 incident primary liver cancer cases and 1,061 controls. Baseline and demographic characteristics of the subjects are shown and compared in Table 1. The median age of participants was 55 years old. There were no statistically significant differences between cases and controls in terms of BMI, smoking history, alcohol consumption, trial, or anti-HCV. Compared with controls, primary liver cancer cases had a higher prevalence of HBsAg and anti-HBc positivity. Seventy-one percent were below the reference range for normal serum calcium (<2.15 mmol/L; ref. 30) and 91% had inadequate 25(OH)D (<50 nmol/L; ref. 31). The median serum calcium concentration was higher in those who went on to develop primary liver cancer (median = 1.54 mmol/L) than those in the control group (median = 1.48 mmol/L, P < 0.05). The median of serum 25(OH)D levels were not significantly different cases (median = 20.50 nmol/L) than controls (median = 20.99 nmol/L).
Characteristics . | Controls (n = 1,061) . | Liver cancers (n = 226) . |
---|---|---|
Age, M (IQR) | 55 (50–61) | 55 (49–61) |
Gender, n (%) | ||
Female | 490 (46.2) | 91 (40.3) |
Male | 571 (53.8) | 135 (59.7) |
BMI, kg/m2, M (IQR) | 21.6 (20.2–23.1) | 21.8 (20.1–23.1) |
Ever smoking ≥6 months, n (%) | ||
No | 682 (64.3) | 138 (61.1) |
Yes | 379 (35.7) | 88 (38.9) |
Any alcohol consumption in last 12 months, n (%) | ||
No | 806 (76.0) | 171 (75.7) |
Yes | 255 (24.0) | 55 (24.3) |
Trial, n (%) | ||
Dysplasia | 131 (12.3) | 29 (12.8) |
General Population | 930 (87.7) | 197 (87.2) |
HBsAg, n (%)a | ||
Negative | 1,003 (94.5) | 173 (76.5) |
Positive | 58 (5.5) | 53 (23.5) |
Anti-HBc, n (%)a | ||
Negative | 440 (41.5) | 69 (30.5) |
Positive | 621 (58.5) | 157 (69.5) |
Anti-HCV, n (%)a | ||
Negative | 984 (92.7) | 202 (89.4) |
Positive | 77 (7.3) | 24 (10.6) |
Calcium, mmol/L, M (IQR) | 1.48 (1.03–2.23) | 1.54 (1.01–2.46) |
25(OH)D, nmol/L, M (IQR) | 20.99 (13.44–33.38) | 20.50 (13.96–30.26) |
Characteristics . | Controls (n = 1,061) . | Liver cancers (n = 226) . |
---|---|---|
Age, M (IQR) | 55 (50–61) | 55 (49–61) |
Gender, n (%) | ||
Female | 490 (46.2) | 91 (40.3) |
Male | 571 (53.8) | 135 (59.7) |
BMI, kg/m2, M (IQR) | 21.6 (20.2–23.1) | 21.8 (20.1–23.1) |
Ever smoking ≥6 months, n (%) | ||
No | 682 (64.3) | 138 (61.1) |
Yes | 379 (35.7) | 88 (38.9) |
Any alcohol consumption in last 12 months, n (%) | ||
No | 806 (76.0) | 171 (75.7) |
Yes | 255 (24.0) | 55 (24.3) |
Trial, n (%) | ||
Dysplasia | 131 (12.3) | 29 (12.8) |
General Population | 930 (87.7) | 197 (87.2) |
HBsAg, n (%)a | ||
Negative | 1,003 (94.5) | 173 (76.5) |
Positive | 58 (5.5) | 53 (23.5) |
Anti-HBc, n (%)a | ||
Negative | 440 (41.5) | 69 (30.5) |
Positive | 621 (58.5) | 157 (69.5) |
Anti-HCV, n (%)a | ||
Negative | 984 (92.7) | 202 (89.4) |
Positive | 77 (7.3) | 24 (10.6) |
Calcium, mmol/L, M (IQR) | 1.48 (1.03–2.23) | 1.54 (1.01–2.46) |
25(OH)D, nmol/L, M (IQR) | 20.99 (13.44–33.38) | 20.50 (13.96–30.26) |
Abbreviations: anti-HBc, antibody to hepatitis B core antigen; anti-HCV, antibody to hepatitis C virus; HBsAg, hepatitis B surface antigen; M (IQR), median (interquartile range).
aWilcoxon rank-sum test or χ2 test, P < 0.05.
Independent associations between the concentration of serum calcium, serum 25(OH)D concentrations, and primary liver cancer
The restricted cubic spline graphs showed a U-shaped association between serum calcium concentration and primary liver cancer risk, and an inverted U-shaped association between serum 25(OH)D concentrations and risk of primary liver cancer (Supplementary Fig. S1). Table 2 presents the results of the independent associations between serum calcium levels, serum 25(OH)D, and the risk of primary liver cancer adjusted by different variables. In the fully adjusted model 4, compared with the medium group, subjects with low calcium (OR = 1.48, 95% CI = 1.01–2.17) or high calcium (OR = 1.92, 95% CI = 1.34–2.76) had higher risks of primary liver cancer, and those with high 25(OH)D (OR = 0.54, 95% CI = 0.35–0.82) had a reduced risk of primary liver cancer. We found similar results when we analyzed the associations of calcium or 25(OH)D quartiles with primary liver cancer risk (Supplementary Table S1) or when we exclude primary liver cancer cases diagnosed in the first 2 years of follow-up (Supplementary Table S2).
. | Case/control . | Model 1 OR (95% CI)a . | Model 2 OR (95% CI)b . | Model 3 OR (95% CI)c . | Model 4 OR (95% CI)d . |
---|---|---|---|---|---|
Calcium | |||||
Low | 62/255 | 1.54 (1.07–2.20) | 1.46 (0.99–2.15) | 1.53 (1.07–2.20) | 1.48 (1.01–2.17) |
Medium | 86/543 | 1.00 (reference) | 1.00 (reference) | 1.00 (reference) | 1.00 (reference) |
High | 78/264 | 1.87 (1.33–2.62) | 1.93 (1.35–2.77) | 1.91 (1.36–2.69) | 1.92 (1.34–2.76) |
25(OH)D | |||||
Low | 51/264 | 0.78 (0.55–1.11) | 0.86 (0.59–1.25) | 0.74 (0.52–1.06) | 0.83 (0.56–1.21) |
Medium | 132/531 | 1.00 (reference) | 1.00 (reference) | 1.00 (reference) | 1.00 (reference) |
High | 43/265 | 0.65 (0.45–0.95) | 0.53 (0.36–0.81) | 0.64 (0.44–0.94) | 0.54 (0.35–0.82) |
. | Case/control . | Model 1 OR (95% CI)a . | Model 2 OR (95% CI)b . | Model 3 OR (95% CI)c . | Model 4 OR (95% CI)d . |
---|---|---|---|---|---|
Calcium | |||||
Low | 62/255 | 1.54 (1.07–2.20) | 1.46 (0.99–2.15) | 1.53 (1.07–2.20) | 1.48 (1.01–2.17) |
Medium | 86/543 | 1.00 (reference) | 1.00 (reference) | 1.00 (reference) | 1.00 (reference) |
High | 78/264 | 1.87 (1.33–2.62) | 1.93 (1.35–2.77) | 1.91 (1.36–2.69) | 1.92 (1.34–2.76) |
25(OH)D | |||||
Low | 51/264 | 0.78 (0.55–1.11) | 0.86 (0.59–1.25) | 0.74 (0.52–1.06) | 0.83 (0.56–1.21) |
Medium | 132/531 | 1.00 (reference) | 1.00 (reference) | 1.00 (reference) | 1.00 (reference) |
High | 43/265 | 0.65 (0.45–0.95) | 0.53 (0.36–0.81) | 0.64 (0.44–0.94) | 0.54 (0.35–0.82) |
Note: Low, medium, and high were defined as quartile 1, quartiles 2–3, and quartile 4, respectively.
aModel 1: Crude.
bModel 2: Adjusted for age, gender, BMI, smoking, drinking, trial, HBsAg, anti-HBc, and anti-HCV.
cModel 3: For the analysis of serum calcium, adjusted for serum 25(OH)D; for the analysis of serum 25(OH)D, adjusted for serum calcium.
dModel 4: For the analysis of serum calcium, adjusted for serum 25(OH)D, age, gender, BMI, smoking, drinking, trial, HBsAg, anti-HBc, and anti-HCV; for the analysis of serum 25(OH)D, adjusted for serum calcium, age, gender, BMI, smoking, drinking, trial, HBsAg, anti-HBc, and anti-HCV.
In subgroup analyses, the increased risks of primary liver cancer associated with low or high calcium were only found in young participants (≤55 years old at baseline), males, those with normal weight (BMI: 18.5–25.0), smokers, nondrinkers, participants in the General Population Trial, and those who were HBV positive. As for anti-HCV, increased risk of primary liver cancer was found in the high calcium group in both HCV-positive and HCV-negative subjects, and in the low calcium group in HCV-negative participants (all Pinteraction > 0.05; Table 3).
. | Serum calcium . | . | |||||
---|---|---|---|---|---|---|---|
. | Low (62/255) . | Medium (86/543) . | High (78/264) . | . | |||
. | Case/control . | OR (95% CI)a . | Case/control . | OR (reference) . | Case/control . | OR (95% CI)a . | Pinteraction . |
Age | 0.390 | ||||||
≤55 | 36/132 | 1.88 (1.12–3.16) | 43/302 | 1.00 | 43/125 | 2.29 (1.39–3.77) | |
>55 | 26/123 | 1.18 (0.66–2.12) | 43/241 | 1.00 | 35/139 | 1.64 (0.97–2.78) | |
Gender | 0.240 | ||||||
Female | 29/160 | 1.01 (0.63–1.93) | 32/209 | 1.00 | 30/121 | 1.71 (0.97–3.05) | |
Male | 33/95 | 2.00 (1.18–3.40) | 54/334 | 1.00 | 48/143 | 2.16 (1.35–3.47) | |
BMI | 0.728 | ||||||
<18.5 | 4/18 | 1.01 (0.19–5.30) | 6/31 | 1.00 | 9/28 | 2.52 (0.67–9.52) | |
18.5–25.0 | 48/192 | 1.62 (1.07–2.44) | 69/438 | 1.00 | 60/205 | 1.91 (1.29–2.82) | |
≥25.0 | 10/45 | 1.05 (0.21–5.40) | 11/73 | 1.00 | 9/31 | 4.79 (0.83–27.80) | |
Smoking | 0.106 | ||||||
No | 39/191 | 1.12 (0.70–1.81) | 53/310 | 1.00 | 46/182 | 1.55 (0.98–2.44) | |
Yes | 23/64 | 2.44 (1.27–4.69) | 33/232 | 1.00 | 32/82 | 2.88 (1.59–5.23) | |
Drinking | 0.475 | ||||||
No | 47/198 | 1.50 (0.96–2.32) | 61/408 | 1.00 | 63/201 | 2.17 (1.44–3.24) | |
Yes | 15/57 | 1.80 (0.79–4.09) | 25/134 | 1.00 | 15/63 | 1.35 (0.58–3.16) | |
Trial | 0.794 | ||||||
Dysplasia | —/— | — | 10/57 | 1.00 | 19/73 | 2.07 (0.78–5.50) | |
General Population | 62/255 | 1.48 (1.01–2.18) | 76/485 | 1.00 | 59/191 | 1.98 (1.34–2.94) | |
HBV | 0.102 | ||||||
Negative | 18/105 | 1.09 (0.57–2.10) | 31/207 | 1.00 | 20/120 | 1.09 (0.58–2.05) | |
Positive | 44/150 | 1.99 (1.26–3.14) | 55/336 | 1.00 | 58/143 | 2.58 (1.68–3.97) | |
HCV | 0.098 | ||||||
Negative | 58/229 | 1.58 (1.06–2.35) | 78/508 | 1.00 | 66/247 | 1.80 (1.32–2.64) | |
Positive | 4/26 | 0.99 (0.21–4.54) | 8/35 | 1.00 | 12/16 | 7.07 (1.69–29.61) |
. | Serum calcium . | . | |||||
---|---|---|---|---|---|---|---|
. | Low (62/255) . | Medium (86/543) . | High (78/264) . | . | |||
. | Case/control . | OR (95% CI)a . | Case/control . | OR (reference) . | Case/control . | OR (95% CI)a . | Pinteraction . |
Age | 0.390 | ||||||
≤55 | 36/132 | 1.88 (1.12–3.16) | 43/302 | 1.00 | 43/125 | 2.29 (1.39–3.77) | |
>55 | 26/123 | 1.18 (0.66–2.12) | 43/241 | 1.00 | 35/139 | 1.64 (0.97–2.78) | |
Gender | 0.240 | ||||||
Female | 29/160 | 1.01 (0.63–1.93) | 32/209 | 1.00 | 30/121 | 1.71 (0.97–3.05) | |
Male | 33/95 | 2.00 (1.18–3.40) | 54/334 | 1.00 | 48/143 | 2.16 (1.35–3.47) | |
BMI | 0.728 | ||||||
<18.5 | 4/18 | 1.01 (0.19–5.30) | 6/31 | 1.00 | 9/28 | 2.52 (0.67–9.52) | |
18.5–25.0 | 48/192 | 1.62 (1.07–2.44) | 69/438 | 1.00 | 60/205 | 1.91 (1.29–2.82) | |
≥25.0 | 10/45 | 1.05 (0.21–5.40) | 11/73 | 1.00 | 9/31 | 4.79 (0.83–27.80) | |
Smoking | 0.106 | ||||||
No | 39/191 | 1.12 (0.70–1.81) | 53/310 | 1.00 | 46/182 | 1.55 (0.98–2.44) | |
Yes | 23/64 | 2.44 (1.27–4.69) | 33/232 | 1.00 | 32/82 | 2.88 (1.59–5.23) | |
Drinking | 0.475 | ||||||
No | 47/198 | 1.50 (0.96–2.32) | 61/408 | 1.00 | 63/201 | 2.17 (1.44–3.24) | |
Yes | 15/57 | 1.80 (0.79–4.09) | 25/134 | 1.00 | 15/63 | 1.35 (0.58–3.16) | |
Trial | 0.794 | ||||||
Dysplasia | —/— | — | 10/57 | 1.00 | 19/73 | 2.07 (0.78–5.50) | |
General Population | 62/255 | 1.48 (1.01–2.18) | 76/485 | 1.00 | 59/191 | 1.98 (1.34–2.94) | |
HBV | 0.102 | ||||||
Negative | 18/105 | 1.09 (0.57–2.10) | 31/207 | 1.00 | 20/120 | 1.09 (0.58–2.05) | |
Positive | 44/150 | 1.99 (1.26–3.14) | 55/336 | 1.00 | 58/143 | 2.58 (1.68–3.97) | |
HCV | 0.098 | ||||||
Negative | 58/229 | 1.58 (1.06–2.35) | 78/508 | 1.00 | 66/247 | 1.80 (1.32–2.64) | |
Positive | 4/26 | 0.99 (0.21–4.54) | 8/35 | 1.00 | 12/16 | 7.07 (1.69–29.61) |
Note: Low, medium, and high were defined as quartile 1, quartiles 2–3, and quartile 4, respectively.
Abbreviations: HBV positive, either HBsAg positive or anti-HBc positive; HCV positive, anti-HCV positive.
aAdjusted for age, gender, BMI, smoking, drinking, trial, HBsAg, anti-HBc, and anti-HCV.
In subgroup analyses of 25(OH)D, significantly decreased risk of primary liver cancer was found in both strata of smoking and trial, and in older participants (>55 years old at baseline), male participants, those with normal weight, drinkers, HBV-positive subjects, and anti-HCV–negative subjects (all Pinteraction > 0.05; Table 4).
. | Serum 25(OH)D . | . | |||||
---|---|---|---|---|---|---|---|
. | Low (51/264) . | Medium (132/531) . | High (43/265) . | . | |||
. | Case/control . | OR (95% CI)a . | Case/control . | OR (reference) . | Case/control . | OR (95% CI)a . | Pinteraction . |
Age | 0.820 | ||||||
≤55 | 24/116 | 0.93 (0.54–1.60) | 77/316 | 1.00 | 21/127 | 0.58 (0.32–1.05) | |
>55 | 27/148 | 0.81 (0.47–1.40) | 55/215 | 1.00 | 22/138 | 0.48 (0.26–0.89) | |
Gender | 0.398 | ||||||
Female | 30/187 | 0.68 (0.41–1.13) | 55/255 | 1.00 | 6/45 | 0.60 (0.23–1.58) | |
Male | 21/77 | 1.20 (0.67–2.13) | 77/276 | 1.00 | 37/220 | 0.52 (0.32–0.87) | |
BMI | 0.993 | ||||||
<18.5 | 3/12 | 1.42 (0.27–7.46) | 12/42 | 1.00 | 4/23 | 0.63 (0.13–3.11) | |
18.5–25.0 | 38/252 | 0.82 (0.54–1.24) | 104/408 | 1.00 | 35/211 | 0.55 (0.35–0.87) | |
≥25.0 | 10/38 | 0.93 (0.18–4.80) | 16/80 | 1.00 | 4/31 | 0.24 (0.03–2.35) | |
Smoking | 0.324 | ||||||
No | 36/211 | 0.71 (0.45–1.12) | 81/338 | 1.00 | 21/131 | 0.55 (0.31–0.99) | |
Yes | 15/53 | 1.38 (0.69–2.76) | 51/192 | 1.00 | 22/134 | 0.51 (0.27–0.99) | |
Drinking | 0.129 | ||||||
No | 43/215 | 0.90 (0.59–1.37) | 94/417 | 1.00 | 34/172 | 0.77 (0.47–1.26) | |
Yes | 8/49 | 0.83 (0.33–2.12) | 38/113 | 1.00 | 9/93 | 0.17 (0.06–0.46) | |
Trial | 0.379 | ||||||
Dysplasia | 1/9 | 0.49 (0.05–4.95) | 15/39 | 1.00 | 13/83 | 0.20 (0.07–0.60) | |
General Population | 50/255 | 0.88 (0.60–1.30) | 117/492 | 1.00 | 30/182 | 0.62 (0.39–0.99) | |
HBV | 0.880 | ||||||
Negative | 17/111 | 0.79 (0.42–1.48) | 42/223 | 1.00 | 10/97 | 0.57 (0.26–1.25) | |
Positive | 34/153 | 0.89 (0.56–1.42) | 90/307 | 1.00 | 33/168 | 0.54 (0.32–0.87) | |
HCV | 0.786 | ||||||
Negative | 46/243 | 0.83 (0.56–1.25) | 119/491 | 1.00 | 37/248 | 0.52 (0.33–0.82) | |
Positive | 5/21 | 0.93 (0.26–3.27) | 13/39 | 1.00 | 6/17 | 0.21 (0.04–1.19) |
. | Serum 25(OH)D . | . | |||||
---|---|---|---|---|---|---|---|
. | Low (51/264) . | Medium (132/531) . | High (43/265) . | . | |||
. | Case/control . | OR (95% CI)a . | Case/control . | OR (reference) . | Case/control . | OR (95% CI)a . | Pinteraction . |
Age | 0.820 | ||||||
≤55 | 24/116 | 0.93 (0.54–1.60) | 77/316 | 1.00 | 21/127 | 0.58 (0.32–1.05) | |
>55 | 27/148 | 0.81 (0.47–1.40) | 55/215 | 1.00 | 22/138 | 0.48 (0.26–0.89) | |
Gender | 0.398 | ||||||
Female | 30/187 | 0.68 (0.41–1.13) | 55/255 | 1.00 | 6/45 | 0.60 (0.23–1.58) | |
Male | 21/77 | 1.20 (0.67–2.13) | 77/276 | 1.00 | 37/220 | 0.52 (0.32–0.87) | |
BMI | 0.993 | ||||||
<18.5 | 3/12 | 1.42 (0.27–7.46) | 12/42 | 1.00 | 4/23 | 0.63 (0.13–3.11) | |
18.5–25.0 | 38/252 | 0.82 (0.54–1.24) | 104/408 | 1.00 | 35/211 | 0.55 (0.35–0.87) | |
≥25.0 | 10/38 | 0.93 (0.18–4.80) | 16/80 | 1.00 | 4/31 | 0.24 (0.03–2.35) | |
Smoking | 0.324 | ||||||
No | 36/211 | 0.71 (0.45–1.12) | 81/338 | 1.00 | 21/131 | 0.55 (0.31–0.99) | |
Yes | 15/53 | 1.38 (0.69–2.76) | 51/192 | 1.00 | 22/134 | 0.51 (0.27–0.99) | |
Drinking | 0.129 | ||||||
No | 43/215 | 0.90 (0.59–1.37) | 94/417 | 1.00 | 34/172 | 0.77 (0.47–1.26) | |
Yes | 8/49 | 0.83 (0.33–2.12) | 38/113 | 1.00 | 9/93 | 0.17 (0.06–0.46) | |
Trial | 0.379 | ||||||
Dysplasia | 1/9 | 0.49 (0.05–4.95) | 15/39 | 1.00 | 13/83 | 0.20 (0.07–0.60) | |
General Population | 50/255 | 0.88 (0.60–1.30) | 117/492 | 1.00 | 30/182 | 0.62 (0.39–0.99) | |
HBV | 0.880 | ||||||
Negative | 17/111 | 0.79 (0.42–1.48) | 42/223 | 1.00 | 10/97 | 0.57 (0.26–1.25) | |
Positive | 34/153 | 0.89 (0.56–1.42) | 90/307 | 1.00 | 33/168 | 0.54 (0.32–0.87) | |
HCV | 0.786 | ||||||
Negative | 46/243 | 0.83 (0.56–1.25) | 119/491 | 1.00 | 37/248 | 0.52 (0.33–0.82) | |
Positive | 5/21 | 0.93 (0.26–3.27) | 13/39 | 1.00 | 6/17 | 0.21 (0.04–1.19) |
Note: Low, medium, and high were defined as quartile 1, quartiles 2–3, and quartile 4, respectively.
Abbreviations: HBV positive, either HBsAg positive or anti-HBc positive; HCV positive, anti-HCV positive.
aAdjusted for age, gender, BMI, smoking, drinking, trial, HBsAg, anti-HBc, and anti-HCV.
Joint associations between serum calcium, 25(OH)D concentrations, and primary liver cancer
Using medium levels of serum calcium and 25(OH)D as the reference, we found that the combination of medium calcium and high 25(OH)D was associated with a 66% decreased risk of primary liver cancer (OR = 0.34, 95% CI = 0.17–0.67). On the other hand, there was a nearly two-fold increased risk of primary liver cancer in subjects with the combination of high calcium and medium 25(OH)D (OR = 1.89, 95% CI = 1.17–3.05). We did not find significant differences in primary liver cancer risk in the other groups (all P > 0.05; Table 5). When we restricted the subjects to those in the General Population Trial, the results were similar (Supplementary Table S3). And excluding cases that were diagnosed within 2 years, 5 years, or 10 years of baseline did not alter the results of the main analysis (Table 6).
Calcium . | 25(OH)D . | Case/control . | Model 1 OR (95%CI)a . | Model 2 OR (95%CI)b . |
---|---|---|---|---|
Medium | Medium | 54/283 | 1.00 (reference) | 1.00 (reference) |
Medium | Low | 19/111 | 0.90 (0.51–1.58) | 1.02 (0.56–1.84) |
Medium | High | 13/148 | 0.46 (0.24–0.87) | 0.34 (0.17–0.67) |
Low | Medium | 33/131 | 1.32 (0.81–2.13) | 1.27 (0.76–2.11) |
Low | Low | 17/88 | 1.01 (0.56–1.84) | 1.02 (0.54–1.92) |
Low | High | 12/34 | 1.85 (0.90–3.80) | 1.55 (0.72–3.32) |
High | Medium | 45/117 | 2.02 (1.29–3.16) | 1.89 (1.17–3.05) |
High | Low | 15/65 | 1.21 (0.64–2.28) | 1.31 (0.67–2.56) |
High | High | 18/82 | 1.15 (0.64–2.07) | 0.99 (0.51–1.93) |
Calcium . | 25(OH)D . | Case/control . | Model 1 OR (95%CI)a . | Model 2 OR (95%CI)b . |
---|---|---|---|---|
Medium | Medium | 54/283 | 1.00 (reference) | 1.00 (reference) |
Medium | Low | 19/111 | 0.90 (0.51–1.58) | 1.02 (0.56–1.84) |
Medium | High | 13/148 | 0.46 (0.24–0.87) | 0.34 (0.17–0.67) |
Low | Medium | 33/131 | 1.32 (0.81–2.13) | 1.27 (0.76–2.11) |
Low | Low | 17/88 | 1.01 (0.56–1.84) | 1.02 (0.54–1.92) |
Low | High | 12/34 | 1.85 (0.90–3.80) | 1.55 (0.72–3.32) |
High | Medium | 45/117 | 2.02 (1.29–3.16) | 1.89 (1.17–3.05) |
High | Low | 15/65 | 1.21 (0.64–2.28) | 1.31 (0.67–2.56) |
High | High | 18/82 | 1.15 (0.64–2.07) | 0.99 (0.51–1.93) |
Note: Low, medium, and high were defined as quartile 1, quartiles 2–3, and quartile 4, respectively.
aModel 1: Crude.
bModel 2: Adjusted for age, gender, BMI, smoking, drinking, trial, HBsAg, anti-HBc, and anti-HCV.
Calcium . | 25(OH)D . | Case/control . | Model 1 OR (95% CI)a . | Model 2 OR (95% CI)b . |
---|---|---|---|---|
Cases that occurred within 2 years excluded | ||||
Medium | Medium | 50/283 | 1.00 (reference) | 1.00 (reference) |
Medium | High | 12/148 | 0.46 (0.24–0.89) | 0.34 (0.17–0.70) |
High | Medium | 44/117 | 2.13 (1.35–3.37) | 2.00 (1.23–3.25) |
Cases that occurred within 5 years excluded | ||||
Medium | Medium | 41/283 | 1.00 (reference) | 1.00 (reference) |
Medium | High | 9/148 | 0.42 (0.20–0.89) | 0.36 (0.16–0.79) |
High | Medium | 40/117 | 2.36 (1.45–3.84) | 2.43 (1.45–4.05) |
Cases that occurred within 10 years excluded | ||||
Medium | Medium | 33/283 | 1.00 (reference) | 1.00 (reference) |
Medium | High | 5/148 | 0.29 (0.11–0.76) | 0.28 (0.10–0.75) |
High | Medium | 34/117 | 2.49 (1.47–4.21) | 2.67 (1.53–4.66) |
Calcium . | 25(OH)D . | Case/control . | Model 1 OR (95% CI)a . | Model 2 OR (95% CI)b . |
---|---|---|---|---|
Cases that occurred within 2 years excluded | ||||
Medium | Medium | 50/283 | 1.00 (reference) | 1.00 (reference) |
Medium | High | 12/148 | 0.46 (0.24–0.89) | 0.34 (0.17–0.70) |
High | Medium | 44/117 | 2.13 (1.35–3.37) | 2.00 (1.23–3.25) |
Cases that occurred within 5 years excluded | ||||
Medium | Medium | 41/283 | 1.00 (reference) | 1.00 (reference) |
Medium | High | 9/148 | 0.42 (0.20–0.89) | 0.36 (0.16–0.79) |
High | Medium | 40/117 | 2.36 (1.45–3.84) | 2.43 (1.45–4.05) |
Cases that occurred within 10 years excluded | ||||
Medium | Medium | 33/283 | 1.00 (reference) | 1.00 (reference) |
Medium | High | 5/148 | 0.29 (0.11–0.76) | 0.28 (0.10–0.75) |
High | Medium | 34/117 | 2.49 (1.47–4.21) | 2.67 (1.53–4.66) |
Note: Low, medium, and high were defined as quartile 1, quartiles 2–3, and quartile 4, respectively.
aModel 1: Crude.
bModel 2: Adjusted for age, gender, BMI, smoking, drinking, trial, HBsAg, anti-HBc, and anti-HCV.
Discussion
To the best of our knowledge, this is the first study to prospectively examine the independent and joint associations between prediagnostic serum calcium, vitamin D, and the risk of incident primary liver cancer. This case–control study was nested in two nutritional intervention trials (the Dysplasia Trial and the General Population Trial) conducted in parallel in Linxian, China. In both trials, serum samples used for calcium and vitamin D measurements were collected at baseline, before supplementation began.
Calcium and primary liver cancer
Most prior investigations of the relationship between calcium and liver cells have been performed in vitro or in animals (3, 32, 33), and they suggest that calcium is a key mediator of lethal cell injury. Some studies have shown that calcium homeostasis plays a critical role in cell degeneration, aging, and death in various rat tissues (3, 34). The biological processes by which calcium may influence primary liver cancer are largely unknown, and there have been no direct studies of the independent effects of serum calcium concentrations on this cancer. Our study observed an increased primary liver cancer risk associated with both low (OR = 1.48) and high (OR = 1.92) serum calcium levels, especially in men. Additional studies will be required to confirm our findings and learn more about this relationship.
Excess body weight increases risk of liver cancer (35, 36). But this association was not found in our study, possibly because few of our subjects (13.9%) were overweight, and the participants in the NIT cohorts were from a nutrient-deficient population.
Alcohol consumption is a well-known risk factor for primary liver cancer, but we observed no association between it and calcium concentration on the risk of primary liver cancer incidence. One reason may be that during the time the serum was collected (1984–1985), drinking was very uncommon in the study population (37) and the reported intensity of alcohol drinking was not associated with another typically alcohol-related cancer, esophageal squamous cell carcinoma, in the same population (38).
Overwhelming evidence suggests that infection with HBV or HCV causes or exacerbates the development of primary liver cancer (39). Alteration of cytosolic calcium is a basic requirement for HBV replication and is mediated by HBV X (HBx) protein that is important for infection by HBV (40). Also, HBx protein boosts the progression of liver cancer through calcium-dependent cascades (41). In our study, although we did not find a statistically significant interaction between serum calcium and HBV, the subjects with HBV infection (HBsAg or anti-HBc positive) had a higher risk of primary liver cancer in both lower and higher calcium groups. HCV infection is also related to calcium homeostasis because HCV regulates host calcium metabolism for virus propagation (42). Stratified by anti-HCV antibody, our data showed that in subjects with HCV infection, the high calcium group showed more than seven times the odds of developing primary liver cancer compared with those in the medium calcium group.
Vitamin D and primary liver cancer
In contrast to the results of serum calcium, we found a decreased risk of primary liver cancer cases in subjects with high 25(OH)D levels (ORHigh/Medium = 0.54), which is consistent with the findings of previous studies (12, 22, 24). In a study from a Japanese public health center–based prospective study cohort, higher vitamin D concentration was associated with lower risk of liver cancer (HR = 0.45, 95% CI = 0.26–0.79; ref. 22). The findings did not change in strata of smoking or trial. As stated above in the Materials and Methods section, the blood samples from the Dysplasia Trial cohort were drawn just after summer (August to October), and those from the General Population Trial cohort were drawn just after winter (March to May). Thus, the result of stratification by trials was same as that of stratification by season of blood collection. After stratification by other variables, the significant associations were only found in older participants (>55 years old at baseline), males, those with medium weight, alcohol drinkers, HBV-positive subjects, and HCV-negative subjects.
Lower vitamin D levels have been reported in HBV- or HCV-positive participants (43, 44), which is consistent with our study. Several studies have shown protective properties of vitamin D in inhibiting HBV and HCV replication and in retarding clinical progression of HBV- or HCV-related liver diseases (45, 46). We also observed that high 25(OH)D in HBV-positive subjects was protective for primary liver cancer risk. In contrast, this protection was not found in the HCV-positive group, possibly because only a few participants were infected with HCV in our study.
The joint effects of serum calcium and vitamin D levels on the risk of primary liver cancer
When calcium and 25(OH)D concentrations were considered jointly, subjects with high calcium and medium 25(OH)D were nearly two-fold more likely to develop primary liver cancer than those with medium calcium and medium 25(OH)D (OR = 1.89); those with medium calcium and high 25(OH)D had a 66% reduction in primary liver cancer risk (OR = 0.34); and those in other groups had no change in the odds of developing primary liver cancer. The underlying biologic mechanisms of the relationship between serum calcium, serum 25(OH)D levels and primary liver cancer risk need to be elucidated by further research. Excluding the subjects from the Dysplasia Trial cohort indicated that a prior history of cytologic esophageal dysplasia was not a risk factor for primary liver cancer. Excluding cases diagnosed within 2, 5, or 10 years of baseline from the analyses did not substantially alter ORs, which showed that our conclusions were stable and reliable.
Strengths and limitations
Our study had several strengths. Chief among them was the use of a prospective design. The serum used for testing was collected at baseline, before interventions, diseases, or other possible confounders could affect the interpretation of the associations. The questionnaire information was also collected at the beginning of the cohort, by face-to-face interview, avoiding bias in data collection. We also had high-quality follow-up, and the lost to follow-up rate was <1%. Furthermore, our study design and analysis considered the major risk factors for liver disease (HBV and HCV infection) to isolate the relationship of concern in this study. Finally, this study was the first to analyze the independent effects of serum calcium and the combined effects of serum calcium and serum 25(OH)D on primary liver cancer.
There were also several limitations in this study. As serum albumin levels may affect serum calcium levels in patients, some studies suggest that serum calcium levels should be adjusted for serum albumin (15, 47), while other studies have drawn the opposite conclusion and suggested that unadjusted serum total calcium is superior to albumin-adjusted total serum calcium (48). In our study, the blood was collected from an apparently healthy population, with no serious diseases diagnosed at baseline, so the albumin levels should not have varied greatly and they should have had relatively little influence on the levels of total serum calcium. Furthermore, as a conventional component of serum biochemistry test panels, unadjusted total serum calcium remains the most widely used indicator to estimate human calcium levels by clinicians in China. A second limitation of this study was that a large proportion of the primary liver cancer cases were not diagnosed on the basis of histologic evidence, but by the combined evidence from biochemical assays, clinical examination, ultrasound, and CT scan, which could not exclude the possibility of misclassification. However, if that were the case, our reported results would likely be attenuated in magnitude. Another limitation of this study was that serum calcium and vitamin D levels in our population were quite low. Seventy-one percent were below the reference range for normal serum calcium (<2.15 mmol/L; ref. 30), and 91% had inadequate 25(OH)D (<50 nmol/L; ref. 31), which probably reflects a diet poor in calcium-rich, vitamin D–rich foods. Thus, it is not clear whether our results are applicable to other populations with higher calcium and vitamin D levels. And finally, our sample size was limited for detecting modest associations and for examining stratifications. Future studies are needed to address these issues.
In conclusion, in a nutrient-deficient population, we have provided the first prospective evaluation of the independent and joint effects of serum calcium and 25(OH)D concentrations on the risk of incident primary liver cancer. Low and high serum calcium resulted in increased risk, while high serum 25(OH)D was a protective factor. These results may be of potential scientific and clinical significance for primary liver cancer prevention and control. Further studies are needed to confirm these findings in populations with higher calcium and 25(OH)D levels and to elaborate underlying mechanisms.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
J. Yin: Conceptualization, data curation, software, formal analysis, investigation, methodology, writing–original draft, project administration, writing–review and editing. L.-Y. Yin: Conceptualization, data curation, writing–review and editing. N.D. Freedman: Data curation, writing–review and editing. T.-Y. Li: Data curation, investigation. S.M. Dawsey: Data curation, methodology, writing–review and editing. J.-F. Cui: Data curation. P.R. Taylor: Writing–review and editing. B. Liu: Data curation. J.-H. Fan: Resources, supervision. W. Chen: Conceptualization, resources, data curation, supervision, project administration, writing–review and editing. C.C. Abnet: Conceptualization, writing–review and editing. Y.-L. Qiao: Supervision, project administration, writing–review and editing.
Acknowledgments
This work was supported in part by NCI contracts (N01-SC-91030, N01-RC-47701, and N02CP-2017-00047 to Y.-L. Qiao) to the National Cancer Center, Chinese Academy of Medical Sciences, and in part by the National Cancer Center, Chinese Academy of Medical Sciences.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.