Pre-diagnostic urinary 15-F_2t-isoprostane level and liver cancer risk: results from the Shanghai Women’s and Men’s Health Studies

Abstract

Oxidative stress has been hypothesized to affect cancer development via various mechanisms, but the evidence from human is limited and inconclusive. 15-F_2t-isoprostane (15-F_2t-IsoP) is an accurate marker of oxidative stress in humans. Recent studies showed that the evidence of urinary 15-F_2t-IsoP level correlating cancer risk is conflicting. We conducted a case-control study nested within two population-based cohort studies. Pre-diagnosis urine samples, collected at cohort enrollment, from 363 incident liver cancer cases and 725 individually matched controls, were used to determine the level of 15-F_2t-IsoP by enzyme-linked immunosorbent assay. Conditional logistic regression model was used to estimate the odds ratio to measure the association between the urinary 15-F_2t-IsoP level and liver cancer risk. We found that higher pre-diagnostic urinary 15-F_2t-IsoP level was associated with an increased liver cancer risk, with an adjusted OR in males (OR_{4th versus 1st quartile} = 8.84, 95% CI 2.74-28.60), which was significantly higher than that in females (OR_{4th versus 1st qarntile} = 1.75, 95% CI 0.70-4.42). HBsAg carriers with higher 15-F_2t-IsoP had a significantly increased liver cancer risk (OR _female=59.04, 95% CI 12.26, 284.30; OR _male=92.55, 95% CI 34.83, 245.96) compared with non-carriers with lower 15-F_2t-IsoP. High urinary 15-F_2t-IsoP level was associated with high liver cancer risk, suggesting that 15-F_2t-IsoP may be a promising biomarker for liver cancer risk. The result suggest that people with sero-positive HBsAg and higher level of 15-F_2t-IsoP might be given a higher priority on future surveillance program of liver cancer.

Introduction

Liver cancer is the sixth most commonly diagnosed cancer¹ and second leading cause of cancer death worldwide². Prognosis of liver cancer is poor with an overall 5-year survival rate of less than 15% in most population-based studied^{2, 3}. An estimated 782,000 new cases and 745,000 deaths occurred worldwide in 2012⁴. Liver cancer rates are the highest in East and South-East Asia and Northern and Western Africa². During last few decades, the incidence rate of liver cancer has increased in most developed countries⁵. China is the area of world most affected by liver cancer, with an age-standardized incidence rate of 22.3 per 100,000 (males: 33.7 per 100,000, females: 10.9 per 100,000). In China nowadays, liver cancer is the second major cause of cancer deaths, with a mortality rate of 21.4 per 100,000 (males: 32.3 per 100,000, females: 10.7 per 100,000), accounting for 17.4% of all cancer deaths².

The development of liver cancer is not completely understood, but evidence suggests that oxidative stress has a role in liver cancer initiation or promotion^6–9. Most cells in healthy people have adequate antioxidant defenses to protect themselves from the damage caused by reactive oxygen species (ROS)^{10, 11}. However, this capacity can be overwhelmed by an abundance of ROS from exogenous or endogenous sources^12–14. Oxidative stress, mediated by ROS, may result in direct DNA damage as well as in lipid peroxidation, which may lead to cancer¹⁵. ROS has been found to be involved in the progression of chronic liver disease and the occurrence of liver cancer¹⁶.

Accumulating evidence suggests that the F2-isoprostanes, mainly 15-F_2t-isoprostane (15-F_2t-IsoP), produced during peroxidation by free radicals and ROS during lipid peroxidation of arachidonic acid, are more accurate markers of oxidative stress than other available biomarkers^17–19. It has also been suggested that the level of 15-F_2t-IsoP might be used as a biomarker of exposure to relevant carcinogens and may predict cancer risk^{14, 20, 21}. Increasing level of urinary 15-F_2t-IsoP has been shown to be associated with breast cancer in an American retrospective case-control study, and with prostate cancer in an American prospective sub-cohort study, and with hepatocellular carcinoma in a Taiwanese prospective nested case-control study^{14, 20, 21}. In our study, we evaluated the relationship between urinary 15-F_2t-IsoP level and liver cancer in two prospective studies in mainland China.

Materials and Methods

Study population

The Shanghai Women’s Health Study (SWHS) and the Shanghai Men’s Health Study (SMHS) are both prospective, population-based cohort studies in Shanghai, China^{22, 23}. In brief, for the SWHS, 74,941 women aged 40 to 70 years in Shanghai were recruited from 1997 to 2000, with an overall study participation rate of 92.7%. For the SMHS, 61,480 men aged 40 to 74 years with no history of cancer were recruited in Shanghai from 2002 to 2006, with an overall study participation rate of 74.1%. Trained interviewers administered baseline surveys and obtained anthropometric measurements of the participants. The two cohorts were approved by all relevant Institutional Review Boards and an informed consent was obtained from each participant.

Follow-up and cancer identification

Cohort members have been followed for cancer occurrence through in-person follow-up surveys taking place every 3 years and annual record linkage with databases of the population-based Shanghai Cancer Registry, Shanghai Vital Statistics, and Shanghai Resident Registry. For the SWHS, the response rates for the first (2000-2002), second (2002-2004), third (2004-2007), and fourth (2008-2011) in-person follow-up surveys were 99.8%, 98.7%, 96.7%, and 92.3%, respectively. For the SMHS, the response rates for the first (2004-2008) and second (2008-2011) follow-up surveys were 97.6% and 93.7%.

All possible cancer cases were verified through home visits, and diagnosed by review of medical charts by a panel of clinical and pathological experts. Cancers were coded according to the International Classification of Disease, Ninth Revision (ICD-9). Liver cancer was defined as a primary malignant tumor with an ICD-9 code of 155²⁴.

Nested case-control study

Participants of nested case-control study included incident cases of liver cancer diagnosed during the period between recruitment and December 31, 2012. Two controls per case were randomly selected and individually matched to the index cases by gender, age at recruitment (± 2 years), date (± 30 days) of biospecimen collection, time (morning or afternoon) of urine sample collection, interval since last meal (less than 2h) and menopausal status (in women, yes or no). Since one case failed to find two controls (only one control), thus, 725 matched controls were selected. The present study included 363 incidence cases and 725 eligible controls in the analysis.

Quantification of urinary 15-F_2t-IsoP

Urinary 15-F_2t-IsoP level was determined using a competitive immunoenzymatic assay kit from Abcam Company (Cambs, UK). A quality control sample consisting of urine pooled from ten controls was analyzed with each batch of test samples. Urine samples and standards were assayed in duplicate. Laboratory staff was blinded to case-control status. The intra-day relative standard deviation was 9.84% and the inter-day relative standard deviation was 3.67%. The urinary creatinine (Cr) was measured with commercial kit from the Jiancheng Bio-engineering Institute (Nanjing, China) on the basis of by the Jaffé alkaline picrate procedure²⁵. Urinary 15-F_2t-IsoP level was standardized to urinary Cr level (ng/mg Cr).

Quantification of plasma hepatitis B surface antigen (HBsAg) and hepatitis B e antigen (HBeAg)

The plasma hepatitis B surface antigen (HBsAg) and hepatitis B e antigen (HBeAg) were measured using Architect i2000 (Abbott Diagnostic, USA) and according to the manufacturer’s instructions. The laboratory personnel were blinded as to the disease or control status of study subjects. The design has been previously described in detail²⁶.

Statistical analysis

We conducted analyses separately by gender. For the baseline characteristics between the liver cancer cases and matched controls in women and men, continuous variables were compared using t test and categorical variables were compared using Chi-square test.

We evaluated the associations of urinary 15-F_2t-IsoP level with selected risk factors among the control subjects, including age (40 to 49, 50 to 59, ≥60 years), body mass index (BMI, <25, ≥25 kg/m²), physical activity, income, education, occupation, smoking (men only), drinking (men only), history of hepatitis/chronic liver disease, menopausal status (women only), oral contraceptive use (women only), and HBsAg status (positive/negative). The 15-F_2t-IsoP level was categorized based on quartile distribution in controls. Variables were compared using Chi-square test.

Conditional logistic models were used to calculate the odds ratios (ORs) for liver cancer risk by quartiles of 15-F_2t-IsoP. The final model was adjusted for test batch, BMI, physical activity, smoking (for male), alcohol consumption (for male), income, education, total energy intake, total vegetable intake, total fruit intake, total red meat intake, total egg intake, total fish intake, total soy intake, total vitmin E intake, history of hepatitis/chronic liver disease, family history of liver cancer, diabetes, cholelithiasis, cholecystectomy, oral contraceptive use (for female), HBsAg, and HBeAg. Tests for heterogeneity (Cochran’s Q statistic and its P value)²⁷ were used to assess gender difference for liver cancer risk when comparing the highest quartile of 15-F_2t-IsoP with the lowest quartile. To evaluate the combined effect of urinary 15-F_2t-IsoP level on the risk of liver cancer, subjects were divided into different groups based on the level of urinary15-F_2t-IsoP and HBsAg status. The reference group was those with urinary 15-F_2t-IsoP level below the median value and negative for HBsAg.

Statistical analyses were conducted using SAS 9.3 (SAS Institute, Cary, NC). A two-sided P values of <0.05 was considered statistically significant.

Results

Baseline characteristics of liver cancer cases and matched healthy controls were shown in Table 1. Compared to controls, patients with liver cancer were more likely to have a family history of liver cancer and a history of chronic diseases, including viral hepatitis/chronic liver disease, HBsAg positive, and HBeAg positive. Female cases were more likely to have lower education level than controls.

Table 1.

Baseline characteristics of liver cancer patients and controls (SWHS 1997–2012, SMHS 2002-2012)

	Women			Men
Characteristic1	Cases (n=129)	Controls (n=257)	P2	cases (n=234)	Non-cases (n=468)	P2
Age at recruitment (years)	58.75±8.73	58.62±8.78	0.89	59.43±9.83	59.44±9.80	0.98
Body mass index (BMI) (kg/m2)	24.77±3.77	24.58±3.54	0.62	23.70±3.35	23.81±3.01	0.67
Total physical activity (MET hours/week)	102.25±44.77	100.59±45.71	0.73	64.51±39.31	64.36±35.35	0.96
Family income, per person per year, %			0.38			0.06
Low	24.81	21.40		66.67	57.91
Middle	42.64	38.91		26.50	35.04
High	32.56	39.69		6.84	7.05
Education level, %			0.06			0.13
Elementary school or less	44.96	41.63		11.11	11.32
Middle school	26.36	24.12		37.18	36.97
High school	23.26	19.46		36.32	29.70
College or above	5.43	14.79		15.38	22.01
Ever smoking, %	3.10	2.72	0.83	72.22	65.81	0.09
Ever alcohol drinker, %	2.33	1.95	0.80	30.77	30.56	0.95
History of hepatitis/chronic liver disease, %	21.71	3.89	<0.001	45.30	10.26	<0.001
Family history of liver cancer, %	9.30	1.17	<0.001	11.54	4.70	<0.001
Diabetes, %	11.63	7.00	0.13	11.97	9.19	0.25
Cholelithiasis, %	19.38	14.79	0.25	14.10	10.90	0.22
Cholecystectomy, %	10.08	7.78	0.45	2.56	3.63	0.45
Menopausal status, %	76.74	76.65	0.98
History of oral contraceptive, %	20.93	20.62	0.94
HBsAg, %	41.09	3.57	<0.001	67.52	7.26	<0.001
HBeAg, %	17.83	0	<0.001	24.36	0.43	<0.001

¹Continuous variables are presented as the mean ± the standard deviation.

²The general linear model was used to calculate P for continuous variables: age at recruitment, body mass index, total physical activity. Chi-square tests were used to calculate P for binary variables: ever smoking (yes or no); alcohol drinker (yes or no); history of viral hepatitis/chronic liver disease (yes or no); family history of liver cancer (yes or no); history of diabetes (yes or no); history of cholelithiasis (yes or no), cholecystectomy (yes or no), menopausal status (yes or no), oral contraceptive use(yes or no), HBsAg (yes or no), and HbeAg (yes or no). All statistical tests were two-sided. Cochran–Mantel–Haenszel statistics were used to calculate P for ordinal categorical variables: family income level (low income for <5,000 yuan/year in the SWHS and <12,000 yuan/year in the SMHS; medium income for 5,000 to <10,000 yuan/year in the SWHS and 12,000 to <24,000 yuan/year in the SMHS; and high income for ≥10,000 yuan/year in the SWHS and ≥24 000 yuan/year in the SMHS); education level (elementary school or less, middle school, high school, college or above).

In 468 middle-aged male controls, subjects with higher age (P=0.02), never smoking (P<0.001), and never drinking (P=0.005) had lower level of 15-F_2t-IsoP (Table 2).

Table 2.

The distributions of urinary 15-F_2t-IsoP levels in selected baseline characteristics by gender (in control subjects)

Characteristic	Women (ng/mg Cr)					Men (ng/mg Cr)
	<0.166	0.166-0.327	0.328-0.559	≥0.560	P1	<0.230	0.230-0.411	0.412-0.689	≥0.690	P1
Age, %					0.31					0.02
<50	18.75	28.13	27.69	15.63		13.68	17.95	16.24	27.35
50~59	26.56	23.44	20.00	17.19		28.21	31.62	37.61	36.75
≥60	54.69	48.44	52.31	67.19		58.12	50.43	46.15	35.90
BMI, %					0.55					0.92
<25	35.94	46.88	44.62	46.88		33.33	30.77	31.62	29.06
≥25	64.06	53.13	55.38	53.13		66.67	69.23	68.38	70.94
Physical activity, %					0.43					0.23
Low	29.69	31.25	40.00	31.25		19.66	22.22	26.50	31.62
Middle	29.69	31.25	27.69	42.19		36.75	42.74	36.75	29.06
High	40.63	37.50	32.31	26.56		43.59	35.04	36.75	39.32
Income, %					0.33					0.21
Low	20.31	18.75	18.46	28.13		49.57	55.56	65.81	60.68
Middle	29.69	45.31	44.62	35.94		41.03	36.75	30.77	31.62
High	50.00	35.94	36.92	35.94		9.40	7.69	3.42	7.69
Education, %					0.33					0.19
Elementary school or less	40.63	35.94	35.38	54.69		10.26	10.26	11.11	13.68
Middle school	21.88	31.25	29.23	14.06		32.48	35.90	43.59	35.90
High school	23.44	18.75	16.92	18.75		27.35	31.62	23.93	35.89
College or above	14.06	14.06	18.46	12.50		29.91	22.22	21.37	14.53
Occupation, %					0.67					0.32
Professional	31.25	32.81	29.23	20.31		32.48	33.33	28.21	23.08
Clerical	18.75	14.06	20.00	17.19		24.79	18.80	17.09	23.08
Manufacture/labour	50.00	53.13	50.77	62.50		42.74	47.86	54.70	53.85
OC use in female, smoking in male, %	25.00	20.31	20.00	17.19	0.75	54.70	60.68	66.67	81.20	<0.001
Menopause in female, Alcohol consumption in male, %	75.00	70.31	75.38	85.94	0.20	20.51	29.91	29.91	41.88	0.005

¹Using chi-square test

We observed a significant heterogeneity by gender on the association of liver cancer risk with different level of 15-F_2t-IsoP (P for heterogeneity=0.033) (Table 3). Higher 15-F_2t-IsoP level was strongly associated with an increased risk of liver cancer, with an adjusted OR in men (OR_{4th quartile versus 1st quartile}=8.84, 95% CI 2.74-28.60), which was significantly higher than that in women (OR_{4th qarntile versus 1st qarntile}=1.75, 95% CI 0.70-4.42). Furthermore, risk of liver cancer was positively related to an increasing level of 15-F_2t-IsoP in a dose-response manner in men but not in women (P_trend for male<0.001, P_trend for female=0.206).

Table 3.

Odds ratios for liver cancer by quartiles of urinary15-F_2t-IsoP levels in two nested case-control studies (SWHS and SMHS)

Urinary 15-F2t-IsoP1	Women				Men				P heterogeneity
	Range	Cases(n=129)/ controls(n=257)	OR (95% CI)2	OR (95% CI)3	Range	Cases(n=234)/ controls(n=468)	OR(95% CI)2	OR (95% CI)3
Quartiles	ng/mg Cr				ng/mg Cr				0.033
1	<0.166	13/64	1.00 (reference)	1.00 (reference)	<0.230	29/117	1.00 (reference)	1.00 (reference)
2	0.166-0.327	32/64	2.18 (1.04,4.56)	1.21 (0.47,3.15)	0.230-0.411	43/117	1.80 (1.01,3.21)	2.75 (1.00,7.59)
3	0.328-0.559	34/65	2.12 (1.01,4.47)	1.53 (0.52,4.48)	0.412-0.689	53/117	2.33 (1.32,4.13)	2.92 (0.99,8.64)
4	≥0.560	50/64	2.92 (1.46,5.84)	1.75 (0.70,4.43)	≥0.690	109/117	5.52 (3.08,9.89)	8.84 (2.74, 28.57)
Ptrend			0.006	0.272			<0.001	<0.001

¹Matched by age (±2 years), sex, date (±30 days) of biospecimen collection, interval since last meal (<2 hr), time (morning or afternoon) of urine sample collection, menopausal status (for female)

²Results were adjusted for batch

³Results were adjusted for batch, BMI, physical activity, smoking (for male), alcohol consumption (for male), income, education, total energy intake, total vegetable intake, total fruit intake, total red meat intake, total egg intake, total fish intake, total soy intake, total vitmin E intake, history of hepatitis/chronic liver disease, family history of liver cancer, diabetes, cholelithiasis, cholecystectomy, oral contraceptive use (for female), HBsAg, HBeAg

The combined effects of urinary 15-F_2t-IsoP level and HBsAg were given in Table 4 and Table 5. The HBsAg carriers with urinary 15-F_2t-IsoP level above the median value had an significantly increased risk of liver cancer (OR_female=59.04, 95% CI 12.26, 284.30; OR_male=92.55, 95% CI 34.83, 245.96) compared with non-carriers with urinary 15-F_2t-IsoP level below the median value (P_interaction for female=0.016, P_interaction for male=0.689).

Table 4.

Combined effect of HBV infection and urinary 15-F_2t-IsoP levels on risk of liver cancer in women

HBsAg1	Urinary 15-F2t-IsoP (ng/mg creatinine) 1	Case	Control	OR (95% CI) 2	OR (95% CI) 3
Negative	<0.363	32	121	1.00 (reference)	1.00 (reference)
Negative	≥0.363	44	127	1.04(0.58,1.85)	1.03(0.56,1.88)
Positive	<0.363	13	7	6.60(2.07,21.00)	5.98(1.81,19.82)
Positive	≥0.363	40	2	57.96(12.17,276.07)	59.04(12.26,284.30)
Pinteraction				0.022	0.016

¹Matched by age (±2 years), sex, date (±30 days) of biospecimen collection, interval since last meal (<2 hr), time (morning or afternoon) of blood sample collection, menopausal status

²Results were adjusted for batch

³Results were adjusted for batch, BMI, physical activity, income, education, oral contraceptive use

Table 5.

Combined effect of HBV infection and urinary 15-F_2t-IsoP levels on risk of liver cancer in men

HBsAg1	Urinary 15-F2t-IsoP (ng/mg creatinine) 1	Case	Control	OR (95% CI) 2	OR (95% CI) 3
Negative	<0.481	26	248	1.00 (reference)	1.00 (reference)
Negative	≥0.481	50	186	2.97(1.61,5.50)	3.01(1.59,5.68)
Positive	<0.481	58	19	48.78(17.32,137.37)	50.33(17.54,144.43)
Positive	≥0.481	100	15	85.46(33.76,216.31)	92.55(34.83,245.96)
Pinteraction				0.670	0.689

¹Matched by age (±2 years), sex, date (±30 days) of biospecimen collection, interval since last meal (<2 hr), time (morning or afternoon) of blood sample collection

²Results were adjusted for batch

³Results were adjusted for batch, BMI, physical activity, smoking, alcohol consumption, income, education

Discussion

In this nested case-control study of women and men in Shanghai, we found that the increased level of urinary 15-F_2t-IsoP was strongly associated with higher risk of liver cancer in a dose-response manner. As the 15-F_2t-IsoP level rising, the risk of liver cancer in men increased more steeply, which imply a quite different risk effects respectively in men and women. The HBsAg carriers with high urinary 15-F_2t-IsoP level had much higher liver cancer risk than non-carriers with low urinary 15-F_2t-IsoP level.

The association between level of 15-F_2t-IsoP and liver cancer risk seems to be biologically plausible. The 15-F_2t-IsoP is an accurate marker of oxidative stress in humans^17–19. Accumulating evidence has shown that oxidative stress plays an important role in the development of liver cell carcinogenesis through disrupting either normal cell function or genetic materials^28–31. Liver injury can be either an acute or a chronic inflammatory process. In the environment of local inflammation, many types of liver cells, such as liver sinusoidal endothelial cells, hepatic stellate cells, dendritic cells, and kupffer cells, are activated³². These cells produce many kinds of immune mediators, cytokines, and chemokines^33–35. On the other hand, DNA damage induced by oxygen free radicals and DNA repair of the adaptation disorder lead to accumulation of cancer-related gene mutations³². There is much evidence that chronic inflammation is one of causes for human cancer³⁶. Oxidative stress and accumulation of DNA damage play an important role in the process of virus-induced cancer³⁷. The association of 15-F_2t-IsoP and liver cancer has not been deeply investigated. The only study that prospectively evaluated the urinary 15-F_2t-IsoP and hepatocellular carcinoma (HCC) risk included 74 incident HCC cases and 290 matched controls, suggesting that elevated levels of urinary 15-F_2t-IsoP may be related to an increasing level of aflatoxin exposure and are associated with an increased risk of HCC¹⁴. Our study, the largest study on the topic, valid the finding of the early study and further we found liver cancer risk rises more steeply in men than in women and the HBsAg carriers with higher urinary 15-F_2t-IsoP level had even higher liver cancer risk.

A notable gender disparity was observed on the dose-response association between 15-F_2t-IsoP level and liver cancer risk in present study. Increasing evidence suggested that sex hormones such as androgens and estrogens played an important role in the development of liver cancer and regulating the level of oxidative stress³⁸. It was supposed that androgen may stimulate the androgen signaling pathway and cooperate to the increased transcription and replication of HBV genes³⁸. In contrast, estrogen may play a protecting role against the development of liver cancer through decreasing HBV RNA transcription and inflammatory cytokines levels³⁸. Systemic oxidative stress may originate from accumulated fat or androgen-stimulated circulating mononuclear cells^{39, 40}. ROS generation from mononuclear cells is increased in polycystic ovarian syndrome women independent of obesity but correlated to testosterone levels³⁹. Thus, excess androgen in females may provoke systemic oxidative stress via androgen receptor action on mononuclear cell³⁹. In contrast, increasing evidence suggested that estrogens played an important role in the process of antioxidant stress⁴¹. It was supposed that estrogens might exert cardioprotective effects through up-regulation of cystathionine-γ-lyase expression and hydrogen sulfide generation, which were associated with decreased oxidative stress and inflammatory status⁴¹.

Chronic HBV infection is a major risk factor for liver cancer. The continuous generation of ROS due to chronic inflammation may be one of the mechanisms that HBV induces liver cancer^{42, 43}. In our study, individuals with a higher urinary 15-F_2t-IsoP level and chronic HBV infection had a much higher risk of developing liver cancer than those with a lower urinary 15-F_2t-IsoP level and chronic HBV infection subjects when comparing to individuals with a lower urinary 15-F_2t-IsoP level and without HBV infection, indicating a synergetic effect. Even though we detect effect modification of HBV and 15-F_2t-IsoP in women, the number of subjects in controls with HBsAg positive is too small (nine controls with HBV infection) (Table 4). So the effect for HBsAg carriers with higher urinary 15-F_2t-IsoP level and HBsAg carriers with lower urinary 15-F_2t-IsoP level comparing with non-carriers with lower urinary 15-F_2t-IsoP level, and the effect modification, is not relatively convinced in women.

In the present study, we have found a positive association between smoking or alcohol intake and the level of urinary 15-F_2t-IsoP. Alcohol consumption may result in increased oxidative stress inducing the development of alcoholic liver damage due to ROS generation through stimulation of cytochrome P450 2E1⁴⁴. Substantially higher doses of alcohol consumption have been found required to produce a statistically significant elevation in urinary 15-F_2t-IsoP⁴⁵. Smoking is found to be associated with an increased level of 15-F_2t-IsoP in some prior studies^{20, 46}, which was consistent with the results of our study.

Although there are some advanced detection methods which can measure the level of urinary 15-F_2t-IsoP, such as gas chromatography mass spectrometry (GC-MS), liquid chromatograph mass spectrometer (LC-MS)⁴⁷, the enzyme-linked immune-sorbent assay (ELISA) technique is more reliable and suitable for large size of population-based study⁴⁸. Previous study has shown an acceptable reproducibility between GC-MS and ELISA for measuring urinary 15-F_2t-IsoP⁴⁹.

Our study has a number of important strengths. First, the SWHS and SMHS are both rigorously designed cohort studies with high participation and retention rates. Second, all the covariates used in analysis were assessed before the development of any cancer which decreases potential reverse bias. Third, our nested case-control study incorporated both biomarkers of HBsAg and HBeAg into the analysis to minimize the possibility of hepatitis B virus infection affecting the association.

The present study has some limitations. First, our quantification of 15-F_2t-IsoP used a single urinary sample obtained at study entry; thus, the changes in level of 15-F_2t-IsoP over time during follow-up could not be assessed. Second, potential confounders such as intake of aflatoxin B1, hepatitis C virus infection and acceptance of antiviral therapy were not considered in the present study, which may bias the association between 15-F_2t-IsoP level and liver cancer risk. However, exposure rates for these two factors are low in our study participants^{22, 23, 26}.

In summary, we found a strong positive association between pre-diagnostic urinary level of 15-F_2t-IsoP and risk of liver cancer in the population-based cohorts of residents in urban Shanghai, indicating that higher level of oxidative stress may lead to higher liver cancer risk. With 15-F_2t-IsoP level increasing, liver cancer risk rises more steeply in men than in women, for both with and without HBV infection. The HBsAg carriers with higher urinary 15-F_2t-IsoP level had even higher liver cancer risk, suggesting that combine sero-positive HBsAg and urinary assay for 15-F_2t-IsoP may identify high risk population for liver cancer for a close surveillance and primary prevention.

Novelty and Impact.

Currently, the role of oxidative stress in the development of liver cancer is unclear. Our study investigated the association between 15-F_2t-isoprostane, an accurate marker of oxidative stress, and liver cancer risk, using sectional data from two large population-based cohorts in China. It was found that higher urinary level of 15-F_2t-isoprostane may lead to higher liver cancer risk. The result may have important implications for a close surveillance and primary prevention of liver cancer.

Abbreviations

SMHS

Shanghai Men’s Health Study

SWHS

Shanghai Women’s Health Study

15-F_2t-IsoP

15-F_2t-isoprostane

ROS

reactive oxygen species

ICD-9

the International Classification of Disease, Ninth Revision

HBsAg

hepatitis B surface antigen

HBeAg

hepatitis B e antigen

OR

odds ratio

CI

Confidence interval

HCC

hepatocellular carcinoma

GC-MS

gas chromatography mass spectrometry

LC-MS

liquid chromatograph mass spectrometer

BMI

body mass index

ELISA

enzyme-linked immuno sorbent assay