Plasma cholesterol and lipoprotein levels in relation to tumor aggressiveness and survival in HCC patients

Abstract

Background and Aims:

Hepatocellular carcinoma is associated with several chronic liver diseases, especially chronic hepatitis B virus, hepatitis C virus, and alcoholism. It is increasingly appreciated that obesity/metabolic syndrome is also associated with chronic liver disease and subsequent hepatocellular carcinoma.

Methods:

We retrospectively investigated the serum lipid profiles in a large hepatocellular carcinoma cohort, associated predominantly with the hepatitis B virus, hepatitis C virus, alcohol or nonalcoholic steatohepatitis. The cohort was examined both as a whole, as well as stratified by etiology.

Results:

We found significant associations between parameters of hepatocellular carcinoma biology such as maximum tumor diameter, portal vein thrombosis, tumor multifocality or alpha-fetoprotein levels and individual lipid components, including total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, triglycerides and body mass index. In a final multiple linear regression model considering all lipid variables together, only high-density lipoprotein cholesterol was significantly associated with the tumor Tumor Aggressiveness Index. High-density lipoprotein cholesterol was found to have a statistically higher hazard ratio for death than low high-density lipoprotein cholesterol levels (Cox). On examination by etiological group, alpha-fetoprotein levels were significantly higher in patients with hepatitis C virus compared to those with alcohol or nonalcoholic steatohepatitis, but maximum tumor diameter, tumor multifocality and portal vein thrombosis were similar across etiological groups. Nonalcoholic steatohepatitis patients had significantly less cirrhosis than other groups and hepatitis B virus patients had significantly higher cholesterol and low-density lipoprotein cholesterol levels than hepatitis C virus patients.

Conclusions:

This is the first report, to our knowledge, of a relationship between serum lipid parameters and indices of hepatocellular carcinoma growth, invasion and aggressiveness, as well as with survival.

Keywords

Hepatocellular carcinoma lipids aggressiveness survival

Introduction

Alterations in blood lipid profiles and metabolism have been described in the presence of chronic hepatitis infection, cirrhosis, and hepatocellular carcinoma (HCC), and changes in metabolic profiling have even been suggested to be useful for HCC screening.^1-7 Multiple changes in the lipid composition of cancer cells, including HCC have also been well described.^1,8-13 Several approaches to the correction of altered lipid levels in non-alcoholic fatty liver disease have been reported.^14-17 The global epidemic of obesity has recently shifted attention to the increased incidence of obesity-associated and metabolic syndrome¹⁸-associated non-alcoholic fatty liver disease (NAFLD) and obesity-driven cancers and HCC,^19-23 that may occur in the presence or in the absence of cirrhosis.^22-24

In the present study, we report an analysis of blood lipid profile parameters from a large cohort of well-studied HCC patients who also had survival information and we show an association between lipid profile parameters and HCC aggressiveness. Furthermore, the blood lipid profile patterns differ HCC patients associated with the hepatitis B virus (HBV), the hepatitis C virus (HCV), alcohol, or nonalcoholic steatohepatitis (NASH).

Methods

Data collection

We retrospectively analyzed prospectively collected data in the Italian Liver Cancer (ITA.LI.CA) study group database of 521 HCC patients accrued till 2008 at 11 centers²⁵ who had full baseline tumor parameter data, including computed tomography (CT) scan information on the maximum tumor diameter (MTD); the number of tumor nodules; the presence of portal vein thrombosis (PVT) and plasma alpha-fetoprotein (AFP) levels; blood counts; routine blood liver function tests, (total bilirubin, gamma glutamyl transpeptidase (GGTP), albumin); demographics and survival information; and those who had lipid profile data and a single known etiological cause. ITA.LI.CA database management conforms to Italian legislation on privacy, and this study conforms to the ethical guidelines of the Declaration of Helsinki. Approval for this retrospective study on de-identified HCC patients was obtained by the Institutional Review Board of participating centers.

Tumor Aggressiveness Index

The Tumor Aggressiveness Index²⁶ was calculated as the sum of scores:

MTD (in tertiles): MTD<4.5; 4.5 ≤ MTD ≤ 9.6; MTD>9.6; for scores 1, 2, 3, respectively;

AFP (cutoff): AFP<100; 100 ≤ AFP ≤ 1000; AFP>1000; for scores 1, 2, 3, respectively;

PVT (No/Yes): PVT(No); PVT(Yes); for scores 1 and 3 respectively; Nodules (number): Nodules ≤ 3; Nodules>3; for scores 1 and 3 respectively.

Statistical methods

Characteristics of HCC patients were described by using mean (±SD) or proportions (%) for continuous or categorical variables, respectively.

For analytic purposes some variables were dichotomized: glycemia (< 100 or ≥100 mg/dL); total cholesterol (<200 or ≥200 (mg/dL), low-density lipoprotein (LDL) cholesterol (<160 or ≥160 mg/dL), high-density lipoprotein (HDL) cholesterol (>50(F), >40(M) or <50(F),<40(M) mg/dL), triglycerides (<150 or ≥150 mg/dL) and body mass index (BMI) (<30 or ≥30 kg/m²). Differences between means were tested by using Wilcoxon rank-sum, whereas a proportions test was used to test differences for the categorical variable.

Multiple linear regression on dichotomized variables was applied to the data.

To explore differences in metabolic parameters an Tumor Aggressiveness Index was built. The index reflects the sum of each single score as follows: MTD (in terciles): MTD<4.5; 4.5≤MTD≤9.6; MTD>9.6; scores 1, 2, 3, respectively; AFP (cutoff): AFP<100; 100≤AFP≤1000; AFP>1000 ng/mL; scores 1, 2, 3, respectively; PVT (No/Yes): PVT (No); PVT (Yes); scores 1 and 3, respectively; Tumor Nodules (number): Nodules≤3; Nodules>3; scores 1 and 3, respectively.

For survival analysis, the HDL cholesterol ≥50(F), ≥40(M) or <50(F), <40(M) mg/ was used.

The non-parametric Kaplan–Meier method was applied to explore survival probabilities for each Metabolic Index category and to provide estimates of the probability of death. Cox’s Model was then fitted to the data. The proportional hazard assumption was evaluated by means of Schoenfeld residuals (SRT). Model fitting was evaluated by means of Akaike Information Criteria (AIC) and Bayesian information criterion (BIC). Risk estimators are expressed as hazard ratios (HR) and 95% confidence interval (95% CI).

The final multiple linear regression model of the Tumor Aggressiveness Index score or the Cox proportional hazard models for death were used in the stepwise method, and all variables were examined as categorical.

All Cox proportional hazard models for death were adjusted for age and gender.

All statistical analyses were performed by using Stata statistical software, version 14.2 (StataCorp LP, College Station, TX, USA).

Results

Metabolic parameters in relation to tumor aggressiveness parameters

The focus of this study was on 521 HCC patients who had lipid profile data collected and a known single etiological cause of their disease. Child Pugh A cirrhosis was evident in 74.8% of the patients, and 14.9% had Child Pugh B cirrhosis. An examination was made for any association between each of the six metabolic profile components (glycemia, total cholesterol, HDL, LDL, triglycerides and BMI) and each or all of the four parameters of HCC aggressiveness (Tumor Aggressiveness Index), namely MTD, the percentage of patients with PVT, mean AFP levels, or percentage of patients with tumor multifocality (Table 1). MTD was significantly associated with change in total cholesterol and LDL cholesterol. PVT was significantly associated only with HDL cholesterol. AFP levels were significantly associated with both HDL cholesterol and LDL cholesterol. Thus, each of the four tumor parameters was associated with a different pattern of lipids. The combination of all four tumor aggressiveness parameters together (the Tumor Aggressiveness Index^25,26) was only significantly associated with HDL cholesterol levels (Table 1).

Table 1.

Plasma glycemia, HDL, triglycerides, total cholesterol, LDL cholesterol, HDL cholesterol, and BMI in HCC patients in relation to tumor parameters.

Variables ^ψ	Glycemia (mg/dL)		P ^#
Variables ^ψ	<100	≥100	P ^#
MTD (cm)	3.85±2.75	3.97±2.73	0.27
Vascular invasion (%)	13.42	11.89	0.56 ^{^}
AFP (ng/mL)	3554.54±25697.34	1466.54±7629.56	0.57
Multifocality (%)	43.21	42.33	0.83 ^{^}
Tumor Aggressiveness Index (4<Score≤12) (%)	43.42	50.18	0.11 ^{^}
	HDL(mg/dL)
Variables ^ψ	[>=50(F)]/[>=40(M)]	[<50(F)]/[<40(M)]	P ^#
MTD (cm)	3.69±2.20	4.31±3.34	0.14
Vascular invasion (%)	8.51	19.03	0.0002 ^{^}
AFP (ng/mL)	1946.69±21744.49	3165.52±12334.72	0.001
Multifocality (%)	41.42	45.49	0.33 ^{^}
Tumor Aggressiveness Index (4<Score≤12) (%)	43.43	53.10	0.02 ^{^}
	Triglycerides, (mg/dL)
Variables ^ψ	<150	≥150	P ^#
MTD (cm)	3.85±2.42	4.58±4.26	0.62
Vascular invasion (%)	13.26	11.76	0.69 ^{^}
AFP (ng/mL)	2415.76±19415.16	2930.66±13320.33	0.56
Multifocality (%)	43.92	37.66	0.29 ^{^}
Tumor Aggressiveness Index (4<Score≤12) (%)	47.13	47.37	0.97 ^{^}
	Total Cholesterol (mg/dL)
Variables ^ψ	<200	≥200	P ^#
MTD (cm)	3.83±2.58	4.53±3.38	0.02
Vascular invasion (%)	13.08	11.22	0.60 ^{^}
AFP (ng/mL)	2194.41±19081.13	3843.89±15098.95	0.43
Multifocality (%)	43.50	40.22	0.56 ^{^}
Tumor Aggressiveness Index (4<Score≤12) (%)	46.25	53.01	0.26 ^{^}
	LDL Cholesterol (mg/dL)
Variables ^ψ	<160	≥160	P ^#
MTD (cm)	3.88±2.62	5.37±4.37	0.07
Vascular invasion (%)	12.88	14.29	0.84 ^{^}
AFP (ng/mL)	2170.44±18325.79	8916.07±24240.56	0.03
Multifocality (%)	42.86	50.00	0.48 ^{^}
Tumor Aggressiveness Index (4<Score≤12) (%)	46.89	50.00	0.77 ^{^}
	BMI (kg/m²)
Variables ^ψ	<30	≥30	P ^#
MTD (cm)	3.71±2.45	4.09±4.04	0.77
Vascular invasion (%)	35 (12.96)	6 (11.54)	0.77 ^{^}
AFP (ng/mL)	4085.97±28161.51	4206.30±1901.02	0.83
Multifocality (%)	100 (39.53)	16 (33.33)	0.41 ^{^}
Tumor Aggressiveness Index (4<Score≤12) (%)	91 (40.27)	19 (40.43)	0.98 ^{^}

AFP, alpha-fetoprotein; BMI: body mass index; HDL: high-density lipoprotein; LDL: low-density lipoprotein; MTD, maximum tumor diameter; PVT, portal vein thrombosis.

All values: M±SD or No. of Patients (%).

Wilcoxon rank-sum (Mann–Whitney) test; ^{^} Test for proportions;

Tumor Aggressiveness Index as sum of scores:

MTD (in terciles): MTD<4.5; 4.5≤MTD≤9.6; MTD>9.6; scores 1, 2, 3, respectively;

AFP (cutoff): AFP<100; 100≤AFP≤1000; AFP>1000 ng/ml; scores 1, 2, 3, respectively;

PVT (No/Yes): PVT (No); PVT (Yes); scores 1 and 3, respectively;

Tumor nodules (number): Nodules≤3; Nodules>3; scores 1 and 3, respectively.

A linear regression model of the Tumor Aggressiveness Index was then calculated on single metabolism variables of glycemia, HDL cholesterol, LDL cholesterol, total cholesterol, triglycerides, and BMI. Significance was found only for HDL cholesterol, both as a single variable (Table 2A) and in the final linear regression model, using the stepwise method (Table 2B). The calculation was repeated after excluding patients with total bilirubin levels >2.0mg/dL, with similar results (Table 2C).

Table 2.

Linear regression model of the Tumor Aggressiveness Index score on single variables glycemia, HDL, triglycerides, total cholesterol, LDL, and BMI (A).

Final multiple linear regression model in the stepwise method of the Tumor Aggressiveness Index score on glycemia, HDL, triglycerides, total cholesterol, LDL, and BMI, all in categories and included together in the model (B), and the final multiple linear regression model in subjects with Bilirubin<2.0 (mg/dL) (C).

	β	se(β)	p	95% CI
(A)
Glycemia (mg/dL)
<100 (Ref. category)	–
≥100	−0.11	0.20	0.59	−0.50, 0.28
HDL(ng/mL)
[≥50(F)]/[≥40(M)] (Ref. category)	–
[<50(F)]/[<40(M)]	0.39	0.20	0.05	−0.01, 0.79
Triglycerides, (mg/dL)
<150 (Ref. category)	–
≥150	−0.002	0.29	0.99	−0.57, 0.57
Total Cholesterol (mg/dL)
<200 (Ref. category)	–
≥200	0.14	0.26	0.59	−0.38, 0.66
LDL Cholesterol (mg/dL)
<160 (Ref. category)	–
≥160	0.43	0.45	0.34	−0.45, 1.32
BMI (kg/m²)
<30 (Ref. category)	–
≥30	−0.05	0.26	0.85	−0.57, 0.47
(B)
HDL(ng/mL)
[≥50(F)]/[≥40(M)] (Ref. category)	–
[<50(F)]/[<40(M)]	0.39	0.20	0.05	−0.01, 0.79
(C)
HDL(ng/mL)
[≥50(F)]/[≥40(M)] (Ref. category)	–
[<50(F)]/[<40(M)]	0.25	0.22	0.25	−0.18, 0.68

AFP: alpha-fetoprotein; β: coefficient; BMI: body mass index; F: female; HDL, high-density lipoprotein; LDL, low-density lipoprotein; M: male; MTD, maximum tumor diameter; se(β): standard error of coefficient.

Tumor Aggressiveness Index as the sum of scores:

MTD (in terciles): MTD<4.5; 4.5≤MTD≤9.6; MTD>9.6; scores 1, 2, 3, respectively;

AFP (cut-off): AFP<100; 100≤AFP≤1000; AFP>1000 ng/ml; scores 1, 2, 3, respectively;

PVT (No/Yes): PVT (No); PVT (Yes); scores 1 and 3, respectively;

Tumor Nodules (number): Nodules≤3; Nodules>3; scores 1 and 3, respectively.

Survival and HDL cholesterol

The survival of patients with all of the lipid parameters as single variables was examined, using the Cox proportional hazard method for death (Table 3). The highest HR for death was found for high HDL cholesterol (HR 1.52), and was significant at the P<0.001 level. In the final model, HDL cholesterol was significant for death, HR=1.69, P=0.003. Furthermore, after excluding patients with total bilirubin levels >2.0 mg/dL, HDL cholesterol was still found to be significant for death (Table 3C).

Table 3.

Cox proportional hazard models for death, on single variables (A).

Final multiple Cox proportional hazard models for death, in the stepwise method, on all variables included together in the model (B), and the final multiple Cox proportional hazard models in subjects with bilirubin<2.0 (mg/dL) (C).

All models were adjusted for age and gender, and all variables in categories.

	HR	se(HR)	P	95% CI
(A)
Glycemia (mg/dL)
< 100 [Ref. category]	1	–	–	–
≥ 100	1.03	0.11	0.77	0.83, 1.28
HDL Cholesterol (mg/dL) *
[≥50(F)] / [≥40(M)] [Ref. category]	1	–	–	–
[<50(F)] / [<40(M)]	1.52	0.17	<0.001	1.22, 1.89
Triglycerides, (mg/dL)
< 150 [Ref. category]	1	–	–	–
≥ 150	0.90	0.15	0.53	0.64, 1.25
Total Cholesterol (mg/dL)
< 200 [Ref. category]	1	–	–	–
≥ 200	0.80	0.12	0.15	0.58, 1.09
LDL cholesterol (mg/dL)
< 160 [Ref. category]	1	–	–	–
≥ 160	1.11	0.28	0.68	0.68, 1.81
BMI (kg/m²)
< 30 [Ref. category]	1	–	–	–
≥ 30	0.84	0.21	0.48	0.52, 1.36
(B)
HDL cholesterol (mg/dL) *
[≥50(F)] / [≥40(M)] [Ref. category]	1	–	–	–
[<50(F)] / [<40(M)]	1.69	0.30	0.003	1.20, 2.38
(C)
HDL cholesterol (mg/dL) *
[≥50(F)] / [≥40(M)] [Ref. category]	1	–	–	–
[<50(F)] / [<40(M)]	1.38	0.30	0.13	0.91, 2.11

BMI: body mass index; F: female; HR: hazard ratio; HDL: high-density lipoprotein; LDL: low-density lipoprotein; M: male.

Dichotomization: HDL Cholesterol: 0, [≥50(F)]/[≥40(M)]; 1, [<50(F)]/[<40(M)].

Likewise, in a Kaplan–Meier survival probability curve (Figure 1), low HDL cholesterol was shown to be significant for enhanced survival (Log-rank test, P=0.0009 and Wilcoxon test, P=0.0001). The sensitivity and specificity of HDL for death was 41.76% (CI 37.98%, 45.55%) and 61.41% (CI 57.68%, 45.55%) in the total cohort. Similar calculations for patients with bilirubin <2.0mg/dL were: sensitivity and specificity of HDL for death: 36.14% (CI 32.03%, 40.26%) and 62.04% (CI 57.88%, 66.20%). The positive predictive value in the total cohort was 54.2% and the negative predictive value was 49.1%.

Figure 1.

Kaplan–Meier survival probability among HCC patients categorized according to HDL (ng/mL). Log-rank test: P=0.002; Wilcoxon (Breslow) test: P=0.0003; test for the trend of survivor functions across groups: P=0.004.

Patient parameters according to HCC etiology

Since there may be different patterns of lipid metabolism amongst the various etiological types of HCC and conversely, and since lipids can also alter HCC biology, we examined the clinical and lipid profiles of the same patients as above, but stratified them according to single disease of etiology, as shown in Table 4 (patients with two diseases together, such as HBV plus HCV, were excluded). NASH patients had a lower percentage of cirrhosis (80.33%) compared with HBV, HCV, and alcohol patients (all >90%), and had a greatly increased percentage of patients with diabetes type II (NASH 63.33, alcohol 43.59, HCV 26.65, and HBV 18.18), P<0.001. NASH patients had a similarly much higher percentage of patients with obesity (NASH 54.55, alcohol 17.50, HCV 6.72, HBV 8.57), P <0.001. NASH and alcohol-associated patients both had significant glycemia compared with patients with HBV or HCV. HBV patients had significantly higher levels of both total cholesterol and LDL cholesterol than HCV patients, P <0.001. However, neither HDL levels nor triglyceride levels differed significantly among the etiological groups.

Table 4.

Hepatocellular carcinoma patient parameters according to etiology.

Variable ^ψ	Etiology				p^*	Comparisons p^§
	HBV(n=69)	HCV (n=272)	Alcohol (n=119)	NASH/other (n=61)		Comparisons p^§
	(a)	(b)	(c)	(d)		(a)vs(b)	(a)vs(c)	(a)vs(d)	(b)vs(c)	(b)vs(d)	(c)vs(d)
Age (y)	61.65±12.13	70.47±10.16	66.92±9.31	69.72±9.46	0.0001	<0.0001	0.003	<0.0001	0.0002	0.31	0.04
Sex (Men) (%)	61 (88.41)	167 (61.40)	106 (89.08)	55 (90.16)	<0.001	<0.0001	0.89	0.75	<0.0001	<0.0001	0.82
Cirrhosis (%)	61 (91.04)	264 (97.78)	112 (94.12)	49 (80.33)	<0.001	0.06	0.46	0.08	0.12	0.0009	0.01
HCC (%)					0.18
Uninodular	35 (53.03)	155 (57.41)	52 (43.70)	34 (55.74)		–	–	–	–	–	–
Multinodular	25 (37.88)	97 (35.93)	54 (45.38)	23 (37.70)		–	–	–	–	–	–
Infiltrating	5 (7.58)	16 (5.93)	8 (6.72)	4 (6.56)		–	–	–	–	–	–
Massive	1 (1.52)	2 (0.74)	5 (4.20)	0 (0.00)		–	–	–	–	–	–
Vascular invasion (%)	7 (10.14)	31 (11.44)	17 (14.41)	8 (13.33)	0.79	–	–	–	–	–	–
MTD (cm)	4.25±3.68	3.52±2.06	4.16±2.82	4.06±2.07	0.10	0.24	0.95	0.51	0.13	0.02	0.41
AFP (ng/mL)	3147.12±14761.47	3426.30±26928.59	1653.90±7715.28	2031.67±8454.68	0.0004	0.18	0.47	0.12	0.002	0.0002	0.26
Extrahepatic spread (%)	0 (0.00)	12 (4.56)	6 (5.41)	7 (12.28)	0.02	0.0004	0.01	0.005	0.74	0.09	0.16
Obesity (yes) (%)	3 (8.57)	9 (6.72)	7 (17.50)	18 (54.55)	<0.001	0.72	0.25	<0.0001	0.10	<0.0001	0.0006
Diabetes Type 2 (%)	12 (18.18)	69 (25.65)	51 (43.59)	38 (63.33)	<0.001	0.17	0.0001	<0.0001	0.0008	<0.0001	0.01
Glycemia (mg/dL)	104.37±31.78	108.47±37.60	126.50±52.15	126.96±55.40	0.0001	0.36	0.002	0.001	0.0009	0.002	0.62
Cholesterol (mg/dL)	168.09±40.03	150.64±38.67	159.27±46.47	171.01±49.72	0.001	0.0008	0.15	0.97	0.14	0.004	0.18
HDL (mg/dL)	47.19±15.14	46.72±17.29	47.49±22.31	46.34±19.45	0.85	–	–	–	–	–	–
LDL (mg/dL)	100.56±29.52	84.50±34.63	98.99±41.89	99.86±35.44	0.0001	0.0002	0.40	0.68	0.004	0.002	0.66
Triglycerides (mg/dL)	102.76±40.53	98.85±40.60	101.67±54.21	123.12±71.42	0.15	–	–	–	–	–	–
Bilirubin (mg/dL)	1.61±2.84	1.51±2.08	2.21±2.78	1.32±1.08	0.0001	0.20	<0.0001	0.53	<0.0001	0.61	0.0001

AFP: alpha-fetoprotein; HBV: hepatitis B virus; HCC: hepatocellular carcinoma; HCV: Hepatitis C virus; HDL: high-density lipoprotein; LDL: low-density lipoprotein; MTD: maximum tumor diameter; NASH: nonalcoholic steatohepatitis.

All values: M±SD or no. of patients (%);

Kruskal–Wallis rank test; ^§Wilcoxon rank-sum (Mann–Whitney) test.

With regard to tumor characteristics, tumor multifocality, percentage PVT, and MTD were similar across etiological groups. However, AFP levels were significantly higher in patients with HCV compared with either alcohol or NASH, but the ranges in values were large.

Discussion

Plasma lipid changes have been reported in the presence of many cancer types and have given rise to the idea that they may reflect an altered metabolism of the host milieu, which is involved in the carcinogenic process, such as inflammation-associated growth cytokines. Changes in plasma lipid profile levels have been reported in association with hepatitis infection, cirrhosis, and hepatitis-associated HCCs.^1-3,7,15,27 Furthermore, there are multiple reports of altered plasma lipid profiles in obesity- or metabolic syndrome-associated HCCs^8,9,19,37,39 Thus, both hepatitis-associated HCC and, separately, obesity/NAFLD-associated HCC, seem to occur in the presence of altered lipid metabolism.

Also, metabolic and lipid profiles are found to be altered not only in the plasma of HCC patients, but also in the tumor lipid droplets plus tumor mitochondria and tumor membranes,^16,17 in the presence or absence of obesity. Alterations in lipids have been thought to influence not only metabolism, but also gene and tumor biology.^28,29 Thus, several studies have attempted to alter the lipid composition of tumors, both experimentally and in patients,^30-32 as well as indirectly, by targeting the obesity-associated inflammation.³³ There may be different patterns of lipid metabolism amongst the various etiological types of HCC^7,8,14,15,27 and conversely, since lipids can also alter HCC biology,^6,28,29 we examined the clinical and lipid profiles of a cohort of HCC patients with defined single etiologies, to determine whether there might be any relationship between HCC biology and levels of several commonly measured plasma lipids.

In this HCC cohort, 52.21% of the patients had HCV-associated HCC, 13.24% had HBV, 22.84% had alcoholism, and 11.71% had NASH. We examined the six metabolic profile components of glycemia, total cholesterol, HDL, LDL, triglycerides, and BMI in relation to each of the four parameters of HCC aggressiveness, namely MTD, percentage patients with PVT, percentage tumor multifocality and mean AFP levels, as well as all four parameters together, called the Tumor Aggressiveness Index (Table 1). MTD was significantly associated with changes in total cholesterol and LDL cholesterol, and PVT was significantly associated only with HDL cholesterol. AFP levels were significantly associated with both HDL and LDL cholesterol. In both a linear and a final multiple linear regression model, using the stepwise method, only HDL cholesterol was found to be significantly associated with the Tumor Aggressiveness Index (Table 2). We examined the survival of patients, using the Cox proportional hazard models for death, and found that the highest HR for death was found for HDL cholesterol on both the single lipid variables and in the final model with all the variables included together. We then examined the HCC patients separately according to the four etiologies of HCV, HBV, alcohol, and NASH. NASH patients had significantly less cirrhosis than in the other etiologies, although the tumor characteristics were similar amongst different etiologies, except for AFP levels, which were higher in the HCV than in the alcohol- or NASH-associated patients (Table 4). However, the lipid profiles had significant differences among the four etiologies. Thus, NASH- and alcohol-associated patients both had significant glycemia compared with patients with either HBV or HCV, and HBV patients had significantly higher levels of both total cholesterol and LDL cholesterol than HCV patients (P=0.0001). Tumor aggressiveness was associated with HDL cholesterol in HCV patients, but tumor aggressiveness was instead significantly associated with both LDL cholesterol and BMI in NASH patients. Thus, for the first time, we have shown a relationship between lipid parameters and indices of HCC growth.

There might several possible lipid changes that are involved in HCC biological characteristics such as growth (MTD). Changes in tumor membrane lipids alter membrane fluidics and thus growth factor signaling, as well as HCC properties such as tumor invasiveness.^6,15 Postulated mechanisms have also included chronic inflammation as well as signaling via adipokines, such as leptin. A role for lipids and obesity in altering the intestinal microbiome, with subsequent effects on tumorigenesis, is being intensively studied. It is becoming clearer that free radicals are also involved in tumor biology, possibly through obesity-associated inflammation or through lipid-related changes.^34-36 Thus, although obesity and metabolic syndrome, and their attendant lipid changes, have been shown to be associated with HCC development, it is increasingly clear that in non-obesity/NASH- associated HCC, there are also many associated lipid changes,^37-40 which are likely to contribute to HCC biology and are thus coming to be viewed as a target for potential HCC therapies.^{8-11,31-33,36}

There were some clear limitations to this study (n=521), including the relatively small number of patients in the different subgroups—especially in the etiological subgroups. Furthermore, although we found associations for HDL and tumor parameters, whether they were cause or consequence cannot be definitively established. The fact that the HCC–HDL relationships held up even when patients with elevated blood bilirubin levels were excluded from the analysis (Tables 2C and 3C), indicated to us that the lipid changes were not a consequence of compromised liver function, as suggested in other studies.² Furthermore, some patients from the larger dataset were excluded from this analysis due to missing lipids data. Thus, these findings might not be useful in screening, but they do have potential for prognosis considerations.

Conclusions

Both the Tumor Aggressiveness Index and the PVT were significantly associated with HDL levels. Significance for HDL was also found in a final linear regression model of the tumor aggressiveness index, regardless of bilirubin levels. The highest HR for death was also found for high HDL. When different etiological causes were compared, there were no significant differences for HDL.

Footnotes

Author contributions

Brian I. Carr, Gianluigi Giannelli, and Vito Guerra contributed equally to this study.

Declaration of conflicting interest

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Jiang

Nilsson-Ehle

Influence of liver cancer on lipid and lipoprotein metabolism. Lipids Health Dis 2006; 5: 4.

Arain

Talpur

Channa

et al . Serum lipid profile as a marker of liver impairment in hepatitis B Cirrhosis patients. Lipids Health Dis 2017; 16: 51. doi: 10.1186/s12944–017–0437–2.

Arain

Talpur

Channa

NA.

A comparative study of serum lipid contents in pre and post IFN-alpha treated acute hepatitis C patients. Lipids Health Dis 2015; 14: 117.

Siddiqui

Fuchs

Idowu

et al . Severity of nonalcoholic fatty liver disease and progression to cirrhosis are associated with atherogenic lipoprotein profile. Clin Gastroenterol Hepatol 2015; 13: 1000–1008.e3.

Shariff

Tognarelli

Lewis

et al . Plasma lipid profiling in a rat model of hepatocellular carcinoma: potential modulation through quinolone administration. J Clin Exp Hepatol 2015; 5: 286–294.

Lin

Ding

Wang

et al . Functional lipidomics: Palmitic acid impairs hepatocellular carcinoma development by modulating membrane fluidity and glucose metabolism. Hepatology 2017; 66 (2): 432–448. doi: 10.1002/hep.29033. Epub 2017 Jun 16.

Passos-Castilho

Carvalho

Cardozo

et al . Serum lipidomic profiling as a useful tool for screening potential biomarkers of hepatitis B-related hepatocellular carcinoma by ultraperformance liquid chromatography-mass spectrometry. BMC Cancer 2015; 15: 985.

Abel

Smuts

de Villiers

et al . Changes in essential fatty acid patterns associated with normal liver regeneration and the progression of hepatocyte nodules in rat hepatocarcinogenesis. Carcinogenesis 2001; 22: 795–804.

Diren

Beyoğlu

Sandrine

Imbeaud

Olivier

Maurhofer

et al . Tissue metabolomics of hepatocellular carcinoma: tumor energy metabolism and the role of transcriptomic classification. Hepatology 2013; 58: 229–238.

10.

Saito

Ojima

Ichikawa

et al . Molecular background of alpha-fetoprotein in liver cancer cells as revealed by global RNA expression analysis. Cancer Sci 2008; 99: 2402–2409.

11.

Budhu

Roessler

Zhao

et al . Integrated metabolite and gene expression profiles identify lipid biomarkers associated with progression of hepatocellular carcinoma and patient outcomes. Gastroenterology 2013; 144: 1066–1075.e1.

12.

Forte

McCall

Knowles

et al . Isolation and characterization of lipoproteins produced by human hepatoma-derived cell lines other than HepG2. J Lipid Res 1989; 30: 817–829.

13.

Weber

Cantero

Phospholipid content in Novikoff hepatoma, regenerating liver and in liver of fed and fasted normal rats. Exp Cell Res 1957; 13: 125–131.

14.

Miura

Ohnishi

Nonalcoholic fatty liver disease: from lipid profile to treatment. Clin J Gastroenterol 2012; 5: 313–321.

15.

Wierzbicki

Oben

Nonalcoholic fatty liver disease and lipids. Curr Opin Lipidol 2012; 23: 345–352.

16.

Nseir

Mograbi

Ghali

Lipid-lowering agents in nonalcoholic fatty liver disease and steatohepatitis: human studies. Dig Dis Sci 2012; 57: 1773–1781.

17.

Musso

Cassader

Gambino

Cholesterol-lowering therapy for the treatment of nonalcoholic fatty liver disease: an update. Curr Opin Lipidol 2011; 22: 489–496.

18.

Huang

PL.

A comprehensive definition for metabolic syndrome. Dis Model Mech 2009; 2: 231–237.

19.

Kant

Hull

MA.

Excess body weight and obesity—the link with gastrointestinal and hepatobiliary cancer. Nat Rev Gastroenterol Hepatol 2011; 8: 224–238.

20.

Khan

Perumpail

Wong

et al . Advances in hepatocellular carcinoma: Nonalcoholic steatohepatitis-related hepatocellular carcinoma. World J Hepatol 2015; 7: 2155–2161.

21.

Younossi

Koenig

Abdelatif

et al . Global epidemiology of nonalcoholic fatty liver disease-Meta-analytic assessment of prevalence, incidence, and outcomes. Hepatology 2016; 64: 73–84.

22.

Mittal

El-Serag

Sada

et al . Hepatocellular carcinoma in the absence of cirrhosis in United States veterans is associated with nonalcoholic fatty liver disease. Clin Gastroenterol Hepatol 2016; 14: 124–131.

23.

Michelotti

Machado

Diehl

AM.

NAFLD, NASH and liver cancer. Nat Rev Gastroenterol Hepatol 2013; 10: 656–665.

24.

Piscaglia

Svegliati-Baroni

Barchetti

et al . Clinical patterns of hepatocellular carcinoma in nonalcoholic fatty liver disease: A multicenter prospective study. Hepatology 2016; 63: 827–838.

25.

Carr

Guerra

Giannini

et al . An HCC tumor aggressiveness index and blood GTP, bilirubin and platelet levels. J Integr Oncol 2016; 5: 2.

26.

Carr

Guerra

a hepatocellular carcinoma tumor aggressiveness index and its relationship to liver enzyme levels. Oncology 2016; 90: 215–220.

27.

Jiang

Nilsson-Ehle

Influence of liver cancer on lipid and lipoprotein metabolism. Lipids Health Dis 2006; 3 (5): 4.

28.

Sauer

Dauchy

Blask

DE.

Dietary linoleic acid intake controls the arterial blood plasma concentration and the rates of growth and linoleic acid uptake and metabolism in hepatoma 7288CTC in Buffalo rats. J Nutr 1997: 127: 1412–1421.

29.

Lin

Ding

Wang

et al . Functional lipidomics: Palmitic acid impairs hepatocellular carcinoma development by modulating membrane fluidity and glucose metabolism. Hepatology 2017; 66: 432–448.

30.

Liu

Luo

Halim

et al . Targeting lipid metabolism of cancer cells: A promising therapeutic strategy for cancer. Cancer Letters 2017; 10: 39–45.

31.

Bhalla

Hwang

Dewi

et al . Metformin prevents liver tumorigenesis by inhibiting pathways driving hepatic lipogenesis. Cancer Prev Res 2012; 5: 544–552.

32.

Yang

Wang

Feng

et al . Aspirin suppresses the abnormal lipid metabolism in liver cancer cells via disrupting an NFκB-ACSL1 signaling. Biochem Biophys Res Commun 2017; 486: 827–832.

33.

Shimizu

Tanaka

Moriwaki

Obesity and hepatocellular carcinoma: targeting obesity-related inflammation for chemoprevention of liver carcinogenesis. Semin Immunopathol 2013; 35: 191–202.

34.

Rohr-Udilova

Stolze

Sagmeister

et al . Lipid hydroperoxides from processed dietary oils enhance growth of hepatocarcinoma cells. Mol Nutr Food Res 2008; 52: 352–359.

35.

Rohr-Udilova

Sieghart

Eferl

et al . Antagonistic effects of selenium and lipid peroxides on growth control in early hepatocellular carcinoma. Hepatology 2012; 55: 1112–1121.

36.

Liu

Yang

et al . Inhibition of oxidative stress-elicited AKT activation facilitates PPARγ agonist-mediated inhibition of stem cell character and tumor growth of liver cancer cells. PLoS One 2013; 8: e73038.

37.

Nderitu

Bosco

Garmo

et al . The association between individual metabolic syndrome components, primary liver cancer and cirrhosis: A study in the Swedish AMORIS cohort. Int J Cancer 2017; 141 (6): 1148–1160. doi: 10.1002/ijc.30818. Epub 2017 Jun 21.

38.

Khattab

Eslam

Mousa

et al . Association between metabolic abnormalities and hepatitis C-related hepatocellular carcinoma. Ann Hepatol 2012; 11: 487–494.

39.

Skill

Scott

et al . Hepatocellular carcinoma associated lipid metabolism reprogramming. J Surg Res 2011; 169: 51–56.

40.

Ooi

Shiraki

Sakurai

et al . Clinical significance of abnormal lipoprotein patterns in liver diseases. Int J Mol Med 2005; 15: 655–660.