Statin Use Decreases the Incidence of Hepatocellular Carcinoma: An Updated Meta-Analysis.

Statins can decrease hepatocellular carcinoma (HCC) occurrence, but the magnitude and the predictors of these effects remain unclear. This meta-analysis provides a pooled estimate of the impact of statin use on HCC occurrence. Pooled effects were calculated using a random-effects model by means of the DerSimonian and Laird test. Primary endpoint was the time-dependent correlation between statin use and HCC incidence expressed as hazard ratio (HR), both crude and adjusted. The crude and adjusted odds ratios (OR) for HCC occurrence between statin users and non-users were analyzed. Twenty-five studies with 1,925,964 patients were included. Crude OR for HCC incidence was 0.59 (95% CI: 0.47–0.74), confirmed in adjusted analysis (OR: 0.74, 95% CI: 0.70–0.78). Adjusted HR was 0.73 (95% CI: 0.69–0.76). This effect was more pronounced in HBV patients (HR: 0.46, 95% CI: 0.36–0.60) and with a cumulative daily dose beyond 365 (HR: 0.27, 95% CI: 0.11–0.67). Lipophilic statins were associated with reduced HCC incidence (HR: 0.49, 95% CI: 0.39–0.62). Atorvastatin determined the greater magnitude of effect (HR: 0.43, 95% CI: 0.28–0.65). This meta-analysis demonstrates the beneficial chemopreventive effect of statins against HCC occurrence. This effect is dose-dependent and more pronounced with lipophilic statins.


Introduction
Hepatocellular carcinoma (HCC) is the fifth most commonly occurring type of cancer and the leading cause of mortality in cirrhotic patients [1]. Despite the recent improvement in diagnosis and screening programs in cirrhotics, a great number of patients are still diagnosed in the advanced stage, thus being unsuitable to curative treatments, such as surgery, orthotopic liver transplantation, or radiofrequency ablation [2][3][4]. Therefore, the identification of prognostic factors for HCC occurrence is of paramount importance in order to decrease the burden of this disease, in particular in high-risk populations, such as cirrhotic or chronic hepatitis B virus (HBV) patients.
Three-hydroxy-3-methyl-glutaryl-coenzyme A (HMG co-A) reductase inhibitors (statins) are effective and commonly used worldwide as a treatment for dyslipidemia, and increasing evidence shows that statins also have anti-inflammatory and anti-oncogenic effects [5,6].
A previous meta-analysis, based on 10 studies, shed light on the beneficial role of statins in preventing HCC occurrence [7]; however, the relatively low number of studies included did not allow subgroup analyses based on the specific agent administered (type of statin) or to correlate the anticancer effect with the dose.
In the last years, several cohort and case-control studies have been published in the field, hence the need to update the previous data in order to better define the eventual chemopreventive role of statins in hepato-oncology.
The aim of this meta-analysis was to provide a pooled estimate of the impact of statin use on HCC occurrence. Primary endpoint was defined as the time-dependent correlation between statin use and HCC incidence (in terms of both crude and adjusted hazard ratio [HR] for several baseline variables). Secondary outcome was the correlation between the overall incidence of HCC and statin administration expressed in terms of the odds ratio (OR), again both crude and adjusted.

Characteristics of Included Studies
As shown in Figure 1, out of 5562 studies initially identified, after preliminary exclusion of manuscripts not fulfilling inclusion criteria, 95 potentially relevant articles were examined. Among these studies, 23 overlap series and 47 studies not reporting OR or HR (or data useful for their calculation) were further excluded. Finally, 25 studies reporting 21,576 cases of HCC in 1,925,964 patients were included in the meta-analysis .
Five studies [10,11,20,26,27] were conducted in the same population (Taiwan National Health Insurance and Research Database) but they reported different outcomes or data from different subgroups, therefore were considered separately in different analyses. Nine studies were conducted in Asia [10,11,15,16,18,20,[26][27][28], and the remaining studies were conducted in western countries. The two aforementioned individual patient data analyses concerned multicenter RCTs [31,32].
Baseline clinical and demographical characteristics were well balanced between statin users and the control group. Variables adjusted for were mainly age, sex, and etiology of the underlying liver disease; other comorbidities; and the use of other medications.
Details on methodological characteristics and quality of included articles are shown in Table S1.

Risk of HCC
As reported in Figure 2A, overall crude OR for HCC incidence was 0.59 (95% CI: 0.47-0.74), thus highlighting a significant protective role of statins against HCC occurrence (p < 0.001). Of note, high evidence of heterogeneity was observed (I 2 = 92%).
Adjusted analysis, considering the aforementioned baseline confounders, confirmed the anti-oncogenic effect of statins, with a reported adjusted OR (aOR) as high as 0.74 (95% CI: 0.70-0.78). Heterogeneity slightly decreased to 79% in adjusted analysis ( Figure 2B).
There was no evidence of publication bias (data not shown).

Subgroup Analysis
The aHR for HCC occurrence was confirmed as significantly in favor of statins in HBV patients (0.46, 95% CI: 0.36-0.60; I 2 = 0%) while only a non-significant benefit was observed in HCV patients, although this result should be interpreted with caution due to the low number of studies and the high heterogeneity (Table 2).
Statins were proved to be effective in both diabetic and non-diabetic patients, while the magnitude of the chemopreventive effect was found to be linearly correlated to the dose, with an aHR of 0.51 (95% CI: 0.30-0.88) and 0.27 (95% CI: 0.11-0.67) in patients administered a cumulative defined daily dose (cDDD) below or beyond 365, respectively ( Table 2).

Sensitivity Analysis
The results of several sensitivity analyses are reported in Table 3. Table 3. Sensitivity analysis of the main diagnostic outcome (odds ratio for hepatocellular carcinoma occurrence) performed based on: (a) study design (observational versus RCT); (b) study location (Asia versus western); and (c) study quality (high versus low).

Variable Subgroup Studies (n) Summary Estimate (95% CI) Within-Group Heterogeneity (I 2 )
Study The findings of main analysis were confirmed in sensitivity analysis performed according to study quality (high versus low), design (RCT versus observational), and location (Asia versus western).
The only exception was the sub-analysis restricted to RCTs, where an OR around 1 was observed; of note, two out of three studies included as RCTs represented individual patient data analysis of patients enrolled in prospective controlled trials of cholesterol in heart disease (hence, non-cirrhotic) [31,32] where the incidence of HCC was very low.
Heterogeneity in the sensitivity analysis was mainly moderate.

Discussion
HCC is one of the most common cancer types and the leading cause of tumor-related deaths in cirrhotic patients [1].
Statins have been shown to consistently reduce liver fibrosis progression, mainly due to their immunomodulatory effects [33], to mitigate portal hypertension, and to upregulate transcription factors that exert vasoprotective effects in the liver and inhibit stellate cells, thus potentially further decreasing fibrosis [34]. On the other hand, the enthusiasm towards statins has been tempered by fears about their safety profile in cirrhotic patients because of the risk of dose-dependent hepatic injury [35].
A previous meta-analysis found a 37% decreased risk of HCC occurrence in statin users [7] but a specific analysis aiming to identify higher-risk settings was unfeasible due to the low number of included studies.
Through a meta-analysis of 25 studies, we made several key observations. First, we found a 26% decreased incidence of HCC in patients treated with statins, after adjustment for several variables. When considering a time-dependent outcome, such as HR, not influenced by the imbalance in follow-up length between the studies, we confirmed a 27% decreased incidence of HCC when adjusting for several clinical and demographical parameters.
Second, this effect was more pronounced and consistent in HBV patients (56% decrease in HCC incidence) and it was found to be linearly correlated to the dose, with a 73% decreased HCC risk in patients administered a cDDD beyond 365.
Third, as already observed in previous individual studies [24,27], lipophilic statins (atorvastatin, lovastatin, and simvastatin) were associated with significantly reduced HCC (51% compared to 27% in hydrophilic statin users). Among several agents tested, the more pronounced chemopreventive effect was observed with atorvastatin (57% reduction in HCC occurrence), although further studies are warranted to provide definitive results in this regard.
This preventive effect of statins is likely to be independent of its lipid-lowering effects, because lipid-lowering agents other than statins were not associated with reduction in the risk of HCC in previous reports [26,30].
The chemopreventive effect of statins in Asian populations, mainly affected by HBV hepatitis/cirrhosis, is well recognized and it was strongly confirmed in our analysis. In fact, HBV genome integration determines several DNA modifications and microdeletions that can target cancer-relevant genes, potentially providing hepatocytes with a growth advantage [36]. Statins, by inhibiting the mevalonate pathway, can prevent potential detrimental effects of these growth signaling proteins [37]. Statins also exert pro-apoptotic effects by activating several caspases and decreasing Bcl-2 [37]. Moreover, statins inhibit the activation of the proteasome pathway, limiting the breakdown of some molecules with growth-inhibitory effects, such as p21 and p27 [37].
On the other hand, the anti-oncogenic effect of statins in HCV patients is less evident, probably due to modification of metabolic syndrome, insulin-mediated cell proliferation, and obesity-associated inflammation [33,38]. A clear relationship between statin use and HCC incidence in HCV patients was not found in our meta-analysis, but this result might be due to the low number of studies specifically evaluating this subset of patients.
One of the novel findings in our study is the clear dose-dependent effect of statins in decreasing HCC occurrence, with 365 cDDDs as the cut-off to observe the highest preventive effects, as already reported in previous individual studies [28]. Several biological properties of statins, such as anti-angiogenetic or anti-fibrotic effects, were found to be strictly associated to the dose used in in vivo studies [39,40].
The greater chemoprotective effect of lipophobic statins, clearly outlined by our analysis, is likely due to greater lipid solubility and membrane permeability which enhance their pharmacological effects [41]. Further studies are needed to confirm these findings.
Of note, sensitivity analysis restricted to RCTs did not show a significant preventive benefit with statins although the result should be interpreted with caution due to the low number of RCTs included. It is evident that the limited number of RCTs was insufficient to detect a significant effect of statins, in particular considering that these RCTs were conducted in non-cirrhotic patients, hence with a very low HCC incidence.
There are some limitations to our study. First, the limited number of studies in many subgroups does not allow a strong comparison between statins users and non-users in several subsets of patients, in particular concerning single pharmacological agents or based on the use of specific antiviral treatments. Second, several comparisons were weakened by the high heterogeneity. We performed different sensitivity analyses that confirmed the main results and, noteworthy, the heterogeneity decreased when several subgroups were considered separately. To take into account the potential baseline confounders, we considered in our primary analysis the aHR. However, even if the heterogeneity decreased, it remained significant, probably none of the studies adjusted for the same confounders. Other eventual source of heterogeneity could be represented by the different populations enrolled in the included studies, with different risk of HCC occurrence and probably uneven screening campaigns in the different geographic areas. Finally, most of the included studies were retrospective series, hence prone to selection bias.
In conclusion, despite these weaknesses, our meta-analysis demonstrates the beneficial chemopreventive effect of statins against HCC occurrence. This effect is dose-dependent and more pronounced with lipophilic statins.
Further studies are warranted to confirm these results and to identify the exact setting where this anti-oncogenic effect could be enhanced.

Inclusion and Exclusion Criteria
Only studies meeting the following criteria were included: (1) RCTs or observational studies recruiting >10 patients with clear exposure to statin therapy; (2) studies published in English; (3) articles reporting HCC occurrence; and (4) studies reporting OR, HR, or data useful for their calculation. Case reports, non-clinical studies, review articles, and animal models were excluded. Figure 1 reports the search strategy followed in the meta-analysis. Bibliographic research was conducted on PubMed, EMBASE, Cochrane Library, and Google Scholar including all studies fulfilling inclusion criteria published until December 2019.

Search Strategy
The search was conducted by two study investigators (AF and MAA) independently and keywords used were "HMG-CoA reductase inhibitor(s)," "statin(s)," "atorvastatin," "fluvastatin," "lovastatin," "pravastatin," "rosuvastatin," or "simvastatin" combined with "liver cancer" or "neoplasm(s)." Relevant reviews and meta-analyses on the use of statins and HCC occurrence were examined for potential suitable studies. Authors of included studies were contacted to obtain full text or further information when needed.
The quality of included studies was assessed by two authors independently (AF and MAA) according to the Cochrane Collaboration's tool for assessing the risk of bias [42] for RCTs and the Newcastle-Ottawa scale [43] for non-randomized studies. Disagreements were solved by discussion and following a third opinion (RS).

Statistical Analysis
Primary endpoint of the current meta-analysis was the comparison of HCC occurrence between statin users and non-statin users. Data of HCC occurrence were compared through a random-effects model based on the DerSimonian and Laird test, and summary estimates were expressed in terms of both HR and OR along with their relevant 95% CIs.
To partially obviate the bias due to the different follow-up length among the studies and, within each study, between the two treatment arms and to consider not only the number of events but also their timing and the follow-up of censored patients, HRs were considered in the analysis when reported in the included studies.
Two separate analyses were conducted for crude and adjusted summary estimates (both OR and HR) and aHR was considered the primary endpoint in the meta-analysis.
Chi-square and I 2 tests were used for across studies comparison of the percentage of variability attributable to heterogeneity beyond chance. p < 0.10 for chi-square test and I 2 <20% were interpreted as low-level heterogeneity.
Probability of publication bias was assessed using funnel plots and with Begg and Mazumdar's test. Sensitivity analysis was conducted according to the quality of included studies (high versus low), location of the studies (Asia versus western), and study design (RCT versus observational).
A subgroup analysis based on several statin molecule and class (lipophilic versus hydrophilic), etiology of liver disease, presence of diabetes, and cumulative defined daily dose (cDDD: ≤365 versus >365) was performed.
All statistical analyses were conducted using RevMan 5.3 software (the Cochrane Collaboration, Oxford, UK). For all calculations a two-tailed p-value of <0.05 was considered statistically significant.

Conclusions
Our meta-analysis demonstrates the beneficial chemopreventive effect of statins against HCC occurrence. This effect is dose-dependent and more pronounced with lipophilic statins. Further studies are warranted to confirm these results and to identify the exact setting where this anti-oncogenic effect could be enhanced.