Characteristics And Etiologies Of Hepatocellular Carcinoma In ... - PLOS

Discussion

In the present study, nearly 26% of the patients were non-cirrhotic. The non-cirrhotic patients were younger, included a higher proportion of male patients, and had a lower mean BMI than the cirrhotic patients. This phenomenon could be explained by HBV being the leading etiology among the non-cirrhotic patients. HBV-associated HCC is seen in younger and predominately male patients compared with HCCs with other etiologies [4,13–15]. Relatedly, obesity is a risk factor for fibrosis progression, not only in NAFLD, but also in other chronic liver diseases [16].

The leading etiology of HCC among the non‐cirrhotic patients was HBV, as expected, followed by HCV. HBV infection is a risk factor for non-cirrhotic HCC in East Asia [17]. Notably, 162 (20.7%) of the patients who tested anti-HCV positive in the present study were non-cirrhotic. We did not have information regarding whether these patients had viremia. HCV-associated HCC usually arises from cirrhotic liver [4,13–15], and the finding that a significant proportion of the HCV patients in the present study were non-cirrhotic could have been due to the study having included patients with past resolved HCV infections and patients who received antiviral therapy with sustained virological responses which led to cirrhosis regression [18].

The overall rate of non‐cirrhotic HCC in this study (25.7%) was higher than that in the previous study from the US (11.7%) [9]. This discrepancy between the two studies could have been due to the different definitions of cirrhosis used and referral bias. There are no universally accepted criteria for defining cirrhosis in the absence of histology. Previous studies have used images, laboratory results, and the presence or absence of liver‐related complications to define cirrhosis in those without histological evidence. In contrast, we used imaging alone to define cirrhosis in those without histological evidence. The aforementioned previous study from the US was a multi-center study, and referral is needed in the US. As such, there was referral bias in the previous study [9]. In contrast, referral is not needed in Taiwan.

In the present study, the average tumor size was significantly larger in the non-cirrhotic patients than in the cirrhotic patients, possibly due to a lack of surveillance in the non-cirrhotic patients. 66.0% of the non-cirrhotic patients underwent liver resection. In contrast, only 17.4% of the patients with cirrhosis underwent liver resection. Major liver resections can be performed with low rates of life-threatening complications in non-cirrhotic patients [13]. The better survival in non-cirrhotic patients may be due to liver resection being associated with improved survival across BCLC stages [19]. Furthermore, cirrhosis is a major risk factor for HCC recurrence, and the relatively decreased incidence of recurrence in those without cirrhosis also leads to better survival. Finally, non-cirrhotic patients are less likely to develop liver decompensation after multiple liver-directed therapies than cirrhotic patients, and cirrhosis is the most important competing risk of death in patients with HCC [4,13].

Old age, advanced tumor stage, and non-curative treatments were associated with mortality in the non-cirrhotic patients. The results of the present study are consistent in this regard with those of previous studies [4,13–15].

Our HCC registry database only records etiologies including HBV, HCV, and alcohol use disorder; other etiologies, including NAFLD, are not recorded. Current guidelines recommend that NAFLD be diagnosed based on evidence of steatosis, whether provided by histology or imaging, in the absence of an alternative liver disease [16].

The gold standard for diagnosing steatosis is liver biopsy or magnetic resonance spectroscopy. However, liver biopsy is invasive and magnetic resonance spectroscopy is expensive. Therefore, ultrasound is the most commonly used imaging modality for diagnosing steatosis [3]. Ultrasound has good sensitivity (85%) and specificity (94%) for diagnosing moderate to severe steatosis [20], but it is less reliable for diagnosing mild steatosis. Furthermore, over time, steatosis may disappear when cirrhosis develops (i.e. burnout non-alcoholic steatohepatitis) [16].

Thus, previous studies used different criteria to define NAFLD-associated HCC in the absence of histology. In the absence of an alternative liver disease as a precondition, Mittal et al. defined NAFLD as the presence of metabolic syndrome [21]. Bengtsson et al. defined NAFLD as BMI ≥25 kg/m2 and Type 2 diabetes, or BMI ≥30 kg/m2 [22]. Kanwal et al. defined NAFLD as elevated alanine aminotransferase values [23]. Younossi et al. defined NAFLD by using the International Classification of Diseases codes, and including the codes for NAFLD and cryptogenic liver disease [24]. A lifestyle history, including factors such as a sedentary lifestyle, eating habits, and the trajectories of weight change since young adulthood and waist expansion, is a prerequisite for diagnosing NAFLD [25].

The limited number of patients with alcohol use disorder (n = 71) in the present study suggests that the reported alcohol use of the patients may have been underestimated. Thus, the patients with apparently non-viral- and non-alcohol-related etiologies in the present study may actually have been composed in part of some with etiologies including NAFLD, unreported alcohol use disorder, rare chronic liver diseases, and HBV with HBsAg seroclearance [26] in an HBV endemic area, namely Taiwan. Among the entire cohort, only 94 patients (4.6%) in the present study were found to have non-cirrhotic, non-viral-, non-alcohol-related HCC, which indicated that the number of patients with NAFLD-associated non-cirrhotic HCC should have been less than 94.

Our previous study enrolled 5613 consecutive patients with HCC who were treated at our institution between 1986 and 2002, including 4287 (76.4%) patients with HCC associated with viral hepatitis [27]. In the present study, 1681 (81.8%) patients had HCC associated with viral hepatitis. As time goes by, the proportion of patients with non-viral-associated HCC has not significantly changed. According to a recent review article, although the potential impact of NAFLD on HCC incidence in North America is major, the potential impact of NAFLD in the epidemiology of HCC in Asia is minor at present but may be growing [28].

In Taiwan, surveillance for HCC in those with chronic viral hepatitis and cirrhosis of any etiology is reimbursed by the National Health Insurance program. Therefore, those patients with non-viral, non-cirrhotic HCC may be diagnosed incidentally or during a symptom work-up. It is impossible to surveil (or even screen) for NAFLD patients without cirrhosis. NAFLD-associated HCC in non-cirrhotic liver was therefore likely underestimated in the present study and in other studies [9,21–24].

One strength of the present study is that it had no referral bias. The patients enrolled in the present study could thus be representative of the general HCC population in Taiwan. Our institution is one of the largest academic medical cancer treatment centers in Taiwan, and has 2724 beds and 861 physicians (https://en.wikipedia.org/wiki/Chang_Gung_Medical_Foundation#Kaohsiung_Chang_Gung_Memorial_Hospital). The current healthcare system in Taiwan, known as the National Health Insurance program, was instituted in 1995, and the population coverage had reached 99% as of 2004. Under the system, citizens can choose hospitals and physicians without referral, and regular office visits have co-payments as low as US$5 per visit. Therefore, most patients are quick to visit medical centers if they feel some need to. The main island of Taiwan, measuring 35808 square kilometers, making it smaller in size to Switzerland, is a highly urbanized island with 26 academic medical centers, all located on west side of the island (https://en.wikipedia.org/wiki/Healthcare_in_Taiwan). Thus, there is no barrier to healthcare for citizens who live on the west side of the main island. A second strength of the present study is that we used an authoritative source to check vital statuses of the patients enrolled. We could thus make sure of the vital status of every single patient enrolled in the study.

There were also some limitations in the present study. First, this was a retrospective study. Second, there are no specific diagnostic criteria for defining the presence vs. absence of cirrhosis on imaging in our clinical practice. Relatedly, the diagnosis of cirrhosis on imaging was subjective in the present study. Third, there was a lack of etiology data on patients without HBV, HCV, or alcohol-related liver disease (such as NAFLD). Diabetes, hypertension, dyslipidemia, and other risk factors typically associated with NAFLD (e.g. metabolic syndrome, truncal obesity defined by waist circumference) were not recorded in our HCC registry data. However, the HCC registry data of our institution do include BMI data. Obesity is defined as BMI ≥27(kg/m2) by the Health Promotion Administration, Ministry of Health and Welfare, Taiwan (https://health99.hpa.gov.tw/onlinkhealth/onlink_bmi.aspx). Among those with HBV, HCV, or alcohol-related liver disease, there were 443 (25.29%) patients with BMI ≥27(kg/m2). Among those with unknown etiology of liver disease, there were 95 (31.35%) patients with BMI ≥27(kg/m2). The proportion of patients with BMI ≥27 (kg/m2) was, therefore, significantly higher among those with unknown etiology of liver disease compared with those with HBV, HCV, or alcohol-related liver disease (p = 0.002).

The limitations of our HCC registry data were as follows: (1) only the first-line therapy was recorded, (2) the patients’ reported alcohol consumption levels may have been underestimated. The HCC registry data for our institution do not, however, include daily alcohol intake or data provided by the use of screening tools [e.g. the AUDIT (Alcohol Use Disorders Inventory Test)] to identify alcohol use disorders [29] and (3) data regarding medical comorbidities, abstinence from alcohol, antiviral therapy, and weight change during the follow-up period, which might affect the prognosis, were not available [30–33].

In conclusion, around 26% of the patients with HCC in our large cohort of patients from an academic medical center in East Asia were non-cirrhotic. HBV and HCV were the leading etiologies of the investigated patients with HCC, regardless of whether they did or did not have cirrhosis. Although obesity rates are rising worldwide [34], the contribution of NAFLD-associated non‐cirrhotic HCC to the overall burden of HCC was not significant enough to result in an increase in the total burden of HCC in Taiwan as of the end of this study in 2017.