Increasing mortality in the United States from cholangiocarcinoma: an analysis of the National Center for Health Statistics Database

Previous literature is sparse and not conclusive regarding the combined mortality from ICC and ECC. The current study uniquely examines combined ICC and ECC mortality as a group based on the latest national data, and shows a consistent increase in CCA mortality across age and gender since 1999. In our analysis, the increasing trend concentrated in ICC, not ECC. Our study showed that the number of CCA deaths has increased substantially since 1999, reaching over 7000 cases a year in 2013. The majority of patients with CCA develop a recurrence after resection [17], and there is considerable perioperative mortality [18]; therefore, mortality from this disease is a strong indicator of incidence. Our data suggest that the widely quoted rate of 2000–3000 new cases of CCA per year is an under-estimate [19‐23], and that there has been a true increase in incidence of CCA. Our results parallel a recent study of data up to 2009 in the 33 Cancer Registries that participate in the North American Association of Central Cancer Registries [12] and are consistent with a trend demonstrated in the incidence of CCA among the SEER population up to 2012 [15], but are derived from more complete national data.

We observed significant variations in CCA mortality across race and gender after adjusting for age distributions. Of particular note, increased risk was associated with male gender overall, Asian ethnicity for both genders, Hispanic women, and advanced age. Notably, while the risk of dying of all cancers combined is highest among African Americans [24], we found that risk of CCA mortality was lower, although increasing, among African Americans compared to other ethnic groups.

The NCHS mortality database does not include data on potential risk factors of CCA such as primary sclerosing cholangitis (PSC), so we cannot quantify the contribution of various risk factors on CCA mortality. However, external data sources show that the higher risk of CCA mortality in males is likely to be related to increased risk in hepatitis C, cirrhosis and PSC. Epidemiological studies have found that risk of CCA increases 27-fold with cirrhosis (adjusted odds ratio, 27.2; P <.0001) [25], six-fold with hepatitis C virus infection, (adjusted odds ratio, 6.1; P <.0001) [25, 26], and 1560 times with PSC (HR 1560; 95 % CI = 780, 2793; p <0.0001) [27]. Furthermore, compared with females, males have higher risk of hepatitis C (176 vs. 105 per 100,000) [28], chronic liver disease (305 vs 206 per 100,000) [28], and PSC (0.45 vs. 0.37 per 100,000) [29]. The higher risk of CCA mortality among males in our study may be related in part to the fact that men are at higher risk for both cirrhosis and PSC [6]. PSC is, however, is a rare diagnosis affecting at most 16.2 per 100,000 people in some studies.

Extensive evidence implicates inflammation and cholestasis as key factors in the pathogenesis of CCA [8, 30, 31]. Inflammatory markers are also universally elevated in metabolic syndrome [32]. In the face of the increasing burden of CCA we have demonstrated and given the increasing incidence of the metabolic syndrome in the United States [33], a recent large study examining preexisting metabolic syndrome as a risk factor for primary liver tumors (not differentiating between ICC and HCC) bears attention [34]. In addition, obesity itself is becoming established as an independent risk factor for CCA [35]. While there are multiple mechanisms by which the metabolic syndrome may be linked to the pathogenesis of gastrointestinal malignancies [36], a putative etiologic link between obesity and CCA is leptin, the hormone regulating homoeostasis which is increased in obese patients, and has been shown to stimulate grown, migration and prevent apoptosis of a CCA cell model [37]. Thus, the underlying etiology of this increase in ICC across gender and race may possibly be related to the prevalence of metabolic syndrome and obesity [34, 35, 37, 38].

In addition to inflammatory risk factors that may cross ethnicities, there are well-established risk factors for CCA specific to ethnicities that merit notice given the significant racial and ethnic variations in CCA mortality in our study. The high risk of CCA mortality among Asians is unsurprising as CCA is more common in Southeast Asia [39, 40], potentially because of the prevalence of infections such hepatitis B and C virus [41], and hepatobiliary fluke infection, prevalent in Asia [42], both inflammatory [43] risk factors for CCA [3, 8, 44‐46]. In addition, hepatolithiasis is more commonly noted in Asia than in Western countries and is associated with a 10 % incidence of CCA [8, 47‐49]. Another potential contributing factor to the increased incidence among Asians is Type 2 diabetes mellitus (T2DM) [46]. Diabetes has been associated with CCA in a Taiwanese population for both ICC (OR = 2.0) and ECC (OR = 1.8) [46]. The mortality and incidence rate from T2DM increased more in Asians than in their Caucasian or African American counterparts [50], despite on average a substantially lower BMI in the Asian population [51]. African Americans have 2.6 times the mortality due to diabetes compared with those of Asian descent [50], and although the mortality and incidence rates for most other cancers are higher for African Americans [24, 52], and African Americans have the highest increase in CCA mortality, they have substantially lower overall risk of CCA mortality in our study and in the earlier study of ICC [53]. That study also found that although hepatobiliary cancers were highly prevalent in Asian Pacific Islanders, the prevalence of ICC was not significantly different from that of other racial groups [53]. Those results, in combination with the data we present here, suggest that the increased mortality in Asians in the United States may be due to ECCs.

Another notable finding in our study was that the rate of death from CCA increased substantially with age (Fig. 2). This finding is consistent with the trend observed in the SEER databases for HCC, which notes that among persons 75 to 84 years, increases in HCC incidence were seen among all men and white women (P ≤0.05) in the United States from 1975 to 2005 [54]. Given that some authors have found decreased mortality from ICC, and have attibuted the decrease to better detection [14], the increase in CCA mortality with age may be in part related to surgical mortality. The incidence of severe and non-surgical postoperative complications is higher in older compared to younger patients undergoing hepatic resection [55] and in patients undergoing surgery for Klatskin Tumors [56]. Because of changes in the US population, this age-dependence may also contribute to increase in total CCA deaths.

In contrast to our findings in CCA, in a review of HCC from the SEER registries and liver cancer mortality data from the National Center for Health Statistics, Altekruse et al. found that HCC incidence rates in SEER registries did not significantly increase during 2007–2010 [57] but the US liver cancer mortality rates did increase [57]. These results suggest that the increased liver cancer mortality in SEER registries may be driven by CCA, not HCC (since ICC was combined with HCC in the SEER registries).

The results of this study must be interpreted in the context of some limitations. First, given that the NCHS mortality database does not include data on comorbidities that are potential risk factors for CCA such as PSC, we cannot quantify the contribution of these risk factors on CCA mortality. Secondly, the mortality data was extracted from death certificates and misclassification might occur in some cases [58, 59]. Misclassification of the cause of death on the death certificate may occur between CCA, pancreatic cancer, gallbladder cancer and hepatocellular cancer [60]. Another potential source of error is that the liver is a common site of adenocarcinoma metastasis, and thus some secondary liver cancers could be mistakenly over-counted as primary liver adenocarcinoma, or ICC [57]. Conversely, CCA can be misdiagnosed as a metastatic adenocarcinoma to the liver rather than a primary liver cancer [15, 61].

Our study has major strengths including the use of recently updated nationally representative NCHS data through 2014 and the provision of data on both ICC and ECC mortality. The inclusion of the entire US data minimizes potential selection bias or referral bias that are commonly encountered in institution-based studies or age limitations of the SEER-Medicare database [25]. Furthermore, the large sample provides a unique opportunity to evaluate how CCA mortality varies with age, gender, and race.