Independent of Primary Sclerosing Cholangitis and Cirrhosis, Early Adulthood Obesity Is Associated with Cholangiocarcinoma

Cholangiocarcinoma (CCA) represents a rare but heterogenous group of tumors that can occur anywhere along the biliary duct. CCAs are generally divided into two distinct anatomic groups: intrahepatic (ICC) and extrahepatic (ECC), which includes both perihilar (Klatskin tumor) and distal CCA. Altogether, CCAs are the second most common primary liver tumor, representing 10% to 15% of all primary liver tumors and 3% of all gastrointestinal malignancies (1). In the United States, CCA, ICC, and ECC have increased in incidence (per 100,000 person-years) by 43.8% (3.08–4.43), 148.8% (0.80–1.99), and 7.5% (2.28–2.45), respectively from 2001 to 2017 using the NCI Surveillance, Epidemiology, and End Results cancer registry (2). Although the rising incidence of CCA in the United States is likely in part due to advances in diagnosis and reclassification of previous liver carcinomas or cancers of unknown primary origin, the continued increase in incidence points to an actual rise in cases (3). A similar trend in ICC incidence and mortality has also been observed globally, although ECC incidence has largely stabilized (4). The reasons for this apparent increase in ICC incidence and mortality have not been well studied. Furthermore, the divergent trends between ICC and ECC potentially suggest differing etiologies and emphasize the importance of exploring the epidemiology of these diseases separately.

Although primary sclerosing cholangitis (PSC) and choledochal cysts are established risk factors for CCA, these diseases remain rare (5, 6). Less established potential risk factors for CCA include cirrhosis, type 2 diabetes mellitus, obesity, alcohol consumption, tobacco use, non-alcoholic fatty liver disease (NAFLD), and viral hepatitis (7, 8). The epidemiology of CCA in the United States is not well characterized; approximately half of the patients diagnosed with CCA have no identifiable risk factor and only 6% to 20% of all cases are explained by PSC (9, 10). Understanding the risk factors of CCA remains key for identifying high-risk populations and developing effective prevention and therapeutic interventions. Accordingly, we performed a case–control study to identify risk factors for CCA and to determine the differences in risk factors between ICC and ECC.

This hospital-based case–control study was approved by the Institutional Review Board of The University of Texas MD Anderson Cancer Center (Houston, TX). Written informed consent was obtained from each study participant for an interview and a blood sample donation.

Case patients were prospectively recruited from new patients evaluated and treated for CCA at the MD Anderson gastrointestinal medical oncology outpatient clinic. Inclusion criteria for case patients were a pathologically or radiologically confirmed diagnosis of ICC or ECC and the ability to communicate in English. Exclusion criteria were the presence of other types of primary liver cancer (i.e., hepatocellular carcinoma and fibrolamellar hepatocellular carcinoma), concurrent or past diagnosis of cancer at another site, or diagnosis of primary tumor of unknown origin. A total of 308 patients diagnosed with ICC and 121 patients diagnosed with ECC were enrolled in the study from May 2014 through March 2020.

Control participants were healthy (cancer-free) and genetically unrelated family members (i.e., spouses or in-laws) of patients who were diagnosed with cancers other than liver and gastrointestinal recruited from the same institution as the case patients. To prevent selection bias associated with shared environmental factors (due to similar lifestyle and similar genetic factors), we excluded spouses of patients who had been diagnosed with liver cancer, lung cancer, and head and neck cancer. Control participants were recruited from central diagnostic radiology clinics, where patients with cancer were directed for their initial diagnosis or posttreatment follow-up appointments. All cases and controls were simultaneously recruited in the study. Cases and controls were frequency-matched by age (±5 years) and sex for CCA, ICC, and ECC. A total of 1,061 control participants were enrolled, along with 308 ICC and 121 ECC cases, and thereby eliciting a power of 90% for testing analyses. The CCA case patient-to-control participant ratio was 1:2.5.

Patients with CCA and control participants were personally interviewed for 25–30 minutes by trained research coordinators using a structured and validated questionnaire (11, 12) for collecting demographic features and risk factors for CCA (13–16). Interviewers were blinded for the scope of the study. Potential risk factors included cigarette smoking, alcohol consumption, type 2 diabetes mellitus (both presence and duration), body weight, height, presence of gallbladder stones, prior history of chronic medical conditions, and family history of cancer.

During our interviews with patients with CCA and control participants, we collected detailed information regarding the history of first-degree relatives with cancer (yes or no), family history of cancer in type of relatives (parents, siblings, and offspring), including number of parents (fathers and mothers) with cancer, number of siblings (sisters and brothers) with cancer, number of offspring (daughters and sons) with cancer, and the total number of first-degree relatives with cancer. Patients were evaluated for PSC by outside gastroenterologists prior to their diagnosis of CCA at MD Anderson.

For our assessment of risk factors, we defined cigarette smokers as those who had smoked ≥100 cigarettes during their lifetime. Former smokers were defined as subjects who had quit smoking at least 1 year before study enrollment. Former and current smokers provided information regarding average number of cigarettes per day they smoked, the age at which they began smoking, and the duration of smoking. Former smokers were questioned about the age at which they stopped smoking. Former smokers were classified as ex-smokers by the number of years since they last smoked cigarettes: (i) 5 years and (ii) 10 years. Heavy smokers were defined as those who had >20 pack-years of smoking, in which pack-year was calculated by multiplying the number of years the subject smoked and the number of packs of cigarettes they smoked per day. We also examined the status and amount of alcohol consumption for cases and controls. Participants were defined as ever-alcohol consumers if they had consumed at least four alcoholic drinks per month for 6 months in their lifetime. Former drinkers were defined as subjects who quit drinking before enrollment, and we noted the age at which they stopped drinking. For current and former drinkers, we calculated the total lifetime volume of alcohol consumption in milliliters by examining the frequency of drinking, the type of servings (glass, bottle, or can), type of alcohol (beer, wine, or hard liquor), and number and size of servings during the entire duration of alcohol use. As a result, participants defined as heavy alcohol consumers had consumed at least 60 mL of ethanol per day during the subject's period of drinking alcohol (17).

A detailed assessment of obesity was included during the interview to obtain information about self-reported height (inches) and weight (pounds) at different intervals before cancer diagnosis for patients with CCA or before recruitment for control participants. Specifically, we asked for information regarding their current weight, as well as their weights during their mid-20s, mid-30s, mid-40s, mid-50s, and mid-60s, along with reported body size during the same age periods using the validated Stunkard pictograms (18). Body mass index (BMI) was calculated using the following formula: |${\rm{BMI}} = \frac{{{\rm{Weight}}\,( {{\rm{kilograms}}} )}}{{{{( {{\rm{Height}}\,{\rm{in}}\,{\rm{meters}}} )}}^2}}$|⁠. The calculated BMI was then classified into four levels: underweight (<18.5 kg/m²), normal weight (18.5–24.9 kg/m²), overweight (25–29.9 kg/m²), and obese (≥30 kg/m²).

All participants provided blood samples that were processed and analyzed at the MD Anderson Gastrointestinal Laboratory. The blood samples were tested for the presence of antibodies to hepatitis C antigen (anti-HCV) using a third-generation ELISA (Abbott Laboratories). The samples were also tested for the presence of hepatitis B surface antigen (HBsAg) and antibodies to hepatitis B core antigen (anti-HBc) using the ELISA.

Stata software version 13 was used for all statistical analyses. We conducted multivariable unconditional logistic regression analyses to identify significant independent risk factors. For each risk factor, the adjusted OR (AOR) and its corresponding 95% confidence interval (CI) were calculated using maximum likelihood estimation. All AORs and 95% CIs were adjusted for age, sex, race, state of residency, viral hepatitis infection, cigarette smoking, alcohol consumption, type 2 diabetes mellitus, family history of cancer, and presence of gallbladder stones. Analysis was performed in presence and absence of PSC and cirrhosis. Linear regression was used to estimate the mean differences in age of CCA onset associated with BMI. For all analyses, P < 0.05 was considered statistically significant.

Raw data that support the findings of this study are available from the corresponding author, upon request. Results output of the statistical analysis can also be shared.

Demographic characteristics of patients with CCA at diagnosis and control participants at enrollment are summarized in Table 1. Most patients with ICC and ECC were non-Hispanic White and older than 60 years of age. There was a significant difference in the mean age at diagnosis or enrollment between patients with CCA (60.97 ± 12.34 years) and control participants (59.66 ± 10.85 years; P = 0.043). Stratifying the age of diagnosis by sex, in ICC and ECC, showed no significant variation in the age of diagnosis between cases and controls in men and women separately. Table 1 shows that the majority of patients with CCA (56.8%) were referred from the Southern region of the United States. We found no significant difference in the geographic regions between cases and controls (P = 0.167).

Multivariable logistic regression showed that being Asian, having type 2 diabetes mellitus, family history of cancer, or gallbladder stones were associated with higher odds of developing ICC and ECC (Table 2). Being male (ICC AOR = 1.58) and having infection with viral hepatitis was associated with higher odds of developing ICC, but these risk factors were not associated with odds of developing ECC. We observed 50% and 65% reduction in the risk of developing ICC (P = 0.001) and ECC (P = 0.003), respectively, among cigarette smokers (≤20 pack-years) compared with nonsmokers. However, this protective effect among heavy smokers (>20 pack-years) was not statistically significant for ICC (P = 0.752) or ECC (P = 0.898). We continued to observe this protective effect among cigarette smokers when the groups were stratified by sex (ICC in women: AOR 0.392, 95% CI: 0.228–0.674; ICC in men: AOR 0.687, 95% CI: 0.337–1.399; ECC in women: AOR 0.478, 95% CI: 0.231–0.992; ECC in men: AOR 0.564, 95% CI: 0.356–0.896). We also observed this protective effect among current cigarette smokers stratified by sex, although the effect was not statistically significant (ICC in women: AOR 0.321, 95% CI: 0.089–1.157; ICC in men: AOR 0.490, 95% CI: 0.201–1.19; ECC in women: AOR 0.341, 95% CI: 0.088–1.313; ECC in men: AOR 0.679, 95% CI: 0.214–2.154). Examining the association of former smoking and obesity status in males and females did not meaningfully change the results. Interestingly, we observed that women who are not obese and smoked ≤20 pack-years continued to have a protective effect against the risk of CCA (AOR = 0.26, P < 0.001). Alcohol consumption was not associated with the odds of developing either ICC or ECC.

The association between obesity and odds of developing CCA is summarized in Table 3. Obesity at various periods of life was associated with higher odds of developing ICC in both men and women. Early adulthood obesity (ages 20–40 years) was associated with approximately three times higher odds of developing ICC in men and women, separately. Similarly, early adulthood obesity was associated with higher odds of developing ECC in both men and women. The mean difference in age at CCA diagnosis between obese patients and patients with normal BMI was statistically significant after adjustment for risk factors associated with age at CCA onset. Using linear regression, we estimated that each 1 unit of increase in BMI during early adulthood (mid-20s to mid-40s) before CCA diagnosis was associated with a 6.7-month (95% CI: −9.7 to −3.6) decrease in the age at CCA diagnosis (P < 0.001). The estimated coefficient was −6.1 (95% CI: −9.6 to −2.5) for ICC and −8.2 (95% CI: −14.13 to −2.3) for ECC after controlling for confounding factors. We observed significant correlation between self-reported weight and body size based on the Stunkard pictograms. The correlation coefficients ranged from 0.82 to 0.87 (P < 0.001) from teenage to diagnostic age.

Family history of cancer was also associated with odds of developing CCA (Table 4). The AORs for CCA, given a positive family history of any cancer, were 1.11 (95% CI: 1.06–1.48) for ICC and 1.32 (95% CI: 1.01–2.00) for ECC. We did not find an association between the number of affected family members (multiple members or single member) and the odds of developing either ICC or ECC. However, when looking at the type of relationship (parents, siblings, or offspring), we found that patients with CCA with a parent, particularly a father, with any type of cancer had higher odds of developing both ICC and ECC. Moreover, we analyzed a subgroup of subjects who had a positive first-degree family member (parent, sibling or offspring) with primary liver cancer. We identified 5 ICC, 8 ECC, and 14 control subjects that met these criteria. A positive family history of primary liver cancer, the number of affected family members, and the type of relationship in this subgroup were associated with the odds of developing ECC but not with ICC (Table 4).

Figure 1 illustrates the proportion of the most common cancers found in first-degree relatives of control participants, patients with ICC, and patients with ECC. Statistically significant differences were observed in the proportion of first-degree relatives with breast (P = 0.008), prostate (P = 0.009), and pancreatic (P < 0.001, possibly due to accrual bias) cancers between patients with ICC, patients with ECC, and control participants.

The most common self-reported medical conditions in patients with ICC and ECC are shown in Fig. 2A. Only thyroid condition (P = 0.02) and pancreatitis (P = 0.003) statistically differed between patients with ICC and ECC. Pathologic or radiologic evidence of PSC was observed significantly more often in patients with ECC than in ICC patients (P < 0.001; Fig. 2B). No such differences were observed between patients with ICC and ECC for cirrhosis (P = 0.074) or fibrosis/steatosis (P = 0.224). Excluding patients with PSC and cirrhosis from the case population did not meaningfully change the overall results of this report.

In this large hospital-based case–control study, we highlighted important independent risk factors for ICC and ECC, including race/ethnicity, obesity, type 2 diabetes mellitus, family history of cancer, and presence of gallbladder stones, independent of PSC and cirrhosis. Male patients and patients with viral hepatitis had higher odds of developing ICC. Our multivariable analysis showed a negative association of smoking and development of CCA among smokers. Our study also showed that having a first-degree relative with primary liver cancer was associated with higher odds of developing ECC. A prior history of obesity particularly in early adulthood (mid-20s to mid-40s) was associated with higher odds of developing ECC, as well as younger age at diagnosis of ICC or ECC, in both men and women.

To the best of our knowledge, this large-scale study is the first to highlight the association between early adulthood obesity and CCA in United States. The mean age at diagnosis of CCA in Western countries is 50 years, and most patients are older than 65 years at diagnosis (19). A recent study by Boonstra and colleagues showed that the age-adjusted incidence of ICC for patients younger than 65 years doubled during 2005–2014 compared with 1995–2004 (5). The association between obesity and CCA has been previously identified through meta-analyses and systematic reviews (20–22). A recent European study evaluating 2,200 patients with CCA from 11 European countries showed that more than 50% of patients were overweight or obese and 20% were diabetic at the time of diagnosis (23). One limitation of these studies was that they did not adjust for known risk factors, including viral hepatitis infection and diabetes mellitus.

Although the incidence of CCA does increase with age, the age-adjusted incidence of ICC doubled for patients younger than 65 years in 2005–2014 compared with 1995–2004 (5, 7). It is possible that such drop in the diagnostic age is attributable to obesity. In this study, early adulthood obesity was associated with early diagnosis of ICC. In the United States, obesity prevalence has increased from 30.5% in 1999–2020 to 41.9% in 2017–2020, and the prevalence of severe obesity has increased from 4.7% to 9.2% (24).

Our data showed that type 2 diabetes mellitus was associated with higher odds of developing CCA, for both ICC and ECC, supporting the findings of previous meta-analyses and population-based studies (7, 23, 25). Understanding the association of type 2 diabetes mellitus, age-related obesity, metabolic syndrome, and NAFLD with CCA, remain to be elucidated. Obesity is one of the underlying etiologies of non-alcoholic steatohepatitis (NASH), thus it can be associated with cirrhosis and has a well-known effect (26) and may play a role in cell growth, proliferation, fatty degeneration, and tumorigenesis of the hepatocytes and epithelial cells of bile duct (27). However, cirrhosis is not the only possible explanation for the association between obesity and CCA. In our cohort, 19.2% of patients with ICC and 12.4% of patients with ECC had underlying cirrhosis. However, subgroup analysis of these patients did not reveal any significant differences in BMI. Obese individuals also often develop some degree of insulin resistance, along with elevated insulin-like growth factors, which could also contribute to the development of CCA (28). Overall, the mechanisms involved in obesity and cirrhosis are still unknown. One potential explanation could be adipose tissue dysfunction and portal hypertension (29).

In Western countries, approximately 80% of CCAs are ECC (30). In the current study, we found that having a first-degree relative with primary liver cancer was associated with higher odds of developing ECC. These results support the idea that genetic factors play a role in risk of developing CCA. Although patients with Lynch syndrome are known to have increased lifetime risk of biliary tract cancers (5, 31), there are limited data regarding other hereditary predispositions to CCA. Recent studies using next-generation sequencing have identified genetic heterogeneity between ECC and ICC, and preliminary data suggest that potentially clinically relevant somatic and germline mutations may be prominent in CCA (32). Although we do not have information about germline genetic testing in this cohort as it is standard of care for CCA, prior studies results (33, 34) could explain our findings in Fig. 1 showing a higher proportion of patients with ECC with a family history of breast cancer compared with control participants. The findings of the current study, along with ongoing efforts showing the prevalence of germline mutations, support early and universal germline testing in patients with CCA, and this may have implications for the screening of family members, as well as potential therapeutic targets for patients harboring actionable germline mutations.

Despite the positive association between alcohol use and ICC, statistical analysis yields nonsignificant AORs. Our results were consistent with previously published American and Japanese studies (35, 36). Moreover, consistent with previously reported meta-analysis of 11 case–control studies (37), we conclude no association between alcohol consumption and ECC in our study.

One of the interesting findings by our study is the negative association between cigarette smoking and both types of CCA. Previous epidemiologic studies have also reported protective associations between smoking cigarettes and the risk of other malignancies, particularly in hormonal-related cancers, such as endometrial, endometrioid, clear cell ovarian, breast, and thyroid cancers (38–40). Studies for estrogen-related cancers have reported that smoking has an antiestrogenic effect on circulating estrogen concentrations, thereby resulting in a lower relative BMI and an earlier age of menopause (38, 39). The relatively lower BMI may partially explain the inverse association between cigarette smoking and estrogen (38, 39). For thyroid cancer, lower levels of thyroid-stimulating hormone, along with a lower BMI, may influence the protective effect of tobacco against the risk of thyroid cancer (41). In our study, we observed a similarly protective effect of smoking against CCA in both men and women, when analyzed as separate groups.

Despite the inherent vulnerability of case–control studies, it is unlikely that the observed protective effect of cigarette smoking against CCA is due to the observational bias of the high prevalence of smoking in our controls and the inclusion of former smokers in our cigarette smoker category (i.e., anyone who had smoked ≥100 cigarettes during their lifetime). We continued to see a protective effect among current smokers for both women and men in the ICC and ECC groups. Furthermore, the prevalence of current cigarette smoking in our control group (11.4%) is similar to that of the general population in the United States reported in 2020 (12.5%; ref. 42). In addition, controls participating in our study have previously participated in research for hepatocellular carcinoma and pancreatic cancer, where smoking cigarettes has been reported to be a significant risk factor (43, 44).

Although this area remains to be investigated, another potential explanation for the protective effect of cigarette smoking against CCA could be related to the fact that cumulative smoking may lead to a decrease in serum bilirubin, an endogenous antioxidant that can be scavenged by free radical species found in cigarette smoke (45). Elevated bilirubin has also been associated with the development of CCA in patients with PSC (46). Although cigarette smoking has been shown to increase the risk of chronic inflammatory diseases, such as PSC and inflammatory bowel disease (26, 29, 47, 48), the magnitude of this effect could not be identified in the current study because we did not collect this information from the control participants. Choi and colleagues showed that cigarette smoking was associated with CCA (AOR 1.3, 95% CI: 1.1–1.5) after adjusting for PSC (29, 47). The exact mechanism of action for nicotine and cigarette smoking in these patient populations remains unknown, and any inference on their possible mechanisms remains hypothetical.

Consistent with previous reports (49, 50), our study showed that gallbladder stones are a risk factor for CCA. Although the carcinogenic mechanisms of CCA are not well understood, chronic inflammation near the site of the gallbladder stones may be the underlying culprit (51). The presence of a network of cytokines and molecules present in high concentrations can trigger and maintain the chronic inflammatory process, and ultimately transforms the biliary epithelium into cholangiocarcinogenesis (52).

One limitation of our study is relying on hospital-based case referral, where all patients were ascertained from a large cancer center. However, due to disease rarity (1.26 per 100,000 persons per year; ref. 53) and the challenge of accurate diagnosis of CCA, we believe that hospital base-case-control design is the ultimate method for large-scale studies without misclassification bias. Moreover, to avoid selection bias, our case patients had pathologic and clinical confirmed diagnosis of CCA. Consistent with the natural history of CCA (53), approximately 72.9% of our patients with CCA were diagnosed with advanced-stage disease (tumor–node–metastasis III–IV) at time of initial evaluation with rapid fatality. Control subjects were healthy individuals selected from genetically unrelated family members (spouses), of patients with non-liver cancer at our institution, whose reasons for accompanying patients (altruism, care, love) were not related to risk factors being studied; therefore, these control subjects represented the study base from which our cases were selected (54).

Another limitation of our study was relying on hospital-based recruited patients who recalled prior history of weight and height. However, we observed consistent agreement between self-reported weight and body size (Stunkard pictograms) across different ages for patients with CCA and control participants. Moreover, we found no discrepancy between information obtained in interviews and patient records with respect to CCA risk factors, such as type 2 diabetes mellitus, indicating reliability and validity of self-reported diseases and medical conditions (55–57). In addition, prior studies have reported high correlations between recalled and measured weight and height in young adulthood among middle-aged and older men and women (58–61).

Control participants were selected to represent the population from which patients with CCA were ascertained. Only patients and controls from the United States were included, and the geographic distribution of their residential states was similar. Therefore, it is unlikely that our results were confounded by selection bias of cases or controls. The prevalence of obesity in control participants was 30.6%, which is consistent with the prevalence of obesity in the United States estimated by Ng and colleagues (62).

In summary, our findings indicate that CCA is multifactorial in origin and that environmental factors and history of chronic medical conditions may contribute to the disease etiology. Given the significant impact of positive family history of cancer, especially primary liver cancer, on the risk of CCA, a role for genetic factors in CCA development cannot be excluded. Future collaborations between different U.S. sites may be warranted to highlight the epidemiology of CCA in United States.

A. Rashid reports grants from NIH during the conduct of the study; grants from NIH outside the submitted work. P.K. Jalal reports grants from Dora Roberts Foundation during the conduct of the study. C.I. Amos reports grants from NCI during the conduct of the study. R.T. Shroff reports grants from Seagen, Rafael Pharm, Pieris, NUCANA, Novocure, LOXO, IMV Inc., Exelixis Pharm, Bristol-Myers Squibb, Bayer, BMS; non-financial support from AstraZeneca, Boehringer Ingelheim Pharm, Clovis, Genentech, Incyte, Zymeworks Biopharm, Astellas, SYROS, Natera, AbbVie, Hookipa Pharm, Due Oncology, and Ability Pharmaceuticals; grants and non-financial support from Taiho outside the submitted work. No disclosures were reported by the other authors.

R.I. Hatia: Conceptualization, resources, data curation, formal analysis, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. M. Eluri: Conceptualization, resources, data curation, formal analysis, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. E.T. Hawk: Funding acquisition, validation, investigation, writing–review and editing. A. Shalaby: Funding acquisition, validation, investigation, writing–review and editing. E. Karatas: Funding acquisition, validation, investigation, writing–review and editing. A. Shalaby: Funding acquisition, validation, investigation, writing–review and editing. A. Abdelhakeem: Funding acquisition, validation, investigation, writing–review and editing. R. Abdel-Wahab: Funding acquisition, validation, investigation, writing–review and editing. P. Chang: Funding acquisition, validation, investigation, writing–review and editing. A. Rashid: Funding acquisition, validation, investigation, writing–review and editing. P.K. Jalal: Funding acquisition, validation, investigation, writing–review and editing. C.I. Amos: Funding acquisition, validation, investigation, writing–review and editing. Y. Han: Funding acquisition, validation, investigation, writing–review and editing. T. Armaghany: Funding acquisition, validation, investigation, writing–review and editing. R.T. Shroff: Funding acquisition, validation, investigation, writing–review and editing. D. Li: Funding acquisition, validation, investigation, writing–review and editing. M. Javle: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. M.M. Hassan: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing.

This work was supported by the NIH RO1 grant CA098380 (D. Li), Linda Blum Fund, Mike Lu Foundation, and Stewart Mather Fund.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.