We assessed population incidence rates 1998–2002 and 5-year survival rates of 317 primary gallbladder cancer (GBC) entered in the population-based cancer registry in Valdivia. We analyzed GBC incidence (Poisson regression) and GBC survival (Cox regression). Cases were identified by histology (69.4%), clinical work-up (21.8%), or death certificate only (8.8%). Main symptoms were abdominal pain (82.8%), jaundice (53.6%) nausea (42.6%), and weight loss (38.2%); at diagnosis, 64% had Stage TNM IV. In the period, 4% of histopathological studies from presumptively benign cholecystectomies presented GBC. GBC cases were mainly females (76.0%), urban residents (70.3%), Hispanic (83.7%) of low schooling <4 years (64.0%). GBC standardized incidence rate per 100,000 (SIR) were all 17.5 (95%CI: 15.5–19.4), women 24.3, and men 8.6 (p < 0.00001); Mapuche 25.0, Hispanic 16.2 (p = 0.09). The highest SIRs were in Mapuche (269.2) and Hispanic women (199.6) with <4 years of schooling. Lowest SIRs were among Hispanic men (19.8) and women (21.9) with >8 years of schooling. Low schooling, female and urban residence were independent risk factors. By December 31, 2007, 6 (1.9%) cases were living, 280 (88.3%) died from GBC, 32 (10.1%) were lost of follow-up. Kaplan Meier Global 5-year survival was: 10.3%, 85% at stage I and 1.9% at stage IV; median survival: 3.4 months. Independent poor prognostic factors were TNM IV, jaundice and nonincidental diagnoses. Our results suggest that women of Mapuche ancestry with low schooling (>50 years) are at the highest risk of presenting and dying from GBC and should be the target for early detection programs.
Previous reports have demonstrated that gallbladder cancer (GBC) is the main cancer killer for Chilean women also frequent among men.1–4 It has a striking heterogeneous geographical distribution with most cases occurring in Southern Chile. At population level, main population risk factors are a high proportion of Mapuche ancestry, high poverty rates, high prevalence of gallbladder stones and historically high rates of typhoid fever.5 GBC mortality has not decreased in Chile, even though the country has experienced important socioeconomic development in the last decades.1 Worldwide, GBC is considered a poor prognostic cancer. Hospital-based studies have indeed demonstrated that Chilean survival rates are similar to those of developed countries.6–8 Thus, we hypothesized that the high-mortality rate of GBC in Chile is a reflection of a high baseline population incidence and the late stage at consultation.
Our aim was to characterize the GBC cases from a high-risk province in Southern Chile and to measure its population-based survival. The existence of the population-based Cancer Registry in Valdivia9 gave us the opportunity to conduct the first GBC population-based incidence and survival study in Chile.
Material and Methods
The Cancer Registry of Valdivia, initiated in 1982, covers 356,396 people in an area of 18,429 square kilometers in Southern Chile. Its organization and procedures have been described elsewhere.9, 10 In short, this is a systematic review of private and public records to identify every cancer case or cancer death occurring among residents in the Valdivia province, reaching almost 100% case ascertainment in the catchment area.
Cases and base population
A case of the study was an individual diagnosed with GBC (ICD-O code 23.9), identified in the Cancer Registry of Valdivia between 1998 and 2002. In total, 317 new cases were identified; 32 cases had information from a death certificate or autopsy only (DCO). We estimated GBC survival with and without the DCO cases. Deaths occurring within 30 days of surgery with curative intention (n = 5) were considered deaths from other causes.11 They were included but censored in the survival analysis. All histopathological studies of gallbladder from gallbladder and biliary tract surgery performed between 1998 and 2002 in the Province were analyzed to identify GBC. Stage of the tumor was based on TNM classification system.12 We registered number and size of gallbladder stones as described in the ultrasound or in the surgical report.
Standardized incidence rates using the world population, cumulative risk of developing cancer before the age of 75 and rate ratios were estimated according to the methodology proposed by IARC.13 Population data was obtained from the National Institute of Statistics and the Chilean census of 2002.14 Risk of developing GBC was estimated by sex, age, ethnicity (considering anyone with at least one Mapuche surname to be of Mapuche ethnicity, and the rest of the population to be Hispanic), education level (0–3 years, 4–8 years and >8 years of education), urban or rural residency. Significant variables in univariate analysis were included in a Poisson multivariate analysis, for the whole group and separately for men and women.
Specific survival rates were measured at 3 and 5 years using the Kaplan Meier analysis and survival curves for selected characteristics were compared using the Log-rank test.15 Those variables significantly associated with survival were later included in a Cox multivariate model.15 Survival was studied in relation to demographics, clinical presentation, laboratory results and medical management of the cases. Time under observation was counted from the date of first diagnoses to the date when a patient was last seen or date of death or the end of the study, December 31, 2004. SPSS 15.0 (SPSS, Inc., Chicago, IL) was used for the data analysis.
The Cancer Registry of Valdivia did not collect systematically data about personal risk factors such as nutrition, smoking, history of infectious diseases or parental history of cancer. However, we were able to collect data on gallbladder stones and ethnicity of each patient, which are major risk factors of GBC.
We identified a total of 317 GBC cases among residents in the Valdivia province whose first diagnosis occurred between 1998 and 2002, with a crude incidence rate of 17.8 per 100,000 and age standardized incidence rate per 100,000 (SIR) of 17.5 (95%CI: 15.5–19.4). Of all cases, 76% were women, which had a crude incidence rate and SIR of 26.7 and 24.3 (95%CI: 21.2–27.4), respectively. Men had a crude incidence rate and a SIR of 8.4 and 8.6 (95%CI: 6.7–10.6) respectively; thus, women had a relative risk of GBC of 2.82 (95%CI: 2.3–3.7) compared with men (Table 1). We calculated a cumulative risk of developing GBC before age 75 of 3.1% and 1.2% for women and men, respectively. Females were younger than males (mean age was 65 and 70 years, respectively, p = 0.006). GBC incidence rates increased steeply with age, among women peaked over 80 years for men at 70–79 years (207.8 and 110.1, respectively). Highest sex differential was under 60 years with 7.9 times higher incidence for women than men (Table 1). In all, 16.0% of cases were of Mapuche ethnicity, with an incidence rate of 25.0; this was a significant risk factor for women but not for men (Table 1). In multivariate Poisson regression analysis adjusted by age, schooling and place of residence (urban/rural), ethnicity was reached the limit of significance among women (Mapuche RR: 1.39 (0.96–1.97) but not among men (Mapuche RR: 1.11 (0.54–2.09). Even though only 30% of cases occurred in rural areas, rural residence was a protective factor in the multivariate analyses (RR 0.59, 95%CI: 0.46–0.76). Most cases (70%) belonged to low-socioeconomic status as measured by low education level attained (<8 years of schooling). Risk of GBC had an inverse linear trend by education years in both sexes, but it was much more marked for women. The risk differential from low to high schooling was 1.98 among men and 6.43 among women (Table 1). GBC risk was the same for women and men in higher education group (Table 1). All women in the high education group were Hispanic, i.e., there were no Mapuche women in this stratum. Mapuche women had an excess risk compared with Hispanic women when adjusted by level of education. The highest incidence—at age 50–79—occurred among Mapuche women with low schooling (269.2), being higher among women from urban than rural areas (IR: 425.1 and 217.4, respectively). The lowest IR occurred among people with >8 years of schooling, of Hispanic origin, either women 21.9 (22.9 urban, 15.6 rural) or men 19.8 (urban 18.8, rural 24.3). Data not presented. In the multivariate analysis, low schooling, feminine gender and urban residence were independent risk factors.
Table 1. Standardized incidence rate (SIR) Of GBC by selected characteristics, Valdivia, Chile, 1998–2002
We assessed the prevalence of GBC in the routinely histopathological studies of gallbladder surgery performed in Valdivia in the study period. A total of 5,390 histopathological studies were performed in Valdivia, 76.9% of them in women. GBC was found in 4.1% of the cases, with no difference by sex, but with a linear increase with age from 2.2% at 30–39 years to 24.6% at 50–59 years (data not presented). The precise location of the tumor was identified in 67.7% of the cases, for the remaining cases, the gallbladder could not be identified because of the massive involvement of the area with the tumor lesion.
Clinical characteristics and survival
Main ascertainment method (69.4%) was histological confirmation of the primary tumor or of metastasis; 8.8% of cases were identified by death certificate only; In 69 cases (21.8%), GBC was diagnosed based on clinical data. Among them, 56 (17.8%) were diagnosed as GBC Stage IV based on their clinical characteristics (silent jaundice), poor clinical evolution, and an imaging confirmation (abdominal ultrasonography, computed tomography or magnetic resonance cholangiopancreatography), which showed a complex mass in the gallbladder. The remaining 13 patients (4.1%) were diagnosed as GBC at advanced stage based on the medical history, the physical examination (weight loss, jaundice, palpable abdominal tumor) and blood and urine exams. Main clinical symptom at diagnoses was abdominal pain (83%) and in only 16.4% of cases, the diagnosis was incidental (Table 2). Gallbladder stones were diagnosed in 78.0% of cases (DCO excluded) without difference by gender or age. The majority of cases were at advanced stages of the disease at diagnoses (Stage IV 68%); while only 14.5% of cases were diagnosed in Stages I or II, being an incidental diagnoses most of these early stage cases. Most histological types (91.7%) were adenocarcinomas either tubular, papillar or tubulo-papilar, without gender differences, and 90% were poorly or moderately differentiated (Table 2). Global survival estimated with the direct survival method were 10.1% (95%CI: 8.9–11.3) and 8.5% (95%CI: 7.1–9.9), for 3-year and 5-year, respectively. With the “actuarial” survival method, the 3-year and 5-year global survival were 13.2% (95%CI: 9.5–16.9) and 9.0% (95%CI: 5.2–12.7%), respectively. Excluding DCO cases, did not change the global survival much at 3 years (14.5%, 95%CI: 10.4–18.7) nor at 5 years (9.9%, 95%CI: 5.7–14.0%). Similar results were obtained with Kaplan-Meier method, 3-year and 5-year survivals were 13.4% (95%CI: 9.5–17.2) and 10.3% (95%CI: 5.6–13.0), respectively. Median survival was 3.0 months including all cases and 3.4 cases excluding DCO cases. Most deaths occurred in the first 3 years, with a survival of 15.6% in women and 10.0% in men, similar to survival after 5 years of diagnosis. Age, ethnicity, years of education and place of residence were not significant for survival (Table 3). Main prognostic factor in the univariate or unadjusted analysis were incidental diagnosis of GBC, 55.9% survived 3 years, and TNM Stage—Stage I 85% and Stage IV 1.9% survived 3 years respectively (Table 3). When survival was estimated adjusting for all significant factors in a multivariate model, only staging, type of diagnoses and jaundice persisted as independent significant factors (Table 3). Compared with stage I, the relative risk of death increased 7 times in Stage III, and 27 times in Stage IV. Nonincidental diagnoses and the presence of jaundice each increased 3 times the risk of dying compared with incidental and no jaundice, adjusted by TNM (Table 4 and Figs. 1 and 2).
Table 2. Clinical characteristics of the GBC cases entered in the population-based cancer registry of Valdivia (1998–2002)
Table 4. Multivariate analysis of prognostic factors of GBC survival population-based registry of Valdivia. Kaplan Meier
This is the first population-based incidence and survival report in Chile, the country with the highest rates of incidence and mortality worldwide. Rates in the province of Valdivia are higher than the previously reported in population-based cancer registries worldwide.2–4, 16 Current IARC incidence data shows that women and men in Valdivia have the highest GBC incidence rate reported to date (24.3 and 8.6 per 100,000 population, respectively).17 In terms of GBC mortality, according to current IARC estimates, Chile ranks first among females with an age standardized mortality rate of 12.7 followed by the Czech Republic and Republic of Korea with 4.5 and 4.4, respectively; among males, Korea, Chile and Japan occupy first place with rates of 6.5, 6.2 and 5.1, respectively.18
In developed countries, GBC is uncommon, representing 2–3% of all malignant neoplasm and occurs mainly in Latin descendent and Native Americans of the South West; this is rare in Caucasian and exceptional in blacks.2, 3 GBC is more frequent in populations with a high prevalence of gallstones, such as the natives of North,19 Central and South America4 and residents of northeast Europe and Israel.2, 20 By contrast, this neoplasm is uncommon in populations with low incidence of stones, like most Asian, African and American blacks.3 Gallstones are strongly related with Amerindians genes21, 22 and have been associated with an increase in bile acid synthesis, polymorphisms in the lipid metabolism,23 and a high rate of cholesterol lithogenesis.21 Gallstones and GBC could be related to obesity and insulin resistance, which are very common in Amerindians in urban areas.24, 25 To support this, recent reports have associated gallstones and GBC with polymorphisms in lipid metabolism and insulin sensitivity in Chinese populations.26, 27
In our study, the GBC risk differential between Mapuche and Hispanic populations was lower than expected based on ecological studies.1, 5 In the study area of Southern Chile, most of the population has some Amerindian admixture even those who do not have a Mapuche name may have Mapuche genes. In fact, the Chilean population presents a very high Mapuche admixture estimated at 84% using the mitochondrial deoxyribonucleic acid polymorphism analysis.28 Because we classified Mapuche based on names an important misclassification of ethnicity is expected that will decrease the relative risks measured for this characteristic. Better classifications to more efficiently measure the ethnic variable will require molecular markers of ethnicity. Also in Chile, the Mapuche admixture has a high negative correlation with the socioeconomic level; among the highest socio-economic level, the population Mapuche admixture is less than 5%.29 Thus the high-relative risk observed among the low school years group in our study may in part be explained by a higher Mapuche admixture than among the high schooling group. Thus, a better classification of Mapuche ethnicity will also permit a better measurement of the socio-economic variable per se. The disappearance of the excess of GBC risk in women in the highest school years group could be explained by the very low Mapuche admixture in this stratum. The excess of GBC risk among women of low schooling could be explained by genetic predisposition probably interacting with environmental factors. In our series, those residing in rural areas were protected and this is in agreement with ecological studies,5 suggesting that some environmental risk factors are more prevalent in urban areas, i.e., rate of typhoid infection,30 hormone use and obesity,31, 32 or some protective factors that are present in rural areas, i.e., higher intake of fruit and vegetables31 and higher physical activity.32 The protective effect of rural residence overcomes the fact that a rural population has a higher rate of poverty.33
We determined that GBC survival was very low with no differences by age, sex or schooling, coherent with previous reports showing that as of today the main factor for survival is stage at diagnoses in the United States,34 Japan,35 Mexico36 and India.37
The main strengths of our study are the complete case ascertainment in a very high-risk area of GBC with almost complete follow-up to measure survival. Availability of adequate data bases of population allowed us to calculate population incidence risk and mortality rates. In our series, 22% of cases lack histological confirmation and had a GBC diagnosis based on clinical and imaging findings; all were cases at advanced stage of the disease. Although clinical work-up is inaccurate to identify early GBC, it is accurate for advanced stages of the disease.38, 39 On the other hand, 28 patients (8.8%) lack both histology and clinical work-up and were diagnosed at death, based on medical history. In this situation, the GBC diagnoses is uncertain thus we estimated survival rates with and without these cases, with practically no change in the 5-year survival estimates (9.0% and 9.9%, respectively).
Main limitations are inherent to this type of study based on routinely collected data with insufficient information on risk factors like the classification of ethnicity already discussed. Our work provides an overview of the impact of GBC in a high-risk area, even though we did not explore pathogenesis or etiology, we expect our manuscript will facilitate and motivate the initiation of such highly needed studies.
Early detection of GBC is the main factor that could be modified to improve survival. Because incidental diagnoses of GBC have a high survival, increasing the rate of surgery of gallbladder disease may be an option for early detection. Nevertheless, because surgery is not free of risks (rate of surgical mortality varied between 0.09 to 0.12% in patients <50 years and 1.0 to 3.1% in patients >80 years40, 41), it is necessary to focus the preventive cholecystectomies on people at high risk of GBC. We demonstrated that Mapuche women of low schooling aged >50 yearsare at the highest risk of GBC and should be the target of any early detection program.
The authors thank Ms. Angélica Dominguez for her advice and collaboration in data analysis and Dr. Solana Terrazas for her assistance in the edition of the final document.