Gastroenterology 2014;146:989–994
CLINICAL—PANCREAS Mortality, Cancer, and Comorbidities Associated With Chronic Pancreatitis: A Danish Nationwide Matched-Cohort Study Ulrich Christian Bang, 1 Thomas Benfield,2,3 Lars Hyldstrup,1,3 Flemming Bendtsen, 3,4 and Jens–Erik Beck Jensen1,3 1
Department of Endocrinology, 2Department of Infectious Diseases, 4Department of Gastroenterology, Copenhagen University Hospital of Hvidovre, Hvidovre, Denmark; 3Department of Clinical Medicine, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
BACKGROUND & AIMS: We aimed to assess the risk of death, cancer, and comorbidities among patients with alcoholic and nonalcoholic chronic pancreatitis (CP). METHODS: We performed a nationwide retrospective cohort study, collecting data from Danish registries from 1995 through 2010. We evaluated the prevalences and incidences of death, cancers, and comorbidities among subjects with CP (cases) compared with ageand sex-matched individuals (controls). In total, 11,972 cases (71,814 person-years) and 119,720 controls (917,436 personyears) were included in the analysis. Hazard ratios (HR) were estimated by Cox proportional hazards regression. RESULTS: Forty-six percent of the cases died during the follow-up period, compared with 13.0% of controls (mean age, 63.7 vs 72.1 y; P < .0001), corresponding to a HR of 5.0 for CP (95% confidence interval [CI], 4.8–5.2). Cancer was a frequent cause of death among cases (10.2%) and controls (3.3%). Cancer (particularly pancreatic cancer) was a frequent cause of death among cases; the HR was 6.9 (95% CI, 7.5–11.8). Alcoholic CP did not produce a higher risk for cancer or death than nonalcoholic CP. Cerebrovascular disease (HR, 1.3; 95% CI, 1.2–1.4), chronic pulmonary disease (HR, 1.9; 95% CI, 1.8–2.1), ulcer disease (HR, 3.6; 95% CI, 3.3–3.9), diabetes (HR, 5.2; 95% CI, 5.0–5.6), and chronic renal disease (HR, 1.7; 95% CI, 1.5–1.9) occurred more frequently among patients with CP, but myocardial infarction did not (HR, 0.9; 95% CI, 0.8–1.0). CONCLUSIONS: Based on a Danish nationwide cohort study, individuals with CP are at higher risk for death from cancer (particularly pancreatic cancer) and have a higher incidence of comorbidities than people without CP. Keywords: Pancreas; Inflammation; Malignancy; Epidemiology.
C
hronic pancreatitis (CP) is a disease with a variety of etiologies that throughout the Western world is triggered most often by alcohol.1 Fibrosis of the pancreas often is accompanied by fat malabsorption, secondary diabetes, and an increased risk of pancreatic cancer.2,3 Earlier studies on mortality resulting from CP have documented an increased risk of death but have been based on cohorts with 82–249 patients, which have limited the statistical precision.1,4–6 In general, previous studies have included patients from tertiary referral hospitals and the aims of this study
were to evaluate the causes of death and types of cancer in patients with CP in a nationwide population-based cohort. Furthermore, we investigated the risk of other comorbidities.
Methods Study Population We performed a matched retrospective cohort study. Patients diagnosed with CP (cases) between 1995 and 2010 were identified from the Danish National Patient Register, which contains discharge diagnoses from all inpatient and outpatient contacts.7 We used the International Classification of Disease 10th edition codes K86.0 (alcoholic CP) and K86.1 (nonalcoholic CP). Each case was compared with 10 age- and sex-matched controls without CP, who were retrieved from the Danish Civil Registration System.8 CP cases could not enter the control group.
Outcomes Our primary outcome was risk of death and we retrieved all mortality diagnoses from the Danish Register of Causes of Death.9 The following International Classification of Disease 10th edition codes were used for mortality diagnoses: Cxx: malignancies, Kxx: alimentary tract, Ixx: circulatory system, Jxx: respiratory system, Exx: endocrine, Fxx: psychiatric, A0/A2–A9/Bxx: infectious, X6/ X7/X81–X84/Y870: suicide, and V/X1–X5/Y1–Y7/Y80–Y86/ Y872/Y8: accidents. Furthermore, we retrieved all discharge diagnoses assigned to the participants throughout the time of follow-up evaluation. We evaluated the incidences of the following gastrointestinal cancer types: C15: esophagus, C16: gastric, C17: small intestine, C18: colon, C22: hepatic, C25: pancreatic, together with C34: lung, and C43: melanoma. We further distinguished between cancer of the pancreatic head (C250), body (C251), tail (C252), and ducts of the pancreas (C253). When studying the timedependent relationship between the diagnosis of CP and pancreatic cancer, we only wanted to include newly diagnosed cases of CP and therefore omitted the prevalent cases of CP on
Abbreviations used in this paper: AMI, acute myocardial infarction; CI, confidence interval; CP, chronic pancreatitis; CPD, chronic pulmonary disease; HR, hazard ratio; IR, incidence rate; PY, person-year. © 2014 by the AGA Institute 0016-5085/$36.00 http://dx.doi.org/10.1053/j.gastro.2013.12.033
CLINICAL PANCREAS
See Covering the cover Synopsis on page 875.
990
Bang et al
January 1, 1997. We further disregarded the cases of pancreatic cancer that occurred within 1 year after the diagnosis of CP to account for pancreatic cancers that were present at the time of CP diagnosis. We evaluated the incidence of the following other comorbidities: I21–I23: acute myocardial infarction (AMI), I60–I69/ G45/G46: cerebrovascular disease, J40–J47/J60–J67/J684/J701/ J703/J841/J920/J961/J982/J983: chronic pulmonary disease (CPD), K221/K25–K28: ulcer disease, E100/E101/E109/E110/ E111/E119: diabetes, and I12/I13/N00–N05/N07/N11/N14/ N17–N19/Q61: renal disease.
Covariates
CLINICAL PANCREAS
We retrieved information on date of birth, sex, and socioeconomic status from the Danish Civil Registration System, which is updated daily and tracks changes in demographic characteristics of Danish residents.8 In the analysis of mortality and malignancy, we included the Charlson comorbidity index as a covariate, which was based on discharge diagnoses registered for each participant (Supplementary Table 1).10 In the analysis of malignancy, the Charlson index was based solely on morbidity diagnoses reported before the diagnosis of cancer.
Statistics Baseline characteristics were presented as means with standard deviation (SD) or medians with interquartile range (Q1–Q3) where appropriate. Cohort entry was the time of first CP diagnosis within the years 1995–2010 and each control was assigned a cohort entry date identical to the matching case’s date of CP diagnosis. Risk time was expressed in person-years (PYs) and was defined as the time from cohort entry until occurrence of an event, death, or end of follow-up evaluation (December 31, 2010). The results were reported in prevalences, incidence rates (IRs) in numbers per 1000 PYs, or hazard ratio (HR) with 95% confidence interval (CI). We used univariate and multivariate Cox proportional hazard models to assess the HR. The risk time was split into periods of 4 years and included as a covariate to account for nonproportional hazards throughout the follow-up period (extended Cox regression model). Persons with missing data were excluded from the analyses (0.03%). All analyses were performed using SAS 9.2 (SAS Institute, Inc, Cary, NC). The study was approved by the Danish Health and Medicines Authority and followed the regulations set up by the Danish Data Protection Agency.
Results In total, 11,972 persons (33.5% women) with CP were identified and matched with 119,720 controls with a median age of 54 years (Q1–Q3, 45–64 y). Alcohol as the etiology to CP was reported for 6306 (52.7%) of the CP cases. Baseline values are presented in Table 1.
Mortality All mortality diagnoses among the cases and controls are presented in Table 2. Of the 11,972 persons with CP, 5560 (46.4%) died during the follow-up period compared with 15,528 (13.0%) persons in the control group. Age at time of death was significantly lower in the CP group (63.7 y)
Gastroenterology Vol. 146, No. 4 Table 1.Baseline Characteristics of Cases and Controls
N Women, % Age at entry, mean (SD), y Deceased, n (%) Age at death, mean (SD), y Socioeconomic status, % Working Unemployed Retired Charlson index, median (Q1–Q3) 0 1–2 >2
CP
Controls
11,972 33.5 54.5 (14) 5560 (46) 63.8 (13)
119,720 33.5 54.5 (14) 15,528 (13) 73.1 (12)
28 14 58 2 (1–5) 22% 34% 44%
58 7 35 0 (0–1) 62% 25% 13%
compared with the control group (72.1 y; P < .0001). Mortality rates were 77.4 (95% CI, 75.4–79.5) in the CP cohort and 16.9 (95% CI, 16.7–17.2) per 1000 PYs in the control cohort, yielding an adjusted HR of 5.0 (95% CI, 4.8–5.2). Mortality rates increased considerably with age in both cases and controls (Supplementary Figure 1), but the adjusted relative risks of death were significantly higher for the younger CP cases than among older patients (P < .0001) (Figure 1). The excess mortality risk in CP patients was comparable between alcoholic and nonalcoholic CP (Cox regression analysis, P ¼ .7). The most frequently reported mortality diagnoses among the patients with CP were diseases of the alimentary tract (10.6%)—mainly resulting from CP and alcoholic cirrhosis—followed by cancer (10.2%) and circulatory system diseases (5.5%). In comparison, only 0.4% of the controls Table 2.Causes of Mortality Associated With CP Compared With Controls
Number Person-years
Death from all causes Malignancies Alimentary tract Circulatory system Respiratory system Endocrine disorder Psychiatric disorder Infectious disease Suicide Accident Missing diagnosis Other diagnosis
HRa
95% CI
13.0
5.0
4.8–5.2
3.3 0.4 3.2 1.0 0.4 0.4 0.1 0.1 0.3 2.7 1.1
1.4 26.1 1.9 3.3 4.2 6.3 4.4 3.5 4.1 N/A 1.3
1.3–1.5 23.1–29.4 1.7–2.1 2.8–3.8 3.6–4.9 5.4–7.5 3.2–6.0 2.6–4.7 3.5–5.0 N/A 1.1–1.4
CP
Controls
11,972 71,814
119,720 917,436
% of total
% of total
46.4 10.2 10.6 5.5 2.8 2.2 2.1 0.6 0.4 1.5 7.8 2.1
HR, hazard ratio. a Adjusted for Charlson index and socioeconomic status.
Cancer and Mortality in Chronic Pancreatitis
Figure 1. Risk of death for patients with CP compared with age- and sex-matched controls. The excess in mortality rate in the patients with CP compared with controls decreased significantly with age (P < .0001).
died of diseases of the alimentary tract, 3.3% died of malignancies, and 3.2% died of circulatory disorders. We found a significantly increased relative risk for each particular cause of death in the models adjusted for Charlson index and socioeconomic status, and cancer-related death was associated with an HR of 1.4 (95% CI, 1.3–1.5). Among the cases, pancreatic cancer was the most frequently reported cancer-related cause of death (37%), followed by lung cancer (21%). For the controls, pancreatic cancer was the cause in 5% and lung cancer was the cause in 26% of the cancer-related deaths (Supplementary Table 2).
Cancer: CP Cases vs Controls The prevalence of cancers occurring during the followup period were significantly higher among CP cases compared with controls (Table 3). Among CP cases, 13.6% experienced a cancer compared with 7.9% of the controls
(P < .0001), resulting in an adjusted HR of 1.2 (95% CI, 1.1–1.3). Cancers of the pancreas and liver were related with the highest excess risks in CP cases compared with controls with adjusted HRs of 6.9 (95% CI, 5.6–8.6) and 2.0 (95% CI, 1.3–3.1), respectively. Cancer of the colon occurred with a lower incidence among CP cases (adjusted HR, 0.6; 95% CI, 0.5–0.8), with a median diagnosis age of 67 years (Q1–Q3, 58–76 y) among CP cases and 71 years (Q1–Q3, 63–77 y) among controls (P ¼ .01). However, when we omitted socioeconomic status and Charlson index from the adjusted model we found a HR of 1.2 (95% CI, 1.0–1.5). A total of 795 pancreatic cancers were registered among both patients and controls. In total, 695 (87%) of these patients died during the follow-up period of 2491 PYs, corresponding to a mortality IR of 279 per 1000 PYs. Only 7.0% of the patients with pancreatic cancer underwent surgery, and patients with cancer of the pancreatic head were more likely to receive surgical treatment compared with patients with other pancreatic cancers (10.5% vs 5.3%; c2 test: P ¼ .007). The distribution of pancreatic cancer sites were comparable between CP cases and controls (P ¼ .8). The median survival time after the diagnosis of pancreatic cancer until death was 86 days (35–235 days). However, patients who underwent pancreatic resection survived substantially longer (499 days; Q1–Q3, 245–865 days; P < .0001). The survival time of patients who did not undergo surgery (P ¼ .4) and patients who did undergo surgery (P ¼ 1.0) was independent of the site of pancreatic cancer. To determine whether the duration of CP had an impact on the risk of pancreatic cancer, we stratified the risk time into 2–4, 5–8, and 9–16 years after the diagnosis of CP. During the first year after the diagnosis of CP a total of 338 patients were diagnosed with pancreatic cancers and these were ignored, as described earlier. The remaining 172 patients with pancreatic cancers that occurred among CP cases yielded an IR of 4.5 (95% CI, 3.7–5.4) per 1000 PYs during years 2–4 after CP diagnosis. That was significantly higher than the IR of 1.6 (95% CI, 1.1–2.2) during years 5–8 and the IR of 2.0 (95% CI, 1.4–2.8) during years 9–16 (P ¼ .007).
Table 3.Risk of Malignancy During Follow-up Period for CP vs Non-CP and for Alcoholic CP vs Nonalcoholic CP CP vs controlsa
Alcoholic vs nonalcoholic CPa
Cancer site
CP cases
Controls
HR
95% CI
HR
95% CI
Person-years Any cancer, N Any cancer, % Esophagus Stomach Small intestine Colon Liver Pancreas Lungs Hematologic Melanoma
68,245 1622 13.6 0.46 0.33 0.09 0.82 0.38 4.26 1.00 1.00 0.23
888,573 9488 7.9 0.21 0.21 0.03 0.83 0.08 0.24 0.44 0.79 0.35
1.2 1.4 1.0 2.9 0.6 2.0 6.9 1.5 0.9 0.6
1.1–1.3 1.0–1.9 0.7–1.5 1.4–6.1 0.5–0.8 1.3–3.1 5.6–8.6 1.2–1.8 0.7–1.1 0.4–0.9
0.9 1.4 0.9 0.9 0.8 1.1 0.7 0.7 0.5 0.4
0.8–1.0 0.8–2.5 0.5–1.7 0.3–2.9 0.5–1.2 0.5–2.1 0.5–1.0 0.5–1.0 0.3–0.7 0.2–0.9
a
Adjusted for Charlson index and socioeconomic status.
991
CLINICAL PANCREAS
April 2014
992
Bang et al
Gastroenterology Vol. 146, No. 4
The relative risk of pancreatic cancer among patients with CP compared with controls was significantly higher during the initial 2–4 years after cohort entry (adjusted HR, 14.6; 95% CI, 10.9–19.6) compared with the periods of 5–8 years (adjusted HR, 4.8; 95% CI, 3.2–7.1) and 9–16 years (adjusted HR, 4.8; 95% CI, 3.2–7.4).
Cancer: Alcoholic CP vs Nonalcoholic CP
CLINICAL PANCREAS
The rates of any cancer were not different between patients with alcoholic (n ¼ 6306; PY, 39,998) and nonalcoholic CP (n ¼ 5666; PY, 30,247), with a HR of 0.9 (95% CI, 0.8–1.0). Only hematologic cancers and melanoma were less frequent among alcoholic CP (Table 3). The distribution of pancreatic cancer sites was identical in alcoholic and nonalcoholic CP (P ¼ .3), and the survival time after diagnosis of pancreatic cancer was comparable (median, 68 vs 105 days; P ¼ .1). Only 3% of the patients with alcoholic CP underwent pancreas resection owing to pancreatic cancer compared with 9% of the nonalcoholic CP cases (P ¼ .01).
Other Comorbidities The prevalences of other complications are presented in Table 4. The proportion of patients with any morbidity excluding CP was higher among the CP cases (78%) compared with the controls (38%) (c2, P < .0001). In a model adjusted for socioeconomic status, diabetes, and CPD, we found the HR of AMI among CP cases to be 0.9 (95% CI, 0.8–1.0). We found that cerebrovascular diseases and CPD occurred more often among CP cases compared with controls, but the diseases related to the highest risks were ulcers (HR, 3.6; 95% CI, 3.3–3.9) and diabetes (HR, 5.2; 95% CI, 5.0–5.6) (Table 4).
Discussion This nationwide population-based cohort study shows a highly increased mortality rate with a life expectancy that was approximately 8 years less among the CP cases compared with population controls. Our results illustrate the great impact the presence of this disease has on the accompanying complications. Malignancies of the alimentary system added substantially to the observed higher risk
of death. Chronic pancreatitis was associated with an increased risk of comorbidity during follow-up evaluation.
Mortality We showed a mortality rate of 77 persons per 1000 PYs and a 5-fold increased risk of death among patients with CP compared with sex- and age-matched controls. The large sample size allowed us to stratify the mortality estimate with the finding of a highly increased relative mortality risk in younger individuals. Until now, only a few smaller studies have compared the mortality risk among CP cases with the background mortality rate. Among 189 patients with hereditary CP, 19 patients died during follow-up evaluation and a lower relative mortality risk was found in younger individuals in contrast to the present study.4 A Danish study reported an HR of 4.3–4.5 among 249 patients with mainly alcoholic CP.1 Earlier studies with smaller cohorts agree on cancer being the main cause of death but have reported mortality rates with great variations. In a study with 1018 CP patients who had undergone endoscopy, 124 died during a mean follow-up time of 4.9 years, yielding a death rate of 29 persons per 1000 PYs. Pancreatic cancer was the cause of death in 24 (19%) of the deceased.11 In a later study with 107 patients suffering from mainly alcoholic CP, 27 patients died (25%) during a mean follow-up time of 10 years from the onset of CP, which corresponds to a death rate of 25 persons per 1000 PYs. Furthermore, it was shown that cancer was the most frequent mortality diagnosis (29% of the deceased), and a mortality odds ratio of 5.0 among patients with alcoholic CP compared with nonalcoholic CP was found.6 In a smaller study with 82 patients with mainly alcoholic CP, 14 patients died (17%) during an average follow-up time of 25 months, which results in a mortality rate of 82 per 1000 PYs. In that study, no difference in mortality rate between patients with alcoholic CP and other CP patients was found. Three patients of the cohort (21%) died of pancreatic cancer.5 It is interesting that the mortality rates vary so much between earlier studies and the present study. We showed a large increase in mortality rate with age, and thus age could be the reason for the discrepancy reported in earlier studies.
Table 4.Comorbidities During Follow-up Period for CP Compared With Controls Cohort CP Myocardial infarction Cerebrovascular disease Chronic pulmonary diseasea Ulcer disease Diabetesb Chronic renal disease
245 1011 1107 1015 1822 508
(3.1%) (8.4%) (9.3%) (8.5%) (15.2%) (4.2%)
Unadjusted HR Controls 1970 6182 5102 2342 3883 1831
(2.3%) (5.2%) (4.3%) (2.0%) (3.2%) (1.5%)
Adjusted HR
HR
95% CI
HR
95% CI
1.6 2.2 2.9 5.7 6.7 3.7
1.4–1.9 2.0–2.3 2.7–3.1 5.3–6.2 6.3–7.1 3.4–4.1
0.9 1.3 1.9 3.6 5.2 1.7
0.8–1.0 1.2–1.4 1.8–2.1 3.3–3.9 5.0–5.6 1.5–1.9
NOTE. Values were adjusted for socioeconomic status, chronic pulmonary disease, and diabetes. a Chronic pulmonary disease was not included as a covariate. b Diabetes was not included as a covariate.
However, the mean age ranged from 49 to 52 years in the earlier-mentioned studies, and thus, age is hardly the explanation. The fractions of men varied, ranging from 59% to 86%, and the highest mortality rate actually was found in the study with the highest fraction of men. In our study, men had an increased mortality risk of 10%, which is nevertheless too small to explain the large discrepancies in mortality rates reported. The prevalences of diabetes were evenly present in the earlier studies. However, resection of the pancreas was reported more frequently in the study with the lowest death rate, which also found lower death rates among the patients who actually underwent surgery for CP.6 Likewise, in the study with the lowest frequency of surgery, the highest mortality rate was reported.5 In contrast, in an earlier Danish study, no association between mortality and surgery for CP was found.1 It is thus beyond a doubt that CP is associated with increased mortality risk when compared with the rates of a background population. That goes for both alcoholic CP and hereditary CP.1,4 Our study showed dramatically higher mortality risks among younger patients and we speculate whether this was driven mainly by lifestyle factors, which we were unable to adjust for in this epidemiologic set-up. The older age groups may be able to compensate for more advanced CP accompanied by fat malabsorption and secondary diabetes by adopting a healthier lifestyle including tobacco and alcohol cessation, as well as better compliance with routine follow-up evaluation and medication.
Cancer With regard to cancer we were able to show that pancreatic cancer is the main cause of death among CP patients. Pancreatic cancer is a very severe clinical condition, which often is diagnosed too late to allow for surgical intervention. Because symptoms of pancreatic cancer may resemble symptoms and clinical findings in CP, the diagnosis of pancreatic cancer often will be delayed further. This is well illustrated by the low fraction of patients with pancreatic cancer who received surgical treatment in our cohort. We showed that CP is associated strongly with pancreatic cancer, and this is in concordance with earlier epidemiologic studies. Lowenfels et al12 published a study that included 2015 patients with CP, of whom 56 (2.8%) developed pancreatic cancer. When comparing the rate with the background incidence of pancreatic cancer, they found an HR of 14.4. A much lower relative risk was found in a Swedish population-based study from 1994, in which 7956 persons with discharge diagnoses of CP and acute pancreatitis were included. A total of 46 (0.6%) developed pancreatic cancer and they found an HR of 3.8. Similar to our finding, a more pronounced excess risk of pancreatic cancer during the first 2–4 years after the CP diagnosis was shown.13 In later published smaller cohort studies, the reported HRs have shown great variability: 13.314 and 27.2,15 but these studies suffered from low numbers of cases with pancreatic cancer. Recently, a large retrospective study showed a clear relationship between CP and pancreatic cancer with an HR of 11.8, together with a lower risk in
Cancer and Mortality in Chronic Pancreatitis
993
patients who underwent resection of the pancreas.16 Thus, the present study confirms that CP is closely related to the risk of pancreatic cancer. However, we were further able to show that the incidence was highest during the first years after the diagnosis of CP and thereafter remained stable. The link between CP and pancreatic cancer is probably the local influence of the inflammatory process, although this relationship may not be mirrored in the inflammatory markers in the blood.17 We could not show any increased risk of pancreatic cancer among patients with alcoholic CP compared with nonalcoholic CP, which is in contrast to several earlier studies on pancreatic cancer.18,19 However, this study was not designed to validate the association between alcohol intake and cancer risk similar to a recent study that showed a dose-response relationship between alcohol consumption and cancer risk.18 The seemingly reduced risk of colon cancer among the CP cases probably was owing to the high death rate in this group. Colon cancer is most frequent in persons older than age 50, and with a lifetime that is approximately 8 years shorter among CP patients, colon cancer is less likely to develop.
Comorbidities We found increased risks of several comorbidities. Diabetes was expectedly the disease with the highest risk. The risk of CPD also was higher among CP cases. Smoking has a great impact on the risk of CP and therefore we expected a higher incidence of CPD in this group compared with controls.20 With regard to the development of arteriosclerotic diseases, it is evident from our results that CP cases are not at a greater risk compared with controls. The risk of AMI was higher among CP cases, but when we adjusted for the prevalence of CPD and diabetes that difference disappeared, and cases in fact had a lower risk of myocardial infarction during the follow-up period. The Danish registers are maintained by the Danish Health and Medicines Authority and the reporting of discharge diagnoses to the registers is statutory. Hence, the registers comprise all Danish inpatients and outpatients. The limitations lie within the reliability of the diagnoses reported to the registers. The criteria for CP may vary from hospital to hospital but nevertheless are comparable because the diagnoses are established by specialists, who follow national guidelines. The same considerations apply to the mortality and cancer diagnoses but we expect no systemic bias in our material. Regarding the control group it can be argued that this consisted mainly of persons who do not undergo regular physical examinations as do patients with CP. This could result in under-reporting among the controls. Unfortunately, the databases do not include information on lifestyle factors such as smoking and alcohol consumption, which was a weakness of our study. However, we were able to adjust for a number of comorbidities at baseline including diabetes and chronic pulmonary disease using the Charlson comorbidity index. We expected a higher prevalence of smokers among the cases and this was reflected in the prevalence of CPD, which we accordingly adjusted for in our analyses. The diagnosis of CPD will not comprise all
CLINICAL PANCREAS
April 2014
994
Bang et al
smokers, however, but should be useful to mirror the difference of exposure to tobacco between the cases and controls. The strengths of this study included the clinical data, the long duration of the observation periods, the size of the study, and the nationwide covering.
Conclusions We conclude that patients with CP have a highly increased mortality rate and that malignancy-related death, especially from pancreatic cancer, was the most common reason. In our cohort, younger patients in particular had a higher risk of death. Comorbidities occurred more often during follow-up evaluation for patients with chronic pancreatitis, contributing to overall increased morbidity and mortality.
Supplementary Material CLINICAL PANCREAS
Note: To access the supplementary material accompanying this article, visit the online version of Gastroenterology at www.gastrojournal.org, and at http://dx.doi.org/10.1053/ j.gastro.2013.12.033.
References 1. Nojgaard C, Bendtsen F, Becker U, et al. Danish patients with chronic pancreatitis have a four-fold higher mortality rate than the Danish population. Clin Gastroenterol Hepatol 2010;8:384–390. 2. Kudo Y, Kamisawa T, Anjiki H, et al. Incidence of and risk factors for developing pancreatic cancer in patients with chronic pancreatitis. Hepatogastroenterology 2011; 58:609–611. 3. Pezzilli R, Morselli Labate AM, Fantini L, et al. Quality of life and clinical indicators for chronic pancreatitis patients in a 2-year follow-up study. Pancreas 2007;34:191–196. 4. Rebours V, Boutron-Ruault MC, Jooste V, et al. Mortality rate and risk factors in patients with hereditary pancreatitis: uni- and multidimensional analyses. Am J Gastroenterol 2009;104:2312–2317. 5. Seicean A, Tantau M, Grigorescu M, et al. Mortality risk factors in chronic pancreatitis. J Gastrointest Liver Dis 2006;15:21–26. 6. Thuluvath PJ, Imperio D, Nair S, et al. Chronic pancreatitis. Long-term pain relief with or without surgery, cancer risk, and mortality. J Clin Gastroenterol 2003;36:159–165. 7. Lynge E, Sandegaard JL, Rebolj M. The Danish National Patient Register. Scand J Public Health 2011;39:30–33. 8. Pedersen CB, Gotzsche H, Moller JO, et al. The Danish Civil Registration System. A cohort of eight million persons. Dan Med Bull 2006;53:441–449. 9. Helweg-Larsen K. The Danish Register of Causes of Death. Scand J Public Health 2011;39:26–29.
Gastroenterology Vol. 146, No. 4 10. Charlson ME, Pompei P, Ales KL, et al. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987; 40:373–383. 11. Rosch T, Daniel S, Scholz M, et al. Endoscopic treatment of chronic pancreatitis: a multicenter study of 1000 patients with long-term follow-up. Endoscopy 2002; 34:765–771. 12. Lowenfels AB, Maisonneuve P, Cavallini G, et al. Pancreatitis and the risk of pancreatic cancer. International Pancreatitis Study Group. N Engl J Med 1993; 328:1433–1437. 13. Ekbom A, McLaughlin JK, Karlsson BM, et al. Pancreatitis and pancreatic cancer: a population-based study. J Natl Cancer Inst 1994;86:625–627. 14. Talamini G, Falconi M, Bassi C, et al. Incidence of cancer in the course of chronic pancreatitis. Am J Gastroenterol 1999;94:1253–1260. 15. Wang W, Liao Z, Li G, et al. Incidence of pancreatic cancer in Chinese patients with chronic pancreatitis. Pancreatology 2011;11:16–23. 16. Ueda J, Tanaka M, Ohtsuka T, et al. Surgery for chronic pancreatitis decreases the risk for pancreatic cancer: a multicenter retrospective analysis. Surgery 2013;153: 357–364. 17. Dite P, Hermanova M, Trna J, et al. The role of chronic inflammation: chronic pancreatitis as a risk factor of pancreatic cancer. Dig Dis 2012;30:277–283. 18. Anderson MA, Zolotarevsky E, Cooper KL, et al. Alcohol and tobacco lower the age of presentation in sporadic pancreatic cancer in a dose-dependent manner: a multicenter study. Am J Gastroenterol 2012;107:1730–1739. 19. Brand RE, Greer JB, Zolotarevsky E, et al. Pancreatic cancer patients who smoke and drink are diagnosed at younger ages. Clin Gastroenterol Hepatol 2009; 7:1007–1012. 20. Andriulli A, Botteri E, Almasio PL, et al. Smoking as a cofactor for causation of chronic pancreatitis: a metaanalysis. Pancreas 2010;39:1205–1210.
Author names in bold designate shared co-first authors. Received April 24, 2013. Accepted December 20, 2013. Reprint requests Address requests for reprints to: Ulrich Bang, MD, PhD, Department of Endocrinology, Cophenhagen Hvidovre Hospital of Hvidovre, Kettegård Allé 30, 2650 Hvidovre, Denmark. e-mail: [email protected]; fax: (45) 3632-6109. Conflicts of interest The authors disclose no conflicts. Funding The study was supported by the research fund of the Department of Osteoporosis, University Hospital of Hvidovre, which took part in the design and interpretation of data.
April 2014
Cancer and Mortality in Chronic Pancreatitis 994.e1 Supplementary Table 2.Cancer-Related Mortality CP
Supplementary Figure 1. Mortality rates in patients with CP and age- and sex-matched controls.
Supplementary Table 1.Charlson Comorbidity Index Based on ICD-10 Codes Charlson comorbidity index Disorder
Weight
ICD-10
Myocardial infarction Congestive heart failure Peripheral vascular disease Cerebrovascular disease Dementia Chronic pulmonary disease
1 1 1 1 1 1
Connective tissue disease
1
Ulcer disease Mild liver disease
1 1
Diabetes type 1/2
1
Hemiplegia Moderate to severe renal disease Diabetes type 1/2 with end-organ damage Any tumor Leukemia Lymphoma Moderate to severe liver disease Metastatic solid tumor AIDS
2 2
I21, I22, I23 I50, I11.0, I13.0, I13.2 I70, I71, I72, I73, I74, I77 I60–I69, G45, G46 F00–F03, F05.1, G30 J40–J47, J60–J67, J68.4, J70.1, J70.3, J84.1, J92.0, J96.1, J98.2, J98.3 M05, M06, M08, M09, M30, M31, M32, M33, M34, M35, M36, D86 K22.1, K25–K28 K70.0–K70.3, K70.9, K71, K73, K74, K76.0 E10.0, E10.1, E10.9, E11.0, E11.1, E11.9 G81, G82 I12, I13, N00–N05, N07, N11, N14, N17–N19, Q61 E10.2–E10.8, E11.2–E11.8
2 2 2 2 2 6 6
C00–C43, C45–C75 C91–C95 C81–C85, C88, C90 B15.0, B16.0, B16.2, B19.0, K70.4, K72, K76.6, I85 C76–C80 B21–B24
AIDS, acquired immune deficiency syndrome; ICD-10, International Classification of Disease, 10th edition.
Controls
Cancer site
n
%
n
%
Stomach Esophagus Colon Pancreas Liver Lungs Bladder Unspecified location Miscellaneous Sex specific Breasta Ovariesa Prostateb Total
19 34 45 453 29 250 26 76 221 68 29 10 29 1221
1.6 2.4 3.7 37.1 2.4 20.5 2.1 6.2 18.1 5.6 6.7 2.3 3.7 100
127 120 337 206 64 1000 148 198 1059 666 227 85 354 3925
3.2 3.1 8.6 5.3 1.6 25.5 3.8 5.0 27.0 1.7 17.3 6.5 13.5 100
NOTE. The prevalences were stratified into different cancer sites among the persons who died of malignancy. a Only women were included. b Only men were included.
CLINICAL—PANCREAS Mortality, Cancer, and Comorbidities Associated With Chronic Pancreatitis: A Danish Nationwide Matched-Cohort Study Ulrich Christian Bang, 1 Thomas Benfield,2,3 Lars Hyldstrup,1,3 Flemming Bendtsen, 3,4 and Jens–Erik Beck Jensen1,3 1
Department of Endocrinology, 2Department of Infectious Diseases, 4Department of Gastroenterology, Copenhagen University Hospital of Hvidovre, Hvidovre, Denmark; 3Department of Clinical Medicine, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
BACKGROUND & AIMS: We aimed to assess the risk of death, cancer, and comorbidities among patients with alcoholic and nonalcoholic chronic pancreatitis (CP). METHODS: We performed a nationwide retrospective cohort study, collecting data from Danish registries from 1995 through 2010. We evaluated the prevalences and incidences of death, cancers, and comorbidities among subjects with CP (cases) compared with ageand sex-matched individuals (controls). In total, 11,972 cases (71,814 person-years) and 119,720 controls (917,436 personyears) were included in the analysis. Hazard ratios (HR) were estimated by Cox proportional hazards regression. RESULTS: Forty-six percent of the cases died during the follow-up period, compared with 13.0% of controls (mean age, 63.7 vs 72.1 y; P < .0001), corresponding to a HR of 5.0 for CP (95% confidence interval [CI], 4.8–5.2). Cancer was a frequent cause of death among cases (10.2%) and controls (3.3%). Cancer (particularly pancreatic cancer) was a frequent cause of death among cases; the HR was 6.9 (95% CI, 7.5–11.8). Alcoholic CP did not produce a higher risk for cancer or death than nonalcoholic CP. Cerebrovascular disease (HR, 1.3; 95% CI, 1.2–1.4), chronic pulmonary disease (HR, 1.9; 95% CI, 1.8–2.1), ulcer disease (HR, 3.6; 95% CI, 3.3–3.9), diabetes (HR, 5.2; 95% CI, 5.0–5.6), and chronic renal disease (HR, 1.7; 95% CI, 1.5–1.9) occurred more frequently among patients with CP, but myocardial infarction did not (HR, 0.9; 95% CI, 0.8–1.0). CONCLUSIONS: Based on a Danish nationwide cohort study, individuals with CP are at higher risk for death from cancer (particularly pancreatic cancer) and have a higher incidence of comorbidities than people without CP. Keywords: Pancreas; Inflammation; Malignancy; Epidemiology.
C
hronic pancreatitis (CP) is a disease with a variety of etiologies that throughout the Western world is triggered most often by alcohol.1 Fibrosis of the pancreas often is accompanied by fat malabsorption, secondary diabetes, and an increased risk of pancreatic cancer.2,3 Earlier studies on mortality resulting from CP have documented an increased risk of death but have been based on cohorts with 82–249 patients, which have limited the statistical precision.1,4–6 In general, previous studies have included patients from tertiary referral hospitals and the aims of this study
were to evaluate the causes of death and types of cancer in patients with CP in a nationwide population-based cohort. Furthermore, we investigated the risk of other comorbidities.
Methods Study Population We performed a matched retrospective cohort study. Patients diagnosed with CP (cases) between 1995 and 2010 were identified from the Danish National Patient Register, which contains discharge diagnoses from all inpatient and outpatient contacts.7 We used the International Classification of Disease 10th edition codes K86.0 (alcoholic CP) and K86.1 (nonalcoholic CP). Each case was compared with 10 age- and sex-matched controls without CP, who were retrieved from the Danish Civil Registration System.8 CP cases could not enter the control group.
Outcomes Our primary outcome was risk of death and we retrieved all mortality diagnoses from the Danish Register of Causes of Death.9 The following International Classification of Disease 10th edition codes were used for mortality diagnoses: Cxx: malignancies, Kxx: alimentary tract, Ixx: circulatory system, Jxx: respiratory system, Exx: endocrine, Fxx: psychiatric, A0/A2–A9/Bxx: infectious, X6/ X7/X81–X84/Y870: suicide, and V/X1–X5/Y1–Y7/Y80–Y86/ Y872/Y8: accidents. Furthermore, we retrieved all discharge diagnoses assigned to the participants throughout the time of follow-up evaluation. We evaluated the incidences of the following gastrointestinal cancer types: C15: esophagus, C16: gastric, C17: small intestine, C18: colon, C22: hepatic, C25: pancreatic, together with C34: lung, and C43: melanoma. We further distinguished between cancer of the pancreatic head (C250), body (C251), tail (C252), and ducts of the pancreas (C253). When studying the timedependent relationship between the diagnosis of CP and pancreatic cancer, we only wanted to include newly diagnosed cases of CP and therefore omitted the prevalent cases of CP on
Abbreviations used in this paper: AMI, acute myocardial infarction; CI, confidence interval; CP, chronic pancreatitis; CPD, chronic pulmonary disease; HR, hazard ratio; IR, incidence rate; PY, person-year. © 2014 by the AGA Institute 0016-5085/$36.00 http://dx.doi.org/10.1053/j.gastro.2013.12.033
CLINICAL PANCREAS
See Covering the cover Synopsis on page 875.
990
Bang et al
January 1, 1997. We further disregarded the cases of pancreatic cancer that occurred within 1 year after the diagnosis of CP to account for pancreatic cancers that were present at the time of CP diagnosis. We evaluated the incidence of the following other comorbidities: I21–I23: acute myocardial infarction (AMI), I60–I69/ G45/G46: cerebrovascular disease, J40–J47/J60–J67/J684/J701/ J703/J841/J920/J961/J982/J983: chronic pulmonary disease (CPD), K221/K25–K28: ulcer disease, E100/E101/E109/E110/ E111/E119: diabetes, and I12/I13/N00–N05/N07/N11/N14/ N17–N19/Q61: renal disease.
Covariates
CLINICAL PANCREAS
We retrieved information on date of birth, sex, and socioeconomic status from the Danish Civil Registration System, which is updated daily and tracks changes in demographic characteristics of Danish residents.8 In the analysis of mortality and malignancy, we included the Charlson comorbidity index as a covariate, which was based on discharge diagnoses registered for each participant (Supplementary Table 1).10 In the analysis of malignancy, the Charlson index was based solely on morbidity diagnoses reported before the diagnosis of cancer.
Statistics Baseline characteristics were presented as means with standard deviation (SD) or medians with interquartile range (Q1–Q3) where appropriate. Cohort entry was the time of first CP diagnosis within the years 1995–2010 and each control was assigned a cohort entry date identical to the matching case’s date of CP diagnosis. Risk time was expressed in person-years (PYs) and was defined as the time from cohort entry until occurrence of an event, death, or end of follow-up evaluation (December 31, 2010). The results were reported in prevalences, incidence rates (IRs) in numbers per 1000 PYs, or hazard ratio (HR) with 95% confidence interval (CI). We used univariate and multivariate Cox proportional hazard models to assess the HR. The risk time was split into periods of 4 years and included as a covariate to account for nonproportional hazards throughout the follow-up period (extended Cox regression model). Persons with missing data were excluded from the analyses (0.03%). All analyses were performed using SAS 9.2 (SAS Institute, Inc, Cary, NC). The study was approved by the Danish Health and Medicines Authority and followed the regulations set up by the Danish Data Protection Agency.
Results In total, 11,972 persons (33.5% women) with CP were identified and matched with 119,720 controls with a median age of 54 years (Q1–Q3, 45–64 y). Alcohol as the etiology to CP was reported for 6306 (52.7%) of the CP cases. Baseline values are presented in Table 1.
Mortality All mortality diagnoses among the cases and controls are presented in Table 2. Of the 11,972 persons with CP, 5560 (46.4%) died during the follow-up period compared with 15,528 (13.0%) persons in the control group. Age at time of death was significantly lower in the CP group (63.7 y)
Gastroenterology Vol. 146, No. 4 Table 1.Baseline Characteristics of Cases and Controls
N Women, % Age at entry, mean (SD), y Deceased, n (%) Age at death, mean (SD), y Socioeconomic status, % Working Unemployed Retired Charlson index, median (Q1–Q3) 0 1–2 >2
CP
Controls
11,972 33.5 54.5 (14) 5560 (46) 63.8 (13)
119,720 33.5 54.5 (14) 15,528 (13) 73.1 (12)
28 14 58 2 (1–5) 22% 34% 44%
58 7 35 0 (0–1) 62% 25% 13%
compared with the control group (72.1 y; P < .0001). Mortality rates were 77.4 (95% CI, 75.4–79.5) in the CP cohort and 16.9 (95% CI, 16.7–17.2) per 1000 PYs in the control cohort, yielding an adjusted HR of 5.0 (95% CI, 4.8–5.2). Mortality rates increased considerably with age in both cases and controls (Supplementary Figure 1), but the adjusted relative risks of death were significantly higher for the younger CP cases than among older patients (P < .0001) (Figure 1). The excess mortality risk in CP patients was comparable between alcoholic and nonalcoholic CP (Cox regression analysis, P ¼ .7). The most frequently reported mortality diagnoses among the patients with CP were diseases of the alimentary tract (10.6%)—mainly resulting from CP and alcoholic cirrhosis—followed by cancer (10.2%) and circulatory system diseases (5.5%). In comparison, only 0.4% of the controls Table 2.Causes of Mortality Associated With CP Compared With Controls
Number Person-years
Death from all causes Malignancies Alimentary tract Circulatory system Respiratory system Endocrine disorder Psychiatric disorder Infectious disease Suicide Accident Missing diagnosis Other diagnosis
HRa
95% CI
13.0
5.0
4.8–5.2
3.3 0.4 3.2 1.0 0.4 0.4 0.1 0.1 0.3 2.7 1.1
1.4 26.1 1.9 3.3 4.2 6.3 4.4 3.5 4.1 N/A 1.3
1.3–1.5 23.1–29.4 1.7–2.1 2.8–3.8 3.6–4.9 5.4–7.5 3.2–6.0 2.6–4.7 3.5–5.0 N/A 1.1–1.4
CP
Controls
11,972 71,814
119,720 917,436
% of total
% of total
46.4 10.2 10.6 5.5 2.8 2.2 2.1 0.6 0.4 1.5 7.8 2.1
HR, hazard ratio. a Adjusted for Charlson index and socioeconomic status.
Cancer and Mortality in Chronic Pancreatitis
Figure 1. Risk of death for patients with CP compared with age- and sex-matched controls. The excess in mortality rate in the patients with CP compared with controls decreased significantly with age (P < .0001).
died of diseases of the alimentary tract, 3.3% died of malignancies, and 3.2% died of circulatory disorders. We found a significantly increased relative risk for each particular cause of death in the models adjusted for Charlson index and socioeconomic status, and cancer-related death was associated with an HR of 1.4 (95% CI, 1.3–1.5). Among the cases, pancreatic cancer was the most frequently reported cancer-related cause of death (37%), followed by lung cancer (21%). For the controls, pancreatic cancer was the cause in 5% and lung cancer was the cause in 26% of the cancer-related deaths (Supplementary Table 2).
Cancer: CP Cases vs Controls The prevalence of cancers occurring during the followup period were significantly higher among CP cases compared with controls (Table 3). Among CP cases, 13.6% experienced a cancer compared with 7.9% of the controls
(P < .0001), resulting in an adjusted HR of 1.2 (95% CI, 1.1–1.3). Cancers of the pancreas and liver were related with the highest excess risks in CP cases compared with controls with adjusted HRs of 6.9 (95% CI, 5.6–8.6) and 2.0 (95% CI, 1.3–3.1), respectively. Cancer of the colon occurred with a lower incidence among CP cases (adjusted HR, 0.6; 95% CI, 0.5–0.8), with a median diagnosis age of 67 years (Q1–Q3, 58–76 y) among CP cases and 71 years (Q1–Q3, 63–77 y) among controls (P ¼ .01). However, when we omitted socioeconomic status and Charlson index from the adjusted model we found a HR of 1.2 (95% CI, 1.0–1.5). A total of 795 pancreatic cancers were registered among both patients and controls. In total, 695 (87%) of these patients died during the follow-up period of 2491 PYs, corresponding to a mortality IR of 279 per 1000 PYs. Only 7.0% of the patients with pancreatic cancer underwent surgery, and patients with cancer of the pancreatic head were more likely to receive surgical treatment compared with patients with other pancreatic cancers (10.5% vs 5.3%; c2 test: P ¼ .007). The distribution of pancreatic cancer sites were comparable between CP cases and controls (P ¼ .8). The median survival time after the diagnosis of pancreatic cancer until death was 86 days (35–235 days). However, patients who underwent pancreatic resection survived substantially longer (499 days; Q1–Q3, 245–865 days; P < .0001). The survival time of patients who did not undergo surgery (P ¼ .4) and patients who did undergo surgery (P ¼ 1.0) was independent of the site of pancreatic cancer. To determine whether the duration of CP had an impact on the risk of pancreatic cancer, we stratified the risk time into 2–4, 5–8, and 9–16 years after the diagnosis of CP. During the first year after the diagnosis of CP a total of 338 patients were diagnosed with pancreatic cancers and these were ignored, as described earlier. The remaining 172 patients with pancreatic cancers that occurred among CP cases yielded an IR of 4.5 (95% CI, 3.7–5.4) per 1000 PYs during years 2–4 after CP diagnosis. That was significantly higher than the IR of 1.6 (95% CI, 1.1–2.2) during years 5–8 and the IR of 2.0 (95% CI, 1.4–2.8) during years 9–16 (P ¼ .007).
Table 3.Risk of Malignancy During Follow-up Period for CP vs Non-CP and for Alcoholic CP vs Nonalcoholic CP CP vs controlsa
Alcoholic vs nonalcoholic CPa
Cancer site
CP cases
Controls
HR
95% CI
HR
95% CI
Person-years Any cancer, N Any cancer, % Esophagus Stomach Small intestine Colon Liver Pancreas Lungs Hematologic Melanoma
68,245 1622 13.6 0.46 0.33 0.09 0.82 0.38 4.26 1.00 1.00 0.23
888,573 9488 7.9 0.21 0.21 0.03 0.83 0.08 0.24 0.44 0.79 0.35
1.2 1.4 1.0 2.9 0.6 2.0 6.9 1.5 0.9 0.6
1.1–1.3 1.0–1.9 0.7–1.5 1.4–6.1 0.5–0.8 1.3–3.1 5.6–8.6 1.2–1.8 0.7–1.1 0.4–0.9
0.9 1.4 0.9 0.9 0.8 1.1 0.7 0.7 0.5 0.4
0.8–1.0 0.8–2.5 0.5–1.7 0.3–2.9 0.5–1.2 0.5–2.1 0.5–1.0 0.5–1.0 0.3–0.7 0.2–0.9
a
Adjusted for Charlson index and socioeconomic status.
991
CLINICAL PANCREAS
April 2014
992
Bang et al
Gastroenterology Vol. 146, No. 4
The relative risk of pancreatic cancer among patients with CP compared with controls was significantly higher during the initial 2–4 years after cohort entry (adjusted HR, 14.6; 95% CI, 10.9–19.6) compared with the periods of 5–8 years (adjusted HR, 4.8; 95% CI, 3.2–7.1) and 9–16 years (adjusted HR, 4.8; 95% CI, 3.2–7.4).
Cancer: Alcoholic CP vs Nonalcoholic CP
CLINICAL PANCREAS
The rates of any cancer were not different between patients with alcoholic (n ¼ 6306; PY, 39,998) and nonalcoholic CP (n ¼ 5666; PY, 30,247), with a HR of 0.9 (95% CI, 0.8–1.0). Only hematologic cancers and melanoma were less frequent among alcoholic CP (Table 3). The distribution of pancreatic cancer sites was identical in alcoholic and nonalcoholic CP (P ¼ .3), and the survival time after diagnosis of pancreatic cancer was comparable (median, 68 vs 105 days; P ¼ .1). Only 3% of the patients with alcoholic CP underwent pancreas resection owing to pancreatic cancer compared with 9% of the nonalcoholic CP cases (P ¼ .01).
Other Comorbidities The prevalences of other complications are presented in Table 4. The proportion of patients with any morbidity excluding CP was higher among the CP cases (78%) compared with the controls (38%) (c2, P < .0001). In a model adjusted for socioeconomic status, diabetes, and CPD, we found the HR of AMI among CP cases to be 0.9 (95% CI, 0.8–1.0). We found that cerebrovascular diseases and CPD occurred more often among CP cases compared with controls, but the diseases related to the highest risks were ulcers (HR, 3.6; 95% CI, 3.3–3.9) and diabetes (HR, 5.2; 95% CI, 5.0–5.6) (Table 4).
Discussion This nationwide population-based cohort study shows a highly increased mortality rate with a life expectancy that was approximately 8 years less among the CP cases compared with population controls. Our results illustrate the great impact the presence of this disease has on the accompanying complications. Malignancies of the alimentary system added substantially to the observed higher risk
of death. Chronic pancreatitis was associated with an increased risk of comorbidity during follow-up evaluation.
Mortality We showed a mortality rate of 77 persons per 1000 PYs and a 5-fold increased risk of death among patients with CP compared with sex- and age-matched controls. The large sample size allowed us to stratify the mortality estimate with the finding of a highly increased relative mortality risk in younger individuals. Until now, only a few smaller studies have compared the mortality risk among CP cases with the background mortality rate. Among 189 patients with hereditary CP, 19 patients died during follow-up evaluation and a lower relative mortality risk was found in younger individuals in contrast to the present study.4 A Danish study reported an HR of 4.3–4.5 among 249 patients with mainly alcoholic CP.1 Earlier studies with smaller cohorts agree on cancer being the main cause of death but have reported mortality rates with great variations. In a study with 1018 CP patients who had undergone endoscopy, 124 died during a mean follow-up time of 4.9 years, yielding a death rate of 29 persons per 1000 PYs. Pancreatic cancer was the cause of death in 24 (19%) of the deceased.11 In a later study with 107 patients suffering from mainly alcoholic CP, 27 patients died (25%) during a mean follow-up time of 10 years from the onset of CP, which corresponds to a death rate of 25 persons per 1000 PYs. Furthermore, it was shown that cancer was the most frequent mortality diagnosis (29% of the deceased), and a mortality odds ratio of 5.0 among patients with alcoholic CP compared with nonalcoholic CP was found.6 In a smaller study with 82 patients with mainly alcoholic CP, 14 patients died (17%) during an average follow-up time of 25 months, which results in a mortality rate of 82 per 1000 PYs. In that study, no difference in mortality rate between patients with alcoholic CP and other CP patients was found. Three patients of the cohort (21%) died of pancreatic cancer.5 It is interesting that the mortality rates vary so much between earlier studies and the present study. We showed a large increase in mortality rate with age, and thus age could be the reason for the discrepancy reported in earlier studies.
Table 4.Comorbidities During Follow-up Period for CP Compared With Controls Cohort CP Myocardial infarction Cerebrovascular disease Chronic pulmonary diseasea Ulcer disease Diabetesb Chronic renal disease
245 1011 1107 1015 1822 508
(3.1%) (8.4%) (9.3%) (8.5%) (15.2%) (4.2%)
Unadjusted HR Controls 1970 6182 5102 2342 3883 1831
(2.3%) (5.2%) (4.3%) (2.0%) (3.2%) (1.5%)
Adjusted HR
HR
95% CI
HR
95% CI
1.6 2.2 2.9 5.7 6.7 3.7
1.4–1.9 2.0–2.3 2.7–3.1 5.3–6.2 6.3–7.1 3.4–4.1
0.9 1.3 1.9 3.6 5.2 1.7
0.8–1.0 1.2–1.4 1.8–2.1 3.3–3.9 5.0–5.6 1.5–1.9
NOTE. Values were adjusted for socioeconomic status, chronic pulmonary disease, and diabetes. a Chronic pulmonary disease was not included as a covariate. b Diabetes was not included as a covariate.
However, the mean age ranged from 49 to 52 years in the earlier-mentioned studies, and thus, age is hardly the explanation. The fractions of men varied, ranging from 59% to 86%, and the highest mortality rate actually was found in the study with the highest fraction of men. In our study, men had an increased mortality risk of 10%, which is nevertheless too small to explain the large discrepancies in mortality rates reported. The prevalences of diabetes were evenly present in the earlier studies. However, resection of the pancreas was reported more frequently in the study with the lowest death rate, which also found lower death rates among the patients who actually underwent surgery for CP.6 Likewise, in the study with the lowest frequency of surgery, the highest mortality rate was reported.5 In contrast, in an earlier Danish study, no association between mortality and surgery for CP was found.1 It is thus beyond a doubt that CP is associated with increased mortality risk when compared with the rates of a background population. That goes for both alcoholic CP and hereditary CP.1,4 Our study showed dramatically higher mortality risks among younger patients and we speculate whether this was driven mainly by lifestyle factors, which we were unable to adjust for in this epidemiologic set-up. The older age groups may be able to compensate for more advanced CP accompanied by fat malabsorption and secondary diabetes by adopting a healthier lifestyle including tobacco and alcohol cessation, as well as better compliance with routine follow-up evaluation and medication.
Cancer With regard to cancer we were able to show that pancreatic cancer is the main cause of death among CP patients. Pancreatic cancer is a very severe clinical condition, which often is diagnosed too late to allow for surgical intervention. Because symptoms of pancreatic cancer may resemble symptoms and clinical findings in CP, the diagnosis of pancreatic cancer often will be delayed further. This is well illustrated by the low fraction of patients with pancreatic cancer who received surgical treatment in our cohort. We showed that CP is associated strongly with pancreatic cancer, and this is in concordance with earlier epidemiologic studies. Lowenfels et al12 published a study that included 2015 patients with CP, of whom 56 (2.8%) developed pancreatic cancer. When comparing the rate with the background incidence of pancreatic cancer, they found an HR of 14.4. A much lower relative risk was found in a Swedish population-based study from 1994, in which 7956 persons with discharge diagnoses of CP and acute pancreatitis were included. A total of 46 (0.6%) developed pancreatic cancer and they found an HR of 3.8. Similar to our finding, a more pronounced excess risk of pancreatic cancer during the first 2–4 years after the CP diagnosis was shown.13 In later published smaller cohort studies, the reported HRs have shown great variability: 13.314 and 27.2,15 but these studies suffered from low numbers of cases with pancreatic cancer. Recently, a large retrospective study showed a clear relationship between CP and pancreatic cancer with an HR of 11.8, together with a lower risk in
Cancer and Mortality in Chronic Pancreatitis
993
patients who underwent resection of the pancreas.16 Thus, the present study confirms that CP is closely related to the risk of pancreatic cancer. However, we were further able to show that the incidence was highest during the first years after the diagnosis of CP and thereafter remained stable. The link between CP and pancreatic cancer is probably the local influence of the inflammatory process, although this relationship may not be mirrored in the inflammatory markers in the blood.17 We could not show any increased risk of pancreatic cancer among patients with alcoholic CP compared with nonalcoholic CP, which is in contrast to several earlier studies on pancreatic cancer.18,19 However, this study was not designed to validate the association between alcohol intake and cancer risk similar to a recent study that showed a dose-response relationship between alcohol consumption and cancer risk.18 The seemingly reduced risk of colon cancer among the CP cases probably was owing to the high death rate in this group. Colon cancer is most frequent in persons older than age 50, and with a lifetime that is approximately 8 years shorter among CP patients, colon cancer is less likely to develop.
Comorbidities We found increased risks of several comorbidities. Diabetes was expectedly the disease with the highest risk. The risk of CPD also was higher among CP cases. Smoking has a great impact on the risk of CP and therefore we expected a higher incidence of CPD in this group compared with controls.20 With regard to the development of arteriosclerotic diseases, it is evident from our results that CP cases are not at a greater risk compared with controls. The risk of AMI was higher among CP cases, but when we adjusted for the prevalence of CPD and diabetes that difference disappeared, and cases in fact had a lower risk of myocardial infarction during the follow-up period. The Danish registers are maintained by the Danish Health and Medicines Authority and the reporting of discharge diagnoses to the registers is statutory. Hence, the registers comprise all Danish inpatients and outpatients. The limitations lie within the reliability of the diagnoses reported to the registers. The criteria for CP may vary from hospital to hospital but nevertheless are comparable because the diagnoses are established by specialists, who follow national guidelines. The same considerations apply to the mortality and cancer diagnoses but we expect no systemic bias in our material. Regarding the control group it can be argued that this consisted mainly of persons who do not undergo regular physical examinations as do patients with CP. This could result in under-reporting among the controls. Unfortunately, the databases do not include information on lifestyle factors such as smoking and alcohol consumption, which was a weakness of our study. However, we were able to adjust for a number of comorbidities at baseline including diabetes and chronic pulmonary disease using the Charlson comorbidity index. We expected a higher prevalence of smokers among the cases and this was reflected in the prevalence of CPD, which we accordingly adjusted for in our analyses. The diagnosis of CPD will not comprise all
CLINICAL PANCREAS
April 2014
994
Bang et al
smokers, however, but should be useful to mirror the difference of exposure to tobacco between the cases and controls. The strengths of this study included the clinical data, the long duration of the observation periods, the size of the study, and the nationwide covering.
Conclusions We conclude that patients with CP have a highly increased mortality rate and that malignancy-related death, especially from pancreatic cancer, was the most common reason. In our cohort, younger patients in particular had a higher risk of death. Comorbidities occurred more often during follow-up evaluation for patients with chronic pancreatitis, contributing to overall increased morbidity and mortality.
Supplementary Material CLINICAL PANCREAS
Note: To access the supplementary material accompanying this article, visit the online version of Gastroenterology at www.gastrojournal.org, and at http://dx.doi.org/10.1053/ j.gastro.2013.12.033.
References 1. Nojgaard C, Bendtsen F, Becker U, et al. Danish patients with chronic pancreatitis have a four-fold higher mortality rate than the Danish population. Clin Gastroenterol Hepatol 2010;8:384–390. 2. Kudo Y, Kamisawa T, Anjiki H, et al. Incidence of and risk factors for developing pancreatic cancer in patients with chronic pancreatitis. Hepatogastroenterology 2011; 58:609–611. 3. Pezzilli R, Morselli Labate AM, Fantini L, et al. Quality of life and clinical indicators for chronic pancreatitis patients in a 2-year follow-up study. Pancreas 2007;34:191–196. 4. Rebours V, Boutron-Ruault MC, Jooste V, et al. Mortality rate and risk factors in patients with hereditary pancreatitis: uni- and multidimensional analyses. Am J Gastroenterol 2009;104:2312–2317. 5. Seicean A, Tantau M, Grigorescu M, et al. Mortality risk factors in chronic pancreatitis. J Gastrointest Liver Dis 2006;15:21–26. 6. Thuluvath PJ, Imperio D, Nair S, et al. Chronic pancreatitis. Long-term pain relief with or without surgery, cancer risk, and mortality. J Clin Gastroenterol 2003;36:159–165. 7. Lynge E, Sandegaard JL, Rebolj M. The Danish National Patient Register. Scand J Public Health 2011;39:30–33. 8. Pedersen CB, Gotzsche H, Moller JO, et al. The Danish Civil Registration System. A cohort of eight million persons. Dan Med Bull 2006;53:441–449. 9. Helweg-Larsen K. The Danish Register of Causes of Death. Scand J Public Health 2011;39:26–29.
Gastroenterology Vol. 146, No. 4 10. Charlson ME, Pompei P, Ales KL, et al. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987; 40:373–383. 11. Rosch T, Daniel S, Scholz M, et al. Endoscopic treatment of chronic pancreatitis: a multicenter study of 1000 patients with long-term follow-up. Endoscopy 2002; 34:765–771. 12. Lowenfels AB, Maisonneuve P, Cavallini G, et al. Pancreatitis and the risk of pancreatic cancer. International Pancreatitis Study Group. N Engl J Med 1993; 328:1433–1437. 13. Ekbom A, McLaughlin JK, Karlsson BM, et al. Pancreatitis and pancreatic cancer: a population-based study. J Natl Cancer Inst 1994;86:625–627. 14. Talamini G, Falconi M, Bassi C, et al. Incidence of cancer in the course of chronic pancreatitis. Am J Gastroenterol 1999;94:1253–1260. 15. Wang W, Liao Z, Li G, et al. Incidence of pancreatic cancer in Chinese patients with chronic pancreatitis. Pancreatology 2011;11:16–23. 16. Ueda J, Tanaka M, Ohtsuka T, et al. Surgery for chronic pancreatitis decreases the risk for pancreatic cancer: a multicenter retrospective analysis. Surgery 2013;153: 357–364. 17. Dite P, Hermanova M, Trna J, et al. The role of chronic inflammation: chronic pancreatitis as a risk factor of pancreatic cancer. Dig Dis 2012;30:277–283. 18. Anderson MA, Zolotarevsky E, Cooper KL, et al. Alcohol and tobacco lower the age of presentation in sporadic pancreatic cancer in a dose-dependent manner: a multicenter study. Am J Gastroenterol 2012;107:1730–1739. 19. Brand RE, Greer JB, Zolotarevsky E, et al. Pancreatic cancer patients who smoke and drink are diagnosed at younger ages. Clin Gastroenterol Hepatol 2009; 7:1007–1012. 20. Andriulli A, Botteri E, Almasio PL, et al. Smoking as a cofactor for causation of chronic pancreatitis: a metaanalysis. Pancreas 2010;39:1205–1210.
Author names in bold designate shared co-first authors. Received April 24, 2013. Accepted December 20, 2013. Reprint requests Address requests for reprints to: Ulrich Bang, MD, PhD, Department of Endocrinology, Cophenhagen Hvidovre Hospital of Hvidovre, Kettegård Allé 30, 2650 Hvidovre, Denmark. e-mail: [email protected]; fax: (45) 3632-6109. Conflicts of interest The authors disclose no conflicts. Funding The study was supported by the research fund of the Department of Osteoporosis, University Hospital of Hvidovre, which took part in the design and interpretation of data.
April 2014
Cancer and Mortality in Chronic Pancreatitis 994.e1 Supplementary Table 2.Cancer-Related Mortality CP
Supplementary Figure 1. Mortality rates in patients with CP and age- and sex-matched controls.
Supplementary Table 1.Charlson Comorbidity Index Based on ICD-10 Codes Charlson comorbidity index Disorder
Weight
ICD-10
Myocardial infarction Congestive heart failure Peripheral vascular disease Cerebrovascular disease Dementia Chronic pulmonary disease
1 1 1 1 1 1
Connective tissue disease
1
Ulcer disease Mild liver disease
1 1
Diabetes type 1/2
1
Hemiplegia Moderate to severe renal disease Diabetes type 1/2 with end-organ damage Any tumor Leukemia Lymphoma Moderate to severe liver disease Metastatic solid tumor AIDS
2 2
I21, I22, I23 I50, I11.0, I13.0, I13.2 I70, I71, I72, I73, I74, I77 I60–I69, G45, G46 F00–F03, F05.1, G30 J40–J47, J60–J67, J68.4, J70.1, J70.3, J84.1, J92.0, J96.1, J98.2, J98.3 M05, M06, M08, M09, M30, M31, M32, M33, M34, M35, M36, D86 K22.1, K25–K28 K70.0–K70.3, K70.9, K71, K73, K74, K76.0 E10.0, E10.1, E10.9, E11.0, E11.1, E11.9 G81, G82 I12, I13, N00–N05, N07, N11, N14, N17–N19, Q61 E10.2–E10.8, E11.2–E11.8
2 2 2 2 2 6 6
C00–C43, C45–C75 C91–C95 C81–C85, C88, C90 B15.0, B16.0, B16.2, B19.0, K70.4, K72, K76.6, I85 C76–C80 B21–B24
AIDS, acquired immune deficiency syndrome; ICD-10, International Classification of Disease, 10th edition.
Controls
Cancer site
n
%
n
%
Stomach Esophagus Colon Pancreas Liver Lungs Bladder Unspecified location Miscellaneous Sex specific Breasta Ovariesa Prostateb Total
19 34 45 453 29 250 26 76 221 68 29 10 29 1221
1.6 2.4 3.7 37.1 2.4 20.5 2.1 6.2 18.1 5.6 6.7 2.3 3.7 100
127 120 337 206 64 1000 148 198 1059 666 227 85 354 3925
3.2 3.1 8.6 5.3 1.6 25.5 3.8 5.0 27.0 1.7 17.3 6.5 13.5 100
NOTE. The prevalences were stratified into different cancer sites among the persons who died of malignancy. a Only women were included. b Only men were included.
