Increased Risk of Pancreatic Adenocarcinoma After Acute Pancreatitis

Clinical Gastroenterology and Hepatology 2014;12:1143–1150

PANCREAS, BILIARY TRACT, AND LIVER Increased Risk of Pancreatic Adenocarcinoma After Acute Pancreatitis Satish Munigala,* Fasiha Kanwal,* Hong Xian,‡,§ Jeffrey F. Scherrer,‡,k and Banke Agarwal* *Division of Gastroenterology and Hepatology, and kDepartment of Family and Community Medicine, Saint Louis University School of Medicine, St Louis; ‡Veterans Affairs, St Louis Health Care System, St Louis; §Department of Biostatistics, Saint Louis University School of Public Health, St Louis, Missouri BACKGROUND & AIMS:

Acute pancreatitis (AP) is often the initial presentation of pancreatic cancer (pancreatic adenocarcinoma [PaCa]). We evaluated the risk of PaCa after AP.

METHODS:

We performed a retrospective study of patients with AP who sought care in the Veterans Health Administration from 1998 through 2007. We excluded patients with pre-existing PaCa or recurrent AP and those who had the first episode of acute pancreatitis, from 1998 through 2000.

RESULTS:

Of 495,504 patients with Veterans Health Administration inpatient and outpatient records, 5720 were diagnosed with AP (1.15%) and 710 were diagnosed with PaCa (0.14%), from 2000 through 2007. Seventy-six patients had AP within 2 years before being diagnosed with PaCa (10.7% of all patients with cancer diagnosed during that period). The risk for PaCa was greatest in the first year after AP (14.5 per 1000 patient-years) and then decreased rapidly. Risk for PaCa was negligible in patients 24 months for 10 patients.

CONCLUSIONS:

A significant number of patients with PaCa initially present with AP; the diagnosis of cancer is often delayed by up to 2 years. We suggest that PaCa be routinely considered as a potential etiology of AP in patients ‡40 years old.

Keywords: Pancreatic Cancer; Early Diagnosis; Acute Pancreatitis; Cancer Risk.

cute pancreatitis (AP) is often the initial clinical presentation of pancreatic adenocarcinoma (PaCa) and can precede the diagnosis of PaCa by several weeks or months.1,2 It is plausible that diagnostic evaluation of patients with AP for PaCa could potentially help in diagnosing more patients with early stage PaCa. However, it is currently not possible to make this determination because there are no published studies that have evaluated the incidence of PaCa diagnosis after an attack of AP. From 7 published studies that reported a detailed list of the etiology of AP in their study cohort, we estimated that 1.7%–3.6% of patients with AP were finally diagnosed to have PaCa; the pooled average was 2.03% in 2945 patients in these studies.3–9 Because these studies were cross-sectional, the temporal relationship between AP episode and subsequent PaCa diagnosis is not discernible. In this study we determined the proportion of patients with PaCa who had AP before a cancer diagnosis. We also estimated the proportion of patients with AP who were subsequently diagnosed to have PaCa and

A

identified patient characteristics that were associated with a high risk of PaCa diagnosis after an AP attack.

Methods Study Design and Participants A retrospective cohort study was conducted. Data for this study were obtained from Veterans Health Administration (VHA) inpatient and outpatient records by using the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnoses Abbreviations used in this paper: AP, acute pancreatitis; CBD, common bile duct; CI, confidence interval; EUS, endoscopic ultrasound; ICD-9-CM, International Classification of Diseases, Ninth Revision, Clinical Modification; PaCa, pancreatic adenocarcinoma; RR, relative risk; VA, Veterans Administration; VHA, Veteran Health Administration. © 2014 by the AGA Institute 1542-3565/$36.00 http://dx.doi.org/10.1016/j.cgh.2013.12.033

1144 Munigala et al

maintained by the VHA national medical care data sets from fiscal year 1998.10 Initial cohort identified for the study was 534,956 patients. In an attempt to identify only initial episode of AP, a 2-year washout period (October 1998–September 2000) was applied to rule out patients who had a preexisting PaCa (n ¼ 361), preexisting history of AP (single episode or recurrent episodes between 1998 and 2000, n ¼ 3046), and patients in the database who were lost to follow-up before October 2000 (n ¼ 28,859). All VA patients from the study cohort older than the age of 25 years were eligible for the study. This study was approved by Veterans Administration (VA), St Louis Heath Care System Institutional Review Board (protocol number 1153766).

Exclusion Criteria Patients younger than the age of 25 years (n ¼ 2157) and patients diagnosed with pancreatic cystic lesions (n ¼ 994) were excluded. Patients diagnosed to have PaCa within the first 7 days of entry into the study (n ¼ 78), patients who were diagnosed to have PaCa either before or within the first 7 days of diagnosis of AP (n ¼ 22), patients with a diagnosis of chronic pancreatitis before AP (n ¼ 551), and patients with prior diagnosis of chronic pancreatitis (n ¼3384) were also excluded from the study. Final cohort of VA patients included in the study was 495,504 (Figure 1). Follow-up period was from October 1, 2000–September 30, 2007.

Measurements Primary dependent variable. The primary outcome of interest was adenocarcinoma of the pancreas; this was defined on the basis of at least one ICD code 157.0–157.3, 157.9. For patients with AP, duration of the follow-up was calculated from the time of diagnosis of AP to the diagnosis of PaCa. For the patients without AP, duration of the follow-up was calculated from the time of entry into the study up to the diagnosis of PaCa. Primary independent variable. The primary predictor of interest was AP (ICD code 577.0). The number of episodes of AP before diagnosis of PaCa and the time interval between the first episode of AP and subsequent diagnosis of PaCa were calculated. AP diagnoses less than 15 days apart were considered as a single episode of pancreatitis. Covariates. These included sociodemographic variables including age at the time of entry into the study, race, and sex. Other covariates included alcohol abuse/ dependence (ICD codes 303, 3030, 3039, 3050), presence of gallstones (ICD codes 574, 5741, 5743, 5745, 5747, 5748, 5749), and nicotine dependence (ICD codes 305.1 or V1582) defined on the basis of >1 ICD codes before PaCa diagnosis or censorship.

Clinical Gastroenterology and Hepatology Vol. 12, No. 7

Statistical Analysis Demographic characteristics (race, sex) and secondary predictors (history of alcohol, smoking, and presence of gallstones) were assessed by using frequencies and proportions. Age was used as a continuous variable (mean  standard deviation). Proportion of patients with PaCa who had AP before cancer diagnosis and the proportion of patients with AP who were subsequently diagnosed to have PaCa were calculated. Incidence rates for PaCa (per 1000 person-years) among patients with and without AP were estimated. Rate ratios with 95% confidence intervals (CIs) were evaluated by using Poisson regression after adjusting for age, sex, smoking, alcohol status, and history of gallstones. Follow-up time for the study participants who developed PaCa ended on the date of first diagnosis of PaCa, and for those who did not develop PaCa it was censored at the time of loss to follow-up, death, or end of the study period. Stratified analyses were performed to determine the effect of alcohol abuse, use of tobacco, and presence of gallstone on AP and PaCa. Percentages of patients with number of episodes of AP in PaCa group and those without PaCa were also determined. All analyses were conducted by using SAS version 9.2 (SAS Inc, Cary, NC). Significant tests were done by using 2-tailed hypothesis, and the level of significance (a) was set to 0.05.

Results As shown in Figure 1, the final cohort included 495,504 patients. Five thousand seven hundred twenty (1.15%) had 1 or more attacks of AP, and PaCa was diagnosed in 710 patients (0.14%). Median follow-up was 60 months, 26 months, and 60 months for the entire cohort, patients with PaCa, and patients without PaCa, respectively. The demographic information of the veterans in our cohort is summarized in Table 1. The annual incidence of PaCa for the entire cohort was 0.31 per 1000 patient-years. The mean age of all the veterans in our database was 56  13.7 years. As expected from a VA-based database, male patients were predominant (89%). About 34% of patients were smokers, 25% were heavy alcohol drinkers, and 3% of the patients had gallstone disease. The mean age of patients with AP was 57  11.6 years, and for those with PaCa diagnosis it was 64  11.3 years. Among 710 patients with PaCa, 86 patients (12.1%; 95% CI, 9.7–14.5) had at least 1 or more episodes of AP before PaCa diagnosis, with 69 patients (9.7%; 95% CI, 7.5–11.9) and 76 patients (10.7%; 95% CI, 8.6–13.2) having AP within 1 and 2 years, respectively, before the diagnosis of PaCa. The number of patients diagnosed to have PaCa each year after an attack of AP is shown in Table 2. The number of patients with PaCa diagnoses after AP was significantly higher than that expected on the basis of the incidence of PaCa in remaining patients in the database. The incidence and

July 2014

Acute Pancreatitis and Pancreatic Cancer Risk 1145

Figure 1. Study cohort. CP, chronic pancreatitis.

relative risk (RR) of PaCa diagnosis after an episode of AP compared with remaining subjects in this database are summarized in Table 2. After adjusting for age, sex, race, smoking status, alcohol status, and history of gallstones, the higher risk after AP was most marked in the first year (RR, 66; 95% CI, 47.24–92.23) and rapidly

declined in the second year and became statistically similar to that in control group in the third year. The incidence of PaCa per 1000 patient-years within the first year after the first episode of AP stratified by patients’ age is summarized in Table 3. The risk of PaCa was negligible in patients younger than 40 years of age

1146 Munigala et al

Clinical Gastroenterology and Hepatology Vol. 12, No. 7

Table 1. Patient Characteristics One or more episodes of AP

Patients without AP

Patients with PaCa

Patients without PaCa

All study patients

5720 (100) 57  11.6

489,784 (100) 56  13.7

710 (100) 64  11.3

494,794 (100) 56  13.7

495,504 (100) 56  13.7

5328 (93%) 392 (7%) 0

435,308 (89%) 54,469 (11%) 7

691 (97%) 19 (3%) 0

439,945 (89%) 54,842 (11%) 7

440,636 (89%) 54,861 (11%) 7

3997 1548 175 3233 3248 1225

360,043 85,928 43,813 167,061 121,397 12,986

558 130 22 270 170 46

363,482 87,346 43,966 170,024 124,475 14,165

364,040 87,476 43,988 170,294 124,645 14,211

n (%) Age (y) (mean  standard deviation) Sex Male Female Unspecified Race White Black Other Smoking Alcohol History of gallstones

(70%) (27%) (3%) (57%) (57%) (21%)

(74%) (18%) (8%) (34%) (25%) (3%)

(79%) (18%) (3%) (38%) (24%) (6%)

(73%) (18%) (9%) (34%) (25%) (3%)

(73%) (18%) (9%) (34%) (25%) (3%)

NOTE. Age, average age at time of entry into study period; Other race includes Asian, American Indian, Pacific Islander, or unknown.

and increased gradually in each decade from 7.69 per 1000 patient-years in fifth decade of life and reaching 28.67 per 1000 patient-years after 70 years of age. The absolute risk of PaCa diagnosis after AP was highest in patients older than 70 years, but RR was most marked in the fifth decade of life (RR, 104.8; 95% CI, 43.4–252.8). Overall, in patients 40 years of age, the incidence of PaCa diagnosis within 1 year after an attack of AP was 14.48 per 1000 patient-years; RR was 66 (95% CI, 47.23–92.22). Figure 2A illustrates the proportion of patients with 1, 2, or 3 episodes of AP before diagnosis of PaCa in our cohort. Forty-one patients (47.7%; 95% CI, 37.4–58.1) had only a single preceding episode of AP, 19 patients (22.1%; 95% CI, 14.6–32.0) had 2 episodes, and 26 patients (30.2%; 95% CI, 21.5–40.6) had 3 episodes of AP before PaCa was diagnosed. Figure 2B summarizes the time to diagnosis of PaCa after AP in study patients. Thirty-four patients were diagnosed to have PaCa within 60 days of presentation with AP. PaCa was diagnosed 3–12 months later in 35 patients, 13–24 months later in 7 patients, and later than 24 months after AP in 10 patients. Table 4 summarizes RR of PaCa diagnosis within 1 year after AP in patients with history of heavy tobacco

use, alcohol abuse, and in those without gallstones. Presence of gallstones was used as a surrogate for gallstone pancreatitis because it was not possible to reliably identify patients with gallstone pancreatitis because of the nature of the database. The likelihood of PaCa diagnosis was lower in patients with a history of alcohol abuse (RR, 0.37; 95% CI, 0.22–0.62). Presence of gallstones (as potential etiology of AP) and history of tobacco use did not influence the likelihood of PaCa diagnosis after AP. PaCa was diagnosed within 2 years in 1.29% of patients (11 of 851) with heavy alcoholism, 2.63% of patients (22 of 836) with heavy smoking, 0.96% of patients (5 of 516) with gallstones, and 2.12% of patients (37 of 1744) without gallstones, alcoholism, or smoking history.

Discussion PaCa is recognized to be an important cause of AP.1–9 However, insufficient published data are available regarding the prevalence and clinical characteristics of patients with PaCa-associated AP to aid in clinical management. In this retrospective study that used VA-based

Table 2. Incidence of PaCa by Year and AP Status Patients with AP

Patients without AP

Incidence rate per 1000 person-years

Patients at Patients diagnosed Patients at Patients diagnosed Remaining risk (N) with PaCa [n (%)] risk (N) with PaCa [n (%)] AP patients without AP RRa Year Year Year Year

a

1 2 3 4

5270 4250 3372 2526

69 7 3 1

(1.2) (0.16) (0.09) (0.04)

489,784 455,628 433,494 412,117

120 131 139 125

(0.02) (0.03) (0.03) (0.03)

14.48 1.83 1.01 0.46

Adjusted for age in years, sex, race, smoking status, alcohol status, and history of gallstones.

0.25 0.29 0.32 0.31

95 % CI

P value

66.01 47.24–92.23