Advertisement

Signs and symptoms of pancreatic cancer: a population-based case-control study in the San Francisco Bay area

  • Elizabeth A. Holly
    Correspondence
    Address requests for reprints to: Elizabeth A. Holly, Ph.D., M.P.H., Professor, Epidemiology and Biostatistics, School of Medicine, University of California, San Francisco, 3333 California Street, Suite 280, San Francisco, California 94118-1944 USA. Fax: (415) 563-4602
    Affiliations
    Department of Epidemiology and Biostatistics, University of California San Francisco, San Francisco, California, USA

    Department of Health Research and Policy, Stanford University School of Medicine, Stanford, California, USA
    Search for articles by this author
  • Indranushi Chaliha
    Affiliations
    Department of Epidemiology and Biostatistics, University of California San Francisco, San Francisco, California, USA
    Search for articles by this author
  • Paige M. Bracci
    Affiliations
    Department of Epidemiology and Biostatistics, University of California San Francisco, San Francisco, California, USA
    Search for articles by this author
  • Manjushree Gautam
    Affiliations
    Department of Epidemiology and Biostatistics, University of California San Francisco, San Francisco, California, USA
    Search for articles by this author

      Abstract

      Math Eq Pancreatic cancer usually does not cause definitive symptoms until survival is severely compromised. Prevention and early detection are urgently needed. Our aim was to collect and analyze data in a population-based study on signs and symptoms of disease reported by patients with pancreatic cancer and control participants to contribute to earlier detection and better prognosis. Math Eq A supplemental symptoms questionnaire was administered to 120 consecutive patients with pancreatic cancer who were part of a larger population-based case-control study conducted in the San Francisco Bay Area between 1994 and 2001. One hundred eighty age- and sex-matched population-based control participants also were queried about the same symptoms reported by at least 5% of patients with pancreatic cancer. Math Eq Most signs and symptoms occurred within 3 years before diagnosis with pancreatic cancer (cases) and interview (controls). Many signs and symptoms were more likely to have been reported by patients compared with control participants and included appetite loss (odds ratio [OR], 41; 95% confidence interval [CI], 14–120), pale stools (OR, 31; 95% CI, 7.3–134), abdominal pain (OR, 30; 95% CI, 9.1–101), jaundice (OR, 20; 95% CI, 8.0–49), unusual bloating (OR, 20; 95% CI, 5.9–67), unusual belching (OR, 17; 95% CI, 3.9–75), weight loss (OR, 12; 95% CI, 5.2–28), dark urine (OR, 10; 95% CI, 2.9–36), constipation (OR, 7.3; 95% CI, 2.0–26), diarrhea (OR, 5.6; 95% CI, 2.0–16), itching (OR, 5.0; 95% CI, 2.3–11), fatigue (OR, 3.8; 95% CI, 2.0–7.3), altered ability to sleep (OR, 2.9; 95% CI, 1.3–6.3), and unusual heartburn (OR, 2.3; 95% CI, 1.2–4.5). Math Eq Our results show that signs and symptoms likely to be indicators of pancreatic cancer occur substantially more often among patients with pancreatic cancer than among population-based controls. The large magnitude of the risk estimates indicate that common gastrointestinal symptoms may assist clinicians in earlier diagnosis of pancreatic cancer and perhaps affect survival.

      Abbreviations:

      CI (confidence interval), OR (odds ratio), SEER (Surveillance Epidemiology and End Results)
      Pancreatic cancer is the fourth leading cause of cancer death in the United States and the second most common cause of cancer death for all gastrointestinal-related carcinomas.
      American Cancer Society
      In 2004 in the United States, an estimated 31,860 patients will be diagnosed with pancreatic cancer and 31,270 will die of the disease.
      American Cancer Society
      In 2000, worldwide figures for pancreatic cancer were projected at 216,400 new cases and 213,500 deaths.
      • Parkin D.M.
      • Bray F.
      • Ferlay J.
      • Pisani P.
      Estimating the world cancer burden Globocan 2000.
      The 5-year survival rate of 4% for pancreatic cancer, estimated by the Surveillance Epidemiology and End Results (SEER) program, is the lowest for all cancers, and mortality is nearly 50% greater in African Americans than Caucasians.
      • Ries L.
      • Eisner M.
      • Kosary C.
      • Hankey B.
      • Miller B.
      • Clegg L.
      • Mariotto A.
      • Fay M.
      • Feuer E.
      • Edwards B.E.
      Late diagnosis resulting in low resection rates is a major reason for poor survival. In light of the grim prognosis, symptom assessment and early diagnosis should be approached aggressively.
      The etiology of pancreatic adenocarcinoma remains unknown, and clinical presentation is heterogeneous and nonspecific. During the past decade, there have been few studies published that considered pancreatic cancer symptoms in detail.
      • Krech R.L.
      • Walsh D.
      Symptoms of pancreatic cancer.
      ,
      • Gullo L.
      • Tomassetti P.
      • Migliori M.
      • Casadei R.
      • Marrano D.
      Do early symptoms of pancreatic cancer exist that can allow an earlier diagnosis?.
      ,
      • Nix G.A.
      • Schmitz P.I.
      • Wilson J.H.
      • Van Blankenstein M.
      • Groeneveld C.F.
      • Hofwijk R.
      Carcinoma of the head of the pancreas. Therapeutic implications of endoscopic retrograde cholangiopancreatography findings.
      Symptoms often are insidious and are present for many months before diagnosis. The insidious onset has frustrated efforts at early detection, and diagnosis rarely is achieved while the tumor can be treated successfully.
      • Bouvet M.
      • Binmoeller K.
      • Moossa A.
      Diagnosis of adenocarcinoma of the pancreas.
      This is largely because the symptoms are vague and may be ignored by patients and physicians, the spread is rapid, and initial diagnostic tests lack sensitivity and specificity for early detection. Our population-based clinical substudy was conducted to determine signs and symptoms commonly reported by patients with pancreatic cancer to contribute to earlier detection of this deadly malignancy.

      Patients and methods

      This signs-and-symptoms study was part of a large population-based case-control study of 532 patients with pancreatic cancer and 1701 control participants that was conducted between 1994 and 2001 in the San Francisco Bay Area.
      • Hoppin J.A.
      • Tolbert P.E.
      • Holly E.A.
      • Brock J.W.
      • Korrick S.A.
      • Altshul L.M.
      • Zhang R.H.
      • Bracci P.M.
      • Burse V.W.
      • Needham L.L.
      Pancreatic cancer and serum organochlorine levels.
      ,
      • Hoppin J.A.
      • Tolbert P.E.
      • Taylor J.A.
      • Schroeder J.C.
      • Holly E.A.
      Potential for selection bias with tumor tissue retrieval in molecular epidemiology studies.
      ,
      • Slebos R.J.
      • Hoppin J.A.
      • Tolbert P.E.
      • Holly E.A.
      • Brock J.W.
      • Zhang R.H.
      • Bracci P.M.
      • Foley J.
      • Stockton P.
      • McGregor L.M.
      • Flake G.P.
      • Taylor J.A.
      K-ras and p53 in pancreatic cancer association with medical history, histopathology, and environmental exposures in a population-based study.
      ,
      • Duell E.J.
      • Holly E.A.
      • Bracci P.M.
      • Liu M.
      • Wiencke J.K.
      • Kelsey K.T.
      A population-based, case-control study of polymorphisms in carcinogen-metabolizing genes, smoking, and pancreatic adenocarcinoma risk.
      ,
      • Duell E.J.
      • Holly E.A.
      • Bracci P.M.
      • Wiencke J.K.
      • Kelsey K.T.
      A population-based study of the Arg399Gln polymorphism in X-ray repair cross-complementing group 1 (XRCC1) and risk of pancreatic adenocarcinoma.
      ,
      • Holly E.A.
      • Eberle C.A.
      • Bracci P.M.
      Prior history of allergies and pancreatic cancer in the San Francisco Bay area.
      Eligibility and methods of recruitment have been described in detail elsewhere.
      • Holly E.A.
      • Eberle C.A.
      • Bracci P.M.
      Prior history of allergies and pancreatic cancer in the San Francisco Bay area.
      Briefly, patients with pancreatic cancer aged 21–85 years were identified by the Northern California Cancer Center’s rapid case ascertainment within 1 month of diagnosis in hospitals in 6 Bay Area counties. To confirm diagnoses, SEER abstracts were obtained from the Northern California Cancer Center that identified all pancreatic cancer cases in the Bay Area ∼18–24 months after diagnosis. Control participants were identified by using random digit dial and Health Care Financing Administration files to supplement the recruitment of those ≥65 years. All control participants were frequency matched to patients with pancreatic cancer by sex and age within 5 years.
      In addition to interviews administered in person as part of the main study, a supplemental questionnaire was designed to collect clinical data pertaining to the 5 years before diagnosis or interview. It included signs and symptoms of pancreatic cancer previously reported in the literature and the diagnostic tests and procedures conducted to evaluate the cause of the reported symptoms. A series of open-ended questions also was included to probe for additional symptoms and medical experiences. No proxy interviews were conducted. One hundred twenty consecutive population-based patients with pancreatic cancer were administered the supplemental questionnaire during in-person interviews. To exclude symptoms unlikely to be related to pancreatic cancer and reduce interview time, the supplemental clinical questionnaire administered to the control group included signs and symptoms reported by at least 5% of the patients with pancreatic cancer. The minimum duration required to consider a symptom present in the control population was based on symptom duration reported by the patients with pancreatic cancer. The metric used to set the symptom-specific minimum duration for controls was one half the median duration reported by patients for each symptom (except fatigue and altered sleep). The clinical questionnaire was administered as a telephone interview to 180 age- and sex-matched population-based control participants who were interviewed in person for the main study.

       Statistical analysis

      Data were analyzed using SAS software (version 8; SAS Institute, Cary, NC). Descriptive statistics were computed for each symptom, including symptom duration (computed in weeks) and participant’s age at the time the symptom occurred. Symptom duration cutoff points for inclusion in the analyses among control participants were: dark urine, ≥1 week; pale-colored stool, ≥2 weeks; itching, ≥2 weeks; jaundice, ≥1 week; heartburn, ≥4 weeks; unusual bloating, ≥4 weeks; belching, ≥5 weeks; nausea or vomiting, ≥5 weeks; diarrhea, ≥8 weeks; constipation, ≥6 weeks; fatigue, ≥8 weeks; appetite loss, ≥4 weeks; altered sleep pattern, ≥5 weeks; unintended weight loss, ≥6 weeks; and abdominal pain, ≥6 weeks. Symptoms reported for less than the aforementioned defined minimum duration were not used in analyses for either cases or controls. Tumor extent and treatment history obtained from SEER abstracts were available for 116 of 120 patients. Tumor extent was defined as “confined to the pancreas,” “regional spread,” “distant metastasis,” and “unknown.” Treatment history that included specific surgical procedures, such as the Whipple procedure, localized pancreatic resection, and pancreatectomy, was used to identify resectable tumors.
      χ2 analysis was used to evaluate associations among extent of tumor, surgical resection, and self-reported signs and symptoms. Data pertaining to diagnosis with diabetes that lasted for ≥1 year and age at diagnosis of diabetes were collected during in-person interviews. Based on these questionnaire data, recent diagnosis with diabetes was defined as diagnosis with diabetes ≤3 years before pancreatic cancer diagnosis (or interview for controls). Unconditional logistic regression was used to obtain age- and sex-adjusted odds ratios (ORs) and associated 95% confidence intervals (CIs) as estimates of the association between symptoms and pancreatic cancer. Multivariable logistic regression analysis was used to identify self-reported symptoms that discriminated patients with pancreatic cancer from control participants. All symptoms were included in an age-and sex-adjusted logistic model and backward elimination that selected out the least statistically significant factor. These data were used to produce a final age- and sex-adjusted model. Symptoms included in the final model had P < 0.10. In addition, multiple-correspondence analysis was used to explore relationships among symptoms. Graphical presentations of the multiple correspondence analysis results were assessed to determine closely related symptoms. Results from the multiple-correspondence analysis supplemented results from the final logistic model obtained by backward elimination methods and were used to determine a set of signs and symptoms to use in analyses to fit the best logistic models. The best logistic models for 2 symptoms, 3 symptoms, etc. were based on the score χ2 statistic from each model. The effect of each additional symptom was evaluated by comparing the computed difference in the score χ2 statistics from the model with x factors and the model with x + 1 factors with the critical values for χ2 with 1 df. Results from statistical tests were considered significant for a 2-sided P ≤ 0.05.

      Results

      There were slightly more men than women with pancreatic cancer (1.3:1) in our symptoms study, and this was consistent with the ratio of men to women in the main study. Controls were somewhat more educated than cases, whereas marital status and race or ethnicity were similar between groups (Table 1). Symptoms were categorized under 4 broad subgroups of bile duct obstruction, upper gastrointestinal, lower gastrointestinal, and other constitutional symptoms, and percentages by case and control status are shown in Figure 1. The 3 most commonly reported symptoms by patients were loss of appetite, jaundice, and abdominal pain, whereas those most commonly reported by control participants were unusual heartburn, unusual fatigue, and itching, all substantially less common than among cases (Table 2).
      Table 1Characteristics of 120 Population-Based Patients With Pancreatic Cancer and 180 Randomly Selected Age- and Sex-Matched Control Participants in the San Francisco Bay Area
      Patient characteristicsPatients (N = 120)Controls (N = 180)
      Sex
      Men68 (57)94 (52)
      Women52 (43)86 (48)
      Education (yr)
      ≤1255 (46)43 (24)
      13–1641 (34)86 (48)
      >1624 (20)51 (28)
      Marital status
      Single, never married7 (6)11 (6)
      Married, living as married80 (67)115 (64)
      Widowed17 (14)23 (13)
      Divorced or separated16 (13)31 (17)
      Race
      Caucasian103 (86)159 (88)
      African American6 (5)9 (5)
      Asian9 (8)10 (6)
      Other2 (2)2 (1)
      Hispanic ethnicity4 (3)19 (11)
      Age (yr)
      < 5013 (11)17 (9)
      50–5924 (20)32 (18)
      60–6934 (28)46 (26)
      70–7938 (32)68 (38)
      80+11 (9)17 (9)
      NOTE. Values expressed as number (percent).
      Figure thumbnail GR1
      Figure 1Bar graph showing the percentage of symptoms self-reported by 120 cases and 180 age- and sex-matched controls in a population-based pancreatic cancer study in the San Francisco Bay Area. Symptoms were categorized under 4 broad subgroups of bile duct obstruction, upper gastrointestinal, lower gastrointestinal, and other constitutional symptoms, and percentages are presented by case and control status. Bile duct obstruction symptoms included jaundice, pale stools, dark urine, and itching. Upper gastrointestinal symptoms included unusual bloating, unusual belching, unusual heartburn, nausea, and vomiting. Lower gastrointestinal symptoms included constipation and diarrhea. Other constitutional symptoms included appetite loss, abdominal pain, weight loss, fatigue, and altered ability to sleep. ▩, Cases; □, controls.
      Table 2Frequency of Signs and Symptoms Reported by 120 Population-Based Patients With Pancreatic Cancer and 180 Randomly Selected Age- and Sex-Matched Control Participants in the San Francisco Bay Area
      Reported symptomsPatientsControlsOdds ratio
      Age and sex adjusted.
      95% confidence interval
      Age and sex adjusted.
      No.%Median age (yr)No.%Median age (yr)
      Bile duct obstruction
      Jaundice4941606363208.0–49
      Pale stools3126652167317.3–134
      Itching272267106625.02.3–11
      Dark urine1815683263102.9–36
      Upper gastrointestinal
      Unusual bloating3025663272205.9–67
      Unusual belching1916702166173.9–75
      Unusual heartburn2622671911632.31.2–4.5
      Lower gastrointestinal
      Diarrhea16135753745.62.0–16
      Constipation13117132547.32.0–26
      Other constitutional symptoms
      Appetite loss54456842774114–120
      Abdominal pain4033663266309.1–101
      Weight loss3932667472125.2–28
      Fatigue322762169713.82.0–7.3
      Altered ability to sleep191662116642.91.3–6.3
      NOTE. Odds ratios and 95% confidence intervals: ≥10 are presented as whole numbers.
      a Age and sex adjusted.

       Bile duct obstruction

      Jaundice was more likely to have been reported by case participants than controls, with 41% of patients compared with 3% of controls reporting this condition within the 5 years before diagnosis or interview (OR, 20; P < 0.0001; Table 2). Other symptoms common to bile duct obstruction also were reported by significantly greater proportions of patients than control participants and included pale-colored stools (26% of patients, 1% of controls; OR, 31; P < 0.0001), itching (22% of patients, 6% of controls; OR, 5; P < 0.0001), and dark urine (15% of patients, 2% of controls; OR, 10; P = 0.0003). Among controls, itching was reported more often by older participants as related to allergies or dry skin.

       Upper gastrointestinal symptoms

      Unusual and sustained bloating was the upper gastrointestinal symptom most commonly reported by patients (25% of patients, 2% of controls; OR, 20; P < 0.0001; Table 2), whereas heartburn was the symptom commonly reported by control participants. Unusual belching was reported by a significantly greater proportion of patients than controls (16% of patients, 1% of controls; OR, 17; P = 0.0002). Twice the proportion of patients reported heartburn (22% of patients, 11% of controls; OR, 2.3; P = 0.01). Nausea or vomiting (≥5 wk) was reported by 12% of patients, but none of the control participants (data not shown in table).

       Lower gastrointestinal symptoms

      Diarrhea (13% of patients, 3% controls; OR, 5.6; P = 0.001) and constipation (11% of cases, 2% controls; OR, 7.3; P = 0.002) were reported by significantly greater proportions of patients with pancreatic cancer than control participants (Table 2).

       Other constitutional symptoms

      Among the other constitutional symptoms, loss of appetite for ≥4 weeks was reported by 45% of patients and 2% of control participants (OR, 41; P < 0.0001; Table 2). Abdominal pain (OR, 30; P < 0.0001) and unintended weight loss (OR, 12; P < 0.0001) each were reported by one third of patients, who were substantially more likely to have reported a history of these symptoms than were their control counterparts (Table 2). Although fatigue (27% of patients, 9% of controls; OR, 3.8; P < 0.0001) and altered ability to sleep (16% of patients, 6% of controls; OR, 2.9; P = 0.008) were reported by nearly thrice the proportion of patients as controls, they were the other constitutional symptoms most often reported by control participants.

       Extent of tumor and symptoms

      Results from analyses of the association between tumor extent and symptoms are listed in Table 3. Tumor extent was significantly associated with symptoms of bile duct obstruction (jaundice, P = 0.02; dark urine, P = 0.004) and abdominal pain (P = 0.02). Surgical resection was reported for 56% of patients with tumors confined to the pancreas and 54% of patients with tumors with regional spread (data not shown in tables). Surgical resection was associated with symptoms of bile duct obstruction (jaundice, P = 0.0002; pale stools, P = 0.09; itching, P = 0.07; dark urine, P < 0.0001), abdominal pain (P = 0.06), and unusual bloating (P = 0.04; data not shown in tables). In hierarchical modeling that used factors identified in the initial analyses of symptoms taken singly, the model that best differentiated resected tumors from nonresected tumors based on the model score chi-square statistic included tumor extent, jaundice, and dark urine. In 18 of 120 patients with pancreatic cancer (15%), diabetes mellitus was diagnosed; in 4 of these patients (3%), diabetes was diagnosed within the 3 years before their pancreatic cancer diagnosis. Of the 4 patients with recent diabetes, 1 patient had a tumor confined to the pancreas, 2 patients had tumors with regional spread, and 1 patient had a tumor with distant metastasis. None of these 4 patients had undergone surgical resection. In 15 of 180 control participants (8%), diabetes mellitus was diagnosed; in 5 of these patients (3%), diabetes was diagnosed within 3 years of the study interview. Recently diagnosed diabetes was not associated with pancreatic cancer diagnosis (3% of patients, 3% of controls). Graphical results from exploratory multiple correspondence analysis among patients were examined to assess the relationship between signs and symptoms and extent of the tumor. The graphical presentation showed the greatest delineation between resectable and nonresectable tumors and between confined/regional tumor extent and distant/unknown tumor extent. Consistent with contingency table analyses, surgical resection was closely related to tumors defined as confined or regional. Graphical results also indicated that jaundice and dark urine were related to confined and regional tumors, whereas the closely related symptoms of abdominal pain, appetite loss, weight loss, and unusual bloating were related to distant and unknown tumors.
      Table 3Extent of Tumor in Relation to Surgical Resection and Symptoms of Pancreatic Cancer in Population-Based Patients With Pancreatic Cancer in the San Francisco Bay Area
      FactorTumor extent
      Percentage computed as the number of patients within the specific tumor extent category over the total number of patients with the factor.
      Confined to pancreas (n = 16)Regional spread (n = 46)Distant metastasis (n = 35)Unknown (n = 19)
      Surgical resection
      Whipple procedure, localized pancreatic resection, pancreatectomy.
      9 (24)25 (66)2 (5)2 (5)
      Jaundice10 (21)23 (48)8 (17)7 (15)
      Dark urine1 (6)14 (78)1 (6)2 (11)
      Abdominal pain6 (16)10 (26)18 (47)4 (11)
      Appetite loss7 (44)22 (48)15 (43)8 (42)
      Weight loss7 (44)11 (24)13 (37)8 (42)
      Unusual bloating5 (31)9 (20)10 (29)5 (26)
      NOTE. Values expressed as number (percent).
      a Percentage computed as the number of patients within the specific tumor extent category over the total number of patients with the factor.
      b Whipple procedure, localized pancreatic resection, pancreatectomy.

       Models

      Symptoms that remained in the age- and sex-adjusted logistic model that best differentiated patients from control participants were abdominal pain (P < 0.0001), appetite loss (P < 0.0001), jaundice (P < 0.0001), pale-colored stools (P = 0.002), unusual belching (P = 0.005), weight loss (P = 0.06), and unusual bloating (P = 0.09). Graphical results from multiple correspondence analyses of symptoms reported by patients and control participants identified that jaundice and pale-colored stools, appetite loss and weight loss, and unusual belching and unusual bloating were closely related symptoms. Based on these results, jaundice, abdominal pain, appetite loss, unusual belching, and heartburn were included in logistic analyses to determine the best models with an increasing number of symptoms selected using score chi-square statistics (Table 4). Abdominal pain, appetite loss, and jaundice were the 3 symptoms that best differentiated patients from control participants. Addition of the symptom “unusual belching” increased the model score χ2 significantly (P = 0.007), indicating that this symptom also is important in differentiating patients from controls.
      Table 4Age- and Sex-Adjusted Logistic Regression Models Selected by the Magnitude of the Model Score χ2 Statistic to Identify Self-Reported Symptoms That Discriminate Patients With Pancreatic Cancer From Control Participants in the San Francisco Bay Area
      No. of factors in the model
      In addition to sex and age group.
      Symptoms in the model
      In addition to sex and age group.
      Score χ2
      The χ2 P for each additional factor, computed as described, is <0.01.
      2Abdominal pain, jaundice117.7
      3Abdominal pain, appetite loss, jaundice141.5
      4Abdominal pain, appetite loss, jaundice, unusual belching148.8
      5Abdominal pain, appetite loss, jaundice, unusual belching, heartburn151.4
      NOTE. Score χ2 statistics are computed for each model. The significance of each additional factor in the models is evaluated by computing the difference between the model score χ2 of the model with X factors to the model with X + 1 factors and comparing the difference with the critical value for χ2 with 1 of p = 0.05 (χ2 1df, α = 0.05 = 3.841). SAS software, version 8.0, was used. N = 120 patients with pancreatic cancer. N = 180 control participants.
      a In addition to sex and age group.
      b The χ2 P for each additional factor, computed as described, is <0.01.

      Discussion

      The purpose of this study was to evaluate the signs and symptoms commonly reported by patients with pancreatic cancer in a population-based setting, with a goal toward promoting earlier diagnosis. Unlike most symptom studies, our participants were part of a larger population-based study that included a consecutive group of patients with pancreatic cancer and population-based control participants to better compare and highlight symptoms likely to define pancreatic cancer in the general population. Structured and open-ended questions were used to obtain from patients and controls a more complete and detailed history of symptoms within the 5 years before their diagnosis of pancreatic cancer (or interview for controls). We focused on signs and symptoms that were present for a defined minimum duration to exclude those related to short-term illness and rule out other diseases. Also, clinical data were collected directly from participants to eliminate biases and inaccuracies associated with data collected from proxy respondents.
      A weakness of this study is the potential for selection bias associated with the loss of patients who died before we could interview them. However, our use of rapid case ascertainment, designed to identify patients within a month of diagnosis, served to diminish this bias. Comparison of basic demographic data available from the tumor registry abstracts showed that interviewed patients were more likely to be men, more likely to be white, and slightly younger, and their tumor characteristics more often were unknown compared with noninterviewed patients. As in all case-control studies, data also were subject to potential recall and reporting bias. Patients may have reported and recalled more signs and symptoms of illness regardless of their duration than did control participants. To limit this reporting bias, we asked controls about the unusual and prolonged occurrence of specific symptoms and used a symptom-specific minimum duration to analyze the data on the presence or absence of symptoms in all study participants. This would tend to diminish the overall reporting bias and also allowed us to exclude symptoms likely to be related to transient food-borne gastrointestinal or other intermittent problems. Minimum duration restrictions also were useful to reduce differential recall of symptoms between cases and controls because illness symptoms of longer duration would be more memorable to both groups.
      Certain signs and symptoms of pancreatic cancer may be indicative of earlier disease stage and better prognosis.
      • Kalser M.H.
      • Barkin J.
      • MacIntyre J.M.
      Pancreatic cancer. Assessment of prognosis by clinical presentation.
      ,
      • Mannell A.
      • van Heerden J.A.
      • Weiland L.H.
      • Ilstrup D.M.
      Factors influencing survival after resection for ductal adenocarcinoma of the pancreas.
      The most promising of these is jaundice, although this has not been reported consistently. In our study, jaundice and other bile duct obstruction symptoms, including pale-colored stools, pruritus, and dark urine, were common in pancreatic cases and occurred more frequently among patients than control participants. Our results were consistent with those from several studies that reported jaundice in a large proportion of patients with pancreatic cancer.
      • Kalser M.H.
      • Barkin J.
      • MacIntyre J.M.
      Pancreatic cancer. Assessment of prognosis by clinical presentation.
      ,
      • Alvarez C.
      • Livingston E.H.
      • Ashley S.W.
      • Schwarz M.
      • Reber H.A.
      Cost-benefit analysis of the work-up for pancreatic cancer.
      ,
      • Manabe T.
      • Miyashita T.
      • Ohshio G.
      • Nonaka A.
      • Suzuki T.
      • Endo K.
      • Takahashi M.
      • Tobe T.
      Small carcinoma of the pancreas. Clinical and pathologic evaluation of 17 patients.
      Similar to our results, a retrospective review that included 393 patients with histologically verified adenocarcinoma of the pancreas reported that jaundice occurred in nearly 50% of study patients and varied by lesion location.
      • Kalser M.H.
      • Barkin J.
      • MacIntyre J.M.
      Pancreatic cancer. Assessment of prognosis by clinical presentation.
      The association between lesion location and jaundice has been documented, with jaundice reported as the solitary presenting symptom in 7%–30% of all patients with pancreatic cancer
      • Moossa A.R.
      Pancreatic cancer approach to diagnosis, selection for surgery and choice of operation.
      and found in 65%–90% of patients with carcinomas of the head of the pancreas and generally in <20% of patients with body or tail lesions.
      • Moossa A.R.
      Pancreatic cancer approach to diagnosis, selection for surgery and choice of operation.
      ,
      • Gullick H.
      Carcinoma of the pancreas a review and critical study of 100 cases.
      Jaundice as an earlier sign of disease is supported by data that showed these patients were more likely to have had resectable lesions compared with nonjaundiced patients.
      • Kalser M.H.
      • Barkin J.
      • MacIntyre J.M.
      Pancreatic cancer. Assessment of prognosis by clinical presentation.
      ,
      • Moossa A.R.
      Pancreatic cancer approach to diagnosis, selection for surgery and choice of operation.
      Conversely, another study reported that 123 patients with pancreatic cancer had jaundice (89%), pale stools (75%), dark urine (78%), and pruritus (41%) as late manifestations of disease.
      • Nix G.A.
      • Schmitz P.I.
      • Wilson J.H.
      • Van Blankenstein M.
      • Groeneveld C.F.
      • Hofwijk R.
      Carcinoma of the head of the pancreas. Therapeutic implications of endoscopic retrograde cholangiopancreatography findings.
      Because most patients with lesions located in the head of the pancreas, proximal to the bile duct and ampulla, present with jaundice,
      • Nix G.A.
      • Schmitz P.I.
      • Wilson J.H.
      • Van Blankenstein M.
      • Groeneveld C.F.
      • Hofwijk R.
      Carcinoma of the head of the pancreas. Therapeutic implications of endoscopic retrograde cholangiopancreatography findings.
      ,
      • Manabe T.
      • Miyashita T.
      • Ohshio G.
      • Nonaka A.
      • Suzuki T.
      • Endo K.
      • Takahashi M.
      • Tobe T.
      Small carcinoma of the pancreas. Clinical and pathologic evaluation of 17 patients.
      these lesions may be more likely to physically obstruct the bile duct, resulting in the presentation of symptoms earlier than in lesions in the body or tail, more distant from the pancreatic duct. The occurrence of symptoms indicating pancreatic disturbances before symptoms of pain and jaundice were investigated in a study that assumed pain and jaundice to occur late in patients with pancreatic cancer.
      • Gullo L.
      • Tomassetti P.
      • Migliori M.
      • Casadei R.
      • Marrano D.
      Do early symptoms of pancreatic cancer exist that can allow an earlier diagnosis?.
      Results showed that among the nearly 50% of patients who experienced some disturbance before pain or jaundice, most of their other symptoms presented within 6 months of pain or jaundice.
      • Gullo L.
      • Tomassetti P.
      • Migliori M.
      • Casadei R.
      • Marrano D.
      Do early symptoms of pancreatic cancer exist that can allow an earlier diagnosis?.
      Additional characterization of jaundice also has been useful in pancreatic cancer prognosis. Painless jaundice has been ascribed a relatively more favorable prognosis because it may be more likely to correspond to a resectable tumor.
      • Kalser M.H.
      • Barkin J.
      • MacIntyre J.M.
      Pancreatic cancer. Assessment of prognosis by clinical presentation.
      Despite the inconsistency in reports of whether jaundice is an early or late symptom of pancreatic cancer, the combined results indicate that patients with bile duct obstructive symptoms should be thoroughly investigated to rule out pancreatic cancer.
      In our study, all upper gastrointestinal symptoms were reported more frequently by patients than control participants. Our results are consistent with those of another study that found patients with pancreatic cancer commonly reported bloating (23%), belching (23%), dyspepsia (36%), and vomiting (31%).
      • Krech R.L.
      • Walsh D.
      Symptoms of pancreatic cancer.
      An early gastrointestinal symptom of pancreatic cancer arguably is generalized dyspepsia that may be attributed to delayed gastric emptying.
      • Barkin J.S.
      • Goldstein J.A.
      Diagnostic and therapeutic approach to pancreatic cancer.
      Two studies that used scintigraphic techniques to evaluate delays in gastric emptying in patients with pancreatic cancer reported delays in up to 60% of patients.
      • Barkin J.S.
      • Goldberg R.I.
      • Sfakianakis G.N.
      • Levi J.
      Pancreatic carcinoma is associated with delayed gastric emptying.
      ,
      • Sikora S.S.
      • Mital B.R.
      • Prasad K.R.
      • Das B.K.
      • Kaushik S.P.
      Functional gastric impairment in carcinoma of the pancreas.
      Additional data showed that dyspepsia associated with hiccups, flatulence, and regurgitation may be early symptoms of pancreatic cancer, preceding more suggestive signs and symptoms of pancreatic cancer by many months.
      • Dubois A.
      Clinical features.
      Unexplained dyspepsia, among other symptoms in those >40 years, particularly among heavy cigarette smokers, should raise clinical suspicion for pancreatic cancer.
      • Bouvet M.
      • Binmoeller K.
      • Moossa A.
      Diagnosis of adenocarcinoma of the pancreas.
      Patients with unexplained dyspepsia and accompanying severe midabdominal pain or radiating pain that interferes with sleep also should be evaluated for pancreatic disease.
      The occurrence of other upper gastrointestinal symptoms has not received as much focus as dyspepsia, possibly because of their ubiquitous presentation associated with benign conditions. The delayed gastric emptying that often accompanies pancreatic cancer may account for the common reporting of nausea and early satiety.
      • Barkin J.S.
      • Goldberg R.I.
      • Sfakianakis G.N.
      • Levi J.
      Pancreatic carcinoma is associated with delayed gastric emptying.
      Although clinical data indicate that these symptoms often occur at late stages of disease, patients are likely to present with accompanying symptoms that should be considered in the symptom profile for pancreatic cancer. Vomiting may be seen in relatively late stages when the tumor has invaded or compressed the second portion of the duodenum, creating a partial or complete obstruction.
      • Barkin J.S.
      • Goldstein J.A.
      Diagnostic and therapeutic approach to pancreatic cancer.
      In our study, nausea and vomiting were reported by patients, but not by control participants, and always were accompanied by a history of at least 2 other symptoms of interest. In addition, consistent with evidence that nausea and bloating occur in late disease, >85% of patients in our study who reported these symptoms reported their first occurrence within ∼1 month of their diagnosis of pancreatic cancer.
      Heartburn in association with pancreatic cancer is supported by evidence from animal studies showing duodenogastric reflux enhances growth and carcinogenesis in the rat pancreas.
      • Watanapa P.
      • Flaks B.
      • Oztas H.
      • Deprez P.H.
      • Calam J.
      • Williamson R.C.
      Duodenogastric reflux enhances growth and carcinogenesis in the rat pancreas.
      ,
      • Taylor P.R.
      • Dowling R.H.
      • Palmer T.J.
      • Hanley D.C.
      • Murphy G.M.
      • Mason R.C.
      • McColl I.
      Induction of pancreatic tumours by longterm duodenogastric reflux.
      Given the results from animal studies and our analyses, onset of these symptoms should raise the suspicion of pancreatic cancer.
      Diarrhea and constipation were reported more commonly by patients with pancreatic cancer than control participants in our study. Both constipation and diarrhea have been variously cited as the predominant alteration in bowel habits in carcinoma of the pancreas.
      • Gullick H.
      Carcinoma of the pancreas a review and critical study of 100 cases.
      Constipation may only be important in case of a recent manifestation of pancreatic cancer, whereas diarrhea is rare and steatorrhea is inconstant.
      • Dubois A.
      Clinical features.
      However, it also has been reported that fat malabsorption occurs only in patients with severe pancreatic insufficiency, defined as <10% output of the pancreatic enzymes lipase and trypsin.
      • DiMagno E.P.
      • Malagelada J.R.
      • Go V.L.
      • Moertel C.G.
      Fate of orally ingested enzymes in pancreatic insufficiency. Comparison of two dosage schedules.
      In our study, pain was categorized as abdominal pain at least 6 weeks in duration. Pain also was one of the most frequent symptoms reported by patients in other pancreatic cancer studies. Abdominal pain was present regardless of tumor location,
      • Braganza J.M.
      • Howat H.T.
      Cancer of the pancreas.
      although it was reported by more patients with cancer in the body and tail of the pancreas (90%) compared with those with cancer in the head of the pancreas (70%).
      • Furukawa H.
      • Okada S.
      • Saisho H.
      • Ariyama J.
      • Karasawa E.
      • Nakaizumi A.
      • Nakazawa S.
      • Murakami K.
      • Kakizoe T.
      Clinicopathologic features of small pancreatic adenocarcinoma. A collective study.
      Other studies also reported that abdominal pain is one of the most common presenting symptoms of patients with pancreatic cancer (32%–80%),
      • Mannell A.
      • van Heerden J.A.
      • Weiland L.H.
      • Ilstrup D.M.
      Factors influencing survival after resection for ductal adenocarcinoma of the pancreas.
      ,
      • Alvarez C.
      • Livingston E.H.
      • Ashley S.W.
      • Schwarz M.
      • Reber H.A.
      Cost-benefit analysis of the work-up for pancreatic cancer.
      ,
      • Manabe T.
      • Miyashita T.
      • Ohshio G.
      • Nonaka A.
      • Suzuki T.
      • Endo K.
      • Takahashi M.
      • Tobe T.
      Small carcinoma of the pancreas. Clinical and pathologic evaluation of 17 patients.
      ,
      • Furukawa H.
      • Okada S.
      • Saisho H.
      • Ariyama J.
      • Karasawa E.
      • Nakaizumi A.
      • Nakazawa S.
      • Murakami K.
      • Kakizoe T.
      Clinicopathologic features of small pancreatic adenocarcinoma. A collective study.
      even when the tumor was small (<2 cm).
      • Manabe T.
      • Miyashita T.
      • Ohshio G.
      • Nonaka A.
      • Suzuki T.
      • Endo K.
      • Takahashi M.
      • Tobe T.
      Small carcinoma of the pancreas. Clinical and pathologic evaluation of 17 patients.
      ,
      • Furukawa H.
      • Okada S.
      • Saisho H.
      • Ariyama J.
      • Karasawa E.
      • Nakaizumi A.
      • Nakazawa S.
      • Murakami K.
      • Kakizoe T.
      Clinicopathologic features of small pancreatic adenocarcinoma. A collective study.
      Pain in patients with pancreatic cancer can have several contributing sources. Perineural spread or invasion by the tumor, capsular stretching, and pancreatic cancer ductal obstruction have been proposed as mechanisms.
      • Moossa A.R.
      • Gamagami R.A.
      Diagnosis and staging of pancreatic neoplasms.
      The majority of patients with exocrine pancreatic cancer have reported pain as an initial symptom, and practically all experienced substantial pain before they died of the disease.
      • Moossa A.R.
      • Gamagami R.A.
      Diagnosis and staging of pancreatic neoplasms.
      However, some researchers suggested that pain may be a late symptom, and the presence of pain denotes advanced carcinomas that may be associated with reduced survival time.
      • Kelsen D.P.
      • Portenoy R.
      • Thaler H.
      • Tao Y.
      • Brennan M.
      Pain as a predictor of outcome in patients with operable pancreatic carcinoma.
      Conversely, some clinicians reported that pain is not necessarily a sign of advanced unresectable disease,
      • Grahm A.L.
      • Andren-Sandberg A.
      Prospective evaluation of pain in exocrine pancreatic cancer.
      and the presence of pain should encourage additional staging.
      Among all constitutional symptoms analyzed in our study, loss of appetite was the most common symptom to occur in patients with pancreatic cancer and was rare in the population controls. One study of weight loss in patients with pancreatic cancer reported an association between self-reported loss of appetite and the presence of cachexia,
      • Brown D.R.
      • Berkowitz D.E.
      • Breslow M.J.
      Weight loss is not associated with hyperleptinemia in humans with pancreatic cancer.
      whereas more general studies of pancreatic cancer symptoms have reported high frequencies of anorexia and early satiety.
      • Krech R.L.
      • Walsh D.
      Symptoms of pancreatic cancer.
      ,
      • Gullo L.
      • Tomassetti P.
      • Migliori M.
      • Casadei R.
      • Marrano D.
      Do early symptoms of pancreatic cancer exist that can allow an earlier diagnosis?.
      Weight loss, fatigue, or altered ability to sleep reported by patients in our study also were reported in other investigations.
      • Krech R.L.
      • Walsh D.
      Symptoms of pancreatic cancer.
      ,
      • Mannell A.
      • van Heerden J.A.
      • Weiland L.H.
      • Ilstrup D.M.
      Factors influencing survival after resection for ductal adenocarcinoma of the pancreas.
      These are among several symptoms consistent with locally advanced disease. Weight loss has been reported to occur several months before diagnosis
      • Mannell A.
      • van Heerden J.A.
      • Weiland L.H.
      • Ilstrup D.M.
      Factors influencing survival after resection for ductal adenocarcinoma of the pancreas.
      and also has correlated with higher stage of disease.
      • Bakkevold K.E.
      • Arnesjo B.
      • Kambestad B.
      Carcinoma of the pancreas and papilla of Vater—assessment of resectability and factors influencing resectability in stage I carcinomas. A prospective multicentre trial in 472 patients.
      Although cancer often has been associated with weight loss and/or cachexia, weight loss in patients with pancreatic and biliary cancers also may be attributed to gastrointestinal disruption that results in malabsorption.
      In conclusion, pancreatic cancer typically is diagnosed at a relatively advanced stage. In the past, few studies have considered symptoms that commonly are experienced by patients with pancreatic cancer. Most patients diagnosed with pancreatic cancer are symptomatic, but early pancreatic cancer is notoriously difficult to detect, partly because the pancreas is a relatively inaccessible organ for physical examination. Some of the common presenting symptoms, pain, dyspepsia, and weight loss, have an insidious onset and easily are mistaken for other illnesses or functional disorders.
      • Hermann R.E.
      • Cooperman A.M.
      Current concepts in cancer cancer of the pancreas.
      Abdominal pain and jaundice as the primary symptoms of pancreatic cancer often appear late, when the tumor is advanced.
      • Nix G.A.
      • Schmitz P.I.
      • Wilson J.H.
      • Van Blankenstein M.
      • Groeneveld C.F.
      • Hofwijk R.
      Carcinoma of the head of the pancreas. Therapeutic implications of endoscopic retrograde cholangiopancreatography findings.
      ,
      • Warshaw A.L.
      • Fernandez-Del Castillo C.
      Pancreatic carcinoma.
      Therefore, pancreatic cancer remains a diagnostic challenge to clinicians and researchers. At present, a prompt diagnosis of pancreatic cancer comes from careful clinical history and appropriate tests. The lack of common knowledge about specific symptoms and signs that could induce the general population to visit a physician while the tumor still is in an early stage currently is an impediment to early diagnosis. Research is essential to illuminate facets of clinical diagnosis that will improve our understanding of the behavior of this disease. Our population-based results are consistent with those from clinical series and other studies and emphasize the urgent need for more directed research and education regarding symptoms and diagnosis for prevention and early detection of this deadly malignancy.

      Acknowledgements

      The authors thank Dr. Andrew Ko for his thoughtful suggestions.

      References

        • American Cancer Society
        Cancer facts & figures 2004. American Cancer Society, Atlanta2004
        • Parkin D.M.
        • Bray F.
        • Ferlay J.
        • Pisani P.
        Estimating the world cancer burden.
        Int J Cancer. 2001; 94: 153-156
        • Ries L.
        • Eisner M.
        • Kosary C.
        • Hankey B.
        • Miller B.
        • Clegg L.
        • Mariotto A.
        • Fay M.
        • Feuer E.
        • Edwards B.E.
        SEER cancer statistics review, 1975–2000. National Cancer Institute, Bethesda, MD2003
        • Krech R.L.
        • Walsh D.
        Symptoms of pancreatic cancer.
        J Pain Symptom Manage. 1991; 6: 360-367
        • Gullo L.
        • Tomassetti P.
        • Migliori M.
        • Casadei R.
        • Marrano D.
        Do early symptoms of pancreatic cancer exist that can allow an earlier diagnosis?.
        Pancreas. 2001; 22: 210-213
        • Nix G.A.
        • Schmitz P.I.
        • Wilson J.H.
        • Van Blankenstein M.
        • Groeneveld C.F.
        • Hofwijk R.
        Carcinoma of the head of the pancreas. Therapeutic implications of endoscopic retrograde cholangiopancreatography findings.
        Gastroenterology. 1984; 87: 37-43
        • Bouvet M.
        • Binmoeller K.
        • Moossa A.
        Diagnosis of adenocarcinoma of the pancreas.
        in: Cameron J. Pancreatic cancer. Decker, Hamilton, ON2001: 67-85
        • Hoppin J.A.
        • Tolbert P.E.
        • Holly E.A.
        • Brock J.W.
        • Korrick S.A.
        • Altshul L.M.
        • Zhang R.H.
        • Bracci P.M.
        • Burse V.W.
        • Needham L.L.
        Pancreatic cancer and serum organochlorine levels.
        Cancer Epidemiol Biomarkers Prev. 2000; 9: 199-205
        • Hoppin J.A.
        • Tolbert P.E.
        • Taylor J.A.
        • Schroeder J.C.
        • Holly E.A.
        Potential for selection bias with tumor tissue retrieval in molecular epidemiology studies.
        Ann Epidemiol. 2002; 12: 1-6
        • Slebos R.J.
        • Hoppin J.A.
        • Tolbert P.E.
        • Holly E.A.
        • Brock J.W.
        • Zhang R.H.
        • Bracci P.M.
        • Foley J.
        • Stockton P.
        • McGregor L.M.
        • Flake G.P.
        • Taylor J.A.
        K-ras and p53 in pancreatic cancer.
        Cancer Epidemiol Biomarkers Prev. 2000; 9: 1223-1232
        • Duell E.J.
        • Holly E.A.
        • Bracci P.M.
        • Liu M.
        • Wiencke J.K.
        • Kelsey K.T.
        A population-based, case-control study of polymorphisms in carcinogen-metabolizing genes, smoking, and pancreatic adenocarcinoma risk.
        J Natl Cancer Inst. 2002; 94: 297-306
        • Duell E.J.
        • Holly E.A.
        • Bracci P.M.
        • Wiencke J.K.
        • Kelsey K.T.
        A population-based study of the Arg399Gln polymorphism in X-ray repair cross-complementing group 1 (XRCC1) and risk of pancreatic adenocarcinoma.
        Cancer Res. 2002; 62: 4630-4636
        • Holly E.A.
        • Eberle C.A.
        • Bracci P.M.
        Prior history of allergies and pancreatic cancer in the San Francisco Bay area.
        Am J Epidemiol. 2003; 158: 432-441
        • Kalser M.H.
        • Barkin J.
        • MacIntyre J.M.
        Pancreatic cancer. Assessment of prognosis by clinical presentation.
        Cancer. 1985; 56: 397-402
        • Mannell A.
        • van Heerden J.A.
        • Weiland L.H.
        • Ilstrup D.M.
        Factors influencing survival after resection for ductal adenocarcinoma of the pancreas.
        Ann Surg. 1986; 203: 403-407
        • Alvarez C.
        • Livingston E.H.
        • Ashley S.W.
        • Schwarz M.
        • Reber H.A.
        Cost-benefit analysis of the work-up for pancreatic cancer.
        Am J Surg. 1993; 165 (discussion, 58–60.): 53-58
        • Manabe T.
        • Miyashita T.
        • Ohshio G.
        • Nonaka A.
        • Suzuki T.
        • Endo K.
        • Takahashi M.
        • Tobe T.
        Small carcinoma of the pancreas. Clinical and pathologic evaluation of 17 patients.
        Cancer. 1988; 62: 135-141
        • Moossa A.R.
        Pancreatic cancer.
        Cancer. 1982; 50: 2689-2698
        • Gullick H.
        Carcinoma of the pancreas.
        Medicine (Baltimore). 1959; 38: 47-84
        • Barkin J.S.
        • Goldstein J.A.
        Diagnostic and therapeutic approach to pancreatic cancer.
        Biomed Pharmacother. 2000; 54: 400-409
        • Barkin J.S.
        • Goldberg R.I.
        • Sfakianakis G.N.
        • Levi J.
        Pancreatic carcinoma is associated with delayed gastric emptying.
        Dig Dis Sci. 1986; 31: 265-267
        • Sikora S.S.
        • Mital B.R.
        • Prasad K.R.
        • Das B.K.
        • Kaushik S.P.
        Functional gastric impairment in carcinoma of the pancreas.
        Br J Surg. 1995; 82: 1112-1113
        • Dubois A.
        Clinical features.
        in: Baumel H. Deixonne B. Exocrine pancreatic cancer. Springer-Verlag, Berlin1986: 46-54
        • Watanapa P.
        • Flaks B.
        • Oztas H.
        • Deprez P.H.
        • Calam J.
        • Williamson R.C.
        Duodenogastric reflux enhances growth and carcinogenesis in the rat pancreas.
        Br J Surg. 1992; 79: 791-794
        • Taylor P.R.
        • Dowling R.H.
        • Palmer T.J.
        • Hanley D.C.
        • Murphy G.M.
        • Mason R.C.
        • McColl I.
        Induction of pancreatic tumours by longterm duodenogastric reflux.
        Gut. 1989; 30: 1596-1600
        • DiMagno E.P.
        • Malagelada J.R.
        • Go V.L.
        • Moertel C.G.
        Fate of orally ingested enzymes in pancreatic insufficiency. Comparison of two dosage schedules.
        N Engl J Med. 1977; 296: 1318-1322
        • Braganza J.M.
        • Howat H.T.
        Cancer of the pancreas.
        Clin Gastroenterol. 1972; 1: 219-237
        • Furukawa H.
        • Okada S.
        • Saisho H.
        • Ariyama J.
        • Karasawa E.
        • Nakaizumi A.
        • Nakazawa S.
        • Murakami K.
        • Kakizoe T.
        Clinicopathologic features of small pancreatic adenocarcinoma. A collective study.
        Cancer. 1996; 78: 986-990
        • Moossa A.R.
        • Gamagami R.A.
        Diagnosis and staging of pancreatic neoplasms.
        Surg Clin North Am. 1995; 75: 871-890
        • Kelsen D.P.
        • Portenoy R.
        • Thaler H.
        • Tao Y.
        • Brennan M.
        Pain as a predictor of outcome in patients with operable pancreatic carcinoma.
        Surgery. 1997; 122: 53-59
        • Grahm A.L.
        • Andren-Sandberg A.
        Prospective evaluation of pain in exocrine pancreatic cancer.
        Digestion. 1997; 58: 542-559
        • Brown D.R.
        • Berkowitz D.E.
        • Breslow M.J.
        Weight loss is not associated with hyperleptinemia in humans with pancreatic cancer.
        J Clin Endocrinol Metab. 2001; 86: 162-166
        • Bakkevold K.E.
        • Arnesjo B.
        • Kambestad B.
        Carcinoma of the pancreas and papilla of Vater—assessment of resectability and factors influencing resectability in stage I carcinomas. A prospective multicentre trial in 472 patients.
        Eur J Surg Oncol. 1992; 18: 494-507
        • Hermann R.E.
        • Cooperman A.M.
        Current concepts in cancer.
        N Engl J Med. 1979; 301: 482-485
        • Warshaw A.L.
        • Fernandez-Del Castillo C.
        Pancreatic carcinoma.
        N Engl J Med. 1992; 326: 455-465