Abstract
Aims:
To determine the relationship of diabetes with pancreatic cancer incidence among African American and Whites of similar socioeconomic status.
Methods:
Using the Southern Community Cohort Study, we conducted a follow-up during 2002–2015 of pancreatic cancer incidence of 73,378 mostly low-income participants aged 40–79 years; 15,913 reported diabetes at baseline. Multivariable Cox analysis controlling for sex, family history of pancreatic cancer, BMI, smoking status, alcohol consumption, education, income and other important covariates, and with age as the time scale was used.
Results:
265 incident pancreatic cancer cases were observed. Pancreatic cancer risk was increased among those with diabetes (HR 1.54, CI 1.16–2.05), with similar increases among African Americans (HR 1.51, CI 1.08–2.11) and Whites (HR 1.78, CI 1.00–3.16). No trend in risk was observed for diabetes duration among those with diabetes, with HRs of 1.39 (0.91–2.11), 2.31 (1.51–3.54) and 1.23 (0.80–1.89) for <5, 5–9 and 10+ years duration, respectively. African Americans were at increased risk of pancreatic cancer (HR=1.40, 95% CI 1.05–1.87), which persisted after adjusting for diabetes (HR 1.36, CI 1.02–1.81). The effect sizes for other pancreatic cancer risk factors with pancreatic cancer were similar by diabetes status, although a stronger association with low BMI was evident among those with diabetes.
Conclusions:
Diabetes increases pancreatic cancer risk similarly among African Americans and Whites in this Southern U.S. cohort.
Keywords: diabetes, pancreatic cancer, African Americans, racial disparities
Introduction
Pancreatic cancer is one of the most fatal malignancies, with a 5-year survival of <11%1 and is the 3rd leading cause of cancer death in the United States.1, 2 Pancreatic cancer is also one of the cancers most tightly linked to diabetes, with impaired glucose tolerance often reported in the majority of pancreatic cancer patients3–5. While diabetes sometimes occurs secondary to pancreatic cancer, i.e. as a complication of the cancer, there is evidence that diabetes also increases pancreatic cancer risk, with meta analyses suggesting a nearly doubled incidence of pancreatic cancer among type 2 diabetes patients6, 7. Shared risk factors, such as hyperinsulinemia or elevated circulating insulin levels, obesity, and hyperglycemia may contribute to the diabetes-pancreatic cancer risk association.
The highest rates of pancreatic cancer are found among African American residents in the US.8 African Americans also have elevated levels of insulin9, 10 and a higher prevalence of obesity,11 also shared risk factors for diabetes and pancreatic cancer. However, epidemiologic studies of diabetes and pancreas cancer among African American people are few. Despite the higher prevalence of diabetes among African American compared to White American residents12, little is known about the interrelationship between diabetes and pancreatic cancer among African American adults and whether this accounts for or contributes to the observed racial disparity. We thus investigated the relationship of diabetes with the incidence of pancreatic cancer in a large, low socioeconomic status population and whether this relationship varied between African American and White participants. Our aims were to 1) examine whether diabetes is associated with incident pancreatic cancer in African Americans and in White Americans, and 2) examine whether diabetes contributed to the higher incidence of pancreatic cancer in African American versus White Americans.
Methods
The Southern Community Cohort Study (SCCS) is a population-based prospective study designed to investigate causes of health disparities in the incidence and mortality of cancer and other chronic diseases. Details of the rationale, study design, and methods have been described previously 13. Briefly, between 2002 and 2009, nearly 85,000 participants aged 40 to 79 were recruited from community health centers (85%) and surrounding communities (15%) from twelve states in the southeastern United States. Community health centers (CHC) are government-funded health care facilities offering basic health care services to the medically underserved. All participants provided informed consent and all study procedures were approved by the Institutional Review Boards of Vanderbilt University Medical Center and Meharry Medical College.
After obtaining informed consent, information was collected on medical history, lifestyle, and demographic factors by in-person interview from CHC-recruited participants and by completion of the study questionnaire in the mail survey recruited participants. If a participant answered “yes” to the question “Has a doctor ever told you that you have diabetes or high blood sugar?” the participant was then asked about age at diabetes diagnosis and medications currently taken to treat diabetes. Women were specifically asked not to include gestational diabetes in their reporting. A total of 17,750 individuals reported having a physician diagnosis of diabetes (prevalent diabetes cases) at cohort entry, of whom 421 had a missing age of diabetes diagnosis and 504 had missing data on race, and an additional 1,546 who did not have full covariate data resulting in 15,913 included in the analysis. During the three routine follow-up surveys respectively conducted at averages of 4.6, 7.7 and 11 years after cohort entry, 8,656 additional participants not reporting diabetes at study enrollment subsequently reported a diagnosis of diabetes (incident diabetes cases).
Using information reported on medications used to treat diabetes, we classified those with prevalent diabetes into the following five non-overlapping categories: (1) users of insulin alone; (2) users of metformin alone; (3) users of either insulin or metformin plus one additional anti-hyperglycemic medication; (4) users of other antihyperglycemic medications, but not insulin or metformin; and (5) not on anti-hyperglycemic medication. Medication type was not obtained at the first follow-up survey, so this analysis was limited to those with prevalent diabetes.
To be included in this current study, participants had to be cancer free at study baseline. Pancreatic cancer (ICD10 code C25) cases were ascertained by linkage of the cohort with the state cancer registries in each of the twelve states included in the recruitment of the cohort or from deaths with underlying cause coded as pancreatic cancer. Although lags in reporting occur, all but a small minority of incident cancer diagnoses are captured through the linkages. Deaths and causes of death were identified via linkage with both the Social Security Administration (SSA) and the National Death Index (NDI), and expected to capture nearly all deaths 14, 15.
The t-test and general linear models were used to test for differences in continuous variables across the comparison groups and chi square tests were used to test for differences in categorical data. Cox proportional hazards regression with age as the time scale was used to compute hazard ratios and their 95% confidence intervals (CIs) for the relative risk of incident pancreatic cancer according to status of diabetes and other covariates. The multivariate modeling was carried out for African American and White participants separately and then combined including a term for self-reported race in addition to terms for diabetes, sex, body mass index (BMI) categories, hypercholesterolemia, smoking status, alcohol consumption, physical activity, education, annual household income, and baseline insurance coverage. Due to sample size limitations in other racial/ethnic groups, effect modification by race, i.e. a race by diabetes interaction, was assessed and race-specific analyses conducted among African American and White participants only. Analyses were conducted both with and without the incident diabetes cases included in the analyses, with incident cases included with the non-diabetes cases at study baseline and then again with these incident cases completely excluded from the analyses. We also conducted additional sensitivity analyses in which we began counting follow up time among those with diabetes six, twelve, eighteen months after study enrollment (i.e., excluding pancreatic cancer cases occurring within 6/12/18 months after enrollment) at baseline, the first and then the second follow-up periods to examine the possibility of reverse causation of secondary diabetes resulting from pancreatic cancer. Too few pancreatic cancer cases existed for analyses restricted to the third follow-up period. Follow-up was calculated as time between study enrollment to time of diagnosis of pancreatic cancer, death or December 31, 2015, whichever occurred first. The criterion for statistical significance was a two-tailed P-value <0.05. Statistical analyses were conducted using SAS version 9.4 (Cary, North Carolina).
Results
Diabetes was reported at baseline for 15,913 (22%) participants, representing 23% of African American and 20% of White SCCS participants. Characteristics of participants at study enrollment by baseline diabetes status are presented in Table 1. Compared to those without diabetes, those with diabetes at study baseline were older, more likely to be female and non-White participants. Persons with diabetes also had a higher BMI, were more likely to have a history of high cholesterol, were less physically active, but were also less likely to be current smokers and consumed less alcohol. Socioeconomic indices were also poorer among those with diabetes. No differences in family history of pancreatic cancer were observed.
Table 1.
Baseline Characteristics of the Southern Community Cohort Study by Baseline Diabetes Status, mean ± std, median (IQR) or % (n).
| Baseline diabetes N=15,913 | No diabetes N=57,219 | |
|---|---|---|
| Age, years | 54.8 ± 8.9 | 51.2 ± 8.5d |
| Sex, women | 65.2 (10,538) | 57.6 (32,964)d |
| Race† | ||
| Black | 68.4 (11,055) | 64.9 (37,114)d |
| White | 27.4 (4,425) | 31.3 (17,908) |
| Other races | 4.2 (679) | 3.8 (2,197) |
| Diabetes duration, years | 6.0 (2–13) | |
| Antihyperglycemic medication therapy | ||
| On insulin therapy alone | 17.2 (2,783) | ------------------------------ |
| Metformin alone | 5.5 (891) | |
| Insulin or metformin plus another anti-hyperglycemic medication |
16.2 (2,618) | ------------------------------ |
| Antihyperglycemic therapy other than insulin or metformin |
39.2 (6,322) | ------------------------------ |
| Not on antihyperglycemic therapy |
21.9 (3,530) | |
| Family history of pancreatic cancer | 1.6 (252) | 1.5 (842) |
| BMI, kg/m2* | 33.9 ± 8.0 | 29.2 ± 7.1d |
| BMI category† | ||
| Underweight | 0.52 (84) | 1.5 (842) |
| Normal weight | 10.4 (1,675) | 28.9 (16,534) |
| Overweight | 23.6 (3,818) | 31.5 (18,012) |
| Obese class 1 | 26.4 (4,262) | 20.3 (11,609) |
| Obese class 2+ | 39.1 (6,320) | 17.9 (10,222) |
| History of high cholesterol† | ||
| None | 44.9 (7,250) | 73.0 (41,767) |
| Treated | 39.4 (6,367) | 13.0 (7,455) |
| Not treated | 15.7 (2,542) | 14.0 (7,997) |
| Smoking status† | ||
| Never | 41.4 (6,692) | 34.6 (19,806) |
| Current | 29.6 (4,786) | 45.2 (25,875) |
| Former | 28.9 (4,674) | 20.1 (11,511) |
| Pack Years* | 4.0 (0–22.5) | 7.3 (0–22.5)d |
| Alcohol consumption, drinks /day* | 0.63 ± 2.7 | 1.65 ± 4.39d |
| Total physical activity, mets/hr* | 14.2 (6.9–24.9) | 18.9 (9.8–32.8)d |
| Less than high school graduate† | 34.5 (5,582) | 26.9 (15,374)d |
| Annual household income <$15,000† | 61.3 (9,908) | 53.9 (30,831)d |
| Insurance coverage† | 67.6 (10,929) | 57.6 (32,946)d |
P<0.05
p<0.01
p<0.001
p<0.0001
P<0.0001 for global differences between diabetes groups
Natural logarithmically transformed before analysis.
During an average of 10.5 years follow-up, there were 265 incident cases of pancreatic cancer, with 81 among those with and 184 among those without diabetes at cohort entry. The 10-year age specific incidence rates, stratified by race and diabetes status, are presented in Appendix Table A1. For most age strata, particularly for African Americans, incidence rates were higher among those without diabetes. However, as shown in Table 1, women were more likely to have diabetes at study baseline, a potential confounder. The multivariable relationships of diabetes and other potential risk factors for pancreatic cancer are presented in Table 2. In analyses controlling for age, sex, race, BMI category, hypercholesterolemia, smoking pack years, alcohol consumption, physical activity, education, income and insurance coverage, diabetes was associated with an approximately 50% increased risk of pancreatic cancer (HR = 1.54, 95% CI 1.16–2.05). Other significant risk factors for pancreas cancer included being male, a BMI less than 20 kg/m2, and smoking. Risk was also increased among African American participants, suggesting that diabetes was a confounder in the relationship between African American race and pancreatic cancer. Included in the analyses among those without diabetes were participants who developed diabetes during follow-up. Appendix Table A2 shows that when these participants were excluded from the analysis, the relationship of diabetes (HR=1.48, 95% CI=1.10–1.99) and the other covariates with pancreatic cancer were essentially the same as that of the full model presented in Table 2.
Table 2.
Multivariabl-adjusted Cox regression on the association of diabetes, race and other variables with incident pancreatic cancer
| HR (95% CI) | |
|---|---|
| Diabetes | 1.54 (1.16 –2.05) |
| Sex, women | 0.78 (0.60–1.02) |
| Race | |
| White | Ref |
| African American | 1.36 (1.02–1.81) |
| Other races | 0.99 (0.48–2.06) |
| Family history of pancreatic cancer | 1.03 (0.38–2.77) |
| BMI | |
| Normal weight | Ref |
| Underweight | 3.16 (1.62–6.14) |
| Overweight | 0.94 (0.68–1.30) |
| Obese class 1 | 0.63 (0.43–0.94) |
| Obese class 2+ | 0.91 (0.62–1.33) |
| History of high cholesterol | |
| None | Ref |
| Treated | 0.74 (0.53–1.03) |
| Untreated | 0.89 (0.62–1.28) |
| Smoking pack yearsa | 1.16 (1.07–1.26) |
| Alcohol consumptiona | 0.94 (0.78–1.14) |
| Physical activitya | 0.91 (0.80–1.03) |
| Less than high school graduate | 1.06 (0.80–1.39) |
| Annual household income <$15,000 | 0.96 (0.74–1.26) |
| Insurance coverage | 0.97 (0.73–1.27) |
Natural logarithmically transformed after the addition of 1 to each value.
BMI categories were defined as follows: Underweight: BMI < 20 kg/m2; Normal weight: 20 ≤ BMI < 25 kg/m2; Overweight: 25 ≤ BMI < 30 kg/m2; Obese class 1: 30 ≤ BMI < 35 kg/m2; Obese class 2+: BMI ≥ 35 kg/m2
The table includes all variables included in the multivariable adjusted analyses. Age was used as the time scale in these Cox regression analyses.
Race-specific analyses of diabetes with incident pancreatic cancer are presented in Table 3. Diabetes was associated with higher risks of pancreatic cancer for both African American (HR=1.51, 95% CI=1.08–2.11) and White (HR=1.78, 95% CI=1.00–3.16) participants. Other risk factors for pancreatic cancer were essentially similar between African American and White cohort members. Results by sex-race group showed no significant sex-by race interaction between diabetes and pancreatic cancer, with the stratified results as following: White women (HR=1.92, 0.91–4.04): Black men (HR=1.55, 0.94–2.56); Black women (HR=1.43, 0.93–2.21); and White men (HR=1.34, 0.57–3.16 (data not depicted).
Table 3.
Multivariable-adjusted Cox regression on the association of diabetes and other variables with incident pancreatic cancer stratified by race
| African American participants HR (95% CI) | White participants HR (95% CI) | |
|---|---|---|
| Diabetes | 1.51 (1.08–2.11) | 1.78 (1.00–3.16) |
| Sex, women | 0.80 (0.58–1.11) | 0.73 (0.44–1.21) |
| Family history of pancreatic cancer | 0.91 (0.23–3.69) | 1.30 (0.32–5.33) |
| BMI | ||
| Underweight | 3.65 (1.73–7.72) | 2.35 (0.54–10.20) |
| Normal weight | Ref | Ref |
| Overweight | 0.85 (0.57–1.25) | 1.06 (0.57–1.96) |
| Obese, class 1 | 0.59 (0.37–0.95) | 0.63 (0.28–1.38) |
| Obese, class 2+ | 0.83 (0.53–1.31) | 1.06 (0.50–2.23) |
| History of high cholesterol | ||
| None | Ref | Ref |
| Treated | 0.75 (0.50–1.12) | 0.74 (0.41–1.35) |
| Untreated | 0.74 (0.46–1.18) | 1.11 (0.59–2.08) |
| Smoking pack yearsa | 1.17 (1.06–1.30) | 1.07 (0.92–1.23) |
| Alcohol consumptiona | 0.89 (0.71–1.12) | 1.22 (0.85–1.74) |
| Physical activitya | 0.91 (0.78–1.05) | 0.91 (0.71–1.19) |
| Less than high school graduate | 0.97 (0.70–1.33) | 1.31 (0.75–2.29) |
| Annual household income <$15,000 | 0.98 (0.71–1.35) | 0.91 (0.53–1.55) |
| Insurance coverage | 0.99 (0.71–1.37) | 1.03 (0.58–1.83) |
Natural logarithmically transformed after the addition of 1 to each value.
BMI categories were defined as follows: Underweight: BMI < 20 kg/m2; Normal weight: 20 ≤ BMI < 25 kg/m2; Overweight: 25 ≤ BMI < 30 kg/m2; Obese class 1: 30 ≤ BMI < 35 kg/m2; Obese class 2+: BMI ≥ 35 kg/m2.
p-value for interaction of race by diabetes status=0.65.
The table includes all variables included in the multivariable adjusted analyses. Age was used as the time scale in these Cox regression analyses.
Table 4 shows that among those with diabetes at study baseline, no clear association with pancreatic cancer risk was observed for diabetes duration. Compared to those with a diabetes duration of less than five years, HRs (95% CIs) for those with a duration of five to 9 years and ≥ 10 years were 1.75 (1.01–3.03) and 0.93 (0.52–1.66). Risk factors for pancreatic cancer tended to be similar among those with vs. without diabetes, although a stronger association with BMI, with highest risk among the underweight, was evident among those with diabetes (Table 4).
Table 4.
Multivariable Adjusted Hazard Ratios (and 95% Confidence Intervals) for Incident Pancreatic Cancer among those with and without Diabetes
| Diabetes HR (95% CI) | Without diabetes HR (95% CI) | |
|---|---|---|
| Diabetes duration | ----------------------- | |
| No diabetes | --------------------------- | ------------------------ |
| 0 -< 5 years | Ref | ------------------------- |
| 5 - <10 years | 1.75 (1.01–3.03) | ------------------------- |
| ≥ 10 years | 0.93 (0.52–1.66) | ------------------------- |
| Sex, women | 0.88 (0.54–1.46) | 0.78 (0.56–1.06) |
| Race | ||
| White | Ref | Ref |
| African American | 1.32 (0.77–2.26) | 1.37 (0.97–1.93) |
| Other races | 0.40 (0.05–3.00) | 1.30 (0.59–2.87) |
| Antihyperglycemic medications | ||
| None | Ref | ---------------------- |
| Insulin alone | 1.13 (0.57–2.25) | ---------------------- |
| Metformin alone | 0.68 (0.20–2.32) | ---------------------- |
| Insulin or metformin plus one other anti-hyperglycemic medication | 0.53 (0.22–1.29) | ----------------------- |
| Antihyperglycemic medication other than insulin or metformin | 0.81 (0.46–1.43) | ------------------------- |
| Family history of pancreatic cancer | 2.45 (0.77–7.86) | 0.38 (0.05–2.69) |
| BMI category | ||
| Normal weight | Ref | Ref |
| Underweight | 6.39 (1.33–30.64) | 2.66 (1.27–5.59) |
| Overweight | 1.14 (0.50–2.57) | 0.89 (0.62–1.28) |
| Obese class 1 | 1.07 (0.47–2.44) | 0.50 (0.30–0.83) |
| Obese class 2+ | 1.10 (0.49–2.49) | 0.95 (0.61–1.50) |
| History of high cholesterol | ||
| None | Ref | Ref |
| Treated | 0.68 (0.41–1.13) | 0.81 (0.52–1.26) |
| Not treated | 0.77 (0.39–1.50) | 0.96 (0.63–1.47) |
| Smoking pack years* | 1.20 (1.04–1.39) | 1.13 (1.02–1.25) |
| Alcohol consumption* | 0.80 (0.46–1.37) | 0.98 (0.80–1.20) |
| Physical activity* | 0.99 (0.78–1.26) | 0.88 (0.76–1.02) |
| Less than high school graduate | 0.83 (0.50–1.36) | 1.21 (0.87–1.67) |
| Annual household income <$15,000 | 0.80 (0.50–1.30) | 1.04 (0.76–1.45) |
| Insurance coverage | 0.61 (0.37–1.01) | 1.17 (0.84–1.62) |
Natural logarithmically transformed after the addition of 1 to each value.
The table includes all variables included in the multivariable adjusted analyses. Age was used as the time scale in these Cox regression analyses.
In sensitivity analyses (Appendix Table A3), restricting analyses to after the first six, twelve or eighteen months of follow-up also had no material effect on the relationship between pancreatic cancer risk and the presence of diabetes at cohort entry or incident diabetes captured during the first or second follow-up period. The hazard ratios for incident diabetes tended to be higher during the second follow-up period compared to baseline, but the CIs were wide. Finally, in another sensitivity analysis, excluding the first 3 years after diagnosis of diabetes increased the relationship of prevalent diabetes at study enrollment with subsequent pancreatic risk (HR=2.31, 95% CI=1.32–4.05), data not depicted. Further competing risks analysis (Appendix Table A4) showed negligible to no differences in the association of diabetes, diabetes duration, or race accounting the potential competing risk of dying from other causes prior to being able to develop pancreatic cancer.
Discussion
In this population-based prospective study of over 80,000 predominately low-income African American and White participants, diabetes was associated with increased risk of pancreatic cancer by 50%. Pancreatic cancer risk factor profiles tended to be similar among both races. Our data suggest that among low-income African American and White adults, diabetes increases pancreatic cancer risk in both races.
The risk of incident pancreatic cancer was 36% higher among African American than White SCCS participants. From Table 1 it can be seen that the prevalence of diabetes at baseline was higher among African American than among White participants, but the difference was small (23% vs 20%) and not enough to account for the 36% higher risk of pancreatic cancer in the SCCS population. We have previously reported that diabetes incidence after entry into the cohort was nearly twice as high among African American than White participants of normal weight, with the relative increase declining with rising BMI and nearly disappearing among the morbidly obese with BMI > 40 kg/m2.11 Incorporating a higher prevalence plus incidence of diabetes among African American than White participants would raise the fraction of pancreatic cancer among African American participants in the SCCS attributable to diabetes, but this higher fraction would be offset by the lower hazard ratio of pancreatic cancer associated with diabetes for African American (HR 1.51) than White (HR 1.78) participants shown in Table 3. The overall result is that diabetes does not seem to account for the higher rate of pancreatic cancer among African American than White participants in the SCCS. African American race was associated with increased risk of pancreatic cancer even among people without diabetes (Table 4), suggesting that factors other than racial differences in propensity for diabetes contributed to racial differences in risk of pancreatic cancer.
Because of its very high 5-year mortality rate, pancreatic cancer death is often used as a proxy for cancer incidence. In the Cancer Prevention Study (CPS) II, 48,525 African American and 1,011,864 White American adults were followed for pancreatic cancer death, of which there were 360 cases among African American and 5,983 cases among White American participants.16 Diabetes was a risk factor for pancreatic cancer among White (HR=1.36, 95 % CI=1.22–1.51) but not Black (HR=0.97, 95% CI=0.68–1.38) participants in this CPS II study, contrary to the similarly increased risk among both African American and White participants in our study. In a multicenter case-control study which included 526 pancreatic cases and 2,153 population controls, diabetes was associated with a two-fold excess risk among African American and White men, with similar population attributable risks of approximately 10%, but among women diabetes was associated with pancreatic cancer risk only among White participants.15 Diabetes has been linked to increased pancreatic cancer in numerous epidemiological studies among predominantly non-African ancestry populations, with excess risk approximately two-fold in those with compared to those without diabetes.6, 17
Mechanisms by which diabetes is hypothesized to increase the risk of pancreas cancer include endogenous hyperinsulinemia and exogenous insulin therapy,18, 19 glucose toxicity, high iron diet, and excess circulating and body iron stores.20, 21 Other shared risk factors include smoking, obesity, low socioeconomic status, and African American race.
Hyperglycemia or overt diabetes is present in the majority of pancreatic cancer cases.3, 4, 22 In the majority of pancreatic cancer cases with comorbid diabetes, diagnosis of diabetes occurred within the 2 years preceding pancreatic cancer diagnosis.23 One explanation for the increased risk of pancreatic cancer often observed with diabetes is reverse causality where Type 3c diabetes, misdiagnosed as Type 2 diabetes, appears to increase the risk of pancreatic cancer. Type 3c diabetes is a form of secondary diabetes caused by pancreatic cancer or other exocrine pancreatic disorders. In the majority of the cases diabetes is a comorbidity of pancreatic cancer, with the diagnosis of diabetes occurring within 2 to five years of pancreatic cancer diagnosis24, 25. While HRs for pancreatic cancer associated with prevalent diabetes at cohort entry and incident diabetes subsequent to cohort entry were similar in our population, restricting follow up to at least 3 years following diabetes diagnosis tended to strengthen the association of baseline diabetes with pancreatic cancer, although no clear pattern of association was observed with diabetes duration. However, sensitivity analyses revealed that risk among those with diagnosis of diabetes first reported at the second follow-up survey were greater than those with diabetes prevalent at baseline or first reported at the first follow-up. This suggests that in recently diagnosed diabetes, diabetes duration is inversely related to pancreatic cancer risk. While this inverse relationship between diabetes duration and pancreatic cancer has been observed by others,7 duration of diabetes was not associated with survival in a pooled analyses of 2,792 pancreatic cancer cases.26
We did not find an association between obesity and pancreatic cancer risk, even though obesity is an established mild risk factor.27 A portion (62 cases) of our pancreatic cancer cases were included in a pooled analysis of BMI and pancreas cancer among African American adults27. That study found about a 30% increased risk among those with BMI > 35 kg/m2 who had been followed at least 5 years. In the current analysis, a higher BMI compared to those of normal weight was not associated with an increased risk of pancreatic cancer. The highest risk was observed among those with a BMI of less than 20 kg/m2, who were at a nearly 3-fold increase of pancreatic cancer. While this may to a certain extent represent weight loss or cachexia in those with subclinical undiagnosed pancreatic cancer, excluding the first three years since diagnosis of diabetes did not change this 3-fold increased risk. Smoking, another established risk factor for both diabetes and pancreatic cancer28, was linked to increased risk for pancreatic cancer in our population.
Strengths of our study include its large sample size, including the largest traditional cohort study of African American adults with diabetes to date, and a population of African American and White persons fairly homogenous with respect to SES status. Another strength is that by recruiting largely from federally qualified health centers, African American and White participants also had generally similar access to health care. The similar SES status of our population, by both race and diabetes status, enabled us to account for a major confounder in the association of both diabetes and race with pancreatic cancer.
A study limitation was our reliance on self-reported diabetes status. However, separate validation efforts based on review of medical records and/or A1c levels for samples of SCCS participants confirmed over 97% of the self-reports 29. We also did not have data on HbA1c for the majority of the participants, so the relationship of hyperglycemia with pancreatic cancer risk could not be assessed. Power to address potential interactions was limited, as was power to evaluate associations between medications used in diabetes treatment. We computed hazard ratios associated with use of insulin, metformin, and other diabetes medications and found an increased risk for insulin use and non-significant protective association for metformin, but confidence limits were wide. A study by Carstensen et al 30describing the association between cancer incidence and type 1 diabetes found a smaller excess risk among persons with type 1 diabetes than had been previously observed for type 2 diabetes, suggesting that insulin therapy does not contribute to elevated diabetes risk among people with diabetes. We did not data on diabetes type and given that the majority of the study participants were diagnosed with diabetes after the age of 40 years, most are presumed to have type 2 diabetes, thus we could not further investigate this in detail in our study. We also did not have data on fasting glucose or a history of pancreatitis, known risk factors for diabetes.
In conclusion, in this prospective study of over 70,000 middle-aged and elderly adults, diabetes was associated with a moderately increased risk of pancreatic cancer, increased risk similarly in both African American and White participants, but did not account for the excess risk of pancreatic cancer in African American people. Further investigation into factors accounting for the diabetes and socioeconomic status mediating excess risk of pancreatic cancer among African American adults is warranted.
Supplementary Material
Highlights.
Diabetes increases pancreatic cancer risk similarly in African and White Americans
Diabetes contributes to but does not account for the increased pancreatic cancer risk in African Americans
The low BMI as a stronger risk factor in those with compared to those without diabetes may be due to reverse causality
Acknowledgements
Research reported in this publication was supported by the National Cancer Institute of the National Institutes of Health under Award Number U01CA202979. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. SCCS data collection was performed by the Survey and Biospecimen Shared Resource which is supported in part by the Vanderbilt-Ingram Cancer Center (P30 CA68485). No potential conflicts of interest relevant to this article were reported.
RBNC wrote the manuscript and analyzed and interpreted the data. HM analyzed the data and reviewed the manuscript for scientific content. AGH, DF, and DAM reviewed the manuscript, helped with the interpretation of the data and contributed to the discussion. WJB collected the data, contributed to the research design and methods and discussion, and reviewed and edited the manuscript for scientific content. All authors gave final approval for this version of the manuscript to be published.
Parts of this study were presented in abstract and poster form at the 78th and 80th Scientific Sessions of the American Diabetes Association, June 2018 and June 2020.
Abbreviations:
- (SCCS)
Southern Community Cohort Study
- (CHC)
Community health centers
- (SSA)
Social Security Administration
- (NDI)
National Death Index
- (CPS)
Cancer Prevention Study
Data Availability
Data are available upon an approved request to the Southern Community Cohort Study biospecimen Committee.
References
- 1.Cancer Stat Facts: Pancreatic Cancer. Accessed March 3, 2022. National Cancer Institute: Surveillance, Epidemiology, and End Results Program. [Google Scholar]
- 2.Scarton L, Yoon S, Oh S, et al. Pancreatic Cancer Related Health Disparities: A Commentary. Cancers. Jul 18 2018;10(7)doi: 10.3390/cancers10070235 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Permert J, Ihse I, Jorfeldt L, von Schenck H, Arnqvist HJ, Larsson J. Pancreatic cancer is associated with impaired glucose metabolism. The European journal of surgery = Acta chirurgica. Feb 1993;159(2):101–7. [PubMed] [Google Scholar]
- 4.Pannala R, Leirness JB, Bamlet WR, Basu A, Petersen GM, Chari ST. Prevalence and clinical profile of pancreatic cancer-associated diabetes mellitus. Gastroenterology. Apr 2008;134(4):981–7. doi: 10.1053/j.gastro.2008.01.039 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Tsuchiya R, Noda T, Harada N, et al. Collective review of small carcinomas of the pancreas. Annals of surgery. Jan 1986;203(1):77–81. doi: 10.1097/00000658-198601000-00013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Everhart J, Wright D. Diabetes mellitus as a risk factor for pancreatic cancer. A meta-analysis. Jama. May 24–31 1995;273(20):1605–9. [PubMed] [Google Scholar]
- 7.Huxley R, Ansary-Moghaddam A, Berrington de Gonzalez A, Barzi F, Woodward M. Type-II diabetes and pancreatic cancer: a meta-analysis of 36 studies. British journal of cancer. Jun 6 2005;92(11):2076–83. doi: 10.1038/sj.bjc.6602619 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Gordon-Dseagu VL, Devesa SS, Goggins M, Stolzenberg-Solomon R. Pancreatic cancer incidence trends: evidence from the Surveillance, Epidemiology and End Results (SEER) population-based data. International journal of epidemiology. Apr 1 2018;47(2):427–439. doi: 10.1093/ije/dyx232 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Haffner SM, D’Agostino R, Saad MF, et al. Increased insulin resistance and insulin secretion in nondiabetic African-Americans and Hispanics compared with non-Hispanic whites. The Insulin Resistance Atherosclerosis Study. Diabetes. Jun 1996;45(6):742–8. doi: 10.2337/diab.45.6.742 [DOI] [PubMed] [Google Scholar]
- 10.Osei K, Schuster DP, Owusu SK, Amoah AG. Race and ethnicity determine serum insulin and C-peptide concentrations and hepatic insulin extraction and insulin clearance: comparative studies of three populations of West African ancestry and white Americans. Metabolism. Jan 1997;46(1):53–8. doi: 10.1016/s0026-0495(97)90167-0 [DOI] [PubMed] [Google Scholar]
- 11.Conway BN, Han X, Munro HM, et al. The obesity epidemic and rising diabetes incidence in a low-income racially diverse southern US cohort. PLoS One. 2018;13(1):e0190993. doi: 10.1371/journal.pone.0190993 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.National Diabetes Statistics Report, 2020 (Centers for Disease Control and Prevention, U.S. Department of Health and Human Services) (2020). [Google Scholar]
- 13.Signorello LB, Hargreaves MK, Blot WJ. The Southern Community Cohort Study: Investigating health disparities. J Health Care Poor Underserved. 2010;21:26–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Calle EE, Terrell DD. Utility of the National Death Index for ascertainment of mortality among cancer prevention study II participants. American journal of epidemiology. Jan 15 1993;137(2):235–41. doi: 10.1093/oxfordjournals.aje.a116664 [DOI] [PubMed] [Google Scholar]
- 15.Lash TL, Silliman RA. A comparison of the National Death Index and Social Security Administration databases to ascertain vital status. Epidemiology (Cambridge, Mass). Mar 2001;12(2):259–61. doi: 10.1097/00001648-200103000-00021 [DOI] [PubMed] [Google Scholar]
- 16.Arnold LD, Patel AV, Yan Y, et al. Are racial disparities in pancreatic cancer explained by smoking and overweight/obesity? Cancer epidemiology, biomarkers & prevention : a publication of the American Association for Cancer Research, cosponsored by the American Society of Preventive Oncology. Sep 2009;18(9):2397–405. doi: 10.1158/1055-9965.Epi-09-0080 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Song S, Wang B, Zhang X, et al. Long-Term Diabetes Mellitus Is Associated with an Increased Risk of Pancreatic Cancer: A Meta-Analysis. PloS one. 2015;10(7):e0134321. doi: 10.1371/journal.pone.0134321 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Wolpin BM, Bao Y, Qian ZR, et al. Hyperglycemia, insulin resistance, impaired pancreatic β-cell function, and risk of pancreatic cancer. Journal of the National Cancer Institute. Jul 17 2013;105(14):1027–35. doi: 10.1093/jnci/djt123 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Fisher WE, Boros LG, Schirmer WJ. Insulin promotes pancreatic cancer: evidence for endocrine influence on exocrine pancreatic tumors. The Journal of surgical research. Jun 1996;63(1):310–3. doi: 10.1006/jsre.1996.0266 [DOI] [PubMed] [Google Scholar]
- 20.Barton JC, Acton RT. Diabetes in HFE Hemochromatosis. Journal of diabetes research. 2017;2017:9826930. doi: 10.1155/2017/9826930 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Bao W, Rong Y, Rong S, Liu L. Dietary iron intake, body iron stores, and the risk of type 2 diabetes: a systematic review and meta-analysis. BMC Med. Oct 10 2012;10:119. doi: 10.1186/1741-7015-10-119 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Wakasugi H, Funakoshi A, Iguchi H. Clinical observations of pancreatic diabetes caused by pancreatic carcinoma, and survival period. International journal of clinical oncology. Feb 2001;6(1):50–4. [DOI] [PubMed] [Google Scholar]
- 23.Chari ST, Klee GG, Miller LJ, Raimondo M, DiMagno EP. Islet amyloid polypeptide is not a satisfactory marker for detecting pancreatic cancer. Gastroenterology. Sep 2001;121(3):640–5. [DOI] [PubMed] [Google Scholar]
- 24.Andersen DK, Korc M, Petersen GM, et al. Diabetes, Pancreatogenic Diabetes, and Pancreatic Cancer. Diabetes. May 2017;66(5):1103–1110. doi: 10.2337/db16-1477 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Aggarwal G, Rabe KG, Petersen GM, Chari ST. New-onset diabetes in pancreatic cancer: a study in the primary care setting. Pancreatology Mar-Apr 2012;12(2):156–61. doi: 10.1016/j.pan.2012.02.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Jeon CY, Li D, Cleary S, et al. The Association of Recently Diagnosed Diabetes and Long-term Diabetes With Survival in Pancreatic Cancer Patients: A Pooled Analysis. Pancreas. Mar 2018;47(3):314–320. doi: 10.1097/mpa.0000000000000989 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Bethea TN, Kitahara CM, Sonderman J, et al. A pooled analysis of body mass index and pancreatic cancer mortality in african americans. Cancer epidemiology, biomarkers & prevention : a publication of the American Association for Cancer Research, cosponsored by the American Society of Preventive Oncology. Oct 2014;23(10):2119–25. doi: 10.1158/1055-9965.Epi-14-0422 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Maisonneuve P, Lowenfels AB. Risk factors for pancreatic cancer: a summary review of meta-analytical studies. Int J Epidemiol Feb 2015;44(1):186–98. doi: 10.1093/ije/dyu240 [DOI] [PubMed] [Google Scholar]
- 29.Huizinga M, Elasy T, Villegas R, Signorello L, Blot W, Cavanaugh K. Validation of diabetes self-report and characteristics of undiagnosed diabetes in the Southern Community Cohort Study. Diabetes. 2009;58(S1):A279. [Google Scholar]
- 30.Carstensen B, Read SH, Friis S, et al. Cancer incidence in persons with type 1 diabetes: a five-country study of 9,000 cancers in type 1 diabetic individuals. Diabetologia. May 2016;59(5):980–8. doi: 10.1007/s00125-016-3884-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data are available upon an approved request to the Southern Community Cohort Study biospecimen Committee.