Relationships of Serum Ferritin, Transferrin Saturation, and HFE Mutations and Self-Reported Diabetes in the Hemochromatosis and Iron Overload Screening (HEIRS) Study

  1. Ronald T. Acton, PHD1,
  2. James C. Barton, MD2,
  3. Leah V. Passmore3,
  4. Paul C. Adams, MD4,
  5. Mark R. Speechley, PHD5,
  6. Fitzroy W. Dawkins, MD6,
  7. Phyliss Sholinsky, MSPH7,
  8. David M. Reboussin, PHD3,
  9. Gordon D. McLaren, MD89,
  10. Emily L. Harris, PHD, MPH10,
  11. Thomas C. Bent, MD8,
  12. Thomas M. Vogt, MD10 and
  13. Oswaldo Castro, MD6
  1. 1Department of Microbiology, University of Alabama at Birmingham, Birmingham, Alabama
  2. 2Southern Iron Disorders Center, Birmingham, Alabama
  3. 3Department of Public Health Sciences, Wake Forest University School of Medicine, Winston-Salem, North Carolina
  4. 4Department of Medicine, London Health Sciences Centre, London, Ontario, Canada
  5. 5Department of Epidemiology and Biostatistics, University of Western Ontario, London, Ontario, Canada
  6. 6Department of Medicine, Howard University, Washington, DC
  7. 7Epidemiology and Biometry Program, National Heart, Lung, and Blood Institute, Bethesda, Maryland
  8. 8Department of Medicine, University of California, Irvine, California
  9. 9Veterans Affairs Long Beach Healthcare System, Long Beach, California
  10. 10Kaiser Permanente Center for Health Research, Portland, Oregon, and Honolulu, Hawaii
  1. Address correspondence and reprint requests to Ronald T. Acton, PhD, MCLM 265, 1530 3rd Ave. South, Birmingham, AL 35294-0005. E-mail: acton{at}uab.edu

Abstract

OBJECTIVE—We evaluated the associations of self-reported diabetes with serum ferritin concentration, transferrin saturation (TfSat), and HFE C282Y and H63D mutations in six racial/ethnic groups recruited at five field centers in the Hemochromatosis and Iron Overload Screening (HEIRS) study.

RESEARCH DESIGN AND METHODS—Analyses were conducted on 97,470 participants. Participants who reported a previous diagnosis of diabetes and/or hemochromatosis or iron overload were compared with participants who did not report a previous diagnosis.

RESULTS—The overall prevalence of diabetes was 13.8%; the highest prevalence was in Pacific Islanders (20.1%). Of all participants with diabetes, 2.0% reported that they also had hemochromatosis or iron overload. The mean serum ferritin concentration was significantly greater in women with diabetes in all racial/ethnic groups and in Native-American men with diabetes than in those without diabetes. The mean serum ferritin concentration was significantly lower in Asian men with diabetes than in those without diabetes. Mean TfSat was lower in participants with diabetes from all racial/ethnic groups except Native-American women than in those without diabetes. There was no significant association of diabetes with HFE genotype. The mean serum ferritin concentration was greater (P < 0.0001) in women with diabetes than in those without diabetes for HFE genotypes except C282Y/C282Y and C282Y/H63D. Log serum ferritin concentration was significantly associated with diabetes in a logistic regression analysis after adjusting for age, sex, racial/ethnic group, HFE genotype, and field center.

CONCLUSIONS—Serum ferritin concentration is associated with diabetes, even at levels below those typically associated with hemochromatosis or iron overload.

Mean serum ferritin concentrations in individuals with type 2 diabetes or glucose intolerance are significantly higher than serum ferritin concentrations in control subjects (1). There are positive associations of serum ferritin concentration with fasting glucose and insulin concentrations, glucose intolerance, HbA1c (A1C) levels, insulin resistance, metabolic syndrome, and BMI (14). The serum ferritin concentration has been reported to predict the occurrence of type 2 diabetes (5,6).

Diabetes is a complication of hemochromatosis in whites (7). The prevalence of hemochromatosis or iron overload in persons with diabetes is equal to or greater than that in the general white population (7). The prevalence of type 2 diabetes is the same or greater in blacks with primary iron overload than in whites with hemochromatosis (8). Hemochromatosis is less prevalent in Hispanics than in non-Hispanic whites, although the prevalence of self-reported diabetes is similar in hemochromatosis and iron overload patients in both racial/ethnic groups (9). Missense mutations of the HFE gene are associated with susceptibility to iron overload (10). The prevalence of common HFE mutations is greater in some (11), but not all, diabetes cohorts (1214) than in corresponding control groups.

The relationships of diabetes to serum ferritin concentration, transferrin saturation (TfSat), and HFE mutations in subjects stratified by age and in some racial/ethnic groups are not well defined. Thus, we evaluated these variables in 97,470 Hemochromatosis and Iron Overload Screening (HEIRS) study participants representing six racial/ethnic groups (12,15) who self-reported diabetes at their initial screening.

RESEARCH DESIGN AND METHODS—

The institutional review board of each field center approved the study protocol (15). The HEIRS study recruited participants ≥25 years of age who gave informed consent. During the interval from February 2001 to February 2003, 103,118 participants were recruited from five field centers (12,15). Volunteers who reported that they participated because a family member also participated were excluded from the present analysis (12,15).

At initial screening, participants completed a questionnaire that included questions about their race/ethnicity and provided a blood specimen as previously described (15). Participants who reported multiple or unknown race/ethnicity were not included. The present cohort consisted of participants for whom serum ferritin concentration, TfSat, and HFE C282Y and H63D mutation data were available. Participants were also asked, “Has a doctor ever told you that you have 1) too much iron in your body, iron overload, or hemochromatosis or 2) diabetes?” Self-reports of diabetes were not confirmed by medical record review, obtaining information on participant use of diabetes medications, or measuring blood or serum levels of glucose or A1C. All TfSat and serum ferritin concentration measurements and HFE mutation analyses were performed at the Central Laboratory (located at Fairview-University Medical Center, Minneapolis, MN) except TfSat testing of London Health Sciences Centre participants (performed by MDS Laboratory Services, Toronto, Canada, using an identical method) (12,15,16). The interlaboratory correlation coefficient of original participant values and blind replicates was 0.98. Participants without C282Y or H63D are designated as having the HFE wild-type genotype (wt/wt).

Statistical methods—

Statistical analyses were performed using SAS (version 9.0; SAS Institute, Cary, NC). Serum ferritin concentration data are displayed as antilogs of mean loge-transformed serum ferritin concentration. Participants who reported having diabetes were compared with all available participants who did not report having diabetes. Logistic regression modeling was used to determine whether loge serum ferritin concentration was an independent predictor of diabetes and to test the hypothesis that the prevalence of diabetes increases with increasing serum ferritin concentration. The model included age, sex, racial/ethnic group, HFE genotype, and field center. Adjusted proportions and least square means were compared through regression modeling. Age was recorded as a discrete variable with categories for each 5 years from 25 to 90 and a single category for participants ≥90 years of age. A value of P < 0.05 was defined as significant.

RESULTS—

The prevalence of diabetes was lower in women than in men (odds ratio [OR] 0.882 [95% CI 0.849–0.914], P < 0.0001) (Table 1). When each decade increment for participants aged ≤60 years was compared with the others, the prevalences of diabetes were different (P < 0.0001); there was no difference between 60- to 69-year-old and ≥70-year-old participants. The prevalence of diabetes was greatest in Pacific Islanders (P < 0.0001). The prevalences of diabetes in white and Asian participants were similar but were lower than those in other racial/ethnic groups (P < 0.0001).

The highest (15.81%) and lowest (8.69%) prevalences of diabetes across all racial/ethnic groups were reported by participants at the Howard University and London, Ontario, field centers, respectively. The highest prevalences of diabetes in Pacific Islanders and Hispanics were reported by participants at the California field center (29.63 and 17.45%, respectively). The highest prevalences of diabetes in Asians (14.81%) and blacks (23.17%) were reported by participants at the Oregon/Hawaii field centers. The highest prevalence of diabetes in Native- American participants was observed at the London, Ontario, field center (18.68%). Whites at the Alabama field center reported the highest prevalence of diabetes (13.45%) among all whites.

The prevalences of hemochromatosis or iron overload reported by all participants and for whites who also reported having diabetes were 2.01 and 2.2%, respectively. There were no significant differences between men (2.18%) and women (1.89%) or among the various racial/ethnic groups except between Hispanics and whites (P = 0.0108). None of the 102 Native-American participants who reported having diabetes also reported having hemochromatosis or iron overload. The overall age effect for the prevalence of hemochromatosis or iron overload was not significant (P = 0.06).

Native-American men with self-reported diabetes had significantly greater mean serum ferritin concentrations than those without self-reported diabetes (Table 2). Asian men who reported that they had diabetes had significantly lower mean serum ferritin concentrations than those without self-reported diabetes (Table 2). In women, however, the mean serum ferritin concentration was significantly greater in those with self-reported diabetes than in those without self-reported diabetes. This characteristic was observed across all racial/ethnic groups (Table 2).

Mean TfSat values were significantly lower in participants with self-reported diabetes than in those without self-reported diabetes. This characteristic was observed in white, black, and Asian men and women and in Hispanic men (Table 2).

The odds of men and women in the various racial/ethnic groups having diabetes with increasing serum ferritin concentrations are displayed in Fig. 1. In an analysis using serum ferritin concentration quintiles, there was a significant interaction for the overall serum ferritin concentration effect among different race/ethnic groups compared with whites for women (P = 0.0185) but not men (P = 0.0986). There was a significant increasing linear relationship effect of serum ferritin concentration in white (P < 0.0001, OR 1.26), Asian (P < 0.0001, 1.37), black (P < 0.0001, 1.25), and Hispanic (P < 0.0001, 1.34) women and in white (P = 0.0002, 1.12), black (P = 0.0068, 1.13), and Hispanic (P = 0.0003, 1.30) men.

Logistic regression analyses were conducted to assess the interaction between serum ferritin concentration and Tfsat. For a fixed loge serum ferritin concentration, the risk of having diabetes decreases as the TfSat increases. However, the higher the loge serum ferritin concentration, the less the increase in risk of diabetes is due to lower levels of TfSat.

Frequencies of HFE genotypes were similar in participants with or without self-reported diabetes when data were stratified by race/ethnicity. The overall HFE C282Y and H63D frequencies among participants stratified by race/ethnicity are reported elsewhere (12). In all genotype categories, men had higher mean serum ferritin concentration than women. Across all HFE genotypes, mean serum ferritin concentration was similar in men with and without self-reported diabetes. Mean serum ferritin concentration values were greater (P < 0.0001) in women with diabetes than in women without diabetes for all HFE genotypes except C282Y/C282Y and C282Y/H63D. Women who had diabetes and C282Y/C282Y had lower mean serum ferritin concentrations than women who had C282Y/C282Y and no diabetes (P = 0.0051). In women with HFE C282Y/H63D, mean serum ferritin concentration values were similar in those with and without diabetes. Additional analyses were conducted to determine that HFE genotype had no effect on the proportion of self-reported diabetes. Similarly, the interaction between iron overload and HFE genotype was also a nonsignificant predictor. TfSat was significantly associated with reported diabetes but in a direction opposite of loge serum ferritin concentration. An increase in TfSat decreased the odds of self-reported diabetes.

In a multivariate logistic regression model that adjusted for age, sex, racial/ethnic group, HFE genotype, and field center, the loge serum ferritin concentration was an independent predictor of diabetes (OR 1.24 [95% CI 1.2–1.3], P < 0.0001).

CONCLUSIONS—

This is the largest study of multiple racial/ethnic groups to evaluate the associations of self-reported diabetes with serum ferritin concentration, TfSat, and HFE C282Y and H63D mutations. The overall prevalences of self-reported diabetes in HEIRS study participants were 13.80 and 7.9% in the Behavioral Risk Factor Surveillance System (BRFSS), a telephone survey of 195,005 white, black, and Hispanic adults ≥18 years of age (17). The two studies asked their respective participants similar questions about diabetes (15). The prevalence of diabetes in blacks was 17.48% in the HEIRS study and 11.2% in the BRFSS study (17). In the HEIRS study, prevalences of diabetes were similar in all racial/ethnic participants combined from Alabama, California, and District of Columbia field centers (15.79, 15.54, and 15.81%, respectively). In the BFRSS study, the highest prevalence of diabetes was observed in Alabama residents (10.5%) (17). Prevalences of diabetes in the HEIRS study are generally higher than those in the BRFSS study (17). This result could be explained by higher prevalences of diabetes in primary care clinics used for HEIRS study recruitment than in the population evaluated by the BRFSS study. In a pilot study of iron status and glycemic control among men attending the Veterans Affairs Primary Care Clinics in Jackson, Mississippi, the prevalences of diabetes among black and white participants were 26.7 and 22.7%, respectively (3). The highest prevalence of diabetes in the HEIRS study was observed in Pacific Islanders (20.06%), a racial/ethnic group that was not evaluated by the BRFSS study. In both studies, the prevalence of diabetes increased with age until 70 years in men and women. In the HEIRS study, the prevalence of diabetes was greater in men than in women; in the BRFSS study the highest prevalence of diabetes was observed in women.

In the HEIRS study, serum ferritin concentration was significantly higher in women with diabetes than in those without diabetes across all racial/ethnic groups. Among men, only Native Americans with diabetes had significantly greater mean serum ferritin concentration than men without diabetes. Asian men with diabetes had significantly lower mean serum ferritin concentration than Asian men without diabetes. Thus, Asian men are an exception to the relationship between serum ferritin concentration and diabetes observed in other groups. The probability of having diabetes also increased significantly with an increase in loge serum ferritin concentration for racial/ethnic groups except Pacific Islanders and Native Americans. In the National Health and Nutrition Examination Survey III (NHANES III) study, greater serum ferritin concentration was associated with increased diabetes risk in white men and women but not in blacks or Hispanics (1). In other studies, greater serum ferritin concentration was associated with diabetes in whites and blacks (3), Chinese, Malays, and Asian Indians (18) and Koreans (19). In Singapore Asians (18), mean serum ferritin concentration was significantly greater in men and women with diabetes than in nondiabetic control subjects. This result differs from observations of Asians in the present study. Other investigators observed that a relationship exists between serum ferritin concentration and insulin resistance in Chinese women but not men (20). In longitudinal studies, increased serum ferritin concentration predicted the development of diabetes in Finnish men (5) and in American white and nonwhite women (6).

Serum ferritin concentration is related directly to body iron stores in healthy individuals. Serum ferritin is also an acute-phase protein, levels of which are often increased out of proportion to body iron stores during inflammation. However, determining the relationship of serum ferritin concentration to inflammation was beyond the scope of the HEIRS study. In the NHANES III study (1) and the Nurses’ Health Study (6), the positive association of increased serum ferritin concentration and risk of developing diabetes was not changed significantly by cohort adjustment for inflammation by including serum C-reactive protein levels in the multivariate statistical model. This finding suggests that the relationship of serum ferritin concentration to diabetes (or to risk of developing diabetes) is unrelated to serum ferritin concentration elevation due to inflammation.

In the HEIRS study, mean TfSat was lower in many participant groups who reported having diabetes than in those without self-reported diabetes. Other investigators have reported that TfSat values were not elevated in patients with newly diagnosed diabetes (1). In a retrospective study, elevated TfSat was not associated with the development of diabetes (21). In contrast, the prevalence of elevated TfSat was higher in individuals with diabetes than the prevalence of elevated TfSat in historical control subjects and was independently associated with male sex (22).

Among HEIRS study participants, 2.07% of those who reported having diabetes also reported that they had hemochromatosis or iron overload. Among whites who reported having diabetes, 2.21% also reported that they had hemochromatosis or iron overload. The prevalences of hemochromatosis or iron overload in HEIRS study participants with self-reported diabetes are similar to those previously reported (7).

The HEIRS study is among the few large cohorts that has investigated the association of HFE mutations with diabetes in racial/ethnic groups other than whites (1214). A significant association of self-reported diabetes and HFE C282Y or H63D stratified by race/ethnicity was not observed in the present study; this finding is consistent with the results of some (1214) but not all studies (11). Although HFE mutations are often associated with increased serum ferritin concentration and TfSat values (12), mean serum ferritin concentration was similar in men with or without self-reported diabetes across all HFE genotypes. Serum ferritin concentration values were significantly greater in women with diabetes than in those without diabetes for HFE genotypes other than C282Y/C282Y and C282Y/H63D. Thus, HFE mutations may contribute to elevated serum ferritin concentration and TfSat values, but the HFE mutations are not associated with diabetes.

Obesity and mean BMI are risk factors for diabetes (17). In whites with HFE wt/wt, age-adjusted BMI was positively correlated with serum ferritin concentration and negatively correlated with TfSat (23). Height and weight measurements were not obtained at the time of initial HEIRS study screening (12,15), thus precluding analyses to determine whether greater mean BMI in HEIRS study participants could partly account for the association of mean serum ferritin concentration with diabetes in some of the racial/ethnic groups. However, in the NHANES III study (1) and the Nurses’ Health Study (6), the relationship between serum ferritin concentration and risk of developing diabetes remained significant after adjustment for BMI.

We conclude that higher mean serum ferritin concentration is significantly associated with self-reported diabetes in the six racial/ethnic groups evaluated in the HEIRS study. In contrast, there is a significant negative association of TfSat with diabetes. The association of serum ferritin concentration with self-reported diabetes is most consistent among women. The significant associations of serum ferritin concentration with self-reported diabetes in Native Americans and Pacific Islanders defined herein extend observations regarding the relationships of serum ferritin concentration and diabetes established in other racial/ethnic groups. The assessment of diabetes by self-report is a possible limitation of the present study. However, self-reporting probably creates a bias toward underestimating diabetes because diabetes in approximately one-third of individuals who have it is undiagnosed (24). Moreover, our results are consistent with those from longitudinal studies indicating that serum ferritin concentration is significantly associated with diabetes, even at serum ferritin concentration levels less than those typically associated with hemochromatosis or iron overload.

Figure 1—

Odds of having type 2 diabetes with increasing serum ferritin concentration (SF). Log ORs were estimated using logistic regression of self-reported type 2 diabetes on serum ferritin concentration quintile, age category, race/ethnicity, field center, and race/ethnicity by serum ferritin concentration quintile. The serum ferritin concentration range for each quintile is 7.5–47, 41–79, 79–134, and ≥239 μg/l, respectively. Asians and Pacific Islanders were combined due to the low numbers in each racial/ethnic group. A/PI, Asian/Pacific Islander.

Table 1—

Prevalence of self-reported diabetes in HEIRS study participants

Table 2—

Serum ferritin concentration and transferrin saturation in participants with and without self-reports of diabetes

Acknowledgments

The HEIRS study was initiated and funded by the National Heart, Lung, and Blood Institute (NHLBI), in conjunction with the National Human Genome Research Institute (NHGRI). The study was supported by Contracts N01-HC-05185 (University of Minnesota), N01-HC-05186 (Howard University), N01-HC-05188 (University of Alabama at Birmingham), N01-HC-05189 (Kaiser Permanente Center for Health Research), N01-HC-05190 (University of California, Irvine, CA), N01-HC-05191 (London Health Sciences Centre), and N01-HC-05192 (Wake Forest University). Additional support was provided by the Howard University General Clinical Research Center (GCRC) Grant M01-RR10284 and from the NHLBI and Office of Minority Health Grant UH1-HL03679-05, grant M01 RR00827 from the National Center for Research Resources (University of California, Irvine, CA), and GCRC grant M01-RR00032 from the National Center for Research Resources awarded to the University of Alabama at Birmingham.

We acknowledge the participation of these additional HEIRS study investigators: University of Alabama at Birmingham: Deborah Dixon, Susan Ferguson, Richard Jones, Jerry McKnight, Charles A. Rivers, Diane Tucker, and Janice C. Ware; University of California, Irvine: Christine E. McLaren, Hoda Anton-Culver, Jo Ann A. Baca, Lance C. Brunner, Michael M. Dao, Korey S. Jorgensen, Julie Kuniyoshi, Huan D. Le, Miles K. Masatsugu, Frank L. Meyskens, David Morohashi, Huan P. Nguyen, Sophocles N. Panagon, Chi Phung, Virgil Raymundo, Thomas Ton, Ann P. Walker, Lari B. Wenzel, and Argyrios Ziogas; London Health Sciences Center: Erin Bloch, Subrata Chakrabarti, Arlene Fleischhauer, Helen Harrison, Bonnie Hogan, Kelly Jia, John Jordan, Sheila Larson, Edward Lin, and Melissa Lopez; MDS Laboratories: Godfrey Moses, Lien Nguyen, Corry Pepper, Tara Power, Donald Sun, and Diane Woelfle; Kaiser Permanente Center for Health Research, Northwest and Hawaii, and Oregon Health and Science University: Mikel Aickin, Elaine Baker, Marjorie Erwin, Joan Holup, Carol Lloyd, Nancy Press, Richard D. Press, Jacob Reiss, Cheryl Ritenbaugh, Aileen Uchida, and Dwight Yim; Howard University: Victor R. Gordeuk, Margaret Fadojutimi-Akinsiku, Debra White-Coleman, Melvin Gerald, Barbara W. Harrison, Ometha Lewis-Jack, Robert F. Murray, Shelley McDonald-Pinkett, Angela Rock, Juan Romagoza, and Robert Williams; University of Minnesota and Fairview-University Medical Center: John H. Eckfeldt, Catherine Leiendecker-Foster, Ronald C. McGlennen, Greg Rynders, and Michael Y. Tsai; Wake Forest University: Beverly M. Snively, Roger Anderson, Elease Bostic, Brenda L. Craven, Shellie Ellis, Curt Furberg, Jason Griffin, Mark Hall, Darrin Harris, Leora Henkin, Dr. Sharon Jackson, Tamison Jewett, Mark D. King, Kurt Lohman, Laura Lovato, Joe Michaleckyj, Shana Palla, Tina Parks, Pradyumna D. Phatak, Stephen Rich, Andrea Ruggiero, Mara Vitolins, Gary Wolgast, and Daniel Zaccaro; National Institutes of Health/NHLBI: Ebony Bookman, Henry Chang, Richard Fabsitz, Cashell Jaquish, Teri Manolio, and Lisa O’Neill; National Institutes of Health/NHGRI: Elizabeth Thomson and Jean MacCluer. The Southwest Foundation for Biomedical Research also contributed to the design of this study.

Footnotes

  • A table elsewhere in this issue shows conventional and Système International (SI) units and conversion factors for many substances.

    The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

    • Accepted May 29, 2006.
    • Received August 24, 2005.

References

| Table of Contents