Diabetes Care Volume 42, November 2019 2075 Sanyog G. Shitole,1,2 Mary L. Biggs,3 Soluble CD14 and CD14 Variants, Alexander P. Reiner,2 fl Kenneth J. Mukamal,4,5 Luc Djousse,´ 5,6,7 Other In ammatory Markers, and Joachim H. Ix,8,9 Joshua I. Barzilay,10,11 Russell P. Tracy,12 David Siscovick,13 and Glucose Dysregulation in Older Jorge R. Kizer1,2 Adults: The Cardiovascular Health Study Diabetes Care 2019;42:2075–2082 | https://doi.org/10.2337/dc19-0723 EPIDEMIOLOGY/HEALTH SERVICES RESEARCH OBJECTIVE Experimental studies have implicated soluble (s)CD14, an effector of lipopolysac- charide-induced inflammation, in insulin resistance, but its role in human metabolic endotoxemia has not been studied. We evaluated sCD14 in relation to dysglycemia 1 in older adults and how this compares to other markers of inflammation. San Francisco VA Health Care System, San Francisco, CA 2 RESEARCH DESIGN AND METHODS University of California, San Francisco, San Francisco, CA Weinvestigated associations of sCD14, interleukin-6 (IL-6),CRP, and whiteblood cell 3University of Washington, Seattle, WA (WBC) count with insulin resistance (quantitative insulin-sensitivity check index and 4Beth Israel Deaconess Medical Center, Boston, HOMA 2 of insulin resistance) and incident type 2 diabetes in a population-based MA 5Harvard Medical School, Boston, MA cohort of older adults. We also assessed the causal role of sCD14 in insulin resistance 6Brigham and Women’s Hospital, Boston, MA using an instrumental variable approach by Mendelian randomization. 7VA Boston Healthcare System, Boston, MA 8University of California San Diego School of RESULTS Medicine, La Jolla, CA 9 After adjustment for conventional risk factors, each of the four biomarkers showed VA San Diego Healthcare System, San Diego, CA 10Kaiser Permanente Georgia Region, Atlanta, positive cross-sectional associations with both insulin resistance measures. These GA associations persisted after mutual adjustment for all markers except sCD14. Over a 11Emory University School of Medicine, Atlanta, median follow-up of 11.6 years, 466 cases of diabetes occurred. All biomarkers GA 12 except sCD14 were positively associated with diabetes, although only WBC count University of Vermont, Burlington, VT 13New York Academy of Medicine, New York, NY remained associated (hazard ratio 1.43 per doubling [95% CI 1.07, 1.90]) after Corresponding author: Jorge R. Kizer, jorge mutual adjustment. Instrumental variable analysis did not support a causal role for [email protected] sCD14 in insulin resistance. Received 10 April 2019 and accepted 6 August 2019 CONCLUSIONS This article contains Supplementary Data online Among older adults, sCD14 was associated with insulin resistance, but this at http://care.diabetesjournals.org/lookup/suppl/ disappeared after adjustment for other biomarkers, showed no evidence of a doi:10.2337/dc19-0723/-/DC1. causal basis, and was not accompanied by a similar association with diabetes. IL-6, © 2019 by the American Diabetes Association. CRP, and WBC count were each associated with insulin resistance and diabetes, WBC Readers may use this article as long as the work fi is properly cited, the use is educational and not count most robustly. These ndings do not support a central role for sCD14, but fi fl for pro t, and the work is not altered. More infor- they highlight the preeminence of WBC count as an in ammatory measure of mation is available at http://www.diabetesjournals diabetes risk in this population. .org/content/license. 2076 Inflammatory Markers in Glucose Dysregulation Diabetes Care Volume 42, November 2019 The role of chronic low-grade inflamma- CD14 on monocytes/macrophages (14). 5,296 individuals had measures of both tion in insulin resistance and subsequent Previous work by our group in CHS has fasting glucose and fasting insulin at development of diabetes has been well shown that elevated sCD14 is associated baseline, who were included in analyses documented experimentally (1). Consis- with incident cardiovascular disease and of insulin resistance measures. For these tent with such experimental findings, mortality in older adults (15). The link analyses, we had baseline measures of epidemiologic studies have linked vari- between sCD14 level and insulin resis- sCD14 in 4,999 participants and of IL-6, ous inflammatory biomarkers, including tance has been examined in small clinical CRP, and WBC count in 4,927, 5,302, and CRP and interleukin-6 (IL-6), to risk of studies, which have suggested an inverse 5,305 participants, respectively. After incident diabetes in women (2), middle- association (16) or one that differs by further exclusion of 515 individuals aged European cohorts (3), and middle- obesity status (17), but findings have with prevalent type 2 diabetes at base- aged to older multiethnic cohorts (4). been inconclusive. The relationship be- line by fasting glucose ($126 mg/dL), This also includes previous work from the tween sCD14 level and incident diabetes random glucose ($200 mg/dL), or anti- Cardiovascular Health Study (CHS), which has not been examined to date. In this hyperglycemic medication, our study found a significant association in older study, we sought to determine the asso- sample comprised 4,865 participants adults between CRP and development of ciations of sCD14 with abnormal glucose for assessment of the association with diabetes 3 to 4 years later (5). Addition- metabolism in a large cohort of older incident type 2 diabetes (hereafter re- ally, a meta-analysis in 2010 combined adults, both alone and in the context of ferred to as diabetes). For this set of data from 20 studies to conclude that an other inflammatory biomarkers not pre- analyses, we had baseline measures of increased total white blood cell (WBC) viously evaluated in relation to long-term sCD14 in 4,558 participants and of IL-6, count is associated with a higher risk of incidence of type 2 diabetes in this age- CRP, and WBC count in 4,502, 4,836, and diabetes (6). The pathways implicated in group. We also undertook an instrumen- 4,850 participants, respectively. For in- such chronic inflammation and conse- tal variable analysis using Mendelian strumental variable analyses of sCD14 in quent dysregulation of glucose metabo- randomization to investigate whether relation to insulin resistance measures lism remain incompletely understood. observed associations of sCD14 with in- using Mendelian randomization, which CD14, a pattern-recognition recep- sulin resistance may have a causal basis. were stratified by race, genotypic data tor that has specificity for lipopolysac- were available in 3,387 white and 422 charides (LPSs) and other bacterial African American participants. wall–derived components, is expressed RESEARCH DESIGN AND METHODS primarily by myeloid cells but also by a Study Population Inflammatory Measures range of nonmyeloid tissues (7). The The design and rationale of CHS have sCD14 was measured in baseline plasma receptor is anchored to the plasma mem- been described before (18). Briefly, CHS specimens using a commercial ELISA brane by a glycosylphosphatidylinositol is a prospective population-based cohort (R&D Systems, Minneapolis, MN) with moiety but is also detectable in plasma study of older adult men and women an interassay coefficient of variation of as a soluble form lacking this membrane- recruited from four U.S. field centers: 5.3–12.4% (20). CRP was measured using anchoring component (7). Such soluble Forsyth County, NC; Sacramento County, an ultra-sensitive ELISA developed at the (s)CD14 mainly arises from monocytes/ CA; Washington County, MD; and Pitts- CHS Central Laboratory at the University macrophages following cleavage of the burgh, PA. Candidates were randomly of Vermont with an interassay coefficient membrane receptor by proteases or from sampled from Medicare eligibility lists, of variation of 5.5% (20). IL-6 was mea- direct secretion of the anchor-free form and those who were institutionalized, sured using commercial ELISA kits (Quan- by hepatocytes and adipocytes (8). wheelchair-bound, in hospice, receiving tikine IL-6; R&D Systems, Minneapolis, Experimental data suggest that the treatment for cancer, not expected to MN) with an interassay coefficient of composition of the gut microbiome remain in the area for the next 3 years, or variation of 6.3%. Complete blood counts and translocation of bacterial compo- unable to give informed consent were (including WBC counts) were measured nents, namely LPSs, contribute impor- excluded. A total of 5,201 participants at each of the participating clinical cen- tantly to obesity-related inflammation, (original cohort) were recruited in ters using the following instruments: so-called metabolic endotoxemia (9,10). 1988–1989 from random samples of Coulter Stack S cell counter (Coulter, As a key factor in signal transduction of Medicare eligibility lists. An additional Inc., Hialeah, FL) (University of Pitts- LPS-related proinflammatory cascades 687 primarily black participants were burgh, Pittsburgh, PA, and University by Toll-like receptors, CD14 has been recruited in 1992–1993 (supplemen- of California, Davis, Davis, CA) and the shown to play a pivotal role in this pro- tal cohort) for a total cohort of 5,888 Sysmex NE8000 counter (Toa Electronics, cess (11,12). sCD14 is an acute-phase individuals. During in-person visits, par- Inc., Chicago, IL) (Wake Forest University, reactant, but it can have differing effects ticipants underwent evaluation for de- Winston-Salem, NC, and Johns Hopkins on LPS-related inflammatory signaling mographic and lifestyle