Validity of Postmortem Glycated Hemoglobin to Determine Status of Diabetes Mellitus in Corneal Donors

CLINICAL SCIENCE

Mark C. Soper, BS, CEBT,* Santica M. Marcovina, PhD, DSc,† Caroline K. Hoover, MBA, CEBT,‡ Peter M. Calhoun, MA,§ Kristen E. McCoy, CEBT, CTBS,¶ Christopher G. Stoeger, MBA, CEBT,k Gregory A. Schmidt, BS, CEBT,** Baha M. Arafah, MD,†† Marianne O. Price, PhD,‡‡ Loretta B. Szczotka-Flynn, OD, PhD,‡‡§§¶¶ and Jonathan H. Lass, MD§§¶¶kk

Key Words: diabetes mellitus, glycated hemoglobin, donor cornea, Purpose: To examine the stability of postmortem glycated postmortem hemoglobin (HbA1c) measurement and its relationship to premortem glycemia. (Cornea 2017;36:942–947) Methods: Postmortem blood samples were obtained from 32 donors (8 known diabetic) and shipped on ice to a central he use of donor corneas from individuals with diabetes laboratory to examine the stability of HbA1c measurements Thas long been considered a risk that could impair graft during the first 9 postmortem days. Thirty-nine other suspected success and increase endothelial cell loss after keratoplasty, diabetic donors underwent comparison of premortem and post- and the acceptance of tissue from these donors has been at the mortem HbA1c measurements. discretion of the surgeon. Many surgeons are requesting not Results: Postmortem HbA1c measurements remained stable after 9 to obtain tissue from donors with diabetes, particularly for 6 Descemet membrane endothelial keratoplasty (DMEK) postmortem days (all measurements within 0.2% from baseline fi with a mean difference of 0.02% 6 0.10%). Of the premortem because it is more dif cult and potentially traumatic to 1–3 measurements obtained within 90 days before death, 79% were prepare diabetic tissue, causing concern about graft 6 survival. An attempt to address loss of donor tissue during within 1.0% of the postmortem measurements compared with 40% fi for measurements more than 90 days apart. Three of the postmortem DMEK preparation with a classi cation system based on 3 HbA1c measurements exceeded 6.5% (considered a threshold for diabetes severity and complications has been described. diabetes diagnosis), although the medical histories did not indicate However, this approach has not been widely applied and does any previous diabetes diagnosis. not include the premortem or postmortem glycated hemoglobin (HbA1c) measurement, which is indicative of the average Conclusions: Postmortem HbA1c testing is feasible with current blood glucose level over the previous 2 to 3 months.4 An eye bank procedures and is reflective of glycemic control of donors HbA1c level $6.5% is considered diagnostic for diabetes.4 during 90 days before death. HbA1c testing could potentially be In 2015, endothelial keratoplasty comprised 57% a useful adjunct to review of the medical history and records for (27,208 of 47,580) of all keratoplasty procedures performed donor assessment for endothelial keratoplasty suitability and long- in the United States.5 Descemet stripping endothelial kerato- term graft success. plasty (DSEK) remained the most common endothelial keratoplasty technique (82%), whereas DMEK was 18% of procedures but growing more rapidly than DSEK. Thus, the Received for publication January 13, 2017; revision received February 27, 2017; accepted March 1, 2017. Published online ahead of print May 23, assessment of diabetes duration and severity in corneal tissue 2017. donors is of increasing importance. From the *Indiana Lions Eye Bank, Indianapolis, IN; †Department of Prevalence of diabetes is increasing in the United States Medicine, University of Washington and the Northwest Lipid Metabolism and worldwide,4 with a parallel increase in the proportion of ‡ and Diabetes Research Laboratories, Seattle, WA; SightLife, Seattle, WA; corneas from diabetic donors that meet the Eye Bank Association §Jaeb Center for Health Research, Tampa, FL; ¶Eversight Illinois, Chicago, 6 IL; kLions VisionGift, Portland, OR; **Iowa Lions Eye Bank, Iowa City, of America suitability guidelines for corneal transplantation. IA; ††Department of Medicine, University Hospitals Cleveland Medical Although the Eye Bank Association of America does not report Center and Case Western Reserve University, Cleveland, OH; ‡‡Cornea the percentage of diabetic donors, 18% of the donors in the Research Foundation of America, Indianapolis, IN; §§University Hospitals Cornea Donor Study (CDS) obtained between 2000 and 2002 Eye Institute, Cleveland, OH; ¶¶Department of Ophthalmology and Visual 7 Sciences, Case Western Reserve University, Cleveland, OH; and had a history of diabetes. By 2015, data from eye banks kkEversight Ohio, Cleveland, OH. supported by Midwire (Ann Arbor, MI), a large software Supported by Research to Prevent Blindness, Inc. (J.H.L.). network for donor distribution, found that 4306 of 11,834 The authors have no conflicts of interest to disclose. (36%) donors recovered for keratoplasty had a history of diabetes Reprints: Jonathan H. Lass, MD, University Hospitals Cleveland Medical ’ Center, 11100 Euclid Avenue, Cleveland, OH 44106 (e-mail: jonathan. (personal communication, Michael O Keefe), doubling relative to [email protected]). the CDS from data obtained 15 years previously. Similarly, Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved. between 2012 and 2015, 8552 of 27,948 (31%) donors from

942 | www.corneajrnl.com Cornea Volume 36, Number 8, August 2017

Copyright Ó 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. Cornea Volume 36, Number 8, August 2017 Validity of Postmortem HbA1c in Corneal Donors

SightLife were diabetic as determined historically.8 The percent- identify donors with this information, each of the participat- age of donors with diabetes may in fact be significantly ing eye banks prospectively reviewed the donor history for underreported because the donor medical history form does not HbA1c measurement, as well as body mass index, diagnosis have a specific field for recording a history of diabetes and data and duration of diabetes, diabetes medications, multiple are retrieved from a write-in field. The type of diabetes, duration, elevated blood glucose levels, and possible comorbidities treatment, and complications also are not captured routinely by associated with diabetes systemically (eg, neuropathy, all eye banks. Finally, it is estimated that approximately 28% of nephropathy, peripheral vascular disease) and ocularly (eg, individuals with diabetes are undiagnosed.4 pan or focal retinal photocoagulation, vitrectomy). Both With an increasing pool of donors with diabetes and the inpatient and outpatient records (when available) were growth of DMEK, a better approach to determine diabetes in reviewed for HbA1c measurements up to 1 year before death. the donor and define the best quality tissue for its intended After this review, the most recent HbA1c measurement purpose is of significant interest. This study is the first effort before death was collected by 5 eye banks (Eversight Illinois to examine the practical determination of glycated hemoglo- and Ohio, Indiana Lions Eye Bank, Iowa Lions Eye Bank, bin postmortem, whether the value is stable with shipment SightLife, Lions VisionGift). In the record review, the greatest and testing, and how the postmortem measurement compares effort was to determine whether a measurement was obtained with the premortem measurement. within 90 to 120 days premortem, considering that the life span and turnover of the red blood cells range between 90 and 120 days.9 The measurement and the date on which it was obtained MATERIALS AND METHODS were recorded. If the initial history and chart review were This study is exempted from institutional review board suitable for the study, and if death to preservation time was in monitoring, as determined by the University Hospitals range (up to 23 hours if refrigerated within 10 hours of death Cleveland Medical Center’s Institutional Review Board, and up to 12 hours if not refrigerated), usual eye bank because the study material was obtained postmortem as part procedures were followed in donor and tissue evaluation for of the anatomical donation process. Eye banks participating in tissue recovery including drawing of blood for usual serologies the study obtained appropriate authorization for blood sample and a blood sample drawn for HbA1c testing at the NLMDRL, collection and testing for research purposes. Information with shipping and testing as described above. analyzed as part of the study was obtained as part of the routine eye bank medical screening process. HbA1c Measurement HbA1c Stability Study HbA1c analysis was performed at the NLMDRL by a high-performance liquid chromatography method on the To examine the stability of HbA1c testing under normal Tosoh G8 autoanalyzer (Tosoh Bioscience, S. San Francisco, eye banking procedures, additional blood samples were obtained CA). The NLMDRL is a level 1 certified laboratory by the at the time of recovery. Thirty-two donors from 2 eye banks National Glycosylation Standardization Program. (SightLife, Seattle, WA; Indiana Lions Eye Bank, Indianapolis, IN) were evaluated for suitability for keratoplasty. An additional 20 mL of blood was drawn within 20 hours postmortem where Statistical Analysis the donor was refrigerated within 10 hours or blood was drawn For the assessment of postmortem HbA1c stability, within 10 hours postmortem if the body was not refrigerated. each measurement was paired with its baseline measurement. The blood was injected into 10-mL purple top tubes (Becton The difference and absolute difference between HbA1c Dickinson, Franklin Lakes, NJ) containing ethylenediaminetetra measurements for each pair were recorded, and mean and acetic acid (EDTA) and then inverted 8 to 10 times for mixing. SD were reported. For the comparison of the premortem and Each blood specimen was then placed on ice, refrigerated at postmortem HbA1c measurements, the most recent premor- 4°C, and either transferred (SightLife) or shipped on ice tem and postmortem measurements were compared and the (Indiana) to the Northwest Lipid Metabolism and Diabetes difference, absolute difference, relative difference, and rela- Research Laboratory (NLMDRL) in Seattle, WA. Days from tive absolute difference were recorded. Because of the small draw to receipt at the NLMDRL varied from less than 1 to sample sizes, no formal testing or confidence intervals were 7 days because of immediate transfer within the city (Seattle) or implemented for this study. Analyses were performed using bulk shipping (Indiana). Once received at the NLMDRL, the SAS 9.4 (SAS Institute, Inc, Cary, NC). HbA1c measurement was determined on each specimen daily, except on weekends, to 9 days after receipt. During this period, the specimens continued to be stored at 4°C. RESULTS HbA1c Stability Study Comparison Study of Premortem and There were 32 donors with at least one pair of postmortem Postmortem HbA1c Measurements and HbA1c measurements. Mean death to refrigeration hours (n = Donor Evaluation 31), hours refrigerated (n = 31), and hours postmortem (n = 32) Eye banks do not normally retrieve premortem HbA1c were 5.0 6 3.9, 5.8 6 3.9, and 11.3 6 4.2 hours, respectively. measurements during routine review of the donor history. To Based on the review of the inpatient record and interviewing

Copyright Ó 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. Soper et al Cornea Volume 36, Number 8, August 2017 relatives, 24 of the 32 donors were classified not to have a blood sample to determine a postmortem HbA1c and that diabetes. However, 3 of these 24 donors subsequently were this sample, if properly refrigerated, can be shipped with the determined to have had diabetes based on an HbA1c greater measurement being stable up to 9 days after the sample was than 6.5% postmortem (measurements 6.7, 6.9, and obtained. This procedure could then be included in normal 6.7, respectively). serology testing as part of the corneal donor assessment. Figure 1 gives the mean difference between the first When coupled with a planned determination of diabetes postmortem HbA1c measurement and each following post- premortem including documentation of comorbidities related mortem HbA1c measurement. The mean absolute difference did to diabetes (eg, neuropathy, nephropathy, peripheral vascular not change over the 9 days ranging from 0.02% to 0.08% (Table disease, intraocular complications), the HbA1c measurement 1). After 9 days, the differences were always within 60.2% with could be used to provide additional information to influence a mean difference of 0.02% 6 0.10%. It should be noted that the donor selection for DSEK and/or DMEK according to the stability results did not vary based on whether the specimen was recently proposed classification of donors with diabetes.3 received within 24 hours of the blood draw (n = 15, SightLife) Another important question addressed was the relation or up to 7 days after the blood draw (n = 17, Indiana Lions). of the postmortem HbA1c to the premortem HbA1c, given the potential impact from possible comorbidities (microvas- cular and macrovascular complications) surrounding death. Comparison Study of Premortem and Reassuringly, the median absolute difference between the 2 Postmortem HbA1c Measurements and values was remarkably only 0.5% as long as the premortem Donor Evaluation measurement was obtained within 90 days of death. For Figure 2 gives the scatterplot of the premortem and measurements within 90 days, 79% of pairs were within postmortem HbA1c measurements for donors who had paired 61.0% compared with 40% for measurements more than 90 data. The mean difference for the 39 donors was 20.23% 6 days apart. This is consistent with the life of the red blood 1.51% (Table 2). The median absolute difference was smaller cell9 and unless there have been massive transfusions before for measurements within 90 days compared with measurements death, there should be a good correlation. more than 90 days apart (0.5% vs. 1.1%). For measurements To our knowledge, this is the first report of determina- within 90 days, 79% of pairs were within 61.0% compared tion of HbA1c postmortem in the eye donor population. The with 40% for measurements more than 90 days apart. forensic literature has described its measurement primarily to determine the cause of death related to diabetes, in particular as it relates to ketoacidosis.10–12 Our study confirms that DISCUSSION postmortem HbA1c is representative of the glycemic condi- Our study has shown that the eye bank technician can tion within several months of death and that the test values practically include as part of their normal procedure drawing were stable. None of the forensic studies compared

FIGURE 1. Mean 6 SD differences between initial and subsequent postmortem HbA1c measurements as a function of the time between the measurements (up to 9 days).

Copyright Ó 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. Cornea Volume 36, Number 8, August 2017 Validity of Postmortem HbA1c in Corneal Donors

TABLE 1. Change in HbA1c During First 9 Postmortem Days % of Pairs With AD HbA1c Time, h* No. of Pairs Absolute Difference (AD), Mean 6 SD Identical Within 0.1% Within 0.2% Within 0.3% 24 27 0.07 6 0.09 52 89 93 100 36 23 0.06 6 0.08 61 87 96 100 48 18 0.02 6 0.05 89 94 100 100 72 9 0.04 6 0.05 56 100 100 100 96 19 0.07 6 0.10 53 84 95 95 120 32 0.06 6 0.08 56 84 100 100 144 32 0.08 6 0.09 44 88 97 97 168 21 0.05 6 0.07 57 90 100 100 196 19 0.04 6 0.07 68 89 100 100 216 16 0.06 6 0.08 56 81 100 100

*The HbA1c measured at each target time is compared with its baseline HbA1c value. Measurements were taken daily, except on weekends, which accounts for the varying number of pairs. premortem with postmortem HbA1c measurements based on a biomarker for long-term donor tissue performance. Diabetes the timing of premortem testing. Our findings, thus, could be has pervasive effects on the cornea including impaired wound valuable for the forensic pathologists as much as for eye healing, recurrent erosions, and diminished corneal sensa- banks. It should be noted, however, that events preceding tion.13 In particular, the corneal endothelium differs death may result in a premortem and postmortem glycemic functionally, morphometrically, and in response to cataract level that differs from the typical level over the course of the surgery.14–16 In patients with type 2 diabetes, a higher HbA1c individual’s lifetime and thus may not accurately reflect the is associated with lower endothelial cell density (ECD).15 potential risk to the donor tissue acquired over the preceding However, in a retrospective study of 27,948 donors recovered decades. Therefore, the medical records and history remain of and evaluated between 2012 and 2015, 8552 had a history of key importance in the donor assessment process. diabetes, and donor diabetes status was not significantly HbA1c testing may be of adjunctive value in predicting associated with technician-induced endothelial damage, trans- short-term DMEK preparation issues and could be used as plant suitability, or central ECD.8 Similarly in another

FIGURE 2. Scatterplot of premortem and postmortem HbA1c measurements. Color of point rep- resents the number of days between premortem and postmortem HbA1c measurements.

Copyright Ó 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. Soper et al Cornea Volume 36, Number 8, August 2017

TABLE 2. donors would be of particular concern when being transplanted Comparison Between Premortem and Postmortem into a diabetic recipient. HbA1c The principal study limitation was the sample size. Within More Than Metric Overall 90 d Apart 90 d Apart Although the HbA1c stability study was conclusive with a relatively small sample size, a larger sample would more No. of subjects 39 24 15 strongly address the relationship of the premortem and post- Difference mortem HbA1c measurements within 90 days of each other. Mean 6 SD 20.23 6 1.51 20.01 6 0.97 20.58 6 2.11 In summary, postmortem HbA1c testing is feasible Min, max 23.3, 6.1 22.5, 2.2 23.3, 6.1 with current eye bank procedures, and test results are Absolute difference stable more than a week postmortem if the sample is (AD) appropriately shipped and stored. The postmortem HbA1c Median (Q1, Q3) 0.7 (0.3, 1.2) 0.5 (0.2, 0.8) 1.1 (0.6, 1.9) potentially adds important information to understanding % of pairs with AD the diabetic state of the donor, especially when combined Identical 5 8 0 with an additional medical history. The utility of this Within 0.2% 23 29 13 measurement to predict DMEK preparation success and Within 0.5% 44 58 20 long-term keratoplasty success coupled with usual param- Within 1.0% 64 79 40 eters for donor tissue evaluation (slit-lamp evaluation, Within 1.5% 82 88 73 central ECD) remains to be determined. Relative difference (RD), % Median (Q1, Q3) 24(213, 4) 0 (27, 6) 212 (218, 1) Relative absolute ACKNOWLEDGMENTS difference (RAD), % The authors acknowledge Michael Payne MS from Median (Q1, Q3) 9 (4, 18) 7 (2, 12) 13 (8, 21) the UH Eye Institute/CWRU Vision Research Coordinat- ing Center for coordination and data collection, Jessica Harting from the NLMDRL for her coordination of HbA1c retrospective study examining 20,026 nondiabetic and 13,617 sample acquisition and analyses, and all the eye bank diabetic donor eyes historically determined and recovered technicians and referral center staff from the contributing between 2007 and 2014, the diabetic donors had a slightly eye banks (Eversight, Indiana Lions Eye Bank, Iowa Lions lower ECD (229 cells/mm2), statistically significant but Eye Bank, Lions VisionGift, and SightLife) for their clinically not important.17 execution of the protocol. In turn, retrospective analysis of clinical results after REFERENCES keratoplasty has not detected an adverse diabetes effect. In the CDS data, a history of donor diabetes was not associated 1. Greiner MA, Rixen JJ, Wagoner MD, et al. Diabetes mellitus increases risk of unsuccessful graft preparation in Descemet Membrane Endothelial with an increased risk of penetrating keratoplasty (PKP) Keratoplasty: a multicenter study. Cornea. 2014;33:1129–1133. failure through 10 years.7 No difference in endothelial cell 2. Vianna LMM, Stoeger CG, Galloway JD, et al. Risk factors for eye bank loss was noted in the 2 groups in this same study. Likewise, preparation failure of Descemet Membrane Endothelial Keratoplasty a 2-center retrospective study of 183 eyes treated with (DMEK) tissue. Am J Ophthalmol. 2015;159:829–834. fi 3. Williams RS, Mayko ZM, Friend DJ, et al. Descemet Membrane DSEK or PKP did not nd a history of donor diabetes to be Endothelial Keratoplasty (DMEK) tissue preparation: a donor diabetes 18 associated with increased risk of graft failure. Although mellitus categorical risk stratification scale for assessing tissue suitability these studies did not show a long-term effect of donor and reducing tissue loss. Cornea. 2016;35:927–931. diabetes on keratoplasty success, key limitations included 4. American Diabetes Association. Diagnosing Diabetes and Learning underestimation of diabetes prevalence and severity in the about Pre-diabetes, and Statistics About Diabetes. Available at: http:// www.diabetes.org/diabetes-basics/diagnosis/ and http://www.diabetes.org/ control populations based on the limited medical history; diabetes-basics/statistics/. Accessed February 1, 2017. lack of any information regarding the type, duration, and 5. Eye Bank Association of America. 2015 Eye Banking Statistical Report. level of glycemic control; and lack of randomization based Available at: http://restoresight.org/wp-content/uploads/2016/03/2015- on diabetes status. Interestingly, Price et al19 did find an Statistical-Report.pdf. Accessed December 21, 2016. 6. Eye Bank Association of America. Medical standards. Int J Eye Banking. effect of recipient diabetes on the rate of PKP endothelial 2016;4:1–39. decompensation (in the absence of rejection) at 5 years with the 7. Lass JH, Riddlesworth TD, Gal RL, et al. The effect of donor diabetes diabetic group at 5.7% versus the nondiabetic group at 2.4%. A history on graft failure and endothelial cell density 10 years after prospective study examining the relation of the postmortem penetrating keratoplasty. Ophthalmology. 2015;122:448–456. HbA1c level and the presence of comorbid diabetic complica- 8. Margo JA, Munir WM, Brown CH, et al. Association between endothelial cell density and transplant suitability of corneal tissue tions in the donor, as well as recipient diabetes status on long- with type 1 and type 2 diabetes. JAMA Ophthalmol. 2017 [epub term keratoplasty (PKP, DSEK, and DMEK) success and ahead of print]. endothelial cell loss is most needed. Most diabetic donors may 9. Peterson KP, Pavlovich JG, Goldstein D, et al. What is hemoglobin A1c? be successfully used, but there may be a subset of donors with An analysis of glycated hemoglobins by electrospray ionization mass spectrometry. Clin Chem. 1998;44:1951–1958. high HbA1c measurements, perhaps over 10%, associated with 10. John WG, Scott KW, Hawcroft DM. Glycated haemoglobin and glycated significant diabetic complications that may result in a higher protein and glucose concentrations in necropsy blood samples. J Clin graft failure rate and should not be used as donors. These Pathol. 1988;41:415–418.

Copyright Ó 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. Cornea Volume 36, Number 8, August 2017 Validity of Postmortem HbA1c in Corneal Donors

11. Palmiere C, Bardy D, Mangin P, et al. Postmortem diagnosis of 16. Urban B, Peczynska J, Glowinska-Olszewska B, et al. Evaluation of unsuspected diabetes mellitus. Forensic Sci Int. 2013;226:160–167. central corneal thickness in children and adolescents with type I diabetes 12. Winecker RE, Hammett-Stabler CA, Chapman JF, et al. HbA1c as a post- mellitus [in Polish]. Klin Oczna. 2007;109:418–420. mortem tool to identify glycemic control. JForensicSci.2002;47:1373–1379. 17. Chen Y, Tsao SW, Heo M, et al. Age-stratiﬁed analysis of diabetes and 13. Lutty GA. Effects of diabetes on the eye. Invest Ophthalmol Vis Sci. pseudophakia effects on corneal endothelial cell density: a retrospective 2013;54:ORSF81–ORSF87. eye bank study. Cornea. 2017;36:367–371. 14. Dhasmana R, Singh IP, Nagpal RC. Corneal changes in diabetic patients 18. Vislisel JM, Liaboe CA, Wagoner MD, et al. Graft survival of diabetic after manual small incision cataract surgery. J Clin Diagn Res. 2014;8: versus nondiabetic donor tissue after initial keratoplasty. Cornea. 2015; VC03–VC06. 34:370–374. 15. Storr-Paulsen A, Singh A, Jeppesen H, et al. Corneal endothelial 19. Price MO, Thompson RW Jr, Price FW Jr. Risk factors for various morphology and central thickness in patients with type II diabetes causes of failure in initial corneal grafts. Arch Ophthalmol. 2003;121: mellitus. Acta Ophthalmol. 2014;92:158–160. 1087–1092.