Predicting non-diabetic renal disease in typvalue of glycated hemoglobin

ld

UKacktal, B

a r t i c l e i n f o

Article history:Received 30 October 2014Received in revised form 28 November 2014

rch Group, 1993; UK98).biochemical analysespersistent microalbu-

minuria. Estimated glomerular ltration rate (eGFR) declines prior tomore severe macroalbuminuria in type 2 diabetes (T2D); hence acombination of eGFR and albuminuria can be used to stage andmonitor patients (American Diabetes Association, 2013; Kidney

Journal of Diabetes and Its Complications xxx (2015) xxxxxx

5SS Birmingham, UK. Tel.: +44 1214242158; fax: +44 1214241159.E-mail addresses: maria.pallayova@upjs.sk (M. Pallayova),

Contents lists available at ScienceDirect

Journal of Diabetes an

j ourna l homepage: WWWazharuddin@doctors.org.uk (A. Mohammed), gerald.langman@heartofengland.nhs.uk(G. Langman), staheri@me.com (S. Taheri), indranil.dasgupta@heartofengland.nhs.ukDiabetes Control & Complications Trial ReseaProspective Diabetes Study (UKPDS) Group, 19

The diagnosis of DN is usually made throughof urine and blood. An early manifestation is

Conict of Interest: MP has received grant/research support from Slovakian DiabetesAssociation/Lilly Diabetes Clinical Research Initiative. AM, GL, ST and ID declare that theyhave no competing interests. Corresponding author at: Renal Unit Heartlands Hospital Bordesley Green East B91. Introduction

Diabetic nephropathy (DN) is a major microvascular complicationof diabetes mellitus associated with end-stage renal disease requiringrenal replacement therapy. A major contributor to development andprogression of DN is glycemic control as shown by major diabetes

studies (Stratton, Adler, Neil, et al., 2000; The MicroalbuminuriaCollaborative Study Group, 1999). Other modiable factors for DNinclude hypertension, obesity, smoking, and dyslipidemia (Gross etal., 2005). There is encouraging evidence suggesting that timely andlong-term tight glycemic control effectively delays the onset andslows the progression of DN in both type 1 and type 2 diabetes (The(I. Dasgupta).1 The permanent address of Maria Pallayova: Depar

Faculty of Medicine, Pavol Jozef Safarik University, Kosic2 The present address of Azharuddin Mohammed: Ren

Uttoxeter Road, Derby, DE22 3NE, UK.3 The present address of Shahrad Taheri: Departme

Medical College in Qatar, PO Box 24144, Doha, Qatar.

http://dx.doi.org/10.1016/j.jdiacomp.2014.12.0051056-8727/ 2014 Elsevier Inc. All rights reserved.

Please cite this article as: Pallayova, M.,hemoglobin, Journal of Diabetes and Its Comdeterioration inHbA1cprior tobiopsypredictedNDRD inT2D. Further studies are needed tovalidate thendings. 2014 Elsevier Inc. All rights reserved.Accepted 8 December 2014Available online xxxx

Keywords:Glycated hemoglobinRenal biopsyRenal diseaseType 2 diabetes mellitusNon-diabetic renal disease

51) over 10 years for nephrotic-range proteinuria, microscopic hematuria, or rapidly declining renal function.Results: Thirty-ve (68.6%) biopsies were diagnostic of NDRD, and 16 (31.4%) revealed isolated diabeticnephropathy. The most common NDRDs were interstitial nephritis (20%), progressive crescentic glomerulo-nephritis (14%), membranous nephropathy (11%), and focal segmental glomerulosclerosis (11%). The odds forNDRDdeclined by 97% in the presence of diabetic retinopathy (P b 0.001). The deterioration of HbA1c during theyear before biopsy predicted NDRD even after adjusting for diabetic retinopathy (OR, 7.65; 95% CI, 1.36123.04;P = 0.003). A model based on the interaction between the HbA1c values 12 months before biopsy and theabsolute change in these values during the preceding year predicted NDRD with 73.7% sensitivity and 75%specicity (AUC, 0.77; 95% CI, 0.590.94).Conclusions: This study demonstrated a considerably high prevalence of NDRD in T2D adults undergoing renalbiopsy. The absence of diabetic retinopathy, lower HbA1c values 12 months before biopsy and greatera b s t r a c t

Aims: The indications for renal biopsy in type 2 diabetesmellitus (T2D) are notwell established.We investigatedthe prevalence, spectrum, and predictors of biopsy-proven non-diabetic renal disease (NDRD) in T2D.Methods:An observational, single-center, retrospective study of T2D adults who underwent renal biopsies (N =Maria Pallayova a,1, Azharuddin Mohammed a,2, GeraShahrad Taheri c,d, Indranil Dasgupta a,a Renal Unit, Heartlands Hospital, Bordesley Green East, Birmingham, UKb Department of Histopathology, Heartlands Hospital, Bordesley Green East, Birmingham,c Collaborations for Leadership in Applied Health Research and Care for Birmingham and Bld Diabetes Centre, Heart of England NHS Foundation Trust, Birmingham Heartlands Hospitment of Human Physiology,e, Slovak Republic.al Unit, Royal Derby Hospital,

nt of Medicine, Weill Cornell

et al., Predicting non-diabetplications (2015), http://dx.e 2 diabetic adults: The

Langman b,3,

Country/National Institute for Health Research, University of Birmingham, Birmingham, UKirmingham, UK

d Its Complications

.JDCJOURNAL.COMDisease: Improving Global Outcomes (KDIGO) CKD Work Group,2013). While diabetes is the major cause of renal disease in patientswith diabetes, in about a third of patients, renal dysfunction is due toother causes (non-diabetic renal disease; NDRD) (Chong, Keng, Tan,et al., 2012; Das, Dakshinamurty, Prayaga, & Uppin, 2012; Harada,Akai, Sumida, et al., 2013; Kleinknecht, Bennis, & Altman, 1992; Lee,

ic renal disease in type 2 diabetic adults: The value of glycateddoi.org/10.1016/j.jdiacomp.2014.12.005

2 M. Pallayova et al. / Journal of Diabetes and Its Complications xxx (2015) xxxxxxChung, & Choi, 1999; Mazzucco, Bertani, Fortunato, et al., 2002; Olsen& Mogensen, 1996; Richards et al., 1992; Soni, Gowrishankar, Kishan,& Raman, 2006; Yaqub, Kashif, & Hussain, 2012). In the NDRD group,the treatment of renal disease may require a different strategy. A renalbiopsy is helpful indetermining theunderlyingpathophysiology inNDRD.

The selection criteria for renal biopsy in diabetic patients are notwell established. In type 1 diabetes, the presence of proteinuria withshort diabetes duration and/or rapidly declining renal function,especially in the absence of diabetic retinopathy, has been suggestedas a signal for the need for renal biopsy (Mauer, Fioretto,Woredekal, et al.,2001). In T2D, the criteria are less clear since dysglycemia is present formany years prior to diagnosis. Commonly, proteinuria N1 g/24 hours,renal involvement without diabetic retinopathy, or unexplainedhematuria has been used as indicators for renal biopsy (Wong, Choi,Szeto, et al., 2002). Identication of novel predictors of renal diseasewillimprove the current selection criteria for renal biopsy and facilitate earlydetection of NDRD in T2D. Early diagnosis and appropriate treatmentmay help slow progression to end stage renal disease. In this study, wesought to investigate the prevalence, spectrum, and predictors ofbiopsy-proven NDRD in adults with T2D.

2. Subjects, material and methods

This was a retrospective observational study of T2D patients whounderwent renal biopsies over 10 years in our center. As this was anaudit of retrospective data, the local research ethics committee feltthat no formal ethics approval was required.

Fifty-one native renal biopsies obtained from 51 adults with adocumented diagnosis of T2D, referred to our center between 2002and 2012, were analysed. In our center, as a policy, DN is diagnosed onclinical grounds and kidney biopsies are only carried out if there areatypical clinical features. Indications for biopsy, in this cohort,included nephrotic range proteinuria (N3 g/24 hours), signicantmicroscopic hematuria ( ++), or rapidly declining renal function.Renal biopsy specimens were examined by light microscopy, directimmunouorescence, and electron microscopy, where indicated.

The biopsy report, biochemical results, and clinical information atthe time of renal biopsy and follow-up were studied. Glomerularltration rate estimates were calculated using the 4-variableModication of Diet in Renal Disease Study equation (Levey et al.,1999). Glycemic control was assessed by glycated hemoglobin(HbA1c) levels, measured using National Glycohemoglobin Standar-dization Program (NGSP) certied method, standardized to theDiabetes Control and Complications Trial assay. HbA1c values arereported in both NGSP percentage units with International Federationof Clinical Chemistry (IFCC) units (mmol/mol) in parentheses.

The primary outcome measure was the prevalence and nature ofhistologically-proven NDRD. The secondary outcome measuresincluded predictors of NDRD vs. DN and the risk factors for adverserenal outcome. Adverse renal outcomes included reaching end stagerenal disease requiring renal replacement therapy or chronic kidneydisease (CKD) leading to death, or a composite of the two.

2.1. Statistical analyses

The ShapiroWilk test was applied to assess normality of datadistribution. Continuous variables with normal distribution arepresented as means standard deviation (SD) and compared usingthe Student's t test. Continuous variables with non-normal distribu-tion are presented as medians and interquartile ranges (IQR) andcompared using the Wilcoxon rank-sum test or the Wilcoxon--matched pairs signed-ranks test. The chi-square test was applied toexamine patterns between categorical variables. Univariate andmultivariate standard and exact logistic regression modeling wereemployed to identify the association between biopsy-proven NDRDand potential predictors. The logistic regression models were ttedPlease cite this article as: Pallayova, M., et al., Predicting non-diabethemoglobin, Journal of Diabetes and Its Complications (2015), http://dx.using a stepwise bidirectional elimination algorithm, with inclusionand exclusion criteria of P 0.15 and P 0.2 respectively.

The number in whom complete data were available is mentioned inthe text. Absolute changes () in eGFR and in HbA1c were calculated bysubtracting values 12 months before biopsy from those at the time ofbiopsy. The follow-up eGFR were calculated by subtracting the eGFRvalues at the time of biopsy from the values 12 months after.

Findings were considered to be statistically signicant at the 5%level. Statistical calculations were performed using Stata 11.2 SpecialEdition (StataCorp LP, College Station, TX).

3. Results

3.1. Sample characteristics and renal biopsy ndings

Theentire cohortof renal biopsypatients (N = 51)waspredominantlymale (64.7%), of mixed ethnicity (28White Europeans, 18 South-Asians, 3African Caribbean, 2 with not stated ethnicity), aged 61 12 years(mean SD). The median (IQR) duration of T2D was 9 years (215 years) (n = 45). At the time of renal biopsy, the NGSP HbA1c was7.2% (6.47.8%), the IFCC HbA1c 55 (4662) mmol/mol (n = 42).Twenty-two of 39 (56.4%) patients had diabetic retinopathy. The median(IQR) eGFR was 23.5 ml/min/1.73 m2 (1139.5 ml/min/1.73 m2), and60.4% of patients had stage 45 CKD. Twenty-one of 36 (58.3%) proteinuricpatients had nephrotic range proteinuria (albumin/creatinine ratioN250 mg/mmol or protein/creatinine ratio N300 mg/mmol), and 16 of35 (45.7%) had microscopic hematuria. Thirty-four of 46 (73.9%) patientswere on reninangiotensinaldosterone system (RAAS) blockade therapy.

Thirty-ve (68.6%) biopsies were diagnostic of NDRD, 16 (31.4%)revealed isolated DN, and 4 (7.8%) showed NDRD superimposed onDN. The spectrum of NDRD was as follows: interstitial nephritis 7(20%), progressive crescentic glomerulonephritis 5 (14.3%), membra-nous nephropathy 4 (11.4%), focal segmental glomerulosclerosis 4(11.4%), acute tubular necrosis 3 (8.6%), immunoglobulin A nephrop-athy 2 (5.7%), ischemic nephropathy 2 (5.7%), minimal changenephropathy 1 (2.9%), minimal change nephropathy + interstitialnephritis 1 (2.9%), mesangiocapillary glomerulonephritis 1 (2.9%),amyloidosis 1 (2.9%), oxalate nephropathy 1 (2.9%), myeloma castnephropathy 1 (2.9%), brillary glomerulonephritis 1 (2.9%), andcollagenobrotic glomerulopathy 1 (2.9%). In more than 50% of NDRD,histology prompted alteration in therapeutic management.

Table 1 shows characteristics of the two subgroups based on renalbiopsy ndings. The patients with NDRD had better glycemic control12 months before biopsy (n = 32), a smaller decrease in HbA1c duringthe year before biopsy (n = 31; Fig. 1a, b), and a lower rate of diabeticretinopathy than thosewith isolated DN. Therewas a trend towardsmoreSouth-Asians in the DN subgroup. We did not nd any difference in eGFRprior to biopsy, at the time of biopsy, and 12 months after biopsy betweenthe two subgroups. Following the therapeutic adjustments based on renalbiopsy ndings, the patients with NDRD had signicantly greaterimprovement in eGFR 12 months after biopsy (n = 38) than patientswith isolated DN (Table 1). Similar numbers received RAAS blockadetherapy in the twosubgroups. Therewerenodifferences indurationofT2D(n = 45) and in pharmacological treatment of diabetes (n = 47; Table 1).

By the end of 2012, overall mortality was 29.4% with no differencebetween the subgroups. There was no difference in number ofpatients requiring renal replacement therapy. However, a compositeadverse renal outcome (reaching end stage renal disease requiringrenal replacement therapy + CKD leading to death; n = 48) washigher in isolated DN subgroup (Table 1).

3.2. Predictors of renal disease and risk factors for adverserenal outcomes

Table 2 presents the results of a univariate exact logistic regressionanalysis for unadjusted (crude) associations between NDRD andic renal disease in type 2 diabetic adults: The value of glycateddoi.org/10.1016/j.jdiacomp.2014.12.005

ND

; +0; +3

)

.8

15)

3M. Pallayova et al. / Journal of Diabetes and Its Complications xxx (2015) xxxxxxTable 1Characteristics of patients (N = 51) according to renal biopsy ndings.

Variable Patients with

Age (years) 63 11Sexfemale (%) 40South-Asian ethnicity (%) 30.3Duration of diabetes (years) 7.5 (113)Diabetes treatmentdiet only (%) 9.4Diabetes treatmentOHA/GLP-1 agonists (%) 53.1Diabetes treatmentinsulin/insulin + OHA (%) 37.5HbA1c at the time of biopsy- NGSP HbA1c (%) 7.2 0.88- IFCC HbA1c (mmol/mol) 55 9.6

HbA1c 6 months before biopsy- NGSP HbA1c (%) 7.4 (6.3-8.0)- IFCC HbA1c (mmol/mol) 57 (4564)

HbA1c 12 months before biopsy- NGSP HbA1c (%) 6.9 (6.5-8.0)- IFCC HbA1c (mmol/mol) 52 (4864)

HbA1c during the year before biopsy- NGSP HbA1c (%) 0.1 (0.5- IFCC HbA1c (mmol/mol) 1.1 (5.5

Diabetic retinopathy (%) 34.6eGFR 12 months before biopsy (ml/min/1.73 m2) 58 (4077)eGFR 6 months before biopsy (ml/min/1.73 m2) 45 (3981)eGFR at the time of biopsy (ml/min/1.73 m2) 16 (1036)eGFR 6 months after biopsy (ml/min/1.73 m2) 33 (1747.5eGFR 12 months after biopsy (ml/min/1.73 m2) 37.8 22eGFR during the year before biopsy (ml/min/1.73 m2) 31.3 25eGFR 12 months after biopsy (ml/min/1.73 m2) 6.8 17.4Urine PCR at the time of biopsy (mg/mmol) 367 (26510RAAS blockade therapy at the time of biopsy (%) 74.2Adverse renal outcome (%) 34.4

18.2independent variables with odds ratios (OR), 95% condence intervals(95% CI), and P-values. For subjects with diabetic retinopathy, theexpected log odds of having NDRD decreased by 3.39, and the odds forNDRD declined by 97% (P b 0.001). The unadjusted OR of theassociation between NDRD and HbA1c 12 months before biopsy was0.60; for every 1% (10.9 mmol/mol) increase in the HbA1c 12 monthsbefore biopsy, the odds for NDRD declined by 40% (P = 0.018). Therewas a trend towards higher odds for NDRD with deterioration ofHbA1c during the year before biopsy (P = 0.054). We also observed atrend towards higher odds for NDRDwith the increasing INTHbA1c (avariable consistent with the interaction between HbA1c 12 monthsbefore biopsy and HbA1c during the year before biopsy); P = 0.069.

The association between NDRD and deterioration of HbA1c duringthe year before biopsy remained signicant even after adjusting fordiabetic retinopathy in the nal multivariate regression model (OR7.65; 95% CI, 1.36 to 123.04; P = 0.003). Our results further showed astatistically signicant joint distribution of diabetic retinopathy andINTHbA1c. After adjusting for the diabetic retinopathy, the expectedodds for NDRD increased by 1.28 for every one unit (1%) increase inINTHbA1c (OR 1.28; 95% CI, 1.03 to 1.82; P = 0.013).

Fig. 2 shows estimated predicted probabilities of having NDRD,by the HbA1c 12 months before biopsy, ranging from an appro-ximately 85% probability of having NDRD in patients with arelatively low preceding HbA1c to a b50% probability in patientswith HbA1c of N8.5% (69 mmol/mol). Of importance, the patientswho experienced deterioration in HbA1c during the year prior tobiopsy had at least 63% probability of having NDRD (Fig. 2, plussymbols). The HbA1c 12 months before biopsy in these patientswas b8% (64 mmol/mol).

- Renal replacement therapy after biopsy (%)- Mortality (%) 22.9- the time until death (months) 29 22.9

Continuous variables with normal distribution are presented as means SD. Continuousnephropathy; eGFR, glomerular ltration rate estimates; eGFR, an absolute change in eGFchange in HbA1c; IFCC, International Federation of Clinical Chemistry; NDRD, non-diabetichypoglycemic agents; PCR, protein to creatinine ratio; RAAS, reninangiotensinaldosterone

Please cite this article as: Pallayova, M., et al., Predicting non-diabethemoglobin, Journal of Diabetes and Its Complications (2015), http://dx.RD (n = 35) Patients with isolated DN (n = 16) P

58 14 0.16725 0.29861.5 0.0519 (419) 0.4336.7 0.75766.7 0.38226.7 0.465

7.7 1.7661 19.2 0.188

8.9 (6.6-12.2)74 (48110) 0.288

9 (7.2-10.6)75 (5592) 0.047

.3) 0.9 (1.75; 0.3)

.3) 9.8 (19.1; 3.3) 0.023100 b0.00155 (4663) 0.64841 (2856) 0.24429 (1749) 0.15828.5 (1836) 0.55032.1 24.4 0.31821.1 16 0.2337.1 16.4 0.015441 (130.5-871) 0.68573.3 0.95068.8 0.02425 0.579To further dene the importance of HbA1c in predicting the NDRD,we performed receiver operating characteristic (ROC) curve analysisof three HbA1c-based predictive models for NDRD (Fig. 3). In the rstmodel, the HbA1c 12 months before biopsy was used. The cut point ofthe HbA1c 12 months before biopsy of6.2% (44 mmol/mol) was thebest predictor of NDRD with 100% sensitivity. However, the 17%specicity reected high rate (83%) of false positives (area under thecurve AUC, 0.29; 95% CI, 0.07 to 0.51). In the second model, wemeasured predictive accuracy of the HbA1c during the year priorto biopsy. The HbA1c during the year prior to biopsy of 0.4%(4.4 mmol/mol) was the best predictor of NDRD (73.7% sensitivity,66.7% specicity; AUC, 0.75; 95% CI, 0.56 to 0.93). Since the precedingHbA1c and the HbA1c during the year before biopsy are closelyinterrelated, the third predictive ROC curve model included aninteraction term between the two aforementioned predictors(INTHbA1c; INTHbA1c = HbA1c 12 months before biopsy * HbA1cduring the year before biopsy). The ROC curve analysis showed that theINTHbA1c cut-off of 2.48% (27.1 mmol/mol) was the bestpredictor of NDRD with 73.7% sensitivity and 75% specicity (AUC,0.77; 95% CI, 0.59 to 0.94; Fig. 3).

4. Discussion

There were three major ndings in the present study. First, renalbiopsies with histological conrmation of renal involvement revealeda considerably high prevalence of NDRD in a cohort of T2D patientswith signicant dipstick hematuria, nephrotic range proteinuria, and/or rapidly declining renal function. Second, this study conrmed thepreviously recognized associations between poor long-term glucose

43.8 0.129

24 27.4 0.738

variables with non-normal distributions are presented as medians (IQR). DN, diabeticR; GLP-1, glucagon-like peptide-1; HbA1c, glycated hemoglobin; HbA1c, an absoluterenal disease; NGSP, National Glycohemoglobin Standardization Program; OHA, oralsystem.

ic renal disease in type 2 diabetic adults: The value of glycateddoi.org/10.1016/j.jdiacomp.2014.12.005

Table 2Unadjusted (crude) associations between non-diabetic renal disease and inde-pendent variables.

Variable P-value OR 95% CI

Age 0.171 1.04 0.99 to 1.09Sex female 1.00 (reference) male 0.474 0.51 0.10 to 2.14

Ethnicity White European 1.00 (reference) South-Asian 0.107 0.28 0.06 to 1.25

Duration of diabetes 0.493 0.97 0.89 to 1.06Diabetes treatment Diet only 1.00 (reference) OHA/GLP-1 agonists 0.607 0.53 0.05 to 5.86 insulin/insulin + OHA 0.950 1.08 0.09 to 13.54

HbA1c 12 months before biopsy 0.018 0.60 0.35 to 0.93HbA1c at the time of biopsy 0.200 0.71 0.41 to 1.19HbA1c during the year before biopsy 0.054 1.89 0.99 to 4.46INTHbA1c 0.069 1.06 1.00 to 1.14Diabetic retinopathy b0.001 0.03 0.00 to 0.24eGFR 12 months before biopsy 0.401 1.02 0.98 to 1.06eGFR at the time of biopsy 0.374 0.99 0.96 to 1.02eGFR during the year before biopsy 0.211 0.98 0.94 to 1.01Microscopic hematuria 0.116 4.86 0.75 to 56.13Nephrotic proteinuria 1.000 1.02 0.19 to 5.17

4 M. Pallayova et al. / Journal of Diabetes and Its Complications xxx (2015) xxxxxx

N

GSP

HbA

1c [%

]

P=0.047

N=12 N=8 N=15

P=0.02

N=20 N=19 N=2756

78

910

1112

1314

15Isolated diabetic nephropathy Non-diabetic renal disease

HbA1c 12 months before renal biopsyHbA1c 6 months before renal biopsyHbA1c at the time of renal biopsy

P=0.023

-2

02

Isolated diabetic nephropathy Non-diabetic renal disease

HbA

1c [%

]

a

bcontrol and presence of diabetic retinopathy and DN in T2D (Strattonet al., 2000). Third, besides the previously identied predictors of NDRDthat include new onset nephrotic range proteinuria, shorter duration ofdiabetes, and the absence of diabetic retinopathy (Chang, Park, Kim,et al., 2011; Pham, Sim, Kujubu, Liu, & Kumar, 2007; Sharma, Bomback,Radhakrishnan, et al., 2013), this is the rst study to demonstrate that inT2D, the HbA1c 12 months before biopsy and the absolute change inHbA1c during the year before biopsy predict the type of renal disease.This study also shows that those with biopsy-proven NDRD have abetter composite outcome of end stage renal disease and death.

Our nding of high prevalence of NDRD in T2D is consistent withobservations made by others (Chong et al., 2012; Das et al., 2012;Harada et al., 2013; Kleinknecht et al., 1992; Lee et al., 1999;Mazzuccoet al., 2002; Olsen & Mogensen, 1996; Richards et al., 1992; Soni et al.,2006; Yaqub et al., 2012). These studies showed a variable prevalenceof NDRD of 1878%. This is attributed to selection criteria for renalbiopsy and to the geographical and ethnic differences in the incidenceof various NDRD. The nding of high prevalence of NDRD in thiscohort of T2D patients with renal impairment and atypical featureshas important clinical implications; at least 50% of patients had NDRDthat was treatable with steroids and immunosuppressive agents. Thetherapeutic adjustments based on histology and resultant modica-tion of course of NDRDmay explain the improvement in eGFR and thebetter composite adverse renal outcome in the NDRD subgroup. Bycontrast, there is no specic treatment for isolated DN (AmericanDiabetes Association, 2013; Gross et al., 2005; Kidney Disease:

N=31

-6

-4

Two-sample Wilcoxon rank-sum (Mann-Whitney) test

N

GSP

Fig. 1. a: Long-term glucose control in type 2 diabetic patients with isolated diabeticnephropathy vs. non-diabetic renal disease. HbA1c, glycated hemoglobin; NGSP,National Glycohemoglobin Standardization Program. b: Changes in HbA1c during theyear before renal biopsy in patients with isolated diabetic nephropathy vs. non-diabeticrenal disease. HbA1c, glycated hemoglobin; HbA1c, an absolute change in HbA1c;NGSP, National Glycohemoglobin Standardization Program.

Please cite this article as: Pallayova, M., et al., Predicting non-diabethemoglobin, Journal of Diabetes and Its Complications (2015), http://dx.Improving Global Outcomes (KDIGO) CKD Work Group, 2013). Earlyand accurate diagnosis of NDRD is important for diabetic patients since

Urine PCR at the time of biopsy 0.384 1.00 1.0 to 1.0RAAS blockade therapy at thetime of biopsy

1.000 1.04 0.19 to 5.03

CI, condence interval; eGFR, glomerular ltration rate estimates; eGFR, an absolutechange in eGFR; GLP-1, glucagon-like peptide-1; HbA1c, glycated hemoglobin; HbA1c,an absolute change in HbA1c; INTHbA1c, a variable consistent with the interactionbetween HbA1c 12 months before biopsy and HbA1c during the year before biopsy;OHA, oral hypoglycemic agents; OR, odds ratio; PCR, protein to creatinine ratio; RAAS,reninangiotensinaldosterone system.treatment and prognosis may vary according to the underlying cause.Several recent cross-sectional and longitudinal studies have

reported associations between HbA1c and kidney function in T2Dwith mixed results (Hsu, Chang, Huang, et al., 2012; Lee, Li, Lin, et al.,2013; Lin, Chen, Chen, et al., 2013; Luk, Ma, Lau, et al., 2013;Rodrguez-Segade, Rodrguez, Garca Lpez, Casanueva, & Camia,2012; Sugawara, Kawai, Motohashi, et al., 2012). Lee and colleagues(Lee et al., 2013) demonstrated a negative effect of preceding HbA1c(recorded 1 year before) on eGFR in T2D patients with CKD stages 3

unknown HbA1c during the year before biopsy

decrease in HbA1c during the year before biopsyincrease in HbA1c during the year before biopsy

.1

.2

.3

.4

.5

.6

.7

.8

.9

5 6 7 8 9 10 11 12 13NGSP HbA1c 12 months before biopsy [%]

pro

babi

lity o

f NDR

D

Fig. 2. Predicted probabilities of having non-diabetic renal disease in type 2 diabetes, bythe HbA1c 12 months before biopsy. HbA1c, glycated hemoglobin; HbA1c, an absolutechange in HbA1c; NDRD, non-diabetic renal disease; NGSP, National GlycohemoglobinStandardization Program.

ic renal disease in type 2 diabetic adults: The value of glycateddoi.org/10.1016/j.jdiacomp.2014.12.005

5M. Pallayova et al. / Journal of Diabetes and Its Complications xxx (2015) xxxxxxand 4. In our study, a trend towards the positive association betweenthe HbA1c 12 months before biopsy and eGFR was observed in thesubgroup with isolated DN (r = 0.56; P = 0.058; n = 12). There wasno association between the preceding HbA1c and eGFR even after thepatients were grouped according to CKD stages. Of importance, ourresults show that both the HbA1c 12 months before biopsy and theabsolute change in HbA1c during the year before biopsy differbetween patients with isolated DN and NDRD with similarlydecreased eGFR at the time of renal biopsy. Our study is consistentwith previous observations that annual variation in HbA1c couldpredict DN in patients with T2D and that long-term variability ofHbA1c predicts microalbuminuria (Hsu et al., 2012; Sugawara et al.,2012) and development/progression of renal and cardiovascularcomplications of T2D (Luk et al., 2013; Rodrguez-Segade et al., 2012).

The ROC curve analysis of the three proposed HbA1c-basedpredictive models showed that although both the HbA1c 12 monthsbefore biopsy and its change during the year before biopsy predictedtype of renal disease, the model based on the interaction between thetwo variables (INTHbA1c) had the best predictive accuracy for NDRD.Since different predictors may be sensitive to different aspects of renaldiseases, this new predictor may improve the overall predictivecapability of the HbA1c-based models.

Fig. 3. ROC curve of the three HbA1c-based predictive models for non-diabetic renaldisease. HbA1c, glycated hemoglobin; HbA1c, an absolute change in HbA1c during theyear prior to biopsy; INTHbA1c, an interaction term between the HbA1c 12 monthsbefore biopsy predictor and theHbA1c during the year before biopsy predictor; NDRD,non-diabetic renal disease; ROC, receiver operating characteristic.In the present study, the patients histologically diagnosed withNDRD had fairly good long-term diabetes control 12 months prior tobiopsy. Our ndings indicate that the HbA1c either remains stable orgradually deteriorates in patients with NDRD despite the decline inrenal function during the year prior to renal biopsy. This observationmay be explained by the adverse impact of inammatory and immuneresponses to progressive NDRD on glucose control. Both acute andchronic inammation leads to stress hyperglycemia, consistent with amaladaptive and detrimental response to stress and inammation(Collier, Dossett, May, & Diaz, 2008). Furthermore, in patients withisolated DN, the elevated preceding HbA1c levels improved signi-cantly during the year before renal biopsy. Decreased renal degrada-tion of insulin typically ensues later in the course of CKD (Rabkin,Ryan, & Duckworth, 1984), which could account for the improvedglucose control prior to biopsy in patients with isolated DN.

There are several limitations to our study that should beconsidered in relation to the ndings. The relatively small samplesize and retrospective nature precluded examination of inuence ofethnicity, arterial hypertension, anemia, smoking, and others. Data onsome clinical variables and characteristics were not available for somepatients and thus could not be included in our analyses. The exactstages of diabetic retinopathy could not be established in each case to

Please cite this article as: Pallayova, M., et al., Predicting non-diabethemoglobin, Journal of Diabetes and Its Complications (2015), http://dx.further correlate with the type of renal disease. A nal limitationconcerns the applicability of the results to practical decisionmaking inthe general diabetic population, as only T2D adults with suspicion ofunderlying NDRD were enrolled.

Despite these limitations, our study demonstrates a considerablyhigh prevalence of histologically-proven NDRD in T2D adults under-going renal biopsy. It also shows theHbA1c levels during the year beforerenal biopsy differ between NDRD and isolated DN. Besides thepredictive value of diabetic retinopathy, we have identied the HbA1c12 months before biopsy and HbA1c as important novel candidatepredictors of NDRD and DN in T2D. While HbA1c remains the bestlong-termmarker of glycemic control in patients with T2D, our ndingssuggest that annual evaluation of HbA1c and its dynamic changes alongwith the assessment of diabetic retinopathy could facilitate earlydetection of NDRD in T2D. Clinical signicance of this nding isemphasized by the fact that the selection criteria for renal biopsy inT2D adults are still not well established and novel predictors of renaldisease are critically needed. Prospective studies are needed to validatethe proposed HbA1c-based predictive models for NDRD with a view torening the current selection criteria for renal biopsy in T2D patients,and identify patients that will benet from specic therapeuticinterventions that will reduce adverse renal outcomes.

References

American Diabetes Association (2013). Standards of medical care in diabetes2013.Diabetes Care, 36(Suppl. 1), S11S66.

Chang, T. I., Park, J. T., Kim, J. K., et al. (2011). Renal outcomes in patients with type 2diabetes with or without coexisting non-diabetic renal disease. Diabetes Researchand Clinical Practice, 92, 198204.

Chong, Y. B., Keng, T. C., Tan, L. P., et al. (2012). Clinical predictors of non-diabetic renaldisease and role of renal biopsy in diabetic patients with renal involvement: Asingle centre review. Renal Failure, 34, 323328.

Collier, B., Dossett, L. A., May, A. K., & Diaz, J. J. (2008). Glucose control and theinammatory response. Nutrition in Clinical Practice, 23, 315.

Das, U., Dakshinamurty, K. V., Prayaga, A., & Uppin, M. S. (2012). Nondiabetic kidneydisease in type 2 diabetic patients: A single center experience. Indian Journal ofNephrology, 22, 358362.

Gross, J. L., de Azevedo, M. J., Silveiro, S. P., Canani, L. H., Caramori, M. L., & Zelmanovitz, T.(2005). Diabetic nephropathy: Diagnosis, prevention, and treatment. Diabetes Care,28, 164176.

Harada, K., Akai, Y., Sumida, K., et al. (2013). Signicance of renal biopsy in patients withpresumed diabetic nephropathy. Journal of Diabetes Investigation, 4, 8893.

Hsu, C. C., Chang, H. Y., Huang, M. C., et al. (2012). HbA1c variability is associated withmicroalbuminuria development in type 2 diabetes: A 7-year prospective cohortstudy. Diabetologia, 55, 31633172.

Kidney Disease: Improving Global Outcomes (KDIGO) CKDWork Group (2013). KDIGO2012 Clinical Practice Guideline for the Evaluation and Management of ChronicKidney Disease. Kidney International. Supplement, 3, 1150.

Kleinknecht, D., Bennis, D., & Altman, J. J. (1992). Increased prevalence of nondiabeticrenal pathology in type 2 diabetes mellitus. Nephrology, Dialysis, Transplantation, 7,12581259.

Lee, E. Y., Chung, C. H., & Choi, S. O. (1999). Non-diabetic renal disease in patients withnon-insulin dependent diabetes mellitus. Yonsei Medical Journal, 40, 321326.

Lee, C. L., Li, T. C., Lin, S. Y., et al. (2013). Dynamic and dual effects of glycatedhemoglobin on estimated glomerular ltration rate in type 2 diabetic outpatients.American Journal of Nephrology, 38, 1926.

Levey, A. S., Bosch, J. P., Lewis, J. B., Greene, T., Rogers, N., & Roth, D. (1999). A moreaccurate method to estimate glomerular ltration rate from serum creatinine: anew prediction equation. Modication of Diet in Renal Disease Study Group. Annalsof Internal Medicine, 130, 461470.

Lin, C. C., Chen, C. C., Chen, F. N., et al. (2013). Risks of diabetic nephropathy withvariation in hemoglobin A1c and fasting plasma glucose. The American Journal ofMedicine, 126, 1017.e11017.e10.

Luk, A. O., Ma, R. C., Lau, E. S., et al. (2013). Risk association of HbA1c variability withchronic kidney disease and cardiovascular disease in type 2 diabetes: Prospectiveanalysis of the Hong Kong Diabetes Registry. Diabetes/Metabolism Research andReviews, 29, 384390.

Mauer, M., Fioretto, P., Woredekal, Y., et al. (2001). Diabetic nephropathy. In R. W.Schrier (Ed.), Diseases of the kidney and urinary tract (pp. 20832116) (7th ed.).Philadelphia, PA: Lippincott Williams & Wilkins.

Mazzucco, G., Bertani, T., Fortunato, M., et al. (2002). Different patterns of renal damagein type 2 diabetes mellitus: A multicentric study on 393 biopsies. American Journalof Kidney Diseases, 39, 713720.

Olsen, S., &Mogensen, C. E. (1996).Howoften isNIDDMcomplicatedwith non-diabetic renaldisease? An analysis of renal biopsies and the literature. Diabetologia, 39, 16381645.

Pham, T. T., Sim, J. J., Kujubu, D. A., Liu, I. L., & Kumar, V. A. (2007). Prevalence ofnondiabetic renal disease in diabetic patients. American Journal of Nephrology, 27,322328.ic renal disease in type 2 diabetic adults: The value of glycateddoi.org/10.1016/j.jdiacomp.2014.12.005

Rabkin, R., Ryan, M. P., & Duckworth, W. C. (1984). The renal metabolism of insulin.Diabetologia, 27, 351357.

Richards, N. T., Greaves, I., Lee, S. J., Howie, A. J., Adu, D., & Michael, J. (1992). Increasedprevalence of renal biopsy ndings other than diabetic glomerulopathy in type 2diabetes mellitus. Nephrology, Dialysis, Transplantation, 7, 397399.

Rodrguez-Segade, S., Rodrguez, J., Garca Lpez, J. M., Casanueva, F. F., & Camia, F.(2012). Intrapersonal HbA(1c) variability and the risk of progression ofnephropathy in patients with type 2 diabetes. Diabetic Medicine, 29, 15621566.

Sharma, S. G., Bomback, A. S., Radhakrishnan, J., et al. (2013). The modern spectrum ofrenal biopsy ndings in patients with diabetes. Clinical Journal of the AmericanSociety of Nephrology, 8, 17181724.

Soni, S. S., Gowrishankar, S., Kishan, A. G., & Raman, A. (2006). Non diabetic renaldisease in type 2 diabetes mellitus. Nephrology (Carlton, Vic.), 11, 533537.

Stratton, I. M., Adler, A. I., Neil, H. A., et al. (2000). Association of glycaemia withmacrovascular and microvascular complications of type 2 diabetes (UKPDS 35):Prospective observational study. BMJ, 321, 405412.

Sugawara, A., Kawai, K., Motohashi, S., et al. (2012). HbA1c variability and thedevelopment of microalbuminuria in type 2 diabetes: Tsukuba Kawai DiabetesRegistry 2. Diabetologia, 55, 21282131.

The Diabetes Control and Complications Trial Research Group (1993). The effect of intensivetreatmentof diabeteson thedevelopment andprogressionof long-termcomplications ininsulin-dependent diabetesmellitus. The New England Journal of Medicine, 329, 977986.

The Microalbuminuria Collaborative Study Group (1999). Predictors of the develop-ment of microalbuminuria in patients with type 1 diabetes mellitus: A seven-yearprospective study. Diabetic Medicine, 16, 918925.

UK Prospective Diabetes Study (UKPDS) Group (1998). Intensive blood-glucose controlwith sulphonylureas or insulin compared with conventional treatment and risk ofcomplications in patients with type 2 diabetes (UKPDS 33). Lancet, 352, 837853.

Wong, T. Y., Choi, P. C., Szeto, C. C., et al. (2002). Renal outcome in type 2 diabetic patientswith or without coexisting nondiabetic nephropathies. Diabetes Care, 25, 900905.

Yaqub, S., Kashif,W., &Hussain, S. A. (2012). Non-diabetic renal disease in patientswith type-2 diabetes mellitus. Saudi Journal of Kidney Diseases and Transplantation, 23, 10001007.

6 M. Pallayova et al. / Journal of Diabetes and Its Complications xxx (2015) xxxxxxPlease cite this article as: Pallayova, M., et al., Predicting non-diabethemoglobin, Journal of Diabetes and Its Complications (2015), http://dx.ic renal disease in type 2 diabetic adults: The value of glycateddoi.org/10.1016/j.jdiacomp.2014.12.005

Predicting non-diabetic renal disease in type 2 diabetic adults: The value of glycated hemoglobin1. Introduction2. Subjects, material and methods2.1. Statistical analyses

3. Results3.1. Sample characteristics and renal biopsy findings3.2. Predictors of renal disease and risk factors for adverse renal outcomes

4. DiscussionReferences