Development and Evaluation of Novel ELISA for Determination of Urinary Pentosidine

J Nutr Sci Vitaminol, 65, 526–533, 2019

Shoji Kashiwabara1, Hiroaki Hosoe1, Rei-ichi Ohno2, Ryoji Nagai2 and Masataka Shiraki3

1 Research & Development, SB Bioscience Co., Ltd., 33–94 Enoki-cho, Suita, Osaka 564–0053, Japan 2 Laboratory of Food and Regulation Biology, Graduate School of Agriculture, Tokai University, Toroku 9–1–1, Higashi-ku, Kumamoto, Kumamoto 862–8652, Japan 3 Research Institute and Practice for Involutional Diseases, Nagano 399–8101, Japan (Received June 12, 2019)

Summary Pentosidine is the most well-characterized advanced glycation end product (AGE). It has been measured by HPLC, although this approach cannot be adapted to analyze many clinical samples and is also time-consuming. Furthermore, the detection of pentosidine using a reported ELISA kit and HPLC system requires pretreatment by heating, which generates artificial pentosidine leading to overestimation. We developed a novel pentosidine ELISA system that don’t require sample pretreatment for analyzing urine samples. We then analyzed the accuracy, precision, and reliability of this system. Urinary samples for analysis were obtained from healthy volunteers and stored urinary samples from the participants of the Nagano cohort study were also used. The LoB and LoD were 4.25 and 6.24 pmol/mL, respectively. Intra- and inter-assay coefficients of variation were less than 5%. The spiking and dilution recoveries were 101.4% and 100.5%, respectively. Analysis of the cross-reac- tivities against seven compounds representative of AGEs and structurally similar to pentosidine showed no significant cross-reactivity. The correlation coefficient between the concentrations of pentosidine obtained from HPLC and ELISA for the same urine samples was r50.815. The urinary excretion of pentosidine upon overnight fasting was lower than that after a meal, suggesting the presence of diurnal variation in urinary pentosidine. In con- trast, day-to-day variation was not observed. These results indicate that the ELISA system has sufficient reliability, accuracy, and precision for measuring urinary pentosidine. Sam- pling of fasting urine is suitable for minimizing variation. In conclusion, this ELISA system is promising to evaluate the effect of AGE on lifestyle-related diseases. Key Words advanced glycation end-product (AGE), cross-reactivity, spiking and dilution recoveries, diurnal variation, least significant change (LSC)

Cardiovascular disease (CVD) and fracture are seri- tosidine is important to evaluate metabolic disorders. ous health problems in the elderly. These morbidities are However, this has not spread widely in a clinical context associated with a reduction in the activities of daily life; because the main approach for assaying pentosidine therefore, there is an urgent need to establish a system has been HPLC, which is time consuming. Furthermore, for their early detection. Although the exact mecha- the detection of pentosidine using a reported ELISA kit nisms behind these morbidities are complex, tissue oxi- and HPLC system requires pretreatment (pronase and dation and glycation are recognized as important causal acid hydrolysis, respectively) by heating, which gener- processes. To prevent CVD or fracture, biomarkers to ates artificial pentosidine leading to overestimation. evaluate tissue glyco-oxidation are urgently required in The application of heat treatment to protein solutions a clinical context. Among several biomarkers to detect containing glucose lead to the generation of (artificial) tissue glyco-oxidation, pentosidine is the most well- AGEs (6). Therefore, an assay method with the ability to characterized advanced glycation end product (AGE). analyze large numbers of samples with sufficient speci- This compound is formed in collagen fibers in bone and ficity and accuracy without heating is required. Here, vessels non-enzymatically and accumulates with age. we describe the development of a new ELISA system to In fact, the urinary excretion of pentosidine has been measure the urinary excretion of pentosidine and its reported to increase with aging (1) and can be predictive application to clinical samples. of future fracture in men (2) and women (3) and dia- MATERIALS AND METHODS betic patients (4). In addition, the association of serum or urinary pentosidine with CVD has been reported (5). Subjects. A test was performed to analyze the cor- These reports indicated that the measurement of pen- relation between the results of HPLC and ELISA for human urine samples (n5105) from the Nagano cohort E-mail: [email protected] study, an ongoing study of outpatients at a primary care

526 Development ELISA for Urinary Pentosidine 527

Fig. 1. Method of sample collection for the determination of LSC. Samples were collected three times a day on three arbitrary days during 1 wk: (m) second voiding urine in the morning, (a) afternoon (after lunch), and (e) evening (after dinner). institute in Nagano Prefecture, Japan (7, 8). To deter- plates were washed with PBS-T and then supplemented mine least significant change (LSC) and day-to-day with secondary antibody [HRP-labeled goat anti-rabbit and diurnal variations, urine samples from six healthy IgG (Scantibodies Laboratory, Santee, CA, USA)] diluted male volunteers were obtained after obtaining written 1 : 1,000 in blocking buffer. After incubation for 1 h at informed consent. room temperature, plates were washed with PBS-T and Chemicals for cross-reactivity evaluation. Pentosidine supplemented with TMB substrate (Agilent Technolo- {aS-amino-2-[(4S-amino-4-carboxybutyl) amino]-4H- gies, Santa Clara, CA, USA) followed by stop solution imidazo(4,5-b)pyridine-4-hexanoic acid, molecular after 30 min. Absorbance (OD450 and OD630 for refer- weight: 378.4} was purchased from Cayman Chemical ence) was measured using an ELx808 microplate reader (Ann Arbor, MI, USA). CML [Ne-(carboxymethyl) lysine] (BioTek Instruments, Winooski, VT, USA) and data were and CMA [N-(carboxymethyl) arginine] were from analyzed using Gen5 data analysis software. Nippi, Inc. (Tokyo, Japan). 3-Deoxyglucosone (3-deoxy- Evaluation of the ELISA. The use of the ELISA for glucosone) was from Dojindo Laboratories (Kumamoto, analyzing human urinary pentosidine was examined Japan). l(1)Arginine, l-histidine, and d(2)ribose were using the limits of blank and detection (LoB, LoD), intra- from FUJIFILM Wako Pure Chemical Corporation and inter-assay variation, recovery, dilution, and cross- (Osaka, Japan). l(1)Lysine was from Tokyo Chemical reactivity test. Industry Co., Ltd. (Tokyo, Japan). The limit of blank (LoB) and limit of detection (LoD) Development of competitive ELISA. Female New Zea- were determined using methods similar to those out- land White SPF rabbits that were 6 wk old were first lined in the classical approach of Clinical Laboratory immunized with pentosidine conjugated with keyhole Standards Institute (CLSI, Wayne, PA, USA) guideline limpet hemocyanin (KLH) in complete Freund’s adju- EP17-A2 (9). Briefly, this involved measurements being vant via subcutaneous injection. Six booster immuni- performed on both a set of blank samples and a set of zations were given at 2-wk intervals via subcutaneous low pentosidine level samples containing a measurand injection with incomplete Freund’s adjuvant. The rab- targeted around the assumed LoD. Depending upon the bits were exsanguinated on day 91, which was 7 d after distributions of the blank and low-level sample results, a the final immunization. Affinity chromatography was nonparametric data analysis option was selected to cal- performed on a Protein A-coupled column (GE Health- culate the LoD and LoD estimates. Specifically, two lots care, Buckinghamshire, England, UK) for purification of of ELISA kits and five urine samples from healthy volun- the polyclonal IgG. teers were used. The healthy samples were used as low- ELISA 96-well Nunc Maxisorp plates (Thermo Fisher level samples. Blank samples were made by performing Scientific, Waltham, MA, USA) were coated with 0.3 mg immunoadsorption for each specimen to remove endog- per well of recombinant streptavidin (Thermo Fisher enous pentosidine. Then, five blank samples for LoB and Scientific) in PBS overnight at 4˚C, were washed with five low-level samples for LoD were measured in quadru- PBS containing 0.05% Tween 20 (PBS-T), and then plicate on three test days. were blocked with Block Ace (Megmilk Snow Brand, To analyze intra-assay variation, eight replicates of Tokyo, Japan) for 3 h at room temperature. Biotin-con- three different samples were run in one assay. For inter- jugated pentosidine in PBS was added to the plate and assay variation, three different samples were run in five incubated at room temperature for 2 h. After the plates independent assays. The intra- and inter-assay variation had been washed with PBS-T, a pentosidine-coated plate was calculated as the coefficient of variation (CV%). was obtained as immobilized pentosidine was left in the The spike recovery test was used to determine whether plates. Then, 200 mL of PBS and 10 mL of test sample analyte detection was affected by a difference between or calibrator (0, 10, 30, 60, 120, or 180 pmol/mL) were the diluent used to prepare the standard curve and added to a microplate for dilution. Moreover, 30 mL of the urine sample matrix. It was performed using three the diluted sample and 70 mL of the polyclonal anti- samples: two with known pentosidine concentrations pentosidine IgG were added to the pentosidine-coated (180 and 700 pmol/mL in calibration 0) and a blank plate and incubated at room temperature for 2 h. The sample (calibration 0). A dilution recovery test was used 528 Kashiwabara S et al.

Table 2. Spiking recovery test for the determination of pentosidine in human urine by ELISA. A spiking recovery test was performed by adding known amounts of pentosidine to three different patients’ urine samples. (a) The spiking recovery test results

Exogenous Measurement Recovered Recovery Sample1 (pmol/mL) (pmol/mL) (pmol/mL) (%)

D 0.0 27.2 — — 18.0 46.5 19.3 107.4 E 70.0 99.3 72.1 103.1 0.0 32.8 — — 18.0 51.9 19.1 105.9 70.0 101.3 68.5 97.9 F 0.0 44.4 — — 18.0 62.5 18.1 100.7 Fig. 2. Calibration curve obtained from the ELISA using 70.0 109.7 65.4 93.4 four-parameter fitting. (b) The dilution recovery test results Table 1. Intra- and inter-assay variation for the deter- Measurement Recovery mination of pentosidine in human urine by ELISA. Sample2 Dilution rate (pmol/mL) (%) Assay Mean6SD CV Test1 Sample (n) (pmol/mL) (%) G 1.0 51.1 — 2.0 23.5 92.0 Intra-assay A 8 28.361.3 4.5 4.0 11.8 92.4 B 8 54.261.5 2.8 H 1.0 85.6 — C 8 105.863.2 3.0 2.0 42.5 99.3 Inter-assay A 5 28.560.5 1.8 4.0 22.3 104.2 I 1.0 91.1 — B 5 52.761.0 1.9 2.0 47.7 104.7 C 5 102.963.5 3.4 4.0 25.2 110.6 Pentosidine in healthy human urine samples was mea- 1 Patients’ urine samples. Concentrations of pentosidine sured repeatedly (n58 for intra-assay, n55 for inter- were 27.2 (D), 32.8 (E), and 44.4 (F). assay) by ELISA. 2 Patients’ urine samples. Concentrations of pentosidine 1 Pentosidine was mixed with healthy human urine were 51.1 (G), 85.6 (H), and 91.1 (I). to give final concentrations of 28.6 (A), 53.2 (B), and 104.4 pmol/mL (C). to determine whether analyte detection was affected by cross-reactivity. a difference between the urine samples matrix and the Correlation between HPLC and ELISA. The correlation diluent in the assay range. It was performed by diluting between HPLC and ELISA was examined using patients’ another three different urine samples containing a high urine samples (n5105). Pentosidine was measured by concentration of pentosidine. The samples were diluted HPLC as described previously (6, 10). The HPLC sys- 1 : 2 and 1 : 4 and assayed for recovery. tem was composed of an LC-10A liquid chromatogra- The stability of pentosidine in human urine speci- phy system (Shimadzu, Kyoto, Japan) with a Capcellpak mens was examined using three patients’ samples at C18 UG80 S-5 column (5 mm, 5034.6 mm I.D.) (Osaka two temperature conditions: 4˚C for 1 wk and 220˚C Soda, Osaka, Japan) and an RF-10A Shimadzu spec- for 1 mo. These samples’ initial values were measured trofluorometric detector (Ex. 335 nm, Em. 385 nm). before storage. The samples stored at 4˚C were measured The mobile phase was solvent A, 16% acetonitrile (v/v) at several points during 1-wk period: 5 h, 3 d and 7 d. containing 0.21% heptafluorobutyric acid (HFBA), and The samples stored at 220˚C were measured after 1 mo. solvent B, 60% acetonitrile containing 0.1% HFBA. The These concentrations were compared with each initial flow rate was maintained at 1.0 mL/min and the column value. Also, the effects of freeze-thaw on three patients’ was kept at 40˚C. Before sample injection for HPLC, the urine samples were evaluated. Freeze (220˚C)/thaw sample was hydrolyzed. Briefly, urine samples (300 mL (25˚C) cycles were repeatedly conducted up to 11 times, each) were mixed with equal volumes of 12 n HCl (final and they were compared with each initial value in the 6 n HCl) and 1 mL of 6 n hydrochloric acid (HCl), fol- same manner. lowed by hydrolysis at 110˚C for 18 h in sealed test Other major AGEs and the compounds with close tubes. The HCl was dried in vacuo and solubilized again structural similarity to pentosidine were evaluated for with 1 mL of distilled water. The samples were applied to Development ELISA for Urinary Pentosidine 529 cation exchange columns (Strata-X-C Polymeric Strong Statistical analysis. JMP for Windows Ver. 14.0.0 Cation, 30 mg/mL) (Phenomenex, Torrance, CA, USA), (SAS Institute Inc., NC, USA) was used to perform sta- eluted by 7% ammonia (Nacalai Tesque, Inc., Kyoto, tistical analyses. The significance of differences of urine Japan), and dried in vacuo. Then, the samples were resolubilized in 300 mL of 16% acetonitrile solution, followed by pentosidine analysis by HPLC. The same samples were adapted to the ELISA and the correlation coefficient of the values obtained by HPLC and ELISA was calculated. LSC, diurnal and day-by-day variation of urinary pentosidine. Least significant change (LSC), minimum significant change (MSC), and diurnal and day-by-day variation of urine pentosidine concentration were determined using urine samples from healthy volunteers (n56, male; age 53.563.1 y). These samples were collected after overnight fasting (n53) or no-fasting (without dietary restrictions) (n53), to evaluate the effect of a meal on urinary pentosidine excretion. The samples were collected three times a day on any 3 d during a week. On a particular day, the collection time was at the second urination in the morning, after lunch, and after dinner. To reduce influence of the individual dif- Fig. 3. Demonstration of the linearity of the urinary ference, the two groups took turns the next week in a pentosidine assay by serial dilution of each test sample. crossover design; therefore, in total, urine was collected Three patients’ urine samples with pentosidine concen- on 6 d (Fig. 1). LSC and MSC at a significance level of trations of 94.1 (d), 85.6 (), and 51.1 pmol/mL () p,0.05 were calculated in accordance with the follow- were serially diluted with blank human urine sample, ing formula (11): which was produced by immunoadsorption. Urinary pentosidine concentrations were measured by ELISA. LSC51.963√23CVt, MSC523CVt 2 2 Each plotted point represents the average of duplicate CVt=√(CVi +CVa ) experiments. The linear regression equations are as fol- CVi: Intra-individual coefficients of variation lows: (d) y592.0x12.0, r51.000, () y584.7x10.7, CVa: Analytical coefficients of variation r51.000, and () y552.8x22.0, r50.999.

Table 3. The stability of pentosidine in human urine samples. (a) Stability: stored at 4˚C and 220˚C

Refrigerated Frozen 4˚C 220˚C Sample Initial 5 h 3 d 7 d 1 mo

J Conc. (pmol/mL) 41.2 39.3 36.6 40.7 39.4 Ratio to initial — 95.4% 88.8% 98.8% 95.6% K Conc. (pmol/mL) 46.2 51.2 46.9 47.4 44.7 Ratio to initial — 110.8% 101.5% 102.6% 96.8% L Conc. (pmol/mL) 127.5 122.3 116.3 120.2 122.4 Ratio to initial — 95.9% 91.2% 94.3% 96.0%

(b) Stability: freeze thaw cycles (freeze: 220˚C/thaw: 25˚C)

Number of freeze thaw cycles Sample Initial 1 3 5 7 9 11

M Conc. (pmol/mL) 36.0 37.9 35.6 36.9 37.9 36.2 36.5 Ratio to initial — 105.3% 98.9% 102.5% 105.3% 100.6% 101.4% N Conc. (pmol/mL) 54.1 54.6 54.1 56.4 57.2 51.6 52.3 Ratio to initial — 100.9% 100.0% 104.3% 105.7% 95.4% 96.7% O Conc. (pmol/mL) 84.4 87.1 86.8 88.5 89.0 88.5 89.1 Ratio to initial — 103.2% 102.8% 104.9% 105.5% 104.9% 105.6% 530 Kashiwabara S et al.

Fig. 4. Compounds for cross-reactivity test of pentosidine. (a) Pentosidine {aS-amino-2-[(4S-amino-4-carboxybutyl) amino]-4H-imidazo(4,5-b) pyridine-4-hexanoic acid}, (b) CML [Ne-(carboxymethyl)lysine], (c) CMA [N-(carboxymethyl)], (d) 3-DG (3-deoxyglucosone), (e) Arg [l(1)arginine], (f) Lys [l(1)lysine], (g) His [l(2)histidine], (h) d(2)ribose.

Table 4. Cross-reactivity among pentosidine and related compounds.

Cross-reactivity Measurement Concentrationof cross-reactant Compound (%) (pmol/mL) (pmol/mL)

Pentosidine 100.3 60.2 60.0 CML 0.00 0.1 1.03105 CMA 0.00 0.5 1.03105 3-DG 0.00 0.0 1.03105 Arg 0.00 1.6 1.03105 Lys 0.00 7.1 1.03107 His 0.00 0.1 1.03107 d(2)Ribose 0.00 0.2 1.03105

The cross-reactivity between closely related compounds and pentosidine was examined in the ELISA system. As indicated in Table 4, there was no significant cross-reactivity among the compounds tested.

pentosidine between the two categories was tested by The dilution recovery study was performed by dilut- Wilcoxon’s rank-sum test. ing three urine samples with pentosidine values from Ethical approval. This study was approved by the eth- 51.1 to 94.1 pmol/mL with sample dilution buffer. The ics committee of DS Pharma biomedical (17-001). samples were diluted 1 : 2 and 1 : 4 and assayed for recovery. There were high correlation coefficients of RESULTS the linear regression for these samples (from r50.999 Analytical performance to 1.000) (Fig. 3). The mean dilution recovery rate Calibration curve is shown in Fig. 2. A four-param- was 100.567.4% (Table 2b). This suggested that there eter fit was applied to obtain the function describing a are no interfering factors in the matrix of the samples sigmoid model. As described in “Materials and Meth- because the sample fluid and the sample diluent dem- ods,” the LoB and LoD values were determined. The onstrated the same behavior in the assay. The present obtained results were processed following the method of ELISA system has sufficient reliability, as judged by the CLSI EP17. LoB and LoD were 4.25 and 6.24 pmol/mL, results of the recovery tests. respectively. All intra- and inter-assay variations were The sample stability test was performed with the below 5% (Table 1). stored samples at 4˚C and 220˚C. The rates of changes The reliability of the measurements was revealed by from the initial values were from 25.7% to 2.6% at 4˚C determining the percentage recoveries of the spike and 7 d and from 24.4% to 23.2% at 220˚C 1 mo (Table dilution tests. The spiking recovery ranged from 97.0% 3a). In the same manner, the freeze-thaw stability of the to 105.2%, with mean recovery of 101.4% (Table 2a). sample was evaluated. The rates of changes from the Development ELISA for Urinary Pentosidine 531

urinary pentosidine concentration was affected by food intake and it should be measured under fasting conditions (Table 5a). The findings also suggest that urine obtained after fasting at the second voiding of the morning is more suitable than that collected at an arbitrary time. According to the results of the diurnal variation in urinary pentosidine concentration, there were significant differences between the evening and morning values only in the OF period. On the other hand, regarding day-to-day variation, no significant change was identified. In addition, the urine obtained at the second voiding of the morning with overnight fasting provided the best LSC and MSC (Table 6). In other words, if judging the urine pentosidine level without fasting, there is a need to take into consideration the effect of food intake. DISCUSSION Fig. 5. Correlation between HPLC and ELISA for the Antecedent ELISA kits are available for measuring urinary pentosidine assays. Urinary samples (n5105) plasma pentosidine levels for the assessment of renal from patients with osteoporosis were measured by HPLC-UV and ELISA assays. The ELISA results showed function. In such kits, pretreatment of samples using a high correlation with HPLC-UV, despite the lack of high temperature acid hydrolysis (12) or enzymatic pretreatment for the hydrolysis of collagen. The lin- digestion using pronase E is required (13, 14). This is ear regression equation is as follows: y51.09x14.91, because plasma/serum proteins are large molecules, r50.815. and therefore, hydrolysis is necessary. Nakano et al. (6) reported that samples that were heated demonstrated 1.1- to 4.2-fold increased pentosidine levels compared with those that were not heated and that the rates of initial values were from 24.6% to 5.7% during the 11 increase were not consistent for each sample. These freeze-thaw cycles (Table 3b). findings indicated that a difference in glycated protein A total of seven compounds were tested for cross- concentrations in each sample may influence the rates reactivity. CMA, CML, and 3DG were adopted in the of increase in pentosidine levels. Conversely, blood is cross-reactivity test as representative AGEs. The other filtered by the renal glomerulus by passage through four compounds with structural similarity to pento- molecular sieves referred to as podocytes. Urinary pro- sidine that were used here were l(1)arginine, l(1) teins are small molecules, and thus, hydrolysis is unnec- lysine, l-histidine, and d(2)ribose (Fig. 4). These com- essary for measuring pentosidine in urine samples. pounds did not show any significant cross-reactivity in Anti-pentosidine antibodies can bind to pentosidine that the ELISA (Table 4), suggesting that the present ELISA exists in urine protein. Based on the above reasons, we system can measure urinary pentosidine with sufficient developed a reliable ELISA system for the detection of specificity. urinary pentosidine. Correlation analysis The LoB and LoD for this system were identified to The correlation between the levels of urinary pento- be 4.25 and 6.24 pmol/mL, respectively. No sample sidine measured by HPLC and ELISA was evaluated in was indicated to be below the LoD, suggesting that the a total of 105 urine samples spanning the assay range ELISA system had sufficient precision to detect urinary (LoD56.2–180 pmol/mL). As Fig. 5 shows, the cor- pentosidine in clinical samples. In addition, the spike relation coefficient between ELISA (y-axis) and HPLC and dilution recovery rates were 101.4% and 100.5%, (x-axis) data was r50.815 (y51.09x14.91). respectively, suggesting that the system had sufficient LSC, diurnal and day-by-day variation of urinary pentosi- accuracy. A comparative study between the ELISA and dine HPLC systems using clinical samples showed a high cor- Urinary pentosidine excretions were compared in the relation between the results obtained with these two samples obtained from the volunteers after overnight assays (r50.815, Fig. 5). Pentosidine is resistant to acid fasting (OF period) and no-fasting (NF period). As Table hydrolysis on the pretreatment of HPLC. It suggests that 5 indicates, the values for the NF period obtained after pentosidine is a very stable chemical substance. Accord- breakfast and after lunch were significantly higher than ing to the urine sample stability examination in this those at the corresponding times in the OF period [morn- study, the 220˚C frozen sample was stable for 1 mo. Fur- ing: 18.0 vs. 20.0 pmol/mgCr, p50.0124, afternoon: ther long-term stability results may be obtained if the 18.2 vs. 20.6 pmol/mgCr, p50.0326, all day (morn- period of sample storage is extended. However, urinary ing, afternoon, evening): 18.8 vs. 20.2 pmol/mgCr, creatinine assay to obtain a creatinine correction value p50.0110]. However, in the evening collection, there before specimen storage is necessary because creatinine were no significant differences (20.1 vs. 19.9 pmol/ stability should be taken into account. Furthermore, the mgCr) between the groups. These results indicate that results on the specificity of the antibody against pento- 532 Kashiwabara S et al.

Table 5. Diurnal (a) and day-to-day (b) variations of urinary pentosidine level with or without fasting. (a) Diurnal variations

Pentosidine level (pmol/mgCr) Diurnal variation (n56) Whole period (6 d) OF period (3 d) NF period (3 d)

Morning Mean6SD 19.063.0 18.062.4 20.063.3* Min 13.3 14.8 13.3 Max 29.8 23.6 29.8 Afternoon Mean6SD 19.463.7 18.262.9 20.664.1† Min 12.1 12.1 13.3 Max 28.7 24.9 28.7 Evening Mean6SD 20.064.1 20.164.3§ 19.964.0 Min 12.1 13.2 12.1 Max 28.8 28.8 26.1

All day Mean6SD 19.563.6 18.863.4 20.263.8‡ Min 12.1 12.1 12.1 Max 29.8 28.8 29.8

(b) Day-to-day variations

Pentosidine level (pmol/mgCr) Day-to-day variation (n56) OF period (3 d) NF period (3 d)

Day 1 Mean6SD 18.863.9 19.965.0 Min 13.2 12.1 Max 28.8 29.8 Day 2 Mean6SD 19.163.1 20.463.2 Min 15.9 13.0 Max 25.7 25.7 Day 3 Mean6SD 18.663.2 20.263.0 Min 12.1 14.7 Max 25.1 27.8

* p50.0124, † p50.0326, ‡ p50.0110, significant differences between the OF period and NF period. § p50.0460, significant difference between the evening and morning in the OF period. (a) Diurnal variation. Pentosidine level was aggregated at each timing of sample collection during the period. Whole period means the values obtained from the entire test period (6 d) including the OF period: overnight fasting period (3 d) and NF period: non-fasting period (3 d). (b) Day-to-day variation. Pentosidine level was aggregated on each day.

Table 6. The MSC and LSC of urine pentosidine measurements.

Second voiding urine in the morning Arbitrary time collection Timing of urine collection OF period NF period OF period NF period Whole period

MSC (%) 20.9 44.8 50.0 44.8 44.8 LSC (%) 29.0 62.0 69.3 62.0 70.5

Urine samples of healthy volunteers (n56) were collected at different times. Urine samples from the second voiding in the morning under OF period and NF period were collected (left column), and were also collected at arbitrary times with or without fasting. Both MSC and LSC at the second voiding urine under OF were smaller than those for the other urine samples. LSC and MSC upon collection of urine from the second voiding in the morning and at an arbitrary time. LSC: least significant change, MSC: minimum significant change. The smallest values of MSC and LSC were observed for the samples obtained in the OF period. Development ELISA for Urinary Pentosidine 533 sidine and other related compounds indicated no signifi- endogenous secretory receptor for advanced glycation cant cross-reactivity between them (Table 4). We also end products to pentosidine predicts fractures in men. J evaluated the diurnal variation of the samples obtained Clin Endocrinol Metab 103: 85–94. under different conditions. The subjects underwent the 3) Tanaka S, Kuroda T, Saito M, Shiraki M. 2011. Urinary pentosidine improves risk classification using fracture diurnal variation test after fasting or consuming break- risk assessment tools for postmenopausal women. J Bone fast, and the urine samples were collected in the morn- Miner Res 26: 2778–2784. ing, afternoon, or evening. A clear effect of food intake 4) Schwartz AV, Garnero P, Hillier TA, Sellmeyer DE, Strot- on the urinary excretion of pentosidine was observed. meyer ES, Feingold KR, Resnick HE, Tylavsky FA, Black Table 6 indicates the MSC and LSC under fasting and DM, Cummings SR, Harris TB, Bauer DC; Health, Aging non-fasting conditions. As Table 6 indicates, both MSC and Body Composition Study. 2009. Pentosidine and and LSC were greater in the samples without fasting increased fracture risk in older adults with type 2 diabe- than in those with fasting. Therefore, we have to con- tes. J Clin Endocrinol Metab 94: 2380–2386. sider the effect of a meal when performing serial sam- 5) Nenna A, Spadaccio C, Lusini M, Ulianich L, Chello pling and be aware of the possibility of wider error than M, Nappi F. 2015. Basic and clinical research against for fasting samples. However, this implies that pentosi- advanced glycation end products (AGEs): New compounds to tackle cardiovascular disease and diabetic dine may be used to evaluate healthy foods having low complications. Recent Adv Cardiovasc Drug Discov 10: AGE content, and this can be assessed in a future study. 10–33. The present study is associated with some limitations. 6) Nakano M, Kubota M, Owada S, Nagai R. 2013. The These included that the clinical samples used here were pentosidine concentration in human blood specimens is limited in number and that we could not evaluate the affected by heating. Amino Acids 44: 1451–1456. effect of diseases on the urinary excretion of pentosi- 7) Shiraki M, Kuroda T, Shiraki Y, Aoki C, Sasaki K, Tanaka dine. Thus, further study is required to establish the S. 2010. Effects of bone mineral density of the lumbar clinical usefulness of the present ELISA system for ana- spine and prevalent vertebral fractures on the risk of lyzing pentosidine. immobility. Osteoporos Int 21: 1545–1551. Despite the above limitations, the ELISA system 8) Urano T, Shiraki M, Ouchi Y, Inoue S. 2012. Association appears to be promising to determine tissue senescence of circulating sclerostin levels with fat mass and metabolic disease—related markers in Japanese postmeno- induced by glyco-oxidation. More extensive application pausal women. J Clin Endocrinol Metab 97: E1473–1477. of this approach to clinical samples is required to evalu- 9) CLSI. 2012. Evaluation of Detection Capability for Clini- ate its clinical usefulness. cal Laboratory Measurements Procedures; Approved In conclusion, we have established a novel ELISA sys- Guideline–Second Edition. CLSI document EP17-A2 tem for analyzing pentosidine that has sufficient speci- Wayne, Clinical and Laboratory Standards Institute, PA. ficity, accuracy, and precision. It is preferable to analyze 10) Ohno R, Moroishi N, Sugawa H, Maejima K, Saigusa fasting urinary pentosidine to avoid large variation in M, Yamanaka M, Nagai M, Yoshimura M, Amakura Y, the obtained results. Nagai R. 2015. Mangosteen pericarp extract inhibits the formation of pentosidine and ameliorates skin elasticity. Disclosure of state of COI J Clin Biochem Nutr 57: 27–32. SK and HH are employees of SB Bioscience Co., Ltd. 11) Hannon R, Blumsohn A, Naylor K, Eastell R. 1998. Response of biochemical markers of bone turnover to All the other authors declared no competing interests. hormone replacement therapy: impact of biological variability. J Bone Miner Res 13: 1124–1133. Acknowledgments 12) Izuhara Y, Miyata T, Ueda Y, Suzuki D, Asahi K, Inagi We thank H. Funaoka for the useful discussions and R, Sakai H, Kurokawa K. 1999. A sensitive and specific M. Hashimoto and H. Koreishi for the technical assis- ELISA for plasma pentosidine. Nephrol Dial Transplant tance. We are grateful to all of the volunteers and 14: 576–580. patients who participated in the present study. 13) Taneda S, Monnier VM. 1994. ELISA of pentosidine, an advanced glycation end product, in biological speci- REFERENCES mens. Clin Chem 40: 1766–1773. 1) Shiraki M, Kuroda T, Tanaka S, Saito M, Fukunaga M, 14) Sanaka T, Funaki T, Tanaka T, Hoshi S, Niwayama J, Nakamura T. 2008. Nonenzymatic collagen cross-links Taitoh T, Nishimura H, Higuchi C. 2002. Plasma pen- induced by glycoxidation (pentosidine) predicts vertebral tosidine levels measured by a newly developed method fractures. J Bone Miner Metab 26: 93–100. using ELISA in patients with chronic renal failure. Neph- 2) Tamaki J, Kouda K, Fujita Y, Iki M, Yura A, Miura M, ron 91: 64–73. Sato Y, Okamoto N, Kurumatani N. 2018. Ratio of