Introduction

The most widely used method for measuring urinary iodine is the Sandell-Kolthoff (SK) assay [1]. This kinetic assay utilizes the iodine-catalyzed redox reaction between colorless arsenic(III) and yellow cerium(IV) to give colorless products (As3+ + 2Ce4+ → As5+ + 2Ce3+) [2]. After a fixed reaction time, the decreased yellow color from cerium(IV) is measured spectrophotometrically and the concentration of iodine is determined from a standard curve [3]. However, the SK reaction is affected by many substances that cause interference [4].

Initially, the SK assay was used to measure iodine in non-biological solutions. Different authors studied interference from inorganic ions; Sandell and Kolthoff’s original paper examined interference from osmium, ruthenium, manganese, permanganate, nitrite, thiocyanate, silver, cyanide, and magnesium [2]. Additionally, Stolc et al. found sulfate, zinc, and fluoride to interfere [5]. While many inorganic interferents to the Sandell-Kolthoff reaction are known, these compounds are not commonly present in urine. Only one organic interferent present in urine, ascorbic acid, has been identified [6].

Human urine contains numerous redox-active organic substances which could cause interference. Substances interfere with the SK reaction either by chelating cerium or by reacting with cerium(IV) or arsenic(III) [7]. Historically, rigorous pretreatment methods were developed to remove all interferents without selectivity, such as overnight digestion in nitric acid followed by heating in a mixture of concentrated sulfuric acid and 70% perchloric acid [8], use of an activated charcoal column [9], and heating in ammonium persulfate [10]. The deployment of such complex sample pretreatment methods has limited urinary iodine analysis to well-equipped laboratory environments, which are scarce in low- and middle-income countries. If an effective, field-friendly pretreatment method can be developed, urinary iodine analysis could be carried out on-site [11].

In order to determine the efficacy of new, milder, and more field-friendly pretreatment methods, it is necessary to know what the major interferents are. Bouatra et al. [12] identified 445 metabolites and quantified 378 unique metabolites found in human urine. We tested 33 metabolites commonly present in urine at physiologically relevant concentrations [12] to determine which ones interfered with the SK reaction.

Materials and Methods

The following chemicals were used as reagents in the SK assay: iodine standard solution (Inorganic Ventures, Lot: H2-IODO1034, Assay: 99%), trace metal grade arsenic(III) oxide (Aldrich), ACS grade ammonium cerium(IV) sulfate dihydrate (Sigma-Aldrich) and trace metal grade sulfuric acid (VWR). The following chemicals were screened for interference: urea, glycine, L-histidine (Sigma-Aldrich), creatinine (MP Biomedicals), hippuric acid, taurine, myoinositol, L-lysine, D-glucuronic acid (TCI), ACS grade citric acid, ascorbic acid (VWR), trimethylamine N-oxide (Fluka), L-cysteine, D-glucose, arabinose (Sigma), creatine, L-cystine, L-alanine (Alfa Aesar), glycolic acid (Thermo-Fisher), L-glutamine, phenol (Acros Organics), methanol (Birch Biotech), ethanolamine, formic acid (Sial), dimethylamine, L-serine, betaine (Aldrich), L-threonine (Beantown Chemical), acetic acid (Fisher Scientific), glycerol (BDH), L-lactic acid (EMD Millipore) oxalic acid (SAJ) and urobilin hydrochloride (Cayman Chemical). For testing a standard pretreatment method: ammonium persulfate (VWR). The following chemical was used in testing potential mechanisms of interference: ethylenediaminetetraacetic tetrasodium salt (Aldrich).

Absorbance data was gathered using one of two plate readers (based on availability): Molecular Devices SpectraMax M5 (SoftMax Pro software version 7.1) and BioTEK Synergy H1N (Gen5 software version 2.09). Data analysis was performed using Microsoft Excel.

Metabolites Tested

We tested 33 of the more abundant, redox-active metabolites commonly present in urine [12]. Ascorbic acid, a metabolite known to interfere with the SK reaction, was also tested as a positive control. Each metabolite (Table S1) was tested at its mean physiological concentration; concentrations were measured by NMR as given in Table 3 of ref [12] and are denoted as 100% in Fig. 1. Once identified as an interferent, lower concentrations were tested to determine an approximate threshold concentration for interference. Each stock solution of metabolite was prepared within 24 h of use to avoid decomposition and analysis was carried out by spiking the metabolite into aqueous iodine solutions.

Microplate SK Assay

Our lab is certified (Fig. S1) through the CDC’s EQUIP program [13] to conduct urinary iodine analysis by the method of ammonium persulfate digestion with spectrophotometric detection of the Sandell-Kolthoff reaction on microplate [1, 14], The microplate SK method has the advantages of increased throughput and reduced waste compared to the original cuvette method, and therefore we adapted it to screen for interference in the SK reaction. Fixed amounts of cerium(IV) and arsenic(III) were allowed to react with a series of concentrations of aqueous iodine ranging from 0 to 300 ppb, in the presence of the metabolite/suspected interferent (test samples) and its absence (controls). The concentrations of the iodine standards are for undiluted solutions, but to mimic the dilution that occurs in the pretreatment step of the SK reaction (where 250 µL of urine would be combined with 1000 µL of ammonium persulfate), each iodine standard was diluted by a factor of five. Initially, interference was screened with a single trial. However, for metabolites that interfered, interference was confirmed by triplicate determinations as shown in Fig. 1.

Fig. 1
figure 1

Microplate setup with different concentrations of iodine added to each row and different concentrations of a selected metabolite added to each column; the mean physiological concentration of metabolite [12] is denoted by 100% and some of the lower concentrations tested are denoted by 50% and 25%, respectively

Full size image

Procedure

To a 96-well plate with a clear and flat bottom, the following reagents were added with a multi-channel pipette: iodine standards (40 µL, added row-wise), metabolite solution (for test samples) or water (for controls) (10 µL, added column-wise) and arsenic(III) solution (0.05 M, 100 µL). Thereafter, cerium(IV) solution (0.02 M, 50 µL) was added quickly (within one minute) to all wells. The total volume in each well was 200 µL. The well plate was placed on an orbital shaker at 200 rpm for 3 min to ensure mixing of the reagents in each well. Exactly 30 min after the cerium addition, the absorbance of each well was recorded at 420 nm using a plate reader.

In Fig. 1, for the 50% columns, 5 µL of metabolite and 5 µL of water were added, such that the concentration of metabolite was 50% of its mean physiological concentration. Similarly, for the 25% columns, 2.5 µL of metabolite and 7.5 µL of water were added. Likewise, even lower physiological concentrations were tested (12.5%, 6.25%, …, etc.) until no interference was observed. These experiments were used to generate the data used in Fig. 3.

Experiments to Test if a Standard Pretreatment Method (Ammonium Persulfate Digestion) Removes Interference

A composite solution of interferents (including ascorbic acid, cysteine, citric acid and glycolic acid, but excluding urobilin) was prepared with final concentrations equivalent to each interferents’ mean physiological concentration. Additionally, another composite solution of interferents was prepared as above, and spiked with urobilin at its lowest physiological concentration. Each solution was divided into aliquots, spiked with different concentrations of iodide, and digested in 1 M ammonium persulfate (APS) at 95 °C for 1 h.

Additional Experiments to Determine Mechanisms of Interference

Three mechanisms of interference were studied as follows: (a) to simulate a reaction of the interferent with cerium(IV), a reduced amount of cerium(IV) was added; (b) to model chelation of cerium by the interferent, EDTA, which is a strong chelator of cerium [15], was added, and (c) to simulate a reaction of the interferent with arsenic, a reduced amount of arsenic was added. The absorbance from a series of iodine solutions was measured in each case.

Access to Data

The raw and processed data are available in the Supporting Information (Spreadsheet S1).

Data Analysis

The data from the plate reader was analyzed using MS Excel. The negative base 10 logarithm of the absorbance of cerium(IV) was the dependent variable (y) and the known iodine concentration of a well was the independent variable (x). The LINEST function in MS Excel was used to fit a linear regression line to each data series and extract the slope and intercept along with their standard errors. To determine whether a metabolite interfered, i.e., to determine if the difference between the slopes and/or intercepts of the control and metabolite was statistically significant, the two-tailed Welch’s t-test [16] at a 98% confidence level was used.

Results and Discussion

When the 33 selected metabolites were tested (Table S1), four new interferents were identified: cysteine, citric acid, glycolic acid and urobilin (Fig. 2). Ascorbic acid, which was previously identified as an interferent in the literature, was also identified as an interferent. During the course of the SK reaction, the yellow color of cerium(IV) fades at a rate that depends on the concentration of iodine present. Urobilin, cysteine, ascorbic acid, and citric acid accelerated the SK reaction, causing positive interference. The effect was particularly strong for urobilin, which induced rapid and complete bleaching of the cerium(IV) color. However, glycolic acid decelerated the reaction, causing negative interference. At their respective mean concentrations in urine, urobilin, ascorbic acid, cysteine, and glycolic acid significantly affected the slope, while citric acid and urobilin affected both the slope and intercept of the standard curve (Fig. 2).

Fig. 2
figure 2

Responses from metabolites identified as interferents, plotted alongside the standard curve (responses from controls). The iodine concentration was for the undiluted standards corresponding to undiluted urine. The concentration of the metabolite is the effective concentration in a well of the microplate SK reaction derived from the metabolite’s mean physiological concentration (see Table S1). The standard error of the mean from three replicate tests has been plotted as error bars, but they are smaller than the data points and hence not visible. Linear fitting parameters show slope (standard error) and y-intercept (standard error)

Full size image

For each interferent, lower concentrations (50%, 25%, 12.5%, etc., of the mean concentration in urine) were tested to better define an approximate threshold (minimum) concentration for interference. These results are shown pictorially in Fig. 3. It is evident that all compounds cause interference even at lower physiological levels, which means that they would be interferents present in nearly all human urine samples.

Fig. 3
figure 3

The yellow bars represent the five interferents and their respective physiological concentrations in urine [12]. The black line indicates the approximate threshold concentration for interference

Full size image

The median urinary iodine level of a population is the statistical figure of merit used to define the iodine nutrition status of a population [17]. According to most recent data [18], the median urinary iodine level of the population of the United States is 121 ppb. Therefore, the percentage change in y-value at an iodine concentration of 121 ppb was used to rank the impact of interferences. According to this method, the ranking of the impact of interferences is: Urobilin (+ 669%) > Cysteine (+ 22%) > Glycolic acid (-20%) > Ascorbic acid (+ 18%) > Citric acid (+12%). Ranking would assist in prioritizing interferents for removal.

When composite solutions of interferents were digested in 1 M APS by heating at 95 °C for 1 h, it was observed that interference from ascorbic acid, cysteine, citric acid, and glycolic acid were completely removed. For the composite solution of interferents including urobilin, APS digestion removed approximately 80% of interference (Fig. S2).

Finally, the mechanism of interference for each applicable compound was hypothesized and investigated. Urobilin caused a large positive interference. When urobilin was added to a solution containing only cerium(IV), no reaction occurred. However, when urobilin was added to a solution containing both cerium(IV) and arsenic(III), the yellow color of cerium(IV) disappeared in under 3 min, while the microplate absorbance measurements were taken at 30 min. The rapid disappearance of color indicates that urobilin is a very potent interferent. Although urobilin is responsible for the yellow color of urine, its interference in the SK reaction is caused by urobilin’s redox activity and not its color.

Cysteine and ascorbic acid also caused positive interference. The literature states that cysteine [19] and ascorbic acid [20] can be oxidized by cerium(IV). It was hypothesized that the resulting reduction in the amount of yellow cerium(IV) was the reason for positive interference. This was simulated by the addition of a lower amount of cerium(IV). The response obtained (Fig. 4a) was a positive interference with a greater slope, similar to the response from ascorbic acid and cysteine (Fig. 2).

Citric acid caused interference with a greater y-intercept but smaller slope (Fig. 2). The greater y-intercept could be due to the carboxylic acid groups on citric acid chelating with cerium(III/IV) which has been shown to increase the reduction potential of cerium(IV) [21]. Chelation with cerium(III/IV) was simulated by the addition of EDTA, and the response obtained (Fig. 4b) was a greater y-intercept, as expected. On the other hand, the decrease in the slope caused by citric acid could be due to a competing reaction: citric acid coordinating with and catalyzing the oxidation of arsenic(III) [22]. This results in negative interference since a decrease in the amount of arsenic(III) has been found to reduce the rate of the SK reaction [23].

Glycolic acid also caused negative interference. It was hypothesized that this was due to the reaction of glycolic acid with arsenic(III) [24]. This was simulated by adding a reduced amount of arsenic(III). The response obtained (Fig. 4c) was a negative interference with a smaller slope, similar to the response from glycolic acid (Fig. 2).

Fig. 4
figure 4

The mechanism of interference was investigated by simulating (a) the removal of cerium(IV) by reduction: addition of a reduced amount of cerium(IV), (b) the removal of cerium by chelation: addition of ethylenediaminetetraacetic acid (EDTA) and (c) the removal of arsenic by oxidation: addition of a reduced amount of arsenic(III). The iodine concentration was for the undiluted standards corresponding to undiluted urine. The change in the concentration of cerium(IV) or arsenic(III) is the effective change in concentration in a well of the microplate SK reaction

Full size image

Conclusion

In this study, we used the microplate SK method to screen 33 of the major organic metabolites present in urine. While many inorganic interferences to the SK assay have been studied, our research fills a literature gap in identifying interference by organic compounds. Knowledge of organic interferences is particularly important because the SK assay is widely used for determining iodine in urine, which is a matrix that contains many organic compounds. In addition to ascorbic acid, the previously known interferent present in urine, we identified four new interferents: urobilin, cysteine, citric acid and glycolic acid. Knowledge of the major interferents would assist researchers in developing targeted pretreatment methods and pave the way for a field-friendly SK test to determine global iodine nutrition.