Optical Online Monitoring of Uremic Toxins beyond Urea

Abstract

This chapter presents origin and physical basis of the optical method for traditional haemodialysis (HD) dose assessment, accepted as a valid bloodless, robust, automatic, in situ and online monitoring technology in clinical praxis. Dialysis dose Kt/V, total removed urea (TRU) and the nutrition parameters PCR, nPCR estimation from ultraviolet (UV) absorbance in the spent dialysate is explained. Since urea, a small water-soluble uremic solute and a surrogate marker for the efficiency of dialysis treatment to clear the blood of toxins and metabolic end products, is not representative for all retained uremic toxins removed with the modern dialysis care, new developments of optical online monitoring of uremic toxins, beyond urea, are discussed. Optical intradialytic monitoring of small-, middle- and protein-bound molecules’ removal, exemplified by marker molecules uric acid, beta-2 microglobulin and indoxyl sulphate, is described. A new concept and sensor technology for multi-component uremic toxins’ intradialytic optical monitoring of spent dialysate with some clinical examples are introduced. Drug interference studies during the optical dialysis monitoring and future directions in optical monitoring are included. Offered benefits will be more patient-centred, integrated and cost-efficient care, as feedback for clinicians helps to improve and personalize the treatment quality, minimizing costly adverse effects.

1. Introduction

The traditional surrogate marker for dialysis dose, urea, which should reflect the clearance of various toxins and metabolic end products is disputed [1]. Removing a sufficient quantity of urea makes it possible to reduce symptoms, morbidity and mortality, and improve quality of life [2]. Urea has over the years been attempted to be measured online in the spent dialysate with different techniques such as enzymatic-, conductivity- and optical sensors. The enzymatic technique measure urea concentrations in the effluent spent dialysate stream, online, using either an ammonium ion (NH4 +) sensor that measures the amount of NH4 + determined directly by an ion-specific electrode or by an electrical potential difference between two electrodes, generated from hydrolysis of urea produces NH4+. A urease membrane catalyzes the chemical reaction when it comes in contact with urea in dialysate [3, 4]. This online technique has disappeared from the market possibly due to cumbersome handling for the staff and high extra costs. There are two still present techniques, first the ionic dialysance method uses a conductivity sensor [5] and is based on the fact that the diffusion coefficients of sodium and urea are similar at 37°C, through a dialysis membrane, therefore sodium dialysance can be used as a marker for urea clearance. The second commercially available technique is the optical method, which will be presented in more detail in this chapter. This technique utilizes the high correlation, in spent dialysate, between urea and UV absorbance at a certain wavelength range, even when urea itself does not absorb UV light [6, 7].

Urea shows a kinetic behavior that is not representative of all retained uremic toxins, including other water-soluble molecules belonging to the group of small molecules. A more comprehensive picture is needed for assessment of uremic solute removal during dialysis involving kinetic profiling and monitoring of the key molecules of all three groups (small-, middle- and protein-bound molecules) of solutes in uremic toxicity [1, 8, 9]. European Renal Best Practice has pointed out that beta-2 microglobulin (β2M) is a potential marker for the middle-size group having a kinetic behavior sufficiently representative of other middle molecules, including peptides of similar size [10, 11]. The protein-bound group, indoxyl sulphate (IS), has received attention because of its link to cardiovascular disease and mortality [12]. Furthermore, analyzing concentrations of these molecules requires today the cumbersome high-performance liquid chromatography (HPLC) method, which, therefore, has limited possibilities to be used in daily clinical practice.

A combined optical online technology utilizing simultaneously both UV absorbance and fluorescence might be a solution for this. This chapter will mainly focus on the latest research and development in that direction and will conclude with the results achieved so far.

2. Optical online monitoring of dialysis dose

2.1 Background of optical dialysis dose monitoring

Optical methods in dialysis dose monitoring started approximately 40 years ago with the development of the HPLC technique, which utilizes UV/Visible (Vis) spectroscopic data for analysis [13, 14, 15, 16]. HPLC was utilized for molecule separation and identification of contents in plasma, urine and spent dialysate [17, 18, 19, 20]. The “era of uremic toxins search” was born.

Applying standard laboratory photometers to measure solute removal during haemodialysis (HD) was first introduced by Boda and coworkers [21] who demonstrated an exponential decline of UV absorption in spent dialysate at wavelength 210 nm. The introduction of light-fiber optics and the developments of monochromator-detector in the mid-1980s and early-1990s, respectively, entailed near infrared spectroscopy (NIRS) becoming more powerful for scientific research. First, at the end of the 20th century, both UV- and NIRS-techniques were developed forward as practical tools applied for HD dose monitoring.

A Hungarian group published the first report about how UV transmittance of the spent dialysate can be measured at 254 nm, believed to be the best wavelength to monitor the efficacy of HD, [22]. Besides, there was concluded that the hardly diffusible components had a higher elimination rate compared to the removal rate of small, more freely diffusible constituents. Nearly two decades later, a work aiming to monitor the dialysis liquid during HD by UV absorbance was presented by Vasilevski et al. [23]. They also discussed UV extinction as an indicator of nucleic acid metabolism [24]. Almost simultaneously, in 2001, independent studies about online monitoring of solutes in dialysate using UV absorbance were published [6, 25], studying the possibility to monitor removal of different uremic solutes in the spent dialysate, and further firstly describing, as an illustrative example, how to estimate urea Kt/V from UV absorbance. The first clinical study, incorporating the UV-technology with a real clinical application, Kt/V calculation, was reported by Uhlin et al. [7]. A fruitful and exciting collaboration between Uhlin, Fridolin and co-workers within the field of optical dialysis dose monitoring has been followed. During the last decade, in connection with commercialization of the UV-technology for dialysis dose monitoring, new interest has appeared in the optical field. Some works, by the groups from Japan, related to spectroscopic analysis of uremic substances in dialysate have been presented [26, 27] and additionally some papers from St. Petersburg [28]. Validation of a clinical prototype device [29] and the commercially available UV absorbance dialysis dose monitor have been also presented [30]. The UV absorbance method as an alternative method to measure small-solute clearance is mentioned in DOQI clinical practice guideline for haemodialysis adequacy [31]. Moreover, the clinical care guidelines are available to interpret the real-time haemodialysate UV absorbance patterns to optimize solute clearance, troubleshoot problematic absorbance patterns and intervene during an individual treatment as needed [32].

2.2 Overview of optical principles in spectroscopy

Optical dialysis monitoring techniques utilize mostly phenomena described by the optics of biological fluids. Biological fluids include all kinds of fluids made by living organisms like urine, lymph, saliva blood, semen, mucus, gastric juice, aqueous humor, etc. Spent dialysate can also be categorized as a biological fluid derived by filtering the compounds usually less in size than 50,000 Dalton from blood through a dialyzer membrane into the pure dialysate containing water and electrolytes.

From the perspective of optics, biological tissues and fluids can be divided into two large classes: strongly scattering (opaque) tissues and fluids, such as skin, brain, vessel walls, blood, milk lymph and eye sclera, and weakly scattering (transparent) tissues and fluids such as crystalline lens, cornea, vitreous humor, aqueous humor of the front chamber of the eye [33] and spent dialysate. For the second class, the Beer–Lambert law is often applicable [34]. In this section, we will present a short description of the electromagnetic spectrum, some basic principles about photon propagation in biological fluids, Beer–Lambert law and examples of how this could be utilized as a measurement.

2.2.1 Electromagnetic radiation

Electromagnetic (EM) radiation used to be classified by wavelength into radio-, microwave, infrared, visible region, ultraviolet and x- and gamma-rays. The characteristics of EM radiation depend on its wavelength. An illustration of the EM spectrum range is shown in Figure 1. Infra-red (IR) is per definition EM radiation with a wavelength range between 760–0.5 mm, Vis 390–770 nm and UV 100–400 nm. The EM spectrum of UV can be subdivided in a number of ways. The draft ISO standard on determining solar irradiances [35, 36] describes the following ranges relevant to dialysis optical monitoring:

• Ultraviolet A (UVA), long wave 400–315 nm

• Ultraviolet B (UVB), or medium wave 315–280 nm

• Ultraviolet C (UVC), short wave 280–100 nm

Figure 1.

Illustration of the EM spectrum range.

“Light” is usually defined as visible EM radiation of the entire EM spectrum where the human eye is sensitive [37, 38, 39, 40], but the term light is often extended to adjacent wavelength ranges that the eye cannot detect [40]. Spectroscopy can detect a much wider region of the EM spectrum than the Vis range and a common laboratory spectrophotometer can detect wavelengths from 200 to 2500 nm.

2.2.2 Photon absorption and fluorescence phenomena

Photons that propagate inside a medium can be absorbed by the molecules in the sample.

Absorption: During absorption spectroscopy, an incident photon can be absorbed by a molecule, which leads to the photon energy being converted into an excitation of that molecule’s electron cloud. This interaction is sensitive to the internal structure of the molecule since the laws of quantum mechanics only allow for the existence of a limited number of excited states of the electron cloud of any given chemical species. Each of these excited states has a defined energy, the absorption of the photon has to bridge the energy gap between the ground state (lowest energy state) and an allowed excited state of the electron cloud. As a consequence, molecules can be identified by their absorption spectrum, their wavelength-dependent capacity for absorbing photons depends on the energy spacing of the states of their electron cloud. If the frequency of the radiation matches the vibration frequency of the molecule, then radiation will be absorbed, causing a change in the amplitude of molecular vibration [41]. Molecules, which strongly absorb Vis light, appear colored to the human eye and are, therefore, called chromophores, that is “carriers of color”.

Fluorescence: In absorption, the signal of molecules in a sample is direct but it can be done with higher sensitivity by using an indirect approach, fluorescence detection. Then the ingoing light will give the absorbing molecules excited states (higher energy) of their electron cloud as described above. From this state, the molecule can shift, “relax”, to the electronic ground state by transforming the excess energy into an outgoing emitted light having longer wavelengths than the ingoing light. Different molecules have different emitted spectrums, which being applied during the fluorescence measurements [42].

2.2.3 Beer-Lambert law

The Beer–Lambert law states that the absorbance of light intensity is proportional to the concentration of the substance. This means that the amount of, for example, UV-light absorbed when passing through a cuvette (manufactured by UV-transparent material such as quartz) with spent dialysate, Figure 2, is linearly dependent on the concentration c [mol/L] of the absorbing solute, the optical pathlength in (l) [m] (depth of the cuvette) and the extinction coefficient ε [m⁻¹ (mol/L)⁻¹], even called the molar absorptivity at a certain wavelength [43].

Figure 2.

Cuvette with a sample containing an absorber with concentration, c, ingoing light, I₀, and outgoing light, I, after absorption in the sample when passing through the cuvette with the optical path length, l.

If I₀ is the intensity of the incident light and I is the intensity transmitted light through the medium, the absorbance (A), dimensionless, is Eq. (1):

$$\mathrm{A}={log}_{10}\left[\frac{{\mathrm{I}}_0}{\mathrm{I}}\right]=\mathrm{\epsilon }·\mathrm{c}·\mathrm{l}$$E1

If ε is known for a substance, the absorbance (A) can be calculated by multiplying the path length and the concentration of the substance. If ε is known for a substance and A is obtained from a measurement, it is possible to derive the concentration as Eq. (2):

$$\mathrm{c}=\frac{\mathrm{A}}{\mathrm{\epsilon }·\mathrm{l}}$$E2

In our case, when the spent dialysate contains several different absorbing compounds, the overall extinction coefficient is the linear sum of the contributions of each compound. However, all the components are not identified and probably there is interference between different substances, which makes it difficult to separate and determine the concentrations of each solute. Absorbance of a solution, obtained by a double beam spectrophotometer, is given by the Lambert–Beer law as [44, 45] Eq. (3):

$$\mathrm{A}=log\frac{{\mathrm{I}}_0}{{\mathrm{I}}_{\mathrm{r}+\mathrm{s}}}-log\frac{{\mathrm{I}}_0}{{\mathrm{I}}_{\mathrm{r}}}=log\frac{{\mathrm{I}}_{\mathrm{r}}}{{\mathrm{I}}_{\mathrm{r}+\mathrm{s}}}$$E3

where I₀ is the intensity of incident light from the light source, I_r is the intensity of transmitted light through the reference solution (e.g. pure dialysate) and I_r + s is the summated intensity of transmitted light through the reference solution mixed with the solution (e.g. pure dialysate + waste products from the blood). The common assumptions to utilize absorbance calculated according to the Beer–Lambert law to determine concentration are: (1) the radiation is monochromatic [34, 43, 46, 47], (2) the irradiating beam is parallel (collimated) across the sample [34, 43], (3) the absorption of radiation for a given species is independent of that of other species [34, 43], (4) only the non-scattered and not-absorbed photons are detected from the medium [40], and (5) the incident radiation and the concentration of the chromophores are not extremely high [46, 48]. The Beer–Lambert law for monochromatic light can be derived by solving a differential equation for a solution with a finite depth containing chromophores and is given in detail in many sources [34, 37, 46]. Analysis of a mixture is based fundamentally on the fact that the absorptions, at each wavelength, of separate components in the mixture are additive, provided that chemical or interfering physical reactions between the components do not occur [43] and the solute concentrations are not very high (not usually found in biological media). In this case, in a medium containing n different absorbing compounds with the concentrations of c₁…c_n [mol/L] and the extinction coefficients of ε₁…ε_n [m⁻¹ (mol/L)⁻¹], the overall extinction coefficient is simply the linear sum of the contributions of each compound (Eq. (4)):

$$\mathrm{A}={log}_{10}\left[\frac{{\mathrm{I}}_0}{\mathrm{I}}\right]=\left({\mathrm{\epsilon }}_1{\mathrm{c}}_1+{\mathrm{\epsilon }}_2{\mathrm{c}}_2+\dots +{\mathrm{\epsilon }}_{\mathrm{n}}{\mathrm{c}}_{\mathrm{n}}\right)\mathrm{l}$$E4

2.2.4 Transmittance and absorbance

The amount of transmitted and absorbed portion of EM radiation in the medium (e.g. in a dialysate sample) can be characterized by the parameter’s transmittance T and absorbance A in the spectrophotometer. In order to utilize EM radiation for measurement of constituents in a fluid, the sample is applied in an optical cuvette, Figure 2. Through the calibration and measurement procedures, one determines the amount of the ingoing light illuminating the sample symbolized by I₀ and the outgoing light I symbolizing the intensity of the light after passing the sample as the remaining ingoing light is partly absorbed by the sample, Figure 2. Having knowledge about those parameters, one can determine transmittance T (Eq. (5)), and absorbance A (Eq. (6)) as:

$$\mathrm{T}=\frac{\mathrm{I}}{{\mathrm{I}}_0}\Rightarrow \%\mathrm{T}=\frac{\mathrm{I}}{{\mathrm{I}}_0}·100\%$$E5

$$\mathrm{A}=-\mathit{log}\mathrm{T}=-\mathit{log}\frac{\mathrm{I}}{{\mathrm{I}}_0}=\mathit{log}\frac{{\mathrm{I}}_0}{\mathrm{I}}$$E6

3. Dialysis monitoring utilizing UV absorbance technique

Several solutes in spent dialysate identified as uremic toxins by Prof. Vanholder and colleagues in the European Uremic Toxin Work Group (EUTox) have been measured by applying UV absorbance in HPLC technique [49, 50, 51, 52]. Currently, two optical techniques for dialysis dose monitoring have been investigated—UV absorbance and NIRS [53, 54]. Other approaches have also been investigated but are more limited so far, for example utilizing the Vis region for measurement. In this chapter, the UV absorbance technique will be discussed that is applied in the commercialized dialysis adequacy monitors ADIMEA [55] and DDM [56]. We will present how clinical parameters based on urea, such as Kt/V, urea reduction ratio (URR), total removed urea (TRU) and protein catabolic rate/protein nitrogen appearance (PCR/PNA), can be estimated optically utilizing UV absorbance.

3.1 Present clinical parameters from optical dialysis dose monitoring

Urea kinetic modeling (UKM), that is where urea is used in differential equations, with the attempt to provide quantitative assessments of dialysis and nutrition adequacy in dialysis patients. The high correlation between urea concentration and UV absorbance values gives consequently the possibility to utilize the UKM equations for UV absorbance similarly. During the UV absorbance and on-line measurements, the pure dialysate was used as the reference, (Eq. (3)) I_r = pure dialysate, and the wavelength was fixed for the entire dialysis. The absorbance baseline level was after the pure flowing dialysate had been stabilized in temperature and conductivity, set to zero when the pure dialysate was flowing through the cuvette prior treatment, see Figure 3 for the schematic set-up.

Figure 3.

The schematic set up during the studies. All spent dialysate was also collected in a collection tank to be able to calculate total removal of solutes. [Reprinted from [6]].

3.1.1 Kt/V estimation from UV absorbance

Assuming that urea is distributed in a single pool volume in the body, that urea generation rate and ultrafiltration are negligible during the session and that the ratio K/V remains constant over the dialysis, the following equation holds [57, 58] (Eq. (7)):

$$\mathrm{Kt}/\mathrm{V}=-ln\frac{{\mathrm{C}}_{\mathrm{t}}}{{\mathrm{C}}_0}$$E7

where C_t is the post and C₀ is the pre-dialysis urea blood concentration, respectively.

From the differential equation, describing urea mass balance during a dialysis session, it can be determined that the average value of the urea clearance (K) in mL/min divided with urea distribution volume in the body (V) in mL (K/V) during a session may be approximated as the slope from the natural logarithm (ln) plot of the urea blood concentration in the blood versus time, S_B. Similarly, but instead of blood urea concentrations, the concentrations of urea in dialysate (S_D) can be used (Eq. (8)). Hence:

$$\mathrm{Kt}/\mathrm{V}\approx -{\mathrm{S}}_B\mathrm{T}\approx -{\mathrm{S}}_D\mathrm{T}$$E8

where T is the dialysis session length in minutes. According to Eq. (8), we obtain Eq. (9):

$$\frac{{\mathrm{C}}_{\mathrm{t}}}{{\mathrm{C}}_0}\approx exp\left(-\mathrm{Kt}/\mathrm{V}\right)\approx exp\left({\mathrm{S}}_{\mathrm{B}}\mathrm{T}\right)\approx exp\left({\mathrm{S}}_{\mathrm{D}}\mathrm{T}\right)\approx exp\left({\mathrm{S}}_{\mathrm{a}}\mathrm{T}\right)$$E9

If the slopes are used instead of the blood urea concentrations. This approximation is equivalent to the equation when two measuring points are used, and the previously mentioned assumptions are fulfilled. This equation would hold strictly if urea obeys fixed volume and single pool kinetics and no urea is generated during the session [59]. In order to calculate Kt/V from the online UV absorbance, the slope of blood or dialysate urea concentration was replaced by the ln slope of the UV absorbance, Sa, see Eq. (9) versus time (Kt/V ≈ − Sa*t, Figure 4), [7].

Figure 4.

Online absorbance curve during a single 4 hours HD treatment, where UV absorbance is plotted against time. The corresponding natural logarithmic (ln) fitting line is also shown and used for Kt/V calculation [Reprinted from [7]].

Using the UV absorbance slope values (Sa), Figure 4, according to Eq. (9), the Daugirdas-based mono compartmental (single pool, sp) Equation (Eq. (10)) [60]:

$$\mathrm{sp}\left(\mathrm{Kt}∕\mathrm{V}\right)=-ln\left(\frac{{\mathrm{C}}_{\mathrm{t}}}{{\mathrm{C}}_0}-0.008\frac{\mathrm{T}}{60}\right)+\left(4-3.5\frac{{\mathrm{C}}_{\mathrm{t}}}{{\mathrm{C}}_0}\right)\frac{\mathrm{UF}}{\mathrm{BW}}$$E10

can be written as [7] Eq. (11):

$$\mathrm{sp}\left(\mathrm{Kt}/\mathrm{Va}\right)=-ln\left(exp\left({\mathrm{S}}_{\mathrm{a}}\mathrm{T}\right)-0.008\frac{\mathrm{T}}{60}\right)+\left(4-3.5exp\left({\mathrm{S}}_{\mathrm{a}}\mathrm{T}\right)\right)\frac{\mathrm{UF}}{\mathrm{BW}}$$E11

where UF and BW are the ultrafiltration volume in liters (L) and the patient’s dry body weight in kg. The equilibrated Kt/V from UV absorbance, eKt/Va, according to the rate adjustment method [60], is predicted from the rate of dialysis (K/V) and the sp(Kt/Va) as Eq. (12):

$$\mathrm{eKt}/\mathrm{Va}=\mathrm{sp}\left(\mathrm{Kt}/\mathrm{Va}\right)-\frac{0.6}{\left(\frac{\mathrm{T}}{60}\right)}\mathrm{sp}\left(\mathrm{Kt}/\mathrm{Va}\right)+0.03$$E12

The rate adjustment method predicts that the urea rebound is related to the rate of dialysis or dialysis efficiency [61].

3.1.2 Estimation of urea removal using UV absorbance

One way to estimate total removed urea (TRU), assuming that the dialysate flow, Qd(t), is constant and the total UF is known, is to use the following equation (Eq. (13)):

$$\mathrm{TRU}\left(\text{mmol}\right)=\overline{\text{Urⅇa}}\left(\mathrm{Qd}\bullet \mathrm{T}+\mathrm{UF}\right)$$E13

where $\overline{\text{Urea}}$ in mmol/L is the mean urea concentration in spent dialysate of a particular HD session [62]. For the TRU calculations, Urea = Dtotal can be utilized as reference, where Dtotal is the urea concentration in the collection tank (all spent dialysate from one session), after the end of dialysis. Qd is the rate of the dialysate flow in L/min, T is the dialysis session length in minutes and UF is the total ultra-filtrated volume in L during the session. Under the condition that a good correlation exists between UV absorbance and concentration of urea, it is possible to utilize this relationship. Therefore, in a similar way, TRU may be calculated from the online UV absorbance curve (Figure 4) as Eq. (14) [63]:

$$\text{TRUa}\left(\text{mmol}\right)=\left(\mathrm{\alpha }·\overline{\mathrm{A}}+\mathrm{\beta }\right)·\left(\mathrm{Qd}\bullet \mathrm{T}+\mathrm{UF}\right)$$E14

where $\overline{\mathrm{A}}$ is the mean of all UV absorbance (A) values from the start to the end of the dialysis. The regression line between the UV absorbance and concentration of urea in spent dialysate from online measurement gives the slope (α) and the intercept (β) inserted in Eq. (14) when determining TRUa from a general model. TRU from the total dialysate collection (TDC), reference, was calculated as Dtotal (mmol/L) multiplied with collected weight (kg), assuming that 1 kg = 1 L of the dialysate [63].

3.1.3 Estimation of nutrition parameters from UV absorbance

High concentration of urea in blood is not necessarily related to a poor dialysis outcome if urea removal is sufficient [64], but also protein-energy malnutrition is frequently present in HD patients. Several studies have suggested that malnutrition is an important risk factor for morbidity and mortality in HD patients [65]. In order to optimize the diet of patients with renal diseases, dietary protein intake has to be controlled. Protein nitrogen appearance (PNA), formerly protein catabolic rate (PCR) [66], is easily obtainable from UKM and in patients who are not markedly catabolic or anabolic, the normalized PNA (nPNA) correlates closely with dietary protein intake [67, 68]. These parameters can be calculated from TRU. The PCR calculation, from TDC and UV absorbance, was based on a theory by Garred et al. (1995), where a calculation of urea removal is expressed as a fraction of the week’s urea generation. The fraction varies with the day of the week and was found to be essentially constant among patients on a given day [69]. The amount of urea could, therefore, be approximated by measuring urea concentration from only one of the three treatments and PCR could be calculated as Eq. (15) [63]:

$$\text{nPCRw}={\text{Factor}}_{1,2\text{or}3}\left(\frac{\mathrm{T}{\mathrm{RU}}_{1,2\text{or}3}}{\mathrm{BW}}\right)+0.17$$E15

where TRU 1, 2 or 3 (expressed in grams of urea nitrogen) is the TRU from the first (1), midweek (2) or last dialysis in week (3) and Factor one, two or three is the fractional factor for the corresponding days; factor one = 2.45; two = 2.89 and three = 3.10 [69]. Obligatory loss of dietary protein in stools and via skin shedding represents the constant term 0.17 (g protein/kg body weight/day). BW was used for normalization of PCR (nPCRw). Observe that these fractional factors relay to a treatment schedule of three times a week. More frequent dialysis treatments are common today whereas the factors are not appropriate.

4. Monitoring uremic toxins beyond urea

The fact that urea is considered generally a nontoxic substance and only a marker for uremic retention solutes, and the EUTox has identified several more relevant uremic toxins utilizing optical analysis methods, arise a question connected to further development of optical techniques “Could some of these identified uremic toxins be measured optically on-line?” Of the 90 compounds that have been identified as uremic toxins by the EUTox group [51, 52, 70], we have, from spectroscopic databases, identified 36 to be UV absorbing and among them approximately 25 to be absorbing near 297 nm. In addition to these 90 compounds mentioned as uremic toxins, there are even more solutes in the dialysate that are optically active at 297 nm, which add to the measured UV absorbance signal. Spent dialysate contains numerous different absorbing compounds and concentrations of solutes decline differently during a dialysis session for individuals. The UV absorbance curve may therefore be an individual “clinical print” of the patient’s sum of several UV absorbing solutes and therefore a possible parameter for monitoring total solute removal during dialysis. Analyses still remain to find the single solute’s individual contribution to the absorbance signal, which is also dependent on which wavelengths are used [71]. Earlier knowledge from the correlation analysis between UV absorbance and a few uremic toxins has shown that it is possible to estimate removal of such solutes, which have a high correlation to UV absorbance. This removal of other uremic retention solutes beyond urea may have stronger impact on dialysis outcomes compared to urea or urea alone.

4.1 Uremic toxins

Uremic toxins used to be divided into three different molecule groups, that is small, water soluble, middle and protein-bound [72], but it needs still to work out their relation to dialysis efficacy and what roles different molecules will have in the development of uremia and which dialysis techniques are best at reducing the elevated levels of these [73]. Proposal for a new classification system of uremic solutes rationale has been made [74]. Declined uremic toxin clearance due to low GFR is not the only cause of toxin accumulation in kidney failure. Excessive production of cytokines and soluble receptors due to local tissue inflammation is a major contributor to the middle molecule accumulation [75], and gut dysbiosis generates a broad spectrum of uremic toxins [76]. Thus, a broader view of uremic solutes that goes beyond simply retention with poor GFR is needed. Recent data regarding the origin of uremic toxins, and the new development of HD methods and new membranes with the ability to clear uremic toxins with specific characteristics, or by using drugs/molecules to facilitate the shift from bound fraction to free fraction [77], have led to propose a new classification beyond the classic physicochemical classification. A study by Vanholder et al. (2018) presents a ranking score list of uremic toxins with known toxicity according to experimental and clinical studies [78], Table 1 shows the highest- and second-highest evidence-scored uremic toxins.

Highest evidence score	Second highest evidence score
p-cresyl sulfate	Advanced glycation end products (AGEs)
β2-microglobulin	Indoxyl sulfate
Asymmetric dimethyl arginine	Uric acid
Carbamylated compounds	Ghrelin
Fibroblast growth factor-23	Indole acetic acid (IAA)
IL-6	Parathyroid hormone
TNFα	Phenyl acetic acid
Symmetric dimethyl arginine	Trimethyl methylamine-N-oxide
	Retinol binding protein
	Endothelin
	Immunoglobulin light chains
	IL-1β
	IL-8
	Neuropeptide Y
	Lipids and lipoprotein

Table 1.

Uremic toxins with the highest toxicity evidence score (modified from Vanholder et al. 2018) [78].

The uremic toxins with the optical monitoring capacity referred to further in this chapter are marked in underline. AGEs and IAA are in italics as the potential targets for the optical monitoring in the future, Table 1.

4.1.1 Optical intradialytic monitoring of small water-soluble molecules’ removal

Uric acid, like urea, a representative molecule from the small group of uremic toxins, has been shown to have a high absorption of UV in the wavelength region 280–310 nm (also a peak around 230–240 nm), Figure 5a. As a consequence, we have been able to show that it is possible to estimate the total removal of uric acid during dialysis [79, 80] and a multi-wavelength and processed signal approach can provide even more accurate results [81]. Figure 5b shows an example of the best-fit regression equation of uric acid vs. UV absorbance in spent dialysate at wavelength 285 nm during four dialysis sessions in the same patient, showing a high correlation of r = 0.99.

Figure 5.

(a) The molar absorptivity for four solutes, β2M, uric acid, creatinine and urea in 24 HDF sessions. (b) An example of the regression line between concentration of uric acid and UV-absorbance [Figure 5a, reprinted from [86]].

Earlier research has demonstrated potential of intradialytic optical monitoring to estimate the removal of low molecular weight uremic solutes other than uric acid as urea [53] and creatinine [82] with their clinical implications. Furthermore, optical monitoring of low molecule weight uremic solutes removal by HD, assessed via the marker molecule urea-related dialysis adequacy parameters, has become a worldwide practice [55, 56, 83]. A possibility for optical monitoring of phosphate and calcium elimination during dialysis has also been presented [84, 85].

4.1.2 Optical intradialytic monitoring of middle molecules’ removal

Using UV absorbance alone to estimate β2M does not appear to be optimal even a high correlation between UV absorbance and β2M in spent dialysate can be achieved for HDF but not for HD [86]. Instead, fluorescence spectra could be a better alternative [87] as it is possible to detect the fluorescence of advanced glycation end products (AGE) modified β2M in spent dialysate [88, 89]. However, the measuring system needs high selectivity and sensitivity for detection due to low contribution of AGE modified β2M to overall fluorescence [89]. The best correlation between the fluorescence of spent dialysate and the concentration of β2M in spent dialysate was found in the wavelength region Ex350–370/Em500–555 nm, with the coefficient of determination R² up to 0.859, Figure 6a.

Figure 6.

a. Wavelength dependence of the correlation between fluorescence intensity and concentration of β2M in spent dialysate for HDF modalities (N = 375). b. Time-series of changing β2M concentration (mean ± SD) in the spent dialysate during HDF dialysis sessions (N = 29) for patients of the validation set [Reprinted from [90]].

A multiwavelength fluorescence approach can yield a high correlation of up to 0.958 between laboratory and optically estimated β2M concentrations in spent dialysate. The main contributors to the optical signal of the middle molecule (MM) fraction were provisionally identified as tryptophan (Trp) in small peptides and proteins and AGEs [90]. Figure 6b visualize the good agreement between β2M concentrations analyzed in laboratory vs. those predicted by an optical model during 29 four-hour HDF sessions.

4.1.3 Optical intradialytic monitoring of protein-bound molecules’ removal

Recent studies have presented that quantification of indoxyl sulphate (IS) in the spent dialysate using fluorescence spectra is possible [91, 92]. Figure 7 shows an example of a HPLC chromatogram of a spent dialysate sample taken 10 min after the start of HD, where the fluorescence was recorded at Ex: 280 nm and Em: 360 nm. In total, 12 clearly resolved chromatographic peaks of fluorophoric compounds were detected in most (82%) of the spent dialysate samples during the HPLC analysis collected at different time moments in the dialysis [92]. Of these, five peaks had a major importance in all samples (peaks no 6, 8, 9, 11 and 12), and six of these 12 peaks were identified as Trp and their metabolites of indole derivates: indoxyl glucuronide (IGluc), IS, 5-hydroxy-indole-3-acetic acid, indole acetyl glutamine (IaG) and indole acetic acid (IAA) [92]. IS is one of the main fluorophores in these measuring conditions (Figure 7, peak 9) [92]. This is in agreement with the earlier studies, where IS has been found as a main contributor to the fluorescence in uremic fluids [93, 94].

Figure 7.

Example of a chromatogram of a spent dialysate sample, where peak 9 is indoxyl sulphate. [Reprinted from [92]].

4.2 Multi-component uremic toxins’ intradialytic optical monitoring of spent dialysate

Several in vitro and online studies towards optical multi-component uremic toxins’ monitoring have been published by our group during the last 10 years [9, 95, 96, 97, 98] and by a group from Taiwan [99]. The studies demonstrate that optical dialysis monitoring, based on UV absorbance and fluorescence of spent dialysate, can simultaneously reveal removal patterns of, for example, urea, β2M and IS during various dialysis treatment modalities without any blood or dialysate sampling, Figure 8.

Figure 8.

An example representing real-time concentration profiles for urea, β2M (B2M) and IS during a single dialysis from optical measurements in the spent dialysate in parallel with the discrete concentration values estimated from the laboratory analyses of spent dialysate samples at different time moments.

A good agreement between chemically and optically estimated solute removal parameters, RR and total removed solute, was achieved. Dialysis modality did not affect the accuracy of optical method, taking into account that β2M was excluded from the analysis in the case of dialysis with low-flux dialyzer [9]. Figure 9 shows the agreement in RR (%) for urea, β2M and IS between measurements from laboratory and the developed online optical dialysis adequacy sensor (OLDIAS).

Figure 9.

A comparison presenting the agreement in RR for urea (Urea), β2M (B2M) and indoxyl sulphate (IS) between measurements from laboratory and the online optical dialysis adequacy sensor (OLDIAS).

4.2.1 Drug interference during the optical dialysis monitoring in spent dialysate

There have been indications that the administration of some drug chromophores, for example paracetamol (Par), to dialysis patients could disturb the accuracy of the optical methods since a noteworthy contribution of Par and its metabolites to the total UV absorbance was determined at three wavelengths 210, 254 and 280 nm [100], where the latter is used in the commercial monitor [56]. Adoberg et al. confirmed that the administration of Par in large amounts increased the UV absorbance of spent dialysate, which can result in overestimation of concentration and the RR of uric acid (UA) when evaluated by UV absorbance of spent dialysate, using the UV region that overlaps with the Par-absorption spectrum [101]. At the same time, the correlation between the IS concentration and fluorescence in the spent dialysate is not affected by Par administration to dialysis patients, neither is the optical assessment of the RR of IS on the basis of the fluorescence of spent dialysate. Figure 10 illustrates a representative chromatogram of the spent dialysate of a patient with high Par intake (twice 1 g of Par before dialysis, 1 g during the dialysis session and four times 1 g on the previous day) and the UV absorbance spectra of Par, Par metabolites and UA peaks on the insert.

Figure 10.

Characteristic HPLC UV 295 nm chromatogram of spent dialysate of one patient with high Par intake. Insert: UV-absorbance spectra of peaks of uric acid (UA), paracetamol glucuronide (ParG), paracetamol (Par), paracetamol sulphate (ParS), and indoxyl sulphate (IS) [Reprinted from [101]].

These limitations could be overcome by using multiparametric optical models that incorporate several UV wavelengths in order to evaluate the removal of UA, and also urea, or using the UV region, such as 295 nm, to minimize the influence of Par. Conventionally prescribed drugs in connection with dialysis treatment did not interfere with the optical monitoring of the treatment [101].

4.2.2 Future directions in optical monitoring

The future vision of the dialysis optical monitoring technology is moving towards the ability to estimate efficacy measures of several important uremic toxins, for example, Figure 8 that are linked to morbidity and survival for dialysis patients.

Attempts of optical intradialytic monitoring of AGE’s have also been carried out within our group, showing no difference between average of free pentosidine concentrations in the spent dialysate measured by HPLC and models, developed from the full fluorescence spectra [102]. Tryptophan, which originates uremic toxins that contribute to end-stage kidney disease (ESKD) patient outcomes, may also be a target for future monitoring. Paats et al. evaluated serum levels and removal during HD and haemodiafiltration (HDF) of tryptophan and tryptophan-derived uremic toxins, indoxyl sulphate (IS) and indole acetic acid (IAA), in ESKD patients in different dialysis treatment settings [103]. High-efficiency HDF resulted in 80% higher Trp losses than conventional low-flux dialysis, despite similar neutral Trp RR values. In conclusion, serum Trp concentrations and RR behave differently from uremic solutes IS, IAA and urea and Trp RR did not reflect dialysis Trp losses. Conventional low-flux dialysis may not adequately clear Trp-related uremic toxins while high-efficiency HDF increased Trp losses [103]. Furthermore, by adding chemical displacers, combining ibuprofen and furosemide, during HDF, the removal of protein-bound uremic toxins can be enhanced [104, 105].

Finally, the technology will be integrated into the dialysis machines of the future for simple handling and easy monitoring over time, Figure 11.

Figure 11.

An example of a future display on the dialysis machine where multiple performance measures (removal ratio, total removed amount, Kt/V) for β2M (B2M) (Courtesy of Optofluid Technologies OÜ, Estonia, with permission).

5. Conclusion

Uremia occurs due to retention of a variety of substances with different properties. Historically, we have had one marker, urea, to describe the total uremic environment in dialysis patients. There is a need to follow a wider spectrum of uremic toxins that are more strongly linked to morbidity and survival than urea alone. Our ambition with this research, using combined optical methods, is to be able to monitor several important uremic toxins online in the future. Patients will benefit from more patient-centred, integrated and cost-efficient care, as feedback for clinicians helps to improve and personalize the treatment quality, minimizing costly adverse effects. Feedback to patients helps to integrate patients into their own treatment, increasing patients’ compliance and well-being.

Acknowledgments

Thanks to all colleagues who have been involved in the projects: Risto Tanner, Merike Luman, Jana Holmar, Jürgen Arund, Sigrid Kalle, Ruth Tomson, Rain Ferenets, Joosep Paats, Kai Lauri, Annika Adoberg, Liisi Leis, Mårten Segelmark, Anders Fernström, Kristjan Pilt, Deniss Karai, Micael Gylling and Kelly Stjernfelt, Martin Magnusson, Lars-Göran Lindberg.