**4. HPLC as an analysis tool for biological samples**

146 Progress in Hemodialysis – From Emergent Biotechnology to Clinical Practice

The examples of two online signals measured at 280 nm from two different dialysis treatments are shown in Figure 2. The UV absorbance is higher at the beginning of treatment because of the high concentration of metabolic waste products in the body fluids. When the waste products are removed from the blood the UV absorbance decreases during the dialysis session. The times at which the blood and dialysate samples were collected are also shown in this figure. Blood samples were drawn before the start of dialysis treatment (Bstart) and immediately after the treatment (Bend). Dialysate samples were taken 10 minutes after the start of the dialysis session (Dstart) and at the end of the treatment (Dend) (210 or 240 minutes).

Fig. 2. Example of two different online UV absorbance measurements (at 280 nm). Time points when the samples were taken for later analysis are as follows: Dstart - dialysate sample collected 10 minutes after start of hemodialysis, Dend- dialysate sample collected at the end of hemodialysis, Bstart – blood sample collected before dialysis session, Bend – blood sample

This figure presents a classic picture of UV absorbance signals obtained by the optical online

As seen from Figure 2, the UV absorbance curves are somewhat different. The exponential decrease of the UV absorbance curve represents the elimination rate of the all UV-absorbing compounds – chromophores – which varies from patient to patient and from treatment to treatment. Good correlation between the UV absorbance measured in the spent dialysate and the concentration of several uremic retention solutes, both in the spent dialysate and in the blood of the dialysis patients, has been previously shown (Fridolin, 2002). For these reasons the origin of the cumulative and integrated UV absorbance arising from the contribution of uremic retention solutes, among them probably several uremic toxins, should be investigated.

dialysis dose monitor, which can be used for estimation of uremic solutes removal.

collected at the end of hemodialysis

High performance liquid chromatography (HPLC) is a technique of analytical chemistry which can separate and identify the components of a mixture of different chemical compounds in liquid solution. The reversed-phase HPLC technique is the most commonly used form of HPLC. This method is recommended as a sensitive, accurate and reproducible tool for qualitative and quantitative analysis of aqueous samples (Vanholder, 2001). Furthermore, the use of ambient temperature in reversed-phase columns makes it possible to investigate the many non-volatile or thermally unstable compounds commonly found in biological samples. The principle involved in HPLC testing enables the separation of compounds in a mixture more efficiently and faster than that of traditional column chromatography.

In general, the HPLC system consists of two essential components – a stationary phase and a mobile phase. The stationary phase is a column packed with small solid sorbent particles and where the separation of different compounds takes place. The mobile phase is a flowing liquid (solvent) that transports the compounds from the sample through the stationary phase. Thus, the compounds of the mixture travel at different rates due to their relative affinities with the solvent and stationary phase. Separation of compounds in the stationary phase occurs with slight differences in chemical properties, such as chemical polarity and size of non-polar groups.

Figure 3 is a simple block diagram illustrating the main components of a modern HPLC. A system consists of several units: pumping, sample-injection, separation (column), detection and data-processing.

Fig. 3. Schematic reversed-phase HPLC method principle

The results of chromatographic analysis are known as chromatograms, where the signal intensity from the detector is recorded on the time axis. For HPLC there are several different detection methods; the most popular are optical. An ultraviolet-visual light (UV-VIS) absorption detector is the main optical detection method. This detector is effective in the detection of components with an absorption wavelength of 400 nm or less in the ultraviolet

Optical Dialysis Adequacy Monitoring:

presence of different-sized molecules in the serum.

the collection and handling of blood and dialysate samples.

Fig. 4. Block diagram of blood and dialysate sample collection and handling

estimated.

Small Uremic Toxins and Contribution to UV-Absorbance Studied by HPLC 149

Ten uremic patients – three females and seven males (mean age 62.6 ± 18.6 years) – participated in the study. All of the patients were investigated over a course of 30 hemodialysis treatments. Three different polysulphone dialysers were used (Fresenius Medical Care, Germany): a low flux membrane dialyser F8 HPS (N=14), a low flux membrane dialyser F10 HPS (N=3) and a high flux membrane dialyser FX 80 (N=11). The elimination of toxins by the different types of semi-permeable membranes was also

Blood samples were drawn before the start of dialysis treatment (Bstart) and immediately after the treatment (Bend) (Figure 2) using the slow flow/stop pump sampling technique. Blood was sampled into BD Vacutainer® Glass Serum Tube (red cup, Beckton Dickinson) and was allowed to clot. After centrifugation at 3000 r.p.m. the serum was ready for clinical chemistry analysis. Small molecules – creatinine (MW=113.12 Da), uric acid (MW=168.11 Da) and urea (MW=60.06 Da) – were measured at the Clinical Chemistry Laboratory of the North Estonia Medical Centre using standardised methods (Hitachi 912 autoanalyser, Roche, Switzerland). The accuracy for creatinine was 5%, for uric acid 2% and for urea 4%. For HPLC analysis the additional pre-treatment of the centrifugated serum was necessary to release the blood proteins. The serum samples were purified from proteins by centrifuging with two different Microcon centrifugal filters (Millipore, USA): a YM-3 with cut-off 3 kDa and a YM-100 with cut-off 70 kDa. The aim of using different filters was to estimate the

Dialysate samples were taken 10 minutes after the start of the dialysis session (Dstart) and immediately after the treatment (Dend) (210 or 240 minutes) (Figure 2). Also, pure dialysate, used as the reference solution, was collected before each dialysis session, when the dialysis machine was being prepared and the conductivity was stable. For HPLC analysis, the dialysate samples were acidified with formic acid to pH 4.0. Sample sizes of 50 μL or 100 μL were used for chromatographic separation. Figure 4 presents a block diagram illustrating

region. There are three types of UV-VIS detectors: a fixed wavelength detector; a variable wavelength detector; and a diode array detector (DAD). However, in modern HPLC systems DAD is commonly used, which in addition to the UV-VIS signal offers the ability to produce an absorption spectrum for every time slice during the chromatogram. A DAD detects the absorption in the UV to VIS region. Using the DAD, absorption on a large number of wavelengths can be measured simultaneously. This enables us to select the best wavelengths for actual analysis. Additionally, one of the advantages of DAD is that it allows us to detect sufficiently pure peaks. Often the peak shape itself does not reveal that it actually corresponds to two or even more components. In such a case, absorbance rationing on several wavelengths is particularly helpful in deciding whether the peak represents a single compound or a composite peak. However, compared with a UV-VIS detector, the DAD is susceptible to changes, such as lamp fluctuations, and noise is large because the amount of light is small.

For those compounds which fluoresce or which exhibit an appropriate fluorescence due to derivatisation, a fluorescence detector is the most sensitive of the existing modern HPLC detectors. Its sensitivity is 10-1000 times higher than that of the UV detector for strong UVabsorbing materials. Fluorescence detectors are very specific and selective compared to other optical detectors. A universal detector, but the least sensitive for non-ionic compound monitoring, is the refractive index detector. To estimate oxidisable and reducible compounds, the most suitable is the electrochemical detector.

Reversed-phase HPLC has been found to be useful in the analysis of uremic biofluids. A number of authors have reported the application of the HPLC method in the analysis of dialysate and serum samples already in several decades ago (Schoots, 1989; Schoots, 1982). Different detectors have been used. Some naturally fluorescent compounds have been separated and identified with a fluorescence detector, in both serum and haemodialysate (Niwa, 1993; Barnett, 1985; Swan, 1983). A liquid chromatographic method including a UV detector for detection of UV-absorbing solutes in uremic serum has been developed (Schoots, 1985; Senftleber, 1976; Knudson 1978). A fixed wavelength at 254 nm was mainly used in these studies. Additionally, several types of chromatographic methods, such as gas chromatography mass spectrometry (GC-MS), for the detection of uremic toxins has also been developed (Niwa, 1997; Niwa, 2009). Mass spectrometry (MS) has been applied to the identification and quantification of uremic toxins and uremia-associated modified proteins (Niwa, 2009). However, GC-MS cannot analyse highly polar, thermally labile and highmolecular weight compounds and usually requires sample preparation, such as extraction or derivation to make non-volatile compounds thermally stable and volatile. Compared with GC-MS, liquid chromatography mass spectrometry (LC-MS) can separate and identify highly polar, thermally labile or high-molecular weight mixture compounds and does not require derivation; sample preparation is also simple. Newly developed LC-MS techniques have been successfully applied to uremic toxin research with the discovery of novel uremic toxins that range from low-molecular weight solutes to small-molecule proteins (Niwa, 2011). This new analytical method is available today, opening up new horizons for uremic toxin identification detected earlier as unidentified HPLC peaks.

### **5. HPLC study of UV absorbance profiles**

In order to explain the origin and potential of the online UV absorbance dialysis dose monitoring method, the HPLC method of analysing UV absorbance profiles was developed.

region. There are three types of UV-VIS detectors: a fixed wavelength detector; a variable wavelength detector; and a diode array detector (DAD). However, in modern HPLC systems DAD is commonly used, which in addition to the UV-VIS signal offers the ability to produce an absorption spectrum for every time slice during the chromatogram. A DAD detects the absorption in the UV to VIS region. Using the DAD, absorption on a large number of wavelengths can be measured simultaneously. This enables us to select the best wavelengths for actual analysis. Additionally, one of the advantages of DAD is that it allows us to detect sufficiently pure peaks. Often the peak shape itself does not reveal that it actually corresponds to two or even more components. In such a case, absorbance rationing on several wavelengths is particularly helpful in deciding whether the peak represents a single compound or a composite peak. However, compared with a UV-VIS detector, the DAD is susceptible to changes, such as lamp fluctuations, and noise is large because the

For those compounds which fluoresce or which exhibit an appropriate fluorescence due to derivatisation, a fluorescence detector is the most sensitive of the existing modern HPLC detectors. Its sensitivity is 10-1000 times higher than that of the UV detector for strong UVabsorbing materials. Fluorescence detectors are very specific and selective compared to other optical detectors. A universal detector, but the least sensitive for non-ionic compound monitoring, is the refractive index detector. To estimate oxidisable and reducible

Reversed-phase HPLC has been found to be useful in the analysis of uremic biofluids. A number of authors have reported the application of the HPLC method in the analysis of dialysate and serum samples already in several decades ago (Schoots, 1989; Schoots, 1982). Different detectors have been used. Some naturally fluorescent compounds have been separated and identified with a fluorescence detector, in both serum and haemodialysate (Niwa, 1993; Barnett, 1985; Swan, 1983). A liquid chromatographic method including a UV detector for detection of UV-absorbing solutes in uremic serum has been developed (Schoots, 1985; Senftleber, 1976; Knudson 1978). A fixed wavelength at 254 nm was mainly used in these studies. Additionally, several types of chromatographic methods, such as gas chromatography mass spectrometry (GC-MS), for the detection of uremic toxins has also been developed (Niwa, 1997; Niwa, 2009). Mass spectrometry (MS) has been applied to the identification and quantification of uremic toxins and uremia-associated modified proteins (Niwa, 2009). However, GC-MS cannot analyse highly polar, thermally labile and highmolecular weight compounds and usually requires sample preparation, such as extraction or derivation to make non-volatile compounds thermally stable and volatile. Compared with GC-MS, liquid chromatography mass spectrometry (LC-MS) can separate and identify highly polar, thermally labile or high-molecular weight mixture compounds and does not require derivation; sample preparation is also simple. Newly developed LC-MS techniques have been successfully applied to uremic toxin research with the discovery of novel uremic toxins that range from low-molecular weight solutes to small-molecule proteins (Niwa, 2011). This new analytical method is available today, opening up new horizons for uremic

In order to explain the origin and potential of the online UV absorbance dialysis dose monitoring method, the HPLC method of analysing UV absorbance profiles was developed.

compounds, the most suitable is the electrochemical detector.

toxin identification detected earlier as unidentified HPLC peaks.

**5. HPLC study of UV absorbance profiles** 

amount of light is small.

Ten uremic patients – three females and seven males (mean age 62.6 ± 18.6 years) – participated in the study. All of the patients were investigated over a course of 30 hemodialysis treatments. Three different polysulphone dialysers were used (Fresenius Medical Care, Germany): a low flux membrane dialyser F8 HPS (N=14), a low flux membrane dialyser F10 HPS (N=3) and a high flux membrane dialyser FX 80 (N=11). The elimination of toxins by the different types of semi-permeable membranes was also estimated.

Blood samples were drawn before the start of dialysis treatment (Bstart) and immediately after the treatment (Bend) (Figure 2) using the slow flow/stop pump sampling technique. Blood was sampled into BD Vacutainer® Glass Serum Tube (red cup, Beckton Dickinson) and was allowed to clot. After centrifugation at 3000 r.p.m. the serum was ready for clinical chemistry analysis. Small molecules – creatinine (MW=113.12 Da), uric acid (MW=168.11 Da) and urea (MW=60.06 Da) – were measured at the Clinical Chemistry Laboratory of the North Estonia Medical Centre using standardised methods (Hitachi 912 autoanalyser, Roche, Switzerland). The accuracy for creatinine was 5%, for uric acid 2% and for urea 4%.

For HPLC analysis the additional pre-treatment of the centrifugated serum was necessary to release the blood proteins. The serum samples were purified from proteins by centrifuging with two different Microcon centrifugal filters (Millipore, USA): a YM-3 with cut-off 3 kDa and a YM-100 with cut-off 70 kDa. The aim of using different filters was to estimate the presence of different-sized molecules in the serum.

Dialysate samples were taken 10 minutes after the start of the dialysis session (Dstart) and immediately after the treatment (Dend) (210 or 240 minutes) (Figure 2). Also, pure dialysate, used as the reference solution, was collected before each dialysis session, when the dialysis machine was being prepared and the conductivity was stable. For HPLC analysis, the dialysate samples were acidified with formic acid to pH 4.0. Sample sizes of 50 μL or 100 μL were used for chromatographic separation. Figure 4 presents a block diagram illustrating the collection and handling of blood and dialysate samples.

Fig. 4. Block diagram of blood and dialysate sample collection and handling

Optical Dialysis Adequacy Monitoring:

sample collected at the end of hemodialysis

Small Uremic Toxins and Contribution to UV-Absorbance Studied by HPLC 151

Fig. 5. Characteristic HPLC profiles of spent dialysate monitored on a wavelength of 280 nm; Dstart - dialysate sample collected 10 minutes after start of hemodialysis, Dend *-* dialysate

Fig. 6. Characteristic HPLC profiles of serum monitored on different wavelengths. Blood samples were collected before the dialysis session (Bstart ) and monitored on wavelengths of

254 nm and 280 nm. Detected HPLC peaks – uremic toxins – are presented.

The HPLC instrumentation and tools were as follows: a diode array spectrophotometric detector (DAD, Perkin Elmer, USA); a manual injector (Rheodyne, USA); and a Zorbax C8 4.6 x 250 mm column (Du Pont Instruments, USA) with security guard KJO-4282 (Phenomenex, USA). The eluent was mixed with 0.05 M formic acid (pH 4.0), HPLC grade methanol and HPLC-S grade acetonitrile (Rathburn, Scotland), with a six-step gradient programme. The total flow rate of 1 mL/min was used continuously and the column temperature was adjusted to 30ºC. The method followed has been described previously (Lauri, 2010). The chromatographic peaks were detected by the UV detector at wavelengths of 254 and 280 nm. The data was processed by means of Turbochrom WS and Turboscan 200 software from Perkin Elmer.

The decrease of the uremic retention solutes was estimated as the reduction ratio (RR) and assessed as a percentage (%). Thus, the RR of compounds was defined as a function of predialysis concentration (Cpre) and concentration at the end of hemodialysis (Cpost) and calculated as:

$$RR = \frac{\mathcal{C}\_{pre} - \mathcal{C}\_{post}}{\mathcal{C}\_{pre}} \mathbf{100\%} \tag{3}$$

### **5.1 Results**

The results from the studies obtained by the HPLC method in order to analyse the UV absorbance profiles of the serum and dialysate samples are presented below. The results are presented as mean ± standard deviation (SD). A student's t-test was used to compare groups of values wherein p < 0.05 was considered significant. Pearson's correlation coefficient (r) between the UV absorbance from HPLC and online monitoring versus concentration of the substances in the blood was investigated. Samples taken at times coinciding with the selftests or alarms of the dialysis machine were excluded (3 of 60 dialysate samples). Some sessions were excluded due to the technical failure of the spectrophotometer (3 of 30 sessions) and due to laboratory errors (3 of 30 sessions). The data analyses were performed in Microsoft Excel 2003 (for Windows).

The characteristic HPLC profiles of the spent dialysate at the start of the dialysis session (Dstart) and at the end (Dend) are presented in Figure 5. When comparing the start and end values, the decrease in the height of the peaks (due to solute removal from the blood and into dialysate during dialysis) can be clearly seen.

As can be seen from this figure, a number of higher prevalent peaks can be observed from the HPLC profiles indicating the presence of chromophores, which are the main cause of cumulative and integrated UV absorbance. The highest peak on the HPLC profile is detected as uric acid causing a substantial amount of UV absorbance.

Figure 6 shows the HPLC chromatograms of the serum (filter YM-3 with cut-off 3 kDa) measured on wavelengths of 254 nm and 280 nm. Identified HPLC peaks, such as creatinine, uric acid (the highest contribution), hypoxanthine, indoxyl sulphate and hippuric acid, are shown. Absorbing spectra of two unknown persistent peaks (P1 and P2) were identified at retention times (RT) of 15.46 and 15.82 min.

Some chromophores, such as P2, creatinine, hypoxanthine and hippuric acid, give higher peaks on a wavelength of 254 nm, while P1 and indoxyl sulphate are better identified on a wavelength of 280 nm. This is due to the absorbing spectrum characteristics of the UV chromophores. This confirms the results obtained by the spectrophotometric analysis in this UV region.

The HPLC instrumentation and tools were as follows: a diode array spectrophotometric detector (DAD, Perkin Elmer, USA); a manual injector (Rheodyne, USA); and a Zorbax C8 4.6 x 250 mm column (Du Pont Instruments, USA) with security guard KJO-4282 (Phenomenex, USA). The eluent was mixed with 0.05 M formic acid (pH 4.0), HPLC grade methanol and HPLC-S grade acetonitrile (Rathburn, Scotland), with a six-step gradient programme. The total flow rate of 1 mL/min was used continuously and the column temperature was adjusted to 30ºC. The method followed has been described previously (Lauri, 2010). The chromatographic peaks were detected by the UV detector at wavelengths of 254 and 280 nm. The data was processed by means of Turbochrom WS and Turboscan 200 software from Perkin Elmer. The decrease of the uremic retention solutes was estimated as the reduction ratio (RR) and assessed as a percentage (%). Thus, the RR of compounds was defined as a function of predialysis concentration (Cpre) and concentration at the end of hemodialysis (Cpost) and

> 100% *pre post pre*

(3)

*C C*

*C*

The results from the studies obtained by the HPLC method in order to analyse the UV absorbance profiles of the serum and dialysate samples are presented below. The results are presented as mean ± standard deviation (SD). A student's t-test was used to compare groups of values wherein p < 0.05 was considered significant. Pearson's correlation coefficient (r) between the UV absorbance from HPLC and online monitoring versus concentration of the substances in the blood was investigated. Samples taken at times coinciding with the selftests or alarms of the dialysis machine were excluded (3 of 60 dialysate samples). Some sessions were excluded due to the technical failure of the spectrophotometer (3 of 30 sessions) and due to laboratory errors (3 of 30 sessions). The data analyses were performed

The characteristic HPLC profiles of the spent dialysate at the start of the dialysis session (Dstart) and at the end (Dend) are presented in Figure 5. When comparing the start and end values, the decrease in the height of the peaks (due to solute removal from the blood and

As can be seen from this figure, a number of higher prevalent peaks can be observed from the HPLC profiles indicating the presence of chromophores, which are the main cause of cumulative and integrated UV absorbance. The highest peak on the HPLC profile is detected

Figure 6 shows the HPLC chromatograms of the serum (filter YM-3 with cut-off 3 kDa) measured on wavelengths of 254 nm and 280 nm. Identified HPLC peaks, such as creatinine, uric acid (the highest contribution), hypoxanthine, indoxyl sulphate and hippuric acid, are shown. Absorbing spectra of two unknown persistent peaks (P1 and P2) were identified at

Some chromophores, such as P2, creatinine, hypoxanthine and hippuric acid, give higher peaks on a wavelength of 254 nm, while P1 and indoxyl sulphate are better identified on a wavelength of 280 nm. This is due to the absorbing spectrum characteristics of the UV chromophores. This confirms the results obtained by the spectrophotometric analysis in this

*RR*

calculated as:

**5.1 Results** 

UV region.

in Microsoft Excel 2003 (for Windows).

into dialysate during dialysis) can be clearly seen.

retention times (RT) of 15.46 and 15.82 min.

as uric acid causing a substantial amount of UV absorbance.

Fig. 5. Characteristic HPLC profiles of spent dialysate monitored on a wavelength of 280 nm; Dstart - dialysate sample collected 10 minutes after start of hemodialysis, Dend *-* dialysate sample collected at the end of hemodialysis

Fig. 6. Characteristic HPLC profiles of serum monitored on different wavelengths. Blood samples were collected before the dialysis session (Bstart ) and monitored on wavelengths of 254 nm and 280 nm. Detected HPLC peaks – uremic toxins – are presented.

Optical Dialysis Adequacy Monitoring:

dialysate for different types of membranes

dialysate online UV absorbance measurement (280 nm).

280 nm) and spent dialysate online UV absorbance (280 nm)

Small Uremic Toxins and Contribution to UV-Absorbance Studied by HPLC 153

Hippuric acid 75.1 ± 11.5 (N=15) 68.1 ± 9.4 (N=10) Uric acid 67.7 ± 8.5 (N=15) 65.6 ± 6.7 (N=15) Urea 67.0 ± 8.7 (N=15) 63.2 ± 5.07 (N=15) All HPLC peaks, 280 nm 65.2 ± 9.6 (N=15) 60.6 ± 7.9 (N=14) P1 62.1 ± 13.0 (N=12) 61.0 ± 5.3 (N=7) All HPLC peaks, 254 nm 60.2 ± 12.5 (N=14) 57.2 ± 7.7 (N=13) P2 59.2 ± 17.5 (N=13) 51.6 ± 5.9 (N=12) Online UV absorbance, 280 nm 58.1 ± 8.3 (N=15) 57.0 ± 10.4 (N=13) Creatinine 58.2 ± 7.7 (N=15) 56.6 ± 5.4 (N=15) Hypoxanthine 42.6 ± 16.0 (N=10) 46.1 ± 18.5 (N=8) Indoxyl sulphate 42.1 ± 18.0 (N=13) 47.8 ± 14.0 (N=12) Table 1. RR (%) of solutes and total area of all HPLC UV absorbance peaks on wavelengths of 254 nm and 280 nm in serum and RR of online UV absorbance at 280 nm in spent

As the low flux and high flux membranes showed similar RR for every uremic solute, the results can be combined. Figure 8 present an illustrative comparison between the RR of serum uric acid, urea and creatinine measured in the clinical laboratory using standardised methods, all HPLC peaks measured on two different wavelengths (254 and 280 nm) and

Fig. 8. Comparison of RR of small molecular uremic retention solutes (uric acid, urea and creatinine) in serum, RR of all HPLC serum peaks on two different wavelengths (254 and

RR (%) Low flux High flux

As seen from Figure 6, around ten peaks on the HPLC profiles can be selected forming the major part of the total HPLC signal. As a result of this, the total area of the ten main peaks (Top 10 area %) was estimated. The comparison for Top 10 area % in the serum and dialysate is presented in Figure 7. The contribution of ten main peaks forms approximately 80-90% of all HPLC peaks in both spent dialysate and serum. However, there is a difference in the number of peaks on different wavelengths. In comparison with the 254 nm wavelength, the Top 10 area % on the 280 nm wavelength is higher. According to the HPLC profiles obtained (Figure 5 and 6), the largest contribution to Top 10 area % on the 280 nm wavelength originates from the small water-soluble non-protein-bound uremic toxin uric acid. Additionally, there was no significant difference between the serum results filtered with different type of filter cut-off (3 kDa and 70 kDa) (p< 0.05).

Fig. 7. Comparison of Top 10 area % on wavelengths of 254 nm and 280 nm in serum (filter YM-3 with cut-off 3 kDa) and spent dialysate

The RR of the detected chromatographic solutes was estimated. Table 1 presents the comparison of the RR (%) of solutes and all HPLC UV absorbance peaks on wavelengths of 254 nm and 280 nm in serum and for the online UV absorbance at 280 nm in the spent dialysate for different types of membranes: low flux (F8 HPS, F10 HPS) and high flux (FX80). There was no significant difference between the results for RR of low flux and high flux membranes (p < 0.05).

Hippuric acid had the highest RR, while the small water-soluble compound hypoxanthine and protein-bound solute indoxyl sulphate had the lowest RR. The RR of all HPLC Ppeaks at 280 nm was similar to uric acid and urea and higher than the RR of online UV absorbance at 280 nm and creatinine. At the same time, the RR of all HPLC peaks at 254 nm and P2 were lower than uric acid and urea. Online UV absorbance at 280 nm, creatinine and P2 were all removed in a statistically similar way (p < 0.05) and had lower RR than uric acid, urea and all HPLC peaks at 280 nm.

As seen from Figure 6, around ten peaks on the HPLC profiles can be selected forming the major part of the total HPLC signal. As a result of this, the total area of the ten main peaks (Top 10 area %) was estimated. The comparison for Top 10 area % in the serum and dialysate is presented in Figure 7. The contribution of ten main peaks forms approximately 80-90% of all HPLC peaks in both spent dialysate and serum. However, there is a difference in the number of peaks on different wavelengths. In comparison with the 254 nm wavelength, the Top 10 area % on the 280 nm wavelength is higher. According to the HPLC profiles obtained (Figure 5 and 6), the largest contribution to Top 10 area % on the 280 nm wavelength originates from the small water-soluble non-protein-bound uremic toxin uric acid. Additionally, there was no significant difference between the serum results filtered

Fig. 7. Comparison of Top 10 area % on wavelengths of 254 nm and 280 nm in serum (filter

The RR of the detected chromatographic solutes was estimated. Table 1 presents the comparison of the RR (%) of solutes and all HPLC UV absorbance peaks on wavelengths of 254 nm and 280 nm in serum and for the online UV absorbance at 280 nm in the spent dialysate for different types of membranes: low flux (F8 HPS, F10 HPS) and high flux (FX80). There was no significant difference between the results for RR of low flux and high

Hippuric acid had the highest RR, while the small water-soluble compound hypoxanthine and protein-bound solute indoxyl sulphate had the lowest RR. The RR of all HPLC Ppeaks at 280 nm was similar to uric acid and urea and higher than the RR of online UV absorbance at 280 nm and creatinine. At the same time, the RR of all HPLC peaks at 254 nm and P2 were lower than uric acid and urea. Online UV absorbance at 280 nm, creatinine and P2 were all removed in a statistically similar way (p < 0.05) and had lower RR than uric acid, urea and

with different type of filter cut-off (3 kDa and 70 kDa) (p< 0.05).

YM-3 with cut-off 3 kDa) and spent dialysate

flux membranes (p < 0.05).

all HPLC peaks at 280 nm.


Table 1. RR (%) of solutes and total area of all HPLC UV absorbance peaks on wavelengths of 254 nm and 280 nm in serum and RR of online UV absorbance at 280 nm in spent dialysate for different types of membranes

As the low flux and high flux membranes showed similar RR for every uremic solute, the results can be combined. Figure 8 present an illustrative comparison between the RR of serum uric acid, urea and creatinine measured in the clinical laboratory using standardised methods, all HPLC peaks measured on two different wavelengths (254 and 280 nm) and dialysate online UV absorbance measurement (280 nm).

Fig. 8. Comparison of RR of small molecular uremic retention solutes (uric acid, urea and creatinine) in serum, RR of all HPLC serum peaks on two different wavelengths (254 and 280 nm) and spent dialysate online UV absorbance (280 nm)

Optical Dialysis Adequacy Monitoring:

membrane.

Small Uremic Toxins and Contribution to UV-Absorbance Studied by HPLC 155

A number of higher prevalent HPLC peaks representing chromophores can be observed (Figure 5 and 6). This indicates that there is a group of compounds, among them several uremic toxins, which are the main cause of cumulative and integrated UV absorbance. The 10 main peaks formed app. 80-90% of the total area of all HPLC peaks; some of these are small molecular weight uremic toxins such as uric acid, creatinine, hippuric acid and indoxyl sulphate. The variations in the number of HPLC peaks depending on hemodialysis treatments and patients have been demonstrated in earlier studies (Schoots, 1982; Vanholder, 1992). The difference between two dialysis sessions may arise as shown in Figure 2, because of the different composition and removal of the uremic retention solutes contained in spent dialysate. When comparing the HPLC profiles of the spent dialysate in Figure 5 and those of the serum in Figure 6, more peaks are detected in the serum. Thus, not all solutes in serum are transported to dialysate and removed through the semi-permeable

The number of detected HPLC peaks at 254 nm and 280 nm is also demonstrated in Figure 6 and 7. The difference in the number of detected HPLC peaks on the wavelengths of 254 nm and 280 nm arises due to the characteristic absorbing spectra of the UV chromophores. The absorption of many components is higher on the wavelength of 254 nm than 280 nm. This confirms the results obtained via the spectrophotometric analysis in this UV region (Fridolin, 2003). However, the studies of relations between UV and small water soluble molecules such as urea and uric acid indicated that the wavelength of 280 nm may be preferred for online measurements when small water soluble molecules should be estimated. On this wavelength a relatively strong linear relationship exists between UV absorbance and concentrations of urea, creatinine and uric acid (Fridolin, 2002; Uhlin, 2003). While the contribution of uric acid forms a considerable part of the total area of HPLC peaks, uric acid plays an important role in online UV absorbance dialysis dose monitoring. Interestingly, the removal of urea and uric acid was statistically undifferentiated (p > 0.05). This information gives us alternative possibilities to use other components and methods to

Additionally, the low flux and high flux membranes showed no different removal of the studied small molecule uremic toxins as presented in earlier studies (Lesaffer, 2000). In this study it was found that the cellulose triacetate and polysulphone HF membranes removed similarly classical markers and protein-bound liphophilic solutes as an LF polysulphone membrane. Parallel results were obtained even with the concentrations corrected using a correction factor based on the total protein concentration at the start and at the end of dialysis as used by Lesaffer et al. (Lesaffer, 2000). Furthermore, there was no statistical difference between intradialytic start-end values, and removal efficiency for the LF and HF membranes estimated by the total area of HPLC peaks at 254 nm and 280 nm in the serum and online UV absorbance at 280 nm in the spent dialysate. This indicates that UV absorbance follows the behaviour of UV-absorbing compounds – uremic toxins – which are

The RR values of different identified compounds, the total area of all HPLC UV absorbance peaks on the wavelengths of 254 nm and 280 nm in the serum and the RR of online UV absorbance at 280 nm in spent dialysate are presented in Table 1 and Figure 8. Taking into account the removal efficiency, a difference can be observed in the relation of UV absorbance to small water-soluble non-protein-bound solutes and to small protein-bound solutes such as indoxyl sulphate. The small non-protein-bound solutes uric acid and urea showed a far more substantial decrease of concentration than creatinine being statistically

monitor urea reduction (URR) in a single hemodialysis session.

the origin of total UV absorbance in serum and spent dialysate.

The RR of urea and uric acid and the RR of all HPLC peaks at 280 nm are > 60%, while the RR of all HPLC peaks at 254 nm, of online UV absorbance measurement and creatinine are < 60%. Interestingly, the removal of urea and uric acid was statistically undifferentiated (p > 0.05). Figure 8 shows that the RR of serum creatinine is statistically different (p < 0.05) from the RR of serum urea and uric acid concentrations and from the RR of all HPLC peaks in the serum (at 280 nm), but not different from that of 254 nm. The RR of online measurement is comparable with the RR of creatinine and of all HPLC peaks at 254 nm and lower than the RR of urea, uric acid and all HPLC peaks at 280 nm (p < 0.05). At the same time, the RR of serum urea and uric acid are not statistically different; neither are they different from the RR of all HPLC peaks at 254 nm and 280 nm (p < 0.05).

The correlation coefficients between RR UV absorbance at 254 nm and at 280 nm from the total area of the HPLC peaks, from online monitoring at 280 nm and RR for certain substances with different molecular weights in spent dialysate and serum are shown in Table 2.


Table 2. Pearson correlation coefficients between RR of all HPLC peaks (254 nm and 280 nm), online absorbance at 280 nm and RR of uric acid, urea and creatinine in serum. The significance level of the results is P<0.01.

A high correlation for uric acid, urea and creatinine was obtained in both spent dialysate and serum. Some differences regarding 254 nm and 280 nm can be observed. However, urea does not represent as good a correlation as uric acid.
