Up to 60% of critically ill patients develop acute kidney injury, and 12% require dialysis (1). In addition to oliguria, diagnosis of renal impairment is based on serum biomarkers, especially urea and creatinine concentrations (2). However, routine renal biomarkers may be unreliable in critically ill patients because muscle mass is compromised and many are given large enteral protein loads (3). Better blood and urine biomarkers exist but are not readily available. A real-time bedside quantification of renal function would be diagnostically helpful and perhaps also guide hemodialysis management.
Breath gas analysis may meet these requirements as exhaled biomarkers are linked to physiologic processes and metabolic disorders (4–6). Several volatiles in exhaled breath have been described to increase with decreasing renal function (7–9). However, all data refer to spontaneously breathing patients with chronic renal failure and to changes in exhaled breath after outpatient, intermittent dialysis. Nevertheless, so far, there are no data available on both acute kidney injury and ventilated critically ill patients. In addition, the influence of continuous renal replacement therapy on exhaled breath is completely unclear. Multicapillary column ion-mobility spectrometry (MCC-IMS) can be used for real-time clinical breath analysis (6 , 10).
We therefore evaluated volatile organic compounds (VOCs) in expired gas from mechanically ventilated critical care patients with acute kidney injury using MCC-IMS. Specifically, we tested the hypothesis that there is a characteristic altered exhalation pattern in acute kidney injury and that the disturbance is reversed during continuous hemodialysis.
MATERIALS AND METHODS
With approval by the responsible ethics committee (Identification Number 232/14, Ärztekammer Saarland, Saarbrücken, Germany), we studied 20 sedated, intubated and mechanically ventilated adults. We excluded patients who had previous renal replacement therapy. Consent was obtained from legal guardians or patients themselves.
Patients were ventilated with an intensive care respirator (EVITA 4; Dräger, Lübeck, Germany) with ventilation variables and oxygen concentrations adjusted to clinical needs. Breath samples were taken via a t-piece, directly connected to the endotracheal tube. To prevent adsorption of volatile compounds at the sampling point, heat-and-moisture-exchanging filters (Humid-Vent Filter Compact S; Teleflex Medical, Athlone, Ireland) were connected to the outlet port of the ventilator. Pressure-supported or pressure-controlled modes were used for artificial respiration.
Indications for continuous renal replacement therapy included acute kidney injury with either hyperkalemia, hypervolemia, enhanced levels of the renal retention solutes urea and creatinine, or metabolic acidosis. The decision was made based on clinical variables by the treating physicians.
All patients were treated with continuous venovenous hemodialysis (multiFiltrate Ci-Ca; Fresenius Medical Care, Bad Homburg, Germany) using polysulfone membrane hemofilters (Ultraflux AV 1000S; Fresenius Medical Care). The dialysate solution was free of calcium and phosphate (Ci-Ca Dialysate K4; Fresenius Medical Care) and had a potassium concentration of 4 mmol/L. Regional sodium citrate anticoagulation (Natriumcitrat 4%; Fresenius Medical Care) was used before and calcium chloride (Calciumchlorid 100 mL; Fresenius Medical Care) to compensate for calcium losses after the hemofilter. Initial blood flow was 100 mL/min with 2 L/hr dialysate flow. However, filtration rates were thereafter adjusted to optimize patients’ volume status and diuresis.
Glomerular filtration rate was assessed with the Chronic Kidney Disease Epidemiology Collaboration equation as recommended by Kidney Disease Improving Global Outcomes guidelines (3).
VOCs in expired gas were evaluated as described previously using MCC-IMS (BreathDiscovery; B&S Analytik, Dortmund, Germany) (11 , 12). A polytetrafluoroethylene aspiration tube (Bohlender, Grünsfeld, Germany) was connected in proximity to the patient endotracheal tube. Ten-milliliter gas samples were aspirated from the breathing circuit at 20-minute intervals from at least 30 minutes before to at least 7 hours after beginning of continuous venovenous hemodialysis.
Preseparation of volatiles by multicapillary columns resulted in compound retention times, analysis by ion-mobility spectrometry in drift times (1/K0 [K = ion mobility]). Ion-mobility spectrometry (IMS)–Peaks with an intensity of more than 5 mV in at least three consecutive measurements were included. Volatile compounds were detected using the software Visual Now 3.6 (B&S Analytik, Dortmund, Germany). Identification of volatile compounds was carried out via automatic alignment (MIMA, Version 1.1.2) with a known database (BS-MCC/IMS-analytes database, Version 1209; B&S Analytik, Dortmund, Germany) (13). These data bank rely on pure substance measurements of all contained compounds. Peak area overlapping of at least 10% with preexisting reference substance in chromatogram defined alignment. Otherwise, compounds were labeled unknown and only retention and drift times were indicated for identification. However, we confirmed the identification of selected compounds with own MCC-IMS measurements of the pure substances.
VOCs in dialysis patients were compared with those observed in six ventilated critically ill patients with normal renal function at a previous time. Data in control group were a post analysis out of a previous study (14). Gas samples in these patients were averaged over 12 hours.
Statistical evaluation was carried out using SigmaPlot (Version 12.5: Systat Software, Erkrath, Germany). Intensities of VOCs during hemodialysis were normalized to baseline values prior renal replacement therapy and expressed as means (± 95% CI). Data were tested for distribution normality with Kolmogorov-Smirnov test and analyzed using repeated-measures analysis of variance (ANOVA). When indicated, repeated-measures ANOVA on ranks were used.
A two-tailed p value of less than 0.05 was considered statistically significant. Post hoc multiple comparisons were made with the Holm-Sidak method. Significant differences in at least three time points compared with baseline values defined increasing or decreasing course of VOC intensity during hemodialysis.
VOC intensities were tested for normal distribution (Shapiro-Wilk) and compared with control group (Mann-Whitney rank sum test) and with 7 hours of hemodialysis treatment (Wilcoxon signed-rank test).
The 20 participating patients had a mean age of 71 (± 12 SD yr) years, weight of 78 (± 18 SD kg) kg, and height of 170 (± 12 SD cm) cm. Demographic and clinical variables in patients with acute kidney injury before initiation of renal replacement therapy and patients in control group are shown in Table 1 and in Supplemental Table 1 (Supplemental Digital Content 1, http://links.lww.com/CCM/E121). Urea and creatinine were higher in patients with acute kidney injury (106 mg/dL, SD ± 59; 3.0 mg/dL, SD ± 1.7) compared with control group (57 mg/dL, SD ± 28; 1.0 mg/dL, SD ± 0.3; p = 0.07; p = 0.01). The glomerular filtration rate was significantly lower in patients with renal impairment (28 mL/min, SD ± 13) compared with patients with normal renal function (77 mL/min, SD ± 30; p = 0.003).
We analyzed 719 samples of expired air. A total of 60 different signals were observed by MCC-IMS. Forty-four compounds were identified, with 16 remaining unknown. Intensity of 45 compounds were greater in patients with renal impairment, seven were lower, and eight were similar. Thirty-four signals decreased during hemodialysis treatment, whereas 26 volatiles remained constant. No volatile increased during dialysis. Among the 45 signals that were exaggerated by acute kidney impairment, 30 volatiles decreased during 7 hours of hemodialysis (Supplemental Table 2, Supplemental Digital Content 2, http://links.lww.com/CCM/E122).
Volatile cyclohexanol (23 mV; 25–75th, 19–38), 3-hydroxy-2-butanone (16 mV; 25–75th, 9–26), 3-methylbutanal (20 mV; 25–75th, 14–26), and dimer of isoprene (26 mV; 25–75th, 18–32) showed significant higher intensities in acute kidney injury compared with control group (12 mV; 25–75th, 10–16 and 8 mV; 25–75, 7–14 and not detectable and 4 mV; 25–75th, 0–6; p < 0.05) and a significant decline after 7 hours continuous venovenous hemodialysis (16 mV; 25–75th, 13–21 and 7 mV; 25–75th, 6–13 and 9 mV; 25–75th, 8–13 and 14 mV; 25–75th, 10–19) (Fig. 1). After 6 hours of hemodialysis, 3-hydroxy-2-butanone concentrations in exhaled air were comparable with those in control group. Cyclohexanol, 3-methylbutanal, and isoprene did not reach level of patients with normal renal function during 7 hours of hemodialysis session. Compared with baseline values before hemodialysis, intensities declined significantly during dialysis: volatile cyclohexanol after 4.7 hours, 3-hydroxy-2-butanone after 1.7 hours, 3-methylbutanal after 3 hours, and isoprene after 20 minutes (Fig. 2). Examples of 3D ion-mobility spectrometry chromatograms of exhaled cyclohexanol in acute kidney injury and after 7 hours of hemodialysis treatment are shown in Figure 3.
Acute kidney injury leads to accumulation of uremic solutes including nitrogenous waste products and low molecular weight organic compounds (15). Effective elimination by hemodialysis requires that substances meet various conditions. First, low protein binding leads to a high free-substance fraction which promotes permeability through hemodialysis membranes. Second, a small volume-of-distribution corresponds to high concentrations in blood. For example, hydrophilic compounds like 3-hydroxy-2-butanone have a small volume-of-distribution, whereas lipophilic molecules like isoprene exceed the volume of body water. Third, small molecules like VOCs are highly dialyzable. Additional factors contributing to enhanced elimination during hemodialysis include high blood and dialysate flow as well as large pore size of hemodialysis filter.
Two mechanisms might explain why the intensities of so many volatiles decreased during renal replacement therapy. First, small volatile molecules are eliminated directly by hemodialysis with a consequent reduction in plasma concentrations. In contrast, lipophilic compounds have high lung permeability but limited renal excretion. Therefore, a second indirect mechanism is more likely: hydrophilic precursors of lipophilic volatiles accumulate in renal impairment. During dialysis, precursors are cleared by hemodialysis which decreases production of lipophilic volatiles and therefore exhalation of these compounds. The adsorption of compounds on the dialysis filter itself seems unlikely since the drop in intensities can only be determined for selected volatiles. On the other hand, we were able to exclude emissions of the compounds mentioned from dialysis components by head space measurements. Nevertheless, it is unclear what proportion the ventilation itself has in the elimination and thus also in decreasing intensities of volatile substances. We can only speculate but all markers that show no intensity changes during dialysis do not seem to be affected. The repartition of substance elimination between ventilation and dialysis seems to be extremely complex.
The intensities of 45 signals were increased during acute kidney impairment, of which 30 decreased significantly during hemodialysis. Most though are unlikely to facilitate diagnosis of acute kidney injury or management of hemodialysis because the signal intensities were too weak and too close to background noise. However, four VOCs were substantially elevated during renal impairment. Furthermore, the abnormal exhalation pattern was at least partially reversible during continuous hemodialysis. And finally, these four volatiles were the only compounds showing consistently decreasing intensities during dialysis.
The first volatile that was substantially increased during acute kidney injury and decreased during dialysis was cyclohexanol. The secondary alcohol is formed with cyclohexanone by oxidation from cyclohexane. We detected cyclohexanol and cyclohexanone, both having significant higher intensities during acute kidney impairment. Cyclohexanone has been described as uremic toxin (9). However, oxidation might also contribute to increased cyclohexanol in acute kidney injury. In contrast to cyclohexanol, exhaled cyclohexanone failed to decrease during hemodialysis. Mochalski et al (9) detected cyclohexanone as contaminant emitted by dialyzer materials in blood and breath of patients undergoing hemodialysis treatment (9). Possibly, cyclohexanone emission counter-acted reductions that might otherwise occur during hemodialysis.
The second volatile was 3-hydroxy-2-butanone. Intensities were high in patients with acute kidney injury and intensity decreased substantially during dialysis in about 1.5 hours. High water solubility and excellent permeability may explain why intensity decreased to control levels within 6 hours of hemodialysis. Elevated oxidative activity might be the reason for increased intensity (16). Pagonas et al (8) detected increasing intensities of 3-hydroxy-2-butanone in acute kidney impairment as well. Interestingly, this volatile was completely removed in exhaled air by one intermittent hemodialysis session. We merely observed a decrease by nearly half after 7 hours of hemodialysis. However, we used continuous venovenous hemodialysis suggesting a more slowly elimination of renal retention solutes explaining slower decrease. Interestingly, our controls revealed much higher intensities of exhaled 3-hydroxy-2-butanone compared with Pagonas et al (8), indicating that critical illness might contribute to endogenous production. We can only speculate about endogenous source of 3-hydroxy-2-butanone in acute kidney injury, but sufficient elimination by hemodialysis indicates renal clearance.
The third volatile with substantially higher intensities in renal impairment was 3-methylbutanal. Hemodialysis resulted in a significant drop after 3 hours. This oxygenated aldehyde was found in the human gut (17), is produced from Staphylococcus aureus (18), and metabolized by human hepatocellular carcinoma cells (19). Exogenous sources include cheddar cheese and spirits. 3-methylbutanal was found in human urines explaining our findings and the clearance during hemodialysis (20).
And finally, we detected increased volatile isoprene intensities in patients with acute kidney injury. The unsaturated hydrocarbon is one of the most abundant volatiles in humans breath (21). As a byproduct of cholesterol biosynthesis, isoprene is endogenously produced (22) and associated to oxidative damage (23–25). Isoprene is a known volatile increasing with decreasing renal function (9 , 15 , 26–28). Interestingly, intermittent hemodialysis was associated with further rising isoprene concentrations in exhaled air (7 , 27–30). There could be numerous reasons for this phenomenon: biocompatibility of dialysis membranes, hemodynamic stress, and inflammation. Furthermore, clearance of isoprene precursor mevalonic acid by hemodialysis might trigger compensatory synthesis of mevalonic acid resulting in increased isoprene production (28). However, sham-hemodialysis resulted in elevated breath isoprene as well, suggesting the origin in the extracorporeal circuit (29). In contrast to these findings, continuous ambulatory peritoneal dialysis was not associated with isoprene overproduction suggesting differences in continuous and intermittent dialysis treatments (31). Supporting these findings, we now report a constant course of isoprene monomer and even though a significant decline of isoprene dimer during continuous renal replacement therapy. Possibly, reduced blood flow and smaller hemodynamic effects of continuous hemodialysis might contribute to diminished inflammation, oxidative and hemodynamic stress. The lack of isoprene overproduction might indicate protective mechanisms of continuous compared with intermittent renal replacement therapies.
We used ion-mobility spectrometry for breath gas analysis. Advantages compared with other technologies include bedside online measurements with measurements being completed at 1-minute intervals, low technical and financial expenditure, and high sensitivity with a detection limit down to ng/L and even pg/L ranges (10). Other advantages of ion-mobility spectrometry over gas chromatography mass spectrometry include the ability to evaluate humid exhaled air, minimal weight, small size, low power consumption, and no vacuum requirement—making the system suitable for serial bedside measurements. All our patients were intubated and mechanically ventilated which allowed standardized gas sampling out of a closed system. Nevertheless, we demonstrated contamination of ventilator circuit by ambient air, mechanical ventilators, and gas from central hospital supplies (14). However, cyclohexanol, 3-hydroxy-2-butanone, 3-methylbutanal, and isoprene are not subject to risk of contamination in our setting. Additionally, sample loop of MCC-IMS was flushed with highly purified, synthetic air between serial measurements to rule out spread of contamination.
We did not determine any concentrations in the serum. This is unfortunately not possible but maybe not useful. Correlation will always be difficult, as many substances have no alveolar passage, others have low solubility in the blood. Thus, the concentrations between serum and respiratory air will always differ. Much more important, however, is the best possible correlation of a biomarker with the outcome of kidney function and the efficiency of dialysis. In contrast to serum variables, volatile biomarkers probably allow a similarly sufficient but continuous fingerprint of organ functions.
Our study has several limitations. First, we conducted gas sampling in mixed, inspired, and expired air of the tubing system. Alveolar sampling during patient expiration might result in higher intensities and more reliable results. Second, critically ill patients have a range of metabolic disorders, comorbidities, and medication. Presumably many alter the patients’ exhalome, but at this point, it is unknown how or to what extent. Third, the etiology of acute kidney injury in our population is of huge diversity ranging from trauma, sepsis, malignancy, or transplantation, so correlation of exhalation pattern to the outcome of kidney injury or efficiency of hemodialysis is not possible up to now. And finally, ion-mobility spectrometry intensities are substance specific. Selected volatiles like isoprene show a weak signal at the detector of the IMS leading to a poor response even when concentrations are high. Protonated substances like isoprene do not form hydrates like other volatiles and therefore can pass through the drift tube more rapidly, leading to poor ion detection (32).
The future clinical benefit might be routine quantification of volatile retention variables in the ventilator circuit. Thus, acute kidney injury could be diagnosed in a timely manner, and the replacement therapy might be individually controlled via feedback systems. This could help maximizing the effectiveness of dialysis and optimizing treatment days to reduce morbidity and unnecessary costs. However, future studies will have to validate these results in larger patient populations with the same etiology of acute kidney injury before routine monitoring of renal function and control of hemodialysis are possible. In addition, it would be necessary to correlate volatile with serological retention variables and investigate its influence on patients’ outcome. This might clarify if breath gas analysis in acute kidney injury is superior to previous surrogate variables in patients’ blood.
Many organic compounds increased in expired gas during acute kidney injury. Several were at least partially reversible during continuous hemodialysis. Four volatiles were especially increased during renal impairment and returned to control levels during dialysis: cyclohexanol, 3-hydroxy-2-butanone, 3-methylbutanal, and dimer of isoprene. Analysis of the exhalome may therefore facilitate bedside noninvasive diagnosis of acute kidney injury and management of renal replacement therapy.
1. Hoste EA, Bagshaw SM, Bellomo R, et al. Epidemiology of acute kidney injury in critically ill patients: The multinational AKI-EPI study. Intensive Care Med 2015; 41:1411–1423
2. Kellum JA, Lameire N; KDIGO AKI Guideline Work Group: Diagnosis, evaluation, and management of acute kidney injury: A KDIGO summary (Part 1). Crit Care 2013; 17:204
3. Levey AS, Becker C, Inker LA. Glomerular filtration rate and albuminuria for detection and staging of acute and chronic kidney disease in adults: A systematic review. JAMA 2015; 313:837–846
4. Amann A, Costello Bde L, Miekisch W, et al. The human volatilome: Volatile organic compounds (VOCs) in exhaled breath, skin emanations, urine, feces and saliva. J Breath Res 2014; 8:034001
5. de Lacy Costello B, Amann A, Al-Kateb H, et al. A review of the volatiles from the healthy human body. J Breath Res 2014; 8:014001
6. Fink T, Baumbach JI, Kreuer S. Ion mobility spectrometry in breath research. J Breath Res 2014; 8:027104
7. Schönermarck U, Dengler C, Gmeinwieser A, et al. Exhaled breath volatile organic and inorganic compound composition in end-stage renal disease. Clin Nephrol 2016; 86:132–140
8. Pagonas N, Vautz W, Seifert L, et al. Volatile organic compounds in uremia. PLoS One 2012; 7:e46258
9. Mochalski P, King J, Haas M, et al. Blood and breath profiles of volatile organic compounds in patients with end-stage renal disease. BMC Nephrol 2014; 15:43
10. Baumbach JI. Ion mobility spectrometry coupled with multi-capillary columns for metabolic profiling of human breath. J Breath Res 2009; 3:034001
11. Wolf A, Baumbach JI, Kleber A, et al. Multi-capillary column-ion mobility spectrometer (MCC-IMS) breath analysis
in ventilated rats: A model with the feasibility of long-term measurements. J Breath Res 2014; 8:016006
12. Albrecht FW, Hüppe T, Fink T, et al. Influence of the respirator on volatile organic compounds: An animal study in rats over 24 hours. J Breath Res 2015; 9:016007
13. Maurer F, Hauschild AC, Eisinger K, et al. MIMA - a software for analyte identification in MCC/IMS chromatograms by mapping accompanying GC/MS measurements. Int J Ion Mobil Spectrom 2014; 17:95–101
14. Hüppe T, Lorenz D, Wachowiak M, et al. Volatile organic compounds in ventilated critical care
patients: A systematic evaluation of cofactors. BMC Pulm Med 2017; 17:116
15. Meinardi S, Jin KB, Barletta B, et al. Exhaled breath and fecal volatile organic biomarkers of chronic kidney disease. Biochim Biophys Acta 2013; 1830:2531–2537
16. Phillips M, Cataneo RN, Cummin AR, et al. Detection of lung cancer with volatile markers in the breath. Chest 2003; 123:2115–2123
17. Garner CE, Smith S, de Lacy Costello B, et al. Volatile organic compounds from feces and their potential for diagnosis of gastrointestinal disease. FASEB J 2007; 21:1675–1688
18. Chen J, Tang J, Shi H, et al. Characteristics of volatile organic compounds produced from five pathogenic bacteria by headspace-solid phase micro-extraction/gas chromatography-mass spectrometry. J Basic Microbiol 2017; 57:228–237
19. Mochalski P, Sponring A, King J, et al. Release and uptake of volatile organic compounds by human hepatocellular carcinoma cells (HepG2) in vitro. Cancer Cell Int 2013; 13:72
20. Calejo I, Moreira N, Araújo AM, et al. Optimisation and validation of a HS-SPME-GC-IT/MS method for analysis of carbonyl volatile compounds as biomarkers in human urine: Application in a pilot study to discriminate individuals with smoking habits. Talanta 2016; 148:486–493
21. King J, Koc H, Unterkofler K, et al. Physiological modeling of isoprene dynamics in exhaled breath. J Theor Biol 2010; 267:626–637
22. Miekisch W, Schubert JK, Noeldge-Schomburg GF. Diagnostic potential of breath analysis
–focus on volatile organic compounds. Clin Chim Acta 2004; 347:25–39
23. Yamada YI, Yamada G, Otsuka M, et al. Volatile organic compounds in exhaled breath of idiopathic pulmonary fibrosis for discrimination from healthy subjects. Lung 2017; 195:247–254
24. Foster WM, Jiang L, Stetkiewicz PT, et al. Breath isoprene: Temporal changes in respiratory output after exposure to ozone. J Appl Physiol (1985) 1996; 80:706–710
25. Mendis S, Sobotka PA, Euler DE. Expired hydrocarbons in patients with acute myocardial infarction. Free Radic Res 1995; 23:117–122
26. Grabowska-Polanowska B, Faber J, Skowron M, et al. Detection of potential chronic kidney disease markers in breath using gas chromatography with mass-spectral detection coupled with thermal desorption method. J Chromatogr A 2013; 1301:179–189
27. Davies S, Spanel P, Smith D. A new ‘online’ method to measure increased exhaled isoprene in end-stage renal failure. Nephrol Dial Transplant 2001; 16:836–839
28. Trovarelli G, Brunori F, De Medio GE, et al. Onset, time course, and persistence of increased haemodialysis-induced breath isoprene emission. Nephron 2001; 88:44–47
29. Capodicasa E, Brunori F, De Medio GE, et al. Effect of two-hour daily hemodialysis and sham dialysis on breath isoprene exhalation. Int J Artif Organs 2007; 30:583–588
30. Goerl T, Kischkel S, Sawacki A, et al. Volatile breath biomarkers for patient monitoring during haemodialysis. J Breath Res 2013; 7:017116
31. Capodicasa E, Trovarelli G, Brunori F, et al. Lack of isoprene overproduction during peritoneal dialysis. Perit Dial Int 2002; 22:48–52
32. Mochalski P, Rudnicka J, Agapiou A, et al. Near real-time VOCs analysis using an aspiration ion mobility spectrometer. J Breath Res 2013; 7:026002