Acta Anaesthesiol Scand 2008; 52: 487–492 r 2008 The Authors Printed in Singapore. All rights reserved Journal compilation r 2008 The Acta Anaesthesiologica Scandinavica Foundation ACTA ANAESTHESIOLOGICA SCANDINAVICA doi: 10.1111/j.1399-6576.2007.01551.x

Perioperative concentrations of catecholamines in the cerebrospinal fluid and plasma during spinal anesthesia

1,2 3 4 5 2 2 M. J. OEHMKE ,T.PODRANSKI ,M.MANN ,N.FRICKEY ,D.F.M.KUHN and G. HEMPELMANN 1Department of Special Anaesthesia and Pain Control, Medical University of Vienna, Vienna, Austria, 2Department of Anaesthesiology, Intensive Care Medicine, Pain Therapy, University Hospital Giessen, Giessen, Germany, 3Outcomes Research Consortium, The Cleveland Clinic, Cleveland, OH, USA, 4Institute of Medical Informatics, Working Group Medical Statistics, University of Giessen, Giessen, Germany, 5Department of Anaesthesia, General Intensive Care, and Pain Control, Medical University of Vienna, Vienna, Austria

Background: Catecholamine release is a physiological re- [249;422] to 556 [423;649] pmol/l and remained elevated sponse to stress. The extent to which perioperative stress 24 h after surgery (P 5 0.009). There was no association provokes the central release of catecholamines, which modu- between changes in CSF or plasma norepinephrine or late pain perception in the spinal cord, still remains unknown. epinephrine concentrations and changes in heart rate The perioperative course of catecholamine concentrations in (HR) or mean arterial pressure (MAP). the cerebrospinal fluid (CSF) and plasma was examined. Conclusion: During spinal anesthesia for elective hip joint Methods: Aprospectivestudywasperformedin25pa- replacement, norepinephrine concentrations were greater tients (ASA III, 60–84 years) undergoing elective hip joint in plasma than in CSF. CSF dopamine and epinephrine replacement in spinal catheter anesthesia. The concentra- concentrations were essentially undetectable. The changes tions of dopamine, epinephrine and norepinephrine in the in CSF-norepinephrine concentrations and the changes of CSF and plasma were measured before anesthesia, imme- plasma norepinephrine concentrations showed no associa- diately after surgery, and 6 and 24 h post-operatively. tion with each other; nor were there correlations between Results: In most patients, dopamine and epinephrine clinical stress parameters (HR, MAP) or visual analog scale were not detectable in CSF. CSF–norepinephrine concen- pain, and the changes in CSF norepinephrine concentra- trations decreased from median [interquartile-range] tions. 159 [124;216] pre-anesthesia to 116 [79;152] pmol/l imme- diately post-operatively and were slightly elevated 24 h post-operatively (180 [134;302] pmol/l) (P 5 0.05). Dopa- Accepted for publication 14 October 2007 mine plasma concentrations were not detectable or were Key words: Catecholamines; cerebrospinal fluid; plasma; barely above the detection threshold. Plasma epinephrine perioperative; epinephrine; norepinephrine; dopamine; increased from 61 [28;77] pmol/l pre-anesthesia to 112 concentration. [69;138] pmol/l 6 h post-operatively and returned to base- line 24 h post-operatively (P 5 0.001). Plasma norepinephr- r 2008 The Authors ine concentrations increased intra-operatively from 298 Journal compilation r 2008 The Acta Anaesthesiologica Scandinavica Foundation

NTICIPATION of surgery, even with general an- response to surgical stress, the extent to which Aesthesia, triggers physical and psychological surgical stress provokes the central release of ca- stress in patients (1, 2). This results in an endocrine techolamines remains unknown. Because catecho- response mediated by the hypothalamic–pituitary– lamines modulate pain perception in the spinal adrenal and renin–angiotensin axes, along with cord (10–14), and differences in catecholamine sympathetic nervous system activation (3). levels may affect analgesia, we chose to investigate The most basic physiological response to stress, the catecholamine concentrations in the cerebrosp- including that resulting from surgical tissue injury, inal fluid (CSF). is release of catecholamines (4–7). The major cate- Therefore, this study determined the periopera- cholamines, dopamine, epinephrine, and norepi- tive changes of catecholamine concentrations in the nephrine are released both peripherally and CSF and plasma in elderly patients undergoing centrally (8, 9). major hip surgery. Possible correlations between While it is well established that peripheral cate- CSF and plasma catecholamine concentrations, cholamine concentrations increase markedly in heart rate (HR), mean arterial pressure (MAP), 487 M. J. Oehmke et al. and the visual analog scale (VAS) pain score were Germany), an indirect sympathomimetic agent, analyzed. and the study was discontinued. During anesthesia, patients received midazolam as required in steps of 1 mg i.v. up to a maximum Methods of 6 mg. Patients also received 3 l/min O2 via a nasal tube, volume substitution, and monitoring With the approval of the local Ethics Committee of SaO2. Post-operatively, patients received single and written informed patient consent, 25 patients dosages of 2 mg morphine i.v., as requested, if were enrolled in this prospective study. Inclusion pain was rated 43 on a 10-cm-long VAS with criteria were an age of 60 years or above with ASA each centimeter marked and numbered 0–10, Physical Status III undergoing elective total hip where 0 cm indicates no pain and 10 cm the worst joint replacement with spinal-catheter anesthesia. possible pain. The surgeon used cemented MS-30 stems and All values recorded for statistical analysis (HR, polyethylene screw-in cups (Biomet Deutschland MAP, VAS, and catecholamine concentrations) GmbH, Berlin, Germany). Exclusion criteria were were sampled at the measurement points immedi- decompensated cardiac or circulatory insufficiency, ately before administration of the local anesthetic severe renal or hepatic impairment, signs of clot- (pre-LA), immediately postoperative (0 h post-OP), ting disorders, neuromuscular dysfunction, dis- 6 h post-operatively (6 h post-OP), and 24 h post- eases of endocrine organs (diabetes, pancreatitis, operatively (24 h post-OP). Five milliliters of CSF thyroid problems, hepatitis), known tumors/neo- from the spinal catheter and an equal volume of plasmas, alcohol or drug abuse, use of substances blood from a central-venous catheter were ob- acting on the central nervous system, and use of tained in pre-chilled tubes and placed on ice at all calcium antagonists, a2- and b-sympathomimetics, measurement points. For storage purposes, fluid or theophylline. samples were centrifuged (10 min, 4000 U/min, One hour before surgery, 3.75 mg oral midazo- 4 1C) within 15 min and the supernatant frozen lam (Dormicum, Hoffmann-La Roche AG, Basel, at À 20 1C. Switzerland) was administered. During the pre- The concentrations of dopamine, epinephrine, paration for anesthesia, a 16 G central venous and norepinephrine in the CSF and plasma were line (Cavafix Certodyn 375, B. Braun, Melsungen, determined in a single batch for each parameter Germany) was placed in the basilic vein and by means of high-performance liquid chromato- correct placement checked by ECG. A 20 G graphy (HPLC) with electrochemical detection arterial line (B.Braun) was inserted into the radial (Reagent Kit Catecholamines in Plasma, Chromsys- artery. tems Instruments & Chemicals GmbH, Mu¨ nchen, HR from the ECG and MAP were recorded at Germany) (15). The lower limits of detection were 5-min intervals until the end of surgery, and then 1, 20 pmol/l or less. The intra-assay coefficients of 3, 6, 12, and 24 h post-operatively. For anesthesia variation were 5.6% for dopamine, 4.9% for epi- and sampling, a 20 G spinal catheter (Perifix, B. nephrine, and 2.8% for norepinephrine. Glucose Braun) was placed at L2/3 or L3/4. Anesthesia was was measured by photometry (Test-Combination initiated with 2.5 ml of isobaric bupivacainhy- Glucoquant Glukose, Boehringer, Mannheim, Ger- drochloride 0.5% (Carbostesin 0.5%, Astra, Ham- many). pH, pCO2, and pO2 from venous blood burg, Germany). The level of spinal anesthesia was were evaluated by a blood gas analyzer (Ciba determined by checking sensibility in both direc- Corning 288, Ciba Corning Diagnostics AG, Dietli- tions (upwards and downwards) with a cold test kon, Switzerland). roll (No. 2090, Nizell, Rickenbach, Switzerland) temperated at 4 1C. Return/loss of cold discrimina- tion was defined as the level of the block and Statistical analysis recorded before incision (once the block was estab- Because the variability of the target parameter was lished and the patients hemodynamically stable) as unknown, sample size calculation was not possible. well as immediately post-operative (0 h post-OP). Nonetheless, 25 patients were included in the If blood pressure decreased by more than 20% as study, which seemed likely to be sufficient for compared with baseline values, the patient re- this sort of natural history study. Statistical ana- ceived 5 mg amezinium metilsulfate intravenously lysis was explorative. We calculated the minimum, s (i.v). (Supratonin ,Gru¨ nenthal GmbH, Aachen, maximum, median, arithmetic mean, interquartile 488 Catecholamines in the cerebrospinal fluid range, and standard deviation for each numeric resulting in a median pain score of VAS 3 [2;4] 6 h parameter. Results are presented as median [inter- post-OP, and VAS 2 [1;3] 24 h post-OP. quartile range]. In most patients, dopamine, and epinephrine Because most parameters were not normally were not detectable in the CSF: dopamine was distributed, non-parametric tests were used for not detectable in 11 patients, in three patients it the analyses. First we determined whether the was detected immediately pre-LA, and in two catecholamine concentrations in plasma and CSF patients it was detected 6 h and 24 h post-OP. CSF changed during the observation period using epinephrine concentrations were detectable in just Friedman’s analysis of variance (ANOVA) by a single patient. Norepinephrine was detectable in ranks. Then absolute changes between the differ- all patients in both CSF and plasma. Its concentra- ences of catecholamine concentrations and the tions in the CSF decreased intra-operatively from differences in vital parameters or VAS during 159 [124;216] pmol/l pre-LA to 116 [79;152] pmol/l time intervals were calculated using Spearman’s 0 h post-OP and increased from 111 [71;283] pmol/l rank correlation. A correlation coefficient lower 6 h post-OP to 180 [134;302] pmol/l 24 h post-OP than 0.5 was termed no association. (P 5 0.05). Dopamine plasma concentrations were not de- tectable or were barely above the detection thresh- Results old. Plasma epinephrine increased from 61 Sixteen of the 25 patients were evaluated (six [28;77] pmol/l pre-LA to 112 [69;138] pmol/l 6 h women, 10 men). Nine patients were excluded post-OP and returned to baseline 24 h post-OP from data analysis due to discontinuation of the (P 5 0.001). Plasma norepinephrine concentration study (technical problems in placing spinal cathe- increased from 298 [249;422] pmol/l pre-LA to 556 ter n 5 1, administration of amezinium metilsulfate [423;649] pmol/l 0 h post-OP, and remained nearly n 5 7, dislocation of spinal catheter n 5 1). The 16 stable for up to 24 h (P 5 0.009) (Figs 1–3). patients included in the data analysis were ASA The changes in the detectable CSF catecholamine Physical Status III; their characteristics are listed in concentrations were not associated with changes in Table 1. plasma catecholamine concentrations and there Anesthesia and surgery of the evaluated patients was no association between changes of the vital were performed without complications. The heart parameters (HR, MAP) and VAS. The known clin- rate was (median [interquartile range]) 75 [71;81] ical effects of plasma catecholamine concentrations beats per minute at the beginning of the examina- tion period and 70 [65;74] beats per minute at the end of the period. The mean arterial pressure was 60 Operation 103 [95;107] mmHg versus 93 [84;100] mmHg. SaO2 Plasma level ranged from 95% to 99% during the entire study CSF level period. The upper level of the blockade varied from 40 T6 to T10 immediately before surgery started and from T10 to L1 right after skin closure. Midazolam

was administered intra-operatively to 11 patients. [pmol/L] Pre-LA the median pain score (VAS) was 1 [0;2]. Dopamine 20 Immediately after surgery the median pain score (VAS) was 0 [0;2.5]. During the 24 h post-OP study period, patients received 6–32 mg morphine i.v., 0 pre-LA 0h 6h 24h Table 1 post-OP Patient characteristics and operation data. Fig. 1. Median values and interquartile ranges of dopamine con- Age (years) 68 Æ 6 centrations in pmol/l in cerebrospinal fluid (CSF) and plasma at the Height (cm) 170 Æ 9 measurement points: pre-LA, right before spinal administration of Weight (kg) 78 Æ 12 the local anesthetic; 0 h post-OP, immediately post-operative; 6 h Duration of surgery (min) 128 Æ 36 Volume substitution (ml) 2719 Æ 587 post-OP, 6 h post-operatively; 24 h post-OP, 24 h post-operatively. Blood loss (ml) 881 Æ 351 Interquartile ranges of CSF values are too small for demonstration; both curves are based on 16 patients (median) but only five had a Values are given as mean Æ SD. n 5 16 (six females, 10 males). CSF dopamine value above the detection threshold. 489 M. J. Oehmke et al.

150 Discussion Operation Plasma level A group of elderly patients who were scheduled CSF level for hip surgery with spinal catheter analgesia was 100 examined, giving us continuous access to CSF and the opportunity to describe the perioperative time- course of catecholamines in humans for the first [pmol/L]

Epinephrine 50 time. The study showed that in this group of patients, catecholamine concentrations were higher in plasma than in CSF. Catecholamines are produced in the brain, adre- 0 nal medulla, extramedullar chromaffine tissue, the pre-LA 0h 6h 24h gray matter of the spinal cord, and sympathic post-OP nerve endings. They are stored in vesicles and Fig. 2. Median values and interquartile ranges of epinephrine liberated via central catecholaminergic impulses, concentrations in pmol/l in cerebrospinal fluid (CSF) and plasma, eventually making their way into plasma and CSF. at the measurement points: pre-LA, right before spinal adminis- This is of particular interest because catechola- tration of the local anesthetic; 0 h post-OP, immediately post- mines modulate pain perception in the spinal operative; 6 h post-OP, 6 h post-operatively; 24 h post-OP, 24 h cord (10–14), and differences in catecholamine post-operatively. Both curves are based on 16 patients (median) but only one had a CSF epinephrine concentration above the concentrations may affect analgesia. The mechan- detection threshold. ism of this process is an effect of epinephrine or norepinephrine on a2-receptors causing hyperpo- larization of the transmission cells of the substantia 750 Operation gelatinosa in the spinal cord dorsal horn (14). Plasma level Several clinical studies demonstrated that admin- 600 CSF level istering epinephrine or norepinephrine during epi- dural analgesia causes an elevation of the 450 nociceptive threshold (11–13). However, in this case the vasocontrictive effect that inhibits systemic

[pmol/L] 300 absorption and therefore prolongs the effect of local anesthetics or morphines also plays a role. Norepinephrine 150 Our results indicate that catecholamine concen- trations are greater in plasma than in CSF. Eldrup 0 et al. (16) observed a similar relationship in their control group. However, we could not confirm pre-LA 0h 6h 24h post-OP their conclusion that CSF and plasma norepinephr- ine concentrations are correlated, and Miura et al. Fig. 3. Median values and interquartile ranges of norepinephrine (17) similarly failed to show such a correlation in concentrations in pmol/l in cerebrospinal fluid and plasma at the their work with rats. This is in accordance with measurement points: pre-LA, right before spinal administration of the work of Oldendorf (18) stating that catechola- the local anesthetic; 0 h post-OP, immediately post-operative; 6 h post-OP, 6 h post-operatively; 24 h post-OP, 24 h post-operatively. mines do not pass the blood–brain barrier and must originate in individual compartments, such as the plasma or brain. Plasma catecholamine on the vital parameters (HR, MAP) at each mea- concentrations in rats (17) and humans thus pro- surement point were correlated, as expected. vide little if any estimate of CSF catecholamine Median values of venous blood pH were be- concentrations and possible associated changes in tween 7.37 [7.36;7.39] and 7.41 [7.39;7.45]. The the cerebrum. median values for pO2 were between 4.6 and Elevated perioperative plasma catecholamine 4.9 kPa (34 and 37 mmHg) and the values for concentrations signal a stress reaction associated pCO2 were between 5.3 and 6.1 kPa (40 and with the operation and perioperative procedure. 46 mmHg) in venous blood. Glucose concentrations Plasma norepinephrine remained elevated 24 h remained within the normal range (3.3–3.8 mmol/l after surgery, while plasma epinephrine rapidly in CSF and 4.9–6.0 mmol/l in blood). normalized. There was an intra-operative reduc- 490 Catecholamines in the cerebrospinal fluid tion in CSF norepinephrine, which was statistically sions caused by intra- or post-operative hypo- or significant. However, it remains unclear as to hyper-glycemia. whether this reduction is clinically important and A limitation of the study is the method of whether norepinephrine CSF concentrations are a anesthesia. This investigation reflects the actual sensitive marker of stress. This is especially the situation during spinal anesthesia. But it would case because the reduction in CSF norepinephrine be interesting to evaluate CSF catecholamine con- concentration during surgery was minimal and centrations during general anesthesia. the absolute changes did not correlate with the In conclusion, changes in perioperative catecho- differences in plasma norepinephrine concentra- lamine concentrations were observed in CSF and tions or VAS. plasma norepinephrine as well as in plasma epi- Kanto and Scheinin (5) recommend using basic nephrine and plasma dopamine concentrations. parameters such as MAP and HR to evaluate stress. The changes in CSF norepinephrine concentrations Consistent with their recommendation, we did not and the changes of plasma norepinephrine concen- observe an association between changes in plasma trations showed no association to each other; nor epinephrine concentrations and changes in HR or were there correlations between clinical stress MAP or between changes in plasma norepinephr- parameters (HR, MAP) or pain (VAS) and the ine concentrations and changes in CSF norepi- changes in CSF norepinephrine concentrations. nephrine concentrations. Furthermore, there was no association between the changes in VAS score and the changes of plasma epinephrine concentra- Acknowledgements tions or plasma norepinephrine concentrations or CSF norepinephrine concentrations, respectively. The authors wish to thank Prof. Marcel E. Durieux and Prof. This might be due to an effective analgesia and Daniel I. Sessler for their comments on the manuscript as well as Birgit Weber and Emely Lo¨ffler for their invaluable support the evaluation of pain by VAS where the chosen with the HPLC measurements. number depends on many factors varying from The study was supported by the Department of Anaesthesiol- person to person. ogy, Intensive Care Medicine, Pain Therapy, University Hospi- To rule out age-related differences in catechola- tal Giessen. The study sponsors were not involved in the design of the study, the collection, analysis, or in interpretation of the mine concentration, only older patients (above 60 data, or the preparation of the manuscript. None of the authors years) were included in the study (19). Postural has personal financial interest in this research. changes as the cause of increased catecholamine concentrations were improbable because all pa- tients were supine during the examination phase. References Patients receiving pre-/intra-operative medication that directly affected catecholamine concentration 1. Adams HA, Hempelmann G. The endocrine stress re- (e.g. amezinium metilsulfate) were excluded from sponse during anaesthesia and surgery – it’s origin and meaning. Ana¨sthesiol Intensivmed Notfallmed Schmerzther the study. The administration of benzodiazepines 1991; 26: 294–305. has little (5, 20) or no effect (21, 22) on circulating 2. Crozier TA, Sumpf E. Total intravenous anaesthesia with catecholamine levels, and we deduce that there S-(1)-ketamine/propofol or alfentanil/propofol for labar- was no significant pharmacological influence on otomy: effects on hemodynamic and endocrine stress re- sponses. Anaesthesist 1996; 45: 1015–23. perioperative catecholamine concentrations. Nor 3. Udelsman R, Norton JA, Jelenich SE et al. Responses of the do we think the sampling procedure affected the hypothalamic-pituitary-adrenal and renin-angiotensin axes measured levels, because all samples were centri- and the sympathetic system during controlled surgical and fuged and then frozen at À 20 1C immediately after anesthetic stress. J Clin Endocrinol Metab 1987; 64: 986–94. 4. Hamberger B, Ja¨rnberg PO. Plasma catecholamines during obtaining them and catecholamines are usually surgical stress: differences between neurolept and enflur- stable when frozen (23). As analgesia was effective ane anaesthesia. Acta Anaesthesiol Scand 1983; 27: 307–10. in all cases, pain-induced increase of catecholamine 5. Kanto J, Scheinin M. Biochemical assessment of preopera- concentrations seem highly unlikely. tive stress: a study with diazepam and measurement of monoamine metabolites and catecholamines in cerebrosp- The values for central venous pO2,pCO2, SaO2, inal fluid and plasma. Br J Anaesth 1991; 66: 587–90. and hemodynamic parameters, as signs of general 6. Migneault B, Girard F, Albert C et al. The effect of music on hypoxia or hypercapnia, did not show pathologi- the neurohormonal stress response to surgery under gen- cal alterations during the operation. Furthermore, eral anesthesia. Anesth Analg 2004; 98: 527–32. 7. Norman JG, Fink GW. The effects of epidural anesthesia on CSF and blood glucose concentrations were mon- the neuroendocrine response to major surgical stress: a itored to rule out or detect potential cerebral le- randomized prospective trial. Am Surg 1997; 63: 75–80. 491 M. J. Oehmke et al.

8. Nader ND, Ignatowski TA, Kurek CJ et al. Clonidine tions of brain and plasma catecholamines during near- suppresses plasma and cerebrospinal fluid concentrations complete cerebral ischemia in the rat. Anesth Analg 1999; of TNF-a during the perioperative period. Anesth Analg 88: 787–92. 2001; 93: 363–9. 18. Oldendorf WH. Brain uptake of radiolabeled amino acids, 9. Ratge D, Bauersfeld W, Wisser H. The relationship of free amines, and hexoses after arterial injection. Am J Physiol and conjugated catecholamines in plasma and cerebrosp- 1971; 221: 1629–39. inal fluid in cerebral and meningeal disease. J Neural 19. Elrod R, Peskind ER, DiGiacomo L et al. Effects of Transm 1985; 62: 267–84. Alzheimer’s disease severity on cerebrospinal fluid nore- 10. Curatolo M. Is epinephrine unfairly neglected for post- pinephrine concentration. Am J Psychiatry 1997; 154: operative epidural mixtures? Anesth Analg 2002; 94: 1381–3. 25–30. 11. Goto F, Fujita N, Fujita T. Cerebrospinal norepinephrine 20. Irjala J, Kanto J, Scheinin M. Monoamine metabolite and concentrations and the duration of epidural analgesia. Can J catecholamine measurements in cerebrospinal fluid in Anaesth 1990; 37: 839–43. determining the quality of the pre-operative night’s sleep. 12. Niemi G, Breivik H. Adrenaline markedly improves thor- Eur J Anaesthesiol 1993; 10: 393–6. acic epidural analgesia produced by a low-dose infusion of 21. Kanto J, Scheinin M. Diazepam and atropine as premedi- bupivacaine, fentanyl and adrenaline after major surgery. A cants: no discrimination by monoamine metabolite and randomised, double-blind, cross-over study with and with- catecholamine measurements in cerebrospinal fluid and out adrenaline. Acta Anaesthesiol Scand 1998; 42: 897–909. plasma. Acta Anaesthesiol Scand 1992; 36: 80–3. 13. Niemi G, Breivik H. Epinephrine markedly improves 22. Wikinski S, Lombardo M, Medina JH et al. Lack of anxio- thoracic epidural analgesia produced by a small-dose lytic effect of diazepam in pre-anaesthetic medication. Br J infusion of ropivacaine, fentanyl, and epinephrine after Anaesth 1994; 72: 694–6. major thoracic or abdominal surgery: a randomized, dou- 23. Wisser H. Katecholamine, Katecholaminmetabolite. In: ble-blinded crossover study with and without epinephrine. Thomas L, ed. Labor und Diagnose: Indikation und Bewertung Anesth Analg 2002; 94: 1598–605. von Laborbefunden, 5th edn. Frankfurt: TH-Books-Verl.-Ges., 14. Reddy SV, Maderdrut JL, Yaksh TL. Spinal cord pharma- 2000:1062–75. cology of adrenergic agonist-mediated antinociception. J Pharmacol Exp Ther 1980; 213: 525–33. 15. Maycock PF, Frayn KN. Use of alumina columns to prepare plasma samples for liquid-chromatographic determination of catecholamines. Clin Chem 1987; 33: 286–7. Address: 16. Eldrup E, Mogensen P, Jacobsen J et al. CSF and plasma Tobias Podranski concentrations of free norepinephrine, dopamine, Outcomes Research Consortium 3,4-dihydroxyphenylacetic acid (DOPAC), 3,4-dihydroxy- The Cleveland Clinic phenylalanine (DOPA), and epinephrine in Parkinson’s 9500 Euclid Ave - P77 disease. Acta Neurol Scand 1995; 92: 116–21. Cleveland, OH 44195 17. Miura Y, Mackensen GB, Nellgard B et al. Effects of USA isoflurane, ketamine, and fentanyl/N2O on concentra- www.OR.org

492