Free
Case Reports  |   April 2006
Hemoglobin M (Milwaukee) Affects Arterial Oxygen Saturation and Makes Pulse Oximetry Unreliable
Author Affiliations & Notes
  • Astrid G. Stucke, M.D.
    *
  • Matthias L. Riess, M.D., Ph.D.
    *
  • Lois A. Connolly, M.D.
  • * Resident, Department of Anesthesiology, † Associate Professor of Anesthesiology.
Article Information
Case Reports / Cardiovascular Anesthesia / Respiratory System
Case Reports   |   April 2006
Hemoglobin M (Milwaukee) Affects Arterial Oxygen Saturation and Makes Pulse Oximetry Unreliable
Anesthesiology 4 2006, Vol.104, 887-888. doi:
Anesthesiology 4 2006, Vol.104, 887-888. doi:
HEMOGLOBIN M (Milwaukee)  is a rare dominant hereditary disorder where glutamate replaces valine in position 67 on the beta chain of the hemoglobin molecule.1 This causes a permanently increased level of methemoglobin ranging between 15 and 30%. Patients are cyanotic but do not exhibit any other symptoms, and life expectancy is unaffected.2 Animal studies have shown that high methemoglobin levels alter pulse oximetry readings.3 In addition, in vitro  studies showed that hemoglobin M (Milwaukee) has reduced oxygen affinity.1 These factors have not been methodically evaluated in a clinical setting for patients with hemoglobin M (Milwaukee). Therefore, we sought to investigate how to achieve optimal oxygenation during general anesthesia and to measure how reliable pulse oximetry was in a patient with this disorder. We present a unique set of arterial blood gas analyses with corresponding pulse oximetry readings at different inspiratory oxygen fractions (Fio2) and two different levels of arterial carbon dioxide tension (Paco2).
Case Report
A 50 yr-old, 79-kg patient presented for pancreaticoduodenectomy. He had been diagnosed with hemoglobin M (Milwaukee) disease as an infant and developed cyanosis at the age of 3–4 months. His mother, brother, and son also live with the disease. He had not experienced any negative side effects of the disease throughout his life. Except for prominent cyanosis, the physical examination results were normal, as were his vital signs. Only pulse oximetry (Nellcor Oxismet XL SpO2; Marquette Medical Systems Inc., Milwaukee, WI) showed a baseline saturation of 92%. Baseline hematocrit was 41%, fractional methemoglobin concentration was 16.3%, and fractional oxygen saturation (Sao2)1was 65.2% on room air. Before induction, two peripheral intravenous lines and a radial artery catheter were placed. Anesthetic management and surgery were uneventful. Total surgical blood loss amounted to 600 ml, which did not require transfusion. With approval by the Institutional Review Board of Froedert Memorial Lutheran Hospital (Milwaukee, Wisconsin), we obtained multiple arterial blood gas samples at different Fio2and two different levels of Paco2. Hemoglobin and hematocrit values were compared between the blood gas analyzer in the operating room (Radiometer Copenhagen ABL 700; Radiometer Medical A/S, Brønshøj, Denmark) and that in the central hospital laboratory (Advia 2120; Bayer, Tarrytown, NY).
We made the following observations: (1) Pulse oximetry showed 100% saturation when the available reduced hemoglobin was fully saturated (Sao2= 85%); for lower Sao2levels, pulse oximetry increasingly overestimated saturation, i.e.  , at an Sao2of 65%, pulse oximetry still showed a saturation of 92% (fig. 1). (2) Hemoglobin M (Milwaukee) showed a decreased Hill coefficient of the oxygen–hemoglobin dissociation curve, i.e.  , it has a decreased affinity for oxygen (fig. 2). Full saturation of the available reduced hemoglobin was only reached at a high arterial oxygen tension (Pao2) of 420 mmHg, which equaled an Fio2of 0.9. (3) A change in Paco2within the clinically used range of 32–42 mmHg did not cause a shift in the oxygen–hemoglobin dissociation curve. (4) For the hematocrit values that were measured in our patient, the cooximeter blood gas analyzer showed values that were on average 10% below that of the laboratory analyzer.
Fig. 1. Relation of fractional oxygen saturation (Sao2) and saturation according to pulse oximetry (Spo2). Pulse oximetry systematically overestimates the actual oxygen saturation with an increased discrepancy at lower Sao2values. There is no difference between lower and higher arterial carbon dioxide tension (Paco2). 
Fig. 1. Relation of fractional oxygen saturation (Sao2) and saturation according to pulse oximetry (Spo2). Pulse oximetry systematically overestimates the actual oxygen saturation with an increased discrepancy at lower Sao2values. There is no difference between lower and higher arterial carbon dioxide tension (Paco2). 
Fig. 1. Relation of fractional oxygen saturation (Sao2) and saturation according to pulse oximetry (Spo2). Pulse oximetry systematically overestimates the actual oxygen saturation with an increased discrepancy at lower Sao2values. There is no difference between lower and higher arterial carbon dioxide tension (Paco2). 
×
Fig. 2. Oxygen–hemoglobin dissociation curve for standard hemoglobin (Hb) at pH of 7.4 and base excess of 0, and hemoglobin M (Milwaukee) at arterial carbon dioxide tension (Paco2) of 42 mmHg, pH of 7.34, and base excess of −2.8 in our patient. The equation for the curve fit is given. The Hill coefficient (n) and thus oxygen affinity are reduced in hemoglobin M (Milwaukee) but appear similar for higher and lower Paco2. Note that dependent on the patient’s methemoglobin level of approximately 15% maximal saturation of the available reduced hemoglobin is reached at a fractional oxygen saturation (Sao2) of approximately 85%. Pao2  50  = oxygen tension at Sao2= 50%; Pao2: arterial oxygen tension; Sao2  max  = maximal Sao2= 100%. 
Fig. 2. Oxygen–hemoglobin dissociation curve for standard hemoglobin (Hb) at pH of 7.4 and base excess of 0, and hemoglobin M (Milwaukee) at arterial carbon dioxide tension (Paco2) of 42 mmHg, pH of 7.34, and base excess of −2.8 in our patient. The equation for the curve fit is given. The Hill coefficient (n) and thus oxygen affinity are reduced in hemoglobin M (Milwaukee) but appear similar for higher and lower Paco2. Note that dependent on the patient’s methemoglobin level of approximately 15% maximal saturation of the available reduced hemoglobin is reached at a fractional oxygen saturation (Sao2) of approximately 85%. Pao2  50  = oxygen tension at Sao2= 50%; Pao2: arterial oxygen tension; Sao2  max  = maximal Sao2= 100%. 
Fig. 2. Oxygen–hemoglobin dissociation curve for standard hemoglobin (Hb) at pH of 7.4 and base excess of 0, and hemoglobin M (Milwaukee) at arterial carbon dioxide tension (Paco2) of 42 mmHg, pH of 7.34, and base excess of −2.8 in our patient. The equation for the curve fit is given. The Hill coefficient (n) and thus oxygen affinity are reduced in hemoglobin M (Milwaukee) but appear similar for higher and lower Paco2. Note that dependent on the patient’s methemoglobin level of approximately 15% maximal saturation of the available reduced hemoglobin is reached at a fractional oxygen saturation (Sao2) of approximately 85%. Pao2  50  = oxygen tension at Sao2= 50%; Pao2: arterial oxygen tension; Sao2  max  = maximal Sao2= 100%. 
×
Discussion
Hemoglobin M (Milwaukee) is a cyanotic condition in which methemoglobinemia is persistent as a fixed percentage of non–oxygen-binding ferric hemoglobin. In our patient, methemoglobin was constantly 15–16%. This condition is different from inherited or acquired methemoglobin reductase deficiency and will therefore not respond to reducing agents such as methylene blue or ascorbic acid. In fact, these patients have normal methemoglobin reductase and can reduce artificially increased methemoglobin back to baseline levels.2,4 Abnormality exists in the structure of hemoglobin itself. The biochemical characteristics of hemoglobin M (Milwaukee) have been extensively investigated in vitro  ; however, the clinical implications have never been evaluated in a systematic fashion. Kinetic studies show that the cooperative binding properties of normal hemoglobin are impaired by increased methemoglobin levels.5,6 We created an oxygen–hemoglobin dissociation curve for our patient from multiple blood gas analyses. Assuming that methemoglobin interferes with the cooperativity of the hemoglobin subunits and considering that changes in cooperativity do not influence the P50  of the curve, we used a Pao250  of 27 mmHg for the curve fit. Our results match in vitro  findings that the Hill coefficient of the oxygen–hemoglobin dissociation curve is decreased in hemoglobin M (Milwaukee) (1.1–1.3)1,5 compared with normal hemoglobin (approximately 2).7 That is, oxygen affinity was greatly reduced in hemoglobin M (Milwaukee), and an Fio2of greater than 0.8 was necessary to achieve near-full saturation of the available reduced hemoglobin.
Because a prominent Bohr effect had been described in the literature for a pH range between 6.75 and 7.85,1 we examined whether mild hyperventilation (Paco2= 32 mmHg, pH 7.45) increased oxygen affinity compared with a normal Paco2(42 mmHg, pH 7.34). However, there seemed to be no difference within this clinically used range.
An important factor for patient care is the discrepancy between oxygen saturation as displayed by pulse oximetry and the actual Sao2as can be determined by blood gas analysis (fig. 1). The effects of methemoglobinemia on pulse oximetry have been delineated.8 In traditional pulse oximetry, light absorbance is measured at two wavelengths: 660 and 940 nm. The pulse oximeter then calculates the ratio of absorbance at both wavelengths, compares this number with historic data from healthy volunteers, and thus generates a value for the arterial oxygen saturation.8 An absorption ratio of 1:1 represents an Sao2of 85% when only oxyhemoglobin and reduced hemoglobin are present.8 Incidentally, methemoglobin shows an absorption ratio of approximately 1:1 at these wavelengths, so that high levels of methemoglobin (> 35%) produce a value for pulse oximetry of approximately 85%, which is nearly independent of the actual Sao2.8 In our patient with a methemoglobin level of approximately 15%, pulse oximetry yielded values that overestimated the true Sao2with increasing discrepancy at lower Sao2. The same tendency has also been shown in an animal model.3 Cooximetry measures the light absorbance at four or more discrete wavelengths and can thus distinguish between oxyhemoglobin, reduced hemoglobin, carboxyhemoglobin, and methemoglobin fractions. In patients with significant methemoglobinemia, it is essential to obtain true values for Sao2with cooximetry. Once a ratio is established between pulse oximetry and cooximetry, pulse oximetry is valuable for trending oxygenation.
Last, in our patient, cooximetry underestimated the hematocrit by approximately 10–15%. Because hemoglobin M patients are mildly anemic, a nominal transfusion threshold may be reached earlier, and a control hematocrit value should be obtained from the laboratory before the decision for transfusion is made.
In conclusion, hemoglobin M (Milwaukee) shows greatly decreased oxygen affinity compared with normal hemoglobin. High Fio2may be necessary to achieve optimal saturation of the available reduced hemoglobin. Mild hyperventilation does not influence oxygen affinity. Pulse oximetry values are falsely high, and oxygen saturation should be verified with blood gas analysis.
The authors thank Edward J. Zuperku, Ph.D. (Professor, Department of Anesthesiology, Zablocki Veterans Administration Medical Center, Milwaukee, Wisconsin), for his support with the statistical analysis.
References
Udem L, Ranney H, Bunn H: Some observation on the properties of hemoglobin M Milwaukee-1. J Mol Biol 1970; 48:489–98Udem, L Ranney, H Bunn, H
Pisciotta AV, Ebbe SN, Hinz JE: Clinical and laboratory features of two variants of methemoglobin M disease. J Lab Clin Med 1959; 54:73–87Pisciotta, AV Ebbe, SN Hinz, JE
Barker SJ, Tremper KK, Hyatt J: Effects of methemoglobinemia on pulse oximetry and mixed venous oximetry. Anesthesiology 1989; 70:112–7Barker, SJ Tremper, KK Hyatt, J
Nagai M, Yoneyama Y: Reduction of methemoglobins M Hyde Park, M Saskatoon, and M Milwaukee by ferredoxin and ferredoxin-nicotinamide adenine dinucleotide phosphate reductase system. J Biol Chem 1983; 258:14379–84Nagai, M Yoneyama, Y
Makino N, Sugita Y, Nakamura T: Kinetic studies on the cooperative ligand binding by hemoglobin M Milwaukee. J Biol Chem 1979; 254:2353–7Makino, N Sugita, Y Nakamura, T
Fung LW, Minton AP, Ho C: Nuclear magnetic resonance study of heme-heme interaction in hemoglobin M Milwaukee: Implications concerning the mechanism of cooperative ligand binding in normal hemoglobin. Proc Natl Acad Sci U S A 1976; 73:1581–5Fung, LW Minton, AP Ho, C
Roughton FJ, Severinghaus JW: Accurate determination of O2 dissociation curve of human blood above 98.7 percent saturation with data on O2 solubility in unmodified human blood from 0 degrees to 37 degrees C. J Appl Physiol 1973; 35:861–9Roughton, FJ Severinghaus, JW
Tremper KK, Barker SJ: Pulse oximetry. Anesthesiology 1989; 70:98–108Tremper, KK Barker, SJ
Fig. 1. Relation of fractional oxygen saturation (Sao2) and saturation according to pulse oximetry (Spo2). Pulse oximetry systematically overestimates the actual oxygen saturation with an increased discrepancy at lower Sao2values. There is no difference between lower and higher arterial carbon dioxide tension (Paco2). 
Fig. 1. Relation of fractional oxygen saturation (Sao2) and saturation according to pulse oximetry (Spo2). Pulse oximetry systematically overestimates the actual oxygen saturation with an increased discrepancy at lower Sao2values. There is no difference between lower and higher arterial carbon dioxide tension (Paco2). 
Fig. 1. Relation of fractional oxygen saturation (Sao2) and saturation according to pulse oximetry (Spo2). Pulse oximetry systematically overestimates the actual oxygen saturation with an increased discrepancy at lower Sao2values. There is no difference between lower and higher arterial carbon dioxide tension (Paco2). 
×
Fig. 2. Oxygen–hemoglobin dissociation curve for standard hemoglobin (Hb) at pH of 7.4 and base excess of 0, and hemoglobin M (Milwaukee) at arterial carbon dioxide tension (Paco2) of 42 mmHg, pH of 7.34, and base excess of −2.8 in our patient. The equation for the curve fit is given. The Hill coefficient (n) and thus oxygen affinity are reduced in hemoglobin M (Milwaukee) but appear similar for higher and lower Paco2. Note that dependent on the patient’s methemoglobin level of approximately 15% maximal saturation of the available reduced hemoglobin is reached at a fractional oxygen saturation (Sao2) of approximately 85%. Pao2  50  = oxygen tension at Sao2= 50%; Pao2: arterial oxygen tension; Sao2  max  = maximal Sao2= 100%. 
Fig. 2. Oxygen–hemoglobin dissociation curve for standard hemoglobin (Hb) at pH of 7.4 and base excess of 0, and hemoglobin M (Milwaukee) at arterial carbon dioxide tension (Paco2) of 42 mmHg, pH of 7.34, and base excess of −2.8 in our patient. The equation for the curve fit is given. The Hill coefficient (n) and thus oxygen affinity are reduced in hemoglobin M (Milwaukee) but appear similar for higher and lower Paco2. Note that dependent on the patient’s methemoglobin level of approximately 15% maximal saturation of the available reduced hemoglobin is reached at a fractional oxygen saturation (Sao2) of approximately 85%. Pao2  50  = oxygen tension at Sao2= 50%; Pao2: arterial oxygen tension; Sao2  max  = maximal Sao2= 100%. 
Fig. 2. Oxygen–hemoglobin dissociation curve for standard hemoglobin (Hb) at pH of 7.4 and base excess of 0, and hemoglobin M (Milwaukee) at arterial carbon dioxide tension (Paco2) of 42 mmHg, pH of 7.34, and base excess of −2.8 in our patient. The equation for the curve fit is given. The Hill coefficient (n) and thus oxygen affinity are reduced in hemoglobin M (Milwaukee) but appear similar for higher and lower Paco2. Note that dependent on the patient’s methemoglobin level of approximately 15% maximal saturation of the available reduced hemoglobin is reached at a fractional oxygen saturation (Sao2) of approximately 85%. Pao2  50  = oxygen tension at Sao2= 50%; Pao2: arterial oxygen tension; Sao2  max  = maximal Sao2= 100%. 
×