In the near future, anesthesiologists will administer artificial blood substitutes to their patients as an alternative to conventional allogenic transfusions (1–3). The blood substitutes now being developed consist of either fluorocarbon emulsions or solutions of chemically modified or recombinant hemoglobin. Unfortunately, a serious problem will complicate their introduction into routine clinical use: these blood substitutes are already known to interfere with a wide variety of clinical analyzers (4–16). In a previous communication, we reported the interference that a fluorocarbon-based blood substitute (perflubron emulsion, or Oxygent™) causes in the measurements of cooximeters (15). The purpose of this study was to determine whether hemoglobin-based blood substitutes affect the measurements of cooximeters.
Cooximeters are dedicated multiwavelength spectrophotometers that automate the analysis of blood samples. The various cooximeters now in use typically measure the optical absorbance of blood at 4 up to 128 different wavelengths and automatically compute the fractional concentrations of the four major hemoglobin species (oxy-, deoxy-, carboxy-, and methemoglobin) as well as the total hemoglobin concentration (tHb). Table 1 shows six different ways in which the blood substitutes now in development could possibly interfere with the spectrophotometric measurements of cooximeters.
When expressed as a percentage, the fractional concentration of oxyhemoglobin (%O2Hb) is defined asMATHwhere O2Hb, HHb, COHb, and MetHb are the amounts of oxy-, deoxy-, carboxy-, and methemoglobin, respectively. The fractional concentrations of the other hemoglobin species are defined similarly. The fractional concentration of oxyhemoglobin should not be confused with functional saturation, the definition of which is similar but does not include carboxy- and methemoglobin in the denominator. In contrast to these unitless fractions, the tHb is expressed in absolute units and consists of the sum of these four major species.
The companies now developing hemoglobin-based blood substitutes are altering the hemoglobin molecule chiefly for two purposes: to adjust the molecule’s affinity for oxygen and thus achieve effective oxygen transport to tissue and to maintain a large enough molecular size so that the hemoglobin remains in the blood-vascular compartment for a therapeutically sufficient time, and to prevent renal toxicity. It is not known to what extent the various molecular modifications will alter the optical absorbance spectra of the major hemoglobin species and thus affect the spectrophotometric measurements of cooximeters. Therefore, we have investigated the effects of five hemoglobin-based blood substitutes on the measurements of eight different oximeters and cooximeters.
To obtain the materials for this study, we wrote to eight different companies that are now developing hemoglobin-based blood substitutes for eventual clinical use. Four of them provided samples of their products (Apex Bioscience, Research Triangle Park, NC; Baxter Healthcare Corp., Deerfield, IL; Hemoglobin Therapeutics, a subsidiary of Baxter; and Hemosol, Inc., Etobicoke, Ontario, Canada). The other companies either refused to provide samples or were unwilling to do so under the usual terms and conditions of material transfer agreements that are acceptable to the University of Texas. We wanted to include a blood substitute derived from bovine hemoglobin because some of the companies that declined are developing their products from bovine blood. Therefore, we purchased a product approved for veterinary use (Oxyglobin®, Biopure Corporation, Cambridge, MA). Table 2 contains detailed information about the five hemoglobin-based blood substitutes we studied, the species from which they are derived, and the molecular alterations they exploit.
The eight test instruments used in this study were the AVL Omni 6 (AVL Scientific Corporation, Roswell, GA), the AVOXimeters 1000 and 4000 (Avox Systems, Inc., San Antonio, TX), the CC270 CO-Oximeter (Ciba Corning Diagnostics Corp. [now Bayer Corporation], Medfield, MA), the IL482 and IL682 CO-Oximeters and the IL Synthesis 35 (Instrumentation Laboratory, Lexington, MA), and the OSM3 Hemoximeter (Radiometer America, Inc., Westlake, OH). The detailed specifications and operating characteristics of these instruments have been published elsewhere (17). Each of these instruments was calibrated and operated in accordance with the respective manufacturer’s instructions. Some of the test instruments report the fractional concentration of reduced hemoglobin (%HHb), but others do not. Therefore, in the instruments that do not report %HHb, we calculated %HHb as 100% minus the algebraic sum of the percentages of the other three species. The AVOXimeter 1000E (18–20) is a whole-blood oximeter that measures tHb and the fractional concentration of oxyhemoglobin, but it does not report carboxy- or methemoglobin. Therefore, no carboxy- or methemoglobin data are shown for the AVOXimeter 1000E.
Standardized hemoglobin solutions (Level 2, Multi-4™ CO-Oximetry Controls) were obtained from Instrumentation Laboratory, Lexington, MA. Instrumentation Laboratory’s Level 2 control is a stabilized solution of purified human hemoglobin. This buffered solution, in our experience, has been stable for months at a time under refrigeration, although its exact composition changes slightly from one lot to another. Using this material from one particular lot provided a method for comparing the eight different instruments’ measurements on unaltered human hemoglobin and also for checking the reproducibility of each instrument.
Some of the blood substitutes shown in Table 2 were sent to us frozen and were kept frozen until the day of the experiment. Each material was analyzed in the original, undiluted state. The experimental protocol consisted of analyzing the Level 2 control and each of the five blood substitutes on the eight test instruments. After each container was opened, the material was kept on ice in a sealed container and analyzed the same day on each of the test instruments; if next-day measurements had to be made on some instruments, the material was reanalyzed on the first instruments to make sure that the composition had not changed.
The first series of experiments consisted of taking repeated readings on Level 2 Multi-4 CO-Oximetry Controls. The results are shown in Table 3, which, like the tables that follow, contains the averages of five readings on each of the eight test instruments. The lower portion of the table contains the all-instrument average for each analyte, the standard deviations about the all-instrument mean (sd), and the range of each analyte’s observed readings. The ranges were computed from the two instruments that gave the highest and lowest average readings for a particular analyte.
The insert sheet with the Level 2 Multi-4 Controls listed reference ranges for the AVOX 1000, IL482/682, and Radiometer OSM3. The average readings on all of these instruments, as shown in Table 3, were within the reference ranges shown on the insert sheet for the particular lot that we used. The insert sheet did not list reference ranges for the other instruments we tested. Nevertheless, Table 3 shows that the eight test instruments were in good agreement for all of the fractional concentrations of the four species of hemoglobin. Similarly, most of the instruments gave tHb readings within 0.5 g/dL of the all-instrument mean of 13.61 g/dL; however, there was a 1.82 g/dL difference in the tHb between the lowest average reading on IL Synthesis (12.54 g/dL) and highest average reading on the Radiometer OSM3 (14.36 g/dL).
Because Level 2 Multi-4 CO-Oximetry Controls are made from native, adult human hemoglobin that has not been molecularly altered, we could expect the eight test instruments to show similar good agreement on the hemoglobin-based blood substitutes if the manufacturers’ respective molecular alterations, species differences, and production methods have not altered the optical absorbance spectra of the four species of hemoglobin or introduced exogenous interferents. By contrast, interference in the measurement of a given analyte produces greater instrument-to-instrument variability in the measurements and larger sd about the all-instrument mean.
Table 4 shows the average readings on Apex PHP. The wider ranges and larger sd about the all-instrument means in comparison with Table 3 indicate that interference occurred in the measurements of oxy-, deoxy-, carboxy-, and methemoglobin. Wider ranges in the fractional concentration readings were also apparent with the other blood substitutes, as can be seen in the tables that follow. In contrast to the greater variability in the fractional concentration readings, Apex PHP yielded measurements of tHb that were a bit more consistent than those on Multi-4 Controls.
From Table 4, two other trends in instrument performance can be inferred that generally apply to the other blood substitutes. First, the OSM3 usually gave the highest reading of tHb, whereas the AVL Omni gave the lowest tHb reading. Second, the AVL Omni consistently gave the lowest readings of %O2Hb on all of the blood substitutes we studied. In the case of Apex PHP, the AVL’s reading of 85.78% was 8.43% O2Hb below the all-instrument mean and 9.63% O2Hb less than the average of the other instruments.
In comparison with some of the other blood substitutes we tested, Apex PHP had a relatively small concentration of MetHb. The manufacturer stated that the particular lot we tested had a MetHb concentration of 1.9% as measured on their IL482. As Table 4 shows, our measurements on the IL482 averaged 2.62%, indicating that subsequent handling and storage in the freezer did not appreciably increase the MetHb fraction.
Table 5 shows the average readings on Hemolink. The wider ranges and larger sd about the all-instrument means in comparison with Table 3 indicate that interference occurred in the measurements of oxy-, deoxy-, carboxy-, and methemoglobin. Hemolink, like several other blood substitutes we studied, contained fairly large levels of MetHb. In fact, all instruments gave consistently high readings that averaged 24.87% MetHb. In response to this finding, the manufacturer indicated that release criteria for the product specified a MetHb fraction of 15% or less, with values typically <10%.
Interestingly, all seven of the instruments that report %COHb indicated that Hemolink contained a nonnegligible concentration of COHb. Readings ranged from a low of 6.10% COHb to a high of 12.10%, with an all-instrument average of 8.56% COHb. We were surprised at this finding and decided to equilibrate Hemolink with room air by using a tonometer (Model 237; Instrumentation Laboratory, Lexington, MA). After 2 h of tonometry, the %COHb measurements decreased consistently, but the differences among instruments persisted. In response to the COHb fractions we measured, the manufacturer indicated that the expected concentration was 2.6%; the method of measurement was not stated. Some companies use carbon monoxide in the virus inactivation process; however, we do not know if that explanation applies here.
Table 6 shows the average readings on EQ1 DCLHb. The wider ranges and larger sd in comparison with Table 3 indicate that interference occurred in the measurements of the fractional concentrations of the four major hemoglobin species. Except for a 2.60% COHb reading on the OSM3, all instruments indicated that DCLHb was devoid of COHb. As the table shows, this blood substitute, like Hemolink, contained fairly large levels of MetHb; all instruments gave consistently high readings that averaged 17.87% MetHb. The manufacturer indicated that lower %MetHb levels could be expected in fresher samples of DCLHb.
Table 7 shows the average readings on EQ2 rHb1.1. The wider ranges and larger sd in comparison with Table 3 indicate that interference occurred in the measurements of the fractional concentrations of the four major hemoglobin species. Measurements of %COHb were inconsistent and ranged from −0.64% on the AVOXimeter 4000 to 8.3% on the OSM3. Tonometering rHb1.1 with room air for 3 h lowered %COHb readings on the OSM3, IL482, and AVOX 4000 but did not eliminate the discrepancies among instruments. Of all the materials we studied, rHb1.1 contained the largest concentration of MetHb (38.17%), and the seven instruments were in good agreement on %MetHb. The manufacturer indicated that lower %MetHb levels could be expected in fresher samples of rHb1.1.
Table 8 shows average readings on Oxyglobin. The optical extinction coefficients of the four major hemoglobin species vary slightly from one species to the next. Therefore, we included Oxyglobin in the study because it is derived from bovine blood. The insert sheet with this material stated that it had a tHb of 13 g/dL, which our all-instrument average confirmed (12.78 g/dL). The insert sheet described Oxyglobin as a “clear dark purple solution,” and our initial measurements of the %O2Hb confirmed that this material was highly deoxygenated with a %O2Hb of approximately 6% or less. Because it was difficult to avoid inadvertently oxygenating this material during our tests on eight instruments at three different sites, we deliberately oxygenated the Oxyglobin to approximately 82%. Fortunately, the %O2Hb readings were stable at this level, and we were able to continue the tests, obtaining an all-instrument average of 82.50%.
The larger sds in comparison with Table 3 indicate that interference occurred in the measurements of oxy-, deoxy-, carboxy-, and methemoglobin. Except for the AVL, which again gave the lowest %O2Hb reading (73.88%), the test instruments were in fairly good agreement on %O2Hb. In fact, if the AVL’s %O2Hb reading were deleted, the maximum difference among instruments would be 6.84% rather than 13.04%.
Oxyglobin had the interesting property of causing negative COHb readings on all but one of the test instruments. Of course, fractional concentrations less than zero or more than 100% are physically impossible, but because of the inaccuracy of spectrophotometric measurements, values slightly less than zero or a bit more than 100% are legitimate cooximeter readings that occur occasionally with clinical samples. The AVL has software that suppresses negative readings, so no %COHb data are shown for that instrument.
To determine the extent to which the bovine origin of Oxyglobin affected the results, we reanalyzed Oxyglobin on the OSM3 by using optical extinction coefficients for bovine rather than human hemoglobin. In the bovine mode, the OSM3’s readings of oxy-, carboxy-, and methemoglobin increased by only 0.50%, 0.30%, and 0.54%, respectively, and tHb increased by 0.2 g/dL. Therefore, the errors caused by analyzing Oxyglobin with optical extinction coefficients for human rather than bovine hemoglobin were small and cannot explain, for example, the 13.04% disparity in the %O2Hb readings.
The benefits of new technology never accrue without costs. When compared with donor blood, the potential clinical and economic benefits of blood substitutes are obvious and enormous: increased availability, decreased costs, elimination of infectious agents, no blood types to cross-match, longer shelf-life, etc. (2,3). Unfortunately, reports (4,6–14) already indicate that hemoglobin-based blood substitutes interfere with many common clinical analyzers, particularly those that require hemoglobin-free plasma or serum. An additional hidden cost of hemoglobin-based blood substitutes is that, particularly during their introduction, they will be a potential source of considerable confusion if all personnel who report or interpret clinical data, from clinicians to medical technologists and laboratorians, are not informed that a patient has received a hemoglobin-based blood substitute. To cite a few examples: the well-known 3:1 ratio between hematocrit and tHb often provides a useful rule of thumb for estimating blood’s oxygen-carrying capacity from hematocrit, but measurements of hematocrit will be less meaningful, and this approximation will no longer hold, if a significant fraction of the patient’s hemoglobin is in solution. Similarly, a plasma hemoglobin concentration in the grams per deciliter range would ordinarily indicate a severe hemolytic disorder, but large plasma hemoglobin levels are to be expected after the administration of these products. In addition, unexpectedly high readings of carboxy- or methemoglobin could be merely puzzling or cause serious clinical concern, and inexplicably negative COHb values would undermine both the laboratory’s and the clinician’s confidence in cooximetry results.
The goal of this study was to determine whether or not hemoglobin-based blood substitutes affect the measurements of oximeters and cooximeters. As Tables 3–8 show, hemoglobin-based blood substitutes do indeed interfere with the measurements of oximeters and cooximeters. In the case of some blood substitutes, the interference was obvious and affected all but one of the test instruments, for example, the negative COHb readings in Table 8. In the case of other products, the interference was less obvious. For example, as Table 5 shows, all test instruments indicated that Hemolink contained COHb (which tonometry with room air confirmed), but the instruments gave quantitatively inconsistent COHb results both before and after tonometry. Although the data demonstrate convincingly that these hemoglobin-based blood substitutes interfere with the measurements of the test instruments, the next question is whether or not some of the measurements would still be clinically useful (albeit possibly less accurate).
Two encouraging aspects of this study indicate that some of the measurements in the presence of hemoglobin-based blood substitutes would be clinically useful. First, we analyzed each material in its undiluted state, whereas in the most frequent anticipated clinical applications, such as blood replacement during surgery, these products would replace only a fraction of the patient’s blood volume. Therefore, the results presented here represent the worst-case interference that hemoglobin-based blood substitutes cause in the measurements of the test instruments. However, it can also be anticipated that, in desperate attempts to save lives, clinicians will, in some cases of severe hemorrhage, almost completely replace a patient’s blood volume with one of these products.
The second encouraging aspect of this study is that if clinicians and laboratorians know how a particular hemoglobin-based blood substitute affects the instrument they use, they may be able to extract useful from nonsensical results. In an anesthetized patient receiving a blood substitute, the two analytes of most frequent interest will be %O2Hb and tHb. If one compares the ranges of values reported for tHb in Table 3 with the ranges in Tables 4–8, it is apparent that the test instruments gave tHb readings on the blood substitutes that were just as consistent as those on solutions of unaltered, adult human hemoglobin. In the case of %O2Hb, all of the test instruments agreed with one another, with the possible exception of the AVL Omni. Although the AVL Omni’s average reading on control hemoglobin solutions was only 2.7% below the mean of the other instruments’ readings, the AVL Omni’s readings on blood substitutes were consistently lower than the average of the other instruments’ readings: Apex (−9.6% O2Hb), Hemolink (−8.2% O2Hb), DCLHb (−4.9% O2Hb), rHb1.1 (−7.5% O2Hb), and Oxyglobin (−9.9% O2Hb). However, these values represent the worst-case interference experienced by the instrument that was most affected by the hemoglobin-based blood substitutes. Therefore, with the possible exception of the AVL Omni, the instruments we tested will provide clinically useful measurements of %O2Hb, even when the other measurements are grossly erroneous or nonsensical.
The data presented here leave us convinced that introducing hemoglobin-based blood substitutes into routine clinical use without causing unnecessary confusion will require greater cooperation between instrument manufacturers and the companies developing hemoglobin-based blood substitutes.
The authors express their gratitude to Charles F. Mountain of Instrumentation Laboratory for the loan of an Instrumentation Laboratory Synthesis and to Ana Campa and Rosemary Delarosa of University Hospital and Lynn Aranda at Audie Murphy VA Hospital for their cheerful and expert assistance.
1. Lamy ML, Daily EK, Brichant JF, et al. Randomized trial of diaspirin cross-linked hemoglobin solution as an alternative to blood transfusion after cardiac surgery. Anesthesiology 2000; 92: 646–56.
2. Vlahakes GJ. Hemoglobin solutions come of age. Anesthesiology 2000; 92: 637–8.
3. Dietz NM, Joyner MJ, Warner MA. Blood substitutes: fluids, drugs, or miracle solutions? Anesth Analg 1996; 82: 390–405.
4. Ma Z, Monk TG, Goodnough LT, et al. Effect of hemoglobin- and perflubron-based oxygen carriers on common clinical laboratory tests. Clin Chem 1997; 43: 1732–7.
5. Scott GM, Kucik DF, Goodnough LT, Monk TG. Blood substitutes: evolution and future applications. Clin Chem 1997; 43: 1724–31.
6. Callas DD, Clark TL, Moreira PL, et al. In vitro effects of a novel hemoglobin-based oxygen carrier on routine chemistry, therapeutic drug, coagulation, hematology, and blood bank assays. Clin Chem 1997; 43: 1744–8.
7. Ali AC, Mihas CC, Campbell JA. Interferences of o-raffinose cross-linked hemoglobin in three methods for serum creatinine. Clin Chem 1997; 43: 1738–43.
8. Ali AC, Campbell JA. Interference of o-raffinose cross-linked hemoglobin with routine Hitachi 717 assays. Clin Chem 1997; 43: 1794–6.
9. Balion CM, Champagne PA, Ali AC. Evaluation of HemogloBind for removal of o-raffinose cross-linked hemoglobin (Hemolink) from serum. Clin Chem 1997; 43: 1796–7.
10. Moreira PL, Lansden CC, Clark TL, Gawryl MS. Effect of Hemopure on the performance of Ektachem and Hitachi clinical analyzers. Clin Chem 1997; 43: 1790–1.
11. Kazmierczak SC, Catrou PG, Boudreau D. Simplified interpretative format for assessing test interference: studies with hemoglobin-based oxygen carrier solutions. Clin Chem 1998; 44: 2347–52.
12. Moreira PL, Lansden CC, Clark TL, Gawryl MS. Effect of Hemopure on prothrombin time and activated partial thromboplastin time on seven coagulation analyzers. Clin Chem 1997; 43: 1792.
13. Sarkozi L, Jacobs E, Clark T, et al. Effect of hemoglobin-based oxygen carrier-201 on common chemistry laboratory procedures. Clin Chem 1997; 43: 1792–4.
14. Wolthuis A, Peek D, Scholten R, et al. Effect of the hemoglobin-based oxygen carrier HBOC-201 on laboratory instrumentation: cobas integra, chiron blood gas analyzer 840, Sysmex SE-9000 and BCT. Clin Chem Lab Med 1999; 37: 71–6.
15. Shepherd AP, Steinke JM. CO-Oximetry interference by perflubron emulsion: comparison of hemolyzing and non-hemolyzing instruments. Clin Chem 1998; 44: 2183–90.
16. Cuignet OY, Wood BL, Chandler WL, Spiess BD. A second-generation blood substitute (perfluorodichlorooctane emulsion) generates spurious elevations in platelet counts from automated hematology analyzers. Anesth Analg 2000; 90: 517–22.
17. Oximeters, in vitro, multiwavelength. Plymouth Meeting, PA. ECRI, 1998.
18. Freeman GL, Steinke JM. Evaluation of two oximeters for use in cardiac catheterization laboratories. Cathet Cardiovasc Diagn 1993; 30: 51–7.
19. Gong AK. Near-patient measurements of methemoglobin, oxygen saturation, and total hemoglobin: evaluation of a new instrument for adult and neonatal intensive care. Crit Care Med 1995; 23: 193–201.
20. Bailey SR, Russell EL, Martinez A. Evaluation of the AVOXimeter: precision, long-term stability, linearity, and use without heparin. J Clin Monit 1997; 13: 191–8.