6. Abiko T,Kunikawa M, Higuchi H, Sekino H. Identification and synthesis of a heptapeptide in uremic fluid. Biochem Biophys Res Cornmun 1978;84:184-94.

7. Abiko T, Onodera I, Sekino H. Isolation, structure and biological activity of the Trp-containing pentapeptide from uremic fluid. Biochem Biophys Res Commun 1979;89:813-21. 8. Cueille G, Man NK, Sausse A, Farges JP, Funck-Brentano JL. Technical aspects on middle molecules: separation, isolation, and identification. Artif Organs 1980;4(Suppl):8-12. 9. Le Moel GE, Strecker G, Troupel S, et al. Carbohydrate content of middle molecular weight substances (MMWS) in uremic patients: preliminary results. Artif Organs 1980;4(Suppl):37-40. 10. Zimmerman L, JOrnvall H, Bergstrom J, et al. Characterization of a double conjugate in uremic body fluids. FEBS Lett 1981;129:237-40. 11. Gallice P, Monti JP, Crevat A, Durand C, Murisasco A. A compound from uremic plasma and from normal urine isolated by liquid chromatography and identified by nuclear magnetic resonance. Clin Chem 1985;31:30-4. 12. Monti JP, Gallice P, Crevat A, Murisasco A. Identification by nuclear magnetic resonance and mass spectrometry of a glucuromc acid conjugate of o-hydroxybenzoic acid in normal urine and uremic plasma. Clin Chem 1985;31:1640-2. 13. Monti JP, Gallice P. Braguer D, Durand C, Murisasco A, Crevat A. Identification of two uremic toxins by nuclear magnetic resonance and mass spectrometry. Adv Exp Med Biol 1987; 223:223-6.

14. Braguer D, Gallice P, Monti JP, Murisasco A, Crevat A Inhibition of microtubule formation by uremic toxins: action mech. anism and hypothesis about the active component. Clin Nephrol 1986;25:212-8. 15. Braguer D, Gallice P, Monti JP, Durand C, Murisasco A, Crevat A. A possible regulatory system of microtubule formation among uremic toxins. Adv Exp Med Biol 1987;223:119-23. 16. GalliceP, Fournier N, Crevat A, Briot M, Frayssinet R, Murisasco A. Separationof one uremic middle molecules fraction by high performance liquid chromatography. Kidney Ini
1983;23:764-6. 17. Brenner GS, Hinkley DF, Perkins LM, Weber S.Isomerization of the ascorbic acids. J Org Chem 1964;29:2389-92. 18. Tolbert BM, Dowing M, Carlson RW, Knight MR, Baker EM, Chemistry and metabolism of ascorbic acid and ascorbate sulfate, Ann NY Acad Sci 1975;258:48-69. 19. Shelanski ML, Gaskin F, Cantor CR. Microtubules assembly in the absence of added nucleotides. Proc Natl Acad Sci USA

1973;70:765-8. MD, Lockwood AM, Hwo S, Kirschner NW. Protein factor essential for microtubule assembly.Proc NatI Acad Sci USA 1975;72:1858-62. 21. Laemmli UK. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature (London)
20. Weingarten 1970;227:680-5. 22. Detrich MW, Williams RC. Reversibledissociation of the dimer of tubulin from bovinebrain. Biochemistry1978;17:3900-7.

CLIN. CHEM. 36/7, 1372-1375

(1990)

Do Enzymatic Analyses of Serum Triglycerides Really Need Blanking for Free Glycerol?
Robert H. Jessen,

CindyJ. Dass, and John H. Eckteidt Additional Keyphrases:

of laboratory operations alimentation heparin
.

Distributions of concentrations of free glycerol in clinical plasma obtained for triglyceride assay were compiled to determine the frequency with which increased concentrations of free glycerol posed a potential problem for clinical interpretation of triglyceride results. Clinical histories were studied in patients with increased concentrations of free glycerol to ascertain possible reasons for the increase and to assess the relative clinical importance of glycerol-blankcorrected triglyceride results. Significant increases in free glycerol were very uncommon, usually occurring in patients receiving glycerol-containing hyperalimentation fluids, those receiving heparin (which causes both in vivo and in vitro increases in free glycerol), or those critically ill. Free glycerol was never increased significantly in a large outpatient population. Monitoring lipid metabolism in critically ill patients, or measuring true triglyceride concentrations in patients receiving glycerol-containing fluids, may represent rare exceptions for which glycerol-blank correction is necessary for accurate clinical diagnosis and management. We conclude that there is insufficient justification for the routine expenditure of extra time and reagents to correct most analytical enzymatic triglyceride methods for free glycerol.

economics

The Laboratory Standardization Panel of the National of Health National Cholesterol Education Program has recommended development of analytical methods for triglyceride that correct for free glycerol (1). However, with many automated clinical analyzers, correcting an

Institutes

Department of Laboratory Medicine and Pathology, Box 198 UMHC, Universityof Minnesota, Minneapolis, MN 55455. 1Corresponding author. Received August 14, 1989; accepted April 30, 1990.
1372 CLINICAL CHEMISTRY, Vol. 36, No. 7, 1990

enzymatic triglyceride result for free glycerol can lead to additional analytical costs. Consequently, the clinical nocessity of this practice is not universally accepted, as demonstrated by statistics from the 1989 Comprehensive Chemistry Survey of the College of American Pathologists, which show that <3% of the nearly 4000 participatin laboratories utilize a free-glycerol blank for correction o triglyceride determinations (2). To determine the clinical needs for a free-glycerol correc tion in our patient population, we examined the magnitud and frequency of increased concentrations of free glycero in inpatient and outpatient specimens submitted to th Clinical Laboratories of the University of Minnesota Hos pita! and Clinic for triglyceride determination. For speci mens with increased free glycerol, we then tried to deter mine possible clinical reasons for the increase and evaluate what potential clinical diagnostic or therapeuti errors would have occurred had free-glycerol blanking o the requested triglyceride determination not been per formed.

MaterIals and Methods

To determine glycerol blanks, we used a coupled-enzyme system described by McGowan et al. (3). For total triglycerides, we used a commercially available reagent (Triglycerides GPO, prod. no. 701912; Boehringer Mannheim Diagnostics, Indianapolis, IN 46250) based on a coupledenzyme system similar to that of McGowan et al. We performed all analyses with a Cobas FARA centrifugal analyzer (Roche Diagnostic Systems, Nutley, NJ 07110), using the manufacturers suggested reagent quantities and instrument settings. Briefly, a two-point kinetic calibration curve is constructed from the change in absorbance between 3 s and 5 mm after mixing sample and reagent. The triglyceride and glycerol-blank methods are standardized with 4.52 and 1.13 mmol/L glycerol solutions, respectively. We measured concentrations of triglyceride and free glycerol in 419 inpatient samples and in 339 outpatient samples from two clinics, one a heart-disease prevention clinic and the other an adult medicine clinic, for which triglyceride analysis was requested over a four-week period. We then reviewed the clinical history and hospital course of 20 inpatients whose concentrations of free glycerol exceeded 0.28 mmol/L, to discover possible clinical explanations for the increased free glycerol and to evaluate the clinical relevance of an accurate triglyceride determination in these patients. We specifically sought evidence for previously reported causes of increased free glycerol, e.g., liver disease, diabetes mellitus, hemodialysis, severe stress, and administration of intravenous medications in a glycerol carrier (1). In addition, because the lipid emulsion used at our institution for parenteral nutrition (Intralipid; KabiVitrurn, Inc., Alameda, CA 94501) contains glycerol at 244 mmoIJL, we sought a history of its administration. Heparmn administration causes release of endogenous lipases and consequently in vivo triglyceride hydrolysis. A recent report showed that this hydrolysis continues in vitro (4). To estimate the relative magnitude of the in vivo and in vitro hydrolysis in routinely handled clinical specimens, we administered 5500 USP units of heparmn intravenously to an apparently healthy volunteer, and collected EDTAanticoagulated blood samples from him over the next 80 mm. Blood specimens were centrifuged without delay, and the separated plasma was divided into six parts. One plasma aliquot was immediately frozen on solid CO2; the others were allowed to stand at approximately 23 #{176}C for 30, 60, 120, 240, and 360 mm, after which they were also frozen. All samples were thawed -24 h later and, without delay, were analyzed for free glycerol and total triglyceride by the methods described above.
80

InpetlentS

60
C

40

20

0.55

1.10

Glycerol
100
80

Blank, mmol/L

C 0

:
20 0.55 1.10

Glycerol

Blank, mmol/L (bottom) showed highly variable

Fig.1.Frequency distribution offreeglycerol concentrations in419

inpatients (top) and 339 outpatients The remaining 5% of inpatients

Results
The frequency distributions for free glycerol in our inpatient and outpatient populations are shown in Figure 1. All of our outpatient samples had glycerol-blank values <0.28 mmolIL; in fact, 99% were <0.10 mmol/L. In our inpatients, 97% had <0.28 mmol/L, with the remainder being quite variable. Figure 2 shows the frequency distributions for the potential error in triglyceride concentration that would result if the glycerol blanks were not subtracted. With no free-glycerol blanking for outpatients, 96% would have <10% false increase in triglyceride concentration and 99% would have <15% false increase. With no free-glycerol blanking for inpatients, 88% would have <10% false increase in triglyceride concentration, 92% would have <15% false increase, and 95% would have <20% false increase.

false increases. Although statistically significant (t = 3.97, P <0.001), we found no clinically useful relationship (r = 0.14) between the concentration of free glycerol and total triglyceride, in contrast to a previous report (5). Table 1 lists the clinical diagnoses and the apparent reasons for increases in free glycerol in the 20 patients with a glycerol blank 0.28 mmoIJL. These 20 patients were all inpatients (nine were male), and they ranged in age from less than one year to 81 years. Of the 20 patients, 12(60%) were receiving intravenous lipid infusions at the time the sample for triglyceride assay was collected. Five (25%) of the patients had been receiving heparin therapy at the time of blood collection, in doses ranging from 10000 to 33 000 USP units per day. Of the remaining three patients (15%), one had severe liver dysfunction as a result of venous occlusive disease, and one had poorly controlled insulin-dependent diabetes mellitus. In the third patient, the increased free glycerol was attributed to an epinephrime effect, as this patient had severe postsurgical complications and infection. Figure 3 shows the results of the intravenous heparin study. For this subject the baseline concentration of triglyceride in plasma was -2.15 mmol/L and the free glycerol was only 0.03 mmolJL before heparmn administration. The points on the ordinate of Figure 3 show that intravenous administration of heparin causes a very rapid, but relatively short-lived, in vivo increase in free glycerol. In this particular experiment, free glycerol peaked at 0.42 mmolJL at 24 mm and fell gradually to -0.18 mmol/L at 85 mm post-heparin administration. Figure 3 also demonstrates

the significant in vitro rise in free glycerol and the decrease in triglycerides when plasma from a heparmn-treated subject is allowed to stand at room temperature for even a few hours. CLINICAL CHEMISTRY, Vol. 36, No. 7, 1990
1373

IU

I npett ents

80
-J

C 0 U

40
200 .
0
-

ti

Time InVitro,
40

mm

60
-J

Percent

Error

E 2.20 E

Pre-hepari

.1)

C-, >-

1.35

0 U

#{149}0 L 5,

0.50
L 0 U

100

200 Time In Vitro,

300

40C

mm

20

40

60

Error Fig. 2. Distribution histograms of the percent false increase in triglyceride value due to free glycerol in 419 inpatients (top) and 339 outpatients (bottom) The percent false increase isdefined as [0/(T- 0)1 x 100 where 0 = free glycerol concentration. I = total unblanked triglyceride concentration result (this includes free glycerol), and I - G = true triglyceride concentration

Percent

Fig. 3. Free glycerol concentration (top) and true triglyceride values in a plasma sample corrected for free glycerol concentration (bottom) as a function of the time that the plasma was allowed to stand at room temperature in vitro Each line represents a blood sample drawn at a differenttime after heparin administration:immediatelypre-hepann (#{149}) and 24 mm (#{149}), 37 mm (A), and 67 mm (#{149}) after

Table 1. ClinIcal Diagnoses

Increased
Glycerol
Ag.,

Free Glycerol

and Presumed Causes of In 20 HospItal PatIents

Reason forglycerol

mmelIL years Clinical dIagnosis 1.60 2 Hirschprungs disease 1.19 11 Small bowel Infarction <1 Tnsomy 21 0.72 0.61 14 Rhabdomyosarcoma 0.54 36 Small bowel infarction 72 Gram-negative sepsis 0.47 0.43 <1 Diffuse alveolar damage 57 Myocardlal infarction 0.42 0.32 60 Acute leukemia 0.32 <1 Gastrointestinal mairotation 56 Heart transplant 0.31 0.29 68 Heminephrectomy 55 Myocardial infarction 0.60 0.49 81 Acute pneumonia 0.35 63 Cardiac catheterization 0.34 59 Oral carcinoma 0.28 50 Aspergillus SePSIS Acute liver failure 0.46 0.40 60 Leg wound infection 0.29 50 Type II diabetes

InCrease

Intravenous lipids

Heparin infusion

Liver disease Epinephnne effect

Diabetes mellitus

DiscussIon With many currently

lyzers,
1374 routine

available

automated

clinical

ana-

glycerol-blanking

of triglyceride

results

demands additional reagents and technical time. Consequently, the necessity of making this correction has been questioned (5, 6), and most clinical laboratories do not perform a glycerol blank (2). Evaluation of the distribution of concentrations of free glycerol in our patient population reveals that under most circumstances the vast majority are not of sufficient magnitude to introduce a significant, clinically meaningful error. In outpatients, most triglyceride requests are part of a lipid profile to allow for calculation of low-density lipoprotein cholesterol and evaluation of combined hyperlipidemias as part of a coronary vascular disease risk assessment (7,8). Failure to glycerol-blank the reported triglyceride result would have introduced an erro of >0.11 mmollL in fewer than 1% of our outpatients, and never would have produced an error of >0.28 mmolJL. Errors of this magnitude are relatively insignificant cmi cally for lipid proffling. Triglycerides are also occasionally requested for screen ing for coronary risk inpatients, in which group we sa increased free glycerol much more often; however, thi practice should be discouraged because no baseline lipit status had been established for these patients during hos pitalization. Other inpatient triglyceride requests are fo: metabolic monitoring of critically ill patients, many o whom are receiving intravenous nutritional support. A shown in Table 1, all of our 20 patients with increaset concentrations of free glycerol were critically ill inpatient with multiple, complicated clinical problems, or were re ceiving intravenous lipid solutions that contain glycerol Decreased triglyceride clearance upon administration o lipid-containing solutions may serve as one indicator o progressive organ failure in critically ill patients (9)

CLINICAL CHEMISTRY, Vol. 36, No. 7, 1990

Therefore, in such patients, correction of reported eride concentrations for free glycerol is indicated. Intravenous heparmn administration accounted
next
largest group

triglycfor the

(25%) of patients studied. However, Figure 3 shows that heparin induces rapid hydrolysis of triglycerides both in vivo and in vitro. Subtraction of free

glycerol from the unblanked triglyceride in such a situation would actually lead to underestimation of the subjects true baseline triglyceride. In fact, the in vitro glycerol production often more than doubles the actual in vivo glycerol concentration. Because glycerol is cleared from the plasma rapidly with a half-life of approximately 40 mm (10), neither a glycerol-blanked nor an unblanked triglyceride result will accurately reflect a subjects baseline, pre-heparmn triglyceride concentration. In our opinion, this combined in vivo and in vitro hydrolysis by heparin usually makes triglyceride measurements in patients receiving heparin clinically uninterpretable. Interpretable results can be obtained if one very rapidly freezes the plasma, within minutes of blood collection, or adds lipase inhibitors to the collection tube. These conditions are almost never met routinely in most inpatient critical-care units. The results of this study suggest that routine correction of triglyceride assays for free glycerol is difficult to justify if additional cost or technical effort is involved. Critically ill patients such as those discussed above clearly represent a group that may have increased concentrations of free glycerol and in which glycerol blanking may be appropriate. Although easily recognized clinically, this group of patients may not be easily recognized by the laboratory. The effect of heparin on free-glycerol blanks demonstrates the complexity of clinical interpretation, even when glycerol blanking is performed. In the critically ill patient with complications, interpretation of triglyceride concentrations as a metabolic indicator requires a clinicians complete understanding of all the clinical and analytical variables that ultimately influence the test results. We believe that providing for glycerol-blank correction of triglyceride determination only upon specific request, as we have chosen to offer in our hospital, facilitates the necessary communication between the laboratorian and the clinician. Furthermore, it eliminates the unnecessary practice of routinely measuring a glycerol blank on all samples, the vast majority of which do not need it. As discussed previously, to measure triglycerides without interference from free glycerol, most clinical analyzers must measure glycerol in a separate reaction cuvette.

However, this requires increased time, reagents, and supplies. An alternative solution to the glycerol interference problem that has recently been introduced commercially is use of a reagent designed for instruments capable of a delayed addition of a second reagent (i.e., lipase). This method first preincubates the sample with reagent to consume the preformed glycerol. Although this represents a more economical solution to the free-glycerol interference, it cannot be applied to many of the analyzers in use in clinical laboratories. Consequently, most laboratories do not perform glycerol blanks (2). In summary, we believe that the performance of routine glycerol blanks is clinically unnecessary under most circumstances and, depending on the analyzers a laboratory has available, may not be justifiable economically. Specimens from outpatients do not need glycerol blanking. However, for inpatients there are special, infrequent situations for which measurement of free glycerol along with triglycerides is useful. We believe that glycerol blanking is best performed after some formal communication between the clinician and the clinical laboratory to facilitate accurate interpretation of the results.
References 1. Naito H. Triglycerides,

part H. Clin Chem News 1987;13:8-10. 2. Comprehensive Chemistry Survey, 1989 Set C-D. Northfield, IL: College of American Pathologists, 1990. 3. McGowan MW, Artiss JD, Strandbergh DR, Zak B. A peroxidase-coupled method for the colorimetric determination of serum triglycerides. Clin Chem 1983;29:538-42. 4. Horton G, Cole TG, Gibson DW, Kessler G. Decreased stability of triglycerides and increased free glycerol in serum from heparintreated patients. Clin Chem 1988;34:1847-9. 5. Stinshoff K, Weisshaar D, Staehler F, Hesse D, Gruber W, Steier E. Relation between concentrations of free glycerol and triglycerides in human sera. Clin Chem 1977;23:1029-32. 6. Rautela GS, Slater S, Aryan DA. Assessment of the need for triglyceride blank measurements. Clin Chem 1973;19:1193-5. 7. McNamara JR, Cohn JS, Wilson PW, Schaefer EJ. Calculated values of low-density lipoprotein cholesterol in the assessment of lipid abnormalities and coronary disease risk. Clin Chem 1990;36:36-42. 8. Warnick GR, Knopp RH, Fitzpatrick V, Branson L. Estimating
low-density lipoprotein cholesterol by the Friedewald equation is adequate for classifying patients on the basis of nationally recommended cutpoints. Clin Chem 1990;36:15-9. 9. Cerra FB. Hypermetabolism, organ failure, and metabolicsup-

port. Surgery 1987;101:1-14. 10. Lin ECC. Glycerol utilization and its regulation in mammals. Annu Rev Biochem 1977;46:765-95.

CLINICAL CHEMISTRY, Vol. 36, No. 7,1990 1375