the preanalytic phase an important component of...

Clinical Chemistry / THE PREANALYTIC PHASE

The Preanalytic Phase

An Important Component of Laboratory Medicine

Sheshadri Narayanan, PhD*

Key Words: Preanalytic phase variables; Physiologic; Specimen collection and processing; Analyte stability; Endogenous interference factors

A b s t r a c t

The preanalytic phase is an important componentof total laboratory quality. A wide range of variablesthat affect the result for a patient from whom aspecimen of blood or body fluid has been collected,including the procedure for collection, handling, andprocessing before analysis, constitute the preanalyticphase. Physiologic variables, such as lifestyle, age, and sex, and conditions such as pregnancy andmenstruation, are some of the preanalytic phase factors.Endogenous variables such as drugs or circulatingantibodies might interact with a specific method to yieldspurious analytic results. The preanalytic phasevariables affect a wide range of laboratory disciplines.

In recent years, considerable attention has been given tothe influence of preanalytic effects on laboratory results.Compendia of preanalytic effects and books have addressedthe effect of the preanalytic phase on laboratory results.1,2

Preanalytic variables can be grouped broadly under 3categories: physiologic, specimen collection, and influence orinterference factors. In this review, I address these variablesand provide examples of their effects across a broad range oflaboratory disciplines. Examples of preanalytic variablesaffecting mainly clinical chemistry, hematology, and coagula-tion tests are discussed.

Admittedly, given the comprehensive range of subjectmatter that spans the realm of the preanalytic phase, thisreview must be selective. Hence, a mere passing mention ismade in the concluding section, “Summary Perspectives,” oftopics such as blood gases, molecular biology, tumormarkers, and problem analytes, such as cytokines, homocys-teine, and fibrinolytic activators and inhibitors, all of whichcould be the subject matter for more than 1 review article.

Even so, an attempt is made to provide a comprehensiveoverview of physiologic variables, specimen collection vari-ables, including a discussion of traditional anticoagulants andnewer stabilizing additives for newer problem analytes, andclinically relevant endogenous interferences.

Physiologic Variables

The effects of age, sex, time, season, altitude, condi-tions such as menstruation and pregnancy, and lifestyle aresome of the physiologic variables that affect laboratoryresults.

Am J Clin Pathol 2000;113:429-452 429© American Society of Clinical Pathologists

by guest on Decem

ber 17, 2015http://ajcp.oxfordjournals.org/

Dow

nloaded from

http://ajcp.oxfordjournals.org/

Narayanan / THE PREANALYTIC PHASE

Age

The effect of age ❚Table 1❚ on laboratory results hasbeen well recognized to the point that separate referenceintervals have been used to distinguish pediatric, adolescent,adult, and geriatric populations. Bone growth and develop-ment in a healthy growing child is characterized by increasedactivity of bone-forming cells, the osteoblasts, that secretethe enzyme, alkaline phosphatase. Hence, the activity of thisenzyme is approximately 3 times higher in a growing childcompared with the activity in a healthy adult.3,4

A substantial increase in the RBC count in the neonatescompared with that in adults is the reason glucose is metabo-lized very rapidly in neonates.2 Increased arterial oxygencontent within a few days of birth leads to an increase inhemoglobin levels owing to the destruction of the RBCs.Serum bilirubin levels then increase, because the immatureliver lacks enzymes to convert bilirubin to the water-solublebilirubin diglucuronide.2 The total number of neutrophils areat a maximum for 1 to 2 days after birth. In the neonate, themonocyte count is increased for up to 2 weeks after birth,while the eosinophil count stays elevated for up to 1 weekafter birth.5,6 The lymphocyte count is increased markedly atbirth and remains elevated in children up to 4 years of age.6

In contrast, the basophil count is elevated only transiently forup to 1 day after birth. A substantial decrease in serum uricacid levels is noted between days 1 and 6 after birth.2

Age also affects renal function, as reflected by an effecton creatinine clearance, which decreases progressively witheach decade.7 For example, the average value for creatinineclearance at age 30 years is 140 mL/min (2.33 mL/s) per1.73 m2 of body surface area. In contrast, at age 80 years, thecreatinine clearance value decreases to 97 mL/min (1.62mL/s) per 1.73 m2 of body surface area.7 Beyond the age of50 years, the decrease in creatinine clearance is also relatedto a decrease in muscle mass.7,8 However, when interpreting

changes in renal function solely on the basis of aging, theoverlap of disease processes, such as the presence of renalvascular disease in elderly patients, must be considered.Secondary risk factors, such as genetics and obesity, mayhave a role in the decrease in glucose tolerance with aging.9

Homeostatic control by the hypothalamic-pituitary-adrenal axis is compromised because of aging. Hence, thereis an increase in plasma corticotropin and corticosteroidlevels due to aging.10,11 Usually in young adults, the concen-tration of the cytokine, interleukin (IL)-6, in serum is lowand is undetectable in absence of inflammation.11 However,IL-6 levels are detectable readily during middle age. Also, inhealthy elderly adults, serum IL-6 levels are increased appar-ently due to the loss of the normal regulation of IL-6 levels.The decrease in dehydroepiandrosterone sulfate with age,which normally has a suppressive effect on IL-6, is related tothe increase in IL-6 levels.12 Increased IL-6 levels, in turn,increase the percentage of cells bearing the receptor for thecytokine, transforming growth factor beta, a powerfulimmunosuppressive agent, thus providing an explanation forthe age-related immunosuppressive phenomena.13

Increases in the level of plasma antidiuretic hormone (orarginine vasopressin) have been related to aging with levelssignificantly higher in 1 study of healthy subjects 53 to 87years of age (4.7 – 0.6 pg/mL [4.4 – 0.6 pmol/L] in 68healthy subjects) compared with subjects 21 to 51 years ofage (2.1 – 0.2 pg/mL [1.9 – 0.19 pmol/L] in 45 healthysubjects).14

Thyroid homeostasis apparently is affected duringaging. Thyrotropin levels in serum have been reported to be38% higher in an elderly group (mean age, 79.6 years)compared with a younger group of subjects (mean age, 39.4years).15 In the same study, serum triiodothyronine levelswere 11% lower in the older age group compared with levelsin the younger group of subjects.

430 Am J Clin Pathol 2000;113:429-452 © American Society of Clinical Pathologists

❚Table 1❚Effect of Age on Laboratory Results

Age Group/Effects References

Neonate Increased RBC count leading to increased glucose metabolism Guder et al2Increased neutrophil, monocyte, eosinophil, and lymphocyte counts Monroe et al5; Weinberg et al6Increased hemoglobin and unconjugated bilirubin values, decreased uric acid level Guder et al2Lymphocyte count elevated up to 4 years of age Weinberg et al6

Growing child 3-fold increase in alkaline phosphatase level due to bone growth Narayanan3; McComb et al4

Progressive changes Creatinine clearance decreases with each decade; age-related change in renal function; may Narayanan and Appleton7; Narayanan11

overlap with disease process in elderly persons Decrease in glucose tolerance with age; overlaps with secondary risk factors, such as genetics Pacini et al9and obesity

Homeostatic control by hypothalamic-pituitary-adrenal axis affected by aging; corticotropin and De Kloet10; Narayanan11

corticosteroid levels increase Dehydroepiandrosterone sulfate levels decrease, interleukin-6 levels increase with age Daynes et al12

Antidiuretic hormone and thyrotropin levels increase, triiodothyronine levels decrease with age Crawford et al14; Harman et al15

by guest on Decem


Dow

nloaded from


Clinical Chemistry / REVIEW ARTICLE

While changes in organ function occur due to aging,which in turn affects the circulating level of various analytes,one must also keep in perspective the overlapping effects dueto disease and dietary deficiencies of trace elements, such aszinc and other nutrients.11

Sex Differences

Between 15 and 55 years of age, the levels of total andlow-density lipoprotein cholesterol (LDL-C) increaseprogressively, with levels slightly higher in premenopausalfemales compared with the levels in males ❚Table 2❚. Thus,in female subjects, a change in cholesterol value from 174mg/dL (4.50 mmol/L) at age 25 years to 213 mg/dL (5.51mg/dL) at 55 years has been reported.2 In the corre-sponding age span for men, cholesterol values have beenreported to range from 166 mg/dL (4.29 mmol/L) at age 25years to 232 mg/dL (6.00 mmol/L) at 55 years.2 In contrast,the high-density lipoprotein cholesterol (HDL-C) leveldoes not fluctuate in either sex between the ages of 15 and55 years; levels are slightly higher in females, presumablyowing to the stimulating effect of estrogens.2

Analytes with levels that depend on muscle mass, suchas creatinine and creatine kinase (CK), are generally at ahigher concentration and activity, respectively, in malescompared with females.2,7 However, when increasedmuscle mass is acquired as a consequence of strenuousathletic activities in female subjects, the sex difference inlevels of creatinine and CK is abolished.2 Serum iron levels

generally are lower in premenopausal female subjectscompared with those for male subjects, with the differencedisappearing in subjects older than 65 years.2

Sex differences are seen for several analytes as can benoted by reviewing a current list of reference laboratoryvalues.16 The differences for some of the analytes based on sexcan be explained as due to differences in muscle mass and toendocrine and organ-specific differences.

Time

There are time-related fluctuations in the level of someanalytes (Table 2). Circadian rhythm is responsible for thediurnal changes seen in the circulating levels of someanalytes. A classic example of an analyte subject to diurnalvariation is cortisol, which generally peaks at around 6:00AM, with levels becoming lower toward the evening andmidnight.2 Because cortisol is a powerful immunosuppres-sant, its diurnal rhythm affects proinflammatory cytokineproduction. Indeed the level of proinflammatory cytokines,such as IL-1 alpha, IL-12, tumor necrosis factor alpha andinterferon gamma have been shown to be inversely corre-lated with the level of plasma cortisol.17 The inverse rela-tionship with plasma cortisol levels extends to the numberof circulating T lymphocytes, especially the CD4+ T cells,and to the urinary secretion of neopterin, which is a markerof cellular immune activity.18,19 Even routinely measuredanalytes such as glucose, potassium, and iron exhibitdiurnal variation.1,2


❚Table 2❚Effect of Sex, Time, Season, and Altitude on Laboratory Results

Variable/Effects References

Sex Between 15 and 55 years, progressive increase in total and LDL cholesterol levels; higher in Guder et al2premenopausal females, while HDL level does not fluctuate; HDL level slightly higher in females

Creatinine and CK are muscle mass dependent; generally higher in males Guder et al2Serum iron lower level in premenopausal females Guder et al2Sex differences reflected in separate reference intervals Kratz and Lewandrowski16

Time Cortisol peak, 6:00 AM; lower toward evening and midnight Guder et al2Proinflammatory cytokines (IL-1 alpha, IL-12, TNF-alpha, INF-gamma, CD4+ T cells) and urinary Petrovsky et al17; Miyawakineopterin levels opposite the cortisol rhythm et al18; Auzeby et al19

Glucose tolerance test values higher in afternoon Guder et al2LH, FSH, and testosterone released in short bursts Narayanan8

Potassium, iron, thyrotropin, and growth hormone subject to diurnal effects Young1; Guder et al2; Keffer20; Narayanan21

Season Summer Vitamin D levels higher Narayanan3

Triiodothyronine levels 20% lower Guder et al2; Harrop et al25

Winter Triglyceride level lower Narayanan3; Statland and Winkel24

Slight (2.5%) increase in total cholesterol level Cooper et al22; Narayanan23

Altitude At 1,400 m, hematocrit and hemoglobin values 8% higher Young1; Guder et al2At 3,600 m, 65% increase in C-reactive protein level Guder et al2As altitude increases, levels of renin, transferrin, and estriol and creatinine clearance decrease Young1; Guder et al2

CK, creatine kinase; FSH, follicle stimulating hormone; HDL, high-density lipoprotein; IL, interleukin; INF, interferon; LDL, low-density lipoprotein; LH, luteinizing hormone;TNF, tumor necrosis factor.

by guest on Decem


Dow

nloaded from



Glucose values obtained during an oral glucose toler-ance test tend to be higher when the test is performed in theafternoon than when the test is performed in the morning.The cortisol rhythm apparently may be responsible for thediscrepancy noted in the results of an oral glucose tolerancetest performed in the afternoon.2

Thyrotropin levels peak during the late evening hours,with the lowest values observed around midday.20,21 Growthhormone levels during waking hours are minimal, withincreases noted during sleep.1

Reproductive hormones, such as luteinizing hormone,follicle-stimulating hormone, and testosterone, are releasedin bursts lasting barely 2 minutes, making accurate assess-ment of their concentration problematic.8

Seasonal Changes

Vitamin D levels tend to be higher during the summer(Table 2), apparently due to prolonged exposure tosunlight.3 A slight increase in the total cholesterol level(average, 2.5%) has been observed during the wintercompared with values measured during the summer.22,23

The decrease in the level of triglycerides in serum fromsummer to winter is more striking than the extent of declinenoted between spring and autumn months when the temper-ature tends to be milder.3,24

The levels of the thyroid hormone, triiodothyronine, arereported to be 20% lower during the summer compared withduring the winter.2,25

Altitude

Changes in levels of some of the constituents in bloodoccur when measured at sea level as opposed to measure-ment at a higher altitude (Table 2). For example, hematocritand hemoglobin levels can be up to 8% higher at an altitudeof 1,400 m.1,2 A 65% increase in C-reactive protein has beenreported at 3,600 m. Concentrations of some analytes, suchas plasma renin, serum transferrin, urinary creatinine, andestriol, and the creatinine clearance rate decrease withincreasing altitude.1,2

Menstruation

At the onset of menstruation, a low estrogen level triggersthe release of follicle stimulating hormone from the pituitarygland. The ovaries thus are stimulated to produce estrogen,and the level begins to increase noticeably from the sixth orseventh day after menstruation; the peak level is reached onapproximately the 13th day. A day later, a burst of luteinizinghormone released from the pituitary gland signals ovulation.With the onset of ovulation, the progesterone level continuesto rise until it decreases together with the estrogen level, justbefore the commencement of the next menstrual cycle. Thus,the reference intervals for estradiol, follicle stimulating

hormone, luteinizing hormone, and progesterone are influ-enced by the stages of menstrual cycle (ie, follicular, midcycle,and luteal phases) ❚Table 3❚.16

Coincident with ovulation, serum cholesterol levels arelower than at any other phase of the menstrual cycle.2

During midcycle or the luteal phase, the aldosteroneconcentration is approximately 2-fold higher comparedwith values during the follicular phase.1,2 Renin activitymay increase during the luteal phase of the cycle.1 Incontrast, serum phosphate and iron levels decrease duringmenstruation.1,2

Pregnancy

A dilutional effect is observed due to an increase inthe mean plasma volume, which in turn causes hemodilu-tion (Table 3).2,3 The effect is more pronounced whenmeasuring trace constituents such as trace elements inserum.3 During pregnancy, there is a considerable increasein the glomerular filtration rate. Hence, the creatinineclearance rate is increased by 50% or more over thenormal rate.7 During the third trimester, the urine volumemay increase by approximately 25%.

During the second half of pregnancy, the placentaprogressively begins to produce hormones antagonistic toinsulin. As a result, the levels of hormones, such asestrogen, progesterone, and human placental lactogen (alsocalled chorionic somatomammotropin), increase, leadingto the onset of gestational diabetes.26

The increased metabolic demand during pregnancycauses an increased mobilization of lipids. As a result,especially during the second and third trimesters, theserum levels of apolipoproteins AI, AII, and B, triglyc-erides, and total cholesterol (especially LDL-C) areincreased substantially.22,23,27 The levels of these lipids inserum return to normal within 10 weeks after deliveryunless the mother breast-feeds her infant.28

Human chorionic gonadotropin elaborated by theplacenta has weak thyrotropic activity. Placental humanchorionic gonadotropin levels are at the highest concentra-tion toward the end of the first trimester of pregnancy, thussuppressing serum thyrotropin levels accompanied by aslight increase in the free thyroxine level.21,29 During thesecond and third trimesters of pregnancy, thyrotropinlevels are more reliable. Hence, trimester-based referenceintervals for thyrotropin are needed in pregnancy.

Associated with an increase in the glomerular filtrationrate in pregnancy is an increase in the clearance of iodideand a transfer of iodide and iodothyronines to the fetus. Asa result, the inorganic iodide level in serum decreases tosuch an extent that pregnant women with marginal iodineintakes of less than 50 µg/d develop iodine deficiency and,in turn, enlargement of the thyroid gland.21,29


by guest on Decem


Dow

nloaded from



The clearance by the liver during the first trimester ofpregnancy of the sialylated thyroxine-binding globulin,which is induced by estrogens, is decreased. Hence, the levelof thyroxine-binding globulin increases, leading to the well-known increases in serum thyroxine and triiodothyroninelevels during pregnancy.

Other analytes with increased levels during pregnancyinclude the heat-stable alkaline phosphatase elaborated bythe placenta, alpha-fetoprotein, copper binding proteins suchas ceruloplasmin, and, in turn, copper, acute phase proteins,and coagulation factor VII.2,8 The increased level of acutephase proteins increases the erythrocyte sedimentation rateapproximately 5-fold.2 Plasma concentrations of the fibrinol-ysis inhibitor, plasminogen activator inhibitor type 2, areincreased up to 2 µmol/L during the third trimester of preg-nancy.30,31 An increased metabolic requirement during preg-nancy leads to a deficiency of iron, and the depletion of ironstores is reflected by a decrease in ferritin.2

Lifestyle

Diet influences the results obtained for some analytes❚Table 4❚. Differences between vegetarians and nonvegetariansconsuming a high-protein or a high-purine diet are wellknown.2 Nonvegetarians tend to have increased blood levels ofuric acid, urea, and ammonia compared with those for their

vegetarian counterparts. Diets rich in saturated fatty acids ingeneral are lipogenic; the effect, however, varies depending onthe fatty acid. Thus, while stearic acid has little effect on LDL-C levels, palmitic acid substantially increases plasma choles-terol levels.23,27 Complex carbohydrates and monounsaturatedand polyunsaturated fatty acids, when substituted for saturatedfatty acids, tend, however, to lower LDL-C levels.23,27

A diet rich in fish oils lowers serum triglyceride and very-low-density lipoprotein (VLDL) levels, presumably because ofthe ability of fish oils to inhibit the synthesis of VLDL triglyc-erides.23,32 Even among vegetarians, there is a difference inserum lipid levels. Among strict vegetarians, plasma LDL-Clevels were reported to be 37% lower and HDL-C levels 12%lower compared with a nonvegetarian control population. Incontrast, among lactovegetarians, LDL-C levels were 24%higher, and HDL-C levels 7% higher compared with a groupof strict vegetarians.23,33

Consumption of caffeine influences the results of someanalytes (Table 4). Caffeine inhibits the enzyme phosphodi-esterase, which normally regulates the levels of cellularcyclic adenosine monophosphate (c-AMP) by converting itto the inactive product 5¢-AMP.2,3 The increase in c-AMPlevels due to the caffeine-induced effect promotes glycol-ysis coupled with energy production, and the subject feelsalert, awake, and energetic.


❚Table 3❚Effect of Menstruation and Pregnancy on Laboratory Results

Condition/Effects References

Menstruation Estradiol, FSH, LH, progesterone levels depend on stages of cycle Kratz and Lewandrowski16

Cholesterol level lowest at ovulation Guder et al2

Mid or luteal phase, aldosterone 2-fold higher than in follicular phase Young1; Guder et al2

Renin increased in luteal phase Young1

Decrease in serum iron and phosphate Young1; Guder et al2

Pregnancy Increase in mean plasma volume leads to hemodilution Guder et al2; Narayanan3

Creatinine clearance increased by 50% or more due to increase in glomerular filtration rate Narayanan and Appleton7

Increase in urine volume (25%) during third trimester Narayanan and Appleton7

Estrogens, progesterone, and human placental lactogen levels increase during second half of Narayanan26

pregnancy, all are antagonistic to insulin; gestational diabetes develops Increased mobilization of lipids during second and third trimesters (increase levels of total Cooper et al22,27; cholesterol, LDL-C, triglycerides, and apolipoproteins AI, AII, and B) Narayanan23

Thyrotropin levels suppressed owing to increased placental HCG at end of first trimester Narayanan21; Burrow et al29

Increased iodide clearance; with marginal iodide intake, can affect thyroid function Narayanan21; Burrow et al29

Estrogen effect increases levels of TBG, T3 and T4 Narayanan21

ESR increases 5-fold due to acute phase proteins Guder et al2

Increased levels of factor VII and PAI-2; decreased levels of iron and ferritin Guder et al2; Narayanan8; Sprengers and Kluft30; Narayanan and Hamaski31

ESR, erythrocyte sedimentation rate; FSH, follicle stimulating hormone; HCG, human chorionic gonadotropin; LDL-C, low-density lipoprotein cholesterol; LH, luteinizinghormone; PAI-2, plasminogen activator inhibitor type 2; T3, triiodothyronine; T4, thyroxine; TBG, thyroxine-binding globulin.

by guest on Decem


Dow

nloaded from



Activation of triglyceride lipase by c-AMP results in a3-fold increase in nonesterified or free fatty acid levels.2,3,24

The ability of free fatty acids to compete for binding siteson the albumin molecule can lead to the displacement of adrug or hormone bound to albumin, thus affecting themeasurement of free drug or hormone levels.8 The increasein free fatty acid levels results in a decrease in pH, which, inturn, can strip calcium from the binding protein. Conse-quently, a spurious increase in ionic calcium concentrationcan occur.8 Plasma renin activity and catecholamine levelshave been increased 3 hours after the consumption of 250mg of caffeine.2,34

The effect of caffeine on lipid levels has been contro-versial, and the results obtained seem to be influenced bycoffee-brewing methods.23 The finding in a double-blindstudy replacing caffeine-containing coffee with decaf-feinated coffee for a period of 6 weeks of an insignificantchange in serum triglyceride, total cholesterol, and HDL-Clevels suggests that caffeine is not the substance in coffeethat elevates the total cholesterol level.35

Consumption of ethanol can induce short- and long-term effects altering the results of several analytes (Table4).2 Short-term effects reflected rapidly within the first 2 to4 hours of consumption include a lowered blood glucoselevel and an increased plasma lactate level.2 Uric acid levels

increase owing to the metabolism of ethanol to acetaldehydeand further to acetate. The net effect of lactate formation isto cause metabolic acidosis owing to the consumption ofbicarbonate.2

Sustained consumption of ethanol can, by enzymeinduction, lead to the increase in the activity of liverenzymes, such as gamma-glutamyltransferase. There are,however, marked intraindividual and interindividual varia-tions in the level of serum gamma-glutamyltransferase, asdemonstrated in a study in which subjects were challengedfor 3 days with a 0.75-g of ethanol dose per kilogram ofbody weight.36 Other liver enzymes that become elevatedowing to the direct cytotoxic effect of ethanol on liver areaspartate aminotransferase (AST) and alanine aminotrans-ferase (ALT).2 Thus, consumption of 225 g of ethanol perday for 1 month caused skeletal muscle changes that led tothe release of liver cell enzymes into blood.37

Depending on the extent of consumption, the effect ofethanol on serum lipid levels is variable.23 The effect ofshort-term alcohol intake, especially by subjects whousually do not consume alcohol, is to increase plasmatriglyceride levels, especially the VLDL fraction.38

A moderate intake of ethanol (<40 g/d) is reported toincrease the plasma levels of HDL-C, HDL3-C, and theapolipoproteins AI and AII. However, if alcohol intake


❚Table 4❚Effect of Lifestyle on Laboratory Results

Lifestyle/Effects References

DietHigh-protein and high-purine diet increase levels of uric acid, urea, and ammonia in blood Guder et al2compared with vegetarians

Differences in lipogenicity of saturated fatty acids: stearic acid, less effect; palmitic acid, Narayanan23; Cooper et al27

substantial effect in raising LDL-C Complex carbohydrates, monounsaturated and polyunsaturated fatty acids lower the levels of LDL-C Narayanan23; Cooper et al27

Fish oils lower the levels of triglycerides and VLDL Narayanan23; Harris et al32

Lipid profile (LDL-C, HDL-C) differences for lactovegetarians vs strict vegetarians Narayanan23; Sacks et al33

Caffeine Inhibits phosphodiesterase; raises cAMP; activation of triglyceride lipase causes 3-fold Guder et al2; Narayanan3;increase in free fatty acids Statland and Winkel24

Decreased pH; increased ionic calcium level; free drug or free hormone stripped from albumin Narayanan8

Plasma renin activity and catecholamine levels increase Guder et al2; Robertson et al34

Ethanol Short- vs long-term effects Guder et al2Increased levels of GGT, AST, and ALT subject to interindividual and intraindividual variation Freer and Statland36,37

Moderate intake increases levels of HDL-C and apolipoproteins AI and AII Narayanan23; Freer and Statland37; Taskinin et al38

Smoking Long-term smoking increases carboxyhemoglobin, hemoglobin, RBC, WBC and MCV values; Narayanan3,8; Statland andWBC level related to number of packs smoked Winkel24

Cadmium levels higher Narayanan3,8

Reduction in HDL-C dependent on extent of smoking Narayanan23; Cooper et al27

Increased levels of plasma epinephrine, aldosterone, cortisol, free fatty acids, and free glycerol Young1; Guder et al2within 1 h of smoking

ACE activity reduced owing to destruction of lung endothelial cells Guder et al2; Haboubi et al39

ACE, angiotensin-converting enzyme; ALT, alanine aminotransferase; AST, aspartate aminotransferase; cAMP, cyclic adenosine monophosphate; GGT, gamma-glutamyltransferase; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; MCV, mean corpuscular volume; VLDL, very-low-densitylipoprotein.

by guest on Decem


Dow

nloaded from



exceeds 80 g/d, the synthesis of VLDL is stimulated,together with the activation of the enzyme lipoproteinlipase. Lipoprotein lipase hydrolyzes VLDL triglycerides,with the result that plasma VLDL levels will be apparentlynormal even when there is increased VLDL synthesis.23,27,38

Long-term smoking induces alteration in both biochem-ical and cellular profiles (Table 4).3,8 Carboxyhemoglobinlevels in blood are higher in long-term smokers since ciga-rette smoke contains carbon monoxide, which has a muchgreater affinity for hemoglobin than does oxygen. As aconsequence of long-term smoking, there is an increase inhemoglobin levels and the RBC and WBC counts. TheRBCs become larger, thus increasing the mean corpuscularvolume (MCV). Striking differences in the WBC count havebeen observed between healthy nonsmokers and healthylong-term smokers. Healthy smokers who smoked 1 pack ofcigarettes per day for 1 year had a WBC count of approxi-mately 1,000/µL (1 ·109/L) higher than that for healthynonsmokers. Healthy smokers who smoked 2 packs of ciga-rettes per day for 1 year had a WBC count of approximately2,000/µL (2 · 109/L) higher than that for healthynonsmokers.24

Cadmium levels in blood are higher in healthy long-term smokers compared with levels for healthy nonsmokers,thus requiring separate reference intervals to distinguish the2 populations.3,8

Smoking can reduce HDL-C levels in plasma; theextent of reduction apparently is related to the number ofcigarettes smoked per day.23,27 Short-term changes occur-ring within 1 hour of smoking 1 to 5 cigarettes include anincrease in the plasma levels of epinephrine, aldosterone,cortisol, free fatty acids, and free glycerol.1,2

Angiotensin-converting enzyme (ACE) activity isdecreased in smokers, presumably owing to the destructionof lung endothelial cells with the resultant decrease in therelease of ACE into the pulmonary circulation or, alterna-tively, to inhibition of enzyme.2,39

The extent of smoking (moderate or heavy) parallelsthe blood levels of thiocyanate and cotinine. Compared withnicotine, which is the parent compound, cotinine has alonger half-life, approximating 20 to 28 hours, making it anideal marker for the assessment of the extent of smoking.2

Specimen Collection Variables

The preparation of subjects for blood specimen collec-tion, such as the duration of overnight fast, time of specimencollection, and posture during blood sampling ❚Table 5❚; theeffects of the duration of tourniquet application, infusion,and exercise (Table 5); the effects of anticoagulants ❚Table 6❚

and stabilizing additives ❚Table 7❚ used for blood collection;the anticoagulant/blood ratio; specimen handling andprocessing; and the relative merits of anticoagulants and

other variables need to be delineated and standardized tominimize preanalytic error.

Duration of Fasting

Ideally, subjects should be instructed to fast overnightfor at least 12 hours before specimen collection (Table5).2,23 The reason for the 12-hour stipulation is based on thefact that an increase in the serum triglyceride level after afatty meal can persist for up to 9 hours, but there is littleeffect on the levels of total cholesterol or apolipoproteinsAI and AII.40

The effect of prolonged fasting, however, can affectsome laboratory results. Thus, during a 48-hour fast, thehepatic clearance of bilirubin could be approximately240% more than the nominal value.3,24 With prolongedfasting, there is a decrease in the level of specific proteins,such as the C3 component of complement, prealbumin, andalbumin. Normal levels of these proteins are restoredrapidly with protein supplementation.24

The effect of fasting, however, varies depending onbody mass.23 In a lean person, the use of fat as reflected byan increase in acetoacetic acid concentration in blood isminimal during a 1-day fast but increases rapidly withprolonged fasting.41 In an obese person, however, theacetoacetic acid level increases markedly during a 1-dayfast. In contrast, while lipolysis, as reflected by a 3-foldincrease in the serum glycerol concentration with 3 days offasting is observed in a lean person, little change is noted inan obese person.23,41

Time of Specimen Collection

The section “Time” referred to diurnal changes seen inthe circulating level of some analytes. It is, therefore,important that the time of specimen collection be keptconstant from day to day to rule out variations introducedby diurnal effects (Table 5).

Specimens for therapeutic drug monitoring (TDM)should be obtained after a steady therapeutic concentrationof drug or steady state is achieved in the blood. A bloodsample obtained just before administering a drug dose aftera steady-state level is reached will reflect the trough levelor the lowest concentration to be obtained with the estab-lished drug dose. If only 1 blood sample is to be collectedfor TDM, it is preferred that it reflect the trough level.42

Blood samples obtained when the drug concentrationis at its maximum, will, of course, reflect the peak level.Peak-level specimens need to be obtained within 1 to 2hours after intramuscular administration of the drug, 15 to30 minutes after intravenous administration, or 1 to 5hours after oral administration.42 Timing of specimencollection for TDM, however, depends on the rate of distri-bution of the specific drug. If the drug is infused intra-


by guest on Decem


Dow

nloaded from



venously, approximately 1 to 2 hours after the completionof infusion are needed for the distribution phase to becompleted, with exceptions such as digoxin and digitoxin,which need 6 to 8 hours for the completion of the distribu-tion phase.2

In general, since the rate of oral drug absorption differsfrom person to person, blood specimens obtained for TDMusually are timed to reflect trough levels. For the moni-toring of certain drugs, such as theophylline, antibiotics,and antiarrhythmic drugs, it may be necessary and useful tomeasure peak levels after intravenous administration.2,43

The time for collection of urine specimens depends onthe constituent being measured. For clearance measure-

ments or for the measurement of analytes that have diurnalvariation, a 24-hour urine sample is needed.7 A timedovernight urine sample or the so-called first-morning spec-imen, provides not only an assessment of the concentratingability of the kidney, but also sufficient time for formedelements or sediment to be produced for easy enumeration.

While a random or spot sample may be suitable forroutine screening, such a specimen may have its limitationsbecause it may be too diluted to detect accurately slightincreases in the concentration of analytes such as glucose,protein, cells, and casts. Thus, a spot or random samplemay be of limited value in the initial screening for microal-buminuria.44 However, in a strategy designed to detect and


❚Table 5❚Effects of Specimen Collection Variables on Laboratory Results

Variable/Effects References

Fasting 12-h overnight fast recommended, since increase in triglycerides persist up to 9 h after fatty meal Guder et al2; Narayanan23

After 48-h fast, 240% increase in bilirubin level; decreased levels of prealbumin, albumin, and C3 Narayanan3; Statland and Winkel24

Effects dependent on body mass Narayanan23; Young41

Time Standardize time to rule out diurnal effect Guder et al2Therapeutic drug monitoring trough vs peak levels; timing regimens Guder et al2; Narayanan42;

Pippenger43

Urine, timed overnight or first morning specimen concentrated, ideal for sediment enumeration Narayanan44

24-hour sample for clearance measurements and for analytes with diurnal variation Narayanan and Appleton7

Random or spot urine generally suitable Guder et al45

But, sensitivity to detect slight increases in numbers of cells and casts and glucose and protein Narayanan44

levels affected by dilution PostureIn general, 5%-15% increase for most cellular and large molecules (eg, albumin) in sitting posture Guder et al2; Narayanan8; Statland

and Winkel24

Increase in upright to supine has dilutional effect: 10% and 12% decreases in cholesterol and Narayanan8,23; Young41

triglyceride levels, respectively Change in posture has no effect on free drug and small molecules Guder et al2; Narayanan8;

Young41

Postural effect substantial in patients with reduced plasma volume (eg, congestive heart Narayanan8

failure, cirrhosis) Increased aldosterone, norepinephrine, epinephrine, renin, and atrial natriuretic peptide levels Guder et al2; Tan et al46

Tourniquet application time >1 minute increase in concentration of large molecules; for total cholesterol level, 5% increase Guder et al2; Narayanan8,23;at 2 minutes; up to 15% increase at 5 minutes Cooper et al22; Statland and

Winkel24; Young41

Decreased pH; increased potassium, lactate, ionized calcium, and magnesium levels with Narayanan8

increased application time Repeated fist clenching can increase potassium by 1-2 mmol/L, up to 2.7 mmol/L Don et al47; Skinner et al48

Infusion Spuriously increased blood glucose level if specimen obtained from arm receiving glucose infusion Guder et al2Waiting times before blood specimen collection: subjects receiving fat emulsion, 8 h; carbohydrate, Guder et al2protein, or electrolytes, 1 h Guder et al2

Extent of hemolysis related to age of transfused blood Guder et al2ExerciseStrenuous exercise increases total CK and CK-MB; CK-MB as percentage of total CK normal; Narayanan8; Statland anddecreased haptoglobin value due to intravascular hemolysis Winkel24

Individual variability in CK related to extent of exercise training Guder et al2Mitochondrial size and number increase with vigorous training; mitochondrial CK increases to >8% Guder et al2of total CK

Increased epinephrine, norepinephrine, glucagon, cortisol, corticotropin, and growth hormone levels; Guder et al2decreased insulin level

Increased serum and decreased urine uric acid levels due to increased lactate level Guder et al2Increased apolipoprotein AI and HDL-C levels; decreased LDL-C, apolipoprotein B, and triglyceride Narayanan23; Cooper et al27; levels with long-term strenuous exercise or brisk walking Hardman et al50

CK, creatine kinase; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol.

by guest on Decem


Dow

nloaded from



differentiate prerenal, glomerular, tubular, and postrenalcauses of proteinuria by measurement of specific proteins,a morning spot urine specimen was adequate.45

Posture During Blood Sampling

Changes in posture during blood sampling can affectthe concentration of several analytes measured in serum orplasma (Table 5). Change in posture from a supine to theerect or sitting position can result in a shift in body waterfrom the intravascular to the interstitial compartment. As a

result, the concentration of large molecules that are notfilterable is increased. Hence, albumin levels generally arehigher among healthy subjects attending an outpatient clinicfrom whom blood specimens are obtained with the subjectsin a sitting position compared with levels among healthyhospitalized subjects from whom blood specimens usuallyare obtained with the subjects in a supine position.8,24

Small molecules, such as glucose, however, owing totheir ability to move freely between the interstitial spaceand the circulation, are least affected by posture during


❚Table 6❚Anticoagulants for Blood Collection

Anticoagulant References

Heparin Unfractionated heparin molecular mass 3-30 kd (mean, 14 kd), as lithium, sodium, or ammonium Narayanan and Hamasaki31;salts; lithium heparin preferred for plasma chemistry testing; nominal amount, 14.3 U/mL of blood Novotny et al51; Narayanan52,54;

Guder et al53; Berg et al55; Doumas et al56

For ionic calcium, heparin in blood gas syringe can be calcium-titrated, electrolyte-balanced, or Narayanan52; Toffaletti et al57;zinc lithium–titrated Toffaletti58; Landt et al59

EDTA: salts of EDTA generally used for routine hematology testing (dipotassium or tripotassium Narayanan52; Goosens et al62;EDTA, disodium EDTA); nominal amount, EDTA, 1.5 mg/mL of blood Sacker63; Reardon et al64

Citrate: generally used for routine coagulation testing; trisodium citrate–dihydrate 3.2% (0.109 mol/L) Guder et al2; Narayanan andor 3.8% (0.129 mol/L) Hamasaki31; Narayanan65;

Adcock et al66

❚Table 7❚Stabilizing Additives

Additive References

Glycolytic inhibitors Sodium fluoride (2.5 mg/mL of blood with EDTA [1 mg/mL] or potassium oxalate [2 mg/mL]) Guder et al2; Chan et al67;

Sidebottom et al68

Lithium iodoacetate (0.5 mg/mL of blood) alone or in combination with lithium heparin Guder et al2; Chan et al67

(14.3 U/mL of blood) Sidebottom et al68

Sodium fluoride and lithium iodoacetate require 3 h to fully become effective Guder et al2Adding mannose (3 mg/mL of blood) to sodium fluoride achieves efficient glycolytic inhibition Liss and Bechtel71;

Nakashima et al72

Mannose, however, has concentration-dependent interference Nakashima et al72; Ho et al73

Maintaining blood pH at 5.3-5.9 with mixture of citric acid, trisodium citrate, disodium EDTA, and Uchida et al70

0.2 mg sodium fluoride stabilizes glucose Cell stabilizersAcid-citrate dextrose A and B formulations; B formulation has greater amounts of dextrose available Guder et al2to metabolizing cells

Citrate, theophylline, adenosine, dipyridamole mixture minimizes in vitro platelet Narayanan and Hamasaki31;activation Contant et al74; Narayanan75;

Van Den Besselaar et al76; Kuhne et al77

Proteolytic enzyme inhibitors EDTA (1.5 mg/mL of blood) with aprotinin (500-2,000 KIU/mL) stabilizes several polypeptide hormones Narayanan52,79

Aprotinin also can be used with lithium heparin (14.3 U/mL) Narayanan52

A mixture of EDTA, aprotinin, leupeptin, and pepstatin stabilizes parathyroid hormone–related protein Pandian et al80; Narayanan81

Catecholamine stabilizers An antioxidant such as glutathione, sodium metabisulfite, or ascorbic acid at a concentration of 1.5 Narayanan81

mg/mL is used with egtazic acid, EDTA, or heparin For urine, use 5 mL of a 6-mol/L concentration of hydrochloric acid per liter of urine or 250 mg Boomsma et al82

each of EDTA and sodium metabisulfite per liter of urine Fibrinolytic inhibitorsA mixture of thrombin and soybean trypsin or aprotinin Narayanan and Hamasaki31

Reptilase and aprotinin Connaghan et al83

Thrombin inhibitor: 5-mmol/L concentration of a phenylalanine, proline, arginine chloromethyl ketone Narayanan and Hamasaki31

mixture with 10-mmol/L concentration of citrate or 4.2-mmol/L concentration of EDTA Mohler et al84

KIU, kallikrein inhibitory unit.

by guest on Decem


Dow

nloaded from



blood specimen collection.8,41 In general, molecules evenas large as insulin can diffuse freely into and out of theinterstitial space and, thus, are not subject to variation dueto postural changes.8

While the free fraction of a metabolite, drug, hormone,or metal ion is not subject to postural variation, the fractionbound to protein, such as albumin, is affected by posture.Thus, bilirubin bound to albumin and calcium bound toalbumin are affected by postural changes.2,8

In blood specimens obtained with subjects in the sittingposition, generally an increase in the range of 5% to 15% isseen for most cellular and macromolecular analytescompared with results when blood specimens were obtainedwith subjects in the supine position.2

The effect of postural change from supine to erect isaccentuated in patients with a tendency to edema, such aspatients with cardiovascular insufficiency and cirrhosis of theliver.2 In such patients, reduction in plasma volume in turncauses increased secretion of hormones, such as aldosterone,norepinephrine, and epinephrine, and the enzyme, renin.2 Anincreased level of atrial natriuretic peptide (ANP) is measur-able in plasma as a compensatory mechanism to correct thedecrease in blood pressure.2,46 Indeed, changing from supineto erect can dramatically influence the concentration ofanalytes such as renin and aldosterone, even in healthypersons. Conversely, moving from upright to supine can havea dilutional effect owing to an increase in plasma volumeresulting from the transfer of fluid from the extravascular tothe intravascular compartment.8 Thus, a change from uprightsupine can reduce (after 5 minutes in the supine position) thecholesterol level by 10% and triglyceride levels by 12%.23,41

If blood specimens are to be obtained with the subject ina sitting position, such as in an outpatient setting, the sittingposture should be standardized by instructing the subject toremain seated for 15 minutes before specimen collection.

Duration of Tourniquet Application

The application of the tourniquet, by reducing the pres-sure below the systolic pressure, maintains the effectivefiltration pressure within the capillaries.2 As a result, smallmolecules and fluid are transferred from the intravasal spaceto the interstitium. Application of the tourniquet for longerthan 1 minute and up to 3 minutes can result in hemoconcen-tration, causing an increase in the concentration of largemolecules that are unable to penetrate the capillary wall(Table 5).2,8,24,41

The venous stasis that results from prolonged tourniquetapplication promotes anaerobic glycolysis with the accumu-lation of plasma lactate and a reduction in blood pH.8 Thehypoxic effect drives potassium out of the cell causing aspurious increase in the serum potassium level. This effectcan be accentuated if the phlebotomist requests the subject to

repeatedly clench the fist to permit visualization of the vein.The contraction of forearm muscles during repeated fistclenching causes release of potassium, since there is a reduc-tion in the intracellular electronegativity during the depolar-ization of muscle cells, which favors the release of potassiumrather than its uptake.47 Repeated fist clenching during phle-botomy after application of the tourniquet can cause a 1- to2-mEq/L (1- to 2-mmol/L) increase in potassium. In 1 study,an increase in the potassium level as much as 2.7 mEq/L (2.7mmol/L) was noted in a healthy subject due to fist clenchingduring phlebotomy.48 Application of the tourniquet for morethan 1 minute and up to 2 minutes can cause a 5% increasein the total cholesterol level, which can increase by 10% to15% if the tourniquet is left on for 5 minutes.22,23,41

In 1 study conducted on children, venostasis of 2 minutessubstantially decreased the blood pyruvate concentration byapproximately 18% yet had a negligible effect on the lactateconcentration, with a mean increase of just 2.2%.49

The reduction in blood pH during prolonged tourniquetapplication can alter drug-protein or hormone-proteinbinding, leading to an increase in the free drug or freehormone concentration. Thus, at a low pH caused by anaer-obic glycolysis during venous stasis, ionic calcium andionic magnesium levels increase since they are releasedfrom albumin, the binding protein.8

To minimize the preanalytic effects of tourniquet appli-cation time, the tourniquet should be released as soon asthe needle enters the vein.8 Avoidance of excessive fistclenching during phlebotomy and maintaining tourniquetapplication time to no more than 1 minute can minimizepreanalytic error.2

Effect of Infusion

For persons receiving an infusion, blood should notbe obtained proximal to the infusion site.2 Blood shouldbe obtained from the opposite arm. Very often, exception-ally high glucose values are obtained by collecting bloodfrom the arm in which an infusion of glucose is beingadministered (Table 5). Eight hours must elapse beforeblood is obtained from a subject who has received a fatemulsion. The waiting period for blood sampling frompersons receiving a carbohydrate-rich solution or aminoacid and protein hydrolysate or electrolytes is 1 hour aftercessation of the infusion.2 For subjects receiving bloodtransfusions, the extent of hemolysis and, with it,increased values for potassium, lactate dehydrogenase,and free hemoglobin are related progressively to the ageof the transfused blood.2

Effect of Exercise

Exercise, such as running up and down several flightsof stairs, or strenuous activity the night before specimen


by guest on Decem


Dow

nloaded from



collection, such as a workout in a gymnasium or marathonrunning, can affect the results obtained for several analytes(Table 5).

Strenuous exercise depletes the energy-yieldingcompound, adenosine triphosphate (ATP), from the musclecells. As a result, there is a change in cell membranepermeability, and enzymes are released from the cells.Thus, CK levels can be very high after strenuous exercise(levels approaching 1,000 U/L) with a release of the CK-MB fraction, although CK-MB as a percentage of total CKwould be normal.8

A high degree of individual variability is seen in ahypoxia-mediated increase in CK, which depends on theextent of exercise training.2 Mitochondrial size and numberare increased in persons who maintain a vigorous trainingschedule, allowing their muscles to oxidatively metabolizeglucose, fatty acids, and ketone bodies. As a result, even inthe absence of myocardial damage, mitochondrial CK-MBlevels increase to greater than 8% of total CK activity.2

Training increases muscle mass, thereby increasing theconcentration of creatinine in plasma and the urinary excre-tion of creatinine and creatine, while decreasing the level ofplasma lactate compared with untrained persons whoindulge randomly in exercise.

During strenuous exercise, there is some amount ofintravascular hemolysis. As a result, the level of plasmahaptoglobin, an acute phase reactant, is reduced owing toits “complexation” with free hemoglobin released duringhemolysis and the subsequent clearance of the complexfrom circulation.8,24

As a result of exercise, the concentrations of hormonessuch as epinephrine, norepinephrine, glucagon, cortisol,corticotropin, and growth hormone increase, while insulinlevels decrease, thereby causing an increase in glucoselevels because of the gluconeogenetic effect of hormonesantagonistic to insulin.2

Volume shift from the intravasal to the interstitialcompartment or volume depletion due to sweating duringexercise can increase serum albumin levels.2 Increasedlactate concentration due to anaerobic glycolysis duringexercise can decrease the urinary excretion of uric acid,thereby increasing its serum concentration.2 There are,however, beneficial effects of long-term strenuous exerciseor brisk walking in terms of reducing the levels of serumLDL-C, apolipoprotein B, and triglycerides and increasingthe levels of apolipoprotein AI and HDL-C.23,27,50

To minimize preanalytic variables introduced by exer-cise, subjects should be instructed to refrain from stren-uous activity at least the night before testing and to notexert themselves by walking a long distance or running orclimbing stairs before blood specimen collection.8

Anticoagulants for Blood Collection

Various salts of heparin, EDTA, and sodium citrate areused widely in the clinical laboratory (Table 6).

HeparinHeparin is the preferred anticoagulant for blood specimen

collection intended for determination of electrolyte levels andother routine chemistry values. The standard commercial-grade heparin preparation, also called unfractionated heparin,with a molecular mass ranging from 3 to 30 kd (mean, 14 kd)exerts its anticoagulant effect by binding to antithrombin III.This interaction causes a conformational change in theantithrombin III molecule, thereby accelerating the inhibitionof coagulation factors Xa, IXa, and thrombin.31 The standardcommercial grade heparin also can inhibit thrombin with theaid of a plasma cofactor, called heparin cofactor II. In vivo,heparin can cause the release of a potent coagulation inhibitor,called tissue factor pathway inhibitor, from the endothelialsurface.31,51

Various salts of heparin, such as lithium, sodium, andammonium, have been used as anticoagulants for the collec-tion of blood specimens intended for routine clinical chemistrydeterminations.52 The nominal amount of heparin used inblood specimen collection tubes is 14.3 U/mL of blood,although a range of 12 to 30 U/mL of blood has been deemedsatisfactory.53

Although all of the 3 salts of heparin give comparableresults for electrolyte determinations, lithium heparin has beenfavored over the others, more because of the perception thatsodium heparin might overestimate sodium levels and ammo-nium heparin might spuriously increase serum urea nitrogenconcentrations when measured by the urease procedure.52,54

Just how much sodium is present in a sodium heparinblood collection tube depends not only on the nominalamounts of heparin present (14.3 U/mL of blood), but also onother variables, such as the molecular weight of unfractionatedheparin, the number of dissolvable milliequivalents of sodiumper milligram of heparin, and the consistency of the heparinpreparation from lot to lot. Indeed, in 1 study when blood wasfilled to half of the tube’s nominal volume, thereby increasing2-fold the nominal amounts of sodium heparin present in theblood collection tube, plasma sodium values were notincreased artifactually.54

Obvious differences in results for certain analytesbetween serum and heparinized plasma are related to theconsumption of fibrinogen and the lysis of cellular elementsduring the process of clotting. Thus, potassium values arehigher in serum than in heparinized plasma, while in contrast,total protein values are higher in plasma than in serum. Theeffect of heparinized plasma on some of the enzyme measure-ments may be related, in part, to the bias introduced by theinstrument-reagent combination.55,56


by guest on Decem


Dow

nloaded from



Since the commonly used anticoagulant, such aslithium or sodium heparin, in the syringe to obtain speci-mens for blood gas measurement binds to ionic calcium,with the effect dependent on the concentration of heparin,attempts have been made to eliminate this bias. Oneapproach has been to use calcium-titrated heparin, in whichheparin in an amount less than 50 U/mL of blood is titratedto an ionized calcium concentration of 1.25 mmol/L.52 Theuse of electrolyte-balanced heparin that involves the use ofdry heparin in the amount of 40 U/mL of blood balancedwith calcium, sodium, and potassium decreases bias to lessthan 0.02 mmol/L of calcium in the analytic range of 0.9 to1.8 mmol/L.57 The use of dry heparin eliminates the dilu-tion effect associated with liquid heparin.

Other approaches to minimize interference withionized calcium measurements have involved saturating thehigh-affinity binding sites on heparin to which divalentcations, such as zinc, can bind. Zinc heparin, however, inthe presence of excess zinc ions is reported to cause a posi-tive interference in ionized calcium and total magnesiummeasurements.58 One successful approach is to saturatehigh- and low-affinity binding sites on heparin, the formerwith zinc and the latter with lithium.59 This calcium-neutralized lithium-zinc heparin has been satisfactory notonly for ionized calcium measurements, but also formeasurement of blood gases, total calcium, sodium, andpotassium.59

For TDM, the drug level in heparinized plasma wouldbe expected to be higher than that in serum, especially ifthe drug was bound to fibrinogen. In reality, the serumvalue for drugs such as amiodarone, desethylamiodarone,chloroquine, and desethylchloroquine was higher than theplasma value, apparently due to the release of cell-bounddrug during the clotting process.52,60,61 Thus, resultsobtained for amiodarone were 11.5% to 13.5% lower inheparinized plasma than in serum, while results fordesethylamiodarone were 4.4% to 8.5% lower in plasmathan in serum.60

The concentration of chloroquine was approximately2-fold higher in serum than in plasma, while the concentra-tion of its metabolite, desethylchloroquine, was approxi-mately 4 times higher in serum than in plasma. These 2drugs were present mainly in platelets and granulocytes.The increased levels of these 2 drugs noted in serum appar-ently were related to the release of these drugs fromplatelets during the coagulation process. Thus, for somedrugs, it is important to specify whether plasma or serumwas used for measurement.61

EDTAEDTA is the commonly used anticoagulant for routine

hematologic determinations. It functions as an anticoagu-

lant by chelating calcium ions, which are required for theclotting process.

Potassium and sodium salts of EDTA have been usedfor blood specimen collection. The solubility of potassiumsalts is better than sodium salts of EDTA. The pH of EDTAvaries depending on the salt. As more ions are added to thefree acid, the acidity of EDTA decreases. Thus, while a satu-rated solution of EDTA as free acid has a pH of 2.5 – 1.0, thetripotassium salt (K3EDTA) as 1% solution has a pH of 7.5 –1.0.52 The disodium (Na2EDTA) and dipotassium (K2EDTA)salts of EDTA in a 1% solution have a lower pH. Na2EDTAhas a pH of 5.0 – 1.0, while K2EDTA has a pH of 4.8 – 1.0.This difference in pH affects the size of the RBC. Althoughall salts of EDTA are hyperosmolar, causing the water toleave the cells and the cells to shrink, this shrinking effect isnoticed only with K3EDTA and not with K2EDTA orNa2EDTA.52 This is because although the cells shrink withK2EDTA and Na2EDTA, the lower pH of the anticoagulantcauses the cells to swell as well, thus balancing for theosmotically dependent shrinkage, and the cell size is notaltered. As a result, while the microhematocrit (centrifugedhematocrit) is lowered with K3EDTA, it is not affected byK2EDTA or Na2EDTA. The effect of dilution of the samplein automated analysis restores cells to their native shape. Atan optimum EDTA concentration of 1.5 mg/mL of blood, nodifferences in microhematocrit were noted between samplescollected in K2EDTA and K3EDTA and analyzed between 1and 4 hours after collection.62

In the same study, the MCV measurement and the calcu-lated hematocrit values were influenced not only by the instru-ment used, but also by the salt of EDTA (K2 or K3) used foranticoagulation of the blood specimen. Thus, MCV valueswere not affected with blood anticoagulated with K3EDTA,even at concentrations up to 10 times the nominal value. Incontrast, with 3 of the 4 instruments tested, increasing theconcentration of K2EDTA beyond the nominal value of 1.5mg/mL of blood resulted in an increase in MCV.62 One expla-nation may lie in the ready dissolution of liquid K3EDTAcompared with powdered K2EDTA, in which the effectivenessof mixing might become a variable, perhaps accounting for thediscrepancy in results.

The neutrophils are subjected to morphologic changesupon collection of blood in EDTA. These changes depend onthe time the peripheral blood smear is prepared and theconcentration of EDTA used for blood specimen collection.Minor changes begin to appear in the morphologic featuresof neutrophils even at an optimal EDTA concentration of 1.5mg/mL of blood. These changes include swelling of theneutrophils and loss of structure in the neutrophil lobes.Thereafter, there is a loss of granulation in the cytoplasm andthe appearance of vacuoles in the nucleus or cytoplasm ofthe cells.63 Between 1 and 3 hours after blood collection in


by guest on Decem


Dow

nloaded from



1.5 mg/mL of EDTA, the nucleus swells further, with acrossover of the nuclear chromatin. Beyond 3 hours, themorphologic features of the neutrophils become unclear asthe cell disintegrates. If a peripheral blood smear is madefrom blood anticoagulated with an EDTA concentration of2.5 mg/mL of blood, approximating to a conventional bloodcollection tube filled to one half of its nominal volume, themorphologic features of the neutrophils are disintegratedtotally.

Similar time-related and EDTA concentration–relatedchanges also occur in the mononuclear cells, although to alesser extent than in the neutrophils. Platelets also undergochanges, including swelling, giving rise to giant plateletsthat ultimately disintegrate, producing fragments in thesame size range as the platelets that are mistakenly countedas platelets by electronic counters, thus spuriouslyincreasing the platelet count.

Platelets rapidly change their shape from discoid to spher-ical when blood specimens are collected in EDTA, makingdeterminations of mean platelet volume unreliable. Thelimited stability of the WBC differential count (up to 6 hoursat room temperature storage) achieved in most of the 5-partdifferential analyzers is another drawback with EDTA.64

CitrateTraditionally, a 0.129-mol/L concentration or a 0.109-

mol/L concentration of trisodium citrate has been used as theanticoagulant for collection of blood specimens intended forglobal coagulation tests, such as prothrombin time (PT) andactivated partial thromboplastin time (aPTT).2,31 Citrate hasvirtually replaced sodium oxalate as an anticoagulant forperforming global coagulation tests, since factor V is morestable in citrate than oxalate. Furthermore, citrate rapidlycomplexes with calcium, forming a soluble complex incontrast with the slow formation of the insoluble complex ofcalcium with oxalate.31

The effective molarity of citrate depends on whether thedihydrate or the anhydrous salt has been chosen in the prepa-ration of the citrate solution. Nominally, a 3.2% solution ofthe dihydrate trisodium citrate salt corresponds to a 0.109-mol/L concentration, whereas a 3.8% solution of the samedihydrate salt corresponds to a 0.129-mol/L concentration.However, a 3.2% solution of the anhydrous trisodium citratesolution would correspond to a 0.124-mol/L concentration,whereas a 3.8% anhydrous sodium citrate solution wouldhave a molarity of 0.147 mol/L.65 Hence, it is important thatthe dihydrate salt be used to prepare either the 3.2% or 3.8%solution of trisodium citrate.

A laboratory that has been using one of the concentra-tions (3.2% or 3.8%) to perform PT determinations forpatients receiving oral anticoagulant therapy should notinterchange the formulations. Doing so will affect the inter-

national normalized ratio (INR) units that are used to reportthe results of the PT.66 INR values generally are higherwith a 0.129-mol/L concentration of sodium citrate thanwith a 0.109-mol/L concentration of sodium citrate, espe-cially when a responsive PT reagent, such as a recombinantthromboplastin that has a sensitivity similar to the WorldHealth Organization thromboplastin (with an internationalsensitivity index [ISI] equal to 1), is used.66 The differencesin INR between the 2 concentrations of citrate can varyfrom 0.7 to 2.7 INR units.66

The INR will be in error when a laboratory thatroutinely uses, for example, a 0.109-mol/L concentration ofcitrate, occasionally analyzes a specimen collected with a0.129-mol/L concentration of citrate, but uses the mean ofthe normal range obtained with the 0.109-mol/L concentra-tion to calculate the PT ratio (the patients PT divided bymean of normal range for PT), which together with the ISIare needed for the calculation of INR.66

Stabilizing Additives

Sodium fluoride and lithium iodoacetate have been usedalone or in combination with an anticoagulant such as potas-sium oxalate, EDTA, citrate, or lithium heparin for bloodcollection (Table 7).2 The concentration of sodium fluoridewhen used alone to inhibit glycolysis is approximately 4.3mg/mL of blood. In combination with an anticoagulant, suchas Na2EDTA or K2EDTA (1 mg/mL of blood) or potassiumoxalate (2 mg/mL of blood), the nominal concentration offluoride that is used to inhibit glycolysis is 2.5 mg/mL ofblood (60 mmol/L).2,67

Similarly, lithium iodoacetate has been used alone (0.5mg/mL of blood) or at that concentration in combinationwith lithium heparin (14.3 U/mL of blood).2 Even with theuse of either of these glycolytic inhibitors, at least 3 hoursare needed for these inhibitors to become fully effectiveand stabilize glucose levels.2,67 During this 3-hour period inhealthy subjects, an approximately 9 mg/dL loss of glucoseoccurs.2,67,68 However, in neonates or subjects with anincreased RBC, WBC, or platelet count, the decrease inglucose during the 3-hour period, even in presence ofglycolytic inhibitors, is greater than in healthy subjects.67,69

Once the glycolytic inhibitors become fully effective afterthe initial 3-hour period, glucose levels remain stable for atleast 3 days.2,67,68

In the absence of glycolytic inhibitors, a decrease in theglucose level of as much as 24% can occur in 1 hour afterblood collection in neonates in contrast with a 5% decreasein healthy adults when specimens are stored at roomtemperature. By 5 hours after blood collection in specimensstored at room temperature, as much as 68% of glucose canbe consumed in specimens collected without glycolyticinhibitors from neonates with a high hematocrit.69


by guest on Decem


Dow

nloaded from



The reason that fluoride and iodoacetate take approxi-mately 3 hours to stabilize glucose levels is that they do notact on the first enzyme in the glycolytic pathway, which ishexokinase. Instead, fluoride acts on enolase, which is theeighth enzyme in the glycolytic pathway, while iodoacetateacts on glyceraldehyde-3-phosphate dehydrogenase, whichis the fifth enzyme in the pathway in which glucose ismetabolized.2

Potassium oxalate, which is used in combination withfluoride, inhibits pyruvate kinase in the enzymatic stepimmediately after enolase in the glycolytic pathway and,thus, can effectively retard lactate formation 1 hour afterblood collection.

The drawback with both fluoride and iodoacetate isthat they promote hemolysis, which increases progressivelyas the concentration of the inhibitor used and the timeelapsed after blood specimen collection increases. Appar-ently both inhibitors deplete the organic phosphate (ATP)content of the RBC, thus causing an efflux of potassium outof the cell.

A variety of approaches have been attempted toimprove the efficacy of glycolytic inhibition. In oneapproach, a mixture of citric acid, trisodium citrate,Na2EDTA, and a very small concentration of sodium fluo-ride (0.2 mg) effectively inhibits the enzymes in theglycolytic pathway by maintaining blood pH in the rangeof 5.3 to 5.9.70 With this additive mixture, glycolysis wasinhibited for 8 hours in specimens stored at room tempera-ture, while at 24 hours, a mean – SD decrease of 1.3 mg –1.1 mg/dL (0.07 – 0.06 mmol/L) of glucose was noted.70

Another strategy for achieving efficient glycolytic inhi-bition is to include mannose in addition to fluoride.71 Therationale is that mannose, which is an epimer of glucose(differs structurally from glucose in the configuration aroundjust 1 carbon atom), will compete with glucose for hexoki-nase, the first enzyme in the glycolytic pathway. Sincemannose is a competitive inhibitor, the inhibition is short-lived, lasting just 4 hours. However, by this time, fluoride,which is also in the additive mixture, would have becomeeffective thus achieving a better glucose stabilization.2,71

At a mannose concentration of 3 mg/mL of blood, nointerference was noted in hexokinase and glucose oxidaseprocedures for the measurement of glucose.72 However,when the mannose concentration was increased to 3 timesthe nominal concentration of 3 mg/mL of blood, an overes-timation of glucose by 18 mg/dL (1.0 mmol/L) wasobserved.72 Some of the mannose preparations also may becontaminated with glucose. Negative interference has beenreported in the hexokinase procedure when mannoseconcentrations exceed 3 times the nominal concentration.73

The extent of reduction in glucose levels depends on theconcentration of hexokinase used in the glucose assay, with

less interference noted at higher concentrations of hexoki-nase used in the assay.73

In practical terms, if plasma is separated from cells bycentrifugation promptly after blood specimen collection, theglucose level in plasma is expected to be stable, since theenzymes that metabolize glucose are in the cellular fraction.

Stabilization of RBCs by including a nutrient such asglucose together with an anticoagulant mixture is the basisof 2 acid citrate dextrose (ACD) formulations.2 Theseformulations, labeled ACD A and B, are similar in pH;ACD A has a pH of 5.05, and ACD B has a pH of 5.1. Thedifference between the formulations is the additive/bloodratio; the A formulation has a ratio of 1:5.67. The B formu-lation, however, with a ratio of 1:3, has relatively greateramounts of dextrose available for the metabolizing RBCsand, hence, can preserve them better. While ACD formula-tions can preserve RBCs for 21 days when blood is storedbetween 1 C and 6 C, their stability can be extendedfurther by including additives such as phosphate andadenine (citrate phosphate dextrose and citrate phosphatedextrose adenine, respectively).

The maintenance of increased levels of cyclic adenosinemonophosphate (c-AMP) within the platelets is the basis ofan additive mixture intended for blood specimen collectionto minimize in vitro platelet activation.74,75 The additive is amixture of citric acid, theophylline, adenosine, and dipyri-damole with the final pH adjusted to 5.0.74 Adenosine acti-vates the enzyme adenylate cyclase, thus increasing c-AMPlevels within the platelets. Theophylline and, to a certainextent, dipyridamole prevent the degradation of c-AMP byinhibiting the enzyme phosphodiesterase. Dipyridamole, inaddition, prevents the uptake of adenosine by the RBCs, thuseffectively increasing the amount of adenosine available tothe platelets to activate adenylate cyclase and, in turn, theinhibition of platelet aggregation and release of contents ofplatelet alpha granules.31,74 The citrate in the additivemixture functions as an anticoagulant by chelating calcium.The citric acid, theophylline, adenosine, and dipyridamolemixture is used for monitoring heparin therapy by aPTT orthe chromogenic substrate assay and in measurement ofplatelet activation markers, such as P-selectin (CD62), byflow cytometry.76,77

Stabilizing additives to inhibit proteolytic enzymes areneeded for some analytes. Although EDTA itself is a metal-loprotease inhibitor, sometimes EDTA alone is insufficient toinhibit proteolytic enzymes. For example, when a syntheticprotease inhibitor, such as nafamostat mesylate, is added toEDTA, the mixture can substantially improve the stability ofcomplement components C3a, C4a, and C5a.52,78

A proteinase inhibitor, aprotinin (Trasylol), when usedin combination with an anticoagulant such as EDTA orheparin, has been useful for the stabilization of labile


by guest on Decem


Dow

nloaded from



polypeptide hormones and enzymes.52 Aprotinin inhibitsspecific proteases of kinin and coagulation and fibrinolyticenzyme systems, in addition to inhibiting the enzymesystems of leukocytes and damaged tissue cells. Thus, itinhibits kallikrein, trypsin, chymotrypsin, coagulation factorsin the preliminary phase of blood clotting, the fibrinolyticenzyme plasmin, and lysosomal proteinases. The fact thataprotinin inhibits kallikrein is the basis for the expression ofits potency in terms of kallikrein inhibitory units (KIUs).Aprotinin has been used in combination with an anticoagu-lant such as EDTA or heparin. A mixture of EDTA (1.5mg/mL of blood) and aprotinin in the activity range of 500 to2,000 KIU/mL has been used to stabilize polypeptidehormones such as glucagon and corticotropin, the enzymerenin, and gastrointestinal hormones, such as beta-endorphin,secretin, neurotensin, gut glucagon, somatostatin, vasoactiveintestinal peptide, and peptide yy.52,79

In the absence of aprotinin, spuriously higher valuesmay be obtained apparently because of the fragments of theanalyte produced by proteolytic enzyme degradation aremeasured as intact molecules by the polyclonal antibodyused in the assay. Thus, in a radioimmunoassay for glucagon,a 26% spurious increase was noted in plasma obtained withblood specimens collected with EDTA alone compared withplasma obtained with blood specimens collected in a mixtureof EDTA and aprotinin.52,79

A mixture of lithium heparin (14.3 U/mL) and 1,000KIU/mL of aprotinin also has been used to stabilizeimmunoreactive somatostatin, secretin, cholecystokinin(pancreozymin), glucagon, and C-peptide.52

In addition to aprotinin and an anticoagulant such asEDTA, additional proteolytic enzyme inhibitors may beneeded to stabilize a labile tumor marker, such as parathy-roid hormone related protein (PTH-rP) that has been associ-ated with the humoral hypercalcemia of malignantneoplasm.80,81 Thus, with an additive mixture of EDTA (1.5mg/mL), aprotinin (500 KIU/mL), leupeptin (2.5 mg), andpepstatin (2.5 mg), PTH-rP levels were stable for 1 hourafter blood specimen collection for specimens stored atroom temperature and for 24 hours, if specimens wererefrigerated at 4 C.80 Even when blood is collected with thisadditive mixture, if the sample is stored at room temperaturefor 24 hours, as much as 60% of the PTH-rP activity can belost.80 Without a stabilizing additive mixture, PTH-rP is sounstable that as much as 60% of its activity can be lostwithin 1 hour of specimen collection and storage at roomtemperature.

Analytes that are susceptible to oxidative damage, suchas catecholamines, require stabilization. Typically togetherwith an anticoagulant such as heparin, EDTA, or egtazicacid, an antioxidant such as glutathione, sodium metabisul-fite, or ascorbic acid at a concentration of 1.5 mg/mL should

be included in the blood specimen collection additivemixture.81 The specimen collected with this additivemixture should be centrifuged within 1 hour of collection. Itis preferable, however, if possible to separate plasma within15 minutes of specimen collection and store it in plastictubes or vials at –20 C in case analysis is delayed. Promptprocessing of plasma obtained with an antioxidant-anticoag-ulant mixture stabilizes catecholamines in plasma for 1 dayat room temperature, for 2 days when stored at 4 C to 8 C,and for 1 month at –20 C. Stability at –20 C can beextended to 6 months with added glutathione.2,82

Apparently, at the higher concentrations of cate-cholamines present in urine compared with plasma, thehormones and their metabolites are stable for 4 days, evenwhen stored at 20 C to 25 C without preservatives.2,82

However, for long-term stabilization of catecholamines inurine specimens, collection is recommended in a containerwith 5 mL of a 6-mol/L concentration of hydrochloric acidper liter of urine or in a mixture of 250 mg each of EDTAand sodium metabisulfite per liter of urine, which will extendstability to 3 weeks at 20 C to 25 C, and 4 months to 1 yearat 4 C to 8 C.2,82

The laboratory assessment of fibrinolysis by themeasurement of the nonspecific fibrinogen-fibrin degrada-tion products requires inhibition of the in vitro generation ofthe fibrinolytic enzyme plasmin with an inhibitor, such assoybean trypsin or aprotinin, and the conversion of residualfibrinogen to fibrin with thrombin.31 However, in patientsreceiving heparin therapy, the conversion of residualfibrinogen to fibrin, even in presence of thrombin, will beslow, thus resulting in spuriously high fibrinogen degradationproduct values owing to the remaining unconvertedfibrinogen. In such cases, the addition of snake venom, alsoknown as reptilase, instead of thrombin to the additivemixture, which also contains a plasmin inhibitor such asaprotinin, can result in the conversion of residual fibrinogento fibrin, even in the presence of heparin.83

For monitoring fibrinolytic therapy, the additive used toinhibit plasminogen activation depends on the therapeuticagent administered. Thus, while aprotinin together withEDTA is effective for inhibiting plasminogen activation bythe therapeutic agent urokinase, it is ineffective for inhibitingthe recombinant tissue plasminogen activator (rt-PA) fromactivating plasminogen.75 Specific synthetic peptides of argi-nine chloromethyl ketone (PPACK) are needed to inhibit invitro rt-PA activity. PPACK irreversibly inactivates rt-PA byalkylating the active center amino acid histidine, in additionto functioning as a potent thrombin inhibitor.84 A 5-mmol/Lconcentration of PPACK combined with a 10-mmol/Lconcentration of citrate or a 4.2-mmol/L concentration ofEDTA effectively inhibited rt-PA upon blood collection.75

However, since PPACK is unstable at neutral pH, blood upon


by guest on Decem


Dow

nloaded from



collection should be maintained at 4 C, centrifuged promptlyto obtain plasma, and processed without delay or kept frozenuntil analysis can be performed.31

As new diagnostic markers are proposed, the labilenature of some poses challenges. A case in point is themeasurement of ANP, which is a marker of chronic conges-tive heart failure. An additive mixture of EDTA and apro-tinin is needed to stabilize ANP; the blood specimencollected with this mixture must be centrifuged immedi-ately at 4 C and plasma maintained frozen at –20 C orbelow until ready for analysis.85 Furthermore, hemolysisinvalidates ANP measurements. Alternatively, the measure-ment of the N-terminal fragment of the prohormone pro-atrial natriuretic peptide (proANP) (which is formed whenproANP is split into equimolar amounts of ANP and the N-terminal fragment) can be performed in blood specimenscollected in EDTA alone with storage at room temperaturefor up to 6 hours. At 4 C, N-terminal proANP is stable for3 days and at –20 C for 4 weeks, with hemolysis notmarkedly affecting the results.86

Anticoagulant/Blood Ratio

The anticoagulant/blood ratio is critical for some labora-tory tests. Obviously, if too little blood is obtained, osmoticeffects can be expected to result in cell shrinkage and, thus, tointroduce dilutional changes in the concentration of extracel-lular analytes.

A well-known example of the critical effect of the antico-agulant/blood ratio is for the global coagulation test, aPTT.Traditionally, a 1:9 ratio of anticoagulant (sodium citrate, a0.109-mol/L concentration or a 0.129-mol/L concentration) toblood is used. However, if less blood is collected, simulating a1:7 anticoagulant/blood ratio, there is a significant increase inthe aPTT result compared with that obtained using thenominal 1:9 ratio.31 In 1 study, a 2.4-second increase wasobserved at a 1:7 ratio compared with the aPTT resultobtained at the nominal 1:9 ratio.65

However, for the global coagulation test, PT, the effect ofthe anticoagulant/blood ratio becomes meaningful only whenthe ratio reaches 1:4.5, which would occur when the bloodcollection tube is filled to just less than half of its nominalvolume.65 The anticoagulant/blood ratio also is influenced bythe sensitivity of the thromboplastin reagent that is used. In astudy of the effect of the anticoagulant/blood ratio using 3.2%(0.109-mol/L concentration) buffered citrate, the results for thePT were valid even for specimen collection tubes that werefilled to 65% of the nominal anticoagulant/blood ratio (1:9).However, filling the tube to at least 80% of the nominal ratiowas needed to obtain reliable results in the therapeutic rangeusing a moderately sensitive thromboplastin reagent with anISI of 2.06. Using a highly sensitive thromboplastin reagentwith an ISI of 1.01 required maintaining the

anticoagulant/blood ratio to at least 90% of the nominalratio.87 In the aforementioned study,87 the need to fill bloodcollection tubes to the nominal capacity (1:9) was reinforcedfor the aPTT, since tubes filled to less than 90% of the nominalcapacity caused prolongation of the aPTT result.

In addition to maintaining the nominal anti-coagulant/blood ratio (1:9), the amount of head space abovethe blood also has been reported to be a variable affecting theaPTT result. Thus, collecting less blood (2.7 mL as opposed tothe nominal 4.5 mL) but maintaining the anticoagulant/bloodratio without decreasing the tube size effectively increases thehead space above the blood. This, in turn, generally leads to ashortened aPTT in values outside the upper limit of the refer-ence interval and in the therapeutic range for monitoringunfractionated heparin therapy.88 Thus, in the abnormal range,the mean value obtained with the partially filled tube was 11seconds shorter than the mean value obtained with the fulltube.88 One possible explanation for the head space effect isthe increase in surface area relative to the volume of bloodcollected, favoring platelet activation and the subsequentrelease of platelet factor 4, which in turn neutralizes some ofthe heparin, resulting in a shortened aPTT.88

For polycythemic patients with hematocrit values above60% (0.60), the PT and the aPTT can be prolonged substan-tially, even when the nominal 1:9 anticoagulant/blood ratio ismaintained.89 This is because the plasma is overcitrated to theextent that it complexes much of the calcium chloride added toperform the PT and aPTT, effectively reducing the calciumions needed to clot the plasma and thereby prolonging theclotting time. This effect can be minimized by adjusting thecitrate concentration in accordance with the hematocrit valueby using empiric formulas or by using a 1:19anticoagulant/blood ratio.65,89 The latter ratio dilutes the citrateconcentration to a point that results of PT and aPTT are notaffected across a spectrum of hematocrit values ranging fromanemia to polycythemia.89

In general, collecting blood specimens to less than thenominal volume increases the effective molarity of the antico-agulant and induces osmotic changes affecting cell morpho-logic features as noted for the morphologic features ofneutrophils when the EDTA concentration changes from thenominal 1.5 mg/mL of blood to 2.5 mg/mL of blood. Further-more, the binding of analytes, such as ionic calcium or ionicmagnesium, to heparin can be enhanced when the effectiveconcentration of unfractionated heparin increases beyond thenominal 14.3 U/mL of blood.52

Maintenance of the anticoagulant/blood ratio also iscritical when the inhibitory effects of the anticoagulant areconcentration-dependent. Thus, a 10% reduction in thegentamicin level was noted in an enzyme-multipliedimmunoassay technique when the concentration of heparinused for specimen collection increased to 25 U/mL of


by guest on Decem


Dow

nloaded from



blood, presumably because of the inhibition of the enzymesused in the immunoassay procedure, as the concentration ofheparin increased from the nominal value.90

At the other extreme is exceeding the anticoagulant/bloodratio to a point that the tube is overfilled with blood to such anextent that proper mixing of anticoagulant with blood becomesproblematic. Thus, in 1 study, the hemoglobin values haddoubled compared with a value obtained a week before, andthe WBC and platelet counts were reduced drastically. Reana-lyzing the specimen for the fourth time from the same tubeyielded values similar to values obtained initially the weekbefore. Apparently, proper mixing of the specimen on arocking mixer was not achieved since there was no head spaceor bubble left in the tube to assist in the mixing of the spec-imen. After a sufficient amount of blood was withdrawn byrepeating the analysis 4 times, there was sufficient space in thetube for the air bubble to move and effect thorough mixing onthe rocking-type mixing device.91

For bacteriologic determinations, inhibition of the normalbactericidal properties of blood requires appropriate dilutionof blood in broth. Incorporation of a 0.025% solution ofsodium polyanethole sulfonate in blood culture media isintended to inhibit phagocytosis, complement, and lysozyme.Even when blood culture is supplemented with sodiumpolyanethole sulfonate, at least a 1:10 blood/broth ratio isconsidered optimal.92 However, a blood/broth ratio less than1:10 has been acceptable when at least 2 separate bloodcultures are obtained routinely or when a biphasic medium isused to culture blood (a broth bottle to which an agar slantwith media is attached).93

Since spurious blood culture results can be obtainedfor patients receiving antibiotics, a 1:10 dilution of blood inbroth dilutes many of the antibiotics in the blood to sub-inhibitory levels. The incorporation of sodium polyanetholesulfonate inactivates aminoglycosides. Alternatively,however, devices containing antibiotic-adsorbent resins canbe used to neutralize antibiotics, and cell lysis followed byfiltration or centrifugation neutralizes the antimicrobialproperties or components of blood and provides a concen-trate for culture on antibiotic-free media.93

Specimen Handling and Processing

The time and temperature for storage of the specimenand the processing steps in the preparation of serum orplasma or cell separation using density-gradient techniquescan introduce a preanalytic variable. Analytes subject tocellular metabolism need prompt separation from cells.

Storage of anticoagulated whole blood or clotted bloodin a refrigerator (4 C) inhibits the Na+,K+-ATPase pump,resulting in potassium leaking out of the cells and causing aspurious increase in levels measured in serum or plasma.This effect is seen even at 4 hours of storage of clotted blood

at 4 C.2,8,94 Inhibition of the Na+,K+-ATPase pump alsodrives sodium into the cell. Since sodium is mainly extracel-lular, in contrast with potassium that is mainly intracellular,the decrease in sodium becomes noticeable only when wholeor clotted blood is stored beyond 24 hours at 4 C. Ideally,when separation of plasma or serum from cells is needed,blood should be processed within 1 hour of collection toobtain plasma or serum.2

Storage of clotted blood beyond 8 hours at roomtemperature (23 C) can cause an increase in inorganic phos-phate owing to the hydrolysis of cellular organic phosphateby the enzyme alkaline phosphatase; the effect is morepronounced when clotted blood is stored at 30 C.2 Thus,while a 10% increase was noted in the inorganic phosphatevalue at 24 hours compared with the 8-hour value duringstorage at 23 C, the increase was 120% at a storage tempera-ture of 30 C.2

A time-dependent increase in ammonia is accentuatedin specimens with increased gamma-glutamyltransferaseactivity.95 The time between collection of blood and itsstorage temperature before processing to yield plasma canintroduce a variable. Thus, in the measurement of cate-cholamine levels in specimens collected with a stabilizingadditive mixture and stored at 4 C, there is an increase inthe level of catecholamines owing to a delay in theprocessing of plasma that exceeds 1 hour.81,82 A delay inthe preparation of plasma for 2 hours after blood collectionand storage at 4 C can increase the mean plasma norepi-nephrine level by 13% (N = 8).82 This effect is due to thelysis of some of the RBCs at 4 C, leading to a release ofcatecholamines into plasma and to the slow reuptake ofthese hormones by cells at 4 C. In addition, cryoprecipita-tion of plasma proteins at 4 C also can contribute to theincreased level of catecholamines measured in plasma.81,82

In contrast, storage of specimens at 20 C and a delaybeyond 1 hour in the processing of plasma after specimencollection can lower the catecholamine level measured inplasma owing to the uptake of catecholamines by cells suchas RBCs or platelets.82 Thus, the mean plasma norepineph-rine levels decreased by 14% (N = 8) from the initial valueif blood specimens kept at 20 C were centrifuged 2 hoursafter collection, while mean plasma epinephrine valuesdecreased by 20%.82

The storage temperature of a specimen influences theresults of global coagulation tests such as PT and aPTT.Since factor VII, which is measured by PT, is activatedduring prolonged storage at 4 C, storage of plasma incontact with cells beyond 7 hours shortens the PT.65,96,97

Indeed, when a blood collection tube is kept well stop-pered and stored at room temperature (25 C), the PT isstable for up to 48 hours.89,97 As is well known, the aPTTshould be performed preferably within 2 to 4 hours of


by guest on Decem


Dow

nloaded from



specimen collection, and some investigators maintainplasma on ice (4 C).2 In 1 study, a 10% to 15% increase inaPTT was observed after 24 hours in specimens that weremaintained well stoppered and at room temperature.89

In contrast with refrigerated temperatures, freezingplasma at –20 C and –70 C did not activate factor VII. ThePT and aPTT results were stable in plasma maintainedfrozen at –20 C for 10 days and at –70 C for 21 days.97

Factors II, V, VII, and X are stable in plasma maintainedunder refrigeration for up to 6 hours. All except factor V hadstability for up to 14 days when frozen at –20 C or –70 C.97

During a 7-day period, 20% of factor V activity was lost inplasma stored at –20 C, while in samples stored at –70 C forthe same period, 10% of the activity was lost.

Factor VIII is the most labile factor; 10% of the activity islost in 4 hours during storage at 4 C, while at –20 C for a 3 dayperiod, 20% of the activity was lost.97 Typically, clotted bloodor anticoagulated blood is centrifuged at 1,000g to 1,200g for10 to 15 minutes. Insufficient centrifugation time and lowercentrifugal forces can result in a gradient of platelets in plasmaderived from heparinized blood, thus affecting some of thelaboratory results, such as potassium levels.2

While centrifugation at 2,000g for 15 minutes isrecommended for whole blood to obtain platelet-poorplasma for global coagulation tests, at slightly lower speeds(1,800g), some residual platelet contamination wasreported based on a shortened thrombin clotting time by10% in plasma frozen for 48 hours and thawed to roomtemperature before testing.97

Despite the use of robotics and improved pneumatictube delivery systems, and despite the fact that erstwhilepreanalytic problems such as hemolysis during specimentransportation may have been overcome by improved designof transport systems, the existence of preanalytic problemsshould not be discounted.98,99 A case in point is the likeli-hood of erroneous results for PO2 if the blood has been cont-aminated with air before pneumatic tube transportation.99

Relative Merits of Anticoagulants

Platelets rapidly change their shape from discoid to spher-ical when blood is collected in EDTA, thus making the deter-mination of mean platelet volume unreliable. Approaches tominimize this artifact have been attempted by the use of analternative anticoagulant, which is a mixture of trisodiumcitrate, pyridoxal phosphate and tris(hydroxymethyl)-aminomethane.100 The rationale is that pyridoxal phosphateis an effective platelet antiaggregant and disaggregant. Meanplatelet volume measurement using the aforementioned addi-tive mixture was stable at room temperature for 24 hours.100

ACD, heparin, and EDTA have been used for the sepa-ration of mononuclear cells. However, regardless of whetherACD, EDTA, or heparin was used as the anticoagulant,

the lymphocyte fraction was contaminated with granulo-cytes if the blood specimen was more than 24 hours old.52

RBC contamination is a problem with aged EDTA bloodspecimens to the extent that 9 RBCs may be encounteredfor every lymphocyte separated from a 2-day-old bloodspecimen.101

For flow cytometric analysis, ACD and heparin aresuperior to EDTA for maintaining viable WBCsovernight.102 However, if analysis is to be performed thesame day, EDTA, ACD, and heparin give equivalent results.However, at 24 hours and beyond, a substantial decrease ingranulocyte viability may occur in blood collected in EDTA.K3EDTA also has been reported to be associated with a lossof function in lymphocyte-mitogen stimulation assays.102

Heparin has been reported to be problematic for use inmolecular biology procedures using restriction enzymes andDNA amplification by the polymerase chain reaction(PCR).103 While heparin inhibition can be overcome by avariety of approaches (eg, treatment with heparinase, orseparation of leukocytes by centrifugation followed by aminimum of 2 washings in a saline buffer or use of an appro-priate buffer), many investigators prefer not to use heparinfor DNA amplification by PCR.103

In vitro platelet, monocyte, and neutrophil activationstudies are influenced by the anticoagulant used for bloodcollection.104 Thus, monocyte activation as measured byrelease of tissue factor and tumor necrosis factor activity waslowest with EDTA compared with citrate, heparin, andhirudin. Neutrophil activation as measured by lipopolysac-charide-induced release of lactoferrin also was the lowestwith EDTA. EDTA seems to suppress platelet degranulation.Platelet factor 4 levels as an indicator of platelet activationwas the lowest with EDTA–platelet-poor plasma (217ng/mL) compared with platelet-poor plasma obtained withcitrate (440 ng/mL), hirudin (469 ng/mL), and unfractionatedheparin (1,180 ng/mL). A low-molecular-weight heparinpreparation, however, gave platelet factor 4 values compa-rable to those obtained with hirudin.104

Endogenous Interferences and RelatedVariables

The effect of drugs and their metabolites on laboratorytests is so extensive that compendia of interferences areavailable for consultation.105 Several are discussed in thefollowing text ❚Table 8❚.

Although the subject of endogenous interferences isbroad, some mention should be made of interferences that canaffect clinical interpretation of laboratory data. In that context,patients who are exposed to mouse immunoglobulins throughimaging or therapeutic techniques are prone to develop


by guest on Decem


Dow

nloaded from



antibodies to mouse immunoglobulins. These antibodies,called human antimouse antibodies, or HAMA, can give riseto false-positive or false-negative results in 2-site immuno-metric assays using murine monoclonal antibodies asreagents.81,106

An increased level of a chemical analyte can affect somehematology determinations. Thus, glucose levels more than600 mg/dL (33.3 mmol/L) can lead to a transient increase inthe MCV as water enters the RBC, as the RBC is placed in anisotonic diluent, owing to the osmotic effect of glucose.107

However, as the glucose level slowly equilibrates, water exitsfrom the RBC along with glucose, thus restoring the originalMCV. This transient hypoglycemic effect can be overcome bya microdilution of the blood sample and a 5-minute waitingperiod for the equilibration to be complete before analysis.107

Triglyceride levels exceeding 1,000 mg/dL (11.29mmol/L) and hyperlipidemic specimens can cause a spuri-ously high hemoglobin value. This effect can be overcome byresuspending the RBCs in saline before analysis or by makingcorrections with a plasma hemoglobin blank.107

Turbidity resulting from the precipitation by lysisreagents of high concentrations of monoclonal proteins seenin patients with myeloma or macroglobulinemia may spuri-ously elevate the hemoglobin level and WBC count.107

Endogenous antibodies can affect both coagulation andhematology results. Circulating antibodies directed to plateletmembrane phosphatidylcholine or lecithin are referred to aslupus anticoagulants. Prothrombin is the antigen for mostlupus anticoagulants and, as such, inhibits phospholipid-depen-dent global coagulation assays.31 Antibodies directed to theplatelet membrane phospholipids phosphatidylserine and phos-phatidylinositol, which are called anticardiolipin antibodies,also inhibit phospholipid-dependent global coagulation tests.

An in vitro phenomenon of platelet agglutination seen inblood collected in EDTA is due to the presence of antibodiesin blood that are reactive to platelets. These antibodies aredirected to antigens such as the glycoprotein IIb/IIIa complexthat is hidden within the platelet membrane. EDTA, uponchelating calcium, exposes these antigens. The exposedplatelet antigen becomes modified at low and room tempera-ture, permitting platelet antibodies to agglutinate platelets.This EDTA-induced pseudothrombocytopenia is, however, notseen when blood specimens are warmed to 37 C, since theglycoprotein IIb/IIIa complex is dissociated at the highertemperature. EDTA-induced pseudothrombocytopenia at 4 Cor room temperature can cause a spuriously low platelet countand a spuriously increased WBC count if the platelet clumpsare in the same size range as WBCs. Apparently an epitope onplatelet membrane glycoprotein IIb is recognized by EDTA-dependent IgG antibodies, causing pseudothrombocy-topenia.108

EDTA-dependent IgM antibody, most active at roomtemperature and of low titer, has been reported to cause leuko-cyte agglutination noticeable on a peripheral blood smear butto provide a spuriously low automated WBC count.109 Thisphenomenon was not seen when the same patient’s specimenwas obtained in a tube with heparin or citrate.

Both pseudothrombocytopenia and pseudoleukopeniahave been reported at room temperature owing to the EDTAantibody–induced platelet aggregates not being counted asplatelets and to large neutrophil-platelet clumps falling outsidethe size range for WBCs.110 This phenomenon was abolishedby warming the specimen to 37 C.

An in vitro phenomenon that is referred to as platelet satel-litism is observed when the platelets surround the neutrophilsowing to EDTA dependent IgG autoantibodies that are directed


❚Table 8❚Selected Endogenous Interferences

Interference References

Circulating antibodies Two-site immunometric assays; positive and negative interferences have been reported Narayanan81; Primus et al106

Antibodies to platelet membrane phospholipids (lupus anticoagulant, anticardiolipin antibodies) Narayanan and Hamasaki31

inhibit phospholipid-dependent global coagulation tests Antibodies to glycoprotein IIb/IIIa complex in the platelet membrane; EDTA-induced platelet Fiorin et al108

agglutination at low or room temperature; PTCP; effect abolished at 37°C EDTA-dependent IgM antibody causes leukocyte agglutination at room temperature; spuriously Hillyer et al109

low automated WBC count not seen in specimens collected in heparin or citrateAntibodies to platelet membrane and neutrophil receptors cause platelet satellitism seen at room Bizzaro et al111; Peters et al112

temperature in EDTA-anticoagulated blood; occasionally, also seen in heparinized or citrated blood; severe platelet satellitism causes spurious PTCP; effect abolished at 37°C

Increased level of chemical analytes affecting hematology test results Glucose level >600 mg/dL causes transient increase in MCV; as glucose equilibrates, water leaves Cornbleet107

RBC restoring original MCV Triglyceride level >1,000 mg/dL causes spuriously high hemoglobin value; washing and resuspending Cornbleet107

RBCs in saline overcomes effect

MCV, mean corpuscular volume; PTCP, pseudothrombocytopenia.

by guest on Decem


Dow

nloaded from



to the glycoprotein IIb/IIIa complex in the platelet membraneand neutrophil FC-gamma receptor III of neutrophils (CD16).111

This phenomenon, which is seen at room temperature, is notseen when blood is exposed to a temperature of 37 C or periph-eral blood smears are made immediately with EDTA-anticoagu-lated or capillary blood. Occasionally, the phenomenon also hasbeen noticed with citrated blood, as well as with heparinizedblood.112 Severe platelet satellitism can cause spuriouspseudothrombocytopenia.

A discussion of endogenous interferences would not becomplete without discussing the most common interfer-ences, such as hemolysis and turbidity due to hyperlipi-demia. Apart from a few analytes, there are conflicting datain scientific literature on the effect of hemolysis on a widerange of analytes. The discrepancy is related not only todifferences in methods for the same analyte, but also todifferences in instrumentation. While to some extent instru-ments that can subtract or blank the hemoglobin interfer-ence by taking spectrophotometric measurements at thewavelength of light where hemoglobin absorbs, in additionto the wavelength at which the analyte absorbs (bichromaticmeasurements), can minimize the interference, the hemo-globin interference is not eliminated totally, especially whenthe reagent system used is affected by hemoglobin.

In general, slight hemolysis would have a negligibleeffect on many of the routine clinical chemistry laboratoryprocedures. Severe hemolysis, while having a substantialeffect on constituents that are at a much higher concentrationwithin the RBC than in plasma, could induce a slightly dilu-tional effect on the constituents present at a lower concentra-tion in the RBC than in plasma.

Typical of the enzymes present within the RBC, theconcentration of lactate dehydrogenase (LDH) is 160-foldgreater than that present in the plasma, acid phosphatase is 68-fold greater, and AST is 6.7-fold greater than the plasmalevels.113 The concentration of potassium inside the RBC isapproximately 23 times as much as that found in plasma.Naturally one would expect the aforementioned tests to beinvalidated when the specimen is hemolyzed severely.

The visual evidence of hemolysis is provided when thehemoglobin concentration exceeds 20 mg/dL. It has beenreported that the appearance of serum when 0.1% of theRBCs are lysed was virtually identical to that ofnonhemolyzed serum and, consequently, would go unde-tected by laboratory personnel looking for hemolyzed spec-imens.114 However, the appearance of serum when 1.0% ofRBCs were lysed was clear and cherry red. Such a spec-imen would be characterized as having moderate hemol-ysis. Hemolysis of 1% of the RBCs affected the measure-ment of LDH, potassium, AST, and ALT.114

Hemolysis also has been classified based on measurementof free hemoglobin levels.115 The nonhemolyzed samples had

a mean – SD free hemoglobin value of 4.8 – 3.2 mg/dL, andmoderately hemolyzed serum samples had a value of 43.5 –13.9 mg/dL. Even slight hemolysis invalidated the results forLDH, total acid phosphatase, prostatic acid phosphatase, andpotassium, and such specimens should be rejected.115 That theeffect of hemolysis is method-dependent was demonstrated bythe absence of statistically significant differences for the levelsof AST, ALT, inorganic phosphate, glucose, bilirubin, totalprotein and albumin.115

Adenylate kinase released from RBCs can spuriouslyincrease CK activity measurements unless a sufficientamount of inhibitors of adenylate kinase are incorporated inthe assay mixture.2

The peroxidative effect of heme can cause a positive andnegative interference with various diazotization methods usedfor the measurement of bilirubin, depending on the regentsused in the assay system.116 In classic diazotization proceduresfor the measurement of bilirubin, hemoglobin in thehemolyzed sample competes with nitrite for the sulfanilic acidreagent, leading to spuriously lower results.115

Depressed glucose values (values lower than the truevalue) have been reported with the glucose oxidase–peroxi-dase procedure, which apparently is related to the reagentcomposition. The negative interference has been explained asresulting from premature decomposition by hemoglobin of thehydrogen peroxide generated in the glucose oxidase reactionor from the dilution by cellular components.117

The effect of hemolysis on glucose oxidase proceduresdepends not only on the composition of the glucose oxidase-peroxidase reagent system, but also on how the assay isperformed. Thus, differences on the effect of hemolysis havebeen reported when using a kinetic assay in which the colordevelopment is monitored between 60 and 120 seconds afterinitiating the reaction as opposed to a bichromatic assay inwhich color is monitored at 2 different wavelengths; the addi-tional wavelength is used to blank out the color resulting fromsubstances other than glucose. For example, 2 grosslyhemolyzed pediatric samples with a low true glucose valuewere diagnosed correctly by the kinetic procedure that yieldedglucose values of 29 and 27 mg/dL (1.6 and 1.5 mmol/L;reference range, 70-105 mg/dL [3.9-5.8 mmol/L]). Thebichromatic procedure, on the other hand, overestimated theglucose values on the 2 pediatric samples by reporting glucosevalues of 65 and 68 mg/dL (3.6 and 3.8 mmol/L).117

Hemolysis also interferes with the hexokinase procedurefor the measurement of glucose.118 However, owing to themultiplicity of instrumentation and variations in methods forthe measurement of glucose, the magnitude of the effect ofhemolysis on each glucose procedure is unpredictable. Assuch, for accurate glucose measurements, it is desirable toavoid hemolysis or, at the very least, not exceed levels thatwould make hemolysis visually apparent in serum.


by guest on Decem


Dow

nloaded from



While total acid phosphatase activity was, as expected,affected by hemolysis due to the release of RBC acid phos-phatase, even the prostatic tartrate–sensitive fraction has beenreported to be affected by hemolysis.118 Among the hormonesvery sensitive to hemolysis is insulin.2 Apparently the releaseof proteolytic enzymes during hemolysis degrades insulin.41

While some analytes clearly are affected by hemolysis, ingeneral, method-dependent variations should be consideredwhen assessing the influence of hemolysis.

Hyperlipidemia, by introducing turbidity, can be a sourceof interference that also is method-dependent.119 Interferencecould be due to a variety of mechanisms, such as inhomo-geneity of the sample, displacement of water, and adsorptionof lipophilic constituents.2 A visibly hyperlipidemic samplecan be expected to interfere with the measurement of totalprotein and electrophoretic and chromatographic procedures.2

Summary Perspectives

So much knowledge on the preanalytic phase has accumu-lated during recent years that this review can only highlight keyissues affecting several contemporary topics.

On the subject of blood gases issues such as maintaining asteady state of ventilation with the patient in a supine positionbefore and during arterial blood collection, the effect of thecontainer material (eg, glass vs plastic syringe), dry vs liquidheparin as the anticoagulant, leakage of gases owing to syringematerial and conditions of storage, and time between collectionand analysis need to be recognized and standardized.120,121

Molecular biology is another area in which informationon the preanalytic phase is accumulating constantly. The typeof detergent used for cell lysis, its effect on DNA amplificationby PCR, the effect of RBC contamination due to the effect ofhematin on the DNA polymerase enzyme, taq polymerase,used in PCR, isolation of RNA including steps to eliminateRNase contamination, treatment of tissue, the effect of fixa-tives and duration of fixation, and protection of viral HIVRNA load in plasma from the inhibiting effect of neutrophilsby prompt separation of plasma from cells after blood collec-tion are just a few of a wide range of preanalytic variables thatmerit consideration.103

As our knowledge of cell function advances andstudies on stimulation of cells to secrete cytokines becomediagnostically important, one must ensure that there is noartifactual effect induced by the anticoagulant used forblood collection. A case in point is the endotoxin- orpyrogen-contaminated heparin stimulating monocytes tosecrete tumor necrosis factor, thereby invalidating themeasurement of this cytokine.122

Information related to the preanalytic phase of tumormarker measurements is voluminous. The preservation of

labile tumor markers, including handling of tumor tissue forthe processing of labile hormone receptors, half-life, circadianrhythm, distribution in cellular fractions, the effect of exerciseon markers such as prostate-specific antigen, amplification bymolecular methods, and sample quality for cytogenetics andflow cytometry assessment of DNA histograms are just someof the many issues that come under the preanalytic umbrella.81

As new markers are proposed, preanalytic problemsassociated with their measurements also surface. Forexample, homocysteine has received considerable attentionas a marker of cardiovascular risk. Yet the measurement ofthis amino acid presents challenges since it continues to beformed by cellular enzymes in vitro, thus making theprompt separation of cells from plasma or serum manda-tory. Alternatively, measures should be taken to inhibithomocysteine generation and converting enzymes immedi-ately after blood specimen collection.123

With advances in coagulation and fibrinolysis, includingincreasing use of molecular methods to study mutations thatpredispose a person to hypercoagulability, preanalytic vari-ables affecting these measurements need to be delineated andcontrolled. Some physiologic variables that affect themeasurement of fibrinolytic activators and inhibitors,including the extent of diurnal variation during a 24-hourperiod, the effects of smoking and alcohol need to be recog-nized. Challenges in measurement of free t-PA in plasma bysample treatment to dissociate the t-PA-plasminogen activatorinhibitor I complex should not be underestimated.31

Finally, preanalytic problems that are unique to a methodshould be kept in perspective as we begin to better understandthe complexities of the effect of new technology and therapy,inasmuch as they affect the preanalytic phase.

From the Department of Pathology, New York MedicalCollege–Metropolitan Hospital Center, New York, NY.

Address reprint requests to Dr Narayanan: Dept ofPathology, New York Medical College–Metropolitan HospitalCenter, 1901, First Ave, New York, NY 10029.

*Dr Narayanan is a Becton Dickinson Fellow at BectonDickinson, Franklin Lakes, NJ.

Acknowledgments: I thank Carol Perello for secretarialsupport and preparation of the tables.

References

1. Young DS. Effects of Preanalytical Variables on Clinical Labora-tory Tests. 2nd ed. Washington, DC: AACC Press; 1997.

2. Guder WG, Narayanan S, Wisser H, et al. Samples: From thePatient to the Laboratory: The Impact of Preanalytical Variableson the Quality of Laboratory Results. Darmstadt, Germany:GIT Verlag; 1996:1-149.

3. Narayanan S. Physiological variables in blood sampling. MittKlin Chem. 1993;24:130-134.


by guest on Decem


Dow

nloaded from



4. Mc Comb RB, Bowers GN, Posen S. Alkaline Phosphatase.New York, NY: Plenum Press; 1979:527-528.

5. Monroe BL, Weinbert AG, Rosenfeld CR, et al. Theneonatal blood count in health and disease, I: referencevalues for neutrophilic cells. J Pediatr. 1979;95:89-98.

6. Weinberg AG, Rosenfeld CR, Monroe BL, et al. Neonatalblood count in health and disease, II: values for lympho-cytes, monocytes and eosinophils. J Pediatr. 1985;106:462-466.

7. Narayanan S, Appleton HD. Creatinine: a review. ClinChem. 1980;26:1119-1126.

8. Narayanan S. Pre and post analytical errors. Indian J ClinBiochem. 1996;11:7-11.

9. Pacini G, Valerio A, Beccaro F, et al. Insulin sensitivity andbeta cell responsivity are not decreased in elderly subjectswith normal OGTT. J Am Geriatr Soc. 1988;36:317-323.

10. De Kloet ER. Corticosteroids, stress and aging. Ann N YAcad Sci. 1992;663:357-371.

11. Narayanan S. Laboratory markers as an index of aging. AnnClin Lab Sci. 1996;26:50-59.

12. Daynes RA, Araneo BA, Ershler WB, et al. Altered regula-tion of IL-6 production with normal aging: possible linkageto the age associated decline in dehydroepiandrosterone andits sulfated derivative. J Immunol. 1993;150:5219-5230.

13. Zhou D, Chrest FJ, Adler W, et al. Increased production ofTGF-beta and IL-6 by aged spleen cells. Immunol Lett.1993;36:7-11.

14. Crawford GA, Johnson AG, Gyory AZ, et al. Change inarginine vasopressin concentration with age [letter]. ClinChem. 1993;39:2023.

15. Harman SM, Wehmann RE, Blackman MR. Pituitary-thyroid hormone economy in healthy aging men: basalindices of thyroid function and thyrotropin responses toconstant infusions of thyrotropin releasing hormone. J ClinEndocrinol Metab. 1984;58:320-326.

16. Kratz A, Lewandrowski KB. Normal reference laboratoryvalues. N Engl J Med. 1998;339:1063-1072.

17. Petrovsky N, McNair P, Harrison LC. Diurnal rhythms ofpro-inflammatory cytokines: regulation by plasma cortisoland therapeutic implications. Cytokine. 1998;10:307-312.

18. Miyawaki T, Taga K, Nagaoki T, et al. Diurnal changes of Tlymphocyte subsets in human peripheral blood. Clin ExpImmunol. 1984;55:618-622.

19. Auzeby A, Bogdan A, Krosi Z, et al. Time-dependence ofurinary neopterin, a marker of cellular immune activity. ClinChem. 1988;34:1866-1867.

20. Keffer JH. Preanalytical considerations in testing thyroidfunction. Clin Chem. 1996;42:125-134.

21. Narayanan S. Current concepts of thyroid function andlaboratory evaluation. Indian J Clin Biochem. 1997;12:25-34.

22. Cooper GR, Myers GL, Smith JJ, et al. Standardization oflipid, lipoprotein, and apolipoprotein measurements. ClinChem. 1988;34:B95-105.

23. Narayanan S. Pre and post analytical errors in lipid determi-nation. Indian J Clin Biochem. 1996;11:12-16.

24. Statland BE, Winkel P. Effect of preanalytical factors on theintra-individual variation of analytes in the blood of healthysubjects: consideration of preparation of the subject andtime of venipuncture. Crit Rev Clin Lab Sci. 1977;9:105-144.

25. Harrop IS, Ashwell K, Hopton MR. Circannual and withinindividual variation of thyroid function tests in normalsubjects. Ann Clin Biochem. 1985;22:371-375.

26. Narayanan S. Laboratory monitoring of gestational diabetes.Ann Clin Lab Sci. 1991;21:392-401.

27. Cooper GR, Myers GL, Smith J, et al. Blood lipid measure-ments: variations and practical utility. JAMA.1992;267:1652-1660.

28. Knopp RH, Walden CE, Whal PW, et al. Effect of post-partum lactation on lipoprotein lipids and apoproteins. JClin Endocrinol Metab. 1985;60:542-547.

29. Burrow GN, Fisher DA, Larsen PR. Maternal and fetalthyroid function. N Engl J Med. 1994;331:1072-1078.

30. Sprengers ED, Kluft C. Plasminogen activator inhibitors.Blood. 1987;69:381-387.

31. Narayanan S, Hamasaki N. Current concepts of coagulationand fibrinolysis. Adv Clin Chem. 1999;33:133-168.

32. Harris WS, Connor WE, Illingworth DR, et al. Effects of fishoil on VLDL triglyceride kinetics in humans. J Lipid Res.1990;31:1549-1558.

33. Sacks FM, Ornish D, Rosner B, et al. Plasma lipoproteinlevels in vegetarians. JAMA. 1985;254:1337-1341.

34. Robertson D, Frolich JC, Carr RK, et al. Effects of caffeineon plasma renin activity, catecholamines, and blood pres-sure. N Engl J Med. 1978;298:181-186.

35. Van Dusseldorp M, Katan MB, Demacker PNM. Effect ofdecaffeinated versus regular coffee on serum lipoproteins: a12-week double-blind trial. Am J Epidemiol. 1990;132:33-40.

36. Freer DE, Statland BE. Effects of ethanol (0.75 g/kg bodyweight) on the activities of selected enzymes in sera ofhealthy young adults, II: interindividual variations inresponse of gamma-glutamyltransferase to repeated ethanolchallenges. Clin Chem. 1977;23:2099-2102.

37. Freer DE, Statland BE. The effect of ethanol (0.5 g/kg bodyweight) on the activities of selected enzymes in sera ofhealthy young adults, I, intermediate term effects. ClinChem. 1977;23:830-834.

38. Taskinin MR, Nikkila EA, Valimaki M, et al. Alcohol-induced changes in serum lipoproteins and in their metabo-lism. Am Heart J. 1987;113:458-464.

39. Haboubi NAA, Bignell AHC, Haboubi NY. Serumangiotensin converting enzyme activity in cigarette smokers.Clin Chim Acta. 1986, 154:69-72.

40. Cloey T, Bachorik PS, Becker D, et al. Reevaluation ofserum-plasma difference in total cholesterol concentration.JAMA. 1990;263:2788-2789.

41. Young DS. Biological variability. In: Brown SS, Mitchell FL,Young DS, eds. Chemical Diagnosis of Disease. New York, NY:Elsevier; 1979:1-113.

42. Narayanan S. Considerations in therapeutic drug moni-toring. Ital J Clin Pathol. 1984;4:73-75.

43. Pippenger CE. Quantitation and interpretation of serumdrug concentrations. In: Spiegel HE, ed. Clinical Biochem-istry: Contemporary Theories and Techniques. New York, NY:Academic Press; 1982:2:255-297.

44. Narayanan S. Renal biochemistry and physiology: patho-physiology and analytical perspectives. Adv Clin Chem.1992;29:121-159.

45. Guder WG, Ivandic M, Hofmann W. Pathophysiology ofproteinuria and laboratory diagnostic strategy based onsingle protein analysis. Clin Chem Lab Med. 1998;36:935-939.

46. Tan ACITL, Kloppenborg PWC, Benraad TJ. Influence ofage, posture and intra-individual variation on plasma levelsof atrial natriuretic peptide. Ann Clin Biochem. 1989;26:481-486.


by guest on Decem


Dow

nloaded from



47. Don BR, Sebastian A, Cheitlin M, et al. Pseudohyper-kalemia caused by fist clenching during phlebotomy. N EnglJ Med. 1990;322:1290-1292.

48. Skinner SL, Adelaide MB. A cause of erroneous potassiumlevels. Lancet. 1961;1:478-480.

49. Kollee LAA, Willems JL, de Kort AFM, et al. Bloodsampling technique for lactate and pyruvate estimation inchildren. Ann Clin Biochem. 1977;14:285-287.

50. Hardman AE, Hudson A, Jones PRM, et al. Brisk walkingand plasma high-density lipoprotein cholesterol concentra-tion in previously sedentary women. Br Med J.1989;299:1204-1205.

51. Novotny WF, Palmier MO, Wun TC, et al. Purification andproperties of heparin releasable lipoprotein-associatedinhibitor. Blood. 1991;78:394-400.

52. Narayanan S. Effect of anticoagulants used for blood collec-tion on laboratory tests. Jpn J Clin Pathol. 1996;103:73-80.

53. Guder W, Ehret W, Fonseca-Wollheim FD, et al. Serum,plasma or whole blood? Which anticoagulants to use. J LabMed. 1998;22:297-312.

54. Narayanan S. Relative merits of plasma derived from threeheparin salts and serum for sodium and potassium measure-ments by direct and indirect potentiometry. In: Maas AHJ,Boink FBTJ, Saris NL, et al, eds. Methodology and ClinicalApplications of Ion Selective Electrodes. Copenhagen,Denmark: IFCC Technical Secretariat. 1986;7:209-212.

55. Berg JD, Romano G, Bayley NF, et al. Heparin interfereswith aspartate aminotransferase activity determination inthe Ektachem 700. Clin Chem. 1988;34:174.

56. Doumas BT, Hause LL, Simuncak DM, et al. Differencesbetween values for plasma and serum in tests performed inthe Ektachem 700 XR analyzer and evaluation of “plasmaseparator tubes (PST).” Clin Chem. 1989;35:151-153.

57. Toffaletti J, Ernst P, Hunt P, et al. Dry electrolyte-balancedheparinized syringes evaluated for determining ionizedcalcium and other electrolytes in whole blood. Clin Chem.1991;37:1730-1733.

58. Toffaletti J. Use of novel preparations of heparin to elimi-nate interference in ionized calcium measurements: have allthe problems been solved? Clin Chem. 1994;40:5508-5509.

59. Landt M, Hortin GL, Smith CH, et al. Interference inionized calcium measurements by heparin salts. Clin Chem.1994;40:565-570.

60. Siebers RWL, Chen CT, Ferguson RI, et al. Effect of bloodsample tubes on amiodarone and desethyl amiodaroneconcentrations. Ther Drug Monit. 1988;10:349-351.

61. Berquist Y, Domeji-Nyberg B. Distribution of chloroquineand its metabolite desethylchloroquine in human blood cellsand its implication for the quantitative determination ofthese compounds in serum and plasma. J Chromatogr.1983;272:137-148.

62. Goosens W, Van Duppen V, Verwilghen RL. K2 or K3EDTA: the anticoagulant of choice in routine hematology?Clin Lab Haematol. 1991;13:291-295.

63. Sacker IS. Specimen collection. In: Lewis SM, Coster JF,eds. Quality Control in Haematology: Symposium of the Inter-national Committee for Standardization in Haematology. NewYork, NY: Academic Press; 1975:211-229.

64. Reardon DM, Warner B, Trowbridge EA. EDTA, the tradi-tional anticoagulant of hematology: with increased automa-tion is it time for a review? Med Lab Sci. 1991;48:72-75.

65. Narayanan S. Preanalytical aspects of coagulation testing.Haematologica. 1995;80(2 suppl):1-5.

66. Adcock DM, Kressin DC, Marlar RA. Effect of 3.2% vs3.8% sodium citrate concentration on routine coagulationtesting. Am J Clin Pathol. 1997;107:105-110.

67. Chan AYW, Swaminathan R, Cockram CS. Effectiveness ofsodium fluoride as a preservative of glucose in blood. ClinChem. 1989;35:315-317.

68. Sidebottom RA, Williams PR, Kanarek KS. Glucose deter-minations in plasma and serum: potential error related toincreased hematocrit. Clin Chem. 1982;28:190-192.

69. Meites S, Saniel-Banrey K. Preservation, distribution andassay of glucose in blood, with special reference to thenewborn. Clin Chem. 1979;25:531-534.

70. Uchida K, Matuse R, Toyoda E, et al. A new method ofinhibiting glycolysis in blood samples. Clin Chim Acta.1988;172:101-108.

71. Liss E, Bechtel S. Improvement of glucose preservation inblood samples. J Clin Chem Clin Biochem. 1990;28:689-690.

72. Nakashima K, Takei H, Nasu Y, et al. D-mannose as apreservative of glucose in blood samples. Clin Chem.1987;33:708-710.

73. Ho CS, Fung SL, Chan AY. Interference of D-mannose inglucose measurements by glucose oxidase and hexokinasemethods [letter]. Clin Chem. 1991;37:477.

74. Contant G, Gouavult-Heilmann M, Martinoli JL. Heparininactivation during blood storage: its prevention by bloodcollection in citric acid, theophylline, adenosine, dipyri-damole-CTAD mixture. Thromb Res. 1983;31:365-374.

75. Narayanan S. Inhibition of in vitro platelet aggregation andrelease and fibrinolysis. Ann Clin Lab Sci. 1989;19:260-265.

76. Van Den Besselaar AMH, Meeuwisse-Braun J, Jansen-Gruter R, et al. Monitoring heparin therapy by the activatedpartial thromboplastin time: the effect of preanalyticalconditions. Thromb Haemost. 1987;58:226-231.

77. Kuhne T, Hornstein A, Semple J, et al. Flow cytometricevaluation of platelet activation in blood collected intoEDTA vs Diatube-H, a sodium citrate solution supple-mented with theophylline, adenosine and dipyridamole. AmJ Hematol. 1995;50:40-45.

78. Watkins J, Wild G, Smith S. Nafamostat to stabilize plasmasamples taken for complement measurements. Lancet.1989;1:896-897.

79. Narayanan S. Protection of peptidic substrates by proteaseinhibitors. Biochim Clin. 1987;11:954-956.

80. Pandian MR, Morgan CH, Carlton E, et al. Modifiedimmunoradiometric assay of parathyroid hormone–relatedprotein: clinical application in the differential diagnosis ofhypercalcemia. Clin Chem. 1992;38:282-288.

81. Narayanan S. Quality control in tumor marker analysis:preanalytical, analytical, and post analytical issues. J ClinLigand Assay. 1998;21:11-17.

82. Boomsma F, Alberts G, van Eijk L, et al. Optimum collec-tion and storage conditions for catecholamine measurementsin human plasma and urine. Clin Chem. 1993;39:2503-2508.

83. Connaghan DG, Francis CE, Ryan DH, et al. Prevalenceand clinical implications of heparin-associated false positivetests for serum fibrin(ogen) degradation products. Am J ClinPathol. 1986;86:304-310.

84. Mohler MA, Refino CJ, Chen SA, et al. D-Phe-Pro-Arg-chloromethylketone: its potential use in inhibiting forma-tion of in vitro artifacts in blood collected during tissue-typeplasminogen activator thrombolytic therapy. ThrombHaemost. 1986;56:160-164.


by guest on Decem


Dow

nloaded from



85. Tsuji T, Masuda H, Imagawa K, et al. Stability of human atrialnatriuretic peptide in blood sample. Clin Chim Acta.1994;225:171-177.

86. Numata Y, Dohi K, Furukawa A, et al. Immunoradiometricassay for the N-terminal fragment of proatrial natriureticpeptide in human plasma. Clin Chem.1998;44:1008-1013.

87. Reneke J, Etzell J, Leslie S, et al. Prolonged prothrombin timeand activated partial thromboplastin time due to underfilledspecimen tubes with 109 mmol/L (3.2%) citrate anticoagu-lant. Am J Clin Pathol. 1998;109:754-757.

88. Siegel JE, Bernard DW, Swami VK, et al. Monitoring heparintherapy: aPTT results from partial- vs full-draw tubes. Am JClin Pathol. 1998;110:184-187.

89. Koepke JA, Rodgers JL, Ollivier MJ. Pre-instrumental vari-ables in coagulation testing. Am J Clin Pathol. 1975;64:591-596.

90. O’Connell ME, Heim KL, Halstenson CE, et al. Analyticalaccuracy of determinations of aminoglycoside concentrationsby enzyme multiplied immunoassay, fluorescence polarizationimmunoassay, and radioimmunoassay in the presence ofheparin. J Clin Microbiol. 1984;20:1080-1082.

91. Pewarchuk W, Vanderboom J, Blajchman MA. Pseudopoly-cythemia, pseudothrombocytopenia, and pseudoleukopeniadue to overfilling of blood collection vacuum tubes. ArchPathol Lab Med. 1992;116:90-92.

92. Aranson MD, Bor DH. Blood cultures. Ann Intern Med.1987;106:246-253.

93. Washington JA II, Ilstrup DM. Blood cultures: issues andcontroversies. Rev Infect Dis. 1986;8:792-802.

94. Ono T, Kitaguchi K, Takehara M, et al. Serum constituentsanalyses: effect of duration and temperature of storage ofclotted blood. Clin Chem. 1981;27:35-38.

95. da Fonseca-Wollheim F. Deamidation of glutamine byincreased plasma gamma-glutamyltransferase is a source ofrapid ammonia formation in blood and plasma specimens.Clin Chem. 1990;36(8 Pt 1):1479-1482.

96. Ho C, Wu S. The influence of time, temperature, and packedcell volume on activated partial thromboplastin time andprothrombin time. Thromb Res. 1991;62:625-633.

97. Plumhoff EA, Thompson CK, Fisher PK, et al. Effects of spec-imen storage and handling on coagulation testing [abstract].Thromb Haemost. 1993;69:866.

98. Keshgegian AA, Bull GE. Evaluation of a soft-handlingcomputerized pneumatic tube specimen delivery system:effects on analytical results and turnaround time. Am J ClinPathol. 1992;97:535-540.

99. Astles JR, Lubarsky D, Loun B, et al. Pneumatic transportexacerbates interference of room air contamination in bloodgas samples. Arch Pathol Lab Med. 1996;120:642-647.

100. Lippi U, Schiniella M, Modena N, et al. Advantages of a newanticoagulant in routine hematology on the Coulter counterS-Plus STKR analyzer. Am J Clin Pathol. 1990;93:760-764.

101. Nicholson JKA, Jones BM, Cross D, et al. Comparison of T-and B-cell analysis on fresh and aged blood. J ImmunolMethods. 1984;73:29-40.

102. Carter PH, Resto-Ruiz S, Washington GC, et al. Flow cyto-metric analysis of whole blood lysis, three anticoagulants, andfive cell preparations. Cytometry. 1992;13:68-74.

103. Narayanan S. Preanalytical and analytical pitfalls in molec-ular biology techniques. J Clin Ligand Assay. 1997;20:200-205.

104. Engstad CS, Gutteberg TJ, Osterud B. Modulation of bloodcell activation by four commonly used anticoagulants. ThrombHaemost. 1997;77:690-696.

105. Young DS. Effect of Drugs on Clinical Laboratory Tests. 4thed. Washington, DC: AACC Press; 1995.

106. Primus FJ, Kelley EA, Hansen HJ, et al. “Sandwich-type”immunoassay of carcinoembryonic antigen in patientsreceiving murine monoclonal antibodies for diagnosis andtherapy. Clin Chem. 1988;34:261-264.

107. Cornbleet J. Spurious results from automated hematologycell counters. Lab Med. 1983;14:509-514.

108. Fiorin F, Steffan A, Pradella P, et al. IgG platelet antibodiesin EDTA-dependent pseudothrombocytopenia bind toplatelet membrane glycoprotein IIb. Am J Clin Pathol.1998;110:178-183.

109. Hillyer CD, Knopf AN, Berkman EM. EDTA-dependentleukoagglutination. Am J Clin Pathol. 1990;94:458-461.

110. Moraglio D, Banfi G, Arnelli A. Association ofpseudothrombocytopenia and pseudoleukopenia. Scand JClin Lab Invest. 1994;54:257-265.

111. Bizzaro N, Goldschmeding R, van dem Borne AE. Plateletsatellitism is Fc gamma R III (CD16) receptor mediated. AmJ Clin Pathol. 1995;103:740-744.

112. Peters MJ, Heyderman RS, Hatch DJ, et al. Investigation ofplatelet-neutrophil interactions in whole blood by flowcytometry. J Immunol Methods. 1997;209:125-135.

113. Caraway WT. Chemical and diagnostic specificity of labora-tory tests. Am J Clin Pathol. 1961;37:445-464.

114. Laessig RH, Hassemer DJ, Paskey TA, et al. The effects of0.1 and 1.0 per cent erythrocytes and hemolysis on serumchemistry values. Am J Clin Pathol. 1978;66:639-644.

115. Yucel D, Dalva K. Effect of in vitro hemolysis on 25common biochemical tests. Clin Chem. 1992;38:575-577.

116. Van der Woerd-de Lange JA, Guder WG, Schieicher E, etal. Studies on the interference by hemoglobin in the deter-mination of bilirubin. J Clin Chem Clin Biochem.1983;21:437-443.

117. Leary NO, Pembroke A, Duggan PF. Improving accuracy ofglucose oxidase procedure for glucose determinations ondiscrete analyzers. Clin Chem. 1992;38:298-302.

118. Glick MR, Ryder KW, Jackson SA. Graphical comparisonsof interferences in clinical chemistry instrumentation. ClinChem. 1986;32:470-475.

119. Grafmeyer D, Bondon M, Manchon M, et al. The influenceof bilirubin haemolysis and turbidity on 20 analytical testsperformed on automatic analysers. Eur J Clin Chem ClinBiochem. 1995;33:31-52.

120. Muller-Plathe O, Heyduck S. Stability of blood gases, elec-trolytes and haemoglobin in heparinized whole bloodsamples: influence of the type of syringe. Eur J Clin ChemClin Biochem. 1992;30:349-355.

121. Burnett RW, Covington AK, Fogh-Andersen N, et al. International Federation of Clinical Chemistry (IFCC),Scientific Division, Committee on pH, Blood Gases andElectrolytes: approved IFCC recommendations on wholeblood sampling, transport and storage for simultaneousdetermination of pH, blood gases and electrolytes. Eur J Clin Chem Clin Biochem. 1995;33:247-253.

122. Leroux-Roels G, Offner F, Philippe J, et al. Influence ofblood-collecting systems on concentrations of tumornecrosis factor in serum and plasma. Clin Chem.1988;34:2373-2374.

123. Nygard O, Vollset SM, Refsum H, et al. Total plasma homo-cysteine and cardiovascular risk profile: the Hordalandhomocysteine study. JAMA. 1995;274:1526-1533.


by guest on Decem


Dow

nloaded from


the preanalytic phase an important component of...

Documents