Clinica Chimica Acta Volume 412 Issue 9-10 2011 [Doi 10.1016%2Fj.cca.2011.01.010] Paola Brunori;...

8/13/2019 Clinica Chimica Acta Volume 412 Issue 9-10 2011 [Doi 10.1016%2Fj.cca.2011.01.010] Paola Brunori; Piergiorgio M

1/4

Evaluation of bilirubin concentration in hemolysed samples, is it really impossible?

The altitude-curve cartography approach to interfered assays

Paola Brunori a,, Piergiorgio Masi a, Luigi Faggiani a, Luciano Villani a, Michele Tronchin b, Claudio Galli b,Clarissa Laube a, Antonella Leoni a, Maila Demi a, Antonio La Gioia a

a Azienda USL 6 Livorno, Italyb Abbott Diagnostic Division, Roma, Italy

a b s t r a c ta r t i c l e i n f o

Article history:

Received 18 November 2010

Received in revised form 6 January 2011

Accepted 6 January 2011

Available online 14 January 2011

Keywords:

Bilirubin

Icterus

Hemolysis

Interference

Neonatal jaundice

Background: Neonatal jaundice might lead to severe clinical consequences. Measurement of bilirubin in

samples is interfered by hemolysis. Over a method-depending cut-off value of measured hemolysis, bilirubin

value is not accepted and a new sample is required for evaluation although this is not always possible,

especially with newborns and cachectic oncological patients. When usage of different methods, less prone to

interferences, is not feasible an alternative recovery method for analytical signicance of rejected data might

help clinicians to take appropriate decisions.

Methods: We studied the effects of hemolysis over total bilirubin measurement, comparing hemolysis-

interfered bilirubinmeasurement with the non-interfered value. Interference curves wereextrapolated overa

wide range of bilirubin (030 mg/mL) and hemolysis (H Index 01100).

Results: Interference altitude curves werecalculated and plotted.A bimodal acceptance table was calculated.

Non-interfered bilirubin of given samples was calculated, by linear interpolation between the nearest lower

and upper interference curves.

Conclusions: Rejection of interference-sensitive data from hemolysed samples for every method should be

based not upon the interferent concentration but upon a more complex algorithm based upon the

concentration-dependent bimodal interaction between the interfered analyte and the measured interferent.

The altitude-curve cartography approach to interfered assays may help laboratories to build up their ownmethod-dependent algorithm and to improve the trueness of their data by choosing a cut-off value different

from the one (10% interference) proposed by manufacturers.

When re-sampling or an alternative method is not available the altitude-curve cartography approach might

also represent an alternative recovery method for analytical signicance of rejected data.

2011 Elsevier B.V. All rights reserved.

1. Introduction

Hemolysis (H) is the most common cause of blood sample

inadequacy and interferes in different ways with several assays [1].

As far as bilirubin (B) measurement is concerned, the interference of

hemolysis is caused by two kinds of events:

1. Physical interference (light absorption of the heme group) and

chemical interference (inhibition of diazotization).

2. Physiological catabolism of plasma/serum heme groups into

bilirubin.

When a hemolysed sample arrives in the laboratory physiological

generation of bilirubin from hemoglobin may be easily prevented or,

at least, limited e.g. by immediate measurement of bilirubin, while

physicalchemical interferences are already present and may only be

evaluated in order to accept (or reject) the results from the analyzer.

Evaluation of hemolysis might be performed in a visual, qualitative

way but also in a semiquantitative way using interference assays that

are normally available in modern clinical chemistry analyzers.

Interference assays are to be preferred for better reliability [2]. For

every bilirubin methoda cut-off value for hemolysis(H) is established

Clinica Chimica Acta 412 (2011) 774777

Abbreviations:Bil, bilirubin (total)total bilirubin value; Bil, value of Bil in a non-

hemolytic sample; BilHmeas, measured value of Bil in a hemolytic sample; Bil index,

icterus index value; H, hemolysishemolysis index value; Hhigh, calculated value of H

for the nearest higher interference curve; Hlow, calculated value of H for the nearest

lower interference curve; Hmeas, measured value of H in a hemolytic sample; I%,

interference; I%chosen, chosen interference level for altitude curves; I%low, nearest

lower interference curve; I%high, nearest higher interference curve; l%mis, measured

interference; POCT, point of care testing.

Corresponding author. Laboratorio Ospedale Villamarina, via Forlanini 22, 57025

Piombino (LI), Italy.

E-mail address:[email protected](P. Brunori).

0009-8981/$ see front matter 2011 Elsevier B.V. All rights reserved.

doi:10.1016/j.cca.2011.01.010

Contents lists available at ScienceDirect

Clinica Chimica Acta

j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / c l i n c h i m
http://dx.doi.org/10.1016/j.cca.2011.01.010http://dx.doi.org/10.1016/j.cca.2011.01.010http://dx.doi.org/10.1016/j.cca.2011.01.010mailto:[email protected]://dx.doi.org/10.1016/j.cca.2011.01.010http://www.sciencedirect.com/science/journal/00098981http://www.sciencedirect.com/science/journal/00098981http://dx.doi.org/10.1016/j.cca.2011.01.010mailto:[email protected]://dx.doi.org/10.1016/j.cca.2011.01.010


2/4

by the manufacturer's insert sheet [3] or laboratory experience.

Therefore the Bil value of a given sample might be rejected if the H

value exceeds this threshold; in this case a new sample without

Hemolysis should be obtained for a correct measurement of

bilirubin. In most cases this is not a great problem, but for newborns

a new sample might not be easy to obtain since vein diameter in

babies is very small, the newborn total blood volume may not allow a

second sampling and, nally, newborn spontaneous hemolysis might

be the causative agent of high bilirubin concentration[4].The measurement of bilirubin in neonatology is a critical point,

since high concentration of bilirubinat birth maygenerate permanent

and severe encephalopathy (kernicterus). Classical kernicterus is a

well-described clinical tetrad of (i) abnormal motor control, move-

ments and muscle tone, (ii) an auditory processing disturbance with

or without hearing loss, (iii) oculomotor impairments, especially

impairment of upward vertical gaze, and (iv) dysplasia of the enamel

of deciduous teeth[5].

Measurements must be accurate and with a short TAT since

evolution of bilirubin in newborns may be very fast, and the clinical

protocol must also take into account the patient's age (in hours)and B

increment speed. The clinicians, therefore, are asking for a different

solution in this context[6].

A second source of problems is that, conventionally, the interfer-

ence cutoffpoint is chosen when interference is equal to 10%, which is

very close, but greater, than 9.8%, that represents the European

inaccuracy goal for bilirubin [7] without considering any other

possible source of inaccuracy.

Therefore we decided to investigate in depth the inuence of

hemolysis on bilirubin measurements in order to propose a different

approach.

2. Methods

Pools of sera with different values of bilirubin(Bil) were generated

from blood donors' samples without relevant hemolysis (inclusion

criteria was a free hemoglobin concentration b20 mg/dL) and were

divided into quotes and frozen until usage. Immediately before

testing, red blood cells from donors without increased Bil wereseparated from plasma, manually hemolysed and centrifuged. The

hemolysed supernatant was used to generate interfering samples by

dilution with a phosphate-buffered saline solution. Samples for

measurement were generated adding 1 part of one interfering sample

and 9 part of one icteric sample. Final free hemoglobin concentration

in the above-mentioned samples ranged from 35 to 1050 mg/dL. Non-

interfered samples (nal free hemoglobin concentration ranged from

2 to 12 mg/dL) were generated by adding 1 part of saline solution

instead of the interfering sample. For each sample we measured the

interference index (Hmeas) and total bilirubin (BilHmeas) with the

reagent (total bilirubin LN 8G6220, Abbott Diagnostics, Chicago, IL)

and instruments (ARCHITECT ci8200, Abbott Diagnostics, Chicago, IL)

routinely used in our laboratory. Each point was measured at least

twice with two different instruments. Subsequent data processingwas performed using a standard worksheet (MS Excel) program. Data

were compared with non-interfered bilirubin (Bil) to calculate

interference (I%) as

I% = BilHmeas BilB

= BilB 100:

The theoretical values of I%=5%, 10%, 15%, 20%, 25%,

30%,35%,40%,45%,50%,55%,60%,65%, 70%, 75%, 80%,

85%, 90% and95%were chosen(I%chosen) forcalculatingthe valuesof

experimental interference altitude curves. Data for I%chosen were

calculate from H (HI%) and Bil (BilI%) for any chosen interference

altitudecurve between two experimental points (BilHmeas1; Hmeas1)

and (BilHmeas2; Hmeas2) whose I% were respectively lower (I%meas1)

and greater (I%meas2) than the chosen I% (I%chosen), using classical

interpolation formulas

HI% = Hmeas1 + I%chosen I%meas1 = I%meas2 I%meas1 Hmeas2 Hmeas1

BilI% = BilHmeas1 + I%chosen I%meas1 = I%meas2 I%meas1

BilHmeas2 BilHmeas1 :

Later, second order regression curves with the formula

H = A Bil2

+ B Bil + C

were calculated over (HI%; BilI%) points of the chosen interference

altitudecurves.

After completing this phase a worksheet was build to plot into

altitude curves any experimental (BilHmeas; Hmeas). In this worksheet

H values in the interference curves are calculated for the given value

of (BilHmeas) and compared to (Hmeas). The H value of the nearest

lower altitude curve of (BilHmeas; Hmeas) is dened as Hlowand the H

value of the nearest higher altitude curve of (BilHmeas; Hmeas) is

dened as Hhigh. This procedure allows to calculate the position of

Hmeas in the line (Hlow: Hhigh) and therefore to evaluate the

interference according to the formula

I% = I%low + Hmeas Hlow = Hhigh Hlow

I%high I%low

:

Therefore Bil is calculated with the formula

BilBcalc = Bilmeas= 1 + I% :

Four fresh samples (verication samples), with signicant Bil

values were treated like the former ones to test the worksheet and

compared their measured Bil (1 part of saline instead of interference

sample) with the calculated Bil.

3. Results

Fig. 1A shows average bulk decay curves of B versus H. Visual

analysis of bulk decay curves was used to evaluate the coverage of the

measured area and eventually to check the single experimental points

in order to replicate eventual non-tting points before proceeding

with calculations.

Fig. 1.Bulk bilirubin measurements (A: mg/dL) and interferences (B: %) versus index

measurements (H).

775P. Brunori et al. / Clinica Chimica Acta 412 (2011) 774777
http://www.westgard.com/europe.htmhttp://www.westgard.com/europe.htm


3/4

Fig. 1B shows the interaction of I% and H for the same data. The

manufacturer indication for interference cutoff point (absolute value

of interference 10% which correspond to I%10%) indicates an H

value of 31. The decay trends of Bil=0.99 mg/dL and Bil=1.73 mg/

dL when H=31 show an absolute value of interference greater than

10% (17% and 13% respectively) while two other ones show absolute

values of interference between 5% and 10% (8.5%, and 5% for

Bil= 4.46 mg/dL, Bil= 5.94 mg/dL, respectively), and negligible

(b

5%) for the other ones (Bil=8.43 mg/dL, Bil=12.62 mg/dL andBil= 29.98 mg/dL).

I% =10% was found at H values of 19, 24, 35, 49, 65, 84, and 109

f or Bi l = 0.99 mg/dL, Bi l = 1.73 mg/dL, Bi l = 4.46 mg/dL,

Bil =5.94 mg/dL, Bil =8.43 mg/dL, Bil=12.62 mg/dL, and

Bil=29.98 mg/dL, respectively.

Fig. 2 shows calculated altitude data points obtained with

previous data interpolation as described above, with their second-

order tting polynomial curves. Calculated points for curves 80%,

85%, 90% and 95% were less than 4 for each curve, therefore the

second-ordertting polynomial curve for this data was forced to be a

rst-order polynomial curve.

Square correlation coefcient (R2) ranged from 0.977 for

I%chosen=5%, to 0.999 for I%chosen=80%. Fit parameters were

used to generate data pairs (BilTmisversus Hmis) inTable 1for three

chosen altitudecurves: 2.5%, 5%, 7.5% and 10% respectively.

Fit parameters from 5%, 10%, 15%, 20%, 25%, 30%,

35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%,

90%and95%curves respectivelywere used to generate an altitude

plot, as shown inFig. 3, in a worksheet le.Table 2shows data of four

verication samples that were treated as described above, and were

used to test the worksheet le. Baseline H (samples with saline)

showed free hemoglobin b20 mg/dL for all samples except #4 (free

hemoglobin=66 mg/dL). InTable 2the icterus index value (Bilindex)

is also shown as a comparison.

Data calculation was not possible for sample 1C (Bilmisb0.1 mg/dL).

Bilcalc of all other samples were in agreement with the measured Bil. It

should be considered that, a not-null value of index is also present in

samples with saline. Therefore the correction formula was used also

with Bilmeasto generate a corrected Bilmeas. Comparison of all Bilcalcversus corrected Bilmeas showed a better agreement than the

comparison versus non-corrected Bilmeas. Average bias of Bilcalc versus

corrected Bilmeas was 37% (50% versus non-corrected Bilmeas) for

sample #1.Average bias of Bilcalc versus corrected Bilmeas was 9% (11%

versus non-corrected Bilmeas) for sample #2. Average bias of Bilcalcversus corrected Bilmeas was 2% (4% versus non-corrected Bilmeas)

for sample #3. Average bias of Bilcalc versus corrected Bilmeas was1%

(8% versus non-corrected Bilmeas) for sample #4.

Fig. 4 shows a bias and precision plot from the data of the

verication samples.

4. Discussion

Bilirubin measurement must identify the infants at risk for

permanent brain damages (kernicterus) and other pathological

conditions [6]. It is of concern that many bilirubin automated kit

methods suffer from hemolysis interference. Since hemolysis may be

one of the causes of high bilirubin concentration, many methods often

fail to measure bilirubin.

Interference is not only a matter of hemoglobin concentration but

depends also upon bilirubin concentration since the same value of H

produces different values of I% upon different bilirubin concentrationsand therefore it is a double-factor event. The interaction is

reproducible, therefore a mathematical approach might expand

measuring possibilities beyond their traditional limits. It is clear that

evaluation of bilirubin from the icterus index is a possible approach,

but is not recommended[3] since it may generate untrue, although

precise, measurement. A different approach might be non-invasive

patient-side (POCT) measurement of transcutaneous bilirubin with

multiwavelength spectral reectance analysis[8].

Multiwavelength spectral reectance analysis, is methodologically

similar to the above-mentioned index method, without the need of

blood sampling, and therefore without needle-induced hemolysis.

Fig. 2.

Altitude-curve

points with regression curves.

Table 1

Bimodal acceptance table for data rejection.

Bil Tmeas(mg/dL) H index Hb (mg/dL) values for interference curves

2.5% 5.0% 7.5% 10%

0 1 3 3 3

1 3 7 9 12

2 5 10 16 21

3 7 14 21 29

4 8 17 27 37

5 10 20 32 45

6 12 23 37 52

7 13 26 42 58

8 14 29 47 65

9 16 32 51 70

11 17 34 55 76

12 18 37 59 81

13 19 39 62 86

14 21 41 66 90

15 22 43 69 94

16 23 45 71 98

17 23 47 74 101

18 24 49 76 103

19 25 50 78 106

20 26 51 80 108

21 27 54 82 111

22 27 55 83 111

23 28 56 84 11224 28 57 84 112

25 29 57 84 111

Fig. 3. Usageof altitudecurves: a patient () withBilmeas=15.0 mg/dLand Hmeas=600

is evaluated to be affected by a 39% interference therefore Bilcalc=24.5 mg/dL.

776 P. Brunori et al. / Clinica Chimica Acta 412 (2011) 774777
http://localhost/var/www/apps/conversion/tmp/scratch_2/image%20of%20Fig.%E0%B3%80http://localhost/var/www/apps/conversion/tmp/scratch_2/image%20of%20Fig.%E0%B2%80


4/4

Nevertheless, transcutaneous bilirubin measurement might overesti-

mate blood bilirubin[10], when signicant skin bilirubin is present.

The altitude curve approach produces a reliable estimation of

interference and Bil. This approach may improve traditional data

rejectiondening thehemolysis cutoffpoint no more as a singlepoint,

but asa function ofBilHmeas (Table 1). Indications fordata rejectionare

traditionally indicated by manufacturers when the absolute value of

interference is equal to 10%, which is very close, but greater, than 9.8%,

that represents the European inaccuracy goal for bilirubin[7]without

considering any other possible source of inaccuracy. With the

altitude curve approach more restrictive cutoffs are available,

therefore in Table 1 not only the traditional minimal (10%)

curve is indicated but also the 2.5% (desirable as of the

European total inaccuracy goal), the 5% (acceptable as of the

European total inaccuracy goal) and 7.5% (permissiveas of the

European total inaccuracy goal) curves respectively.

Interference plots like inFigs. 2and3or a commercial worksheetlike inTable 2may help evaluate Bil from BilHmeasand Hmeas if the

laboratory is more condent with data calculation.

The altitude curveapproach has its own limitations, rst of all it

may estimate chemicalphysical interference but not the physiolog-

ical after-sampling production of bilirubin from hemolised samples

[1]. Furthermore, results of this work are limited to the method in use

in our laboratory, but other laboratories that employ different

methods or instruments may use the same approach to build their

own altitude curves. While the choice of different reagents less

prone to interference by hemolysis will be of more help in specialized

settings[9], this method will be useful in most general labs that serve

both in- and outpatients, as it is common in Italy and generally in

Europe; in these settings it is difcult to envision theusage of different

reagents according to the population assayed. If reagents more

sensitive to hemolysis are in use, our method may be useful, and

though it is validated only for the Architectreagent, it may be adapted

also to different ones. The altitude-curve cartography approach may

give new sights and solutions to the problem of interferences,

especially where a new sampling is not possible or convenient, and

the interfered analyte should be quickly determined to drive

important therapeutic decisions, like bilirubin in the neonatology

eld.

References

[1] BradyJ, O'Leary N. Interference due to hemolysis inroutine photometric analysisa survey. Ann Clin Biochem 1998;35:12834.

[2] Glick MR, Ryder KW, Glick SJ, et al. Unreliable visual estimation of the incidenceand amount of turbidity, hemolysis, and icterus in serum from hospitalizedpatients. Clin Chem 1989;35:8379.

[3] Manufacturer's insert sheetAbbott Laboratories 8G62.[4] Cohen RS, Wong RJ, Stevenson DK. Understanding neonatal jaundice a perspective

on causation. Pediatr neonatol 2010;51:1438.[5] Shapiro SM. Chronic bilirubin encephalopathy: diagnosis and outcome. Semin

Fetal Neonatal Med 2010;15:15763.[6] Kirk JM. Neonatal jaundice: a critical review of the role and practice of bilirubin

analysis. Ann Clin Biochem 2008;45:45262.[7] http://www.westgard.com/europe.htm.[8] Bhutani VK, Gourley GR, Adler S, Kreamer B, Dalin C, Johnson LH. Noninvasive

measurement of total serum bilirubin in a multiracial predischarge newbornpopulation to assess the risk of severe hyperbilirubinemia. Pediatrics 2000;106:e17.

[9] Algeciras-Schimnich A, Cook WJ, Milz TC, Saenger AK, Karon BS. Evaluation ofhemoglobin interference in capillary heel-stick samples collected for determina-tion of neonatal bilirubin. Clin Biochem 2007;40:13116.

[10] Ahmed M, Mostafa S, Fisher G, Reynolds TM. Comparison between transcutaneousbilirubinometry and total serum bilirubin measurements in preterm infantsb35 weeks gestation. Ann Clin Biochem 2010;47:727.

Table 2

Effect of addition of saline or hemolysed sera (A, B, C, D, E, F and G) over bilirubin

measurement. Uncorrected Bil is Bilmeasof saline-added samples, and corrected Bil is

Bilcalcof saline-added samples.

Sample Addition

(10%) of

Bilmeas(mg/dL)

Hmeas

(mg/dL of Hb)

Calculated

interference

Bilcalc(mg/dL)

Bilindex(mg/dL)

#1 Saline 0.9 8 9% 1.0 1.5

A 0.6 47 50% 1.2 1.5

B 0.3 85 80% 1.5 1.5

C b0.1 170 N. a. N. a. N. a.#2 Saline 2.9 3 1% 2.9 2.6

A 2.4 42 20% 3.0 2.7

B 1.9 88 44% 3.4 2.6

C 1.1 168 66% 3.2 2.6

#3 Saline 6.7 12 3% 6.9 5.8

B 5.4 94 21% 6.9 5.8

C 4.4 173 39% 7.2 5.8

E 2.7 311 61% 6.9 5.7

#4 Saline 15.7 66 7% 16.9 15.7

D 13.5 223 23% 17.5 15.6

F 10.8 420 38% 17.6 15.4

G 7.7 699 51% 15.7 15.3

Fig. 4.Bias and precision plot from the data of the verication samples.

777P. Brunori et al. / Clinica Chimica Acta 412 (2011) 774777
http://localhost/var/www/apps/conversion/tmp/scratch_2/image%20of%20Fig.%E0%B4%80

Clinica Chimica Acta Volume 412 Issue 9-10 2011 [Doi 10.1016%2Fj.cca.2011.01.010] Paola Brunori;...

Documents

Transcript of Clinica Chimica Acta Volume 412 Issue 9-10 2011 [Doi 10.1016%2Fj.cca.2011.01.010] Paola Brunori;...