See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/316858790
Identification of a Hemolysis Threshold That Increases Plasma and Serum
Zinc Concentration
Article  in  Journal of Nutrition · May 2017
DOI: 10.3945/jn.116.247171
CITATIONS READS
4 578
7 authors, including:
David W Killilea Fabian Rohner
Children's Hospital Oakland Research Institute GroundWork
87 PUBLICATIONS   1,646 CITATIONS    54 PUBLICATIONS   1,655 CITATIONS   
SEE PROFILE SEE PROFILE
Shibani Ghosh Gloria E. Otoo
Tufts University University of Ghana
56 PUBLICATIONS   401 CITATIONS    29 PUBLICATIONS   194 CITATIONS   
SEE PROFILE SEE PROFILE
Some of the authors of this publication are also working on these related projects:
PreNAPs: Prenatal Nutrition and Psycho-social Health Outcomes Study View project
The Nutrition of Whole Grains - A Wealth of Health View project
All content following this page was uploaded by David W Killilea on 11 May 2017.
The user has requested enhancement of the downloaded file.
The Journal of Nutrition. First published ahead of print May 10, 2017 as doi: 10.3945/jn.116.247171. 
The Journal of Nutrition
Methodology and Mathematical Modeling
Identification of a Hemolysis Threshold
That Increases Plasma and Serum Zinc
Concentration1–3
David W Killilea,4* Fabian Rohner,5 Shibani Ghosh,6 Gloria E Otoo,7 Lauren Smith,4
Jonathan H Siekmann,4 and Janet C King4
4Children
s Hospital Oakland Research Institute, Oakland, CA; 5GroundWork, Fläsch, Switzerland; 6Friedman School of Nutrition
Science and Policy, Tufts University, Boston, MA; and 7Department of Nutrition and Food Sciences, University of Ghana, Accra, Ghana
Abstract
Background: Plasma or serum zinc concentration (PZC or SZC) is the primary measure of zinc status, but accurate
sampling requires controlling for hemolysis to prevent leakage of zinc from erythrocytes. It is not established how much
hemolysis can occur without changing PZC/SZC concentrations.
Objective: This study determines a guideline for the level of hemolysis that can significantly elevate PZC/SZC.
Methods: The effect of hemolysis on PZC/SZC was estimated by using standard hematologic variables and mineral
content. The calculated hemolysis threshold was then compared with results from an in vitro study and a population
survey. Hemolysis was assessed by hemoglobin and iron concentrations, direct spectrophotometry, and visual
assessment of the plasma or serum. Zinc and iron concentrations were determined by inductively coupled plasma
spectrometry.
Results: A 5% increase in PZC/SZC was calculated to result from the lysis of 1.15% of the erythrocytes in whole blood,
corresponding to ;1 g hemoglobin/L added into the plasma or serum. Similarly, the addition of simulated hemolysate to
control plasma in vitro caused a 5% increase in PZCwhen hemoglobin concentrations reached 1.186 0.10 g/L. In addition,
serum samples from a population nutritional survey were scored for hemolysis and analyzed for changes in SZC; samples
with hemolysis in the range of 1–2.5 g hemoglobin/L showed an estimated increase in SZC of 6% compared with
nonhemolyzed samples. Each approach indicated that a 5% increase in PZC/SZC occurs at ;1 g hemoglobin/L in plasma
or serum. This concentration of hemoglobin can be readily identified directly by chemical hemoglobin assays or indirectly
by direct spectrophotometry or matching to a color scale.
Conclusions: A threshold of 1 g hemoglobin/L is recommended for PZC/SZC measurements to avoid increases in zinc
caused by hemolysis. The use of this threshold may improve zinc assessment for monitoring zinc status and nutritional
interventions. J Nutr doi: 10.3945/jn.116.247171
Keywords: hemoglobin, hemolysis, human nutrition, mineral, plasma zinc, serum zinc
Introduction
estimated 1.2 billion individuals are at risk of zinc deficiency on
Zinc is an essential nutrient for normal growth, healthy the basis of national food balance sheets, with the highest
pregnancy, and robust immune function. Inadequate zinc leads prevalence in children <5 y of age (3, 4). To evaluate the zinc
to stunting, developmental delays, hypogonadism, susceptibility status of a population, nutritional surveys collect data on plasma
to infections, and cognitive dysfunction (1–3). Globally, an zinc concentration (PZC)8 or serum zinc concentration (SZC)
along with dietary information. Interestingly, these studies have
1 Supported in part by the Bill and Melinda Gates Foundation (JCK and DWK) and indicated that the prevalence of low circulating zinc is nearly
Ajinomoto Co., Inc., and the Japan International Cooperation Agency (SG and twice as high as the previous estimates of zinc deficiency from
GEO). This is an open access article distributed under the CC-BY license dietary or stunting data alone (5–7). Thus, the current estimates
(http://creativecommons.org/licenses/by/3.0/).
2 Author disclosures: DW Killilea, F Rohner, S Ghosh, GE Otoo, L Smith, JH
Siekmann, and JC King, no conflicts of interest.
3 Supplemental Table 1 and Supplemental Figure 1 are available from the ‘‘Online 8 Abbreviations used: AAS, atomic absorption spectrometry; ICP-OES, induc-
Supporting Material’’ link in the online posting of the article and from the same tively coupled plasma optical emission spectrometry; IRB, institutional review
link in the online table of contents at http://jn.nutrition.org. board; PZC, plasma zinc concentration; RGB, red, green, blue; SZC, serum zinc
*To whom correspondence should be addressed. E-mail: dkillilea@chori.org. concentration.
Manuscript received December 30, 2016. Initial review completed January 19, 2017. Revision accepted March 28, 2017. 1 of 8
doi: 10.3945/jn.116.247171
Copyright (C) 2017 by the American Society for Nutrition 
Downloaded from jn.nutrition.org by guest on May 11, 2017
of individuals at risk of zinc deficiency could be considerably In vitro hemolysis study. Blood samples from 14 healthy participants
underestimated. The accurate assessment of zinc status bio- (Supplemental Table 1) were purchased from a commercial laboratory
markers such as PZC and SZC is critical for understanding the (AllCells LLC). The laboratory prescreened donors for lack of infectious
true burden of zinc deficiency and evaluating the response to zinc agents (e.g., HIV, hepatitis B virus, and hepatitis C virus), chronic illness,
and medication or supplement use within the previous 2 wk. Venous
nutrition intervention programs.
whole blood was collected in trace metal–certified Vacutainers containing
Measuring PZC and SZC requires the acquisition of blood by
lithium heparin as anticoagulant (BD), inverted several times, and stored
venipuncture or finger-stick, followed by rapid and careful at room temperature until processing within 2 h. Whole blood was
processing to isolate plasma or serum for later analysis. With centrifuged at 600 3 g for 15 min at room temperature followed by
any study, there is a potential for variation in the processing and immediate removal of the plasma. The plasma was transferred to
handling of blood samples, but especially when studies are large, polypropylene microtubes (tested free of metal contamination) and
multicenter, or depend on field sites with limited resources. further centrifuged at 30003 g for 5 min at room temperature to remove
Suboptimal blood collection and sample handling can increase any remaining cells and debris. The clarified plasma was then transferred
the likelihood of damage to erythrocytes, resulting in some degree to new polypropylene microtubes and stored at 280
C. For concen-
of hemolysis. Hemolysis has the potential to measurably increase trated hemolysate, erythrocyte-rich plasma was allowed to gradually
PZC and SZC concentrations because the zinc concentration in hemolyze over 4 mo at 4
C to reduce the formation of precipitate or
insoluble material that could affect mineral concentration. To generate
erythrocytes is ;10–20 times that of plasma or serum (8–10).
graded hemolysis, defined amounts of concentrated hemolysate were
Hemolysis is widely recognized as a potential source of
added to nonhemolyzed plasma and mixed to uniformity at 6 different
contamination or interference for a variety of hematologic ratios of 0%, 0.1%, 0.25%, 1%, 2.5%, and 10% concentrated
variables, which has been addressed in numerous publications hemolysate volume into control nonhemolyzed plasma volume. Ten
(11–14). It is also acknowledged that hemolysis could lead to experiments were completed from the participant samples for a total of
spurious increases in PZC and SZC (6, 15–19), although there is 60 independent points. Graded hemolysis samples were then analyzed
little information as to what level of hemolysis creates a concern. for hemoglobin and mineral concentrations. The commercial laboratory
Most studies that reported PZC or SZC have limited or no also provided basic anthropometric and hematologic donor information,
detailed description of how hemolysis was categorized. There- but no protected health information, so institutional review board (IRB)
fore, the International Zinc Nutrition Consultative Group and registration was not required.
the Biomarkers of Nutrition for Development Zinc Expert Panel
Population nutritional survey. Briefly, whole bloodwas collected from
have simply stated that obviously hemolyzed samples should be
909 infants at 6–18 mo of age living in Ghana for the 2012–2013 Trial
discarded when measuring PZC and SZC (6). More work is for Reducing Undernutrition through Modified Feeding study (21).
needed to develop evidence-based recommendations on the level Blood was collected in trace metal–certified Vacutainers without
of hemolysis that alters PZC and SZC. This article recommends a anticoagulant (BD) at multiple locations, transported back to a central
threshold of hemolysis for zinc analysis and compares options for storage facility, separated into serum samples, and stored at 280
C. In
assessing the degree of hemolysis. It is important to note that 2016, serum samples were shipped in batch to the Children
s Hospital
although plasma and serum have important physiologic differ- Oakland Research Institute for the determination of SZC. At the
ences, there does not appear to be a meaningful difference in zinc Children
s Hospital Oakland Research Institute, serum samples were
concentration between the 2 (6); consequently, the clinical thawed, mixed on a vortex for 5 s, and prepared for mineral analysis as
reference ranges for PZC/SZC are the same (20). This article described below. Before proceeding with elemental analysis, each sample
therefore treats PZC and SZC as interchangeable from the was compared visually to a hemolysis color scale by the same trained
technicians and assigned to 1 of 6 hemolysis levels from 0 to 10 g
perspective of impact of hemolysis.
hemoglobin/L. Additional description and results from this population-
based study will be described in a future study (S Ghosh, unpublished
data, 2017). The study was monitored by the host institute IRB; no
Methods protected health information was communicated for elemental and
hemolysis analysis, so additional IRB registration was not required.
Calculations for hemolysis estimates. The effects of hemolysis levels
on PZC and SZC were estimated by using standard hematologic and Mineral analysis. Iron and zinc concentrations were determined by
mineral concentration parameters obtained from clinical reference inductively coupled plasma optimal emission spectrometry (ICP-OES) as
ranges, specifically PZC and SZC (0.7–1.2 mg/L), plasma and serum previously described (22, 23). For PZC and SZC, 50- to 100-mL samples
hemoglobin concentration (<0.1mg/L), erythrocyte cell concentration (3.8– were removed from thawed stocks and processed for mineral analysis.
5.73 1012 cells/L), erythrocyte cell volume (80–100 fL/cell), erythrocyte The ICP-OES was calibrated by using National Institute of Standards
cell hemoglobin content (27–31 pg/cell), and erythrocyte cell zinc and Technology traceable elemental standards (Sigma-Aldrich) and
concentration (10–16 mg/L cell volume) (20). The degree of hemolysis validated by using Seronorm Trace Element Levels 1 and 2 reference
needed to increase PZC and SZC values was calculated by using the material (Sero). The detection range for both iron and zinc was between
equation [(A 3 B) O (C 3 D) O E] 3 100 = F, where A is the average 0.005 and 5.000 mg/L and the CV for interassay precision was <5%.
PZC and SZC of the study or population, B is the measure of precision Cesium (50 mg/L) was used for ionization suppression, and yttrium
beyond which zinc values are considered increased over the mean PZC (5 mg/L) was used as an internal standard for all samples. All associated
and SZC, C is the average erythrocyte zinc concentration of the study or reagents and plasticware were certified as trace metal free or tested for
population, D is the average erythrocyte volume (also known as mean trace metal contamination. Iron and zinc concentrations were normal-
corpuscular volume) of the study or population, and E is the average ized per plasma or serum volume.
erythrocyte concentration (also known as RBC count) of the study or
population. The product (F) of this equation is the percentage of erythrocyte Determination of hemoglobin concentration. Hemolysis is com-
lysis required to yield the desired increase in PZC and SZC. The concentration monly quantified by the corresponding amount of hemoglobin released
of hemoglobin that corresponds to the determined degree of hemolysis was into the plasma or serum. The concentrations of hemoglobin were
calculated according to the equationGO [(A3 B)O (C3 D)] =H, where determined by using the following: 1) direct chemical detection of
A–D represent the same values as above andG is the erythrocyte hemoglobin hemoglobin with the use of Drabkin
s assay according to standard
concentration (also known as mean corpuscular hemoglobin content). The protocols (24); 2) indirect chemical detection of hemoglobin with the use
product (H) of this equation is the concentration of hemoglobin that of iron concentration calculated from standard molar ratios (4 mol Fe/1
indicates a significant increase in PZC and SZC. mol hemoglobin); 3) spectroscopic estimation of hemoglobin by directly
2 of 8 Killilea et al.
Downloaded from jn.nutrition.org by guest on May 11, 2017
measuring absorbance of plasma or serum at 540 nm, a major 3. Determine the level of hemolysis that equates to 4.375 3
absorbance peak for hemoglobin; or 4) visual inspection of plasma or 1010 erythrocytes/L. For an average erythrocyte concen-
serumwith the use of a hemolysis color scale generated by adding defined tration of 3.8 3 1012 cells/L, 4.375 3 1010 cells/L
amounts of hemoglobin from a human hemoglobin standard (Point represents 1.15% of the erythrocyte concentration.
Scientific) to fresh plasma or serum containing no apparent hemolysis
4. Determine the concentration of hemoglobin that equates
(hemoglobin <0.2 g/L and no orange-red tint). For spectroscopic
to 1.15% of erythrocytes. For an average erythrocyte
estimation, 100 mL plasma or serum was pipetted into 96-well flat-
bottom polystyrene microplates and measured on a conventional hemoglobin concentration of 27 pg/cell and an erythro-
10
spectrophotometer with path-length correction (Synergy H1; Biotek cyte concentration of 4.375 3 10 cells/L, 1.12 g
Instruments); background absorbance of empty wells was ;0.05 hemoglobin/L would be released into the plasma or
absorbance units. For visual inspection, 3 independent color scales serum. Because normal plasma or serum concentrations of
were generated in optically clear plastic cuvettes and imaged with a hemoglobin are negligible, then 1.12 g hemoglobin/L
digital camera. Color was determined by using Adobe Photoshop represents the actual hemoglobin that would be measured
Elements color picker tool (version 13.1) with all files converted to by lysis of 1.15% of erythrocytes.
Adobe RGB color profile (with a gamut better matched to human visual
perception) on a computer monitor with a calibrated color spectrum.
Effects of hemolysis on PZC in vitro. To empirically
Color values were determined by using the color picker tool to aggregate
determine the level of hemolysis that would increase PZC by
;50% of the color image with the use of an 8-bit red, green, blue (RGB)
color model. RGB values can be converted to other color models for 5%, concentrated hemolysate was added to nonhemolyzed
printing by using printers with appropriate color-monitoring tools. plasma followed by hemoglobin and mineral analysis. Ratios of
hemolyzed to nonhemoylzed plasma were chosen both within
Statistical analysis. Graphing, regression, and statistical analysis were and exceeding the clinical reference range for iron (0.5–
conducted by using Prism software, version 6 (GraphPad Software, Inc.). 1.75 mg/L) and zinc (0.7–1.2 mg/L) concentrations in plasma
For comparisons of 1) in vitro plasma mineral and hemoglobin or serum (20). Iron concentration increased with the addition of
concentrations, 2) in vitro plasma zinc and plasma iron concentrations, hemolysate, reaching nearly 3000% of the control value as
and 3) population serum zinc and zinc and serum iron concentrations, hemoglobin concentrations approached 10 g/L (Figure 1A). The
linear regression was used. For comparison of population serum mineral
data fit a linear function y = 303.9 (x) + 105.4 with an R2 of
concentrations with hemolysis range, 1 data point was removed from
0.918. In contrast, the zinc concentration did not consis-
both the iron and zinc values because they were >3 SDs from the
hemolysis range groupmean, which left 908 values for each mineral. The tently change at lower levels of added hemolysate, but in-
mean serum iron and zinc concentrations for each hemolysis range were creased modestly only after the hemolysate level reached 1 g
compared by using a 1-factor ANOVA with Dunnett
s multiple- hemoglobin/L in the plasma (Figure 1B). The data fit a linear
2
comparison ad hoc test for differences compared with the control function y = 5.233 (x) + 98.80 with an R of 0.698. The best-fit
(nonhemolyzed) group. Serum mineral concentration data did not pass line for the zinc response crossed the level of 5% over baseline
the test for normal distribution (D
Agostino and Pearson omnibus test, at a hemoglobin concentration of 1.18 6 0.10 g/L, similar to
a = 0.05), but sample sizes were large enough for the ANOVA model to findings from the calculated values. However, the correlation
be robust. For all of the tests, significance was accepted at P < 0.05. between iron and zinc values within the same sample was weak;
the linear function fit the equation y = 17.44 (x) 2 11.86 with
an R2 of 0.274 (Figure 1C).
Results
Effects of hemolysis on SZC in a population study. To test
Effects of hemolysis on PZC or SZC by calculation. The
the hemolysis threshold of 1 g hemoglobin/L, the correlation
degree of hemolysis required to increase PZC and SZC was
between apparent hemolysis level and SZC was examined in
estimated by using values within the standard hematologic and
samples obtained in a large population study conducted in
mineral concentration reference ranges. To model these calcula-
Ghana. SZCs ranged from 0.21 to 1.44 mg/L, with a mean6 SD
tions for a population in whom nutrient deficiencies are common,
of 0.616 0.13 mg/L (Figure 2A). SZCs were low compared with
lower values for each parameter were used. In addition, the
the reference ranges in the United States, but this was expected
analytical techniques that measure PZC and SZC, including atomic
because 15–25% of the population in Ghana is estimated to be
absorption spectrometry (AAS) or ICP-OES, typically have a
at risk of zinc inadequacy (4). Mineral analysis was conducted
precision of ;5% CV (25). Therefore, the target amount of
by ICP-OES, so serum iron concentrations were also deter-
hemolysis required to detect an increase in PZC and SZC was
mined. Serum iron concentrations ranged from 0.36 to 72 mg/L,
selected as 5%. In practice, the population values or desired precision
with a mean 6 SD of 3.92 6 3.76 mg/L (Figure 2A). Many
may vary, so the parameters can be easily adjusted as needed.
(;40%) samples exceeded the clinical reference range for serum
The calculations to determine a threshold level of hemolysis
iron concentration and had a visible tint ranging from light
use the equations described in Methods on the basis of the
orange to dark red, suggesting elevated hemolysis levels in those
following descriptions of each step:
samples. When the mineral values were plotted together, the
1. Determine the amount of zinc needed to increase PZC and data fit a linear function y = 6.087 (x) + 0.1793 with an R2 of
SZC by 5%. For an average PZC and SZC of 0.7 mg/L, an 0.104, indicating a weak correlation between serum iron and
additional 0.035 mg Zn/L would be released into the zinc values.
plasma or serum. The samples were then divided into 6 groups on the basis of
2. Determine the fraction of erythrocyte zinc that would apparent hemolysis level: 0–0.2, 0.2–0.5, 0.5–1, 1–2.5, 2.5–5,
increase PZC and SZC by 0.035 mg/L. For an average and 5–10 g hemoglobin/L. Compared with the 0–0.2-g/L
erythrocyte zinc concentration of 10 mg/L and erythro- (baseline) group, mean serum iron concentrations in the other
cyte volume of 80 fL/cells, the erythrocyte zinc content groups increased markedly from 25% to 842%, which is consis-
would be 0.8 fg/cell. To release 0.035 mg Zn into the tent with increasing hemolysis (Figure 2B). Zinc concentrations
plasma or serum, the concentration of lysed erythrocytes were then assessed to determine which hemolysis group would
would be 4.375 3 1010 cells/L. reach the target of 5% above baseline SZC. The 0.2–0.5- and
Hemolysis threshold for plasma and serum zinc 3 of 8
Downloaded from jn.nutrition.org by guest on May 11, 2017
FIGURE 1 Effects of hemolysis on plasma iron and
zinc concentrations in vitro. The trace mineral concen-
tration increased as a function of hemolysis as indicated
by the hemoglobin concentration. (A) Plasma iron
concentrations increased with the addition of hemoly-
sate to nonhemolyzed plasma. Values are means 6
SDs, n = 60. Data were fit to a linear function (solid line,
R2 = 0.918) with a 95% confidence band (dashed line)
also shown. The inset shows data on a log10 scale. (B)
Plasma zinc concentrations increased with the addition of
hemolysate to nonhemolyzed plasma at;1 g hemoglobin/L.
Values are means 6 SDs, n = 60. Data were fit to a
linear function (solid line, R2 = 0.698) with a 95%
confidence band (dashed line) also shown. The inset
shows data on a log10 scale. (C) Iron and zinc concen-
trations were plotted for each plasma sample. Gray
shading indicates the clinical reference range for plasma
iron and zinc concentration as indicated (20). Values are
means 6 SDs, n = 60. Data were fit to a linear function
(solid line, R2 = 0.274).
0.5–1-g/L hemolysis groups had the same mean SZC as the consistent with the trend of elevated SZC in the groups with
baseline group, whereas the 2.5–5- and 5–10-g/L hemolysis greater hemolysis and was consistent with a 5% increase in zinc
groups had mean SZCs 12% and 28% higher than baseline, concentrations from 1 g hemoglobin/L hemolysates as deter-
respectively (Figure 2C). The 1–2.5-g/L hemolysis group showed mined above. It is important to note that this analysis is based on
the closest match to the 5% target, with a mean SZC of 6% over the assumption that mean SZC would be the same between all
baseline, although this did not reach significance. However, the groups without the addition of hemolysis, which was not
measured 6% increase in mean SZC for the 1–2.5-g/L group is possible to test independently.
4 of 8 Killilea et al.
Downloaded from jn.nutrition.org by guest on May 11, 2017
FIGURE 2 Effects of hemolysis on serum iron and zinc
concentrations in a population nutritional survey. (A) Iron
and zinc concentrations were plotted for each serum
sample. Gray shading indicates the clinical reference
range for serum iron and zinc concentrations as indi-
cated (20). Data were fit to a linear function (solid line,
R2 = 0.104). (B) Serum iron concentrations increased as a
function of hemolysis range grouping on the basis of
hemoglobin concentration. Values are means 6 SDs,
n = 46–289. (C) Serum zinc concentrations increased as a
function of hemolysis range grouping on the basis of
hemoglobin concentration. Values are means 6 SDs,
n = 46–288. *P , 0.05. Hb, hemoglobin.
Evaluation of hemoglobin concentrations to assess he- the degree of hemolysis (Figure 3). The apparent color of 1 g
molysis in plasma or serum. Several approaches were used to hemoglobin/L in plasma or serum is distinctly orange-red, which
determine hemoglobin concentrations in plasma or serum. is similar to an RGB color model value of 249, 125, 63. If greater
Chemical detection of hemoglobin has a typical minimum precision is needed, direct measurement of the absorption of
detection limit of ;0.01 g/L (24), but these assays are often not plasma or serum samples was found to be sufficient to detect 1 g
convenient for large, multicenter, and/or field studies. One hemoglobin/L, without chemical modification of the constituent
alternative approach is the use of a simple color scale to estimate hemoglobin (Supplemental Figure 1). Hemoglobin has several
Hemolysis threshold for plasma and serum zinc 5 of 8
Downloaded from jn.nutrition.org by guest on May 11, 2017
FIGURE 3 Methods to determine hemoly-
sis levels in plasma or serum. A representa-
tion of plasma or serum within increasing
levels of hemolysis is shown. Below the
images are the corresponding Hb concentra-
tions (means 6 SDs; n = 3) and approximate
RGB color model values of the corresponding
plasma or serum. The iron values listed for
the lowest category of hemolysis reflect the
clinical reference range in nonhemolyzed
plasma or serum (0.5–1.8 mg/L) (20).
A540nm, absorbance at 540 nm; Hb, hemoglo-
bin; RGB, red, green, blue; ,DL, below the
detection limit of 0.05 absorbance.
major absorption peaks with varying sensitivity. However, details as to how samples with observable hemolysis were
plasma or serum had a significant absorbance at the smaller evaluated. Other studies have stated that hemolyzed samples
wavelength peaks, so the absorbance peak at 540 nm is were recorded and/or discarded, but do not indicate the level of
recommended. The absorbance value for 1 g hemoglobin/L at hemolysis used for threshold. Only a few studies, to our knowl-
540 nm was 0.16 6 0.04. edge, directly investigated the effect of hemolysis on PZC/SZC, but
the methodologies varied and often used descriptive analysis of
hemolysis (e.g., low, moderate, and extensive) without specific
Discussion quantitative values (27, 28). Lofberg and Levrl (15) reported no
significant change in SZC in a small study of ‘‘hemolyzed’’
Hemolysis can affect the constituent analysis of serum or compared with ‘‘unhemolyzed’’ samples, but estimated that a ‘‘1–
plasma. For zinc measurement, this can be a concern because the 2% hemolysis’’ could result in increased zinc. Strand et al. (27)
zinc concentration in erythrocytes is 10–20 times that of plasma measured the changes in zinc from hemolyzed samples but with the
or serum (8, 10, 20). Many studies for measuring PZC and SZC use of qualitative categorization (mild, moderate, or extensive)
simply warn the investigator to avoid hemolysis, without measured by trained technicians. However, the understanding of
providing specific quantitative metrics. Underestimating the these categories may differ between laboratories.
impact of hemolysis can result in the inclusion of samples Measuring hemoglobin concentrations in plasma or serum is
containing plasma or serum contaminated with inflated amounts arguably the most convenient way to determine the level of
of zinc, which adds noise to the analysis of zinc status in a hemolysis in a blood sample, because the normal concentration
population. Overestimating the impact of hemolysis can result in of hemoglobin in plasma or serum is negligible compared with
the rejection of samples that are perfectly acceptable for zinc the amount of hemoglobin released during even mild hemolysis.
measurement, and thus losing study power and introducing There are several ways to measure hemoglobin. The classic
sampling bias. Having a clear threshold of hemolysis for chemical method involves the transformation to methemoglobin
measuring PZC and SZC is important for evaluating zinc status followed by reaction with alkaline potassium cyanide (24).
and response to zinc nutrition interventions. Improved versions of the chemical method for detecting hemo-
In this study, we approached this problem in 3 different ways: globin are now available that avoid the use of toxic cyanide
1) theoretical calculation by using standard hematologic values compounds and extend the sensitivity of the assay (29). In
and mineral concentrations, 2) controlled addition of hemolysate addition, point-of-care devices that directly measure hemoglo-
in vitro, and 3) correlation of zinc concentration to hemolysis bin in small blood samples are available and now commonly
levels in samples from a large population study. All 3 approaches used in field studies (30). Strengths of the chemical and device-
indicated a similar value of ;1 g hemoglobin/L as the level of based measurements include a straightforward protocol and
hemolysis that increased PZC and SZC by;5%. This target level high sensitivity. Weaknesses of this approach include the
of increased zinc was chosen because the analytical techniques additional time and procedural steps, increased costs, and use of
that measure PZC and SZC, such as AAS or ICP-OES, typically toxic chemicals for the chemical assays; these complications are
yield a CV of 5% (25). Therefore, a 5% increase in PZC or SZC particularly inconvenient for large, multicenter, and/or field
would be a minimum value to reliably detect changes in PZC and studies. In addition, the high sensitivity of these assays is not
SZC. However, it would be straightforward to adjust these necessary to measure the relatively high threshold level of 1 g
calculations if a different percentage increase in PZC and SZC hemoglobin/L.
was preferred. For example, the WHO estimates that >1 billion Therefore, alternative ways to assess hemoglobin concentra-
women and children in developing countries suffer from iron tion were investigated that might be more amenable for large
deficiency and anemia (26); therefore, the concentrations of and/or field studies of PZC and SZC. Because hemoglobin has
erythrocyte hemoglobin and other hematologic values in these significant absorbance within the visual spectrum, hemoglobin
areas are likely to be even lower than the example values used in in plasma or serum can be detected by direct spectrophotometry
this analysis. The degree of hemolysis can be determined by using without any chemical modifications. This method is less
the same calculations as above, replacing the specific hematologic sensitive than the chemical methods, but is easily able to detect
and mineral concentration parameters for the desired study hemoglobin at 1 g/L. Strengths of this approach include minimal
population. additional costs, a simple protocol, and a nondestructive method
Several reports measuring PZC and SZC mentioned the that allows the plasma or serum to be used for other measure-
potential confounding effects of hemolysis, yet do not provide ments. Weaknesses of this approach include additional time and
6 of 8 Killilea et al.
Downloaded from jn.nutrition.org by guest on May 11, 2017
procedural steps and inability to resolve color variations in by hemolysis. By using similar approaches to this study, appro-
plasma unrelated to hemolysis, such as infection or high priate thresholds could be reached for all of these nutrients or
concentrations of bilirubin, carotenoids, or ceruloplasmin (31). biomarkers.
Another proposed way to assess hemoglobin concentration is
to take advantage of the constitutive iron within the hemoglobin Acknowledgments
protein. If technologies such as ICP-OES are used to measure We thank Tatiana Cheong, Darryl Chow, Wesley Kwong, and
PZC and SZC, then iron concentration can be measured in the Kathleen Schultz for technical assistance and Kenneth Brown,
same analytical run. The additional iron added into plasma Christine McDonald, and Bradley A Woodruff for helpful
or serum by 1 g hemoglobin/L is substantially greater than comments on this study. The authors
 responsibilities were as
endogenous concentrations, and thus might be useful to identify follows—DWK: designed the research and had primary respon-
samples that exceed the hemolysis threshold. Strengths of this sibility for the final content; DWK, SG, and GEO: conducted
approach include that no additional time, procedural steps, or the research; DWK and FR: analyzed the data; DWK, FR, LS,
costs are needed beyond what is needed to determine PZC and JHS, and JCK: wrote the manuscript; and all authors: read and
SZC. Weaknesses of this approach include the inability to approved the final manuscript.
distinguish between hemolysis and elevated iron resulting from
iron supplementation or contamination of samples. In addition,
ICP-OES and similar technologies are not always available to
researchers; other common technologies such as AAS can also References
measure iron but require separate analyses with additional 1. Prasad AS. Discovery of human zinc deficiency and studies in an ex-
sample material. Furthermore, we found that in the hemolysis perimental human model. Am J Clin Nutr 1991;53:403–12.
analysis in both the in vitro plasma (Figure 1C) and the 2. Shankar AH, Prasad AS. Zinc and immune function: the biological basis
population serum (Figure 2A) samples, there was only a weak of altered resistance to infection. Am J Clin Nutr 1998;68:447S–63S.
correlation between iron and zinc concentrations in the plasma 3. Brown KH, Rivera JA, Bhutta Z, Gibson RS, King JC, Lonnerdal B,
Ruel MT, Sandtrom B, Wasantwisut E, Hotz C; International Zinc
or serum. Therefore, we do not recommend iron concentrations Nutrition Consultative Group. International Zinc Nutrition Consulta-
alone as an approach to screen plasma or serum samples for PZC tive Group (IZiNCG) technical document #1: assessment of the risk of
and SZC analysis. zinc deficiency in populations and options for its control. Food Nutr
The final way to estimate hemoglobin concentration is to Bull 2004;25:S99–203.
visually compare the plasma or serum samples to a hemolysis 4. Wessells KR, Brown KH. Estimating the global prevalence of zinc de-
ficiency: results based on zinc availability in national food supplies and
color scale, based on the fact that hemoglobin is strongly
the prevalence of stunting. PLoS One 2012;7:e50568.
colored. This approach has been reported in other publications
5. Prasad AS. Discovery of human zinc deficiency: 50 years later. J Trace
(11, 14). Strengths of this approach include that it has no Elem Med Biol 2012;26:66–9.
additional procedural steps or costs. This technique is easy to 6. King JC, Brown KH, Gibson RS, Krebs NF, Lowe NM, Siekmann JH,
use, even for large-population studies, and simple to teach the Raiten DJ. Biomarkers of Nutrition for Development (BOND)—zinc
technical staff. Weaknesses of this approach include the need for review. J Nutr 2016;146(Suppl):858S–85S.
some additional time and the lack of quantitative hemoglobin 7. Hess SY. National risk of zinc deficiency as estimated by national sur-
measurement. However, we found that, once trained, technical veys. Food Nutr Bull 2017 Jan 1 (Epub ahead of print; DOI: 10.1177/
0379572116689000).
staff are able to easily and rapidly identify plasma or serum
8. Vallee BL, Gibson JG. The zinc content of normal human whole blood,
samples that exceed the color threshold. To facilitate this plasma, leucocytes, and erythrocytes. J Biol Chem 1948;176:445–57.
approach, we created a simple card containing a hemolysis index 9. Solomons NW. On the assessment of zinc and copper nutriture in man.
with identification of our recommended threshold of hemolysis. Am J Clin Nutr 1979;32:856–71.
This card is available on the International Zinc Nutrition 10. Whitehouse RC, Prasad AS, Rabbani PI, Cossack ZT. Zinc in plasma,
Consultative Group website (www.izincg.org) and can be neutrophils, lymphocytes, and erythrocytes as determined by flameless
digitally displayed (e.g., on a smartphone or tablet) or printed, atomic absorption spectrophotometry. Clin Chem 1982;28:475–80.
assuming that color-management tools are used to correct for 11. Frank JJ, Bermes EW, Bickel MJ, Watkins BF. Effect of in vitro he-
molysis on chemical values for serum. Clin Chem 1978;24:1966–70.
variance in printer output.
12. Lippi G, Salvagno GL, Montagnana M, Brocco G, Guidi GC. Influence
Once the relation between hemolysis and PZC and/or SZC is
of hemolysis on routine clinical chemistry testing. Clin Chem Lab Med
clearly defined, there might be interest in a correction factor that 2006;44:311–6.
could remove the zinc concentrations originating from erythrocyte 13. Koseoglu M, Hur A, Atay A, Cuhadar S. Effects of hemolysis interfer-
leakage from PZC and SZC calculations. For example, this was a ences on routine biochemistry parameters. Biochem Med (Zagreb)
common approach for the measurement of potassium concentra- 2011;21:79–85.
tions in plasma or serum (32). However, we do not recommend 14. Howanitz PJ, Lehman CM, Jones BA, Meier FA, Horowitz GL. Prac-
this approach for PZC and SZC due to the substantial variance tices for identifying and rejecting hemolyzed specimens are highly var-
iable in clinical laboratories. Arch Pathol Lab Med 2015;139:1014–9.
found in erythrocyte zinc concentrations, hemoglobin concentra-
15. Lofberg RT, Levrl EA. Analysis of copper and zinc in hemolyzed serum
tions, and other factors. Recently, this approach has become samples. Anal Lett 1974;7:775–9.
increasingly criticized even for the determination of potassium, 16. Prasad AS. Laboratory diagnosis of zinc deficiency. J Am Coll Nutr
which tends to be less variable than PZC and SZC (33). 1985;4:591–8.
This study recommends a discrete threshold of hemolysis and 17. Smith JC, Holbrook JT, Danford DE. Analysis and evaluation of zinc
hemoglobin concentration beyond which the plasma or serum and copper in human plasma and serum. J Am Coll Nutr 1985;4:627–
concentrations of zinc could be compromised due to leaching of 38.
erythrocyte zinc stores. This conclusion was reached through a 18. Sauberlich HE. Laboratory tests for the assessment of nutritional status.
2nd ed. Boca Raton (FL): CRC Press; 1994.
combination of theoretical calculations, simulated hemolysis
19. Tamura T, Johnston KE, Freeberg LE, Perkins LL, Goldenberg RL.
testing in vitro, and analysis of samples from a large field study.
Refrigeration of blood samples prior to separation is essential for the
There are many other minerals, vitamins, and enzyme markers accurate determination of plasma or serum zinc concentrations. Biol
in which plasma or serum concentrations could be compromised Trace Elem Res 1994;41:165–73.
Hemolysis threshold for plasma and serum zinc 7 of 8
Downloaded from jn.nutrition.org by guest on May 11, 2017
20. Wu AHB. Tietz clinical guide to laboratory tests. 3rd ed. Philadelphia: 26. Pena-Rosas JP, Viteri FE. Effects of routine oral iron supplementation
WB Saunders; 1995. with or without folic acid for women during pregnancy. Cochrane
21. Ghosh S, Tano-Debrah K, Aaron GJ, Otoo G, Strutt N, Bomfeh K, Database Syst Rev 2006;(3):CD004736.
Kitamura S, Suri DJ, Murakami H, Furuta C, et al. Improving comple- 27. Strand TA, Adhikari RK, Chandyo RK, Sharma PR, Sommerfelt H.
mentary feeding in Ghana: reaching the vulnerable through innovative Predictors of plasma zinc concentrations in children with acute diar-
business—the case of KOKO Plus. Ann N YAcad Sci 2014;1331:76–89. rhea. Am J Clin Nutr 2004;79:451–6.
22. Engle-Stone R, Ndjebayi AO, Nankap M, Killilea DW, Brown KH. 28. Zhang-lin J, Yu-nan X, Fa-da Y. The effect analysis of hemolysis of the
Stunting prevalence, plasma zinc concentrations, and dietary zinc in- sample on biochemistry test results. J Trop Med (Guangzhou)
takes in a nationally representative sample suggest a high risk of zinc 2009;9:799–802.
deficiency among women and young children in Cameroon. J Nutr 29. Shah VB, Shah BS, Puranik GV. Evaluation of non cyanide methods for
2014;144:382–91. hemoglobin estimation. Indian J Pathol Microbiol 2011;54:764–8.
23. Zyba SJ, Shenvi SV, Killilea DW, Holland TC, Kim E, Moy A, 30. Sanchis-Gomar F, Cortell-Ballester J, Pareja-Galeano H, Banfi G,
Sutherland B, Gildengorin V, Shigenaga MK, King JC. A moderate in- Lippi G. Hemoglobin point-of-care testing: the HemoCue system. J Lab
crease in diet zinc reduced DNA strand breaks in leukocytes and altered Autom 2013;18:198–205.
plasma proteins without changing plasma zinc concentrations. Am J 31. Elkassabany NM, Meny GM, Doria RR, Marcucci C. Green plasma—
Clin Nutr 2016;105:343–51. revisited. Anesthesiology 2008;108:764–5.
24. Drabkin D, Austin J. Spectrophotometric studies. II. Preparations from 32. Jay DW, Provasek D. Characterization and mathematical correction of
washed blood cells; nitric oxide hemoglobin and sulfhemoglobin. J Biol hemolysis interference in selected Hitachi 717 assays. Clin Chem
Chem 1935;112:51–65. 1993;39:1804–10.
25. Tyler G. ICP-MS, or ICP-AES and AAS? A comparison. Melbourne 33. Mansour MM, Azzazy HM, Kazmierczak SC. Correction factors for
(Australia): Varian; 1994. [cited 2016 Dec 25]. Available from: http:// estimating potassium concentrations in samples with in vitro hemolysis:
concours.univ-lyon1.fr/servlet/com.univ.collaboratif.utils.LectureFichiergw? a detriment to patient safety. Arch Pathol Lab Med 2009;133:
ID_FICHIER=1320397709943. 960–6.
8 of 8 Killilea et al.
View publication stats
Downloaded from jn.nutrition.org by guest on May 11, 2017