Red Blood Cell Morphology in Sickle Cell Anemia as
Determined by Image Processing Analysis: The Relationship
to Painful Crises
MAXWELL P. WESTERMAN, M.D. AND JAMES W. BACUS, PH.D.
Red blood cell morphology was studied in the peripheral blood
of adults with sickle cell anemia to determine if changes occur
during painful crises. Image processing of the cells with an
automated system of red blood cell analysis was used. Four
groups of cells were observed: normocytes, macrocytes, target
cells, and cells with the shape of irreversibly sickled cells.
During asymptomatic periods, the percentages of these cells
differed in each individual but were typical for that individual
and generally were stable. During crises, macrocytosis occurred and the concentration of irreversibly sickled cells
showed greater fluctuation. The macrocytosis most likely reflected a marrow response to increased hemolysis and demonstrated that the increased red blood cell destruction observed
during pain crises may be more extensive than previously considered. Changes in the concentration of irreversibly sickled
cells during crises-were not consistent and could not be used
as an indicator of a crisis. Image processing with automated
red blood cell analysis allows for accurate assessment of all
the morphologic groups of red blood cells in patients with
sickle cell anemia and compares well with standard methods
for measuring the concentration of irreversibly sickled cells.
(Key words: Red blood cell morphology; Sickle cell anemia;
Image processing; Painful crises) Am J Clin Pathol 1983; 79:
667-672
PATIENTS WITH SICKLE CELL ANEMIA have numerous clinical problems, among which are recurrent
painful crises. The crises are most likely related to occlusion of small blood vessels by aggregated sickle red
blood cefls because of rheologic and morphologic changes
in the cells. Various abnormalities, such as changes in
the gelation time of hemoglobin, 10 ' 4 cellular dehydration and abnormalities in the cell membrane, 8 - 5 - 7 "' 9
irreversible sickling of cells,4 or increased adherence of
the cells to vascular endothelium 9 may contribute singly
or together to this process: The changes could be associated with alterations in red blood cell morphology and
might provide important clinical and pathogenetic information about crises.
To determine whether such changes do occur and to
evaluate their significance, we have examined the red
Received August 13, 1982; received revised manuscript and accepted for publication September 7, 1982.
Address reprint requests to Dr. Westerman: Hematology-Oncology
Unit, Mount Sinai Hospital, 15th Street and California Avenue, Chicago, Illinois 60608.
Department of Medicine and Medical Automation Research
Unit, Rush Medical College, Mount Sinai Hospital and
Presbyterian St. Luke's Medical Center, Chicago, Illinois
blood cells in peripheral blood smears of adults with
sickle cell anemia using an image processing system with
an automated method for analysis of the cells. The system identifies cells according to shape, size, and hemoglobin content and provides concentrations of the
different morphologic types of red blood cells. In the
present study we have (1) examined red blood cell morphology in the peripheral blood smears of patients during painful crises, and during asymptomatic periods (2)
compared the counts of irreversibly sickle shaped cells
obtained by image processing analysis with counts obtained by more traditional methods of examination.
Methods
Fifteen patients with sickle cell anemia as confirmed
by hemoglobin electrophoresis and by family study,
when indicated, were examined. The age range was from
21 to 45 years. The patients had no evidence by physical
examination of enlarged spleens. Studies were obtained
while patients were asymptomatic or during painful
crises that were sufficiently severe to require hospitalization. Representative clinical and laboratory data for
each patient are shown in Table 1.
Image processing analysis of red blood cells was done
as described previously.2 Whole blood samples were obtained in Vacutainer® EDTA tubes. Preparation of films
of cells was done by a centrifugal "spinner." This was
followed by image processing analysis which describes
morphologic characteristics of the cells and measures
hemoglobin absorption of the cells at 415 nM using a
system of high speed processing. Six measurements, consisting of (1) size, (2) hemoglobin content; (3) central
pallor, (4) shape (circularity), (5) shape (elongation), and
(6) shape (spicularity), were obtained with subsequent
separation of the cells into 14 possible classes, i.e., the
red blood cell differential count.' The classes are shown
in Table 2. For the purposes of this study, classes 7
0002-9173/83/0600/0667 $01.00 © American Society of Clinical Pathologists
667
I
A.J.C.P. • June 1983
WESTERMAN AND BACUS
668
Table 1. Representative Clinical and Laboratory Data and Red Blood Cell Differential Counts in Patients
with Sickle Cell Anemia during Asymptomatic Periods
RBC Differential:):
Patient
Age
(Years)
Sex
Hemoglobin
(g/dL)
Hematocrit
(%)
Reticulocyte
count (%)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
35
44
35
24
37
25
24
27
30
24
45
21
22
32
26
M
F
F
F
F
F
F
F
F
F
F
M
M
F
M
9.7
9.5
7.8
6.6
9.8
7.6
7.4
6.9
7.6
9.1
8.0
8.7
7.3
7.5
8.3
31.5
29.3
22.5
20.7
28.0
21.7
22.9
19.9
21.7
29.1
24.5
27.2
21.7
22.5
23.6
12.4
6.8
10.1
19.7
11.2
7.0
27.2
33.3
24.5
20.4
8.8
10.8
30.0
17.0
22.0
MCHC* MCVf rNormocytes
(%)
(fl)
' (%)
31.3
32.3
34.7
33.8
34.5
33.2
32.2
34.7
35.0
33.1
31.1
32.0
33.6
33.4
35.3
92
99
91
94
90
160
94
82
100
97
103
76
92
95
97
67.0
63.1
18.4
58.9
66.2
27.7
32.1
42.5
27.4
69.7
18.9
25.8
10.6
58.8
23.6
ISCs§ Target Cells
(%)
(%)
3.5
13.5
43.5
5.8
7.2
19.8
17.7
14.0
40.5
9.4
35.3
13.4
37.5
11.4
16.2
10.5
6.1
11.6
19.5
11.4
18.2
17.9
13.4
10.4
9.3
13.7
28.3
7.8
9.9
25.9
Macrocytes
(%)
15.0
12.3
19.6
7.9
13.3
23.7
18.3
21.6
16.7
7.6
"""'• 27.5
12.2
22.8
15.2
23.8
X Abbreviated red blood cell differential count as described in text.
§ Cells with irreversibly sickled shapes.
• Mean corpuscular hemoglobin concentration,
t Mean corpuscular volume.
(elongated cells) and 8 (irregular shaped cells) were morphologically similar to cells that are considered to be
native irreversibly sickled cells and were counted as
such. One thousand cells were counted.
To determine if macrocytosis was present in the slides
analyzed by image processing, the macrocyte/normocyte ratio was calculated for each film. To obtain the
ratio, the percentage of macrocytes was divided by the
percentage of normocytes. Normocytes included classes
1, 2, and 11, and macrocytes included classes 3, 12, 14
and 9. Target cells (class 9) were included in the macrocytic group because they were large and more closely
resembled the size of macrocytes (target cells 64.0 /iM2
± 1.3; macrocytes 70.0 nM2 ± 0.6; normocytes 54.7 pM2
± 0.07; M ± 1 SE). A mean macrocyte/normocyte ratio
then was determined for each patient. This was derived
from all the films obtained during crises or asymptom-
Table 2. Classes of Red Blood Cells
Class Number
Class Name
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Normocytes
Normocytes without central pallor
Macrocytes without central pallor
Spherocytes
Elongated spiculed cells
Spiculed, irregular shaped cells
Elongated cells
Irregular shaped cells
Target cells
Hypochromic microcytes
Hypochromic normocytes
Hypochromic macrocytes
Normochromic microcytes
Normochromic macrocytes
atic periods. The use of the macrocyte/normocyte ratio,
i.e., comparison between macrocytes and normocytes
in the same patient, avoided artifactual errors that might
have occurred if comparisons in the concentration of
macrocytes had been made between different patients.
The mean corpuscular volume (MCV) of red blood
cells was obtained on a Coulter Model S® and the mean
corpuscular hemoglobin concentration (MCHC) was
derived from measurements of the hemoglobin concentration and the hematocrit. Morphologic study of red
blood cells using standard visual microscopic methods
was done on glutaraldehyde-fixed preparations of cells
(2% glutaraldehyde in buffered saline; 130 mM CI, 20
mM sodium phosphate, pH 7.4) and on films stained
with Wright-Giemsa stain. The glutaraldehyde fixed and
the Wright-stained preparations were obtained from
blood samples that had been prepared like those used
for image processing. Whole blood was diluted to 10%
hematocrit with 0.9% NaCl prior to preparation of the
samples for the glutaraldehyde-fixed and Wright-stained
preparations. The criteria for morphologic diagnosis of
the various types of red blood cells followed standard
practice. Red blood cells that were "sweet potato"
shaped with one or more pointed extremity were considered to be irreversibly sickled cells.5 One thousand
cells were counted on these preparations.
Results
Four morphologic groups of red blood cells were obtained in the image processing analysis of the peripheral
blood cells of the patients. These consisted of normocytes, macrocytes, target cells, and cells with the shape
Vol. 79-No. 6
RBC MORPHOLOGY IN SICKLE CELL ANEMIA
of irreversibly sickled cells. Wide variation in the concentration of these groups of cells was observed among
patients as shown in the red blood cell differential count
(Table 1). Sequential measurements were made during
asymptomatic periods on eight patients on 72 separate
occasions to determine the stability of red blood cell
morphology during these periods. Variation in morphology was generally moderate. Wider variations were
observed in approximately 30% of the slides during the
asymptomatic periods. Fluctuation was greater in normocyte and macrocyte concentrations, while lesser
changes occurred in the concentration of irreversibly
shaped sickle cells and target cells.
Seven patients were examined during 12 painful crises
on 44 separate occasions. The results were compared
with those obtained in the same patients examined on
61 separate occasions during symptom-free periods.
During crises, macrocytosis occurred. The macrocyte/
normocyte ratio was greater during crises in six of the
seven patients than that observed during asymptomatic
periods (Table 3). Analysis of variance to test the difference in the macrocyte/normocyte ratio between crises
and asymptomatic periods was significant (<0.0001).
Examination of histograms for changes of cell size during crises also showed a marked rightward shift of the
area under the curve compatible wjth macrocytosis.
Marked fluctuation in the percentages of the different
cell types regularly occurred during crises.
The mean percentage of irreversibly shaped sickle
cells did not show a consistent change during crises as
compared with asymptomatic periods, but individual
counts did show greater fluctuation during crises. The
mean percentage of target cell concentrations remained
the same during painful and asymptomatic periods.
To find out if the concentration of irreversibly shaped
sickle cells determined by image processing accurately
reflects the ISC concentrations as determined by standard methods, comparisons were made between ISC
counts obtained by image processing and by visual examination of glutaraldehyde-fixed red blood cells. Both
types of cell preparations were prepared simultaneously.
Comparisons were done on five patients on 15 separate
occasions and showed a close correlation between the
percentages of irreversibly shaped sickle cells obtained
by both methods (Fig. 1) (r = 0.95, P = 0.0001).
To determine the reproducibility of results obtained
by image processing, nine pajrs of duplicate, simultaneously prepared slides from four patients were examined and the results compared with each other. A high
correlation between the two slides for the cell types (irreversibly shaped sickle ce}ls; normocytes and macrocytes) was observed (r = 0.81, P = 0.008; r = 0.86,
P = 0.003; r = 0.68, P = 0.04, respectively)
(Figs. 2A-C).
669
Table 3. Macrocyte/Normocyte Ratios in Patients
during Crises and Asymptomatic Periods
Patient
Crises
Asymptomatic
1
2
3
4
"5
6
7
1.98* (8)t
1.49 (6)
1.14 (11)
0.37 (5)
2.00 (6)
1.25 (5)
1.24 (3)
0.43 (10)
0.89(12)
0.59(15)
0.38 (7)
1.05(8)
0.45 (5)
1.06(4)
* Mean ratio of all specimens obtained during crises or asymptomatic periods.
f Number of specimens from which mean macrocyte/normocyte ratios were calculated.
Each specimen was taken on a day of crisis or an asymptomatic day.
Studies of the effects of brief aeration that occurs during film preparation for image processing were done to
determine whether the percentage of cells with irreversibly sickled shapes was affected by aeration. Analyses
were made on eight occasions from six patients in whom
smears were prepared in the standard manner, i.e., without added aeration of the blood sample, and on smears
that were prepared from blood after one hour of aeration. The results showed no change in irreversibly
shaped sickle cell concentrations after aeration (Table
4). Since formalin fixing also might cause reduction of
hemoglobin to the deoxy form with consequent sickling
of cells and is necessary for preparation of films for image processing, comparisons of ISC counts in Wrightstained formalin-fixed and in Wright-stained unfixed
smears were made. Formalin fixing had no effects on
60r
o
z
50
CO
CO
g «°
a.
o.
$ 30
<
I
>
10
20
«o
O
CO
10
"Tt5
215
50
40
50~
So
% ISC's BY VISUAL EXAMINATION
FIG. 1. Comparisons between the image processing method and
visual examination of glutaraldehydefixedred blood cells for determining the concentration of irreversibly shaped sickle cells (ISCs).
WESTERMAN AND BACUS
670
A.J.C.P. • June 1983
80r
60
70
50
^60
UJ
si
g
- 1 40l-
P 50
_l
2
<
t
Z 30
z
§
p
30
CO
O 20
(0
* 20-
10
10
10
30
20
40
50
% ISC's ON DUPLICATE SLIDE
60
10
20
30
4 0 5 0
66
% NORMOCYTES ON DUPLICATE SLOE
70
SO
70
60
LU
Q
w 50
_i
<
tZ
40
z
O
CO
UJ
H
FIG. 2. Comparisons of the concentration of {A, upper left) irreversibly shaped sickle cells (B, upper right) normocytes, and (C. lower
left) macrocytes in duplicate slides as obtained by image processing.
30
O
O
cr
^20
10
10
20
30
40
50
60
70
% MACROCYTES ON DUPLICATES SLIDE
the concentration of ISCs in four patients who were
studied on five occasions (Table 5).
Discussion
In our study, we show that peripheral blood red blood
cells in patients with sickle cell anemia are divided into
four morphologic groups: (1) normal cells, (2) macrocytes, (3) target cells, and (4) cells with the shape of
irreversibly sickled cells. During asymptomatic periods
the percentage of each type of cell is relatively typical
and constant for each patient but varies considerably
from patient to patient. The least change in concentration of cell types during these periods occurs in cells
with irreversibly sickled shapes and in target cells, while
more variability is observed in the percentage of normocytes and macrocytes. Red blood cell production appears to be relatively stable during these periods since
neither the hematocrit nor the percentage of the various
types of peripheral red blood cells show significant
Vol. 79 • No. 6
RBC MORPHOLOGY IN SICKLE CELL ANEMIA
changes. Occasionally, however, an abrupt and marked
change in the concentration of the various groups of red
blood cells did occur, although the hematocrit remained
unchanged. This suggests that the factors affecting the
morphology of peripheral red blood cells during asymptomatic periods may, at times, change although they are
generally stable. Whether these occasional, marked alterations in peripheral red blood cell morphology during
asymptomatic periods represent the effects of less extensive vasocclusive episodes than those observed in typical severe painful crises is not certain. This would, however, be a likely explanation since similar but more protracted changes in red blood cell morphology were
observed during severe crises. The variations in morphology also suggest that a single measurement of the
concentration of any one cell type may not accurately
represent the concentration of that particular cell type
during an entire asymptomatic period.
During crises, several changes occurred in the concentrations of the various groups of cells. Macrocytosis
was observed by the increase in the macrocyte/normocyte ratio in six of seven patients and by the increase
in the number of large cells seen in histograms that
showed the size distribution of cells. The concentration
of irreversibly shaped sickle cells was not consistently
increased or decreased as compared with the concentration observed during asymptomatic periods. Target
cell concentrations were unchanged. The significance of
these changes could be several. The increase in macrocytes in association with a stable hematocrit most likely
represents the delivery of a greater number of young red
blood cells to the periphery in response to more marked
red blood cell destruction. Increased hemolysis during
pain crises has been described previously in patients with
sickle cell anemia; however, it was thought to occur only
to a minor extent and did not indicate a widespread
increase in hemolysis.13 Also, the previously described
increase in hemolysis did not appear to induce a marrow
response since a reticulocyte rise above an already elevated level was not observed.6 Our results show that the
hemolysis that occurs during vasocclusive episodes is
sufficiently severe to induce a marrow response, as
shown by peripheral macrocytosis, and thus may be
more marked than previously noted. This finding which
differs from the earlier result also may be related to the
differences in methods of measurement. Reticulocyte
counts, although very useful, have limited reproducibility and accuracy15 and may be relatively insensitive
to moderate changes in hemolysis. Measurement of cell
size by image processing is an accurate, reproducible
method and would more precisely reflect an increased
concentration of large red cells in the periphery. The
finding of a marrow response is compatible with the
671
Table 4. Concentrations of Irreversibly Shaped Sickle
Cells Observed on Films Prepared Before or After
Aeration* of Blood Samplesf
ISC (%)
Patient
Before
After
1
3.5
4.2
25.1
12.9
22.7
7.3
32.1
28.5
4.2
7.2
29.9
13.7
19.0
4.7
31.2
32.2
2
3
4
5
6
• Blood samples were exposed to room air for 60 minutes prior to preparation of films.
t Image processing analysis was used.
concept that further marrow compensation could occur
in these patients. 12
The relationship between red blood cell morphology
and vasocclusive pain crises in patients with sickle cell
anemia has had limited study. Irreversibly sickled cells
have been considered to be an important precipitating
factor in the development of painful crises since they
contribute significantly to the increased viscosity of oxygenated sickle blood. 4 Because of this, and since ISC
concentrations in peripheral blood may reflect the degree of in vivo sickling,16"18 it would seem that the concentration of peripheral blood ISCs might change during
vasocclusive episodes. Our studies show that the concentration of ISCs does fluctuate much more widely
during crises; however, a consistent increase or decrease
does not occur. Previous studies had similarly shown
that the percentage of ISCs does not show a consistent
change during crises,6 although the percentage of sickled
cells in venous blood did increase during crises.-' These
results suggest that the relationship between in vivo sickling and ISC concentration may not be as close as considered. The findings would not negate the observations
Table 5. Concentrations of Irreversibly Shaped Sickle
Cells Observed in Films Prepared with or
without Formalin Fixing*
ISC (%)
Patient
Without
Formalin Fixing
With
Formalin Fixing
1
2
3
4
"
38
31
16
34
25
41
30
21
33
24
* Counts were done manually on Wright-Gicsma-stained smears.
672
WESTERMAN AND BACUS
A.J.C.P.-June 1983
5. Clark MR, Guatelli JC, Mohandas N, Shohet SB: Influence of red
cell water content on the morphology of sickling. Blood 1980;
55:823-830
6. Diggs LW: The crises in sickle cell anemia: hematologic studies.
Am J Clin Pathol 1956;26:1109-1118
7. Eaton JW, Skelton TD, SwofTord HS, Kolpin CE, Jacob HS: Elevated erythrocyte calcium in sickle cell disease. Nature 1973;
246:105-106
8. Glader BE, Nathan DG: Cation permeability alterations during
sickling: relationship to cation composition and cellular hydration of irreversibly sickled cells. Blood 1978: 51:983-989
9. Hebbel RP, Boogaerts MAB, Eaton JW, Steinberg MH: Erythrocyte adherence to endothelium in sickle cell anemia: a possible determinant of disease severity. N Engl J Med 1980;
302:992-995
10. Hofrichter J, Ross PD, Eaton WA: Kinetics and mechanism of
deoxyhemoglobin S gelation: a new approach to understanding
sickle cell disease. Proc Natl Acad Sci USA 1974; 71:48644868
11. Lux SE, John KM, Karnovsky MJ: Irreversible deformation of
the spectrin-actin lattice in irreversibly sickled cells. J Clin Invest 1976; 58:955-963
12. McCurdy PR: Erythrokinetics in abnormal hemoglobin syndromes. Blood 1962; 20:686-699
13. Naumann HN, Diggs LW, Barreras L, Williams BJ: Plasma hemoglobin and hemoglobin fractions in sickle cell crisis. Am J
Acknowledgment. The authors would like to thank Nijole Dumbrys
Clin
Pathol 1971; 56:137-147
and Margaret Sheedy for their excellent technical assistance.
14. Noguchi CT, Schechter AN: The intracellular polymerization of
sickle hemoglobin and its relevance to sickle cell disease. Blood
1981;58:1057-1068
References
15. Peebles DA, Hochberg A, Clarke TD: Analysis of manual reticulocyte counting. Am J Clin Pathol 1981; 76:713-717
16. Serjeant GR, Serjeant BE, Milner PF: The irreversibly sickled cell;
1. Bacus JW, Weens JH: An automated method of differential red
blood cell classification with application to the diagnosis of
a determinant of haemolysis in sickle cell anaemia. Br J Haeanemia. J Histochem Cytochem 1977; 25:614-632
matol 1969; 17:527-533
2. Bacus J W: Quantitative morphological analysis of red blood cells. 17. Serjeant GR: Irreversibly sickled cells and splenomegaly in sickle
Blood Cells 1980;6:295-314
cell anemia. Br J Haematol 1970; 19:635-641
3. Barreras L, Diggs LW: Bicarbonatcs, pH and percentage of sickled 18. Serjeant GR, Serjeant BE, Condon PI: The conjunctival sign in
sickle cell anemia. JAMA 1972; 219:1428-1431
cells in venous blood of patients in sickle cell crisis. Am J Med
19. Shaklai N, Sharma VS, Ranney HM: Interaction of sickle cell
Sci 1964; 247:710-718
hemoglobin with erythrocyte membranes. Proc Natl Acad Sci
4. Chien S, Usami S, Bertles JF: Abnormal rheology of oxygenated
USA 1981; 78:65-68
blood in sickle cell anemia. J Clin Invest 1970; 49:623-634
that irreversibly sickled cells may be precipitating events
in painful vasocclusive crises but show that peripheral
ISC counts are not useful as an indicator of painful
crises. The variation in individual ISC counts during
crises is also sufficiently large to suggest that single measurements are not a satisfactory indicator of the concentration of ISCs that are present during crises.
Image processing is very appropriate for evaluating
the red blood cells from patients with sickle cell anemia.
This is of some interest since the stickiness of Hb SS red
blood cells, the changes that occur during oxygenation,
and the necessity for cell fixation prior to image processing could create difficulties. The reproducibility of
the method is very good, however, and it does not appear
that significant changes of the cells occur during formalin fixation or slide preparation. Image processing
also readily permits analysis of all types of red blood
cells and not just one type such as the ISC.