Computer-assisted Spectrophotometric Analysis of
Amniotic Fluid in Erythroblastosis Fetalis
R. C. BROWN, B.S., MT(ASCP), AND W. J. BECKFIELD, M.D.
Department
of Clinical Laboratory, Luther Hospital, Eau Claire, Wisconsin 54701
ABSTRACT
Brown, R. C , and Beckfield, W. J.: Computer-assisted spectrophotometric
analysis of amniotic fluid in erythroblastosis fetalis. Am. J. Clin. Pathol. 57:
659-663, 1972. A computer program has been used to assist in the analysis of
absorbance changes in amniotic fluid from Rh-sensitized pregnancies. The
advantages of the program are reduced chance of human error in calculation
or communication, numerical reports in a form similar to those of other
chemistry values on the patient's chart, and increased laboratory efficiency.
Programmed calculation should be combined with a visual inspection of a
wavelength scan for detection of interfering pigments.
MEASUREMENT OF CHANGES in optical density of amniotic fluid at 450 nm. (AA450)
has been accepted as valuable in determining the management of Rh-sensitized pregnancies.2 On the basis of repeated amniocenteses, Liley 1 defined three zones in the
semilogarithmic plot of AA450 VS. week of
gestation. Cases falling in zone 3, if untreated, will have severe disease or intrauterine death. Zone 1 infants are Rh negative or mildly affected if Rh positive. Those
in zone 2 are indeterminate on the first
test, but can be evaluated by the trend
toward the higher or lower zones on repeated examinations.
Received April 19, 1971; received revised manuscript June 29, 1971; accepted for publication July
28, 1971.
The customary laboratory procedure following spectrophotometry of centrifuged or
filtered amniotic fluid consists of a semilogarithmic plot of absorbance vs. wavelength. A baseline is drawn between absorbances at two wavelengths (365 nm.,
550 nm.) and the baseline value at 450 nm.
is subtracted from the measured absorbance at 450 nm. The AA450 must then be
communicated to the obstetrician. The obstetrician may consult Liley's diagrams or
the laboratory may transmit information
as to the zone and trend of this patient's
values.
Errors may occur in the reading of the
graphs or, most seriously, in the placement
of a decimal point, because two decades
of logarithmic scale usually are involved.
A mathematical presentation of data would
659
660
AJ.C.P.—Vol.
BROWN AND BECKFIELD
.20
.10
.09
.08
\
\
•07.06
^V
•05
N^V
•m
Vo - 0 -" 2 '
V
.03
- 0.08(S
DDTEREKCZ * O.OI46
\$o from baseline » 0 .01,6
A
57
rectangular 1 cm. light path cuvettes, and
a distilled water blank. In addition to an
automatic wavelength scan, absorbances
were read at 10 to 15 nm. intervals from
365 to 575 nm.
Figure 1 shows a plot on two-cycle semilog paper of absorbance versus wavelength.
The points at wavelengths 365 nm. and
550 nm. were joined to form a baseline
and the baseline absorbance at 450 nm.
(x = 450) was read from the graph and subtracted from the measured absorbance of
the amniotic fluid at 450 nm. to obtain
A.A450.
.02-
.ni
350
375 ¥ »
l»25
kSo 1)75 500' 525 55o 575
WAVBI2NQTX (nanonsiars)
FIG. 1. Semilogarithmic plot of optical density vs.
wavelength as used for calculation of AA4a>.
not be subject to these errors. Electronic
computation would improve speed, efficiency, and accuracy. We, therefore, wrote
and tested a program for an OlivettiUnderwood Programma 101 for computation of AA450, comparison of this with
Liley's zones, calculation of the zone number in which this value lies, and its relative position within the zone.
Materials and Methods
Fifty specimens from 19 patients were
received for analysis during the period of
this study. These specimens had been collected by transabdominal amniocentesis,
placed directly into light-tight containers,
and sent immediately to the laboratory.
After centrifugation to remove cells and
debris, wavelength scans were performed
on a Coleman 124 double-beam spectrophotometer using a 1 nm. band pass, paired
30
31
32
33
WEEK OF G8STATIOX
FIG. 2. Plots of amniotic fluid optical density increases at 450 nm. for 24 specimens from nine Rhsensitized pregnancies to show relationship of computed b values to the three major zones defined
by Liley.
May 1972
661
COMPUTER ANALYSIS OF AMNIOTIC FLUID
Mathematical treatment of the absorbance data used the equation for any line
not parallel to the log scale in a semilog
coordinate system.3
y = bax
(1)
In logarithmic form:
log y = log b + xlog a
SIDE
*
SDR R
*
E-
c l
s
11
1 •
c 1
31
t 1
s
cv
u
a<
• X
/V
b I
'1
1 •
'.
5
(2)
Because the baseline was drawn between
two points with known coordinates (365,
A365) and (550, A5B0), it was possible to
determine the equation of this line and
solve for the value of y at the desired point
of intersection with (x = 450).
A4B0 = antilog [log AS65 + 85/185 X
(log A550 - log A36B)] (3)
This baseline number was then subtracted
from the measured amniotic fluid absorbance at 450 nm. to obtain AA450.
The same semilog equation 1 was used
to relate the AA450 to the zone-graph of
Liley. Figure 2 shows the AA4S0 plotted
logarithmically versus weeks of gestation.
It can be demonstrated algebraically and
was confirmed by computation that parallel lines in a semilog coordinate system
have identical values for the number a,
which may be considered a slope constant.
The lines which divide Liley's zones are
parallel. The value of the number a was
calculated.
As can be seen from equation 1, the
number b varies directly in proportion to
y. We computed b values for the lines
separating zones 1 and 2, lower and upper
zone 2, zones 2 and 3, and lower and upper
zone 3.
A Programma 101 card was prepared
using the log and exponential routines supplied with the computer.4 Entries are 1000,
A
5so. A305, A450, and week of gestation. The
printout is AA450, identification of the zone
in which this point is located, and the
value of b of equation 1 for a line through
0 X
Ci1
C 1
c I
• V
a 1
'3
1
1
b -
at
8 X
fi 1
El
*i •
c (
•<
• •
ir
It
C 1
!.X
d •
B-
• 1
•1
r S
t 1
.
1 1
4 -
8«
P- I
11
•1
»Y
1
• 1
'. t
<r
« 1
f •
c 1
1
:• i
I 1
9 1
r-
Y
n
«Z
l
a r
.. 1
• 1
n i
» 1
n 1
'1
' 0
•
*
«•
r
• I
IZ
•
£-
•»
c 1
>
to
i
tf
a i
•' ?
a 1
c X
a i
r -
OT: nK.TT,
ei
f
im. : \
i
r;
i
'.:-jjT"Tr
' • "
r.jns
' • i s i m
1 -C " " c i r
e 9
: • *1 « 7 « 1
t 4
p
1 -;•-,- r -
Ft
FIG. 3. Computer program for amniotic fluid
analysis. Entries are 1000, A n , A^B. A4M, and weeks
of gestation. After passage of Side B, computer
prints AA4E0, Liley zone number, and b value.
the point parallel to the zone lines. This
program (Fig. 3) was tested against hypothetical cases and by calculation of 50 patient scans.
Results
Agreement between computer and
graphic values for AA450 was ±0.001 A.
The values of constant a for slope of the
zone lines illustrated in Liley's publications was found to be 0.91509 to five significant figures. This number, raised to the
power corresponding to weeks of gestation,
and substituted into a rearranged equation
662
BROWN AND BECKFIELD
Table 1. Boundaries (b Values) of Liley Zones Computed
from the Equation for Lines in a
Semilog Coordinate System
Zone
Lower
Upper
Lower
Upper
b
I
II
II
III
III
<0.7
0.7-1.4
1.4-3.0
3.0-6.0
>6.0
1, b = y/a", yielded values for b as precise as the absorbance read from the
graphs. The b constants for zone dividing
lines are given in Table 1. Computed b
values from our patients and their graphic
presentation are shown in Figure 2. It will
be noted that successive points have nearly
the same b value when the line connecting
the points is approximately parallel to the
zone lines. Differing b values indicate a
change in relative position for both consecutive and widely separated tests. Liley's
zone lines, which were drawn to include
groups of clinically similar cases, have b
values which are approximate multiples of
2 (Table 1).
Discussion
The use of a programmable calculator
has made practical a mathematic format
comparable to Liley's graphic handling of
data from Rh-sensitized pregnancies. The
mathematic result has no more, and no
less, biologic significance than the graphic.
It is considerably more rapid. It has value
in communication, in that the computer
provides not only absorbance increase but
also zone designation and a value b for the
relative position within the zones which
can be compared in serial specimens. Possible in-laboratory errors in reading baseline absorbance at 450 nm. from the graph,
and errors in placement of the decimal or
in relating AA450 to Liley's graphs by either
the laboratory or the obstetrician are eliminated. Changes in the value of b can be
A.J.C.P. —Vol. 57
appreciated more easily as numbers than
by successive plots on the graph.
Two problems arose in our comparison
of computer calculations with the graphic
method. One of these was the interpretation of specimens obtained prior to the
27 th week of gestation. These computed
readily, but for graphic comparison, it was
necessary to extrapolate Liley's zones to
include areas in which clinical data had
not been published. One patient had specimens drawn at 23.5, 24, 25, and 26.5 weeks,
with corresponding b values of 1.5, 1.6,
1.8, and 2.9. Immediately after the last of
these was obtained the patient was referred
to another institution, where repeated intrauterine transfusions were followed by
delivery at 34 weeks' gestation of an infant
with cord blood hemoglobin of 13.5 Gm.
per 100 ml. The cells in this blood were
all Type O, D negative, and no fetal hemoglobin was demonstrable. Another patient
had b values of 3.3 and 3.4 during the
25th week of gestation and also was referred and managed successfully by intrauterine transfusion. Zone computation before 27 weeks or extrapolation of the graph
seems justified in cases in which severe
disease is expected from the clinical history,
as it was in each of these cases.
The other problem was related to the
presence of pigments other than bilirubin
and bilirubinoid substances. In one case a
patient showed an extremely high Soret
peak at 403 nm. and positive Schumm's test
for methemalbumin on the first amniocentesis. Because of the effect of methemalbumin on the baseline, a negative AA4B0 computed. The Soret peak was readily detected
by wavelength scanning, but might have
been missed if only the three wavelength
absorbances required for calculation had
been measured. I n subsequent specimens
from this Rh-sensitized mother, the methemalbumin concentration decreased, but
no increase in A4B0 was found and a nor-
May
1972
COMPUTER ANALYSIS OF AMNIOTIC FLUID
mal Rh-negative infant was delivered at
term. The presence of other pigments
would be similarly detected by the automatic wavelength scan but might be missed
if only the three absorbance readings were
made.
Automatic wavelength scanning required
about 5 min.; manual absorbance recording at 10 to 15 nm. intervals would take
slightly longer. The Programma 101 required about 2 min. of computation time
because of the logarithmic and exponential
calculations involved. At least 10 min.
more was needed for graphic methods than
for the computation by programmable calculator. A recording spectrophotometer
which would plot logarithms of absorbance
and a computer with logarithmic and ex-
663
ponential circuitry, if available, should
provide even more rapid results. Choice of
other wavelengths for the determination
of baseline would necessitate appropriate
modification of entries and stored constants
in equation 3.
References
1. Liley AW: Liquor amnii analysis in the management of the pregnancy complicated by rhesus sensitization. Am J Obstet Gynecol 82:13591370, 1961
2. Oski FA, Naiinan JL: Hematologic problems in
the newborn. Major Problems in Clinical Pediatrics. Volume IV. Philadelphia, WB Saunders
Company, 1967, pp 153-157
3. Randolph JF: Calculus and Analytical Geometry.
Second edition. Belmont, California, Wadsworth Publishing Company, Inc., 1966, pp 146148
4. Williams JB: Mathematical Subroutines. New
York, Olivetti-Underwood Corporation, 1968,
pp 127, 269
AMERICAN SOCIETY OF CLINICAL
PATHOLOGISTS
2100 West Harrison Street, Chicago, Illinois 60612
FUTURE MEETINGS
October 13-21, 1972—San Francisco Hilton, San Francisco, California
October 12-20, 1973—Palmer House, Chicago, Illinois
October 19-25, 1975—Palmer House, Chicago, Illinois