The gel test: sensitivity and
specificity for unexpected
antibodies to blood group antigens
W.J. JUDD, E.A. STEINER, AND P.C. KNAFL
The recently FDA-licensed anti-IgG gel test for pretransfusion antibody
detection requires crossover validation before implementation. Six
hundred coded samples sent for routine pretransfusion tests were
used to compare GEL (ID-MTS, Ortho Diagnostic Systems Inc., Raritan,
NJ) with Löw and Messeter’s low-ionic-strength saline (LISS). There
were 456 GEL–LISS–, 97 GEL+LISS+, 45 GEL–LISS+, and 2 GEL+LISS–
tests. The 144 positive tests involved 157 antibodies; 67 of these (cold
auto, anti-M, -Le, etc.) were considered harmless with respect to trans-
fusion management. GEL–LISS+ tests included seven samples con-
taining potentially significant antibodies (assumed from specificity):
anti-K(4), -Jka, -Fyb, and -S. Two potentially significant antibodies (anti-
C and -D) were GEL+LISS–. Sensitivity and specificity for potentially
significant antibodies were 92% and 96% for GEL, and 98% and 90%
for LISS, respectively. The seven GEL–LISS+ samples associated with
potentially significant antibodies were from six patients. Five of these
antibodies, all detected in immune-suppressed patients, reacted pre-
dominantly as agglutinins in LISS. None of these seven antibodies
were detected reliably by polyethylene glycol and LISS-additive tube
methods. In light of the immune status of the patients with GEL–LISS+
agglutinins with specificity normally considered potentially significant,
and because other valid methods did not detect these antibodies, their
clinical importance is questionable. Excluding these questionable anti-
bodies, GEL has the same sensitivity and better specificity than LISS.
GEL is a valid method for pretransfusion antibody detection.
Immunohematology 1997;13:132–135.
Key Words: gel test, antibody detection, LISS, pretrans-
fusion testing, method validation
The gel test for detecting red blood cell (RBC) antigen-
antibody interactions has been utilized by blood group
serologists outside the United States since its first descrip-
tion in 1990 by Lapierre and colleagues.1 In 1994, Micro
Typing Systems, Inc. (Pompano Beach, FL), obtained
approval from the U.S. Food and Drug Administration to
manufacture an anti-IgG gel test (GEL) for detection of
unexpected antibodies.2 This product (ID-MTS GEL) is
now available in the United States through Ortho
Diagnostics Systems Inc. (Raritan, NJ).
There are reports of studies comparing the European
version of GEL (DiaMed AG, Morat, Switzerland) with
tube technologies utilized in Europe and Canada.3-5
132
However, there are no reports that compare ID-MTS GEL
with enhancement methods in use in transfusion services
within the United States. Moreover, as with any new
technology under the Clinical Laboratory Improvement
Act of 1988,6 GEL requires crossover validation before
implementation.
In this report, we present serologic findings compar-
ing GEL to our routine antibody detection procedure
based on the LISS procedure of Löw and Messeter.7
Materials and Methods
An ID-Micro Typing System™ for performing anti-IgG
gel tests was provided courtesy of Ortho Diagnostics
Systems Inc. Samples for GEL were predominantly from
EDTA-anticoagulated blood and were from patients on
whom routine pretransfusion tests had been performed
within the preceding 48 hours. Samples were subjected
to GEL within 48 hours of collection. Technologists per-
forming GEL were not privy to the findings by LISS.
Group O reagent CDe/CDe (R1R1) and cDE/cDE
(R2R2) RBCs were from Immucor, Inc. (Norcross, GA).
One of these RBC samples was always Jk(a+b–). The
same lot number of reagent RBCs was used for both GEL
and LISS.
LISS tests were performed by incubating one volume
(ca. 0.1 mL) of 2% reagent RBCs in LISS (Immucor, Löw
and Messeter’s formulation7) with an equal volume of
patient’s serum. The reactants were not warmed to 37°C
before mixing. Incubation was for 10 minutes at 37°C,
after which the tests were centrifuged and examined for
agglutination and hemolysis. The RBCs were then washed
x 4 with saline before testing with polyclonal (rabbit) anti-
IgG (Ortho). Reactions were read macroscopically with
the aid of an illuminated concave mirror and graded as
described previously.8 All negative antiglobulin tests
were verified with IgG-coated RBCs (Immucor).
GEL tests were set up and graded as recommended by
IMMUNOHEMATOLOGY, VOLUME 13, NUMBER 4, 1997
 Gel test validation

the manufacturer. Briefly, 50 μL of 0.8% reagent RBCs Table 2. Discrepancies between GEL and LISS involving antibodies of
presumed potential significance
suspended in a low-ionic-strength diluent were mixed
Test LISS*
with 25 μL of plasma. After a 15–minute incubation at red blood
37°C, the RBCs were centrifuged through a gel column Anti- cells 37°C IgG GEL* Patient’s medical history†
containing anti-IgG. In our laboratory, reactions in GEL Jka Jk(a+b–) 1+ 0 0 allogeneic bone marrow
transplant
were read utilizing an illuminated view box; doubtful K‡ K+k+ 1+ 0 0 chronic myelogenous
reactions were evaluated using a handheld magnification leukemia
S S+s+ 0 1+ 0 craniotomy,
lens. not transfused
Results by GEL were compared with those by LISS. K K+k– 2+ 1+ 0 liver transplant
K K+k– 1+ 0 0 chronic renal failure§
Discrepancies were evaluated by repeat testing by both Fyb Fy(a–b+) 0 + 0 lung transplant
methods, using both serum and EDTA plasma, as avail- C CDe/CDe 0 0 1+ chronic renal failure
able. When apparent significant antibodies were missed D cDE/cDE 0 0 1+ posttransfusion hepatitis
* Reactions graded as described in the text
by GEL, additional testing was undertaken using multiple † All patients multiply transfused except as indicated
LISS additives and polyethylene glycol (PEG).9 Statistical ‡ Two samples tested on different occasions yielded similar results
analysis of test data were performed as described by § Patient also had C. difficile colitis

Annesley.10
-Fyb, and -S), while LISS failed to detect two Rh antibod-
Results ies (anti-C and -D).
Results of GEL and LISS tests for unexpected antibod- Specific details of our findings on the eight patients
ies on 600 samples are shown in Table 1. There were whose serum contained potentially significant antibodies
456 GEL–LISS–, 97 GEL+LISS+, 45 GEL–LISS+, and 2 that yielded nonconcordant results are summarized in
GEL+LISS– samples. The 144 reactive samples involved Table 2. The seven presumed potentially significant
90 antibodies of potential significance (presumed from GEL–LISS+ antibodies involved five that reacted predom-
specificity) and 67 antibodies that were considered harm- inantly as direct agglutinins in LISS. All but one of the
less with respect to transfusion of incompatible blood seven antibodies were from patients who were immune
(cold auto, Paraben®-dependent anti-Jka,11 -M, -Le, etc.) suppressed by disease or medication; the exception was
LISS detected 66/67 harmless antibodies and 88/90 anti- the anti-S, which was from a 14-year-old boy who had
bodies of potential significance. In contrast, GEL detect- never been transfused.
ed 22/67 harmless antibodies and 83/90 potentially Also shown in Table 2 are specific details on two sam-
significant antibodies. GEL did not detect seven poten- ples containing potentially significant Rh antibodies that
tially significant antibodies in six patients (anti-K(4), -Jka, were missed by LISS. It is of note that the anti-D was from
a patient alleged to have anti-E by an outside hospital. In
Table 1. Results of comparative study between GEL and LISS routine antibody identification studies, which included
Significance* Reaction Samples Antibodies Specificity both LISS and ficin-antiglobulin techniques, no Rh anti-
GEL–LISS– 456 body activity was observed but anti-D was clearly demon-
Harmless GEL+LISS+ 20† 21
GEL–LISS+ 38 45 cold auto (18), M(8),
strable by GEL.
Le(8), NOID‡(5†), Table 3 shows the results of additional tests performed
warm auto (2), N, on the six patients with GEL–LISS+ antibodies of poten-
P1, Paraben®-
dependent-Jka§, tial significance. Of relevance is the finding that some
“rouleaux” widely used commercially available LISS-additives, an in-
GEL+LISS– 1 1 auto-Pr
TOTAL 59 67 house LISS preparation based on the formulation of Löw
Significant GEL+LISS+ 77† 81 and Messeter,7 and PEG tube tests9 did not reliably detect
GEL–LISS+ 7 7 K(4), Jka, S, Fyb
GEL+LISS– 1 2 C, D† these antibodies.
Total 85 90 Table 4 shows the results of predictive value analysis of
Grand Total 600 157
the data. GEL was superior to LISS in terms of specificity
* Presumed significance based on specificity and derived from clinical
experience, with respect to transfusion of incompatible blood and therefore has greater test efficiency. LISS was supe-
† One GEL+LISS+ sample contained a GEL+LISS– anti–D and an unidentified rior to GEL in its ability to detect antibodies of specifici-
GEL–LISS+ antibody
‡ Not identified; reaction pattern inconsistent with single or multiple
ties that are considered potentially significant with
alloantibody specificities respect to transfusion of incompatible blood (but see
§ An additional example of anti-Jka dependent upon the presence of Discussion).
Paraben®, an anti-fungicide used in the commercial preparation of LISS11

IMMUNOHEMATOLOGY, VOLUME 13, NUMBER 4, 1997 133

W.J. JUDD ET AL.

Table 3. Results of additional testing on patients with presumed potentially significant antibodies missed by GEL* using an IgG antihuman globulin test (AHG)
LISS-additives†
L1 L2 L3 L4§
PEG‡
Anti- AHGll 37°C AHG 37°C AHG 37°C AHG 37°C AHG Specificity¶
Jka 0 4+ 0 0 0 0 0 0 0 not confirmed**
K†† 0 4+ 1+ 0 0 0 0 0 0 confirmed
S 0 0 0
K 0 3+ 0 0 0 confirmed
K 0 3+ 0 confirmed
Fyb 2+ 0 0 0 0
* Majority of testing performed elsewhere
† Three commercially available LISS-additive reagents
‡ In-house polyethylene glycol solution (PEG)9
§ In-house LISS solution based on the formulation of Löw and Messeter7
ll Antihuman globulin test
¶ Antibody specificity confirmed/not confirmed in subsequently performed antibody identification tests using the most reactive method
** Reacted 4+ at 37°C with one Jk(a+b–) RBC sample using LISS-additive #1; a Jk(a+b+) sample was nonreactive; antibody was nonreactive in subsequent testing
†† Two samples tested on different occasions; second sample not submitted for additional testing

Discussion
This study has shown that the sensitivity and specifici-
ty of GEL is satisfactory for routine pretransfusion and
perinatal antibody detection. Moreover, according to the
results of test efficiency calculations, which indicate how
many times a test yields the right answer, GEL is superior
to LISS.
LISS detected seven antibodies (in six patients) of
potential significance (assumed from specificity) that
were not detected by GEL. In evaluating the relevance of
this observation, the following should be considered.
First, there are no published reports of hemolytic trans-
fusion reactions attributed to the widespread use of those
LISS-additive methods that failed to detect these agglu-
tinins (see Table 3). Second, there are no published
reports of hemolytic transfusion reactions associated with
the fact that the widely utilized PEG technique9 does not
include a reading for agglutination after 37°C incubation.
Third, the absence of reports of hemolytic transfusion
reactions following the widespread implementation of
GEL by transfusion services outside the United States also
supports the contention that these antibodies are not clin-
ically relevant. Indeed, subsequent to this study and in
light of our previously published data,12 we have elimi-
nated the 37°C reading from our routine LISS antibody
detection procedure. This measure improves the pre-
dictive value of a positive LISS test from 57% to 76% (vs.
79% for GEL) but is not associated with an increase in the
number of observed hemolytic transfusion reactions.13
Early in this evaluation, we found accuracy of RBC con-
centration and use of plasma rather than serum to be
advantageous in avoiding “toplines,” thin lines of unag-
glutinated RBCs that appear on top of the gel column
after centrifugation. In addition, it was noted that agglu-
tinins (anti-I, -M, etc.) and rouleaux may yield unique reac-
tion patterns by GEL (mixed-cell, toplines, or a diffuse ver-
tical column of RBCs within the gel). In our experience,
these unusual reactions as well as some weakly reactive
antibodies were more readily discernible using an illumi-
nated view box and magnification lens; how-ever, rec-
ommendations for or against the use of either of these
optical aids are not included in the manufacturer’s pack-
age insert.
As with any new technology, in addition to the sero-
logic evaluation (validation) of GEL for antibody detec-
tion, other factors to consider prior to implementation
include (1) validation of antibody identification and other
antiglobulin test procedures; (2) work process redesign
to facilitate efficient setup of GEL antibody detection with
ABO/Rh tube testing or to perform these tests indepen-
dently; (3) costs per test, including personnel time and
costs for commodities; (4) the ability to batch tests in
order to maximize the cost-effectiveness of GEL; (5) cur-
rent good manufacturing and compliance issues relative
to CLIA '88;6 (6) staff training and competency assess-
ment; and, (7) revisions to standard operating proce-
dures. These issues will be addressed in subsequent
reports. From data presented in this present study, we
conclude that GEL is a valid method for pretransfusion
antibody detection.
Acknowledgment
The ID-Micro Typing System™ used in this investiga-
tion was provided courtesy of Ortho Diagnostic Systems
Inc., Raritan, NJ.

134 IMMUNOHEMATOLOGY, VOLUME 13, NUMBER 4, 1997

 Gel test validation

Table 4. Predictive value (PV) analysis for GEL and LISS

GEL LISS
Sensitivity*
TP 83 88
_______ x 100 ______ x 100 = 92% ______ x 100 = 98%
TP+FN 83+7 88+2

Specificity†
TN 515 515
________ x 100 _______ x 100 = 96% ________ x 100 = 90%
FP+TN 21+515 58+515

PV+*‡
TP 83 88
________ x 100 ________ x 100 = 79% ________ x 100 = 57%
TP+FP 83+22 88+66

PV–†‡
TN 494 457
________ x 100 ________ x 100 = 98.6% ________ x 100 = 99.5%
TN+FN 494+7 457+2

Efficiency†
TP+TN 83+494 88+457
_______________ x 100 _______________ x 100 = 95% _______________ x 100 = 90%
TP+FP+FN+TN 83+21+7+494 88+56+2+457
* Sensitivity and PV+ calculations based on reactions of antibodies encountered during study (see Table 1)
† Specificity, PV–, and efficiency calculation determined from numbers of reactive and nonreactive samples (see Table 1)
‡ PV+ = predictive value of a positive reaction; PV– = predictive value of a negative reaction

References 10. Annesley TD. Analytical test variables. In:

1. Lapierre Y, Rigal D, Adam J, et al. The gel test: a new
way to detect red cell antigen-antibody reactions.
Transfusion 1990;30:109–13.
2. Malyska H, Weiland D. The gel test. Lab Med
1994;25:81–5.
3. Bromilow LM, Adams KE, Hope J, Eggington J,
Duguid JKM. Evaluation of the ID-gel test antibody
screening and identification. Transfus Med
1991;1:159–61.
4. Phillips PK, Whitton CM, Lavin F. The use of the
antiglobulin “gel-test” for antibody detection.
Transfus Med 1992;2:111–3.
5. Pinkerton PH, Ward J, Chan R, Coovadia AS. An eval-
uation of the gel technique for antibody screening
compared with a conventional tube method.
Transfus Med 1993;3:201–5.
6. Code of Federal Regulations, 42 CFR 493.1213.
Federal Register 1992;57(40):7164.
7. Löw B, Messeter L. Antiglobulin tests in low-ionic
strength salt solutions for rapid antibody screening
and crossmatching. Vox Sang 1974;26:53–61.
8. Judd WJ. Methods in immunohematology. 2nd ed.
Durham, NC: Montgomery Scientific Publications,
1994.
9. Nance SJ, Garratty G. A new potentiator of red blood
cell antigen-antibody reactions. Am J Clin Pathol
1987;87:633–5.
McClatchey KD, ed. Clinical laboratory medicine.
Baltimore, MD: Williams and Wilkins, 1994:77–95.
11. Judd WJ, Steiner EA, Cochran RK. Paraben-associat-
ed autoanti-Jka antibodies: three examples detected
using commercially prepared low-ionic-strength
saline containing parabens. Transfusion 1982;22:
31–5.
12. Judd WJ, Steiner EA, Oberman HA, Nance SJ. Can the
reading for serological reactivity following 37°C incu-
bation be omitted? Transfusion 1992;32:304–8.
13. Judd WJ, Fullen DR, Steiner EA, Knafl P. Omitting the
37C reading (abstract). Transfusion 1997;65S.
W. John Judd, FIBMS, MIBiol, Professor of
Immunohematology, Department of Pathology, UH-
2G332, University of Michigan Health Systems, 1500
East Medical Center Drive, Ann Arbor, MI 48109-
0054; E. Ann Steiner, MT(ASCP)SBB, Supervisor,
Reference Laboratory, Blood Bank and Transfusion
Service, University of Michigan Health Systems, Ann
Arbor, MI (current address: Ortho Diagnostics Systems
Inc., Raritan, NJ); and Pamela C. Knafl, MT(ASCP)SBB,
Senior Clinical Technologist, Reference Laboratory,
Blood Bank and Transfusion Service, University of
Michigan Health Systems, Ann Arbor, MI.

IMMUNOHEMATOLOGY, VOLUME 13, NUMBER 4, 1997 135