The ferning test for detection of amniotic fluidcontamination in umbilical blood samples

The ferning test for detection of amniotic fluid contamination in umbilical blood samples Andrew Chao, MD, James P. Herd, MD, and Khalil M. A. Tabsh, MD Los Angeles, California There is currently no standard means of detecting amniotic fluid contamination in fetal blood obtained by percutaneous umbilical blood sampling. The feming test is proposed for this purpose. An in vitro model, using centrifuged serial dilutions of neonatal cord blood and amniotic fluid, showed that feming of the supernatant occurs in proportion to the degree of contamination. A system for grading feming was devised, and tested in a blind trial. For the detection of amniotic fluid contamination of 10% or more, the method showed a sensitivity of 98% and a specificity of 81 %. The method may be useful in percutaneous umbilical blood sampling, and in the diagnosis of spontaneous rupture of membranes with vaginal bleeding. (AM J OBSTET GVNECOL 1990;162:1207-13.)

Key words: Percutaneous umbilical blood sampling, cordocentesis, amniotic fluid, ferning test

Fetal blood specimens obtained by percutaneous umbilical blood sampling (cordocentesis) may be contaminated by maternal blood or amniotic fluid. Although maternal blood contamination can be identified with Kleihauer-Betke stain or mean red cell volume measurement, I there is no standard method to detect partial contamination with amniotic fluid. Amniotic fluid smears form ferning patterns on drying, whereas human serum does not."' 3 We therefore proposed that the presence of ferning in plasma from heparinized cordocentesis specimens indicates amniotic fluid contamination, and that the extent of ferning is proportional to the degree of contamination. This hypothesis was tested in vitro with mixtures of amniotic fluid and neonatal cord blood specimens.

Material and methods Ten amniotic fluid specimens were obtained from third-trimester pregnant women at the University of California Los Angeles Center for Health Sciences by one of the following techniques: (l) intrapartum transcervical needle amniotomy for indicated artificial rupture of membranes, (2) indicated amniocentesis, or (3) needle aspiration through intact membranes at cesarean delivery. Patients were selected serially. Cases of chorioamnionitis were excluded because umbilical blood sampling would not generally be considered with chorioamnionitis present. FI"01Il the Division of Maternal-FI,tal Ml'dicille, Univnsit.lof Calilinnia Los AngelI'S School of Medicine. Supported by University of Califim!ia Los Angeles Prole.lsiollul Development Fund no. 405040-62175. Receivedjilrpublimtion March 20. 1989: rev~ledlanllury 16,1990; Ilccepted Februmy 5,1990. Reprints not available.

For transcervical aspirations, a 20-gauge, 6-inch needle and introducer of the type used for pudendal nerve block were used. Fluid in amounts of 2 to 6 ml was obtained at each sampling. Specimens visibly contaminated with blood were not used to ensure accuracy of the serial dilutions described below. Each amniotic fluid specimen was paired with a cord blood specimen obtained within the same 24-hour period. Blood (2 to 6 ml) was drawn from a cord vessel immediately after placental delivery by means of a needle and syringe wetted with I: 1000 heparin sodium. Blood and amniotic fluid from the same pregnancy were paired when possible; otherwise, blood from another third-trimester delivery was substituted. The amniotic fluid was stored at 4° C until the corresponding blood sample was obtained. Serial dilutions of blood containing 10% to 90% amniotic fluid (by volume) in 10% increments then were prepared with a calibrated micropipette. Net volume per dilution was 100 ILL Aliquots of about 70 ILl were drawn into preheparinized microhematocrit capillary tubes. In addition, one tube each of pure blood and pure amniotic fluid were prepared. The tubes were sealed and centrifuged for 6 minutes at 12,000 rpm. The hematocrit of the pure blood specimen was measured. Each tube was then cut with wire strippers at the cell-supernatant interface, and the cellular fraction was discarded. Two drops of supernatant were shaken onto a microscope slide from the uncut end of the tube. The fluid was smeared with a coverslip, and dried under a 100 W incandescent lamp at a distance of 4 to 6 inches. The entire smear on each slide was examined under low power (x 32), advancing to X 80 or x 200 as

1207

1208 Chao, Herd, and Tabsh

May 1990

Am J Obstet Gynecol

Fig. lAo "Crystalline" pattern, 100% plasma. (Original magnification x 80.)

Fig. lB. "Curling" pattern, 100% plasma. (Original magnification x 80.)

needed to characterize specific patterns. Slides were classified by the degree of complexity of their geometric patterns. After a four-grade classification system was established based on examination of these first 10 specimens, a blind trial was performed to test the accuracy of the system. From each of 8 new plasma-amniotic fluid mixtures, eight slides (two of 0%, one each of 10%, 20%, 30%, 40%,50%, and 60%) were prepared. The labels were

masked with tape, and the slides were read in assorted sets of 16 or 24 by two of us (A. C. and J. H.) working independently. Our predicted grade was then compared to the actual grade, and the specificity, sensitivity, and efficiency of the method in detecting contamination were determined. Results

The gestational ages of the 16 donors contributing to the first 10 mixtures ranged from 31 to 41 weeks,

Ferning and umbilical blood samples

Volume 162 Number 5

1209

Fig. Ie. "Thorn" (left) and "pseudoferning" (right) patterns, 100% plasma. (Original magnification x 32.)

Table I. Characteristics of ferning patterns Grade

o

% Aminiotic fluid

o 10

20

Pattern of foreign particles

Thorn Secondary branching patterns contiguous with particles As in 10%, plus discrete secondary branching near foreign particles

II

30

Discrete tertiary complexes

II

40

Discrete teritary complexes

;;,:50

Discrete tertiary complexes

III

with a median of 39.5 weeks. The cord blood hematocrit range was 42% to 58% (median 46%). Pure plasma most often presents as a background of fine polygonal crystals of under 10 IJ..m diameter. In addition, four other patterns that must be distinguished from true ferning are seen in pure plasma. First, geometric "crystalline" figures of 200 to 400 IJ..m in diameter (Fig. lA) were seen in five of 10 samples. Second, "curling," elongated structures up to 300 IJ..m in length (Fig. IB) were seen in six samples. These figures have a broad central spine and a wide serrated border, and some produce starfish-like patterns. Third, "pseudoferning," branching complexes of 200 to 500 IJ..m diameter (Fig. I C) were seen in five samples. These differ from true ferning by having

Background pattern

Crystalline, curling, and/or pseudoferning Primary branching with :s4 terminal points Primary branching with many 4- and 5-terminalpoint figures Simple secondary complexes, >5 terminal points Simple secondary complexes, few tertiary Multiple fields filled with tertiary complexes

broad serrated branches (rather than narrow ones with smooth edges), and by lacking a narrow central spine that appears alternately bright or dark as the fine-focus is adjusted. Finally, thickly interlaced, linear complexes surrounding foreign particles were seen in seven samples (Fig. 1C). These complexes may have secondary branching, but are always in direct continuity with the foreign particle and do not form freestanding adjacent complexes. The foreign particles were thought to be dust, cellular debris, and glass fragments from cutting the capillary tubes. The preceding patterns were seen with decreasing frequency as amniotic fluid content increased. They were absent from all specimens at 50% contamination or greater.

1210 Chao, Herd, and Tabsh

May 1990

Am J Obstet Gynecol

Fig. 2. Pattern around a foreign particle, 10% amniotic fluid. Note the loose pattern caused by lack of tertiary buds. (Original magnification x 200.)

Fig. 3. Background pattern, 10% amniotic fluid. Note the majority of figures terminate in four or fewer points. Narrow arrow indicates a primary branching pattern with three terminal points. Broad arrow indicates a secondary branching pattern with five terminal points. (Original magnification x80.)

With increasing amniotic fluid content, pnmltlve ferning patterns are first seen around the foreign particles. At any given level of contamination, the pattern of the background (that is, those areas at least one x 200 field away from foreign particles) is less complex than the pattern around foreign particles. The characteristic patterns are summarized in Table

I. A pattern is described only if present in six or more

of the 10 specimens at that dilution. Examples of the ferning patterns at various dilutions are shown in Figs. 2 to 6. A fully developed ferning pattern has primary stems radiating from a central point, secondary branches arising from the primaries, and tertiary buds from the secondary branches.

Ferning and umbilical blood samples

Volume 162 Number 5

1211

Fig. 4. Pattern around a foreign particle. 30% amniotic fluid. Note the discrete, closely packed fern clusters with tertiary buds. (Original magnification x 200.)

Fig. 5. Background pattern, 40% amniotic fluid. Note the prominent secondary branching, with well over five terminal points per complex. Arrow indicates an example of a secondary branching pattern. (Original magnification x 80.)

To simplify the clinical application of this method, a four-grade system was devised. Grade 0 represents 0% contamination, grade I includes 10% and 20%, grade II includes 30% and 40%, and grade III includes 50% and higher. Ferning patterns within each grade are quite similar. The blind trial to evaluate the grading system included two readings from each of 64 slides from eight

samples, for a total of 128 readings. The efficiency for assigning the correct grade to each slide was 80% (102/128). All incorrect diagnoses varied by one grade from the correct grade. To evaluate this method for the detection of any contamination of 10% or more, the grades were transformed into two nominal groups: not contaminated (grade 0), and contaminated (grades I, II, or III). Thus

1212 Chao, Herd, and Tabsh

May 1990 Am J Obstet Gynecol

Fig. 6. Background pattern, 50% amniotic fluid. Two different types of ferning patterns are seen. Arrows indicates an example of a tertiary budding pattern. (Original magnification x 80.)

the SenSItivlty 10 detecting contamination was 98% (94/96); the specificity was 81 % (26/32), and the efficiency was 94% (120/128). Comment

Daffos et al.· reported specimen contamination with amniotic fluid or sodium citrate anticoagulant in 2.4% of 606 fetal blood specimens, although the method of determination was not described. Contamination presumably occurs during aspiration if the needle tip is displaced from the vessel lumen, or if the puncture wound is large enough to permit ingress of fluid. A color change caused by amniotic fluid in the specimen may be hard to detect, particularly if contamination occurs in midaspiration. In this experiment the tint of the mixture was not remarkably lighter until grade III contamination was present. We therefore suggest that the ferning preparation be used as a quality assurance test for each cordocentesis specimen intended for quantitative analysis. The materials required are inexpensive and readily available. Only 50 to 70 fl.l of the specimen would be needed, using a capillary tube applied directly to the syringe tip after gentle mixing. The method has good sensitivity and efficiency in detecting contamination of 10% or more. When evidence of significant contamination accompanies quantitative abnormalities in the blood specimen, a repeat sampling should be considered. The specificity of 80% reflected a tendency to overdiagnose grade 0 slides as grade I. In retrospect, this error was primarily a result of overinterpretation of the

thorn-like pattern around foreign particles. Even when the branching appeared more complex than expected of grade 0, such patterns appeared less frequently (once or twice) on true grade 0 slides than on grade I. It is therefore important that the entire smear be examined. We recommended that clinicians who choose to use this technique make their own slides from in vitro mixtures for practice instead of relying solely on the photomicrographs provided. These unstained slides cannot be used for permanent reference because the patterns disintegrate or become coarse after standing for several days. Centers that use sodium citrate for umbilical blood sampling should substitute that anticoagulant for heparin in making the practice slides. For use in percutaneous umbilical blood sampling, we do not recommend attempts to correct quantitative data on the basis of the estimated degree of contamination because there is no assurance of uniform mixing of amniotic fluid throughout the specimen in the syringe. This method could be useful in second-trimester umbilical blood sampling, particularly in management of the Rh-immunized pregnancy. Amniotic fluid is capable of ferning as early as 9 weeks' gestation." To show that the same grading system is appropriate, the in vitro experiment should preferably be repeated with secondtrimester amniocentesis samples and cord blood obtained directly, as from prostaglandin-induced midtrimester abortuses. An alternative proposal for indirect detection of amniotic fluid contamination involves the observation of concurrent low hematocrit, white blood cell count, and

Ferning and umbilical blood samples

Volume 162 Number 5

platelet count in the specimen. The amount of contamination required to lower all three parameters below the normal limits has not been described in the literature. In this study complete blood counts were not performed on the contaminated specimens. The ferning test may also be useful for the rapid confirmation of suspected ruptured membranes in the presence of vaginal bleeding, as in cases of placenta previa. The method would be much faster and more economical than assaying the vaginal specimen for a-fetoprotein or prolactin. In summary, the ferning test is a simple, inexpensive, rapid, and accurate means of detecting amniotic fluid contamination of blood specimens, and holds promise for quality control application to the technique of percutaneous umbilical blood sampling.

REFERENCES 1. Ludomirski A, Weiner S. Percutaneous fetal umbilical blood sampling. Clin Obstet Gynecol 1988;31:19-26. 2. Brookes C, Shand K, Jones WR. A reevaluation of the ferning test to detect ruptured membranes. Aust N Z J Obstet Gynaecol 1986;26:260-4. 3. Borten M, Friedman EA. Amniotic fluid ferning in early gestation. AM J OSSTET GY»;ECOL 1986; 154:628-30. 4. Daffos F, Capella-Pavlovsky M, Forestier F. Fetal blood sampling during pregnancy with use of a needle guided by ultrasound: a study of 606 consecutive cases. AM J OSSTET GY;\IECOL 1985; 153:655-60. 5. Kovacs D. Crystallization test for the diagnosis of ruptured membranes. AM J Os STET GY1';ECOL 1962;83: 1257-60.

Is pH test paper as accurate as the electronic measurement of the pH of vaginal secretions? Jessica L. Thomason, MD, Sheldon M. Gelbart, PhD, Lisa M. Monagle, RN, Janine A. James, MD, and Fredrick F. Broekhuizen, MD

Milwaukee, Wisconsin A comparison was made of two brands of pH test paper and electronic instrumentation for measuring the pH of vaginal secretions. When the pH of vaginal secretions was >4.5 (abnormal), there was no significant difference between the methods, showing that pH test paper is reliable for pH determination of vaginal secretions. (AM J OSSTET GVNECOL 1990;162:1213-4.)

Key words: pH, vaginal secretions Increased pH of vaginal secretions is an important sign that aids in the diagnosis of several vaginal disorders (e.g., bacterial vaginosis and trichomoniasis).1 Vaginal pH is generally determined by the use of test papers as recommended by Gardner and Dukes' in 1955. Despite vast improvements in technology of both pH instrumentation and test papers, a comparison of these methods for measuring vaginal pH has not been reported since 1955.' The purpose of this study was to compare two types of pH test papers to state-of-the-art pH instrumentation to determine how reliable pH test From the Department of Obstetrics and Gynecology, University of Wisconsin Medical School. Receivedfor publication July 20,1989; accepted December 4.1989. Reprint requests,' J. L. Thomason, M D, Division of Gynecology, Sinai Samaritan Medical Center, P.O. Box342, Milwaukee, W1532010342. 611118610

paper measurements were when used in the clinical situation. Vaginal secretions were collected from 64 randomly selected, non menstruating, premenopausal women, who had given consent. Samples were obtained from the lateral vaginal fornix on three dry cotton-tipped swabs. Care was taken to avoid any cervical mucus. Swabs were unlabeled when presented to the technician. This was done to obtain randomization of sampling because the technician did not know the order in which the swabs were obtained. Vaginal secretions were tested immediately for pH by the following methods: one swab was touched to the surface of Nitrazine pH test paper (E. R. Squibb & Sons Inc., Princeton, N.J.). The pH was read according to the color indicated per manufacturer's instructions. A second swab was probed with an MI415 Microcombination pH probe (Microelectrodes, Inc., Londonderry, N.H.), and the pH was 1213