Received October 24, 2001; accepted February 8, 2002. 5 Nonstandard abbreviations: RDS, respiratory distress syndrome; FLM, fetal lung maturity; L/S, ...
Clinical Chemistry 48:5 761–765 (2002)
General Clinical Chemistry
Assessment of the Diagnostic Accuracy of the TDx-FLM II to Predict Fetal Lung Maturity Corinne R. Fantz,1 Cynthia Powell,2 Brad Karon,1* Curtis A. Parvin,1 Kelly Hankins,1 Molina Dayal,3 Yoel Sadovsky,3 Vandita Johari,4 Fred S. Apple,4 and Ann M. Gronowski1†
Background: Because respiratory distress syndrome (RDS) affects 1% of live births, accurate and rapid assessment of markers of fetal lung maturity is critical to clinicians in deciding whether to deliver a preterm infant. Our objective was to determine the optimal diagnostic cutoff value for the TDx-FLM II assay (Abbott Laboratories) for predicting clinically significant RDS. Methods: Amniotic fluid TDx-FLM II data were collected retrospectively over 4 years. Women were included in the study if they had delivered within 72 h of TDx-FLM II testing and both the mother and infant charts could be reviewed. Women who had been treated with steroids and delivered unaffected infants were excluded from the analysis. The diagnosis of RDS was defined as infants who either were treated with surfactant and/or were placed on a ventilator and/or required continuous positive airway pressure for >1 day. Results: A total of 185 women met all entry criteria (15 RDS, 170 non-RDS). A cutoff value for a mature result of >45 mg/g gave a sensitivity of 100% (95% confidence interval, 82–100%) and a specificity of 90% (95% confidence interval, 78 – 89%). Conclusions: The TDx-FLM II appears to predict clinically significant RDS when a cutoff of >45 mg/g is used
for mature results. Further studies will be required to confirm these findings. © 2002 American Association for Clinical Chemistry
Respiratory distress syndrome (RDS)5 is a major cause of death in the newborn. In 1998, RDS ranked fourth among leading causes of infant deaths, with 1328 deaths from RDS in the United States (1 ). RDS, also referred to as hyaline membrane disease, is associated with insufficient surfactant production by the neonatal lung. Laboratory assessment of fetal lung maturity (FLM) before delivery enables clinicians to either delay delivery and administer steroids to hasten fetal lung development or to elect to deliver a preterm infant whose lungs are mature. The biochemical markers for assessment of FLM are the lecithin-to-sphingomyelin ratio (L/S) and phosphatidylglycerol (PG) in amniotic fluid, determined by thinlayer chromatography (2–7 ). Chromatography, although considered the gold standard, is technically difficult and time-consuming. Alternative, more rapid assays have emerged, including foam stability, lamellar body count, PG agglutination, and fluorescence polarization. The foam stability index test is fast, easy to perform, inexpensive, and sensitive when compared with L/S and PG, but it is less specific, and there are differences in operator interpretation of results (8, 9 ). Similarly, PG agglutination is also fast and easy to perform, but there is inconsistency in test interpretation (3, 6 ). The lamellar body count test takes advantage of the size similarities between platelets and the lamellar bodies in the amniotic fluid and allows for counting of lamellar bodies on automated cell counters. For this reason, this assay is widely available, requires a low sample volume, and is fast, easy, and
1 Department of Pathology and Immunology, Division of Laboratory Medicine, 2 Department of Pediatrics, and 3 Department of Obstetrics and Gynecology, Washington University School of Medicine, 660 South Euclid Ave., Box 8118, St. Louis, MO 63110. 4 Department of Laboratory Medicine and Pathology, Hennepin County Medical Center, Minneapolis, MN 55415. *Current address: Laboratory Medicine Consultants, Sunrise Hospital and Medical Center, Las Vegas, NV 89109. †Author for correspondence. Fax 314-362-1461; e-mail gronowski@ pathology.wustl.edu. Received October 24, 2001; accepted February 8, 2002.
5 Nonstandard abbreviations: RDS, respiratory distress syndrome; FLM, fetal lung maturity; L/S, lecithin-to-sphingomyelin ratio; PG, phosphatidylglycerol; CPAP, continuous positive airway pressure; PG/S, phosphatidylglycerol-to-sphingomyelin ratio; and CI, confidence interval.
761
762
Fantz et al.: Diagnostic Efficiency of TDx-FLM II
inexpensive to perform. However, there are no standardized protocols, quality-control material, or well-established cutoffs for this assay (10 –13 ). The TDx-FLM is an automated fluorescence polarization assay that measures mg surfactant/g of albumin in amniotic fluid and reports quantitative results (14 –18 ). Studies evaluating the first-generation TDx-FLM I assay indicated that the ⱖ70 mg/g cutoff for a mature result suggested by the manufacturer was conservative and could be lowered without compromising clinical sensitivity (14, 15, 19, 20 ). The TDx-FLM II is a second-generation fluorescence polarization assay in which the manufacturer’s cutoffs have been set as follows: mature ⱖ55 mg/g, intermediate 40 –54 mg/g, and immature ⱕ39 mg/g. Preliminary studies examining the TDx-FLM II assay have indicated that the manufacturer’s cutoff for this assay may also be set conservatively (21–23 ). For this reason, we initiated a collaborative evaluation of the TDx-FLM II assay to determine a cutoff that would not miss prediction of clinically significant RDS. This is the largest study of its kind to date.
Materials and Methods This study was a collaborative, retrospective cohort study in which a consecutive series of physician-ordered TDxFLM II (Abbott Laboratories) samples, collected during a 4-year period, were examined. Data were collected from the charts of patients who had TDx-FLM II analysis performed at either Barnes-Jewish Hospital (St. Louis, MO) during 1998 –2000 or Hennepin County Medical Center (Minneapolis, MN) during 1998. During this time, 1104 physician-ordered FLM tests were performed on 881 different patients. FLM testing was ordered on women presenting with signs and symptoms of preterm labor as well as asymptomatic women before scheduled cesarean sections. Inclusion in the study required that delivery occurred within 72 h of amniocentesis and that charts for both mother and infant were reviewable (see below). Women who received steroids before delivery and delivered nonRDS infants were excluded from the study because it is impossible to determine whether the steroids impacted the respiratory status of the infant. Women who received steroids before delivery and delivered RDS infants were included in the study because it is clear that the steroids did not impact the presence of RDS. Indeed, steroids may have changed the severity of RDS, but data were analyzed only for the presence or absence of RDS. Barnes-Jewish Hospital performs FLM testing for a total of 12 hospitals, so we were unable to obtain charts for many of the 881 patients. Charts were available on both the mother and infant for 281 (32%) patients. Nineteen of these patients delivered ⬎72 h from the time of FLM sampling, giving a total of 262 patients. Of these, 77 were treated with steroids and delivered non-RDS infants. Therefore, 185 patients met the inclusion criteria. Institu-
tional Review Board approval was obtained for these studies at both institutions. Infants who were treated with surfactant and/or were placed on a ventilator and/or required continuous positive airway pressure (CPAP) for ⬎24 h with no other congenital anomalies were given a diagnosis of clinically significant RDS. Indications for administering surfactant include prematurity, radiologic signs of hyaline membrane disease, and mechanical ventilation or CPAP. Six sets of twins were included in the study (three RDS, three unaffected). In all six cases only one amniotic fluid sample was obtained. However, in all six cases both twins had the same clinical outcome. Therefore, the data were included in the study although each set of twins was counted only as a single data point. Amniotic fluid was collected by transabdominal amniocentesis or free flowing vaginal pool if the membranes were ruptured. Both types of specimens are considered acceptable for analysis by the manufacturer (24 ). Samples with visible bilirubin, blood, or meconium were rejected. TDx-FLM II analysis was performed according to the manufacturer’s instructions. The upper limit of detection was 160 mg/g, and the lower limit of detection was 10 mg/g. The L/S and the phosphatidylglycerol-to-sphingomyelin ratio (PG/S) were determined by an outside reference laboratory (Quest Diagnostics, Saint Louis, MO) by thin-layer chromatography (4, 25, 26 ). Reference laboratory cutoffs for L/S were as follows: immature, ⬍1.0; premature, 1.0 –1.5; intermediate, 1.5–1.9; caution mature, 2.0 –2.5; mature, ⬎2.5. Reference laboratory cutoffs for the PG/S were as follows: immature, ⬍0.3; mature, ⬎0.3.
statistics TDx-FLM II cutoff values were established with the criterion that there be no false mature results. Sensitivity refers to the probability of an immature result in a patient with RDS. Specificity refers to the probability of a mature result in a patient without RDS. Confidence intervals (CIs) for sensitivity and specificity were computed as exact binomial 95% CIs, using Stata statistical software, release 7.0 (Stata Corp.).
Results A total of 185 women met all the inclusion criteria, of whom 170 delivered unaffected infants and 15 delivered infants with RDS. The gestational ages at time of delivery ranged from 31 to 40 1/7 weeks (median, 37 weeks) for unaffected infants and 28 to 38 6/7 weeks (median, 33 4/7 weeks) for RDS infants. The frequency of all the TDx-FLM II results based on RDS diagnosis is shown in Fig. 1. The data demonstrate that none of the women who delivered infants with RDS had amniotic TDx-FLM II values ⱖ45 mg/g. Women who delivered unaffected infants and had not been treated with steroids had FLM II results ranging from 14.6 to ⬎160 mg/g. As shown in Table 1, the 15 women who delivered infants with RDS had TDx-FLM II values that ranged
763
Clinical Chemistry 48, No. 5, 2002
Fig. 2. Relationships between TDx-FLM II values and L/S. Fig. 1. Frequency of TDx-FLM II results based on infant respiratory diagnosis. f, women who delivered infants with RDS (n ⫽ 15); 䡺, women who delivered infants without RDS (n ⫽ 170).
from ⬍10 to 40.7 mg/g. Eight of these women had L/S and/or PG/S values obtained concomitantly with the TDx-FLM II. The TDx-FLM II values correlated well with the L/S and PG/S data in five of the eight women. However, three of the women who delivered infants with RDS (patients 10, 11, and 14) had L/S in the caution mature or mature range, with TDx-FLM II values in the immature range. Patient 14 also had a mature PG/S (⬎0.3). In 56 cases, L/S and TDx-FLM II were measured concomitantly (Fig. 2). These data demonstrate fair correlation between the two tests (r ⫽ 0.73), although as shown in Table 1, three RDS patients tested falsely mature by L/S but correctly immature by the TDx-FLM II. Listed in Table 2 are the sensitivity (95% CI), specificity (95% CI), predictive value of a mature result, and predic-
RDS (U), n ⫽ 8; non-RDS (E), n ⫽ 47. Patients (pt) 10, 11, and 14 are indicated by arrows. The dashed line represents the mature cutoff value for L/S. L/S cutoffs: immature, ⬍1.0; premature, 1.0 –1.5; intermediate, 1.5–1.9; caution mature, 2–2.5; mature, ⬎2.5. The diagonal solid line represents the linear regression line (y ⫽ 14.6x ⫹ 8.9; r ⫽ 0.73).
tive value of an immature result for cutoff values ranging from the current cutoff of ⱖ55 mg/g down to ⱖ40 mg/g. As shown in the ROC curve (Fig. 3), lowering the mature cutoff value from ⱖ55 mg/g to ⱖ45 mg/g improved specificity, from 72% to 84%, without compromising the sensitivity of 100%. Sensitivity decreased to 93.3% when the cutoff was lowered to ⱖ40 mg/g. In addition, this change in cutoff increased the predictive value of an immature result from 24% to 36% (Table 2).
Discussion In the assessment of women who are at risk for preterm delivery, the laboratory plays an important role in predicting FLM. Following the development of the automated TDx-FLM method, we eliminated routine perfor-
Table 1. TDx-FLM II results for infants with RDS. Patient
1 2 (twins) 3 4 5 6 7 8 (twins) 9 10 11 (twins) 12 13 14 15 a
ND, not determined.
Gestational age, weeks
34 30 34 34 29 34 29 28 31 33 32 36 30 33 38
3/7 4/7
2/7
4/7 3/7 4/7 4/7 6/7
FLM II, mg/g
L/S ratio
PG/S ratio
Treatment, type (days)
⬍10 ⬍10 ⬍10 ⬍10 ⬍10 11.2 12 14.6 14.8 17.5 23.4 23.4 27 29 40.7
NDa 1.2 ND 1.7 0.8 1.2 ND ND 1.6 2.5 2.5 ND ND 4.6 ND
ND ⬍0.3 ND ⬍0.3 ⬍0.3 ⬍0.3 ND ND ND ⬍0.3 ND ND ND ⬎0.3 ND
Surfactant, ventilator (3) Surfactant, ventilator (3 and 6) Surfactant, ventilator (⬎1) Surfactant, CPAP (3) Surfactant, ventilator (⬎1) Ventilator (2) CPAP (2) Surfactant, CPAP (5 and 6) CPAP (1) Surfactant, ventilator (5) Surfactant, ventilator (4 and 4) Ventilator (3) Surfactant, ventilator (1) Ventilator (⬎1) Surfactant, ventilator (⬎7)
764
Fantz et al.: Diagnostic Efficiency of TDx-FLM II
Table 2. Diagnostic efficiency of the TDx-FLM II in identifying clinically significant RDS. Predictive value, % Cutoff, mg/g
Sensitivity, % (CI)
Specificity, % (CI)
Mature result
Immature result
ⱖ55 ⱖ50 ⱖ45 ⱖ40
100 (82–100) 100 (82–100) 100 (82–100) 93 (68–100)
72 (65–79) 78 (71–84) 84 (78–89) 89 (83–93)
100 100 100 99
24 29 36 42
mance of the time-consuming and technically difficult thin-layer chromatography method for the L/S and PG/S. The manufacturer’s recommended cutoff value for prediction of FLM by the TDx-FLM I was originally set at ⱖ70 mg/g. Studies suggested that this cutoff could be safely lowered to ⱖ55 mg/g, thereby increasing specificity without compromising sensitivity (14, 15, 19, 20 ). Preliminary studies of the second-generation TDx-FLM II assay, presented as abstracts, suggested that the manufacturer’s cutoff for this assay could also be lowered to between 40 and 44 mg/g (21–23 ). We conducted a retrospective cohort study of TDx-FLM II results with 4 years of data to evaluate the ability of this assay to detect clinically significant RDS at different cutoff values. New TDx-FLM II cutoff values were determined, with the criterion that there be no false mature results (i.e., 100% predictive value of a positive result). Our results demonstrate that the current manufacturer’s recommended cutoff for the TDx-FLM II assay can be lowered from ⱖ55 to ⱖ45 mg/g. This change does not compromise sensitivity or predictive value of a positive result and at the same time improves specificity and predictive value of an immature result. This cutoff is slightly conservative: according to the ROC curve, the cutoff could be set as low as 41 mg/g without affecting predictive values and results. In 55 cases, a L/S was obtained concomitantly with the TDx-FLM II value. The overall correlation between L/S
Fig. 3. ROC curve for TDx-FLM II results. TDx-FLM II values from women who delivered 15 RDS and 170 non-RDS infants. Cutoffs of ⱖ40 mg/g, ⱖ45 mg/g, and ⱖ55 mg/g are indicated by the arrows. The dashed line represents the random guess line. Area under the curve, 0.98.
and TDx-FLM II values was fair (r ⫽ 0.73). However, in three women who delivered infants with RDS, the L/S was in the caution mature or mature range, and in one of these cases, the PG/S was also mature. In these cases, the TDx-FLM II was clearly a better indicator of FLM than either the L/S or the PG/S. To obtain Food and Drug Administration approval for the TDx-FLM II assay, Abbott Laboratories conducted a study with 306 patients (292 non-RDS and 14 RDS) with the highest FLM II value for an RDS patient equal to 40.33 mg/g. Using a cutoff of ⱖ55 mg/g, they obtained a sensitivity of 100%, specificity of 75%, a predictive value of maturity of 100%, and a predictive value of immaturity of 16% (data obtained through the Freedom of Information Act from the Department of Health and Human Services, Food and Drug Administration). If the revised cutoff of ⱖ45 mg/g is applied to their data, the sensitivity and predictive value of a mature result remain 100%, the specificity increases to 90%, and predictive value of an immature result increases to 33%. These data support our premise that the diagnostic cutoff can be lowered to ⱖ45 mg/g. Although the largest study of its kind to date, it should be noted that our data are based on a relatively small population and that caution should be used in altering the cutoff based on a single study. For this reason, further studies are indicated to confirm our results in a population with a sufficiently broad spectrum of disease. Nevertheless, in our population, no case of clinically significant RDS had a TDx-FLM II result ⬎40.7 mg/g.
This research was funded in part by a grant from Abbott Laboratories (Abbott Park, IL).
References 1. Guyer B, Hoyert DL, Martin JA, Ventura SJ, MacDorman MF, Strobin DM. Annual summary of vital statistics-1998. Pediatrics 1999;104:1229 – 46. 2. Ashwood ER. Standards of laboratory practice: evaluation of fetal lung maturity. Clin Chem 1997;43:211– 4. 3. Garite TJ. Fetal lung maturity testing. Clin Obstet Gynecol 1987; 30:985–91. 4. Gluck L, Kulovich MV, Borer RC, Brenner PH, Anderson GG, Spellacy WN. Diagnosis of the respiratory distress syndrome by amniocentesis. Am J Obstet Gynecol 1971;109:440 –5. 5. Hallman M, Kulovich MV, Kirkpatrick E, Sugarman RG, Gluck L. Phosphatidylinositol and phosphatidylglycerol in amniotic fluid: indices of lung maturity. Am J Obstet Gynecol 1976;125:613–7. 6. American College of Obstetricians and Gynecologists. Assessment of fetal lung maturity [ACOG Educational Bulletin]. Washington, DC: American College of Obstetricians and Gynecologists, 1996:1– 8. 7. Dubin SB. Assessment of fetal lung maturity [Practice Parameter]. Am J Clin Pathol 1998;110:723–32. 8. Schlueter MA, Phibbs RH, Creasy RK, Clements JA, Tooley WH. Antenatal prediction of graduated risk of hyaline membrane disease by amniotic fluid foam test for surfactant. Am J Obstet Gynecol 1979;134:761–7. 9. Clements JA, Platzker AG, Tierney DF, Hobel CJ, Creasy RK,
Clinical Chemistry 48, No. 5, 2002
10.
11.
12.
13.
14.
15. 16.
17.
Margolis AJ, et al. Assessment of the risk of the respiratory distress syndrome by a rapid test for surfactant on amniotic fluid. N Engl J Med 1972;286:1077– 81. Ashwood ER, Palmer SE, Taylor JS, Pingree SS. Lamellar body counts for rapid fetal lung maturity testing. Obstet Gynecol 1993;81:619 –24. Delance CR, Bowie LJ, Dohnal JC, Farrell EE, Neerhof MG. Amniotic fluid lamellar body count: a rapid and reliable fetal lung maturity test. Obstet Gynecol 1995;86:235–9. Lewis PS, Lauria MR, Dzieczkowski J, Utter GO, Dombrowski MP. Amniotic fluid lamellar body count: cost effective screening for fetal lung maturity. Obstet Gynecol 1999;93:387–91. Piazze JJ, Anceschi MM, Maranghi L, Porpora MG, Cosmi EV. The biophysical/biochemical test. A new marker of fetal lung maturity in borderline cases. J Reprod Med 1999;44:611–5. Apple FS, Bilodeau L, Preese LM, Benson P. Clinical implementation of a rapid automated assay for assessing fetal lung maturity. J Reprod Med 1994;39:883–7. Herbert W, Chapman JF, Schooner MM. Role of the TDx FLM assay in fetal lung maturity. Am J Obstet Gynecol 1993;168:808 –12. Blumenfeld TA, Stark RI, James LS, George JD, Dyrenfurth I, Freda VJ, et al. Determination of fetal lung maturity by fluorescence polarization of amniotic fluid. Am J Obstet Gynecol 1978;130: 782–7. Hagen E, Link JC, Arias F. A comparison of the accuracy of the TDx FLM assay, lecithin-sphingomyelin ratio, and phosphatidylglycerol in the prediction of neonatal respiratory distress syndrome. Obstet Gynecol 1993;82:1004 – 8.
765
18. Bonebrake RG, Towers CV, Rumney PJ, Reimbold P. Is fluorescent polarization reliable and cost efficient in a fetal lung maturity cascade? Am J Obstet Gynecol 1997;177:835– 41. 19. Steinfeld JD, Samuels P, Bulley MA, Cohen AW, Goodman DBP, Senior MB. The utility of the TDx test in the assessment of fetal lung maturity. Obstet Gynecol 1992;79:460 – 4. 20. Russell JC, Cooper CM, Ketchem CH, Torday JS, Richardson DK, Holt JA, et al. Multicenter evaluation of TDx test for assessing fetal lung maturity. Clin Chem 1989;35:1005–10. 21. McManamon TG, Mandsager N. Clinical outcome of neonates following the use of the FLM-II as the only test for fetal lung maturity [Abstract]. Clin Chem 1998;44:A157. 22. Moxness M, Lawson G, O’Brien J, Burritt M. Assessment of fetal lung maturity by the Abbott FLM II assay and three other methods [Abstract]. Clin Chem 1995;41:S95. 23. Dunston-Boone R, Chang T, Bernstein LH. Modified TDx assay for surfactant to albumin ratio in the assessment of fetal lung maturity [Abstract]. Clin Chem 1997;43:S196. 24. Edwards RK, Duff P, Ross KC. Predictive value of fetal lung maturity testing in amniotic fluid samples collected vaginally from patients with preterm PROM [Abstract]. Am J Obstet Gynecol 2000;182:S180. 25. Gluck L, Kulovich MV, Borer RC. Estimates of fetal lung maturity. Clin Perinatol 1974;1:125–39. 26. Gluck L, Kulovich MV, Borer RC, Keidel WN. The interpretation and significance of the lecithin/sphingomyelin ratio in amniotic fluid. Am J Obstet Gynecol 1974;120:142–55.