- Email: [email protected]
Amniotic fluid phosphatidylglycerol and phosphatidylcholine phosphorus as predictors of fetal lung maturity
Amniotic fluid phosphatidylglycerol and phosphatidylcholine phosphorus as predictors of fetal lung maturity D. P. Kogon, M.D., M. Oulton, Ph.D., J. H. Gray, R.N., R. M. Liston, M.B., E. R. Luther, M.D., L. J. Peddle, M.D., and D. C. Young, M.D. Halifax, Nova Scotia, Canada The contents of phosphatidylglycerol and phosphatidylcholine phosphorus in amniotic fluid (10,000 x 9 pellets) were studied as predictors of fetal lung maturity. The presence of phosphatidylglycerol predicted the absence of neonatal respiratory distress syndrome with 99% probability. When phosphatidylglycerol was absent, phosphatidylcholine phosphorus was a reliable predictor if measured 3 to 7 days before delivery. The probability that respiratory distress syndrome would not occur was 94% when phosphatidylcholine phosphorus was >6. When measurement was performed within 2 days of delivery, the probability that respiratory distress syndrome would not occur fell to 69%. As measured in amniotic fluid, phosphatidylglycerol and phosphatidylcholine phosphorus are reliable antenatal predictors of fetal pulmonary maturity and, therefore, are useful in the management of a number of obstetric conditions. (AM J OSSTET GVNECOL 1986;154:226-30.)
Key words: Amniotic fluid, fetal lung maturity, phosphatidylglycerol, phosphatidyicholine phosphorus, respiratory distress syndrome
In a variety of obstetric settings delivery of a fetus before term is desirable. However, the benefit sought by preterm delivery must be measured against the risk of neonatal respiratory distress syndrome (RDS) and its related morbidity and mortality. Several methods of amniotic fluid analysis such as the lecithin/sphingomyelin ratio, the shake test, and, more recently, phosphatidylglycerol determinations are currently available for the assessment of fetal lung maturity."" Some investigators 2·5 have reported that a combination of tests, for instance, the lecithin/sphingomyelin ratio and phosphatidylglycerol determination, improve the predictive accuracy of any single test and, therefore, advocate such a combination of testing. In our center we have been performing phospholipid analyses on isolated surfactant fractions of amniotic fluid (10,000 X g pellets) since 1978, and, as previously reported,'O we found that the phosphatidylglycerol determination on these fractions was superior to both the lecithin/sphingomyelin ratio and the shake test in predicting fetal pulmonary maturity as well as immaturity. We have accepted the presence of phosphatidylglycerol in these fractions as being indicative of fetal pulmonary maturity but have also found many instances in which
From the Division of Perinatal Medicine, Department of Obstetrics and Gynaecology, Dalhousie University. Presented at the Forty-first Annual Meeting of The Society of Obstetricians and Gynaecologists of Canada, jasper, Alberta, Canada, june 10-15, 1985. Reprint requests: Dr. D. P. Kogon, Division of Perinatal Medicine, Department of Obstetrics and Gynaecology, Dalhousie University, 5821 University Ave., Halifax, Nova Scotia, Canada B3H 1W3.
226
the absence of phosphatidylglycerol was not predictive of the development of RDS. We subsequently demonstrated that the measurement of the phosphatidyicholine content of the same isolated surfactant fractions of amniotic fluid provides an index for the rate of surfactant accumulation." With this measurement expressed as the phosphatidyicholine phosphorus value (phosphatidyicholine phosphorus content of the surfactant present in 10 ml of amniotic fluid), it was found that this value progressively increased before the appearance of phosphatidylglycero!. The hypothesis that this measurement may be of value in assessing the state of fetal pulmonary maturity in the absence of phosphatidylglycerol was therefore addressed. Material and methods A review of all phospholipid profiles analyzed between 1978 and 1983 was made. Clinical variables such as intrauterine growth retardation, premature rupture of the membranes, diabetes mellitus, and the use of steroids were not excluding factors. Phospholipid profiles were measured on isolated surfactant fractions as previously described. 10. II Briefly, this involves centrifugation of a carefully measured aliquot (preferably 5 to 10 ml) of each amniotic fluid sample for 5 minutes at 140 X g to remove cellular debris followed by a 20minute spin at 10,000 X g to pellet the surfactant particles. Lipids were then extracted from the surfactant pellets and separated by two-dimensional thin-layer chromatography. The individual phospholipids, after visualization with the use of a modified molybdenum
Predictors of fetal lung maturity 227
Volume 154 Number 2
spray reagent, were scraped from the plate and quantitated by measurement of their phosphorus content. When volumes other than 10.0 ml were used, the phosphorus content of phosphatidylcholine was appropriately corrected to give the phosphatidylcholine phosphorus value. Phosphatidylglycerol is routinely calculated as a percentage of total lipid phosphorus but is reported in this communication as either present or absent. By this method the limit of detection for phosphatidylglycerol is 0.2 to 0.5 j.Lg. All amniocenteses were done to assess fetal lung maturity based on clinical indicators. None was done solely to collect data for the study. The decision to perform the amniocentesis was made by the individual attending obstetrician. All patient management decisions were also made by the attending obstetrician. In the earlier years of study, the shake test and the lecithin/sphingomyelin ratio were the standard accepted tests and were the basis for obstetric management decisions. After initial work on phosphatidylglycerol by Bent et aI., 10 phosphatidylglycerol became the standard test of fetal lung maturity at this institution. At no time was a phosphatidylcholine phosphorus value used to direct obstetric management of the patients in the study group. The study group then comprised patients who had fetal pulmonary testing performed ante natally for any indication regardless of that indication or other variables. The only exclusions were in patients with multiple gestations or patients whose infants died so soon after birth that RDS could not possibly develop. Clinical findings, indication for amniocentesis, gestational age, and neonatal outcome were recorded. All data were collected from the Grace Maternity Hospital obstetric population, which accounts for 5000 to 5500 deliveries per year. This hospital is the sole maternity unit for metropolitan Halifax and is the tertiary care obstetric center for the province of Nova Scotia and elsewhere in the Atlantic region. A total of 1678 amniotic fluid samples from 1359 patients were analyzed. Blood and/or meconium in the sample did not affect the assay. To reevaluate phosphatidylglycerol as a predictive test only transabdominal samples of amniotic fluid taken within 2 days prior to delivery were considered. To assess phosphatidylcholine phosphorus, the same criteria were applied. However, only samples where phosphatidylglycerol was absent were considered. Because of the steady rise in phosphatidylcholine phosphorus that occurs prior to the appearance of phosphatidylglycerol, the time frame in this group was extended to include samples collected 3 to 7 days before delivery as well as those within 2 days of delivery. RDS was diagnosed by the following criteria: grunting, tachypnea, sternal and intercostal retractions, and nasal flaring of ~6 hours' duration and requiring treatment with oxygen. Severity was classified by degree of
Table I. RDS in relation to phosphatidylglycerol determination «2 days before delivery) Phosphatidylglycerol
NoRDS
RDS
Present
592
10
Absent
28
Positive predictive value
= _a_ = a + b
Negative predictive value '" Sensltlvlty
_d_ c + d
=
c d ~
36
98.3%.
=
56.3%.
= -a- = 95.5 0170. a + c
Specificity = _d_ = 78.3%. b + d b False positive rate = - - = 21.7%. b + d .
c
False negative rate
Accuracy
=
Prevalence
= -- =
a + c
a+d a+b+c+d
=
a +
4.5%.
= 94.3%.
a + c b d +c +
= 93.1 %.
=
. . LIkelIhood ratio
sensitivity f' .. alse pOSItive rate
=
prevalence
93.1 %.
Pretest probability
=
= 4.4.
Posttest probability (from nomogram) = 99%.
respiratory assistance required in treatment. Severe RDS required assisted mechanical ventilation, moderate RDS required >35% oxygen and/or continuous positive airway pressure, and mild RDS required only an increase in oxygen concentration up to 35%. Diagnosis of RDS and its severity was made by the attending neonatalogist who was not aware of the results of the phospholipid profile. Analysis of the data was made with the use of standard diagnostic test parameters, that is, positive predictive value, negative predictive value, sensitivity, specificity, and accuracy. To further assess the predictive ability of the phosphatidylglycerol and phosphatidylcholine phosphorus measurements, likelihood ratios, and probability of RDS occurring or not occurring were determined. The key to the analysis is the likelihood ratio, which expresses the odds that a given diagnostic test result (presence of phosphatidylglycerol or level of phosphatidylcholine phosphorus) would be expected in a patient with the target outcome (no RDS). This is calculated in a two-by-two table as the true positive rate (sensitivity), divided by the false positive rate (one mi-
228
Kogan et al.
February, 1986 Am J Obstet Gyneco1
Table IV. RDS in relation to phosphatidylcholine phosphorus (3 to 7 days before delivery) (pretest probability 74.5%)
Table II. RDS in relation to phosphatidylcholine phosphorus «2 days before delivery) Phosphatidylcholine phosphorus
No RDS
RDS
2:6
18
10
Phosphatidylcholine phosphorus
<6
10
26
<2 2-6
c d ~
>6 Positive predictive value = 64%. Negative predictive value = 72%. Sensitivity = 64.3%. Specificity = 72%. False positive rate = 28%. False negative rate = 36%. Accuracy = 69%. Prevalence = 43.8%. Pretest probability = 43.8%. Likelihood ratio = 2.3. Posttest probability = 69%.
Table III. Significant RDS* in relation to phosphotidylcholine phosphorus «2 days before delivery) P Iwsphatidylcholine phosphorus
No significant RDS
2:6
25
<6
16
Significant RDS
No RDS
RDS
Likelihood ratio
2 10 29
8 4 2
0.09 0.85 4.9
Postlest probability of no RDS (%) 19 69 94
Table V. Significant RDS in relation to phosphatidylcholine phosphorus (3 to 7 days before delivery) (pretest probability 81.8%) P hosphatidylcholine phosphorus
No significant RDS
<2 2-6
3 11 31
>6
Significant RDS 7 3
o
Likelihood ratio
Posttest probability of no RDS (%)
0.10 0.81 Infinity
30 79 100
3 c d ~
20
Positive predictive value = 89%. Negative predictive value = 56%. Sensitivity = 61 %. Specificity = 87%. False positive rate = 13%. False negative rate = 39%. Accuracy = 70%. Prevalence = 64.1 %. Pretest probability = 64.1 %. Likelihood ratio = 4.7. Posttest probability = 89%. *Significant RDS = moderate or severe RDS.
nus specificity). 12. 1\ It can also be calculated for various levels of diagnostic test result. 1:< To provide a more meaningful expression for clinicians, the likelihood ratio can also be used to determine the probability of the target outcome (no RDS) occurring for a given test result (that is, presence of phosphatidylglycerol or level of phosphatidylcholine phosphorus). To do this, one must have some estimate of the probability of the target outcome occurring prior to obtaining of the test result (pretest probability). In clinical obstetric practice, this might be based on experience of outcomes at similar gestations. For our present calculations, the prevalence of the target outcome in the group under study was used. This pretest probability of not developing RDS and the appropriate likelihood ratio were plotted on a nomogram 11 to determine the posttest probability, that is, the probability of a patient not developing RDS with a given diagnostic test result.
Results A total of 666 profiles were obtained from amniotic
fluid within 2 days before delivery. Phosphatidylglycerol was found to be present in 602. Ten cases of RDS were found in the presence of phosphatidylglycerol.
Only three cases of significant (moderate to severe) RDS were found and these infants were severely asphyxiated at birth. In 64 cases in which phosphatidylglycerol was absent, there were 36 cases of RDS. The pretest probability of not developing RDS was 93.1 % and the likelihood ratio was calculated to be 4.4. The posttest probability of not developing RDS when phosphatidylglycerol was present was therefore 99% (Table I). The phosphatidylcholine phosphorus values were evaluated in the 64 cases in which the phospholipid profile revealed phosphatidylglycerol to be absent within 2 days preceding delivery. When phosphatidylcholine phosphorus was ",,6 (28 cases), II % developed moderate or severe RDS, 25% developed mild RDS, and 64% did not develop RDS. When phosphatidylcholine phosphorus was <6 only 28% did not develop RDS while 17% developed mild RDS and 55% developed moderate or severe RDS. The pretest probability of not developing RDS when phosphatidylglycerol was absent within 2 days of delivery was 43.1 %. The likelihood ratio was 2.3. From the nomogram, the posttest probability of not developing RDS when the phosphatidylcholine phosphorus was ",,6 was 69.0% (Table II). Because of the benign nature of mild RDS, the data were similarly assessed with only moderate and severe RDS considered to be significant disease. The pretest probability of not developing significant RDS was 64.1 %. The likelihood ratio was 4.7 and therefore the posttest probability of not developing significant RDS when phosphatidylcholine phosphorus was ~6 was 89% (Table III). There were 55 cases in which phosphatidylglycerol was absent 3 to 7 days before delivery. These were
Volume 154 Number 2
analyzed for phosphatidylcholine phosphorus and the data for phosphatidylcholine phosphorus values of <2, 2 to 6, and >6 are presented in Table IV. The calculated likelihood ratios were found to increase as phosphatidylcholine phosphorus levels increased. Consequently, the probability of the infant not developing RDS rose from only 19% when phosphatidylcholine phosphorus was <2 to 94% when phosphatidylcholine phosphorus was >6 between 3 and 7 days before delivery. The data pertaining to significant (moderate or severe) RDS are presented in Table V. Since there were no instances of significant RDS in any of the 31 cases where phosphatidylcholine phosphorus was >6 when measured 3 to 7 days before delivery, the likelihood ratio was calculated as infinity and the posttest probability of not developing moderate or severe RDS approached 100%. Of the 64 cases with absent phosphatidylglycerol, 10 were in insulin-dependent diabetic women. Four had phosphatidylcholine phosphorus measured within 2 days of delivery. There was one case of moderate RDS, the phosphatidylcholine phosphorus being 1.98, in a Class C diabetic patient (White's classification). The other three all had phosphatidylcholine phosphorus >6 and none developed RDS. Six patients with diabetes had phosphatidylcholine phosphorus measured 3 to 7 days before delivery. There was one case of moderate RDS occurring in a Class B diabetic patient with a phosphatidylcholine phosphorus value of 1.38 6 days before delivery. The remaining five all had phosphatidylcholine phosphorus >6 and none developed RDS. However, the aim was not the investigation of the status of diabetes in relation to pulmonary maturity in the fetus. The numbers are too small to make any significant observations or conclusions but the group was looked at separately simply as an interesting observation. Predictability of outcome from the test result did not change when the patients were diabetic. Only 21 of 64 patients in the study group received steroids; no trends were noted and no conclusions could be made with these small numbers. As expected, the majority of patients who developed significant RDS were :s;34 weeks' gestation. Ten of 24 patients >34 weeks' gestation developed significant RDS when phosphatidylglycerol was absent. This reaffirmed the need for pulmonary maturity testing. Mode of delivery was assessed as well, but no difference could be demonstrated clinically or statistically among patients who had labor followed by spontaneous vaginal delivery, labor followed by cesarean section, or cesarean section without labor.
Comment Antenatal determination of fetal pulmonary maturity is an important adjunct to optimum obstetric care.
Predictors of fetal lung maturity 229
Good pregnancy dating is aided by first- and secondtrimester ultrasonography. Obviously, if the indication for delivery is such that the risk of significant RDS is less than the risk of intrauterine death, testing for fetal pulmonary maturity is not indicated. However, there are many situations where either gestation is not known or the intrauterine environment is considered to be sufficiently hostile that the best interest of the fetus would be served by delivery before term, provided fetal pulmonary maturity is assured. It is in these circumstances one seeks a predictor of the risk of RDS in the newborn infant. Phosphatidylglycerol is the second most abundant phospholipid in mature surfactant, and, while most investigators"·7.9 agree that its presence in amniotic fluid is indicative of lung maturity, many"' advocate the use of this measurement in combination with some other test of lung maturity, most notably the lecithin/sphingomyelin ratio. Since we have found that the lecithin/ sphingomyelin ratio provided no additional information to our method of phosphatidylglycerol determination,'O the lecithin/sphingomyelin ratio has been discontinued at this institution since 1980. The updated data presented herein support the conclusion that the presence of phosphatidylglycerol in the isolated surfactant fraction accurately predicts an absence of RDS. It is likely that no test will ever be absolutely predictive, in that RDS may develop in mature fetuses for a number of reasons, in particular, asphyxia neonatorum and acidosis. It is noteworthy that all three cases of signif~ icant RDS from the group with phosphatidylglycerol present occurred in infants that were severely asphyxiated at birth. In predicting the occurrence of RDS in the neonate by a lack of phosphatidylglycerol in amniotic fluid, the test appears to be less reliable. This has been demonstrated by other investigators'"'·7 and, although we have found our phosphatidylglycerol method to be more reliable than the lecithin/sphingomyelin ratio in predicting lung immaturity,'O both our previous and updated data indicate instances in which phosphatidylglycerol is absent and the infant does not develop RDS. As mentioned in a previous report," it is important to note that phosphatidylglycerol is present in the surfactant fraction of amniotic fluid and is often present in very low quantities, particularly just after its initial appearance. We have often found that in order to detect these low levels a 10.0 ml aliquot of amniotic fluid must be used. Using smaller volumes of amniotic fluid or using any steps in the methodology that would decrease these already low levels of phosphatidylglycerol, that is, centrifugation beyond 140 x g to clear the sample of cells,15 acetone precipitation (Oulton M., unpublished observation), or quantitation by densitometry, It> would diminish the likelihood of detecting these low levels. These factors pose less of a problem in detecting
230
Kogon et al.
phosphatidylglycerol when it is present in much larger quantities. They do, however, provide an explanation as to why some investigators 2 . 3. 7. 8 often fail to detect phosphatidylglycerol until other tests, such as the lecithin/sphingomyelin ratio, are mature. The high false negative rate of the phosphatidylglycerol test reported by other centers' 57 (which may be as high as 70% to 80%) can be attributed to these effects. It is felt that these problems can be minimized by determining phosphatidylglycerol on the isolated surfactant fractions as described in this and previous reports. 10. II As the appearance of phosphatidylglycerol is somewhat abrupt,11 it is not possible to predict with 100% accuracy exactly when it will appear. Also, there may be some delay in its appearance in amniotic fluid after its production and release to alveoli. When phosphatidylglycerol is not present in amniotic fluid, it would then be useful to have some other index of the lung maturation process. We have previously shown that the measurement of the phosphatidylcholine phosphorus value of the isolated 10,000 x g pellet provides an index of surfactant accumulation and that the initial rise in this value precedes the appearance of phosphatidylglycerol. II Therefore, we assessed the usefulness of this value in determining fetal lung maturity in instances where phosphatidylglycerol was not detectable. If the need for delivery is considered urgent and mild RDS is not a major concern, then a phosphatidylcholine phosphorus vlue of ~6 would be a strong predictor that the fetal outcome will be satisfactory insofar as RDS is concerned. There is an 89% probability that significant RDS will not occur in our setting. If delivery is not so urgent, the phosphatidylcholine phosphorus value may be even more helpful. In our study there was a 94% probability of the neonate not developing RDS if phosphatidylcholine phosphorus was >6 3 to 7 days before delivery. If mild RDS is not a concern, then the probability of avoiding significant RDS approaches 100% when the phosphatidylcholine phosphorus is >6. It is reasonable to assume that if a full 7 days were to elapse from amniocentesis to delivery, the results would only improve. The test would then be of twofold benefit: It would allow a delivery date to be set and it would avoid the need for a repeat amniocentesis with its associated risk. No antenatal test is 100% predictive of the presence or absence of neonatal RDS. Peripartum factors will at times change the outcome from that which is predicted. The presence of phosphatidylglycerol in amniotic fluid is an accurate predictor of fetal pulmonary maturity. To ensure the detection of the very low levels of phosphatidylglycerol that are often present, we recommend its analysis on the isolated surfactant fraction.
February, 1986 Am J Obstet Gynecol
When phosphatidylglycerol is absent, phosphatidylcholine phosphorus is of clinical value as a predictor of fetal pulmonary maturity. The technical assistance of Marlene Fraser is gratefully acknowledged. REFERENCES I. Clements TA, Platzker ACG, Tierney DF, et al. Assessment of the risk of the respiratory distress syndrome by a rapid test for surfactant in amniotic fluid. N Engl] Med 1972;286: 1077. 2. Kulovich MV, Hallman M, Gluck L. The lung profile. I. Normal pregnancy. AM] OBSTET GYNECOL 1979;135:57. 3. Hallman M, Teramo K. Measurement of the lecithin/ sphingomyelin ratio and phosphatidylglycerol in amniotic fluid: an accurate method for the assessment of fetal lung maturity. Br] Obstet Gynaecol 1981 ;88:806. 4. Whittle M], Wilson AI, Whitfield CR, Paton RD, Logan RW. Amniotic fluid phosphatidylglycerol and the lecithin/ sphingomyelin ratio in the assessment of fetal lung maturity. Br] Obstet Gynecol 1982;89:727. 5. Hamilton PR, Hauschild D, Broekhuizen FF, Beck RM. Comparison of lecithin: sphingomyelin ratio, fluorescence polarization and phosphatidylglycerol in the amniotic fluid in the prediction of respiratory distress syndrome. Obstet Gynecol 1984;63:52. 6. Garite T], Yabusaki KK, Moberg LJ, et al. A new rapid slide agglutination test for amniotic fluid phosphatidylglycerol: laboratory and clinical correlation. AM] OBSTET GY:\ECOL 1983;147:681. 7. Tsai MY, Shultz EK, Nelson]A. Amniotic fluid phosphatidylglycerol in diabetic and control pregnant patients at different gestational lengths. AM ] OBSTET GY:'>iECOL 1984; 149:388. 8. Amenta ]S, Silverman ]A. Amniotic fluid lecithin, phosphatidylglycerol, lecithin/sphingomyelin ratio and foam stability test in predicting respiratory distress in the newborn. Am] Clin Pathol 1983;79:52. 9. Moodley S, Liu ]H, Cherkis RC, Miller RH, Merritt TA, Whalen LE. Fetal pulmonary maturity: relationship between optical density (650 nm) to the lecithin/sphingomyelin ratio and phosphatidylglycerol in amniotic fluid. Br] Obstet Gynaecol 1983;21: 199. 10. Bent AE, Gray]H, Luther ER, Oulton M, Peddle LJ. Phosphatidylglycerol determination on amniotic fluid 10,000 x g pellet in the prediction of fetal lung maturity. AM] OBSTET GYNECOL 1981;139:259. II. Oulton M, Bent AE, Gray ]H, Luther ER, Peddle LJ. Assessment of fetal pulmonary maturity by phospholipid analysis of amniotic fluid lamellar bodies. AM] OBSTET GY:'-IECOL 1982;142:684. 12. Department of Clinical Epidemiology and Biostatistics, McMaster University, Hamilton, Ontario, Canada. Clinical Epidemiology rounds. Interpretation of diagnostic data: 5. How to do it with simple maths. Can Med Assoc ] 1983;129:947. 13. Sackett DL, Haynes RB, Tugwell P. Clinical epidemiology. A basic science for clinical medicine. Boston: Little, Brown & Co., 1985:108-26. 14. Fagan TJ. Nomogram for Bayes' theorem (C). N Engl] Med 1975;293:257. 15. Oulton M. The role of centrifugation in the measurement of surfactant in amniotic fluid. AM ] OBSTET GYNECOL 1979;135:337. 16. Skjaeraasen], Stray-Pedersen S. Amniotic fluid phosphatidylinositol and phosphatidylglycerol. I. Normal pregnancies. Acta Obstet Gynecol Scand 1979;58:225.
Key words: Amniotic fluid, fetal lung maturity, phosphatidylglycerol, phosphatidyicholine phosphorus, respiratory distress syndrome
In a variety of obstetric settings delivery of a fetus before term is desirable. However, the benefit sought by preterm delivery must be measured against the risk of neonatal respiratory distress syndrome (RDS) and its related morbidity and mortality. Several methods of amniotic fluid analysis such as the lecithin/sphingomyelin ratio, the shake test, and, more recently, phosphatidylglycerol determinations are currently available for the assessment of fetal lung maturity."" Some investigators 2·5 have reported that a combination of tests, for instance, the lecithin/sphingomyelin ratio and phosphatidylglycerol determination, improve the predictive accuracy of any single test and, therefore, advocate such a combination of testing. In our center we have been performing phospholipid analyses on isolated surfactant fractions of amniotic fluid (10,000 X g pellets) since 1978, and, as previously reported,'O we found that the phosphatidylglycerol determination on these fractions was superior to both the lecithin/sphingomyelin ratio and the shake test in predicting fetal pulmonary maturity as well as immaturity. We have accepted the presence of phosphatidylglycerol in these fractions as being indicative of fetal pulmonary maturity but have also found many instances in which
From the Division of Perinatal Medicine, Department of Obstetrics and Gynaecology, Dalhousie University. Presented at the Forty-first Annual Meeting of The Society of Obstetricians and Gynaecologists of Canada, jasper, Alberta, Canada, june 10-15, 1985. Reprint requests: Dr. D. P. Kogon, Division of Perinatal Medicine, Department of Obstetrics and Gynaecology, Dalhousie University, 5821 University Ave., Halifax, Nova Scotia, Canada B3H 1W3.
226
the absence of phosphatidylglycerol was not predictive of the development of RDS. We subsequently demonstrated that the measurement of the phosphatidyicholine content of the same isolated surfactant fractions of amniotic fluid provides an index for the rate of surfactant accumulation." With this measurement expressed as the phosphatidyicholine phosphorus value (phosphatidyicholine phosphorus content of the surfactant present in 10 ml of amniotic fluid), it was found that this value progressively increased before the appearance of phosphatidylglycero!. The hypothesis that this measurement may be of value in assessing the state of fetal pulmonary maturity in the absence of phosphatidylglycerol was therefore addressed. Material and methods A review of all phospholipid profiles analyzed between 1978 and 1983 was made. Clinical variables such as intrauterine growth retardation, premature rupture of the membranes, diabetes mellitus, and the use of steroids were not excluding factors. Phospholipid profiles were measured on isolated surfactant fractions as previously described. 10. II Briefly, this involves centrifugation of a carefully measured aliquot (preferably 5 to 10 ml) of each amniotic fluid sample for 5 minutes at 140 X g to remove cellular debris followed by a 20minute spin at 10,000 X g to pellet the surfactant particles. Lipids were then extracted from the surfactant pellets and separated by two-dimensional thin-layer chromatography. The individual phospholipids, after visualization with the use of a modified molybdenum
Predictors of fetal lung maturity 227
Volume 154 Number 2
spray reagent, were scraped from the plate and quantitated by measurement of their phosphorus content. When volumes other than 10.0 ml were used, the phosphorus content of phosphatidylcholine was appropriately corrected to give the phosphatidylcholine phosphorus value. Phosphatidylglycerol is routinely calculated as a percentage of total lipid phosphorus but is reported in this communication as either present or absent. By this method the limit of detection for phosphatidylglycerol is 0.2 to 0.5 j.Lg. All amniocenteses were done to assess fetal lung maturity based on clinical indicators. None was done solely to collect data for the study. The decision to perform the amniocentesis was made by the individual attending obstetrician. All patient management decisions were also made by the attending obstetrician. In the earlier years of study, the shake test and the lecithin/sphingomyelin ratio were the standard accepted tests and were the basis for obstetric management decisions. After initial work on phosphatidylglycerol by Bent et aI., 10 phosphatidylglycerol became the standard test of fetal lung maturity at this institution. At no time was a phosphatidylcholine phosphorus value used to direct obstetric management of the patients in the study group. The study group then comprised patients who had fetal pulmonary testing performed ante natally for any indication regardless of that indication or other variables. The only exclusions were in patients with multiple gestations or patients whose infants died so soon after birth that RDS could not possibly develop. Clinical findings, indication for amniocentesis, gestational age, and neonatal outcome were recorded. All data were collected from the Grace Maternity Hospital obstetric population, which accounts for 5000 to 5500 deliveries per year. This hospital is the sole maternity unit for metropolitan Halifax and is the tertiary care obstetric center for the province of Nova Scotia and elsewhere in the Atlantic region. A total of 1678 amniotic fluid samples from 1359 patients were analyzed. Blood and/or meconium in the sample did not affect the assay. To reevaluate phosphatidylglycerol as a predictive test only transabdominal samples of amniotic fluid taken within 2 days prior to delivery were considered. To assess phosphatidylcholine phosphorus, the same criteria were applied. However, only samples where phosphatidylglycerol was absent were considered. Because of the steady rise in phosphatidylcholine phosphorus that occurs prior to the appearance of phosphatidylglycerol, the time frame in this group was extended to include samples collected 3 to 7 days before delivery as well as those within 2 days of delivery. RDS was diagnosed by the following criteria: grunting, tachypnea, sternal and intercostal retractions, and nasal flaring of ~6 hours' duration and requiring treatment with oxygen. Severity was classified by degree of
Table I. RDS in relation to phosphatidylglycerol determination «2 days before delivery) Phosphatidylglycerol
NoRDS
RDS
Present
592
10
Absent
28
Positive predictive value
= _a_ = a + b
Negative predictive value '" Sensltlvlty
_d_ c + d
=
c d ~
36
98.3%.
=
56.3%.
= -a- = 95.5 0170. a + c
Specificity = _d_ = 78.3%. b + d b False positive rate = - - = 21.7%. b + d .
c
False negative rate
Accuracy
=
Prevalence
= -- =
a + c
a+d a+b+c+d
=
a +
4.5%.
= 94.3%.
a + c b d +c +
= 93.1 %.
=
. . LIkelIhood ratio
sensitivity f' .. alse pOSItive rate
=
prevalence
93.1 %.
Pretest probability
=
= 4.4.
Posttest probability (from nomogram) = 99%.
respiratory assistance required in treatment. Severe RDS required assisted mechanical ventilation, moderate RDS required >35% oxygen and/or continuous positive airway pressure, and mild RDS required only an increase in oxygen concentration up to 35%. Diagnosis of RDS and its severity was made by the attending neonatalogist who was not aware of the results of the phospholipid profile. Analysis of the data was made with the use of standard diagnostic test parameters, that is, positive predictive value, negative predictive value, sensitivity, specificity, and accuracy. To further assess the predictive ability of the phosphatidylglycerol and phosphatidylcholine phosphorus measurements, likelihood ratios, and probability of RDS occurring or not occurring were determined. The key to the analysis is the likelihood ratio, which expresses the odds that a given diagnostic test result (presence of phosphatidylglycerol or level of phosphatidylcholine phosphorus) would be expected in a patient with the target outcome (no RDS). This is calculated in a two-by-two table as the true positive rate (sensitivity), divided by the false positive rate (one mi-
228
Kogan et al.
February, 1986 Am J Obstet Gyneco1
Table IV. RDS in relation to phosphatidylcholine phosphorus (3 to 7 days before delivery) (pretest probability 74.5%)
Table II. RDS in relation to phosphatidylcholine phosphorus «2 days before delivery) Phosphatidylcholine phosphorus
No RDS
RDS
2:6
18
10
Phosphatidylcholine phosphorus
<6
10
26
<2 2-6
c d ~
>6 Positive predictive value = 64%. Negative predictive value = 72%. Sensitivity = 64.3%. Specificity = 72%. False positive rate = 28%. False negative rate = 36%. Accuracy = 69%. Prevalence = 43.8%. Pretest probability = 43.8%. Likelihood ratio = 2.3. Posttest probability = 69%.
Table III. Significant RDS* in relation to phosphotidylcholine phosphorus «2 days before delivery) P Iwsphatidylcholine phosphorus
No significant RDS
2:6
25
<6
16
Significant RDS
No RDS
RDS
Likelihood ratio
2 10 29
8 4 2
0.09 0.85 4.9
Postlest probability of no RDS (%) 19 69 94
Table V. Significant RDS in relation to phosphatidylcholine phosphorus (3 to 7 days before delivery) (pretest probability 81.8%) P hosphatidylcholine phosphorus
No significant RDS
<2 2-6
3 11 31
>6
Significant RDS 7 3
o
Likelihood ratio
Posttest probability of no RDS (%)
0.10 0.81 Infinity
30 79 100
3 c d ~
20
Positive predictive value = 89%. Negative predictive value = 56%. Sensitivity = 61 %. Specificity = 87%. False positive rate = 13%. False negative rate = 39%. Accuracy = 70%. Prevalence = 64.1 %. Pretest probability = 64.1 %. Likelihood ratio = 4.7. Posttest probability = 89%. *Significant RDS = moderate or severe RDS.
nus specificity). 12. 1\ It can also be calculated for various levels of diagnostic test result. 1:< To provide a more meaningful expression for clinicians, the likelihood ratio can also be used to determine the probability of the target outcome (no RDS) occurring for a given test result (that is, presence of phosphatidylglycerol or level of phosphatidylcholine phosphorus). To do this, one must have some estimate of the probability of the target outcome occurring prior to obtaining of the test result (pretest probability). In clinical obstetric practice, this might be based on experience of outcomes at similar gestations. For our present calculations, the prevalence of the target outcome in the group under study was used. This pretest probability of not developing RDS and the appropriate likelihood ratio were plotted on a nomogram 11 to determine the posttest probability, that is, the probability of a patient not developing RDS with a given diagnostic test result.
Results A total of 666 profiles were obtained from amniotic
fluid within 2 days before delivery. Phosphatidylglycerol was found to be present in 602. Ten cases of RDS were found in the presence of phosphatidylglycerol.
Only three cases of significant (moderate to severe) RDS were found and these infants were severely asphyxiated at birth. In 64 cases in which phosphatidylglycerol was absent, there were 36 cases of RDS. The pretest probability of not developing RDS was 93.1 % and the likelihood ratio was calculated to be 4.4. The posttest probability of not developing RDS when phosphatidylglycerol was present was therefore 99% (Table I). The phosphatidylcholine phosphorus values were evaluated in the 64 cases in which the phospholipid profile revealed phosphatidylglycerol to be absent within 2 days preceding delivery. When phosphatidylcholine phosphorus was ",,6 (28 cases), II % developed moderate or severe RDS, 25% developed mild RDS, and 64% did not develop RDS. When phosphatidylcholine phosphorus was <6 only 28% did not develop RDS while 17% developed mild RDS and 55% developed moderate or severe RDS. The pretest probability of not developing RDS when phosphatidylglycerol was absent within 2 days of delivery was 43.1 %. The likelihood ratio was 2.3. From the nomogram, the posttest probability of not developing RDS when the phosphatidylcholine phosphorus was ",,6 was 69.0% (Table II). Because of the benign nature of mild RDS, the data were similarly assessed with only moderate and severe RDS considered to be significant disease. The pretest probability of not developing significant RDS was 64.1 %. The likelihood ratio was 4.7 and therefore the posttest probability of not developing significant RDS when phosphatidylcholine phosphorus was ~6 was 89% (Table III). There were 55 cases in which phosphatidylglycerol was absent 3 to 7 days before delivery. These were
Volume 154 Number 2
analyzed for phosphatidylcholine phosphorus and the data for phosphatidylcholine phosphorus values of <2, 2 to 6, and >6 are presented in Table IV. The calculated likelihood ratios were found to increase as phosphatidylcholine phosphorus levels increased. Consequently, the probability of the infant not developing RDS rose from only 19% when phosphatidylcholine phosphorus was <2 to 94% when phosphatidylcholine phosphorus was >6 between 3 and 7 days before delivery. The data pertaining to significant (moderate or severe) RDS are presented in Table V. Since there were no instances of significant RDS in any of the 31 cases where phosphatidylcholine phosphorus was >6 when measured 3 to 7 days before delivery, the likelihood ratio was calculated as infinity and the posttest probability of not developing moderate or severe RDS approached 100%. Of the 64 cases with absent phosphatidylglycerol, 10 were in insulin-dependent diabetic women. Four had phosphatidylcholine phosphorus measured within 2 days of delivery. There was one case of moderate RDS, the phosphatidylcholine phosphorus being 1.98, in a Class C diabetic patient (White's classification). The other three all had phosphatidylcholine phosphorus >6 and none developed RDS. Six patients with diabetes had phosphatidylcholine phosphorus measured 3 to 7 days before delivery. There was one case of moderate RDS occurring in a Class B diabetic patient with a phosphatidylcholine phosphorus value of 1.38 6 days before delivery. The remaining five all had phosphatidylcholine phosphorus >6 and none developed RDS. However, the aim was not the investigation of the status of diabetes in relation to pulmonary maturity in the fetus. The numbers are too small to make any significant observations or conclusions but the group was looked at separately simply as an interesting observation. Predictability of outcome from the test result did not change when the patients were diabetic. Only 21 of 64 patients in the study group received steroids; no trends were noted and no conclusions could be made with these small numbers. As expected, the majority of patients who developed significant RDS were :s;34 weeks' gestation. Ten of 24 patients >34 weeks' gestation developed significant RDS when phosphatidylglycerol was absent. This reaffirmed the need for pulmonary maturity testing. Mode of delivery was assessed as well, but no difference could be demonstrated clinically or statistically among patients who had labor followed by spontaneous vaginal delivery, labor followed by cesarean section, or cesarean section without labor.
Comment Antenatal determination of fetal pulmonary maturity is an important adjunct to optimum obstetric care.
Predictors of fetal lung maturity 229
Good pregnancy dating is aided by first- and secondtrimester ultrasonography. Obviously, if the indication for delivery is such that the risk of significant RDS is less than the risk of intrauterine death, testing for fetal pulmonary maturity is not indicated. However, there are many situations where either gestation is not known or the intrauterine environment is considered to be sufficiently hostile that the best interest of the fetus would be served by delivery before term, provided fetal pulmonary maturity is assured. It is in these circumstances one seeks a predictor of the risk of RDS in the newborn infant. Phosphatidylglycerol is the second most abundant phospholipid in mature surfactant, and, while most investigators"·7.9 agree that its presence in amniotic fluid is indicative of lung maturity, many"' advocate the use of this measurement in combination with some other test of lung maturity, most notably the lecithin/sphingomyelin ratio. Since we have found that the lecithin/ sphingomyelin ratio provided no additional information to our method of phosphatidylglycerol determination,'O the lecithin/sphingomyelin ratio has been discontinued at this institution since 1980. The updated data presented herein support the conclusion that the presence of phosphatidylglycerol in the isolated surfactant fraction accurately predicts an absence of RDS. It is likely that no test will ever be absolutely predictive, in that RDS may develop in mature fetuses for a number of reasons, in particular, asphyxia neonatorum and acidosis. It is noteworthy that all three cases of signif~ icant RDS from the group with phosphatidylglycerol present occurred in infants that were severely asphyxiated at birth. In predicting the occurrence of RDS in the neonate by a lack of phosphatidylglycerol in amniotic fluid, the test appears to be less reliable. This has been demonstrated by other investigators'"'·7 and, although we have found our phosphatidylglycerol method to be more reliable than the lecithin/sphingomyelin ratio in predicting lung immaturity,'O both our previous and updated data indicate instances in which phosphatidylglycerol is absent and the infant does not develop RDS. As mentioned in a previous report," it is important to note that phosphatidylglycerol is present in the surfactant fraction of amniotic fluid and is often present in very low quantities, particularly just after its initial appearance. We have often found that in order to detect these low levels a 10.0 ml aliquot of amniotic fluid must be used. Using smaller volumes of amniotic fluid or using any steps in the methodology that would decrease these already low levels of phosphatidylglycerol, that is, centrifugation beyond 140 x g to clear the sample of cells,15 acetone precipitation (Oulton M., unpublished observation), or quantitation by densitometry, It> would diminish the likelihood of detecting these low levels. These factors pose less of a problem in detecting
230
Kogon et al.
phosphatidylglycerol when it is present in much larger quantities. They do, however, provide an explanation as to why some investigators 2 . 3. 7. 8 often fail to detect phosphatidylglycerol until other tests, such as the lecithin/sphingomyelin ratio, are mature. The high false negative rate of the phosphatidylglycerol test reported by other centers' 57 (which may be as high as 70% to 80%) can be attributed to these effects. It is felt that these problems can be minimized by determining phosphatidylglycerol on the isolated surfactant fractions as described in this and previous reports. 10. II As the appearance of phosphatidylglycerol is somewhat abrupt,11 it is not possible to predict with 100% accuracy exactly when it will appear. Also, there may be some delay in its appearance in amniotic fluid after its production and release to alveoli. When phosphatidylglycerol is not present in amniotic fluid, it would then be useful to have some other index of the lung maturation process. We have previously shown that the measurement of the phosphatidylcholine phosphorus value of the isolated 10,000 x g pellet provides an index of surfactant accumulation and that the initial rise in this value precedes the appearance of phosphatidylglycerol. II Therefore, we assessed the usefulness of this value in determining fetal lung maturity in instances where phosphatidylglycerol was not detectable. If the need for delivery is considered urgent and mild RDS is not a major concern, then a phosphatidylcholine phosphorus vlue of ~6 would be a strong predictor that the fetal outcome will be satisfactory insofar as RDS is concerned. There is an 89% probability that significant RDS will not occur in our setting. If delivery is not so urgent, the phosphatidylcholine phosphorus value may be even more helpful. In our study there was a 94% probability of the neonate not developing RDS if phosphatidylcholine phosphorus was >6 3 to 7 days before delivery. If mild RDS is not a concern, then the probability of avoiding significant RDS approaches 100% when the phosphatidylcholine phosphorus is >6. It is reasonable to assume that if a full 7 days were to elapse from amniocentesis to delivery, the results would only improve. The test would then be of twofold benefit: It would allow a delivery date to be set and it would avoid the need for a repeat amniocentesis with its associated risk. No antenatal test is 100% predictive of the presence or absence of neonatal RDS. Peripartum factors will at times change the outcome from that which is predicted. The presence of phosphatidylglycerol in amniotic fluid is an accurate predictor of fetal pulmonary maturity. To ensure the detection of the very low levels of phosphatidylglycerol that are often present, we recommend its analysis on the isolated surfactant fraction.
February, 1986 Am J Obstet Gynecol
When phosphatidylglycerol is absent, phosphatidylcholine phosphorus is of clinical value as a predictor of fetal pulmonary maturity. The technical assistance of Marlene Fraser is gratefully acknowledged. REFERENCES I. Clements TA, Platzker ACG, Tierney DF, et al. Assessment of the risk of the respiratory distress syndrome by a rapid test for surfactant in amniotic fluid. N Engl] Med 1972;286: 1077. 2. Kulovich MV, Hallman M, Gluck L. The lung profile. I. Normal pregnancy. AM] OBSTET GYNECOL 1979;135:57. 3. Hallman M, Teramo K. Measurement of the lecithin/ sphingomyelin ratio and phosphatidylglycerol in amniotic fluid: an accurate method for the assessment of fetal lung maturity. Br] Obstet Gynaecol 1981 ;88:806. 4. Whittle M], Wilson AI, Whitfield CR, Paton RD, Logan RW. Amniotic fluid phosphatidylglycerol and the lecithin/ sphingomyelin ratio in the assessment of fetal lung maturity. Br] Obstet Gynecol 1982;89:727. 5. Hamilton PR, Hauschild D, Broekhuizen FF, Beck RM. Comparison of lecithin: sphingomyelin ratio, fluorescence polarization and phosphatidylglycerol in the amniotic fluid in the prediction of respiratory distress syndrome. Obstet Gynecol 1984;63:52. 6. Garite T], Yabusaki KK, Moberg LJ, et al. A new rapid slide agglutination test for amniotic fluid phosphatidylglycerol: laboratory and clinical correlation. AM] OBSTET GY:\ECOL 1983;147:681. 7. Tsai MY, Shultz EK, Nelson]A. Amniotic fluid phosphatidylglycerol in diabetic and control pregnant patients at different gestational lengths. AM ] OBSTET GY:'>iECOL 1984; 149:388. 8. Amenta ]S, Silverman ]A. Amniotic fluid lecithin, phosphatidylglycerol, lecithin/sphingomyelin ratio and foam stability test in predicting respiratory distress in the newborn. Am] Clin Pathol 1983;79:52. 9. Moodley S, Liu ]H, Cherkis RC, Miller RH, Merritt TA, Whalen LE. Fetal pulmonary maturity: relationship between optical density (650 nm) to the lecithin/sphingomyelin ratio and phosphatidylglycerol in amniotic fluid. Br] Obstet Gynaecol 1983;21: 199. 10. Bent AE, Gray]H, Luther ER, Oulton M, Peddle LJ. Phosphatidylglycerol determination on amniotic fluid 10,000 x g pellet in the prediction of fetal lung maturity. AM] OBSTET GYNECOL 1981;139:259. II. Oulton M, Bent AE, Gray ]H, Luther ER, Peddle LJ. Assessment of fetal pulmonary maturity by phospholipid analysis of amniotic fluid lamellar bodies. AM] OBSTET GY:'-IECOL 1982;142:684. 12. Department of Clinical Epidemiology and Biostatistics, McMaster University, Hamilton, Ontario, Canada. Clinical Epidemiology rounds. Interpretation of diagnostic data: 5. How to do it with simple maths. Can Med Assoc ] 1983;129:947. 13. Sackett DL, Haynes RB, Tugwell P. Clinical epidemiology. A basic science for clinical medicine. Boston: Little, Brown & Co., 1985:108-26. 14. Fagan TJ. Nomogram for Bayes' theorem (C). N Engl] Med 1975;293:257. 15. Oulton M. The role of centrifugation in the measurement of surfactant in amniotic fluid. AM ] OBSTET GYNECOL 1979;135:337. 16. Skjaeraasen], Stray-Pedersen S. Amniotic fluid phosphatidylinositol and phosphatidylglycerol. I. Normal pregnancies. Acta Obstet Gynecol Scand 1979;58:225.